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 */
57 * These are unique per filesystem on a single machine.
62 u_short fid_len; /* length of data in bytes */
63 u_short fid_data0; /* force longword alignment */
64 char fid_data[MAXFIDSZ]; /* data (variable length) */
68 * filesystem statistics
70 #define MFSNAMELEN 16 /* length of type name including null */
71 #define MNAMELEN 1024 /* size of on/from name bufs */
72 #define STATFS_VERSION 0x20140518 /* current version number */
74 uint32_t f_version; /* structure version number */
75 uint32_t f_type; /* type of filesystem */
76 uint64_t f_flags; /* copy of mount exported flags */
77 uint64_t f_bsize; /* filesystem fragment size */
78 uint64_t f_iosize; /* optimal transfer block size */
79 uint64_t f_blocks; /* total data blocks in filesystem */
80 uint64_t f_bfree; /* free blocks in filesystem */
81 int64_t f_bavail; /* free blocks avail to non-superuser */
82 uint64_t f_files; /* total file nodes in filesystem */
83 int64_t f_ffree; /* free nodes avail to non-superuser */
84 uint64_t f_syncwrites; /* count of sync writes since mount */
85 uint64_t f_asyncwrites; /* count of async writes since mount */
86 uint64_t f_syncreads; /* count of sync reads since mount */
87 uint64_t f_asyncreads; /* count of async reads since mount */
88 uint64_t f_spare[10]; /* unused spare */
89 uint32_t f_namemax; /* maximum filename length */
90 uid_t f_owner; /* user that mounted the filesystem */
91 fsid_t f_fsid; /* filesystem id */
92 char f_charspare[80]; /* spare string space */
93 char f_fstypename[MFSNAMELEN]; /* filesystem type name */
94 char f_mntfromname[MNAMELEN]; /* mounted filesystem */
95 char f_mntonname[MNAMELEN]; /* directory on which mounted */
98 #if defined(_WANT_FREEBSD11_STATFS) || defined(_KERNEL)
99 #define FREEBSD11_STATFS_VERSION 0x20030518 /* current version number */
100 struct freebsd11_statfs {
101 uint32_t f_version; /* structure version number */
102 uint32_t f_type; /* type of filesystem */
103 uint64_t f_flags; /* copy of mount exported flags */
104 uint64_t f_bsize; /* filesystem fragment size */
105 uint64_t f_iosize; /* optimal transfer block size */
106 uint64_t f_blocks; /* total data blocks in filesystem */
107 uint64_t f_bfree; /* free blocks in filesystem */
108 int64_t f_bavail; /* free blocks avail to non-superuser */
109 uint64_t f_files; /* total file nodes in filesystem */
110 int64_t f_ffree; /* free nodes avail to non-superuser */
111 uint64_t f_syncwrites; /* count of sync writes since mount */
112 uint64_t f_asyncwrites; /* count of async writes since mount */
113 uint64_t f_syncreads; /* count of sync reads since mount */
114 uint64_t f_asyncreads; /* count of async reads since mount */
115 uint64_t f_spare[10]; /* unused spare */
116 uint32_t f_namemax; /* maximum filename length */
117 uid_t f_owner; /* user that mounted the filesystem */
118 fsid_t f_fsid; /* filesystem id */
119 char f_charspare[80]; /* spare string space */
120 char f_fstypename[16]; /* filesystem type name */
121 char f_mntfromname[88]; /* mounted filesystem */
122 char f_mntonname[88]; /* directory on which mounted */
124 #endif /* _WANT_FREEBSD11_STATFS || _KERNEL */
127 #define OMFSNAMELEN 16 /* length of fs type name, including null */
128 #define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */
130 /* XXX getfsstat.2 is out of date with write and read counter changes here. */
131 /* XXX statfs.2 is out of date with read counter changes here. */
133 long f_spare2; /* placeholder */
134 long f_bsize; /* fundamental filesystem block size */
135 long f_iosize; /* optimal transfer block size */
136 long f_blocks; /* total data blocks in filesystem */
137 long f_bfree; /* free blocks in fs */
138 long f_bavail; /* free blocks avail to non-superuser */
139 long f_files; /* total file nodes in filesystem */
140 long f_ffree; /* free file nodes in fs */
141 fsid_t f_fsid; /* filesystem id */
142 uid_t f_owner; /* user that mounted the filesystem */
143 int f_type; /* type of filesystem */
144 int f_flags; /* copy of mount exported flags */
145 long f_syncwrites; /* count of sync writes since mount */
146 long f_asyncwrites; /* count of async writes since mount */
147 char f_fstypename[OMFSNAMELEN]; /* fs type name */
148 char f_mntonname[OMNAMELEN]; /* directory on which mounted */
149 long f_syncreads; /* count of sync reads since mount */
150 long f_asyncreads; /* count of async reads since mount */
151 short f_spares1; /* unused spare */
152 char f_mntfromname[OMNAMELEN];/* mounted filesystem */
153 short f_spares2; /* unused spare */
155 * XXX on machines where longs are aligned to 8-byte boundaries, there
156 * is an unnamed int32_t here. This spare was after the apparent end
157 * of the struct until we bit off the read counters from f_mntonname.
159 long f_spare[2]; /* unused spare */
162 TAILQ_HEAD(vnodelst, vnode);
164 /* Mount options list */
165 TAILQ_HEAD(vfsoptlist, vfsopt);
167 TAILQ_ENTRY(vfsopt) link;
176 * Structure per mounted filesystem. Each mounted filesystem has an
177 * array of operations and an instance record. The filesystems are
178 * put on a doubly linked list.
