1/*-
2 * Copyright (c) 1989, 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 *	@(#)vnode.h	8.7 (Berkeley) 2/4/94
30 * $FreeBSD$
31 */
32
33#ifndef _SYS_VNODE_H_
34#define	_SYS_VNODE_H_
35
36#include <sys/bufobj.h>
37#include <sys/queue.h>
38#include <sys/lock.h>
39#include <sys/lockmgr.h>
40#include <sys/mutex.h>
41#include <sys/rangelock.h>
42#include <sys/selinfo.h>
43#include <sys/uio.h>
44#include <sys/acl.h>
45#include <sys/ktr.h>
46
47/*
48 * The vnode is the focus of all file activity in UNIX.  There is a
49 * unique vnode allocated for each active file, each current directory,
50 * each mounted-on file, text file, and the root.
51 */
52
53/*
54 * Vnode types.  VNON means no type.
55 */
56enum vtype	{ VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD,
57		  VMARKER };
58
59/*
60 * Each underlying filesystem allocates its own private area and hangs
61 * it from v_data.  If non-null, this area is freed in getnewvnode().
62 */
63
64struct namecache;
65
66struct vpollinfo {
67	struct	mtx vpi_lock;		/* lock to protect below */
68	struct	selinfo vpi_selinfo;	/* identity of poller(s) */
69	short	vpi_events;		/* what they are looking for */
70	short	vpi_revents;		/* what has happened */
71};
72
73/*
74 * Reading or writing any of these items requires holding the appropriate lock.
75 *
76 * Lock reference:
77 *	c - namecache mutex
78 *	f - freelist mutex
79 *	i - interlock
80 *	m - mount point interlock
81 *	p - pollinfo lock
82 *	u - Only a reference to the vnode is needed to read.
83 *	v - vnode lock
84 *
85 * Vnodes may be found on many lists.  The general way to deal with operating
86 * on a vnode that is on a list is:
87 *	1) Lock the list and find the vnode.
88 *	2) Lock interlock so that the vnode does not go away.
89 *	3) Unlock the list to avoid lock order reversals.
90 *	4) vget with LK_INTERLOCK and check for ENOENT, or
91 *	5) Check for DOOMED if the vnode lock is not required.
92 *	6) Perform your operation, then vput().
93 */
94
95#if defined(_KERNEL) || defined(_KVM_VNODE)
96
97struct vnode {
98	/*
99	 * Fields which define the identity of the vnode.  These fields are
100	 * owned by the filesystem (XXX: and vgone() ?)
101	 */
102	const char *v_tag;			/* u type of underlying data */
103	struct	vop_vector *v_op;		/* u vnode operations vector */
104	void	*v_data;			/* u private data for fs */
105
106	/*
107	 * Filesystem instance stuff
108	 */
109	struct	mount *v_mount;			/* u ptr to vfs we are in */
110	TAILQ_ENTRY(vnode) v_nmntvnodes;	/* m vnodes for mount point */
111
112	/*
113	 * Type specific fields, only one applies to any given vnode.
114	 * See #defines below for renaming to v_* namespace.
115	 */
116	union {
117		struct mount	*vu_mount;	/* v ptr to mountpoint (VDIR) */
118		struct socket	*vu_socket;	/* v unix domain net (VSOCK) */
119		struct cdev	*vu_cdev; 	/* v device (VCHR, VBLK) */
120		struct fifoinfo	*vu_fifoinfo;	/* v fifo (VFIFO) */
121	} v_un;
122
123	/*
124	 * vfs_hash: (mount + inode) -> vnode hash.  The hash value
125	 * itself is grouped with other int fields, to avoid padding.
126	 */
127	LIST_ENTRY(vnode)	v_hashlist;
128
129	/*
130	 * VFS_namecache stuff
131	 */
132	LIST_HEAD(, namecache) v_cache_src;	/* c Cache entries from us */
133	TAILQ_HEAD(, namecache) v_cache_dst;	/* c Cache entries to us */
134	struct namecache *v_cache_dd;		/* c Cache entry for .. vnode */
135
136	/*
137	 * Locking
138	 */
139	struct	lock v_lock;			/* u (if fs don't have one) */
140	struct	mtx v_interlock;		/* lock for "i" things */
141	struct	lock *v_vnlock;			/* u pointer to vnode lock */
142
143	/*
144	 * The machinery of being a vnode
145	 */
146	TAILQ_ENTRY(vnode) v_actfreelist;	/* f vnode active/free lists */
147	struct bufobj	v_bufobj;		/* * Buffer cache object */
148
149	/*
150	 * Hooks for various subsystems and features.
