vnode.h revision 218195
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: head/sys/sys/vnode.h 218195 2011-02-02 16:35:10Z mdf $
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/selinfo.h>
42#include <sys/uio.h>
43#include <sys/acl.h>
44#include <sys/ktr.h>
45
46/*
47 * The vnode is the focus of all file activity in UNIX.  There is a
48 * unique vnode allocated for each active file, each current directory,
49 * each mounted-on file, text file, and the root.
50 */
51
52/*
53 * Vnode types.  VNON means no type.
54 */
55enum vtype	{ VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD,
56		  VMARKER };
57
58/*
59 * Each underlying filesystem allocates its own private area and hangs
60 * it from v_data.  If non-null, this area is freed in getnewvnode().
61 */
62
63struct namecache;
64
65struct vpollinfo {
66	struct	mtx vpi_lock;		/* lock to protect below */
67	struct	selinfo vpi_selinfo;	/* identity of poller(s) */
68	short	vpi_events;		/* what they are looking for */
69	short	vpi_revents;		/* what has happened */
70};
71
72/*
73 * Reading or writing any of these items requires holding the appropriate lock.
74 *
75 * Lock reference:
76 *	c - namecache mutex
77 *	f - freelist mutex
78 *	G - Giant
79 *	i - interlock
80 *	m - mntvnodes mutex
81 *	p - pollinfo lock
82 *	s - spechash mutex
83 *	S - syncer mutex
84 *	u - Only a reference to the vnode is needed to read.
85 *	v - vnode lock
86 *
87 * Vnodes may be found on many lists.  The general way to deal with operating
88 * on a vnode that is on a list is:
89 *	1) Lock the list and find the vnode.
90 *	2) Lock interlock so that the vnode does not go away.
91 *	3) Unlock the list to avoid lock order reversals.
92 *	4) vget with LK_INTERLOCK and check for ENOENT, or
93 *	5) Check for DOOMED if the vnode lock is not required.
94 *	6) Perform your operation, then vput().
95 */
96
97#if defined(_KERNEL) || defined(_KVM_VNODE)
98
99struct vnode {
100	/*
101	 * Fields which define the identity of the vnode.  These fields are
102	 * owned by the filesystem (XXX: and vgone() ?)
103	 */
104	enum	vtype v_type;			/* u vnode type */
105	const char *v_tag;			/* u type of underlying data */
106	struct	vop_vector *v_op;		/* u vnode operations vector */
107	void	*v_data;			/* u private data for fs */
108
109	/*
110	 * Filesystem instance stuff
111	 */
112	struct	mount *v_mount;			/* u ptr to vfs we are in */
113	TAILQ_ENTRY(vnode) v_nmntvnodes;	/* m vnodes for mount point */
114
115	/*
116	 * Type specific fields, only one applies to any given vnode.
117	 * See #defines below for renaming to v_* namespace.
118	 */
119	union {
120		struct mount	*vu_mount;	/* v ptr to mountpoint (VDIR) */
121		struct socket	*vu_socket;	/* v unix domain net (VSOCK) */
122		struct cdev	*vu_cdev; 	/* v device (VCHR, VBLK) */
123		struct fifoinfo	*vu_fifoinfo;	/* v fifo (VFIFO) */
124	} v_un;
125
126	/*
127	 * vfs_hash:  (mount + inode) -> vnode hash.
128	 */
129	LIST_ENTRY(vnode)	v_hashlist;
130	u_int			v_hash;
131
132	/*
133	 * VFS_namecache stuff
134	 */
135	LIST_HEAD(, namecache) v_cache_src;	/* c Cache entries from us */
136	TAILQ_HEAD(, namecache) v_cache_dst;	/* c Cache entries to us */
137	struct namecache *v_cache_dd;		/* c Cache entry for .. vnode */
138
139	/*
140	 * clustering stuff
141	 */
142	daddr_t	v_cstart;			/* v start block of cluster */
143	daddr_t	v_lasta;			/* v last allocation  */
144	daddr_t	v_lastw;			/* v last write  */
145	int	v_clen;				/* v length of cur. cluster */
146
147	/*
148	 * Locking
149	 */
150	struct	lock v_lock;			/* u (if fs don't have one) */
151	struct	mtx v_interlock;		/* lock for "i" things */
152	struct	lock *v_vnlock;			/* u pointer to vnode lock */
153	int	v_holdcnt;			/* i prevents recycling. */
154	int	v_usecount;			/* i ref count of users */
155	u_long	v_iflag;			/* i vnode flags (see below) */
156	u_long	v_vflag;			/* v vnode flags */
157	int	v_writecount;			/* v ref count of writers */
158
159	/*
160	 * The machinery of being a vnode
161	 */
162	TAILQ_ENTRY(vnode) v_freelist;		/* f vnode freelist */
163	struct bufobj	v_bufobj;		/* * Buffer cache object */
164
165	/*
166	 * Hooks for various subsystems and features.