184 * v - vnode freelist mutex
186 * Unmarked fields are considered stable as long as a ref is held.
190 struct mtx mnt_mtx; /* mount structure interlock */
191 int mnt_gen; /* struct mount generation */
192 #define mnt_startzero mnt_list
193 TAILQ_ENTRY(mount) mnt_list; /* (m) mount list */
194 struct vfsops *mnt_op; /* operations on fs */
195 struct vfsconf *mnt_vfc; /* configuration info */
196 struct vnode *mnt_vnodecovered; /* vnode we mounted on */
197 struct vnode *mnt_syncer; /* syncer vnode */
198 int mnt_ref; /* (i) Reference count */
199 struct vnodelst mnt_nvnodelist; /* (i) list of vnodes */
200 int mnt_nvnodelistsize; /* (i) # of vnodes */
201 int mnt_writeopcount; /* (i) write syscalls pending */
202 int mnt_kern_flag; /* (i) kernel only flags */
203 uint64_t mnt_flag; /* (i) flags shared with user */
204 struct vfsoptlist *mnt_opt; /* current mount options */
205 struct vfsoptlist *mnt_optnew; /* new options passed to fs */
206 int mnt_maxsymlinklen; /* max size of short symlink */
207 struct statfs mnt_stat; /* cache of filesystem stats */
208 struct ucred *mnt_cred; /* credentials of mounter */
209 void * mnt_data; /* private data */
210 time_t mnt_time; /* last time written*/
211 int mnt_iosize_max; /* max size for clusters, etc */
212 struct netexport *mnt_export; /* export list */
213 struct label *mnt_label; /* MAC label for the fs */
214 u_int mnt_hashseed; /* Random seed for vfs_hash */
215 int mnt_lockref; /* (i) Lock reference count */
216 int mnt_secondary_writes; /* (i) # of secondary writes */
217 int mnt_secondary_accwrites;/* (i) secondary wr. starts */
218 struct thread *mnt_susp_owner; /* (i) thread owning suspension */
219 #define mnt_endzero mnt_gjprovider
220 char *mnt_gjprovider; /* gjournal provider name */
221 struct mtx mnt_listmtx;
222 struct vnodelst mnt_lazyvnodelist; /* (l) list of lazy vnodes */
223 int mnt_lazyvnodelistsize; /* (l) # of lazy vnodes */
224 struct lock mnt_explock; /* vfs_export walkers lock */
225 TAILQ_ENTRY(mount) mnt_upper_link; /* (m) we in the all uppers */
226 TAILQ_HEAD(, mount) mnt_uppers; /* (m) upper mounts over us*/
227 int __aligned(CACHE_LINE_SIZE) mnt_vfs_ops;/* (i) pending vfs ops */
228 int *mnt_thread_in_ops_pcpu;
230 int *mnt_lockref_pcpu;
231 int *mnt_writeopcount_pcpu;
232 struct vnode *mnt_rootvnode;
236 * Definitions for MNT_VNODE_FOREACH_ALL.
238 struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp);
239 struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp);
240 void __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp);
242 #define MNT_VNODE_FOREACH_ALL(vp, mp, mvp) \
243 for (vp = __mnt_vnode_first_all(&(mvp), (mp)); \
244 (vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp)))
246 #define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp) \
249 __mnt_vnode_markerfree_all(&(mvp), (mp)); \
250 /* MNT_IUNLOCK(mp); -- done in above function */ \
251 mtx_assert(MNT_MTX(mp), MA_NOTOWNED); \
255 * Definitions for MNT_VNODE_FOREACH_LAZY.
257 typedef int mnt_lazy_cb_t(struct vnode *, void *);
258 struct vnode *__mnt_vnode_next_lazy(struct vnode **mvp, struct mount *mp,
259 mnt_lazy_cb_t *cb, void *cbarg);
260 struct vnode *__mnt_vnode_first_lazy(struct vnode **mvp, struct mount *mp,
261 mnt_lazy_cb_t *cb, void *cbarg);
262 void __mnt_vnode_markerfree_lazy(struct vnode **mvp, struct mount *mp);
264 #define MNT_VNODE_FOREACH_LAZY(vp, mp, mvp, cb, cbarg) \
265 for (vp = __mnt_vnode_first_lazy(&(mvp), (mp), (cb), (cbarg)); \
267 vp = __mnt_vnode_next_lazy(&(mvp), (mp), (cb), (cbarg)))
269 #define MNT_VNODE_FOREACH_LAZY_ABORT(mp, mvp) \
270 __mnt_vnode_markerfree_lazy(&(mvp), (mp))
272 #define MNT_ILOCK(mp) mtx_lock(&(mp)->mnt_mtx)
273 #define MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx)
274 #define MNT_IUNLOCK(mp) mtx_unlock(&(mp)->mnt_mtx)
275 #define MNT_MTX(mp) (&(mp)->mnt_mtx)
277 #define MNT_REF(mp) do { \
278 mtx_assert(MNT_MTX(mp), MA_OWNED); \
281 #define MNT_REL(mp) do { \
282 mtx_assert(MNT_MTX(mp), MA_OWNED); \
284 if ((mp)->mnt_vfs_ops && (mp)->mnt_ref < 0) \
285 vfs_dump_mount_counters(mp); \
286 if ((mp)->mnt_ref == 0 && (mp)->mnt_vfs_ops) \
293 * User specifiable flags, stored in mnt_flag.