151	 */
152	struct vpollinfo *v_pollinfo;		/* i Poll events, p for *v_pi */
153	struct label *v_label;			/* MAC label for vnode */
154	struct lockf *v_lockf;		/* Byte-level advisory lock list */
155	struct rangelock v_rl;			/* Byte-range lock */
156
157	/*
158	 * clustering stuff
159	 */
160	daddr_t	v_cstart;			/* v start block of cluster */
161	daddr_t	v_lasta;			/* v last allocation  */
162	daddr_t	v_lastw;			/* v last write  */
163	int	v_clen;				/* v length of cur. cluster */
164
165	int	v_holdcnt;			/* i prevents recycling. */
166	int	v_usecount;			/* i ref count of users */
167	u_int	v_iflag;			/* i vnode flags (see below) */
168	u_int	v_vflag;			/* v vnode flags */
169	int	v_writecount;			/* v ref count of writers */
170	u_int	v_hash;
171	enum	vtype v_type;			/* u vnode type */
172};
173
174#endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
175
176#define	v_mountedhere	v_un.vu_mount
177#define	v_socket	v_un.vu_socket
178#define	v_rdev		v_un.vu_cdev
179#define	v_fifoinfo	v_un.vu_fifoinfo
180
181/* XXX: These are temporary to avoid a source sweep at this time */
182#define v_object	v_bufobj.bo_object
183
184/*
185 * Userland version of struct vnode, for sysctl.
186 */
187struct xvnode {
188	size_t	xv_size;			/* sizeof(struct xvnode) */
189	void	*xv_vnode;			/* address of real vnode */
190	u_long	xv_flag;			/* vnode vflags */
191	int	xv_usecount;			/* reference count of users */
192	int	xv_writecount;			/* reference count of writers */
193	int	xv_holdcnt;			/* page & buffer references */
194	u_long	xv_id;				/* capability identifier */
195	void	*xv_mount;			/* address of parent mount */
196	long	xv_numoutput;			/* num of writes in progress */
197	enum	vtype xv_type;			/* vnode type */
198	union {
199		void	*xvu_socket;		/* socket, if VSOCK */
200		void	*xvu_fifo;		/* fifo, if VFIFO */
201		dev_t	xvu_rdev;		/* maj/min, if VBLK/VCHR */
202		struct {
203			dev_t	xvu_dev;	/* device, if VDIR/VREG/VLNK */
204			ino_t	xvu_ino;	/* id, if VDIR/VREG/VLNK */
205		} xv_uns;
206	} xv_un;
207};
208#define xv_socket	xv_un.xvu_socket
209#define xv_fifo		xv_un.xvu_fifo
210#define xv_rdev		xv_un.xvu_rdev
211#define xv_dev		xv_un.xv_uns.xvu_dev
212#define xv_ino		xv_un.xv_uns.xvu_ino
213
214/* We don't need to lock the knlist */
215#define	VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL ||	\
216	    KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
217
218#define VN_KNOTE(vp, b, a)					\
219	do {							\
220		if (!VN_KNLIST_EMPTY(vp))			\
221			KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
222			    (a) | KNF_NOKQLOCK);		\
223	} while (0)
224#define	VN_KNOTE_LOCKED(vp, b)		VN_KNOTE(vp, b, KNF_LISTLOCKED)
225#define	VN_KNOTE_UNLOCKED(vp, b)	VN_KNOTE(vp, b, 0)
226
227/*
228 * Vnode flags.
229 *	VI flags are protected by interlock and live in v_iflag
230 *	VV flags are protected by the vnode lock and live in v_vflag
231 *
232 *	VI_DOOMED is doubly protected by the interlock and vnode lock.  Both
233 *	are required for writing but the status may be checked with either.
