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.
61 * Note that the offset of fid_data is 4 bytes, so care must be taken to avoid
62 * undefined behavior accessing unaligned fields within an embedded struct.
67 u_short fid_len; /* length of data in bytes */
68 u_short fid_data0; /* force longword alignment */
69 char fid_data[MAXFIDSZ]; /* data (variable length) */
73 * filesystem statistics
75 #define MFSNAMELEN 16 /* length of type name including null */
76 #define MNAMELEN 1024 /* size of on/from name bufs */
77 #define STATFS_VERSION 0x20140518 /* current version number */
79 uint32_t f_version; /* structure version number */
80 uint32_t f_type; /* type of filesystem */
81 uint64_t f_flags; /* copy of mount exported flags */
82 uint64_t f_bsize; /* filesystem fragment size */
83 uint64_t f_iosize; /* optimal transfer block size */
84 uint64_t f_blocks; /* total data blocks in filesystem */
85 uint64_t f_bfree; /* free blocks in filesystem */
86 int64_t f_bavail; /* free blocks avail to non-superuser */
87 uint64_t f_files; /* total file nodes in filesystem */
88 int64_t f_ffree; /* free nodes avail to non-superuser */
89 uint64_t f_syncwrites; /* count of sync writes since mount */
90 uint64_t f_asyncwrites; /* count of async writes since mount */
91 uint64_t f_syncreads; /* count of sync reads since mount */
92 uint64_t f_asyncreads; /* count of async reads since mount */
93 uint64_t f_spare[10]; /* unused spare */
94 uint32_t f_namemax; /* maximum filename length */
95 uid_t f_owner; /* user that mounted the filesystem */
96 fsid_t f_fsid; /* filesystem id */
97 char f_charspare[80]; /* spare string space */
98 char f_fstypename[MFSNAMELEN]; /* filesystem type name */
99 char f_mntfromname[MNAMELEN]; /* mounted filesystem */
100 char f_mntonname[MNAMELEN]; /* directory on which mounted */
103 #if defined(_WANT_FREEBSD11_STATFS) || defined(_KERNEL)
104 #define FREEBSD11_STATFS_VERSION 0x20030518 /* current version number */
105 struct freebsd11_statfs {
106 uint32_t f_version; /* structure version number */
107 uint32_t f_type; /* type of filesystem */
108 uint64_t f_flags; /* copy of mount exported flags */
109 uint64_t f_bsize; /* filesystem fragment size */
110 uint64_t f_iosize; /* optimal transfer block size */
111 uint64_t f_blocks; /* total data blocks in filesystem */
112 uint64_t f_bfree; /* free blocks in filesystem */
113 int64_t f_bavail; /* free blocks avail to non-superuser */
114 uint64_t f_files; /* total file nodes in filesystem */
115 int64_t f_ffree; /* free nodes avail to non-superuser */
116 uint64_t f_syncwrites; /* count of sync writes since mount */
117 uint64_t f_asyncwrites; /* count of async writes since mount */
118 uint64_t f_syncreads; /* count of sync reads since mount */
119 uint64_t f_asyncreads; /* count of async reads since mount */
120 uint64_t f_spare[10]; /* unused spare */
121 uint32_t f_namemax; /* maximum filename length */
122 uid_t f_owner; /* user that mounted the filesystem */
123 fsid_t f_fsid; /* filesystem id */
124 char f_charspare[80]; /* spare string space */
125 char f_fstypename[16]; /* filesystem type name */
126 char f_mntfromname[88]; /* mounted filesystem */
127 char f_mntonname[88]; /* directory on which mounted */
129 #endif /* _WANT_FREEBSD11_STATFS || _KERNEL */
132 #define OMFSNAMELEN 16 /* length of fs type name, including null */
133 #define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */
135 /* XXX getfsstat.2 is out of date with write and read counter changes here. */
136 /* XXX statfs.2 is out of date with read counter changes here. */
138 long f_spare2; /* placeholder */
139 long f_bsize; /* fundamental filesystem block size */
140 long f_iosize; /* optimal transfer block size */
141 long f_blocks; /* total data blocks in filesystem */
142 long f_bfree; /* free blocks in fs */
143 long f_bavail; /* free blocks avail to non-superuser */
144 long f_files; /* total file nodes in filesystem */
145 long f_ffree; /* free file nodes in fs */
146 fsid_t f_fsid; /* filesystem id */
147 uid_t f_owner; /* user that mounted the filesystem */
148 int f_type; /* type of filesystem */
149 int f_flags; /* copy of mount exported flags */
150 long f_syncwrites; /* count of sync writes since mount */
151 long f_asyncwrites; /* count of async writes since mount */
152 char f_fstypename[OMFSNAMELEN]; /* fs type name */
153 char f_mntonname[OMNAMELEN]; /* directory on which mounted */
154 long f_syncreads; /* count of sync reads since mount */
155 long f_asyncreads; /* count of async reads since mount */
156 short f_spares1; /* unused spare */
157 char f_mntfromname[OMNAMELEN];/* mounted filesystem */
158 short f_spares2; /* unused spare */
160 * XXX on machines where longs are aligned to 8-byte boundaries, there
161 * is an unnamed int32_t here. This spare was after the apparent end
162 * of the struct until we bit off the read counters from f_mntonname.
164 long f_spare[2]; /* unused spare */
167 TAILQ_HEAD(vnodelst, vnode);
169 /* Mount options list */
170 TAILQ_HEAD(vfsoptlist, vfsopt);
172 TAILQ_ENTRY(vfsopt) link;
181 * Structure per mounted filesystem. Each mounted filesystem has an
182 * array of operations and an instance record. The filesystems are
183 * put on a doubly linked list.