295 #define MNT_RDONLY 0x0000000000000001ULL /* read only filesystem */
296 #define MNT_SYNCHRONOUS 0x0000000000000002ULL /* fs written synchronously */
297 #define MNT_NOEXEC 0x0000000000000004ULL /* can't exec from filesystem */
298 #define MNT_NOSUID 0x0000000000000008ULL /* don't honor setuid fs bits */
299 #define MNT_NFS4ACLS 0x0000000000000010ULL /* enable NFS version 4 ACLs */
300 #define MNT_UNION 0x0000000000000020ULL /* union with underlying fs */
301 #define MNT_ASYNC 0x0000000000000040ULL /* fs written asynchronously */
302 #define MNT_SUIDDIR 0x0000000000100000ULL /* special SUID dir handling */
303 #define MNT_SOFTDEP 0x0000000000200000ULL /* using soft updates */
304 #define MNT_NOSYMFOLLOW 0x0000000000400000ULL /* do not follow symlinks */
305 #define MNT_GJOURNAL 0x0000000002000000ULL /* GEOM journal support enabled */
306 #define MNT_MULTILABEL 0x0000000004000000ULL /* MAC support for objects */
307 #define MNT_ACLS 0x0000000008000000ULL /* ACL support enabled */
308 #define MNT_NOATIME 0x0000000010000000ULL /* dont update file access time */
309 #define MNT_NOCLUSTERR 0x0000000040000000ULL /* disable cluster read */
310 #define MNT_NOCLUSTERW 0x0000000080000000ULL /* disable cluster write */
311 #define MNT_SUJ 0x0000000100000000ULL /* using journaled soft updates */
312 #define MNT_AUTOMOUNTED 0x0000000200000000ULL /* mounted by automountd(8) */
313 #define MNT_UNTRUSTED 0x0000000800000000ULL /* filesys metadata untrusted */
316 * NFS export related mount flags.
318 #define MNT_EXRDONLY 0x0000000000000080ULL /* exported read only */
319 #define MNT_EXPORTED 0x0000000000000100ULL /* filesystem is exported */
320 #define MNT_DEFEXPORTED 0x0000000000000200ULL /* exported to the world */
321 #define MNT_EXPORTANON 0x0000000000000400ULL /* anon uid mapping for all */
322 #define MNT_EXKERB 0x0000000000000800ULL /* exported with Kerberos */
323 #define MNT_EXPUBLIC 0x0000000020000000ULL /* public export (WebNFS) */
326 * Flags set by internal operations,
327 * but visible to the user.
328 * XXX some of these are not quite right.. (I've never seen the root flag set)
330 #define MNT_LOCAL 0x0000000000001000ULL /* filesystem is stored locally */
331 #define MNT_QUOTA 0x0000000000002000ULL /* quotas are enabled on fs */
332 #define MNT_ROOTFS 0x0000000000004000ULL /* identifies the root fs */
333 #define MNT_USER 0x0000000000008000ULL /* mounted by a user */
334 #define MNT_IGNORE 0x0000000000800000ULL /* do not show entry in df */
335 #define MNT_VERIFIED 0x0000000400000000ULL /* filesystem is verified */
338 * Mask of flags that are visible to statfs().
339 * XXX I think that this could now become (~(MNT_CMDFLAGS))
340 * but the 'mount' program may need changing to handle this.
342 #define MNT_VISFLAGMASK (MNT_RDONLY | MNT_SYNCHRONOUS | MNT_NOEXEC | \
343 MNT_NOSUID | MNT_UNION | MNT_SUJ | \
344 MNT_ASYNC | MNT_EXRDONLY | MNT_EXPORTED | \
345 MNT_DEFEXPORTED | MNT_EXPORTANON| MNT_EXKERB | \
346 MNT_LOCAL | MNT_USER | MNT_QUOTA | \
347 MNT_ROOTFS | MNT_NOATIME | MNT_NOCLUSTERR| \
348 MNT_NOCLUSTERW | MNT_SUIDDIR | MNT_SOFTDEP | \
349 MNT_IGNORE | MNT_EXPUBLIC | MNT_NOSYMFOLLOW | \
350 MNT_GJOURNAL | MNT_MULTILABEL | MNT_ACLS | \
351 MNT_NFS4ACLS | MNT_AUTOMOUNTED | MNT_VERIFIED | \
354 /* Mask of flags that can be updated. */
355 #define MNT_UPDATEMASK (MNT_NOSUID | MNT_NOEXEC | \
356 MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | \
358 MNT_NOSYMFOLLOW | MNT_IGNORE | \
359 MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR | \
360 MNT_ACLS | MNT_USER | MNT_NFS4ACLS | \
361 MNT_AUTOMOUNTED | MNT_UNTRUSTED)
364 * External filesystem command modifier flags.
365 * Unmount can use the MNT_FORCE flag.
366 * XXX: These are not STATES and really should be somewhere else.
367 * XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL,
368 * but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2),
369 * and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2),
372 #define MNT_UPDATE 0x0000000000010000ULL /* not real mount, just update */
373 #define MNT_DELEXPORT 0x0000000000020000ULL /* delete export host lists */
374 #define MNT_RELOAD 0x0000000000040000ULL /* reload filesystem data */
375 #define MNT_FORCE 0x0000000000080000ULL /* force unmount or readonly */
376 #define MNT_SNAPSHOT 0x0000000001000000ULL /* snapshot the filesystem */
377 #define MNT_NONBUSY 0x0000000004000000ULL /* check vnode use counts. */
378 #define MNT_BYFSID 0x0000000008000000ULL /* specify filesystem by ID. */
379 #define MNT_NOCOVER 0x0000001000000000ULL /* Do not cover a mount point */
380 #define MNT_EMPTYDIR 0x0000002000000000ULL /* Only mount on empty dir */
381 #define MNT_CMDFLAGS (MNT_UPDATE | MNT_DELEXPORT | MNT_RELOAD | \
382 MNT_FORCE | MNT_SNAPSHOT | MNT_NONBUSY | \
383 MNT_BYFSID | MNT_NOCOVER | MNT_EMPTYDIR)
385 * Internal filesystem control flags stored in mnt_kern_flag.