234 */
235#define	VI_MOUNT	0x0020	/* Mount in progress */
236#define	VI_AGE		0x0040	/* Insert vnode at head of free list */
237#define	VI_DOOMED	0x0080	/* This vnode is being recycled */
238#define	VI_FREE		0x0100	/* This vnode is on the freelist */
239#define	VI_ACTIVE	0x0200	/* This vnode is on the active list */
240#define	VI_DOINGINACT	0x0800	/* VOP_INACTIVE is in progress */
241#define	VI_OWEINACT	0x1000	/* Need to call inactive */
242
243#define	VV_ROOT		0x0001	/* root of its filesystem */
244#define	VV_ISTTY	0x0002	/* vnode represents a tty */
245#define	VV_NOSYNC	0x0004	/* unlinked, stop syncing */
246#define	VV_ETERNALDEV	0x0008	/* device that is never destroyed */
247#define	VV_CACHEDLABEL	0x0010	/* Vnode has valid cached MAC label */
248#define	VV_TEXT		0x0020	/* vnode is a pure text prototype */
249#define	VV_COPYONWRITE	0x0040	/* vnode is doing copy-on-write */
250#define	VV_SYSTEM	0x0080	/* vnode being used by kernel */
251#define	VV_PROCDEP	0x0100	/* vnode is process dependent */
252#define	VV_NOKNOTE	0x0200	/* don't activate knotes on this vnode */
253#define	VV_DELETED	0x0400	/* should be removed */
254#define	VV_MD		0x0800	/* vnode backs the md device */
255#define	VV_FORCEINSMQ	0x1000	/* force the insmntque to succeed */
256
257/*
258 * Vnode attributes.  A field value of VNOVAL represents a field whose value
259 * is unavailable (getattr) or which is not to be changed (setattr).
260 */
261struct vattr {
262	enum vtype	va_type;	/* vnode type (for create) */
263	u_short		va_mode;	/* files access mode and type */
264	short		va_nlink;	/* number of references to file */
265	uid_t		va_uid;		/* owner user id */
266	gid_t		va_gid;		/* owner group id */
267	dev_t		va_fsid;	/* filesystem id */
268	long		va_fileid;	/* file id */
269	u_quad_t	va_size;	/* file size in bytes */
270	long		va_blocksize;	/* blocksize preferred for i/o */
271	struct timespec	va_atime;	/* time of last access */
272	struct timespec	va_mtime;	/* time of last modification */
273	struct timespec	va_ctime;	/* time file changed */
274	struct timespec	va_birthtime;	/* time file created */
275	u_long		va_gen;		/* generation number of file */
276	u_long		va_flags;	/* flags defined for file */
277	dev_t		va_rdev;	/* device the special file represents */
278	u_quad_t	va_bytes;	/* bytes of disk space held by file */
279	u_quad_t	va_filerev;	/* file modification number */
280	u_int		va_vaflags;	/* operations flags, see below */
281	long		va_spare;	/* remain quad aligned */
282};
283
284/*
285 * Flags for va_vaflags.
286 */
287#define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
288#define	VA_EXCLUSIVE	0x02		/* exclusive create request */
289
290/*
291 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
292 * help with write clustering)
293 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
294 */
295#define	IO_UNIT		0x0001		/* do I/O as atomic unit */
296#define	IO_APPEND	0x0002		/* append write to end */
297#define	IO_NDELAY	0x0004		/* FNDELAY flag set in file table */
298#define	IO_NODELOCKED	0x0008		/* underlying node already locked */
299#define	IO_ASYNC	0x0010		/* bawrite rather then bdwrite */
300#define	IO_VMIO		0x0020		/* data already in VMIO space */
301#define	IO_INVAL	0x0040		/* invalidate after I/O */
302#define	IO_SYNC		0x0080		/* do I/O synchronously */
303#define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
304#define	IO_EXT		0x0400		/* operate on external attributes */
305#define	IO_NORMAL	0x0800		/* operate on regular data */
306#define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
307#define	IO_BUFLOCKED	0x2000		/* ffs flag; indir buf is locked */
308
309#define IO_SEQMAX	0x7F		/* seq heuristic max value */
310#define IO_SEQSHIFT	16		/* seq heuristic in upper 16 bits */
311
312/*
313 * Flags for accmode_t.
314 */
315#define	VEXEC			000000000100 /* execute/search permission */
316#define	VWRITE			000000000200 /* write permission */
317#define	VREAD			000000000400 /* read permission */
318#define	VADMIN			000000010000 /* being the file owner */
319#define	VAPPEND			000000040000 /* permission to write/append */
320/*
321 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
322 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
323 * and 0 otherwise.  This never happens with ordinary unix access rights
324 * or POSIX.1e ACLs.  Obviously, VEXPLICIT_DENY must be OR-ed with
325 * some other V* constant.