189 * v - vnode freelist mutex
191 * Unmarked fields are considered stable as long as a ref is held.
195 struct mtx mnt_mtx; /* mount structure interlock */
196 int mnt_gen; /* struct mount generation */
197 #define mnt_startzero mnt_list
198 TAILQ_ENTRY(mount) mnt_list; /* (m) mount list */
199 struct vfsops *mnt_op; /* operations on fs */
200 struct vfsconf *mnt_vfc; /* configuration info */
201 struct vnode *mnt_vnodecovered; /* vnode we mounted on */
202 struct vnode *mnt_syncer; /* syncer vnode */
203 int mnt_ref; /* (i) Reference count */
204 struct vnodelst mnt_nvnodelist; /* (i) list of vnodes */
205 int mnt_nvnodelistsize; /* (i) # of vnodes */
206 int mnt_writeopcount; /* (i) write syscalls pending */
207 int mnt_kern_flag; /* (i) kernel only flags */
208 uint64_t mnt_flag; /* (i) flags shared with user */
209 struct vfsoptlist *mnt_opt; /* current mount options */
210 struct vfsoptlist *mnt_optnew; /* new options passed to fs */
211 int mnt_maxsymlinklen; /* max size of short symlink */
212 struct statfs mnt_stat; /* cache of filesystem stats */
213 struct ucred *mnt_cred; /* credentials of mounter */
214 void * mnt_data; /* private data */
215 time_t mnt_time; /* last time written*/
216 int mnt_iosize_max; /* max size for clusters, etc */
217 struct netexport *mnt_export; /* export list */
218 struct label *mnt_label; /* MAC label for the fs */
219 u_int mnt_hashseed; /* Random seed for vfs_hash */
220 int mnt_lockref; /* (i) Lock reference count */
221 int mnt_secondary_writes; /* (i) # of secondary writes */
222 int mnt_secondary_accwrites;/* (i) secondary wr. starts */
223 struct thread *mnt_susp_owner; /* (i) thread owning suspension */
224 #define mnt_endzero mnt_gjprovider
225 char *mnt_gjprovider; /* gjournal provider name */
226 struct mtx mnt_listmtx;
227 struct vnodelst mnt_lazyvnodelist; /* (l) list of lazy vnodes */
228 int mnt_lazyvnodelistsize; /* (l) # of lazy vnodes */
229 struct lock mnt_explock; /* vfs_export walkers lock */
230 TAILQ_ENTRY(mount) mnt_upper_link; /* (m) we in the all uppers */
231 TAILQ_HEAD(, mount) mnt_uppers; /* (m) upper mounts over us*/
232 int __aligned(CACHE_LINE_SIZE) mnt_vfs_ops;/* (i) pending vfs ops */
233 int *mnt_thread_in_ops_pcpu;
235 int *mnt_lockref_pcpu;
236 int *mnt_writeopcount_pcpu;
237 struct vnode *mnt_rootvnode;
241 * Definitions for MNT_VNODE_FOREACH_ALL.
243 struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp);
244 struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp);
245 void __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp);
247 #define MNT_VNODE_FOREACH_ALL(vp, mp, mvp) \
248 for (vp = __mnt_vnode_first_all(&(mvp), (mp)); \
249 (vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp)))
251 #define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp) \
254 __mnt_vnode_markerfree_all(&(mvp), (mp)); \
255 /* MNT_IUNLOCK(mp); -- done in above function */ \
256 mtx_assert(MNT_MTX(mp), MA_NOTOWNED); \
260 * Definitions for MNT_VNODE_FOREACH_LAZY.
262 typedef int mnt_lazy_cb_t(struct vnode *, void *);
263 struct vnode *__mnt_vnode_next_lazy(struct vnode **mvp, struct mount *mp,
264 mnt_lazy_cb_t *cb, void *cbarg);
265 struct vnode *__mnt_vnode_first_lazy(struct vnode **mvp, struct mount *mp,
266 mnt_lazy_cb_t *cb, void *cbarg);
267 void __mnt_vnode_markerfree_lazy(struct vnode **mvp, struct mount *mp);
269 #define MNT_VNODE_FOREACH_LAZY(vp, mp, mvp, cb, cbarg) \
270 for (vp = __mnt_vnode_first_lazy(&(mvp), (mp), (cb), (cbarg)); \
272 vp = __mnt_vnode_next_lazy(&(mvp), (mp), (cb), (cbarg)))
274 #define MNT_VNODE_FOREACH_LAZY_ABORT(mp, mvp) \
275 __mnt_vnode_markerfree_lazy(&(mvp), (mp))
277 #define MNT_ILOCK(mp) mtx_lock(&(mp)->mnt_mtx)
278 #define MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx)
279 #define MNT_IUNLOCK(mp) mtx_unlock(&(mp)->mnt_mtx)
280 #define MNT_MTX(mp) (&(mp)->mnt_mtx)
282 #define MNT_REF(mp) do { \
283 mtx_assert(MNT_MTX(mp), MA_OWNED); \
286 #define MNT_REL(mp) do { \
287 mtx_assert(MNT_MTX(mp), MA_OWNED); \
289 if ((mp)->mnt_vfs_ops && (mp)->mnt_ref < 0) \
290 vfs_dump_mount_counters(mp); \
291 if ((mp)->mnt_ref == 0 && (mp)->mnt_vfs_ops) \
298 * User specifiable flags, stored in mnt_flag.