387 * MNTK_UNMOUNT locks the mount entry so that name lookup cannot
388 * proceed past the mount point. This keeps the subtree stable during
389 * mounts and unmounts. When non-forced unmount flushes all vnodes
390 * from the mp queue, the MNTK_UNMOUNT flag prevents insmntque() from
391 * queueing new vnodes.
393 * MNTK_UNMOUNTF permits filesystems to detect a forced unmount while
394 * dounmount() is still waiting to lock the mountpoint. This allows
395 * the filesystem to cancel operations that might otherwise deadlock
396 * with the unmount attempt (used by NFS).
398 #define MNTK_UNMOUNTF 0x00000001 /* forced unmount in progress */
399 #define MNTK_ASYNC 0x00000002 /* filtered async flag */
400 #define MNTK_SOFTDEP 0x00000004 /* async disabled by softdep */
401 #define MNTK_NOMSYNC 0x00000008 /* don't do msync */
402 #define MNTK_DRAINING 0x00000010 /* lock draining is happening */
403 #define MNTK_REFEXPIRE 0x00000020 /* refcount expiring is happening */
404 #define MNTK_EXTENDED_SHARED 0x00000040 /* Allow shared locking for more ops */
405 #define MNTK_SHARED_WRITES 0x00000080 /* Allow shared locking for writes */
406 #define MNTK_NO_IOPF 0x00000100 /* Disallow page faults during reads
407 and writes. Filesystem shall properly
408 handle i/o state on EFAULT. */
409 #define MNTK_VGONE_UPPER 0x00000200
410 #define MNTK_VGONE_WAITER 0x00000400
411 #define MNTK_LOOKUP_EXCL_DOTDOT 0x00000800
412 #define MNTK_MARKER 0x00001000
413 #define MNTK_UNMAPPED_BUFS 0x00002000
414 #define MNTK_USES_BCACHE 0x00004000 /* FS uses the buffer cache. */
415 #define MNTK_TEXT_REFS 0x00008000 /* Keep use ref for text */
416 #define MNTK_VMSETSIZE_BUG 0x00010000
417 #define MNTK_UNIONFS 0x00020000 /* A hack for F_ISUNIONSTACK */
418 #define MNTK_NOASYNC 0x00800000 /* disable async */
419 #define MNTK_UNMOUNT 0x01000000 /* unmount in progress */
420 #define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */
421 #define MNTK_SUSPEND 0x08000000 /* request write suspension */
422 #define MNTK_SUSPEND2 0x04000000 /* block secondary writes */
423 #define MNTK_SUSPENDED 0x10000000 /* write operations are suspended */
424 #define MNTK_NULL_NOCACHE 0x20000000 /* auto disable cache for nullfs
425 mounts over this fs */
426 #define MNTK_LOOKUP_SHARED 0x40000000 /* FS supports shared lock lookups */
427 #define MNTK_NOKNOTE 0x80000000 /* Don't send KNOTEs from VOP hooks */
431 MNT_SHARED_WRITES(struct mount *mp)
434 return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0);
438 MNT_EXTENDED_SHARED(struct mount *mp)
441 return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0);
446 * Sysctl CTL_VFS definitions.
448 * Second level identifier specifies which filesystem. Second level
449 * identifier VFS_VFSCONF returns information about all filesystems.
450 * Second level identifier VFS_GENERIC is non-terminal.
452 #define VFS_VFSCONF 0 /* get configured filesystems */
453 #define VFS_GENERIC 0 /* generic filesystem information */
455 * Third level identifiers for VFS_GENERIC are given below; third
456 * level identifiers for specific filesystems are given in their
457 * mount specific header files.
459 #define VFS_MAXTYPENUM 1 /* int: highest defined filesystem type */
460 #define VFS_CONF 2 /* struct: vfsconf for filesystem given
464 * Flags for various system call interfaces.
466 * waitfor flags to vfs_sync() and getfsstat()
468 #define MNT_WAIT 1 /* synchronously wait for I/O to complete */
469 #define MNT_NOWAIT 2 /* start all I/O, but do not wait for it */
470 #define MNT_LAZY 3 /* push data not written by filesystem syncer */
471 #define MNT_SUSPEND 4 /* Suspend file system after sync */
474 * Generic file handle
477 fsid_t fh_fsid; /* Filesystem id of mount point */
478 struct fid fh_fid; /* Filesys specific id */
480 typedef struct fhandle fhandle_t;
483 * Old export arguments without security flavor list
485 struct oexport_args {
486 int ex_flags; /* export related flags */
487 uid_t ex_root; /* mapping for root uid */
488 struct xucred ex_anon; /* mapping for anonymous user */
489 struct sockaddr *ex_addr; /* net address to which exported */
490 u_char ex_addrlen; /* and the net address length */
491 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
492 u_char ex_masklen; /* and the smask length */
493 char *ex_indexfile; /* index file for WebNFS URLs */
497 * Export arguments for local filesystem mount calls.