326 */
327#define	VEXPLICIT_DENY		000000100000
328#define	VREAD_NAMED_ATTRS 	000000200000 /* not used */
329#define	VWRITE_NAMED_ATTRS 	000000400000 /* not used */
330#define	VDELETE_CHILD	 	000001000000
331#define	VREAD_ATTRIBUTES 	000002000000 /* permission to stat(2) */
332#define	VWRITE_ATTRIBUTES 	000004000000 /* change {m,c,a}time */
333#define	VDELETE		 	000010000000
334#define	VREAD_ACL	 	000020000000 /* read ACL and file mode */
335#define	VWRITE_ACL	 	000040000000 /* change ACL and/or file mode */
336#define	VWRITE_OWNER	 	000100000000 /* change file owner */
337#define	VSYNCHRONIZE	 	000200000000 /* not used */
338
339/*
340 * Permissions that were traditionally granted only to the file owner.
341 */
342#define VADMIN_PERMS	(VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
343    VWRITE_OWNER)
344
345/*
346 * Permissions that were traditionally granted to everyone.
347 */
348#define VSTAT_PERMS	(VREAD_ATTRIBUTES | VREAD_ACL)
349
350/*
351 * Permissions that allow to change the state of the file in any way.
352 */
353#define VMODIFY_PERMS	(VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
354    VDELETE)
355
356/*
357 * Token indicating no attribute value yet assigned.
358 */
359#define	VNOVAL	(-1)
360
361/*
362 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
363 */
364#define VLKTIMEOUT	(hz / 20 + 1)
365
366#ifdef _KERNEL
367
368#ifdef MALLOC_DECLARE
369MALLOC_DECLARE(M_VNODE);
370#endif
371
372/*
373 * Convert between vnode types and inode formats (since POSIX.1
374 * defines mode word of stat structure in terms of inode formats).
375 */
376extern enum vtype	iftovt_tab[];
377extern int		vttoif_tab[];
378#define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
379#define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
380#define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
381
382/*
383 * Flags to various vnode functions.
384 */
385#define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
386#define	FORCECLOSE	0x0002	/* vflush: force file closure */
387#define	WRITECLOSE	0x0004	/* vflush: only close writable files */
388#define	EARLYFLUSH	0x0008	/* vflush: early call for ffs_flushfiles */
389#define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
390#define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
391#define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
392#define	V_CLEANONLY	0x0008	/* vinvalbuf: invalidate only clean bufs */
393#define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
394#define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
395#define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
396#define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
397
398#define	VR_START_WRITE	0x0001	/* vfs_write_resume: start write atomically */
399#define	VR_NO_SUSPCLR	0x0002	/* vfs_write_resume: do not clear suspension */
400
401#define	VS_SKIP_UNMOUNT	0x0001	/* vfs_write_suspend: fail if the
402				   filesystem is being unmounted */
403
404#define	VREF(vp)	vref(vp)
405
406#ifdef DIAGNOSTIC
407#define	VATTR_NULL(vap)	vattr_null(vap)
408#else
409#define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
410#endif /* DIAGNOSTIC */
411
412#define	NULLVP	((struct vnode *)NULL)
413
414/*
415 * Global vnode data.
416 */
417extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
418extern	int async_io_version;		/* 0 or POSIX version of AIO i'face */
419extern	int desiredvnodes;		/* number of vnodes desired */
420extern	struct uma_zone *namei_zone;
421extern	struct vattr va_null;		/* predefined null vattr structure */
422
423#define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
424#define	VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
425#define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
426#define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
427#define	VI_MTX(vp)	(&(vp)->v_interlock)
428
429#define	VN_LOCK_AREC(vp)	lockallowrecurse((vp)->v_vnlock)
430#define	VN_LOCK_ASHARE(vp)	lockallowshare((vp)->v_vnlock)
431#define	VN_LOCK_DSHARE(vp)	lockdisableshare((vp)->v_vnlock)
432
433#endif /* _KERNEL */
434
435/*
436 * Mods for extensibility.
