1/*-
2 * Copyright (c) 1989, 1991, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 *	@(#)mount.h	8.21 (Berkeley) 5/20/95
30 * $FreeBSD$
31 */
32
33#ifndef _SYS_MOUNT_H_
34#define _SYS_MOUNT_H_
35
36#include <sys/ucred.h>
37#include <sys/queue.h>
38#ifdef _KERNEL
39#include <sys/lock.h>
40#include <sys/lockmgr.h>
41#include <sys/_mutex.h>
42#include <sys/_sx.h>
43#endif
44
45/*
46 * NOTE: When changing statfs structure, mount structure, MNT_* flags or
47 * MNTK_* flags also update DDB show mount command in vfs_subr.c.
48 */
49
50typedef struct fsid { int32_t val[2]; } fsid_t;	/* filesystem id type */
51
52/*
53 * File identifier.
54 * These are unique per filesystem on a single machine.
55 */
56#define	MAXFIDSZ	16
57
58struct fid {
59	u_short		fid_len;		/* length of data in bytes */
60	u_short		fid_data0;		/* force longword alignment */
61	char		fid_data[MAXFIDSZ];	/* data (variable length) */
62};
63
64/*
65 * filesystem statistics
66 */
67#define	MFSNAMELEN	16		/* length of type name including null */
68#define	MNAMELEN	88		/* size of on/from name bufs */
69#define	STATFS_VERSION	0x20030518	/* current version number */
70struct statfs {
71	uint32_t f_version;		/* structure version number */
72	uint32_t f_type;		/* type of filesystem */
73	uint64_t f_flags;		/* copy of mount exported flags */
74	uint64_t f_bsize;		/* filesystem fragment size */
75	uint64_t f_iosize;		/* optimal transfer block size */
76	uint64_t f_blocks;		/* total data blocks in filesystem */
77	uint64_t f_bfree;		/* free blocks in filesystem */
78	int64_t	 f_bavail;		/* free blocks avail to non-superuser */
79	uint64_t f_files;		/* total file nodes in filesystem */
80	int64_t	 f_ffree;		/* free nodes avail to non-superuser */
81	uint64_t f_syncwrites;		/* count of sync writes since mount */
82	uint64_t f_asyncwrites;		/* count of async writes since mount */
83	uint64_t f_syncreads;		/* count of sync reads since mount */
84	uint64_t f_asyncreads;		/* count of async reads since mount */
85	uint64_t f_spare[10];		/* unused spare */
86	uint32_t f_namemax;		/* maximum filename length */
87	uid_t	  f_owner;		/* user that mounted the filesystem */
88	fsid_t	  f_fsid;		/* filesystem id */
89	char	  f_charspare[80];	    /* spare string space */
90	char	  f_fstypename[MFSNAMELEN]; /* filesystem type name */
91	char	  f_mntfromname[MNAMELEN];  /* mounted filesystem */
92	char	  f_mntonname[MNAMELEN];    /* directory on which mounted */
93};
94
95#ifdef _KERNEL
96#define	OMFSNAMELEN	16	/* length of fs type name, including null */
97#define	OMNAMELEN	(88 - 2 * sizeof(long))	/* size of on/from name bufs */
98
99/* XXX getfsstat.2 is out of date with write and read counter changes here. */
100/* XXX statfs.2 is out of date with read counter changes here. */
101struct ostatfs {
102	long	f_spare2;		/* placeholder */
103	long	f_bsize;		/* fundamental filesystem block size */
104	long	f_iosize;		/* optimal transfer block size */
105	long	f_blocks;		/* total data blocks in filesystem */
106	long	f_bfree;		/* free blocks in fs */
107	long	f_bavail;		/* free blocks avail to non-superuser */
108	long	f_files;		/* total file nodes in filesystem */
109	long	f_ffree;		/* free file nodes in fs */
110	fsid_t	f_fsid;			/* filesystem id */
111	uid_t	f_owner;		/* user that mounted the filesystem */
112	int	f_type;			/* type of filesystem */
113	int	f_flags;		/* copy of mount exported flags */
114	long	f_syncwrites;		/* count of sync writes since mount */
115	long	f_asyncwrites;		/* count of async writes since mount */
116	char	f_fstypename[OMFSNAMELEN]; /* fs type name */
117	char	f_mntonname[OMNAMELEN];	/* directory on which mounted */
118	long	f_syncreads;		/* count of sync reads since mount */
119	long	f_asyncreads;		/* count of async reads since mount */
120	short	f_spares1;		/* unused spare */
121	char	f_mntfromname[OMNAMELEN];/* mounted filesystem */
122	short	f_spares2;		/* unused spare */
123	/*
124	 * XXX on machines where longs are aligned to 8-byte boundaries, there
125	 * is an unnamed int32_t here.  This spare was after the apparent end
126	 * of the struct until we bit off the read counters from f_mntonname.
127	 */
128	long	f_spare[2];		/* unused spare */
129};
130
131TAILQ_HEAD(vnodelst, vnode);
132
133/* Mount options list */
134TAILQ_HEAD(vfsoptlist, vfsopt);
135struct vfsopt {
136	TAILQ_ENTRY(vfsopt) link;
137	char	*name;
138	void	*value;
139	int	len;
140	int	pos;
141	int	seen;
142};
143
144/*
145 * Structure per mounted filesystem.  Each mounted filesystem has an
146 * array of operations and an instance record.  The filesystems are
147 * put on a doubly linked list.
148 *
149 * Lock reference:
150 *	m - mountlist_mtx
151 *	i - interlock
152 *	v - vnode freelist mutex
153 *
154 * Unmarked fields are considered stable as long as a ref is held.