499 #define MAXSECFLAVORS 5
501 int ex_flags; /* export related flags */
502 uid_t ex_root; /* mapping for root uid */
503 struct xucred ex_anon; /* mapping for anonymous user */
504 struct sockaddr *ex_addr; /* net address to which exported */
505 u_char ex_addrlen; /* and the net address length */
506 struct sockaddr *ex_mask; /* mask of valid bits in saddr */
507 u_char ex_masklen; /* and the smask length */
508 char *ex_indexfile; /* index file for WebNFS URLs */
509 int ex_numsecflavors; /* security flavor count */
510 int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */
514 * Structure holding information for a publicly exported filesystem
515 * (WebNFS). Currently the specs allow just for one such filesystem.
518 int np_valid; /* Do we hold valid information */
519 fhandle_t np_handle; /* Filehandle for pub fs (internal) */
520 struct mount *np_mount; /* Mountpoint of exported fs */
521 char *np_index; /* Index file */
525 * Filesystem configuration information. One of these exists for each
526 * type of filesystem supported by the kernel. These are searched at
527 * mount time to identify the requested filesystem.
529 * XXX: Never change the first two arguments!
532 u_int vfc_version; /* ABI version number */
533 char vfc_name[MFSNAMELEN]; /* filesystem type name */
534 struct vfsops *vfc_vfsops; /* filesystem operations vector */
535 struct vfsops *vfc_vfsops_sd; /* ... signal-deferred */
536 int vfc_typenum; /* historic filesystem type number */
537 int vfc_refcount; /* number mounted of this type */
538 int vfc_flags; /* permanent flags */
539 int vfc_prison_flag; /* prison allow.mount.* flag */
540 struct vfsoptdecl *vfc_opts; /* mount options */
541 TAILQ_ENTRY(vfsconf) vfc_list; /* list of vfscons */
544 /* Userland version of the struct vfsconf. */
546 struct vfsops *vfc_vfsops; /* filesystem operations vector */
547 char vfc_name[MFSNAMELEN]; /* filesystem type name */
548 int vfc_typenum; /* historic filesystem type number */
549 int vfc_refcount; /* number mounted of this type */
550 int vfc_flags; /* permanent flags */
551 struct vfsconf *vfc_next; /* next in list */
565 * NB: these flags refer to IMPLEMENTATION properties, not properties of
566 * any actual mounts; i.e., it does not make sense to change the flags.
568 #define VFCF_STATIC 0x00010000 /* statically compiled into kernel */
569 #define VFCF_NETWORK 0x00020000 /* may get data over the network */
570 #define VFCF_READONLY 0x00040000 /* writes are not implemented */
571 #define VFCF_SYNTHETIC 0x00080000 /* data does not represent real files */
572 #define VFCF_LOOPBACK 0x00100000 /* aliases some other mounted FS */
573 #define VFCF_UNICODE 0x00200000 /* stores file names as Unicode */
574 #define VFCF_JAIL 0x00400000 /* can be mounted from within a jail */
575 #define VFCF_DELEGADMIN 0x00800000 /* supports delegated administration */
576 #define VFCF_SBDRY 0x01000000 /* Stop at Boundary: defer stop requests
577 to kernel->user (AST) transition */
579 typedef uint32_t fsctlop_t;
582 int vc_vers; /* should be VFSIDCTL_VERS1 (below) */
583 fsid_t vc_fsid; /* fsid to operate on */
584 char vc_fstypename[MFSNAMELEN];
585 /* type of fs 'nfs' or '*' */
586 fsctlop_t vc_op; /* operation VFS_CTL_* (below) */
587 void *vc_ptr; /* pointer to data structure */
588 size_t vc_len; /* sizeof said structure */
589 u_int32_t vc_spare[12]; /* spare (must be zero) */
592 /* vfsidctl API version. */
593 #define VFS_CTL_VERS1 0x01
596 * New style VFS sysctls, do not reuse/conflict with the namespace for
598 * All "global" sysctl ops have the 33rd bit set:
600 * Private sysctl ops should have the 33rd bit unset.
602 #define VFS_CTL_QUERY 0x00010001 /* anything wrong? (vfsquery) */
603 #define VFS_CTL_TIMEO 0x00010002 /* set timeout for vfs notification */
604 #define VFS_CTL_NOLOCKS 0x00010003 /* disable file locking */
608 u_int32_t vq_spare[31];
612 #define VQ_NOTRESP 0x0001 /* server down */
613 #define VQ_NEEDAUTH 0x0002 /* server bad auth */
614 #define VQ_LOWDISK 0x0004 /* we're low on space */
615 #define VQ_MOUNT 0x0008 /* new filesystem arrived */
616 #define VQ_UNMOUNT 0x0010 /* filesystem has left */
617 #define VQ_DEAD 0x0020 /* filesystem is dead, needs force unmount */
618 #define VQ_ASSIST 0x0040 /* filesystem needs assistance from external
620 #define VQ_NOTRESPLOCK 0x0080 /* server lockd down */
621 #define VQ_FLAG0100 0x0100 /* placeholder */
622 #define VQ_FLAG0200 0x0200 /* placeholder */
623 #define VQ_FLAG0400 0x0400 /* placeholder */
624 #define VQ_FLAG0800 0x0800 /* placeholder */
625 #define VQ_FLAG1000 0x1000 /* placeholder */
626 #define VQ_FLAG2000 0x2000 /* placeholder */
627 #define VQ_FLAG4000 0x4000 /* placeholder */
628 #define VQ_FLAG8000 0x8000 /* placeholder */
631 /* Point a sysctl request at a vfsidctl's data. */
632 #define VCTLTOREQ(vc, req) \
634 (req)->newptr = (vc)->vc_ptr; \
635 (req)->newlen = (vc)->vc_len; \
646 * vfs_busy specific flags and mask.