437 */
438
439/*
440 * Flags for vdesc_flags:
441 */
442#define	VDESC_MAX_VPS		16
443/* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
444#define	VDESC_VP0_WILLRELE	0x0001
445#define	VDESC_VP1_WILLRELE	0x0002
446#define	VDESC_VP2_WILLRELE	0x0004
447#define	VDESC_VP3_WILLRELE	0x0008
448#define	VDESC_NOMAP_VPP		0x0100
449#define	VDESC_VPP_WILLRELE	0x0200
450
451/*
452 * A generic structure.
453 * This can be used by bypass routines to identify generic arguments.
454 */
455struct vop_generic_args {
456	struct vnodeop_desc *a_desc;
457	/* other random data follows, presumably */
458};
459
460typedef int vop_bypass_t(struct vop_generic_args *);
461
462/*
463 * VDESC_NO_OFFSET is used to identify the end of the offset list
464 * and in places where no such field exists.
465 */
466#define VDESC_NO_OFFSET -1
467
468/*
469 * This structure describes the vnode operation taking place.
470 */
471struct vnodeop_desc {
472	char	*vdesc_name;		/* a readable name for debugging */
473	int	 vdesc_flags;		/* VDESC_* flags */
474	vop_bypass_t	*vdesc_call;	/* Function to call */
475
476	/*
477	 * These ops are used by bypass routines to map and locate arguments.
478	 * Creds and procs are not needed in bypass routines, but sometimes
479	 * they are useful to (for example) transport layers.
480	 * Nameidata is useful because it has a cred in it.
481	 */
482	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
483	int	vdesc_vpp_offset;	/* return vpp location */
484	int	vdesc_cred_offset;	/* cred location, if any */
485	int	vdesc_thread_offset;	/* thread location, if any */
486	int	vdesc_componentname_offset; /* if any */
487};
488
489#ifdef _KERNEL
490/*
491 * A list of all the operation descs.
492 */
493extern struct vnodeop_desc *vnodeop_descs[];
494
495#define	VOPARG_OFFSETOF(s_type, field)	__offsetof(s_type, field)
496#define	VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
497    ((s_type)(((char*)(struct_p)) + (s_offset)))
498
499
500#ifdef DEBUG_VFS_LOCKS
501/*
502 * Support code to aid in debugging VFS locking problems.  Not totally
503 * reliable since if the thread sleeps between changing the lock
504 * state and checking it with the assert, some other thread could
505 * change the state.  They are good enough for debugging a single
506 * filesystem using a single-threaded test.
507 */
508void	assert_vi_locked(struct vnode *vp, const char *str);
509void	assert_vi_unlocked(struct vnode *vp, const char *str);
510void	assert_vop_elocked(struct vnode *vp, const char *str);
511#if 0
512void	assert_vop_elocked_other(struct vnode *vp, const char *str);
513#endif
514void	assert_vop_locked(struct vnode *vp, const char *str);
515#if 0
516voi0	assert_vop_slocked(struct vnode *vp, const char *str);
517#endif
518void	assert_vop_unlocked(struct vnode *vp, const char *str);
519
520#define	ASSERT_VI_LOCKED(vp, str)	assert_vi_locked((vp), (str))
521#define	ASSERT_VI_UNLOCKED(vp, str)	assert_vi_unlocked((vp), (str))
522#define	ASSERT_VOP_ELOCKED(vp, str)	assert_vop_elocked((vp), (str))
523#if 0
524#define	ASSERT_VOP_ELOCKED_OTHER(vp, str) assert_vop_locked_other((vp), (str))
525#endif
526#define	ASSERT_VOP_LOCKED(vp, str)	assert_vop_locked((vp), (str))
527#if 0
528#define	ASSERT_VOP_SLOCKED(vp, str)	assert_vop_slocked((vp), (str))
529#endif
530#define	ASSERT_VOP_UNLOCKED(vp, str)	assert_vop_unlocked((vp), (str))
531
532#else /* !DEBUG_VFS_LOCKS */
533
534#define	ASSERT_VI_LOCKED(vp, str)	((void)0)
535#define	ASSERT_VI_UNLOCKED(vp, str)	((void)0)
536#define	ASSERT_VOP_ELOCKED(vp, str)	((void)0)
537#if 0
538#define	ASSERT_VOP_ELOCKED_OTHER(vp, str)
539#endif
540#define	ASSERT_VOP_LOCKED(vp, str)	((void)0)
541#if 0
542#define	ASSERT_VOP_SLOCKED(vp, str)
543#endif
544#define	ASSERT_VOP_UNLOCKED(vp, str)	((void)0)
545#endif /* DEBUG_VFS_LOCKS */
546
547
548/*
549 * This call works for vnodes in the kernel.