155 *
156 */
157struct mount {
158	struct mtx	mnt_mtx;		/* mount structure interlock */
159	int		mnt_gen;		/* struct mount generation */
160#define	mnt_startzero	mnt_list
161	TAILQ_ENTRY(mount) mnt_list;		/* (m) mount list */
162	struct vfsops	*mnt_op;		/* operations on fs */
163	struct vfsconf	*mnt_vfc;		/* configuration info */
164	struct vnode	*mnt_vnodecovered;	/* vnode we mounted on */
165	struct vnode	*mnt_syncer;		/* syncer vnode */
166	int		mnt_ref;		/* (i) Reference count */
167	struct vnodelst	mnt_nvnodelist;		/* (i) list of vnodes */
168	int		mnt_nvnodelistsize;	/* (i) # of vnodes */
169	struct vnodelst	mnt_activevnodelist;	/* (v) list of active vnodes */
170	int		mnt_activevnodelistsize;/* (v) # of active vnodes */
171	int		mnt_writeopcount;	/* (i) write syscalls pending */
172	int		mnt_kern_flag;		/* (i) kernel only flags */
173	uint64_t	mnt_flag;		/* (i) flags shared with user */
174	struct vfsoptlist *mnt_opt;		/* current mount options */
175	struct vfsoptlist *mnt_optnew;		/* new options passed to fs */
176	int		mnt_maxsymlinklen;	/* max size of short symlink */
177	struct statfs	mnt_stat;		/* cache of filesystem stats */
178	struct ucred	*mnt_cred;		/* credentials of mounter */
179	void *		mnt_data;		/* private data */
180	time_t		mnt_time;		/* last time written*/
181	int		mnt_iosize_max;		/* max size for clusters, etc */
182	struct netexport *mnt_export;		/* export list */
183	struct label	*mnt_label;		/* MAC label for the fs */
184	u_int		mnt_hashseed;		/* Random seed for vfs_hash */
185	int		mnt_lockref;		/* (i) Lock reference count */
186	int		mnt_secondary_writes;   /* (i) # of secondary writes */
187	int		mnt_secondary_accwrites;/* (i) secondary wr. starts */
188	struct thread	*mnt_susp_owner;	/* (i) thread owning suspension */
189#define	mnt_endzero	mnt_gjprovider
190	char		*mnt_gjprovider;	/* gjournal provider name */
191	struct lock	mnt_explock;		/* vfs_export walkers lock */
192	TAILQ_ENTRY(mount) mnt_upper_link;	/* (m) we in the all uppers */
193	TAILQ_HEAD(, mount) mnt_uppers;		/* (m) upper mounts over us*/
194};
195
196/*
197 * Definitions for MNT_VNODE_FOREACH_ALL.
198 */
199struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp);
200struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp);
201void          __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp);
202
203#define MNT_VNODE_FOREACH_ALL(vp, mp, mvp)				\
204	for (vp = __mnt_vnode_first_all(&(mvp), (mp));			\
205		(vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp)))
206
207#define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp)				\
208	do {								\
209		MNT_ILOCK(mp);						\
210		__mnt_vnode_markerfree_all(&(mvp), (mp));		\
211		/* MNT_IUNLOCK(mp); -- done in above function */	\
212		mtx_assert(MNT_MTX(mp), MA_NOTOWNED);			\
213	} while (0)
214
215/*
216 * Definitions for MNT_VNODE_FOREACH_ACTIVE.
217 */
218struct vnode *__mnt_vnode_next_active(struct vnode **mvp, struct mount *mp);
219struct vnode *__mnt_vnode_first_active(struct vnode **mvp, struct mount *mp);
220void          __mnt_vnode_markerfree_active(struct vnode **mvp, struct mount *);
221
222#define MNT_VNODE_FOREACH_ACTIVE(vp, mp, mvp) 				\
223	for (vp = __mnt_vnode_first_active(&(mvp), (mp)); 		\
224		(vp) != NULL; vp = __mnt_vnode_next_active(&(mvp), (mp)))
225
226#define MNT_VNODE_FOREACH_ACTIVE_ABORT(mp, mvp)				\
227	__mnt_vnode_markerfree_active(&(mvp), (mp))
228
229#define	MNT_ILOCK(mp)	mtx_lock(&(mp)->mnt_mtx)
230#define	MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx)
231#define	MNT_IUNLOCK(mp)	mtx_unlock(&(mp)->mnt_mtx)
232#define	MNT_MTX(mp)	(&(mp)->mnt_mtx)
233#define	MNT_REF(mp)	(mp)->mnt_ref++
234#define	MNT_REL(mp)	do {						\
235	KASSERT((mp)->mnt_ref > 0, ("negative mnt_ref"));		\
236	(mp)->mnt_ref--;						\
237	if ((mp)->mnt_ref == 0)						\
238		wakeup((mp));						\
239} while (0)
240
241#endif /* _KERNEL */
242
243/*
244 * User specifiable flags, stored in mnt_flag.