300 #define MNT_RDONLY 0x0000000000000001ULL /* read only filesystem */
301 #define MNT_SYNCHRONOUS 0x0000000000000002ULL /* fs written synchronously */
302 #define MNT_NOEXEC 0x0000000000000004ULL /* can't exec from filesystem */
303 #define MNT_NOSUID 0x0000000000000008ULL /* don't honor setuid fs bits */
304 #define MNT_NFS4ACLS 0x0000000000000010ULL /* enable NFS version 4 ACLs */
305 #define MNT_UNION 0x0000000000000020ULL /* union with underlying fs */
306 #define MNT_ASYNC 0x0000000000000040ULL /* fs written asynchronously */
307 #define MNT_SUIDDIR 0x0000000000100000ULL /* special SUID dir handling */
308 #define MNT_SOFTDEP 0x0000000000200000ULL /* using soft updates */
309 #define MNT_NOSYMFOLLOW 0x0000000000400000ULL /* do not follow symlinks */
310 #define MNT_GJOURNAL 0x0000000002000000ULL /* GEOM journal support enabled */
311 #define MNT_MULTILABEL 0x0000000004000000ULL /* MAC support for objects */
312 #define MNT_ACLS 0x0000000008000000ULL /* ACL support enabled */
313 #define MNT_NOATIME 0x0000000010000000ULL /* dont update file access time */
314 #define MNT_NOCLUSTERR 0x0000000040000000ULL /* disable cluster read */
315 #define MNT_NOCLUSTERW 0x0000000080000000ULL /* disable cluster write */
316 #define MNT_SUJ 0x0000000100000000ULL /* using journaled soft updates */
317 #define MNT_AUTOMOUNTED 0x0000000200000000ULL /* mounted by automountd(8) */
318 #define MNT_UNTRUSTED 0x0000000800000000ULL /* filesys metadata untrusted */
321 * NFS export related mount flags.
323 #define MNT_EXRDONLY 0x0000000000000080ULL /* exported read only */
324 #define MNT_EXPORTED 0x0000000000000100ULL /* filesystem is exported */
325 #define MNT_DEFEXPORTED 0x0000000000000200ULL /* exported to the world */
326 #define MNT_EXPORTANON 0x0000000000000400ULL /* anon uid mapping for all */
327 #define MNT_EXKERB 0x0000000000000800ULL /* exported with Kerberos */
328 #define MNT_EXPUBLIC 0x0000000020000000ULL /* public export (WebNFS) */
331 * Flags set by internal operations,
332 * but visible to the user.
333 * XXX some of these are not quite right.. (I've never seen the root flag set)
335 #define MNT_LOCAL 0x0000000000001000ULL /* filesystem is stored locally */
336 #define MNT_QUOTA 0x0000000000002000ULL /* quotas are enabled on fs */
337 #define MNT_ROOTFS 0x0000000000004000ULL /* identifies the root fs */
338 #define MNT_USER 0x0000000000008000ULL /* mounted by a user */
339 #define MNT_IGNORE 0x0000000000800000ULL /* do not show entry in df */
340 #define MNT_VERIFIED 0x0000000400000000ULL /* filesystem is verified */
343 * Mask of flags that are visible to statfs().
344 * XXX I think that this could now become (~(MNT_CMDFLAGS))
345 * but the 'mount' program may need changing to handle this.
347 #define MNT_VISFLAGMASK (MNT_RDONLY | MNT_SYNCHRONOUS | MNT_NOEXEC | \
348 MNT_NOSUID | MNT_UNION | MNT_SUJ | \
349 MNT_ASYNC | MNT_EXRDONLY | MNT_EXPORTED | \
350 MNT_DEFEXPORTED | MNT_EXPORTANON| MNT_EXKERB | \
351 MNT_LOCAL | MNT_USER | MNT_QUOTA | \
352 MNT_ROOTFS | MNT_NOATIME | MNT_NOCLUSTERR| \
353 MNT_NOCLUSTERW | MNT_SUIDDIR | MNT_SOFTDEP | \
354 MNT_IGNORE | MNT_EXPUBLIC | MNT_NOSYMFOLLOW | \
355 MNT_GJOURNAL | MNT_MULTILABEL | MNT_ACLS | \
356 MNT_NFS4ACLS | MNT_AUTOMOUNTED | MNT_VERIFIED | \
359 /* Mask of flags that can be updated. */
360 #define MNT_UPDATEMASK (MNT_NOSUID | MNT_NOEXEC | \
361 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | \
363 MNT_NOSYMFOLLOW | MNT_IGNORE | \
364 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR | \
365 MNT_ACLS | MNT_USER | MNT_NFS4ACLS | \
366 MNT_AUTOMOUNTED | MNT_UNTRUSTED)
369 * External filesystem command modifier flags.
370 * Unmount can use the MNT_FORCE flag.
371 * XXX: These are not STATES and really should be somewhere else.
372 * XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL,
373 * but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2),
374 * and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2),
377 #define MNT_UPDATE 0x0000000000010000ULL /* not real mount, just update */
378 #define MNT_DELEXPORT 0x0000000000020000ULL /* delete export host lists */
379 #define MNT_RELOAD 0x0000000000040000ULL /* reload filesystem data */
380 #define MNT_FORCE 0x0000000000080000ULL /* force unmount or readonly */
381 #define MNT_SNAPSHOT 0x0000000001000000ULL /* snapshot the filesystem */
382 #define MNT_NONBUSY 0x0000000004000000ULL /* check vnode use counts. */
383 #define MNT_BYFSID 0x0000000008000000ULL /* specify filesystem by ID. */
384 #define MNT_NOCOVER 0x0000001000000000ULL /* Do not cover a mount point */
385 #define MNT_EMPTYDIR 0x0000002000000000ULL /* Only mount on empty dir */
386 #define MNT_CMDFLAGS (MNT_UPDATE | MNT_DELEXPORT | MNT_RELOAD | \
387 MNT_FORCE | MNT_SNAPSHOT | MNT_NONBUSY | \
388 MNT_BYFSID | MNT_NOCOVER | MNT_EMPTYDIR)
390 * Internal filesystem control flags stored in mnt_kern_flag.