648 #define MBF_NOWAIT 0x01
649 #define MBF_MNTLSTLOCK 0x02
650 #define MBF_MASK (MBF_NOWAIT | MBF_MNTLSTLOCK)
652 #ifdef MALLOC_DECLARE
653 MALLOC_DECLARE(M_MOUNT);
654 MALLOC_DECLARE(M_STATFS);
656 extern int maxvfsconf; /* highest defined filesystem type */
658 TAILQ_HEAD(vfsconfhead, vfsconf);
659 extern struct vfsconfhead vfsconf;
662 * Operations supported on mounted filesystem.
670 * N.B., vfs_cmount is the ancient vfsop invoked by the old mount(2) syscall.
671 * The new way is vfs_mount.
673 * vfs_cmount implementations typically translate arguments from their
674 * respective old per-FS structures into the key-value list supported by
675 * nmount(2), then use kernel_mount(9) to mimic nmount(2) from kernelspace.
677 * Filesystems with mounters that use nmount(2) do not need to and should not
678 * implement vfs_cmount. Hopefully a future cleanup can remove vfs_cmount and
681 typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags);
682 typedef int vfs_unmount_t(struct mount *mp, int mntflags);
683 typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp);
684 typedef int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg);
685 typedef int vfs_statfs_t(struct mount *mp, struct statfs *sbp);
686 typedef int vfs_sync_t(struct mount *mp, int waitfor);
687 typedef int vfs_vget_t(struct mount *mp, ino_t ino, int flags,
689 typedef int vfs_fhtovp_t(struct mount *mp, struct fid *fhp,
690 int flags, struct vnode **vpp);
691 typedef int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam,
692 int *extflagsp, struct ucred **credanonp,
693 int *numsecflavors, int **secflavors);
694 typedef int vfs_init_t(struct vfsconf *);
695 typedef int vfs_uninit_t(struct vfsconf *);
696 typedef int vfs_extattrctl_t(struct mount *mp, int cmd,
697 struct vnode *filename_vp, int attrnamespace,
698 const char *attrname);
699 typedef int vfs_mount_t(struct mount *mp);
700 typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op,
701 struct sysctl_req *req);
702 typedef void vfs_susp_clean_t(struct mount *mp);
703 typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp);
704 typedef void vfs_purge_t(struct mount *mp);
707 vfs_mount_t *vfs_mount;
708 vfs_cmount_t *vfs_cmount;
709 vfs_unmount_t *vfs_unmount;
710 vfs_root_t *vfs_root;
711 vfs_root_t *vfs_cachedroot;
712 vfs_quotactl_t *vfs_quotactl;
713 vfs_statfs_t *vfs_statfs;
714 vfs_sync_t *vfs_sync;
715 vfs_vget_t *vfs_vget;
716 vfs_fhtovp_t *vfs_fhtovp;
717 vfs_checkexp_t *vfs_checkexp;
718 vfs_init_t *vfs_init;
719 vfs_uninit_t *vfs_uninit;
720 vfs_extattrctl_t *vfs_extattrctl;
721 vfs_sysctl_t *vfs_sysctl;
722 vfs_susp_clean_t *vfs_susp_clean;
723 vfs_notify_lowervp_t *vfs_reclaim_lowervp;
724 vfs_notify_lowervp_t *vfs_unlink_lowervp;
725 vfs_purge_t *vfs_purge;
726 vfs_mount_t *vfs_spare[6]; /* spares for ABI compat */
729 vfs_statfs_t __vfs_statfs;
731 #define VFS_MOUNT(MP) ({ \
734 TSRAW(curthread, TS_ENTER, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
735 _rc = (*(MP)->mnt_op->vfs_mount)(MP); \
736 TSRAW(curthread, TS_EXIT, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
739 #define VFS_UNMOUNT(MP, FORCE) ({ \
742 _rc = (*(MP)->mnt_op->vfs_unmount)(MP, FORCE); \
745 #define VFS_ROOT(MP, FLAGS, VPP) ({ \
748 _rc = (*(MP)->mnt_op->vfs_root)(MP, FLAGS, VPP); \
751 #define VFS_CACHEDROOT(MP, FLAGS, VPP) ({ \
754 _rc = (*(MP)->mnt_op->vfs_cachedroot)(MP, FLAGS, VPP); \
757 #define VFS_QUOTACTL(MP, C, U, A) ({ \
760 _rc = (*(MP)->mnt_op->vfs_quotactl)(MP, C, U, A); \
763 #define VFS_STATFS(MP, SBP) ({ \
766 _rc = __vfs_statfs((MP), (SBP)); \
769 #define VFS_SYNC(MP, WAIT) ({ \
772 _rc = (*(MP)->mnt_op->vfs_sync)(MP, WAIT); \
775 #define VFS_VGET(MP, INO, FLAGS, VPP) ({ \
778 _rc = (*(MP)->mnt_op->vfs_vget)(MP, INO, FLAGS, VPP); \
781 #define VFS_FHTOVP(MP, FIDP, FLAGS, VPP) ({ \
784 _rc = (*(MP)->mnt_op->vfs_fhtovp)(MP, FIDP, FLAGS, VPP); \
787 #define VFS_CHECKEXP(MP, NAM, EXFLG, CRED, NUMSEC, SEC) ({ \
790 _rc = (*(MP)->mnt_op->vfs_checkexp)(MP, NAM, EXFLG, CRED, NUMSEC,\
794 #define VFS_EXTATTRCTL(MP, C, FN, NS, N) ({ \
797 _rc = (*(MP)->mnt_op->vfs_extattrctl)(MP, C, FN, NS, N); \
800 #define VFS_SYSCTL(MP, OP, REQ) ({ \
803 _rc = (*(MP)->mnt_op->vfs_sysctl)(MP, OP, REQ); \
806 #define VFS_SUSP_CLEAN(MP) do { \
807 if (*(MP)->mnt_op->vfs_susp_clean != NULL) { \
808 (*(MP)->mnt_op->vfs_susp_clean)(MP); \
812 #define VFS_RECLAIM_LOWERVP(MP, VP) do { \
813 if (*(MP)->mnt_op->vfs_reclaim_lowervp != NULL) { \
814 (*(MP)->mnt_op->vfs_reclaim_lowervp)((MP), (VP)); \
818 #define VFS_UNLINK_LOWERVP(MP, VP) do { \
819 if (*(MP)->mnt_op->vfs_unlink_lowervp != NULL) { \
820 (*(MP)->mnt_op->vfs_unlink_lowervp)((MP), (VP)); \
824 #define VFS_PURGE(MP) do { \