245 */
246#define	MNT_RDONLY	0x0000000000000001ULL /* read only filesystem */
247#define	MNT_SYNCHRONOUS	0x0000000000000002ULL /* fs written synchronously */
248#define	MNT_NOEXEC	0x0000000000000004ULL /* can't exec from filesystem */
249#define	MNT_NOSUID	0x0000000000000008ULL /* don't honor setuid fs bits */
250#define	MNT_NFS4ACLS	0x0000000000000010ULL /* enable NFS version 4 ACLs */
251#define	MNT_UNION	0x0000000000000020ULL /* union with underlying fs */
252#define	MNT_ASYNC	0x0000000000000040ULL /* fs written asynchronously */
253#define	MNT_SUIDDIR	0x0000000000100000ULL /* special SUID dir handling */
254#define	MNT_SOFTDEP	0x0000000000200000ULL /* using soft updates */
255#define	MNT_NOSYMFOLLOW	0x0000000000400000ULL /* do not follow symlinks */
256#define	MNT_GJOURNAL	0x0000000002000000ULL /* GEOM journal support enabled */
257#define	MNT_MULTILABEL	0x0000000004000000ULL /* MAC support for objects */
258#define	MNT_ACLS	0x0000000008000000ULL /* ACL support enabled */
259#define	MNT_NOATIME	0x0000000010000000ULL /* dont update file access time */
260#define	MNT_NOCLUSTERR	0x0000000040000000ULL /* disable cluster read */
261#define	MNT_NOCLUSTERW	0x0000000080000000ULL /* disable cluster write */
262#define	MNT_SUJ		0x0000000100000000ULL /* using journaled soft updates */
263#define	MNT_AUTOMOUNTED	0x0000000200000000ULL /* mounted by automountd(8) */
264
265/*
266 * NFS export related mount flags.
267 */
268#define	MNT_EXRDONLY	0x0000000000000080ULL	/* exported read only */
269#define	MNT_EXPORTED	0x0000000000000100ULL	/* filesystem is exported */
270#define	MNT_DEFEXPORTED	0x0000000000000200ULL	/* exported to the world */
271#define	MNT_EXPORTANON	0x0000000000000400ULL	/* anon uid mapping for all */
272#define	MNT_EXKERB	0x0000000000000800ULL	/* exported with Kerberos */
273#define	MNT_EXPUBLIC	0x0000000020000000ULL	/* public export (WebNFS) */
274
275/*
276 * Flags set by internal operations,
277 * but visible to the user.
278 * XXX some of these are not quite right.. (I've never seen the root flag set)
279 */
280#define	MNT_LOCAL	0x0000000000001000ULL /* filesystem is stored locally */
281#define	MNT_QUOTA	0x0000000000002000ULL /* quotas are enabled on fs */
282#define	MNT_ROOTFS	0x0000000000004000ULL /* identifies the root fs */
283#define	MNT_USER	0x0000000000008000ULL /* mounted by a user */
284#define	MNT_IGNORE	0x0000000000800000ULL /* do not show entry in df */
285
286/*
287 * Mask of flags that are visible to statfs().
288 * XXX I think that this could now become (~(MNT_CMDFLAGS))
289 * but the 'mount' program may need changing to handle this.
290 */
291#define	MNT_VISFLAGMASK	(MNT_RDONLY	| MNT_SYNCHRONOUS | MNT_NOEXEC	| \
292			MNT_NOSUID	| MNT_UNION	| MNT_SUJ	| \
293			MNT_ASYNC	| MNT_EXRDONLY	| MNT_EXPORTED	| \
294			MNT_DEFEXPORTED	| MNT_EXPORTANON| MNT_EXKERB	| \
295			MNT_LOCAL	| MNT_USER	| MNT_QUOTA	| \
296			MNT_ROOTFS	| MNT_NOATIME	| MNT_NOCLUSTERR| \
297			MNT_NOCLUSTERW	| MNT_SUIDDIR	| MNT_SOFTDEP	| \
298			MNT_IGNORE	| MNT_EXPUBLIC	| MNT_NOSYMFOLLOW | \
299			MNT_GJOURNAL	| MNT_MULTILABEL | MNT_ACLS	| \
300			MNT_NFS4ACLS	| MNT_AUTOMOUNTED)
301
302/* Mask of flags that can be updated. */
303#define	MNT_UPDATEMASK (MNT_NOSUID	| MNT_NOEXEC	| \
304			MNT_SYNCHRONOUS	| MNT_UNION	| MNT_ASYNC	| \
305			MNT_NOATIME | \
306			MNT_NOSYMFOLLOW	| MNT_IGNORE	| \
307			MNT_NOCLUSTERR	| MNT_NOCLUSTERW | MNT_SUIDDIR	| \
308			MNT_ACLS	| MNT_USER	| MNT_NFS4ACLS	| \
309			MNT_AUTOMOUNTED)
310
311/*
312 * External filesystem command modifier flags.
313 * Unmount can use the MNT_FORCE flag.
314 * XXX: These are not STATES and really should be somewhere else.
315 * XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL,
316 *      but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2),
317 *      and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2),
318 *      it's harmless.
319 */
320#define	MNT_UPDATE	0x0000000000010000ULL /* not real mount, just update */
321#define	MNT_DELEXPORT	0x0000000000020000ULL /* delete export host lists */
322#define	MNT_RELOAD	0x0000000000040000ULL /* reload filesystem data */
323#define	MNT_FORCE	0x0000000000080000ULL /* force unmount or readonly */
324#define	MNT_SNAPSHOT	0x0000000001000000ULL /* snapshot the filesystem */
325#define	MNT_NONBUSY	0x0000000004000000ULL /* check vnode use counts. */
326#define	MNT_BYFSID	0x0000000008000000ULL /* specify filesystem by ID. */
327#define MNT_CMDFLAGS   (MNT_UPDATE	| MNT_DELEXPORT	| MNT_RELOAD	| \
328			MNT_FORCE	| MNT_SNAPSHOT	| MNT_NONBUSY	| \
329			MNT_BYFSID)
330/*
331 * Internal filesystem control flags stored in mnt_kern_flag.
332 *
333 * MNTK_UNMOUNT locks the mount entry so that name lookup cannot proceed
334 * past the mount point.  This keeps the subtree stable during mounts
335 * and unmounts.