392 * MNTK_UNMOUNT locks the mount entry so that name lookup cannot
393 * proceed past the mount point. This keeps the subtree stable during
394 * mounts and unmounts. When non-forced unmount flushes all vnodes
395 * from the mp queue, the MNTK_UNMOUNT flag prevents insmntque() from
396 * queueing new vnodes.
398 * MNTK_UNMOUNTF permits filesystems to detect a forced unmount while
399 * dounmount() is still waiting to lock the mountpoint. This allows
400 * the filesystem to cancel operations that might otherwise deadlock
401 * with the unmount attempt (used by NFS).
403 #define MNTK_UNMOUNTF 0x00000001 /* forced unmount in progress */
404 #define MNTK_ASYNC 0x00000002 /* filtered async flag */
405 #define MNTK_SOFTDEP 0x00000004 /* async disabled by softdep */
406 #define MNTK_NOMSYNC 0x00000008 /* don't do msync */
407 #define MNTK_DRAINING 0x00000010 /* lock draining is happening */
408 #define MNTK_REFEXPIRE 0x00000020 /* refcount expiring is happening */
409 #define MNTK_EXTENDED_SHARED 0x00000040 /* Allow shared locking for more ops */
410 #define MNTK_SHARED_WRITES 0x00000080 /* Allow shared locking for writes */
411 #define MNTK_NO_IOPF 0x00000100 /* Disallow page faults during reads
412 and writes. Filesystem shall properly
413 handle i/o state on EFAULT. */
414 #define MNTK_VGONE_UPPER 0x00000200
415 #define MNTK_VGONE_WAITER 0x00000400
416 #define MNTK_LOOKUP_EXCL_DOTDOT 0x00000800
417 #define MNTK_MARKER 0x00001000
418 #define MNTK_UNMAPPED_BUFS 0x00002000
419 #define MNTK_USES_BCACHE 0x00004000 /* FS uses the buffer cache. */
420 #define MNTK_TEXT_REFS 0x00008000 /* Keep use ref for text */
421 #define MNTK_VMSETSIZE_BUG 0x00010000
422 #define MNTK_UNIONFS 0x00020000 /* A hack for F_ISUNIONSTACK */
423 #define MNTK_NOASYNC 0x00800000 /* disable async */
424 #define MNTK_UNMOUNT 0x01000000 /* unmount in progress */
425 #define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */
426 #define MNTK_SUSPEND 0x08000000 /* request write suspension */
427 #define MNTK_SUSPEND2 0x04000000 /* block secondary writes */
428 #define MNTK_SUSPENDED 0x10000000 /* write operations are suspended */
429 #define MNTK_NULL_NOCACHE 0x20000000 /* auto disable cache for nullfs
430 mounts over this fs */
431 #define MNTK_LOOKUP_SHARED 0x40000000 /* FS supports shared lock lookups */
432 #define MNTK_NOKNOTE 0x80000000 /* Don't send KNOTEs from VOP hooks */
436 MNT_SHARED_WRITES(struct mount *mp)
439 return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0);
443 MNT_EXTENDED_SHARED(struct mount *mp)
446 return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0);
451 * Sysctl CTL_VFS definitions.
453 * Second level identifier specifies which filesystem. Second level
454 * identifier VFS_VFSCONF returns information about all filesystems.
455 * Second level identifier VFS_GENERIC is non-terminal.
457 #define VFS_VFSCONF 0 /* get configured filesystems */
458 #define VFS_GENERIC 0 /* generic filesystem information */
460 * Third level identifiers for VFS_GENERIC are given below; third
461 * level identifiers for specific filesystems are given in their
462 * mount specific header files.
464 #define VFS_MAXTYPENUM 1 /* int: highest defined filesystem type */
465 #define VFS_CONF 2 /* struct: vfsconf for filesystem given
469 * Flags for various system call interfaces.
471 * waitfor flags to vfs_sync() and getfsstat()
473 #define MNT_WAIT 1 /* synchronously wait for I/O to complete */
474 #define MNT_NOWAIT 2 /* start all I/O, but do not wait for it */
475 #define MNT_LAZY 3 /* push data not written by filesystem syncer */
476 #define MNT_SUSPEND 4 /* Suspend file system after sync */
479 * Generic file handle
482 fsid_t fh_fsid; /* Filesystem id of mount point */
483 struct fid fh_fid; /* Filesys specific id */
485 typedef struct fhandle fhandle_t;
488 * Old export arguments without security flavor list
490 struct oexport_args {
491 int ex_flags; /* export related flags */
492 uid_t ex_root; /* mapping for root uid */
493 struct xucred ex_anon; /* mapping for anonymous user */
494 struct sockaddr *ex_addr; /* net address to which exported */
495 u_char ex_addrlen; /* and the net address length */
496 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
497 u_char ex_masklen; /* and the smask length */
498 char *ex_indexfile; /* index file for WebNFS URLs */
502 * Not quite so old export arguments with 32bit ex_flags and xucred ex_anon.
504 #define MAXSECFLAVORS 5
505 struct o2export_args {
506 int ex_flags; /* export related flags */
507 uid_t ex_root; /* mapping for root uid */
508 struct xucred ex_anon; /* mapping for anonymous user */
509 struct sockaddr *ex_addr; /* net address to which exported */
510 u_char ex_addrlen; /* and the net address length */
511 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
512 u_char ex_masklen; /* and the smask length */
513 char *ex_indexfile; /* index file for WebNFS URLs */
514 int ex_numsecflavors; /* security flavor count */
515 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */
519 * Export arguments for local filesystem mount calls.