167	 */
168	struct vpollinfo *v_pollinfo;		/* G Poll events, p for *v_pi */
169	struct label *v_label;			/* MAC label for vnode */
170	struct lockf *v_lockf;			/* Byte-level lock list */
171};
172
173#endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
174
175#define	v_mountedhere	v_un.vu_mount
176#define	v_socket	v_un.vu_socket
177#define	v_rdev		v_un.vu_cdev
178#define	v_fifoinfo	v_un.vu_fifoinfo
179
180/* XXX: These are temporary to avoid a source sweep at this time */
181#define v_object	v_bufobj.bo_object
182
183/*
184 * Userland version of struct vnode, for sysctl.
185 */
186struct xvnode {
187	size_t	xv_size;			/* sizeof(struct xvnode) */
188	void	*xv_vnode;			/* address of real vnode */
189	u_long	xv_flag;			/* vnode vflags */
190	int	xv_usecount;			/* reference count of users */
191	int	xv_writecount;			/* reference count of writers */
192	int	xv_holdcnt;			/* page & buffer references */
193	u_long	xv_id;				/* capability identifier */
194	void	*xv_mount;			/* address of parent mount */
195	long	xv_numoutput;			/* num of writes in progress */
196	enum	vtype xv_type;			/* vnode type */
197	union {
198		void	*xvu_socket;		/* socket, if VSOCK */
199		void	*xvu_fifo;		/* fifo, if VFIFO */
200		dev_t	xvu_rdev;		/* maj/min, if VBLK/VCHR */
201		struct {
202			dev_t	xvu_dev;	/* device, if VDIR/VREG/VLNK */
203			ino_t	xvu_ino;	/* id, if VDIR/VREG/VLNK */
204		} xv_uns;
205	} xv_un;
206};
207#define xv_socket	xv_un.xvu_socket
208#define xv_fifo		xv_un.xvu_fifo
209#define xv_rdev		xv_un.xvu_rdev
210#define xv_dev		xv_un.xv_uns.xvu_dev
211#define xv_ino		xv_un.xv_uns.xvu_ino
212
213/* We don't need to lock the knlist */
214#define	VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL ||	\
215	    KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
216
217#define VN_KNOTE(vp, b, a)					\
218	do {							\
219		if (!VN_KNLIST_EMPTY(vp))			\
220			KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
221			    (a) | KNF_NOKQLOCK);		\
222	} while (0)
223#define	VN_KNOTE_LOCKED(vp, b)		VN_KNOTE(vp, b, KNF_LISTLOCKED)
224#define	VN_KNOTE_UNLOCKED(vp, b)	VN_KNOTE(vp, b, 0)
225
226/*
227 * Vnode flags.
228 *	VI flags are protected by interlock and live in v_iflag
229 *	VV flags are protected by the vnode lock and live in v_vflag
230 *
231 *	VI_DOOMED is doubly protected by the interlock and vnode lock.  Both
232 *	are required for writing but the status may be checked with either.