336 *
337 * MNTK_UNMOUNTF permits filesystems to detect a forced unmount while
338 * dounmount() is still waiting to lock the mountpoint. This allows
339 * the filesystem to cancel operations that might otherwise deadlock
340 * with the unmount attempt (used by NFS).
341 *
342 * MNTK_NOINSMNTQ is strict subset of MNTK_UNMOUNT. They are separated
343 * to allow for failed unmount attempt to restore the syncer vnode for
344 * the mount.
345 */
346#define MNTK_UNMOUNTF	0x00000001	/* forced unmount in progress */
347#define MNTK_ASYNC	0x00000002	/* filtered async flag */
348#define MNTK_SOFTDEP	0x00000004	/* async disabled by softdep */
349#define MNTK_NOINSMNTQ	0x00000008	/* insmntque is not allowed */
350#define	MNTK_DRAINING	0x00000010	/* lock draining is happening */
351#define	MNTK_REFEXPIRE	0x00000020	/* refcount expiring is happening */
352#define MNTK_EXTENDED_SHARED	0x00000040 /* Allow shared locking for more ops */
353#define	MNTK_SHARED_WRITES	0x00000080 /* Allow shared locking for writes */
354#define	MNTK_NO_IOPF	0x00000100	/* Disallow page faults during reads
355					   and writes. Filesystem shall properly
356					   handle i/o state on EFAULT. */
357#define	MNTK_VGONE_UPPER	0x00000200
358#define	MNTK_VGONE_WAITER	0x00000400
359#define	MNTK_LOOKUP_EXCL_DOTDOT	0x00000800
360#define	MNTK_MARKER		0x00001000
361#define	MNTK_UNMAPPED_BUFS	0x00002000
362#define	MNTK_USES_BCACHE	0x00004000 /* FS uses the buffer cache. */
363#define MNTK_NOASYNC	0x00800000	/* disable async */
364#define MNTK_UNMOUNT	0x01000000	/* unmount in progress */
365#define	MNTK_MWAIT	0x02000000	/* waiting for unmount to finish */
366#define	MNTK_SUSPEND	0x08000000	/* request write suspension */
367#define	MNTK_SUSPEND2	0x04000000	/* block secondary writes */
368#define	MNTK_SUSPENDED	0x10000000	/* write operations are suspended */
369#define	MNTK_UNUSED1	0x20000000
370#define MNTK_LOOKUP_SHARED	0x40000000 /* FS supports shared lock lookups */
371#define	MNTK_NOKNOTE	0x80000000	/* Don't send KNOTEs from VOP hooks */
372
373#ifdef _KERNEL
374static inline int
375MNT_SHARED_WRITES(struct mount *mp)
376{
377
378	return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0);
379}
380
381static inline int
382MNT_EXTENDED_SHARED(struct mount *mp)
383{
384
385	return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0);
386}
387#endif
388
389/*
390 * Sysctl CTL_VFS definitions.
391 *
392 * Second level identifier specifies which filesystem. Second level
393 * identifier VFS_VFSCONF returns information about all filesystems.
394 * Second level identifier VFS_GENERIC is non-terminal.
395 */
396#define	VFS_VFSCONF		0	/* get configured filesystems */
397#define	VFS_GENERIC		0	/* generic filesystem information */
398/*
399 * Third level identifiers for VFS_GENERIC are given below; third
400 * level identifiers for specific filesystems are given in their
401 * mount specific header files.
402 */
403#define VFS_MAXTYPENUM	1	/* int: highest defined filesystem type */
404#define VFS_CONF	2	/* struct: vfsconf for filesystem given
405				   as next argument */
406
407/*
408 * Flags for various system call interfaces.
409 *
410 * waitfor flags to vfs_sync() and getfsstat()
411 */
412#define MNT_WAIT	1	/* synchronously wait for I/O to complete */
413#define MNT_NOWAIT	2	/* start all I/O, but do not wait for it */
414#define MNT_LAZY	3	/* push data not written by filesystem syncer */
415#define MNT_SUSPEND	4	/* Suspend file system after sync */
416
417/*
418 * Generic file handle
419 */
420struct fhandle {
421	fsid_t	fh_fsid;	/* Filesystem id of mount point */
422	struct	fid fh_fid;	/* Filesys specific id */
423};
424typedef struct fhandle	fhandle_t;
425
426/*
427 * Old export arguments without security flavor list
428 */
429struct oexport_args {
430	int	ex_flags;		/* export related flags */
431	uid_t	ex_root;		/* mapping for root uid */
432	struct	xucred ex_anon;		/* mapping for anonymous user */
433	struct	sockaddr *ex_addr;	/* net address to which exported */
434	u_char	ex_addrlen;		/* and the net address length */
435	struct	sockaddr *ex_mask;	/* mask of valid bits in saddr */
436	u_char	ex_masklen;		/* and the smask length */
437	char	*ex_indexfile;		/* index file for WebNFS URLs */
438};
439
440/*
441 * Export arguments for local filesystem mount calls.
442 */
443#define	MAXSECFLAVORS	5
444struct export_args {
445	int	ex_flags;		/* export related flags */
446	uid_t	ex_root;		/* mapping for root uid */
447	struct	xucred ex_anon;		/* mapping for anonymous user */
448	struct	sockaddr *ex_addr;	/* net address to which exported */
449	u_char	ex_addrlen;		/* and the net address length */
450	struct	sockaddr *ex_mask;	/* mask of valid bits in saddr */
451	u_char	ex_masklen;		/* and the smask length */
452	char	*ex_indexfile;		/* index file for WebNFS URLs */
453	int	ex_numsecflavors;	/* security flavor count */
454	int	ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */
455};
456
457/*
458 * Structure holding information for a publicly exported filesystem
459 * (WebNFS). Currently the specs allow just for one such filesystem.