522 uint64_t ex_flags; /* export related flags */
523 uid_t ex_root; /* mapping for root uid */
524 uid_t ex_uid; /* mapping for anonymous user */
527 struct sockaddr *ex_addr; /* net address to which exported */
528 u_char ex_addrlen; /* and the net address length */
529 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
530 u_char ex_masklen; /* and the smask length */
531 char *ex_indexfile; /* index file for WebNFS URLs */
532 int ex_numsecflavors; /* security flavor count */
533 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */
537 * Structure holding information for a publicly exported filesystem
538 * (WebNFS). Currently the specs allow just for one such filesystem.
541 int np_valid; /* Do we hold valid information */
542 fhandle_t np_handle; /* Filehandle for pub fs (internal) */
543 struct mount *np_mount; /* Mountpoint of exported fs */
544 char *np_index; /* Index file */
548 * Filesystem configuration information. One of these exists for each
549 * type of filesystem supported by the kernel. These are searched at
550 * mount time to identify the requested filesystem.
552 * XXX: Never change the first two arguments!
555 u_int vfc_version; /* ABI version number */
556 char vfc_name[MFSNAMELEN]; /* filesystem type name */
557 struct vfsops *vfc_vfsops; /* filesystem operations vector */
558 struct vfsops *vfc_vfsops_sd; /* ... signal-deferred */
559 int vfc_typenum; /* historic filesystem type number */
560 int vfc_refcount; /* number mounted of this type */
561 int vfc_flags; /* permanent flags */
562 int vfc_prison_flag; /* prison allow.mount.* flag */
563 struct vfsoptdecl *vfc_opts; /* mount options */
564 TAILQ_ENTRY(vfsconf) vfc_list; /* list of vfscons */
567 /* Userland version of the struct vfsconf. */
569 struct vfsops *vfc_vfsops; /* filesystem operations vector */
570 char vfc_name[MFSNAMELEN]; /* filesystem type name */
571 int vfc_typenum; /* historic filesystem type number */
572 int vfc_refcount; /* number mounted of this type */
573 int vfc_flags; /* permanent flags */
574 struct vfsconf *vfc_next; /* next in list */
588 * NB: these flags refer to IMPLEMENTATION properties, not properties of
589 * any actual mounts; i.e., it does not make sense to change the flags.
591 #define VFCF_STATIC 0x00010000 /* statically compiled into kernel */
592 #define VFCF_NETWORK 0x00020000 /* may get data over the network */
593 #define VFCF_READONLY 0x00040000 /* writes are not implemented */
594 #define VFCF_SYNTHETIC 0x00080000 /* data does not represent real files */
595 #define VFCF_LOOPBACK 0x00100000 /* aliases some other mounted FS */
596 #define VFCF_UNICODE 0x00200000 /* stores file names as Unicode */
597 #define VFCF_JAIL 0x00400000 /* can be mounted from within a jail */
598 #define VFCF_DELEGADMIN 0x00800000 /* supports delegated administration */
599 #define VFCF_SBDRY 0x01000000 /* Stop at Boundary: defer stop requests
600 to kernel->user (AST) transition */
602 typedef uint32_t fsctlop_t;
605 int vc_vers; /* should be VFSIDCTL_VERS1 (below) */
606 fsid_t vc_fsid; /* fsid to operate on */
607 char vc_fstypename[MFSNAMELEN];
608 /* type of fs 'nfs' or '*' */
609 fsctlop_t vc_op; /* operation VFS_CTL_* (below) */
610 void *vc_ptr; /* pointer to data structure */
611 size_t vc_len; /* sizeof said structure */
612 u_int32_t vc_spare[12]; /* spare (must be zero) */
615 /* vfsidctl API version. */
616 #define VFS_CTL_VERS1 0x01
619 * New style VFS sysctls, do not reuse/conflict with the namespace for
621 * All "global" sysctl ops have the 33rd bit set:
623 * Private sysctl ops should have the 33rd bit unset.
625 #define VFS_CTL_QUERY 0x00010001 /* anything wrong? (vfsquery) */
626 #define VFS_CTL_TIMEO 0x00010002 /* set timeout for vfs notification */
627 #define VFS_CTL_NOLOCKS 0x00010003 /* disable file locking */
631 u_int32_t vq_spare[31];
635 #define VQ_NOTRESP 0x0001 /* server down */
636 #define VQ_NEEDAUTH 0x0002 /* server bad auth */
637 #define VQ_LOWDISK 0x0004 /* we're low on space */
638 #define VQ_MOUNT 0x0008 /* new filesystem arrived */
639 #define VQ_UNMOUNT 0x0010 /* filesystem has left */
640 #define VQ_DEAD 0x0020 /* filesystem is dead, needs force unmount */
641 #define VQ_ASSIST 0x0040 /* filesystem needs assistance from external
643 #define VQ_NOTRESPLOCK 0x0080 /* server lockd down */
644 #define VQ_FLAG0100 0x0100 /* placeholder */
645 #define VQ_FLAG0200 0x0200 /* placeholder */
646 #define VQ_FLAG0400 0x0400 /* placeholder */
647 #define VQ_FLAG0800 0x0800 /* placeholder */
648 #define VQ_FLAG1000 0x1000 /* placeholder */
649 #define VQ_FLAG2000 0x2000 /* placeholder */
650 #define VQ_FLAG4000 0x4000 /* placeholder */
651 #define VQ_FLAG8000 0x8000 /* placeholder */
654 /* Point a sysctl request at a vfsidctl's data. */
655 #define VCTLTOREQ(vc, req) \
657 (req)->newptr = (vc)->vc_ptr; \
658 (req)->newlen = (vc)->vc_len; \
669 * vfs_busy specific flags and mask.