550 */
551#define VCALL(c) ((c)->a_desc->vdesc_call(c))
552
553#define DOINGASYNC(vp)	   					\
554	(((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 &&	\
555	 ((curthread->td_pflags & TDP_SYNCIO) == 0))
556
557/*
558 * VMIO support inline
559 */
560
561extern int vmiodirenable;
562
563static __inline int
564vn_canvmio(struct vnode *vp)
565{
566      if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
567		return(TRUE);
568	return(FALSE);
569}
570
571/*
572 * Finally, include the default set of vnode operations.
573 */
574#include "vnode_if.h"
575
576/* vn_open_flags */
577#define	VN_OPEN_NOAUDIT		0x00000001
578#define	VN_OPEN_NOCAPCHECK	0x00000002
579
580/*
581 * Public vnode manipulation functions.
582 */
583struct componentname;
584struct file;
585struct mount;
586struct nameidata;
587struct ostat;
588struct thread;
589struct proc;
590struct stat;
591struct nstat;
592struct ucred;
593struct uio;
594struct vattr;
595struct vnode;
596
597typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
598
599/* cache_* may belong in namei.h. */
600#define	cache_enter(dvp, vp, cnp)					\
601	cache_enter_time(dvp, vp, cnp, NULL, NULL)
602void	cache_enter_time(struct vnode *dvp, struct vnode *vp,
603	    struct componentname *cnp, struct timespec *tsp,
604	    struct timespec *dtsp);
605int	cache_lookup(struct vnode *dvp, struct vnode **vpp,
606	    struct componentname *cnp, struct timespec *tsp, int *ticksp);
607void	cache_purge(struct vnode *vp);
608void	cache_purge_negative(struct vnode *vp);
609void	cache_purgevfs(struct mount *mp);
610int	change_dir(struct vnode *vp, struct thread *td);
611int	change_root(struct vnode *vp, struct thread *td);
612void	cvtstat(struct stat *st, struct ostat *ost);
613void	cvtnstat(struct stat *sb, struct nstat *nsb);
614int	getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
615	    struct vnode **vpp);
616void	getnewvnode_reserve(u_int count);
617void	getnewvnode_drop_reserve(void);
618int	insmntque1(struct vnode *vp, struct mount *mp,
619	    void (*dtr)(struct vnode *, void *), void *dtr_arg);
620int	insmntque(struct vnode *vp, struct mount *mp);
621u_quad_t init_va_filerev(void);
622int	speedup_syncer(void);
623int	vn_vptocnp(struct vnode **vp, struct ucred *cred, char *buf,
624	    u_int *buflen);
625#define textvp_fullpath(p, rb, rfb) \
626	vn_fullpath(FIRST_THREAD_IN_PROC(p), (p)->p_textvp, rb, rfb)
627int	vn_fullpath(struct thread *td, struct vnode *vn,
628	    char **retbuf, char **freebuf);
629int	vn_fullpath_global(struct thread *td, struct vnode *vn,
630	    char **retbuf, char **freebuf);
631struct vnode *
632	vn_dir_dd_ino(struct vnode *vp);
633int	vn_commname(struct vnode *vn, char *buf, u_int buflen);
634int	vn_path_to_global_path(struct thread *td, struct vnode *vp,
635	    char *path, u_int pathlen);
636int	vaccess(enum vtype type, mode_t file_mode, uid_t file_uid,
637	    gid_t file_gid, accmode_t accmode, struct ucred *cred,
638	    int *privused);
639int	vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid,
640	    struct acl *aclp, accmode_t accmode, struct ucred *cred,
641	    int *privused);
642int	vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
643	    gid_t file_gid, struct acl *acl, accmode_t accmode,
644	    struct ucred *cred, int *privused);
645void	vattr_null(struct vattr *vap);
646int	vcount(struct vnode *vp);
647void	vdrop(struct vnode *);
648void	vdropl(struct vnode *);
649int	vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
650int	vget(struct vnode *vp, int lockflag, struct thread *td);
651void	vgone(struct vnode *vp);
652void	vhold(struct vnode *);
653void	vholdl(struct vnode *);
654void	vinactive(struct vnode *, struct thread *);
655int	vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
656int	vtruncbuf(struct vnode *vp, struct ucred *cred, off_t length,
657	    int blksize);
658void	vunref(struct vnode *);
659void	vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
660#define vprint(label, vp) vn_printf((vp), "%s\n", (label))
661int	vrecycle(struct vnode *vp);
662int	vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
663	    struct ucred *cred);
664int	vn_close(struct vnode *vp,
665	    int flags, struct ucred *file_cred, struct thread *td);