825 if (*(MP)->mnt_op->vfs_purge != NULL) { \
826 (*(MP)->mnt_op->vfs_purge)(MP); \
830 #define VFS_KNOTE_LOCKED(vp, hint) do \
832 if (((vp)->v_vflag & VV_NOKNOTE) == 0) \
833 VN_KNOTE((vp), (hint), KNF_LISTLOCKED); \
836 #define VFS_KNOTE_UNLOCKED(vp, hint) do \
838 if (((vp)->v_vflag & VV_NOKNOTE) == 0) \
839 VN_KNOTE((vp), (hint), 0); \
842 #define VFS_NOTIFY_UPPER_RECLAIM 1
843 #define VFS_NOTIFY_UPPER_UNLINK 2
845 #include <sys/module.h>
850 #define VFS_VERSION_00 0x19660120
851 #define VFS_VERSION_01 0x20121030
852 #define VFS_VERSION_02 0x20180504
853 #define VFS_VERSION VFS_VERSION_02
855 #define VFS_SET(vfsops, fsname, flags) \
856 static struct vfsconf fsname ## _vfsconf = { \
857 .vfc_version = VFS_VERSION, \
858 .vfc_name = #fsname, \
859 .vfc_vfsops = &vfsops, \
861 .vfc_flags = flags, \
863 static moduledata_t fsname ## _mod = { \
866 & fsname ## _vfsconf \
868 DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE)
871 * exported vnode operations
874 int dounmount(struct mount *, int, struct thread *);
876 int kernel_mount(struct mntarg *ma, uint64_t flags);
877 int kernel_vmount(int flags, ...);
878 struct mntarg *mount_arg(struct mntarg *ma, const char *name, const void *val, int len);
879 struct mntarg *mount_argb(struct mntarg *ma, int flag, const char *name);
880 struct mntarg *mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...);
881 struct mntarg *mount_argsu(struct mntarg *ma, const char *name, const void *val, int len);
882 void statfs_scale_blocks(struct statfs *sf, long max_size);
883 struct vfsconf *vfs_byname(const char *);
884 struct vfsconf *vfs_byname_kld(const char *, struct thread *td, int *);
885 void vfs_mount_destroy(struct mount *);
886 void vfs_event_signal(fsid_t *, u_int32_t, intptr_t);
887 void vfs_freeopts(struct vfsoptlist *opts);
888 void vfs_deleteopt(struct vfsoptlist *opts, const char *name);
889 int vfs_buildopts(struct uio *auio, struct vfsoptlist **options);
890 int vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w,
892 int vfs_getopt(struct vfsoptlist *, const char *, void **, int *);
893 int vfs_getopt_pos(struct vfsoptlist *opts, const char *name);
894 int vfs_getopt_size(struct vfsoptlist *opts, const char *name,
896 char *vfs_getopts(struct vfsoptlist *, const char *, int *error);
897 int vfs_copyopt(struct vfsoptlist *, const char *, void *, int);
898 int vfs_filteropt(struct vfsoptlist *, const char **legal);
899 void vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...);
900 int vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...);
901 int vfs_setopt(struct vfsoptlist *opts, const char *name, void *value,
903 int vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value,
905 int vfs_setopts(struct vfsoptlist *opts, const char *name,
907 int vfs_setpublicfs /* set publicly exported fs */
908 (struct mount *, struct netexport *, struct export_args *);
909 void vfs_periodic(struct mount *, int);
910 int vfs_busy(struct mount *, int);
911 int vfs_export /* process mount export info */
912 (struct mount *, struct export_args *);
913 void vfs_allocate_syncvnode(struct mount *);
914 void vfs_deallocate_syncvnode(struct mount *);
915 int vfs_donmount(struct thread *td, uint64_t fsflags,
916 struct uio *fsoptions);
917 void vfs_getnewfsid(struct mount *);
918 struct cdev *vfs_getrootfsid(struct mount *);
919 struct mount *vfs_getvfs(fsid_t *); /* return vfs given fsid */
920 struct mount *vfs_busyfs(fsid_t *);
921 int vfs_modevent(module_t, int, void *);
922 void vfs_mount_error(struct mount *, const char *, ...);
923 void vfs_mountroot(void); /* mount our root filesystem */
924 void vfs_mountedfrom(struct mount *, const char *from);
925 void vfs_notify_upper(struct vnode *, int);
926 void vfs_oexport_conv(const struct oexport_args *oexp,
927 struct export_args *exp);
928 void vfs_ref(struct mount *);
929 void vfs_rel(struct mount *);
930 struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *,
932 int vfs_suser(struct mount *, struct thread *);
933 void vfs_unbusy(struct mount *);
934 void vfs_unmountall(void);
935 extern TAILQ_HEAD(mntlist, mount) mountlist; /* mounted filesystem list */
936 extern struct mtx mountlist_mtx;
937 extern struct nfs_public nfs_pub;
938 extern struct sx vfsconf_sx;
939 #define vfsconf_lock() sx_xlock(&vfsconf_sx)
940 #define vfsconf_unlock() sx_xunlock(&vfsconf_sx)
941 #define vfsconf_slock() sx_slock(&vfsconf_sx)
942 #define vfsconf_sunlock() sx_sunlock(&vfsconf_sx)
945 * Declarations for these vfs default operations are located in
946 * kern/vfs_default.c. They will be automatically used to replace
947 * null entries in VFS ops tables when registering a new filesystem
948 * type in the global table.