671 #define MBF_NOWAIT 0x01
672 #define MBF_MNTLSTLOCK 0x02
673 #define MBF_MASK (MBF_NOWAIT | MBF_MNTLSTLOCK)
675 #ifdef MALLOC_DECLARE
676 MALLOC_DECLARE(M_MOUNT);
677 MALLOC_DECLARE(M_STATFS);
679 extern int maxvfsconf; /* highest defined filesystem type */
681 TAILQ_HEAD(vfsconfhead, vfsconf);
682 extern struct vfsconfhead vfsconf;
685 * Operations supported on mounted filesystem.
693 * N.B., vfs_cmount is the ancient vfsop invoked by the old mount(2) syscall.
694 * The new way is vfs_mount.
696 * vfs_cmount implementations typically translate arguments from their
697 * respective old per-FS structures into the key-value list supported by
698 * nmount(2), then use kernel_mount(9) to mimic nmount(2) from kernelspace.
700 * Filesystems with mounters that use nmount(2) do not need to and should not
701 * implement vfs_cmount. Hopefully a future cleanup can remove vfs_cmount and
704 typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags);
705 typedef int vfs_unmount_t(struct mount *mp, int mntflags);
706 typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp);
707 typedef int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg);
708 typedef int vfs_statfs_t(struct mount *mp, struct statfs *sbp);
709 typedef int vfs_sync_t(struct mount *mp, int waitfor);
710 typedef int vfs_vget_t(struct mount *mp, ino_t ino, int flags,
712 typedef int vfs_fhtovp_t(struct mount *mp, struct fid *fhp,
713 int flags, struct vnode **vpp);
714 typedef int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam,
715 uint64_t *extflagsp, struct ucred **credanonp,
716 int *numsecflavors, int *secflavors);
717 typedef int vfs_init_t(struct vfsconf *);
718 typedef int vfs_uninit_t(struct vfsconf *);
719 typedef int vfs_extattrctl_t(struct mount *mp, int cmd,
720 struct vnode *filename_vp, int attrnamespace,
721 const char *attrname);
722 typedef int vfs_mount_t(struct mount *mp);
723 typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op,
724 struct sysctl_req *req);
725 typedef void vfs_susp_clean_t(struct mount *mp);
726 typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp);
727 typedef void vfs_purge_t(struct mount *mp);
730 vfs_mount_t *vfs_mount;
731 vfs_cmount_t *vfs_cmount;
732 vfs_unmount_t *vfs_unmount;
733 vfs_root_t *vfs_root;
734 vfs_root_t *vfs_cachedroot;
735 vfs_quotactl_t *vfs_quotactl;
736 vfs_statfs_t *vfs_statfs;
737 vfs_sync_t *vfs_sync;
738 vfs_vget_t *vfs_vget;
739 vfs_fhtovp_t *vfs_fhtovp;
740 vfs_checkexp_t *vfs_checkexp;
741 vfs_init_t *vfs_init;
742 vfs_uninit_t *vfs_uninit;
743 vfs_extattrctl_t *vfs_extattrctl;
744 vfs_sysctl_t *vfs_sysctl;
745 vfs_susp_clean_t *vfs_susp_clean;
746 vfs_notify_lowervp_t *vfs_reclaim_lowervp;
747 vfs_notify_lowervp_t *vfs_unlink_lowervp;
748 vfs_purge_t *vfs_purge;
749 vfs_mount_t *vfs_spare[6]; /* spares for ABI compat */
752 vfs_statfs_t __vfs_statfs;
754 #define VFS_MOUNT(MP) ({ \
757 TSRAW(curthread, TS_ENTER, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
758 _rc = (*(MP)->mnt_op->vfs_mount)(MP); \
759 TSRAW(curthread, TS_EXIT, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
762 #define VFS_UNMOUNT(MP, FORCE) ({ \
765 _rc = (*(MP)->mnt_op->vfs_unmount)(MP, FORCE); \
768 #define VFS_ROOT(MP, FLAGS, VPP) ({ \
771 _rc = (*(MP)->mnt_op->vfs_root)(MP, FLAGS, VPP); \
774 #define VFS_CACHEDROOT(MP, FLAGS, VPP) ({ \
777 _rc = (*(MP)->mnt_op->vfs_cachedroot)(MP, FLAGS, VPP); \
780 #define VFS_QUOTACTL(MP, C, U, A) ({ \
783 _rc = (*(MP)->mnt_op->vfs_quotactl)(MP, C, U, A); \
786 #define VFS_STATFS(MP, SBP) ({ \
789 _rc = __vfs_statfs((MP), (SBP)); \
792 #define VFS_SYNC(MP, WAIT) ({ \
795 _rc = (*(MP)->mnt_op->vfs_sync)(MP, WAIT); \
798 #define VFS_VGET(MP, INO, FLAGS, VPP) ({ \
801 _rc = (*(MP)->mnt_op->vfs_vget)(MP, INO, FLAGS, VPP); \
804 #define VFS_FHTOVP(MP, FIDP, FLAGS, VPP) ({ \
807 _rc = (*(MP)->mnt_op->vfs_fhtovp)(MP, FIDP, FLAGS, VPP); \
810 #define VFS_CHECKEXP(MP, NAM, EXFLG, CRED, NUMSEC, SEC) ({ \
813 _rc = (*(MP)->mnt_op->vfs_checkexp)(MP, NAM, EXFLG, CRED, NUMSEC,\
817 #define VFS_EXTATTRCTL(MP, C, FN, NS, N) ({ \
820 _rc = (*(MP)->mnt_op->vfs_extattrctl)(MP, C, FN, NS, N); \
823 #define VFS_SYSCTL(MP, OP, REQ) ({ \
826 _rc = (*(MP)->mnt_op->vfs_sysctl)(MP, OP, REQ); \
829 #define VFS_SUSP_CLEAN(MP) do { \
830 if (*(MP)->mnt_op->vfs_susp_clean != NULL) { \
831 (*(MP)->mnt_op->vfs_susp_clean)(MP); \
835 #define VFS_RECLAIM_LOWERVP(MP, VP) do { \
836 if (*(MP)->mnt_op->vfs_reclaim_lowervp != NULL) { \
837 (*(MP)->mnt_op->vfs_reclaim_lowervp)((MP), (VP)); \
841 #define VFS_UNLINK_LOWERVP(MP, VP) do { \
842 if (*(MP)->mnt_op->vfs_unlink_lowervp != NULL) { \
843 (*(MP)->mnt_op->vfs_unlink_lowervp)((MP), (VP)); \
847 #define VFS_PURGE(MP) do { \