666void	vn_finished_write(struct mount *mp);
667void	vn_finished_secondary_write(struct mount *mp);
668int	vn_isdisk(struct vnode *vp, int *errp);
669int	_vn_lock(struct vnode *vp, int flags, char *file, int line);
670#define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
671int	vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
672int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
673	    u_int vn_open_flags, struct ucred *cred, struct file *fp);
674int	vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
675	    struct thread *td, struct file *fp);
676void	vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
677int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
678int	vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
679	    int len, off_t offset, enum uio_seg segflg, int ioflg,
680	    struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
681	    struct thread *td);
682int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
683	    size_t len, off_t offset, enum uio_seg segflg, int ioflg,
684	    struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
685	    struct thread *td);
686int	vn_rlimit_fsize(const struct vnode *vn, const struct uio *uio,
687	    const struct thread *td);
688int	vn_stat(struct vnode *vp, struct stat *sb, struct ucred *active_cred,
689	    struct ucred *file_cred, struct thread *td);
690int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
691int	vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
692	    int flags);
693int	vn_writechk(struct vnode *vp);
694int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
695	    const char *attrname, int *buflen, char *buf, struct thread *td);
696int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
697	    const char *attrname, int buflen, char *buf, struct thread *td);
698int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
699	    const char *attrname, struct thread *td);
700int	vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
701	    struct vnode **rvp);
702int	vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
703	    void *alloc_arg, int lkflags, struct vnode **rvp);
704int	vn_utimes_perm(struct vnode *vp, struct vattr *vap,
705	    struct ucred *cred, struct thread *td);
706
707int	vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
708int	vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
709	    struct uio *uio);
710
711#define	vn_rangelock_unlock(vp, cookie)					\
712	rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp))
713#define	vn_rangelock_unlock_range(vp, cookie, start, end)		\
714	rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), 	\
715	    VI_MTX(vp))
716#define	vn_rangelock_rlock(vp, start, end)				\
717	rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
718#define	vn_rangelock_wlock(vp, start, end)				\
719	rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
720
721int	vfs_cache_lookup(struct vop_lookup_args *ap);
722void	vfs_timestamp(struct timespec *);
723void	vfs_write_resume(struct mount *mp, int flags);
724int	vfs_write_suspend(struct mount *mp, int flags);
725int	vfs_write_suspend_umnt(struct mount *mp);
726int	vop_stdbmap(struct vop_bmap_args *);
727int	vop_stdfsync(struct vop_fsync_args *);
728int	vop_stdgetwritemount(struct vop_getwritemount_args *);
729int	vop_stdgetpages(struct vop_getpages_args *);
730int	vop_stdinactive(struct vop_inactive_args *);
731int	vop_stdislocked(struct vop_islocked_args *);
732int	vop_stdkqfilter(struct vop_kqfilter_args *);
733int	vop_stdlock(struct vop_lock1_args *);
734int	vop_stdputpages(struct vop_putpages_args *);
735int	vop_stdunlock(struct vop_unlock_args *);
736int	vop_nopoll(struct vop_poll_args *);
737int	vop_stdaccess(struct vop_access_args *ap);
738int	vop_stdaccessx(struct vop_accessx_args *ap);
739int	vop_stdadvise(struct vop_advise_args *ap);
740int	vop_stdadvlock(struct vop_advlock_args *ap);
741int	vop_stdadvlockasync(struct vop_advlockasync_args *ap);
742int	vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
743int	vop_stdallocate(struct vop_allocate_args *ap);
744int	vop_stdpathconf(struct vop_pathconf_args *);
745int	vop_stdpoll(struct vop_poll_args *);