233 */
234#define	VI_MOUNT	0x0020	/* Mount in progress */
235#define	VI_AGE		0x0040	/* Insert vnode at head of free list */
236#define	VI_DOOMED	0x0080	/* This vnode is being recycled */
237#define	VI_FREE		0x0100	/* This vnode is on the freelist */
238#define	VI_DOINGINACT	0x0800	/* VOP_INACTIVE is in progress */
239#define	VI_OWEINACT	0x1000	/* Need to call inactive */
240
241#define	VV_ROOT		0x0001	/* root of its filesystem */
242#define	VV_ISTTY	0x0002	/* vnode represents a tty */
243#define	VV_NOSYNC	0x0004	/* unlinked, stop syncing */
244#define	VV_ETERNALDEV	0x0008	/* device that is never destroyed */
245#define	VV_CACHEDLABEL	0x0010	/* Vnode has valid cached MAC label */
246#define	VV_TEXT		0x0020	/* vnode is a pure text prototype */
247#define	VV_COPYONWRITE	0x0040	/* vnode is doing copy-on-write */
248#define	VV_SYSTEM	0x0080	/* vnode being used by kernel */
249#define	VV_PROCDEP	0x0100	/* vnode is process dependent */
250#define	VV_NOKNOTE	0x0200	/* don't activate knotes on this vnode */
251#define	VV_DELETED	0x0400	/* should be removed */
252#define	VV_MD		0x0800	/* vnode backs the md device */
253#define	VV_FORCEINSMQ	0x1000	/* force the insmntque to succeed */
254
255/*
256 * Vnode attributes.  A field value of VNOVAL represents a field whose value
257 * is unavailable (getattr) or which is not to be changed (setattr).
258 */
259struct vattr {
260	enum vtype	va_type;	/* vnode type (for create) */
261	u_short		va_mode;	/* files access mode and type */
262	short		va_nlink;	/* number of references to file */
263	uid_t		va_uid;		/* owner user id */
264	gid_t		va_gid;		/* owner group id */
265	dev_t		va_fsid;	/* filesystem id */
266	long		va_fileid;	/* file id */
267	u_quad_t	va_size;	/* file size in bytes */
268	long		va_blocksize;	/* blocksize preferred for i/o */
269	struct timespec	va_atime;	/* time of last access */
270	struct timespec	va_mtime;	/* time of last modification */
271	struct timespec	va_ctime;	/* time file changed */
272	struct timespec	va_birthtime;	/* time file created */
273	u_long		va_gen;		/* generation number of file */
274	u_long		va_flags;	/* flags defined for file */
275	dev_t		va_rdev;	/* device the special file represents */
276	u_quad_t	va_bytes;	/* bytes of disk space held by file */
277	u_quad_t	va_filerev;	/* file modification number */
278	u_int		va_vaflags;	/* operations flags, see below */
279	long		va_spare;	/* remain quad aligned */
280};
281
282/*
283 * Flags for va_vaflags.
284 */
285#define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
286#define	VA_EXCLUSIVE	0x02		/* exclusive create request */
287
288/*
289 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
290 * help with write clustering)
291 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
292 */
293#define	IO_UNIT		0x0001		/* do I/O as atomic unit */
294#define	IO_APPEND	0x0002		/* append write to end */
295#define	IO_NDELAY	0x0004		/* FNDELAY flag set in file table */
296#define	IO_NODELOCKED	0x0008		/* underlying node already locked */
297#define	IO_ASYNC	0x0010		/* bawrite rather then bdwrite */
298#define	IO_VMIO		0x0020		/* data already in VMIO space */
299#define	IO_INVAL	0x0040		/* invalidate after I/O */
300#define	IO_SYNC		0x0080		/* do I/O synchronously */
301#define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
302#define	IO_EXT		0x0400		/* operate on external attributes */
303#define	IO_NORMAL	0x0800		/* operate on regular data */
304#define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
305
306#define IO_SEQMAX	0x7F		/* seq heuristic max value */
307#define IO_SEQSHIFT	16		/* seq heuristic in upper 16 bits */
308
309/*
310 * Flags for accmode_t.
311 */
312#define	VEXEC			000000000100 /* execute/search permission */
313#define	VWRITE			000000000200 /* write permission */
314#define	VREAD			000000000400 /* read permission */
315#define	VADMIN			000000010000 /* being the file owner */
316#define	VAPPEND			000000040000 /* permission to write/append */
317/*
318 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
319 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
320 * and 0 otherwise.  This never happens with ordinary unix access rights
321 * or POSIX.1e ACLs.  Obviously, VEXPLICIT_DENY must be OR-ed with
322 * some other V* constant.