460 */
461struct nfs_public {
462	int		np_valid;	/* Do we hold valid information */
463	fhandle_t	np_handle;	/* Filehandle for pub fs (internal) */
464	struct mount	*np_mount;	/* Mountpoint of exported fs */
465	char		*np_index;	/* Index file */
466};
467
468/*
469 * Filesystem configuration information. One of these exists for each
470 * type of filesystem supported by the kernel. These are searched at
471 * mount time to identify the requested filesystem.
472 *
473 * XXX: Never change the first two arguments!
474 */
475struct vfsconf {
476	u_int	vfc_version;		/* ABI version number */
477	char	vfc_name[MFSNAMELEN];	/* filesystem type name */
478	struct	vfsops *vfc_vfsops;	/* filesystem operations vector */
479	int	vfc_typenum;		/* historic filesystem type number */
480	int	vfc_refcount;		/* number mounted of this type */
481	int	vfc_flags;		/* permanent flags */
482	struct	vfsoptdecl *vfc_opts;	/* mount options */
483	TAILQ_ENTRY(vfsconf) vfc_list;	/* list of vfscons */
484};
485
486/* Userland version of the struct vfsconf. */
487struct xvfsconf {
488	struct	vfsops *vfc_vfsops;	/* filesystem operations vector */
489	char	vfc_name[MFSNAMELEN];	/* filesystem type name */
490	int	vfc_typenum;		/* historic filesystem type number */
491	int	vfc_refcount;		/* number mounted of this type */
492	int	vfc_flags;		/* permanent flags */
493	struct	vfsconf *vfc_next;	/* next in list */
494};
495
496#ifndef BURN_BRIDGES
497struct ovfsconf {
498	void	*vfc_vfsops;
499	char	vfc_name[32];
500	int	vfc_index;
501	int	vfc_refcount;
502	int	vfc_flags;
503};
504#endif
505
506/*
507 * NB: these flags refer to IMPLEMENTATION properties, not properties of
508 * any actual mounts; i.e., it does not make sense to change the flags.
509 */
510#define	VFCF_STATIC	0x00010000	/* statically compiled into kernel */
511#define	VFCF_NETWORK	0x00020000	/* may get data over the network */
512#define	VFCF_READONLY	0x00040000	/* writes are not implemented */
513#define	VFCF_SYNTHETIC	0x00080000	/* data does not represent real files */
514#define	VFCF_LOOPBACK	0x00100000	/* aliases some other mounted FS */
515#define	VFCF_UNICODE	0x00200000	/* stores file names as Unicode */
516#define	VFCF_JAIL	0x00400000	/* can be mounted from within a jail */
517#define	VFCF_DELEGADMIN	0x00800000	/* supports delegated administration */
518#define	VFCF_SBDRY	0x01000000	/* defer stop requests */
519
520typedef uint32_t fsctlop_t;
521
522struct vfsidctl {
523	int		vc_vers;	/* should be VFSIDCTL_VERS1 (below) */
524	fsid_t		vc_fsid;	/* fsid to operate on */
525	char		vc_fstypename[MFSNAMELEN];
526					/* type of fs 'nfs' or '*' */
527	fsctlop_t	vc_op;		/* operation VFS_CTL_* (below) */
528	void		*vc_ptr;	/* pointer to data structure */
529	size_t		vc_len;		/* sizeof said structure */
530	u_int32_t	vc_spare[12];	/* spare (must be zero) */
531};
532
533/* vfsidctl API version. */
534#define VFS_CTL_VERS1	0x01
535
536/*
537 * New style VFS sysctls, do not reuse/conflict with the namespace for
538 * private sysctls.
539 * All "global" sysctl ops have the 33rd bit set:
540 * 0x...1....
541 * Private sysctl ops should have the 33rd bit unset.
542 */
543#define VFS_CTL_QUERY	0x00010001	/* anything wrong? (vfsquery) */
544#define VFS_CTL_TIMEO	0x00010002	/* set timeout for vfs notification */
545#define VFS_CTL_NOLOCKS	0x00010003	/* disable file locking */
546
547struct vfsquery {
548	u_int32_t	vq_flags;
549	u_int32_t	vq_spare[31];
550};
551
552/* vfsquery flags */
553#define VQ_NOTRESP	0x0001	/* server down */
554#define VQ_NEEDAUTH	0x0002	/* server bad auth */
555#define VQ_LOWDISK	0x0004	/* we're low on space */
556#define VQ_MOUNT	0x0008	/* new filesystem arrived */
557#define VQ_UNMOUNT	0x0010	/* filesystem has left */
558#define VQ_DEAD		0x0020	/* filesystem is dead, needs force unmount */
559#define VQ_ASSIST	0x0040	/* filesystem needs assistance from external
560				   program */
561#define VQ_NOTRESPLOCK	0x0080	/* server lockd down */
562#define VQ_FLAG0100	0x0100	/* placeholder */
563#define VQ_FLAG0200	0x0200	/* placeholder */
564#define VQ_FLAG0400	0x0400	/* placeholder */
565#define VQ_FLAG0800	0x0800	/* placeholder */
566#define VQ_FLAG1000	0x1000	/* placeholder */
567#define VQ_FLAG2000	0x2000	/* placeholder */
568#define VQ_FLAG4000	0x4000	/* placeholder */
569#define VQ_FLAG8000	0x8000	/* placeholder */
570
571#ifdef _KERNEL
572/* Point a sysctl request at a vfsidctl's data. */
573#define VCTLTOREQ(vc, req)						\
574	do {								\
575		(req)->newptr = (vc)->vc_ptr;				\
576		(req)->newlen = (vc)->vc_len;				\
577		(req)->newidx = 0;					\
578	} while (0)
579#endif
580
581struct iovec;
582struct uio;
583
584#ifdef _KERNEL
585
586/*
587 * vfs_busy specific flags and mask.