746int	vop_stdvptocnp(struct vop_vptocnp_args *ap);
747int	vop_stdvptofh(struct vop_vptofh_args *ap);
748int	vop_stdunp_bind(struct vop_unp_bind_args *ap);
749int	vop_stdunp_connect(struct vop_unp_connect_args *ap);
750int	vop_stdunp_detach(struct vop_unp_detach_args *ap);
751int	vop_eopnotsupp(struct vop_generic_args *ap);
752int	vop_ebadf(struct vop_generic_args *ap);
753int	vop_einval(struct vop_generic_args *ap);
754int	vop_enoent(struct vop_generic_args *ap);
755int	vop_enotty(struct vop_generic_args *ap);
756int	vop_null(struct vop_generic_args *ap);
757int	vop_panic(struct vop_generic_args *ap);
758
759/* These are called from within the actual VOPS. */
760void	vop_create_post(void *a, int rc);
761void	vop_deleteextattr_post(void *a, int rc);
762void	vop_link_post(void *a, int rc);
763void	vop_lock_pre(void *a);
764void	vop_lock_post(void *a, int rc);
765void	vop_lookup_post(void *a, int rc);
766void	vop_lookup_pre(void *a);
767void	vop_mkdir_post(void *a, int rc);
768void	vop_mknod_post(void *a, int rc);
769void	vop_remove_post(void *a, int rc);
770void	vop_rename_post(void *a, int rc);
771void	vop_rename_pre(void *a);
772void	vop_rmdir_post(void *a, int rc);
773void	vop_setattr_post(void *a, int rc);
774void	vop_setextattr_post(void *a, int rc);
775void	vop_strategy_pre(void *a);
776void	vop_symlink_post(void *a, int rc);
777void	vop_unlock_post(void *a, int rc);
778void	vop_unlock_pre(void *a);
779
780void	vop_rename_fail(struct vop_rename_args *ap);
781
782#define	VOP_WRITE_PRE(ap)						\
783	struct vattr va;						\
784	int error;							\
785	off_t osize, ooffset, noffset;					\
786									\
787	osize = ooffset = noffset = 0;					\
788	if (!VN_KNLIST_EMPTY((ap)->a_vp)) {				\
789		error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred);	\
790		if (error)						\
791			return (error);					\
792		ooffset = (ap)->a_uio->uio_offset;			\
793		osize = (off_t)va.va_size;				\
794	}
795
796#define VOP_WRITE_POST(ap, ret)						\
797	noffset = (ap)->a_uio->uio_offset;				\
798	if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) {	\
799		VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE			\
800		    | (noffset > osize ? NOTE_EXTEND : 0));		\
801	}
802
803#define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
804
805
806void	vput(struct vnode *vp);
807void	vrele(struct vnode *vp);
808void	vref(struct vnode *vp);
809int	vrefcnt(struct vnode *vp);
810void 	v_addpollinfo(struct vnode *vp);
811
812int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
813void vnode_destroy_vobject(struct vnode *vp);
814
815extern struct vop_vector fifo_specops;
816extern struct vop_vector dead_vnodeops;
817extern struct vop_vector default_vnodeops;
818
819#define VOP_PANIC	((void*)(uintptr_t)vop_panic)
820#define VOP_NULL	((void*)(uintptr_t)vop_null)
821#define VOP_EBADF	((void*)(uintptr_t)vop_ebadf)
822#define VOP_ENOTTY	((void*)(uintptr_t)vop_enotty)
823#define VOP_EINVAL	((void*)(uintptr_t)vop_einval)
824#define VOP_ENOENT	((void*)(uintptr_t)vop_enoent)
825#define VOP_EOPNOTSUPP	((void*)(uintptr_t)vop_eopnotsupp)
826
827/* fifo_vnops.c */
828int	fifo_printinfo(struct vnode *);
829
830/* vfs_hash.c */
831typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
832
833int vfs_hash_get(const struct mount *mp, u_int hash, int flags, struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
834u_int vfs_hash_index(struct vnode *vp);
835int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
836void vfs_hash_rehash(struct vnode *vp, u_int hash);
837void vfs_hash_remove(struct vnode *vp);
838
839int vfs_kqfilter(struct vop_kqfilter_args *);
840void vfs_mark_atime(struct vnode *vp, struct ucred *cred);
841struct dirent;
842int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
843
844int vfs_unixify_accmode(accmode_t *accmode);
845
846void vfs_unp_reclaim(struct vnode *vp);
847
848int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
849int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
850    gid_t gid);
851int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
852    struct thread *td);
853int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
854    struct thread *td);
855
856#endif /* _KERNEL */
857
858#endif /* !_SYS_VNODE_H_ */
859