323 */
324#define	VEXPLICIT_DENY		000000100000
325#define	VREAD_NAMED_ATTRS 	000000200000 /* not used */
326#define	VWRITE_NAMED_ATTRS 	000000400000 /* not used */
327#define	VDELETE_CHILD	 	000001000000
328#define	VREAD_ATTRIBUTES 	000002000000 /* permission to stat(2) */
329#define	VWRITE_ATTRIBUTES 	000004000000 /* change {m,c,a}time */
330#define	VDELETE		 	000010000000
331#define	VREAD_ACL	 	000020000000 /* read ACL and file mode */
332#define	VWRITE_ACL	 	000040000000 /* change ACL and/or file mode */
333#define	VWRITE_OWNER	 	000100000000 /* change file owner */
334#define	VSYNCHRONIZE	 	000200000000 /* not used */
335
336/*
337 * Permissions that were traditionally granted only to the file owner.
338 */
339#define VADMIN_PERMS	(VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
340    VWRITE_OWNER)
341
342/*
343 * Permissions that were traditionally granted to everyone.
344 */
345#define VSTAT_PERMS	(VREAD_ATTRIBUTES | VREAD_ACL)
346
347/*
348 * Permissions that allow to change the state of the file in any way.
349 */
350#define VMODIFY_PERMS	(VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
351    VDELETE)
352
353/*
354 * Token indicating no attribute value yet assigned.
355 */
356#define	VNOVAL	(-1)
357
358/*
359 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
360 */
361#define VLKTIMEOUT	(hz / 20 + 1)
362
363#ifdef _KERNEL
364
365#ifdef MALLOC_DECLARE
366MALLOC_DECLARE(M_VNODE);
367#endif
368
369/*
370 * Convert between vnode types and inode formats (since POSIX.1
371 * defines mode word of stat structure in terms of inode formats).
372 */
373extern enum vtype	iftovt_tab[];
374extern int		vttoif_tab[];
375#define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
376#define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
377#define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
378
379/*
380 * Flags to various vnode functions.
381 */
382#define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
383#define	FORCECLOSE	0x0002	/* vflush: force file closure */
384#define	WRITECLOSE	0x0004	/* vflush: only close writable files */
385#define	DOCLOSE		0x0008	/* vclean: close active files */
386#define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
387#define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
388#define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
389#define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
390#define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
391#define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
392#define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
393
394#define	VREF(vp)	vref(vp)
395
396#ifdef DIAGNOSTIC
397#define	VATTR_NULL(vap)	vattr_null(vap)
398#else
399#define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
400#endif /* DIAGNOSTIC */
401
402#define	NULLVP	((struct vnode *)NULL)
403
404/*
405 * Global vnode data.
406 */
407extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
408extern	int async_io_version;		/* 0 or POSIX version of AIO i'face */
409extern	int desiredvnodes;		/* number of vnodes desired */
410extern	struct uma_zone *namei_zone;
411extern	struct vattr va_null;		/* predefined null vattr structure */
412
413#define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
414#define	VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
415#define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
416#define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
417#define	VI_MTX(vp)	(&(vp)->v_interlock)
418
419#define	VN_LOCK_AREC(vp)	lockallowrecurse((vp)->v_vnlock)
420#define	VN_LOCK_ASHARE(vp)	lockallowshare((vp)->v_vnlock)
421
422#endif /* _KERNEL */
423
424/*
425 * Mods for extensibility.
426 */
427
428/*
429 * Flags for vdesc_flags:
430 */
431#define	VDESC_MAX_VPS		16
432/* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
433#define	VDESC_VP0_WILLRELE	0x0001
434#define	VDESC_VP1_WILLRELE	0x0002
435#define	VDESC_VP2_WILLRELE	0x0004
436#define	VDESC_VP3_WILLRELE	0x0008
437#define	VDESC_NOMAP_VPP		0x0100
438#define	VDESC_VPP_WILLRELE	0x0200
439
440/*
441 * A generic structure.
442 * This can be used by bypass routines to identify generic arguments.