588 */
589#define	MBF_NOWAIT	0x01
590#define	MBF_MNTLSTLOCK	0x02
591#define	MBF_MASK	(MBF_NOWAIT | MBF_MNTLSTLOCK)
592
593#ifdef MALLOC_DECLARE
594MALLOC_DECLARE(M_MOUNT);
595#endif
596extern int maxvfsconf;		/* highest defined filesystem type */
597
598TAILQ_HEAD(vfsconfhead, vfsconf);
599extern struct vfsconfhead vfsconf;
600
601/*
602 * Operations supported on mounted filesystem.
603 */
604struct mount_args;
605struct nameidata;
606struct sysctl_req;
607struct mntarg;
608
609typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags);
610typedef int vfs_unmount_t(struct mount *mp, int mntflags);
611typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp);
612typedef	int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg);
613typedef	int vfs_statfs_t(struct mount *mp, struct statfs *sbp);
614typedef	int vfs_sync_t(struct mount *mp, int waitfor);
615typedef	int vfs_vget_t(struct mount *mp, ino_t ino, int flags,
616		    struct vnode **vpp);
617typedef	int vfs_fhtovp_t(struct mount *mp, struct fid *fhp,
618		    int flags, struct vnode **vpp);
619typedef	int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam,
620		    int *extflagsp, struct ucred **credanonp,
621		    int *numsecflavors, int **secflavors);
622typedef	int vfs_init_t(struct vfsconf *);
623typedef	int vfs_uninit_t(struct vfsconf *);
624typedef	int vfs_extattrctl_t(struct mount *mp, int cmd,
625		    struct vnode *filename_vp, int attrnamespace,
626		    const char *attrname);
627typedef	int vfs_mount_t(struct mount *mp);
628typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op,
629		    struct sysctl_req *req);
630typedef void vfs_susp_clean_t(struct mount *mp);
631typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp);
632typedef void vfs_purge_t(struct mount *mp);
633
634struct vfsops {
635	vfs_mount_t		*vfs_mount;
636	vfs_cmount_t		*vfs_cmount;
637	vfs_unmount_t		*vfs_unmount;
638	vfs_root_t		*vfs_root;
639	vfs_quotactl_t		*vfs_quotactl;
640	vfs_statfs_t		*vfs_statfs;
641	vfs_sync_t		*vfs_sync;
642	vfs_vget_t		*vfs_vget;
643	vfs_fhtovp_t		*vfs_fhtovp;
644	vfs_checkexp_t		*vfs_checkexp;
645	vfs_init_t		*vfs_init;
646	vfs_uninit_t		*vfs_uninit;
647	vfs_extattrctl_t	*vfs_extattrctl;
648	vfs_sysctl_t		*vfs_sysctl;
649	vfs_susp_clean_t	*vfs_susp_clean;
650	vfs_notify_lowervp_t	*vfs_reclaim_lowervp;
651	vfs_notify_lowervp_t	*vfs_unlink_lowervp;
652	vfs_purge_t		*vfs_purge;
653	vfs_mount_t		*vfs_spare[6];	/* spares for ABI compat */
654};
655
656vfs_statfs_t	__vfs_statfs;
657
658#define	VFS_PROLOGUE(MP)	do {					\
659	struct mount *mp__;						\
660	int _prev_stops;						\
661									\
662	mp__ = (MP);							\
663	_prev_stops = sigdeferstop((mp__ != NULL &&			\
664	    (mp__->mnt_vfc->vfc_flags & VFCF_SBDRY) != 0) ?		\
665	    SIGDEFERSTOP_SILENT : SIGDEFERSTOP_NOP);
666
667#define	VFS_EPILOGUE(MP)						\
668	sigallowstop(_prev_stops);					\
669} while (0)
670
671#define	VFS_MOUNT(MP) ({						\
672	int _rc;							\
673									\
674	VFS_PROLOGUE(MP);						\
675	_rc = (*(MP)->mnt_op->vfs_mount)(MP);				\
676	VFS_EPILOGUE(MP);						\
677	_rc; })
678
679#define	VFS_UNMOUNT(MP, FORCE) ({					\
680	int _rc;							\
681									\
682	VFS_PROLOGUE(MP);						\
683	_rc = (*(MP)->mnt_op->vfs_unmount)(MP, FORCE);			\
684	VFS_EPILOGUE(MP);						\
685	_rc; })
686
687#define	VFS_ROOT(MP, FLAGS, VPP) ({					\
688	int _rc;							\