950 vfs_root_t vfs_stdroot;
951 vfs_quotactl_t vfs_stdquotactl;
952 vfs_statfs_t vfs_stdstatfs;
953 vfs_sync_t vfs_stdsync;
954 vfs_sync_t vfs_stdnosync;
955 vfs_vget_t vfs_stdvget;
956 vfs_fhtovp_t vfs_stdfhtovp;
957 vfs_checkexp_t vfs_stdcheckexp;
958 vfs_init_t vfs_stdinit;
959 vfs_uninit_t vfs_stduninit;
960 vfs_extattrctl_t vfs_stdextattrctl;
961 vfs_sysctl_t vfs_stdsysctl;
963 void syncer_suspend(void);
964 void syncer_resume(void);
966 struct vnode *vfs_cache_root_clear(struct mount *);
967 void vfs_cache_root_set(struct mount *, struct vnode *);
969 void vfs_op_barrier_wait(struct mount *);
970 void vfs_op_enter(struct mount *);
971 void vfs_op_exit_locked(struct mount *);
972 void vfs_op_exit(struct mount *);
975 void vfs_assert_mount_counters(struct mount *);
976 void vfs_dump_mount_counters(struct mount *);
978 #define vfs_assert_mount_counters(mp) do { } while (0)
979 #define vfs_dump_mount_counters(mp) do { } while (0)
982 enum mount_counter { MNT_COUNT_REF, MNT_COUNT_LOCKREF, MNT_COUNT_WRITEOPCOUNT };
983 int vfs_mount_fetch_counter(struct mount *, enum mount_counter);
986 * We mark ourselves as entering the section and post a sequentially consistent
987 * fence, meaning the store is completed before we get into the section and
988 * mnt_vfs_ops is only read afterwards.
990 * Any thread transitioning the ops counter 0->1 does things in the opposite
991 * order - first bumps the count, posts a sequentially consistent fence and
992 * observes all CPUs not executing within the section.
994 * This provides an invariant that by the time the last CPU is observed not
995 * executing, everyone else entering will see the counter > 0 and exit.
997 * Note there is no barrier between vfs_ops and the rest of the code in the
998 * section. It is not necessary as the writer has to wait for everyone to drain
999 * before making any changes or only make changes safe while the section is
1002 #define vfs_op_thread_entered(mp) ({ \
1003 MPASS(curthread->td_critnest > 0); \
1004 *(int *)zpcpu_get(mp->mnt_thread_in_ops_pcpu) == 1; \
1007 #define vfs_op_thread_enter(mp) ({ \
1008 bool _retval = true; \
1010 MPASS(!vfs_op_thread_entered(mp)); \
1011 *(int *)zpcpu_get(mp->mnt_thread_in_ops_pcpu) = 1; \
1012 atomic_thread_fence_seq_cst(); \
1013 if (__predict_false(mp->mnt_vfs_ops > 0)) { \
1014 vfs_op_thread_exit(mp); \
1020 #define vfs_op_thread_exit(mp) do { \
1021 MPASS(vfs_op_thread_entered(mp)); \
1022 atomic_thread_fence_rel(); \
1023 *(int *)zpcpu_get(mp->mnt_thread_in_ops_pcpu) = 0; \
1027 #define vfs_mp_count_add_pcpu(mp, count, val) do { \
1028 MPASS(vfs_op_thread_entered(mp)); \
1029 (*(int *)zpcpu_get(mp->mnt_##count##_pcpu)) += val; \
1032 #define vfs_mp_count_sub_pcpu(mp, count, val) do { \
1033 MPASS(vfs_op_thread_entered(mp)); \
1034 (*(int *)zpcpu_get(mp->mnt_##count##_pcpu)) -= val; \
1037 #else /* !_KERNEL */
1039 #include <sys/cdefs.h>
1044 int fhlink(struct fhandle *, const char *);
1045 int fhlinkat(struct fhandle *, int, const char *);
1046 int fhopen(const struct fhandle *, int);
1047 int fhreadlink(struct fhandle *, char *, size_t);
1048 int fhstat(const struct fhandle *, struct stat *);
1049 int fhstatfs(const struct fhandle *, struct statfs *);
1050 int fstatfs(int, struct statfs *);
1051 int getfh(const char *, fhandle_t *);
1052 int getfhat(int, char *, struct fhandle *, int);
1053 int getfsstat(struct statfs *, long, int);
1054 int getmntinfo(struct statfs **, int);
1055 int lgetfh(const char *, fhandle_t *);
1056 int mount(const char *, const char *, int, void *);
1057 int nmount(struct iovec *, unsigned int, int);
1058 int statfs(const char *, struct statfs *);
1059 int unmount(const char *, int);
1061 /* C library stuff */
1062 int getvfsbyname(const char *, struct xvfsconf *);
1065 #endif /* _KERNEL */
1067 #endif /* !_SYS_MOUNT_H_ */