443 */
444struct vop_generic_args {
445	struct vnodeop_desc *a_desc;
446	/* other random data follows, presumably */
447};
448
449typedef int vop_bypass_t(struct vop_generic_args *);
450
451/*
452 * VDESC_NO_OFFSET is used to identify the end of the offset list
453 * and in places where no such field exists.
454 */
455#define VDESC_NO_OFFSET -1
456
457/*
458 * This structure describes the vnode operation taking place.
459 */
460struct vnodeop_desc {
461	char	*vdesc_name;		/* a readable name for debugging */
462	int	 vdesc_flags;		/* VDESC_* flags */
463	vop_bypass_t	*vdesc_call;	/* Function to call */
464
465	/*
466	 * These ops are used by bypass routines to map and locate arguments.
467	 * Creds and procs are not needed in bypass routines, but sometimes
468	 * they are useful to (for example) transport layers.
469	 * Nameidata is useful because it has a cred in it.
470	 */
471	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
472	int	vdesc_vpp_offset;	/* return vpp location */
473	int	vdesc_cred_offset;	/* cred location, if any */
474	int	vdesc_thread_offset;	/* thread location, if any */
475	int	vdesc_componentname_offset; /* if any */
476};
477
478#ifdef _KERNEL
479/*
480 * A list of all the operation descs.
481 */
482extern struct vnodeop_desc *vnodeop_descs[];
483
484#define	VOPARG_OFFSETOF(s_type, field)	__offsetof(s_type, field)
485#define	VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
486    ((s_type)(((char*)(struct_p)) + (s_offset)))
487
488
489#ifdef DEBUG_VFS_LOCKS
490/*
491 * Support code to aid in debugging VFS locking problems.  Not totally
492 * reliable since if the thread sleeps between changing the lock
493 * state and checking it with the assert, some other thread could
494 * change the state.  They are good enough for debugging a single
495 * filesystem using a single-threaded test.
496 */
497void	assert_vi_locked(struct vnode *vp, const char *str);
498void	assert_vi_unlocked(struct vnode *vp, const char *str);
499void	assert_vop_elocked(struct vnode *vp, const char *str);
500#if 0
501void	assert_vop_elocked_other(struct vnode *vp, const char *str);
502#endif
503void	assert_vop_locked(struct vnode *vp, const char *str);
504#if 0
505voi0	assert_vop_slocked(struct vnode *vp, const char *str);
506#endif
507void	assert_vop_unlocked(struct vnode *vp, const char *str);
508
509#define	ASSERT_VI_LOCKED(vp, str)	assert_vi_locked((vp), (str))
510#define	ASSERT_VI_UNLOCKED(vp, str)	assert_vi_unlocked((vp), (str))
511#define	ASSERT_VOP_ELOCKED(vp, str)	assert_vop_elocked((vp), (str))
512#if 0
513#define	ASSERT_VOP_ELOCKED_OTHER(vp, str) assert_vop_locked_other((vp), (str))
514#endif
515#define	ASSERT_VOP_LOCKED(vp, str)	assert_vop_locked((vp), (str))
516#if 0
517#define	ASSERT_VOP_SLOCKED(vp, str)	assert_vop_slocked((vp), (str))
518#endif
519#define	ASSERT_VOP_UNLOCKED(vp, str)	assert_vop_unlocked((vp), (str))
520
521#else /* !DEBUG_VFS_LOCKS */
522
523#define	ASSERT_VI_LOCKED(vp, str)	((void)0)
524#define	ASSERT_VI_UNLOCKED(vp, str)	((void)0)
525#define	ASSERT_VOP_ELOCKED(vp, str)	((void)0)
526#if 0
527#define	ASSERT_VOP_ELOCKED_OTHER(vp, str)
528#endif
529#define	ASSERT_VOP_LOCKED(vp, str)	((void)0)
530#if 0
531#define	ASSERT_VOP_SLOCKED(vp, str)
532#endif
533#define	ASSERT_VOP_UNLOCKED(vp, str)	((void)0)
534#endif /* DEBUG_VFS_LOCKS */
535
536
537/*
538 * This call works for vnodes in the kernel.
539 */
540#define VCALL(c) ((c)->a_desc->vdesc_call(c))
541
542/*
543 * VMIO support inline
544 */
545
546extern int vmiodirenable;
547
548static __inline int
549vn_canvmio(struct vnode *vp)
550{
551      if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
552		return(TRUE);
553	return(FALSE);
554}
555
556/*
557 * Finally, include the default set of vnode operations.