689									\
690	VFS_PROLOGUE(MP);						\
691	_rc = (*(MP)->mnt_op->vfs_root)(MP, FLAGS, VPP);		\
692	VFS_EPILOGUE(MP);						\
693	_rc; })
694
695#define	VFS_QUOTACTL(MP, C, U, A) ({					\
696	int _rc;							\
697									\
698	VFS_PROLOGUE(MP);						\
699	_rc = (*(MP)->mnt_op->vfs_quotactl)(MP, C, U, A);		\
700	VFS_EPILOGUE(MP);						\
701	_rc; })
702
703#define	VFS_STATFS(MP, SBP) ({						\
704	int _rc;							\
705									\
706	VFS_PROLOGUE(MP);						\
707	_rc = __vfs_statfs((MP), (SBP));				\
708	VFS_EPILOGUE(MP);						\
709	_rc; })
710
711#define	VFS_SYNC(MP, WAIT) ({						\
712	int _rc;							\
713									\
714	VFS_PROLOGUE(MP);						\
715	_rc = (*(MP)->mnt_op->vfs_sync)(MP, WAIT);			\
716	VFS_EPILOGUE(MP);						\
717	_rc; })
718
719#define	VFS_VGET(MP, INO, FLAGS, VPP) ({				\
720	int _rc;							\
721									\
722	VFS_PROLOGUE(MP);						\
723	_rc = (*(MP)->mnt_op->vfs_vget)(MP, INO, FLAGS, VPP);		\
724	VFS_EPILOGUE(MP);						\
725	_rc; })
726
727#define	VFS_FHTOVP(MP, FIDP, FLAGS, VPP) ({				\
728	int _rc;							\
729									\
730	VFS_PROLOGUE(MP);						\
731	_rc = (*(MP)->mnt_op->vfs_fhtovp)(MP, FIDP, FLAGS, VPP);	\
732	VFS_EPILOGUE(MP);						\
733	_rc; })
734
735#define	VFS_CHECKEXP(MP, NAM, EXFLG, CRED, NUMSEC, SEC) ({		\
736	int _rc;							\
737									\
738	VFS_PROLOGUE(MP);						\
739	_rc = (*(MP)->mnt_op->vfs_checkexp)(MP, NAM, EXFLG, CRED, NUMSEC,\
740	    SEC);							\
741	VFS_EPILOGUE(MP);						\
742	_rc; })
743
744#define	VFS_EXTATTRCTL(MP, C, FN, NS, N) ({				\
745	int _rc;							\
746									\
747	VFS_PROLOGUE(MP);						\
748	_rc = (*(MP)->mnt_op->vfs_extattrctl)(MP, C, FN, NS, N);	\
749	VFS_EPILOGUE(MP);						\
750	_rc; })
751
752#define	VFS_SYSCTL(MP, OP, REQ) ({					\
753	int _rc;							\
754									\
755	VFS_PROLOGUE(MP);						\
756	_rc = (*(MP)->mnt_op->vfs_sysctl)(MP, OP, REQ);			\
757	VFS_EPILOGUE(MP);						\
758	_rc; })
759
760#define	VFS_SUSP_CLEAN(MP) do {						\
761	if (*(MP)->mnt_op->vfs_susp_clean != NULL) {			\
762		VFS_PROLOGUE(MP);					\
763		(*(MP)->mnt_op->vfs_susp_clean)(MP);			\
764		VFS_EPILOGUE(MP);					\
765	}								\
766} while (0)
767
768#define	VFS_RECLAIM_LOWERVP(MP, VP) do {				\
769	if (*(MP)->mnt_op->vfs_reclaim_lowervp != NULL) {		\
770		VFS_PROLOGUE(MP);					\
771		(*(MP)->mnt_op->vfs_reclaim_lowervp)((MP), (VP));	\
772		VFS_EPILOGUE(MP);					\
773	}								\
774} while (0)
775
776#define	VFS_UNLINK_LOWERVP(MP, VP) do {					\
777	if (*(MP)->mnt_op->vfs_unlink_lowervp != NULL) {		\
778		VFS_PROLOGUE(MP);					\
779		(*(MP)->mnt_op->vfs_unlink_lowervp)((MP), (VP));	\
780		VFS_EPILOGUE(MP);					\
781	}								\
782} while (0)
783
784#define	VFS_PURGE(MP) do {						\
785	if (*(MP)->mnt_op->vfs_purge != NULL) {				\
786		VFS_PROLOGUE(MP);					\
787		(*(MP)->mnt_op->vfs_purge)(MP);				\
788		VFS_EPILOGUE(MP);					\
789	}								\
790} while (0)
791
792#define VFS_KNOTE_LOCKED(vp, hint) do					\
793{									\
794	if (((vp)->v_vflag & VV_NOKNOTE) == 0)				\
795		VN_KNOTE((vp), (hint), KNF_LISTLOCKED);			\
796} while (0)
797
798#define VFS_KNOTE_UNLOCKED(vp, hint) do					\
799{									\
800	if (((vp)->v_vflag & VV_NOKNOTE) == 0)				\
801		VN_KNOTE((vp), (hint), 0);				\
802} while (0)
803
804#define	VFS_NOTIFY_UPPER_RECLAIM	1
805#define	VFS_NOTIFY_UPPER_UNLINK		2
806
807#include <sys/module.h>
808
809/*
810 * Version numbers.