848 if (*(MP)->mnt_op->vfs_purge != NULL) { \
849 (*(MP)->mnt_op->vfs_purge)(MP); \
853 #define VFS_KNOTE_LOCKED(vp, hint) do \
855 if (((vp)->v_vflag & VV_NOKNOTE) == 0) \
856 VN_KNOTE((vp), (hint), KNF_LISTLOCKED); \
859 #define VFS_KNOTE_UNLOCKED(vp, hint) do \
861 if (((vp)->v_vflag & VV_NOKNOTE) == 0) \
862 VN_KNOTE((vp), (hint), 0); \
865 #define VFS_NOTIFY_UPPER_RECLAIM 1
866 #define VFS_NOTIFY_UPPER_UNLINK 2
868 #include <sys/module.h>
873 #define VFS_VERSION_00 0x19660120
874 #define VFS_VERSION_01 0x20121030
875 #define VFS_VERSION_02 0x20180504
876 #define VFS_VERSION VFS_VERSION_02
878 #define VFS_SET(vfsops, fsname, flags) \
879 static struct vfsconf fsname ## _vfsconf = { \
880 .vfc_version = VFS_VERSION, \
881 .vfc_name = #fsname, \
882 .vfc_vfsops = &vfsops, \
884 .vfc_flags = flags, \
886 static moduledata_t fsname ## _mod = { \
889 & fsname ## _vfsconf \
891 DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE)
894 * exported vnode operations
897 int dounmount(struct mount *, int, struct thread *);
899 int kernel_mount(struct mntarg *ma, uint64_t flags);
900 int kernel_vmount(int flags, ...);
901 struct mntarg *mount_arg(struct mntarg *ma, const char *name, const void *val, int len);
902 struct mntarg *mount_argb(struct mntarg *ma, int flag, const char *name);
903 struct mntarg *mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...);
904 struct mntarg *mount_argsu(struct mntarg *ma, const char *name, const void *val, int len);
905 void statfs_scale_blocks(struct statfs *sf, long max_size);
906 struct vfsconf *vfs_byname(const char *);
907 struct vfsconf *vfs_byname_kld(const char *, struct thread *td, int *);
908 void vfs_mount_destroy(struct mount *);
909 void vfs_event_signal(fsid_t *, u_int32_t, intptr_t);
910 void vfs_freeopts(struct vfsoptlist *opts);
911 void vfs_deleteopt(struct vfsoptlist *opts, const char *name);
912 int vfs_buildopts(struct uio *auio, struct vfsoptlist **options);
913 int vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w,
915 int vfs_getopt(struct vfsoptlist *, const char *, void **, int *);
916 int vfs_getopt_pos(struct vfsoptlist *opts, const char *name);
917 int vfs_getopt_size(struct vfsoptlist *opts, const char *name,
919 char *vfs_getopts(struct vfsoptlist *, const char *, int *error);
920 int vfs_copyopt(struct vfsoptlist *, const char *, void *, int);
921 int vfs_filteropt(struct vfsoptlist *, const char **legal);
922 void vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...);
923 int vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...);
924 int vfs_setopt(struct vfsoptlist *opts, const char *name, void *value,
926 int vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value,
928 int vfs_setopts(struct vfsoptlist *opts, const char *name,
930 int vfs_setpublicfs /* set publicly exported fs */
931 (struct mount *, struct netexport *, struct export_args *);
932 void vfs_periodic(struct mount *, int);
933 int vfs_busy(struct mount *, int);
934 int vfs_export /* process mount export info */
935 (struct mount *, struct export_args *);
936 void vfs_allocate_syncvnode(struct mount *);
937 void vfs_deallocate_syncvnode(struct mount *);
938 int vfs_donmount(struct thread *td, uint64_t fsflags,
939 struct uio *fsoptions);
940 void vfs_getnewfsid(struct mount *);
941 struct cdev *vfs_getrootfsid(struct mount *);
942 struct mount *vfs_getvfs(fsid_t *); /* return vfs given fsid */
943 struct mount *vfs_busyfs(fsid_t *);
944 int vfs_modevent(module_t, int, void *);
945 void vfs_mount_error(struct mount *, const char *, ...);
946 void vfs_mountroot(void); /* mount our root filesystem */
947 void vfs_mountedfrom(struct mount *, const char *from);
948 void vfs_notify_upper(struct vnode *, int);
949 void vfs_ref(struct mount *);
950 void vfs_rel(struct mount *);
951 struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *,
953 int vfs_suser(struct mount *, struct thread *);
954 void vfs_unbusy(struct mount *);
955 void vfs_unmountall(void);
956 extern TAILQ_HEAD(mntlist, mount) mountlist; /* mounted filesystem list */
957 extern struct mtx mountlist_mtx;
958 extern struct nfs_public nfs_pub;
959 extern struct sx vfsconf_sx;
960 #define vfsconf_lock() sx_xlock(&vfsconf_sx)
961 #define vfsconf_unlock() sx_xunlock(&vfsconf_sx)
962 #define vfsconf_slock() sx_slock(&vfsconf_sx)
963 #define vfsconf_sunlock() sx_sunlock(&vfsconf_sx)
964 struct vnode *mntfs_allocvp(struct mount *, struct vnode *);
965 void mntfs_freevp(struct vnode *);
968 * Declarations for these vfs default operations are located in
969 * kern/vfs_default.c. They will be automatically used to replace
970 * null entries in VFS ops tables when registering a new filesystem
971 * type in the global table.