558 */
559#include "vnode_if.h"
560
561/* vn_open_flags */
562#define	VN_OPEN_NOAUDIT		0x00000001
563
564/*
565 * Public vnode manipulation functions.
566 */
567struct componentname;
568struct file;
569struct mount;
570struct nameidata;
571struct ostat;
572struct thread;
573struct proc;
574struct stat;
575struct nstat;
576struct ucred;
577struct uio;
578struct vattr;
579struct vnode;
580
581/* cache_* may belong in namei.h. */
582void	cache_enter(struct vnode *dvp, struct vnode *vp,
583	    struct componentname *cnp);
584int	cache_lookup(struct vnode *dvp, struct vnode **vpp,
585	    struct componentname *cnp);
586void	cache_purge(struct vnode *vp);
587void	cache_purge_negative(struct vnode *vp);
588void	cache_purgevfs(struct mount *mp);
589int	change_dir(struct vnode *vp, struct thread *td);
590int	change_root(struct vnode *vp, struct thread *td);
591void	cvtstat(struct stat *st, struct ostat *ost);
592void	cvtnstat(struct stat *sb, struct nstat *nsb);
593int	getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
594	    struct vnode **vpp);
595int	insmntque1(struct vnode *vp, struct mount *mp,
596	    void (*dtr)(struct vnode *, void *), void *dtr_arg);
597int	insmntque(struct vnode *vp, struct mount *mp);
598u_quad_t init_va_filerev(void);
599int	speedup_syncer(void);
600int	vn_vptocnp(struct vnode **vp, struct ucred *cred, char *buf,
601	    u_int *buflen);
602#define textvp_fullpath(p, rb, rfb) \
603	vn_fullpath(FIRST_THREAD_IN_PROC(p), (p)->p_textvp, rb, rfb)
604int	vn_fullpath(struct thread *td, struct vnode *vn,
605	    char **retbuf, char **freebuf);
606int	vn_fullpath_global(struct thread *td, struct vnode *vn,
607	    char **retbuf, char **freebuf);
608int	vn_commname(struct vnode *vn, char *buf, u_int buflen);
609int	vaccess(enum vtype type, mode_t file_mode, uid_t file_uid,
610	    gid_t file_gid, accmode_t accmode, struct ucred *cred,
611	    int *privused);
612int	vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid,
613	    struct acl *aclp, accmode_t accmode, struct ucred *cred,
614	    int *privused);
615int	vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
616	    gid_t file_gid, struct acl *acl, accmode_t accmode,
617	    struct ucred *cred, int *privused);
618void	vattr_null(struct vattr *vap);
619int	vcount(struct vnode *vp);
620void	vdrop(struct vnode *);
621void	vdropl(struct vnode *);
622void	vdestroy(struct vnode *);
623int	vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
624int	vget(struct vnode *vp, int lockflag, struct thread *td);
625void	vgone(struct vnode *vp);
626void	vhold(struct vnode *);
627void	vholdl(struct vnode *);
628int	vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
629int	vtruncbuf(struct vnode *vp, struct ucred *cred, struct thread *td,
630	    off_t length, int blksize);
631void	vunref(struct vnode *);
632void	vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
633#define vprint(label, vp) vn_printf((vp), "%s\n", (label))
634int	vrecycle(struct vnode *vp, struct thread *td);
635int	vn_close(struct vnode *vp,
636	    int flags, struct ucred *file_cred, struct thread *td);
637void	vn_finished_write(struct mount *mp);
638void	vn_finished_secondary_write(struct mount *mp);
639int	vn_isdisk(struct vnode *vp, int *errp);
640int	_vn_lock(struct vnode *vp, int flags, char *file, int line);
641#define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
642int	vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
643int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
644	    u_int vn_open_flags, struct ucred *cred, struct file *fp);
645int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
646int	vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
647	    int len, off_t offset, enum uio_seg segflg, int ioflg,
648	    struct ucred *active_cred, struct ucred *file_cred, int *aresid,
649	    struct thread *td);