811 */
812#define VFS_VERSION_00	0x19660120
813#define VFS_VERSION_01	0x20121030
814#define VFS_VERSION	VFS_VERSION_01
815
816#define VFS_SET(vfsops, fsname, flags) \
817	static struct vfsconf fsname ## _vfsconf = {		\
818		.vfc_version = VFS_VERSION,			\
819		.vfc_name = #fsname,				\
820		.vfc_vfsops = &vfsops,				\
821		.vfc_typenum = -1,				\
822		.vfc_flags = flags,				\
823	};							\
824	static moduledata_t fsname ## _mod = {			\
825		#fsname,					\
826		vfs_modevent,					\
827		& fsname ## _vfsconf				\
828	};							\
829	DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE)
830
831/*
832 * exported vnode operations
833 */
834
835int	dounmount(struct mount *, int, struct thread *);
836
837int	kernel_mount(struct mntarg *ma, uint64_t flags);
838int	kernel_vmount(int flags, ...);
839struct mntarg *mount_arg(struct mntarg *ma, const char *name, const void *val, int len);
840struct mntarg *mount_argb(struct mntarg *ma, int flag, const char *name);
841struct mntarg *mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...);
842struct mntarg *mount_argsu(struct mntarg *ma, const char *name, const void *val, int len);
843void	statfs_scale_blocks(struct statfs *sf, long max_size);
844struct vfsconf *vfs_byname(const char *);
845struct vfsconf *vfs_byname_kld(const char *, struct thread *td, int *);
846void	vfs_mount_destroy(struct mount *);
847void	vfs_event_signal(fsid_t *, u_int32_t, intptr_t);
848void	vfs_freeopts(struct vfsoptlist *opts);
849void	vfs_deleteopt(struct vfsoptlist *opts, const char *name);
850int	vfs_buildopts(struct uio *auio, struct vfsoptlist **options);
851int	vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w,
852	    uint64_t val);
853int	vfs_getopt(struct vfsoptlist *, const char *, void **, int *);
854int	vfs_getopt_pos(struct vfsoptlist *opts, const char *name);
855int	vfs_getopt_size(struct vfsoptlist *opts, const char *name,
856	    off_t *value);
857char	*vfs_getopts(struct vfsoptlist *, const char *, int *error);
858int	vfs_copyopt(struct vfsoptlist *, const char *, void *, int);
859int	vfs_filteropt(struct vfsoptlist *, const char **legal);
860void	vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...);
861int	vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...);
862int	vfs_setopt(struct vfsoptlist *opts, const char *name, void *value,
863	    int len);
864int	vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value,
865	    int len);
866int	vfs_setopts(struct vfsoptlist *opts, const char *name,
867	    const char *value);
868int	vfs_setpublicfs			    /* set publicly exported fs */
869	    (struct mount *, struct netexport *, struct export_args *);
870void	vfs_msync(struct mount *, int);
871int	vfs_busy(struct mount *, int);
872int	vfs_export			 /* process mount export info */
873	    (struct mount *, struct export_args *);
874void	vfs_allocate_syncvnode(struct mount *);
875void	vfs_deallocate_syncvnode(struct mount *);
876int	vfs_donmount(struct thread *td, uint64_t fsflags,
877	    struct uio *fsoptions);
878void	vfs_getnewfsid(struct mount *);
879struct cdev *vfs_getrootfsid(struct mount *);
880struct	mount *vfs_getvfs(fsid_t *);      /* return vfs given fsid */
881struct	mount *vfs_busyfs(fsid_t *);
882int	vfs_modevent(module_t, int, void *);
883void	vfs_mount_error(struct mount *, const char *, ...);
884void	vfs_mountroot(void);			/* mount our root filesystem */
885void	vfs_mountedfrom(struct mount *, const char *from);
886void	vfs_notify_upper(struct vnode *, int);
887void	vfs_oexport_conv(const struct oexport_args *oexp,
888	    struct export_args *exp);
889void	vfs_ref(struct mount *);
890void	vfs_rel(struct mount *);
891struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *,
892	    struct ucred *);
893int	vfs_suser(struct mount *, struct thread *);
894void	vfs_unbusy(struct mount *);
895void	vfs_unmountall(void);
896extern	TAILQ_HEAD(mntlist, mount) mountlist;	/* mounted filesystem list */
897extern	struct mtx mountlist_mtx;
898extern	struct nfs_public nfs_pub;
899extern	struct sx vfsconf_sx;
900#define	vfsconf_lock()		sx_xlock(&vfsconf_sx)
901#define	vfsconf_unlock()	sx_xunlock(&vfsconf_sx)
902#define	vfsconf_slock()		sx_slock(&vfsconf_sx)
903#define	vfsconf_sunlock()	sx_sunlock(&vfsconf_sx)
904
905/*
906 * Declarations for these vfs default operations are located in
907 * kern/vfs_default.c.  They will be automatically used to replace
908 * null entries in VFS ops tables when registering a new filesystem
909 * type in the global table.
910 */
911vfs_root_t		vfs_stdroot;
912vfs_quotactl_t		vfs_stdquotactl;
913vfs_statfs_t		vfs_stdstatfs;
914vfs_sync_t		vfs_stdsync;
915vfs_sync_t		vfs_stdnosync;
916vfs_vget_t		vfs_stdvget;
917vfs_fhtovp_t		vfs_stdfhtovp;
918vfs_checkexp_t		vfs_stdcheckexp;
919vfs_init_t		vfs_stdinit;
920vfs_uninit_t		vfs_stduninit;
921vfs_extattrctl_t	vfs_stdextattrctl;
922vfs_sysctl_t		vfs_stdsysctl;
923
924void	syncer_suspend(void);
925void	syncer_resume(void);
926
927#else /* !_KERNEL */
928
929#include <sys/cdefs.h>
930
931struct stat;
932
933__BEGIN_DECLS
934int	fhopen(const struct fhandle *, int);
935int	fhstat(const struct fhandle *, struct stat *);
936int	fhstatfs(const struct fhandle *, struct statfs *);
937int	fstatfs(int, struct statfs *);
938int	getfh(const char *, fhandle_t *);
939int	getfsstat(struct statfs *, long, int);
940int	getmntinfo(struct statfs **, int);
941int	lgetfh(const char *, fhandle_t *);
942int	mount(const char *, const char *, int, void *);
943int	nmount(struct iovec *, unsigned int, int);
944int	statfs(const char *, struct statfs *);
945int	unmount(const char *, int);
946
947/* C library stuff */
948int	getvfsbyname(const char *, struct xvfsconf *);
949__END_DECLS
950
951#endif /* _KERNEL */
952
953#endif /* !_SYS_MOUNT_H_ */
954