973 vfs_root_t vfs_stdroot;
974 vfs_quotactl_t vfs_stdquotactl;
975 vfs_statfs_t vfs_stdstatfs;
976 vfs_sync_t vfs_stdsync;
977 vfs_sync_t vfs_stdnosync;
978 vfs_vget_t vfs_stdvget;
979 vfs_fhtovp_t vfs_stdfhtovp;
980 vfs_checkexp_t vfs_stdcheckexp;
981 vfs_init_t vfs_stdinit;
982 vfs_uninit_t vfs_stduninit;
983 vfs_extattrctl_t vfs_stdextattrctl;
984 vfs_sysctl_t vfs_stdsysctl;
986 void syncer_suspend(void);
987 void syncer_resume(void);
989 struct vnode *vfs_cache_root_clear(struct mount *);
990 void vfs_cache_root_set(struct mount *, struct vnode *);
992 void vfs_op_barrier_wait(struct mount *);
993 void vfs_op_enter(struct mount *);
994 void vfs_op_exit_locked(struct mount *);
995 void vfs_op_exit(struct mount *);
998 void vfs_assert_mount_counters(struct mount *);
999 void vfs_dump_mount_counters(struct mount *);
1001 #define vfs_assert_mount_counters(mp) do { } while (0)
1002 #define vfs_dump_mount_counters(mp) do { } while (0)
1005 enum mount_counter { MNT_COUNT_REF, MNT_COUNT_LOCKREF, MNT_COUNT_WRITEOPCOUNT };
1006 int vfs_mount_fetch_counter(struct mount *, enum mount_counter);
1009 * Code transitioning mnt_vfs_ops to > 0 issues IPIs until it observes
1010 * all CPUs not executing code enclosed by mnt_thread_in_ops_pcpu.
1012 * This provides an invariant that by the time the last CPU is observed not
1013 * executing, everyone else entering will see the counter > 0 and exit.
1015 * Note there is no barrier between vfs_ops and the rest of the code in the
1016 * section. It is not necessary as the writer has to wait for everyone to drain
1017 * before making any changes or only make changes safe while the section is
1020 #define vfs_op_thread_entered(mp) ({ \
1021 MPASS(curthread->td_critnest > 0); \
1022 *zpcpu_get(mp->mnt_thread_in_ops_pcpu) == 1; \
1025 #define vfs_op_thread_enter(mp) ({ \
1026 bool _retval = true; \
1028 MPASS(!vfs_op_thread_entered(mp)); \
1029 zpcpu_set_protected(mp->mnt_thread_in_ops_pcpu, 1); \
1030 __compiler_membar(); \
1031 if (__predict_false(mp->mnt_vfs_ops > 0)) { \
1032 vfs_op_thread_exit(mp); \
1038 #define vfs_op_thread_exit(mp) do { \
1039 MPASS(vfs_op_thread_entered(mp)); \
1040 __compiler_membar(); \
1041 zpcpu_set_protected(mp->mnt_thread_in_ops_pcpu, 0); \
1045 #define vfs_mp_count_add_pcpu(mp, count, val) do { \
1046 MPASS(vfs_op_thread_entered(mp)); \
1047 zpcpu_add_protected(mp->mnt_##count##_pcpu, val); \
1050 #define vfs_mp_count_sub_pcpu(mp, count, val) do { \
1051 MPASS(vfs_op_thread_entered(mp)); \
1052 zpcpu_sub_protected(mp->mnt_##count##_pcpu, val); \
1055 #else /* !_KERNEL */
1057 #include <sys/cdefs.h>
1062 int fhlink(struct fhandle *, const char *);
1063 int fhlinkat(struct fhandle *, int, const char *);
1064 int fhopen(const struct fhandle *, int);
1065 int fhreadlink(struct fhandle *, char *, size_t);
1066 int fhstat(const struct fhandle *, struct stat *);
1067 int fhstatfs(const struct fhandle *, struct statfs *);
1068 int fstatfs(int, struct statfs *);
1069 int getfh(const char *, fhandle_t *);
1070 int getfhat(int, char *, struct fhandle *, int);
1071 int getfsstat(struct statfs *, long, int);
1072 int getmntinfo(struct statfs **, int);
1073 int lgetfh(const char *, fhandle_t *);
1074 int mount(const char *, const char *, int, void *);
1075 int nmount(struct iovec *, unsigned int, int);
1076 int statfs(const char *, struct statfs *);
1077 int unmount(const char *, int);
1079 /* C library stuff */
1080 int getvfsbyname(const char *, struct xvfsconf *);
1083 #endif /* _KERNEL */
1085 #endif /* !_SYS_MOUNT_H_ */