650int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
651	    size_t len, off_t offset, enum uio_seg segflg, int ioflg,
652	    struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
653	    struct thread *td);
654int	vn_rlimit_fsize(const struct vnode *vn, const struct uio *uio,
655	    const struct thread *td);
656int	vn_stat(struct vnode *vp, struct stat *sb, struct ucred *active_cred,
657	    struct ucred *file_cred, struct thread *td);
658int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
659int	vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
660	    int flags);
661int	vn_writechk(struct vnode *vp);
662int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
663	    const char *attrname, int *buflen, char *buf, struct thread *td);
664int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
665	    const char *attrname, int buflen, char *buf, struct thread *td);
666int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
667	    const char *attrname, struct thread *td);
668int	vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
669	    struct vnode **rvp);
670
671
672int	vfs_cache_lookup(struct vop_lookup_args *ap);
673void	vfs_timestamp(struct timespec *);
674void	vfs_write_resume(struct mount *mp);
675int	vfs_write_suspend(struct mount *mp);
676int	vop_stdbmap(struct vop_bmap_args *);
677int	vop_stdfsync(struct vop_fsync_args *);
678int	vop_stdgetwritemount(struct vop_getwritemount_args *);
679int	vop_stdgetpages(struct vop_getpages_args *);
680int	vop_stdinactive(struct vop_inactive_args *);
681int	vop_stdislocked(struct vop_islocked_args *);
682int	vop_stdkqfilter(struct vop_kqfilter_args *);
683int	vop_stdlock(struct vop_lock1_args *);
684int	vop_stdputpages(struct vop_putpages_args *);
685int	vop_stdunlock(struct vop_unlock_args *);
686int	vop_nopoll(struct vop_poll_args *);
687int	vop_stdaccess(struct vop_access_args *ap);
688int	vop_stdaccessx(struct vop_accessx_args *ap);
689int	vop_stdadvlock(struct vop_advlock_args *ap);
690int	vop_stdadvlockasync(struct vop_advlockasync_args *ap);
691int	vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
692int	vop_stdpathconf(struct vop_pathconf_args *);
693int	vop_stdpoll(struct vop_poll_args *);
694int	vop_stdvptocnp(struct vop_vptocnp_args *ap);
695int	vop_stdvptofh(struct vop_vptofh_args *ap);
696int	vop_eopnotsupp(struct vop_generic_args *ap);
697int	vop_ebadf(struct vop_generic_args *ap);
698int	vop_einval(struct vop_generic_args *ap);
699int	vop_enoent(struct vop_generic_args *ap);
700int	vop_enotty(struct vop_generic_args *ap);
701int	vop_null(struct vop_generic_args *ap);
702int	vop_panic(struct vop_generic_args *ap);
703
704/* These are called from within the actual VOPS. */
705void	vop_create_post(void *a, int rc);
706void	vop_link_post(void *a, int rc);
707void	vop_lock_pre(void *a);
708void	vop_lock_post(void *a, int rc);
709void	vop_lookup_post(void *a, int rc);
710void	vop_lookup_pre(void *a);
711void	vop_mkdir_post(void *a, int rc);
712void	vop_mknod_post(void *a, int rc);
713void	vop_remove_post(void *a, int rc);
714void	vop_rename_post(void *a, int rc);
715void	vop_rename_pre(void *a);
716void	vop_rmdir_post(void *a, int rc);
717void	vop_setattr_post(void *a, int rc);
718void	vop_strategy_pre(void *a);
719void	vop_symlink_post(void *a, int rc);
720void	vop_unlock_post(void *a, int rc);
721void	vop_unlock_pre(void *a);
722
723void	vop_rename_fail(struct vop_rename_args *ap);
724
725#define	VOP_WRITE_PRE(ap)						\
726	struct vattr va;						\
727	int error, osize, ooffset, noffset;				\
728									\
729	osize = ooffset = noffset = 0;					\
730	if (!VN_KNLIST_EMPTY((ap)->a_vp)) {				\
731		error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred);	\
732		if (error)						\
733			return (error);					\
734		ooffset = (ap)->a_uio->uio_offset;			\
735		osize = va.va_size;					\
736	}
737
738#define VOP_WRITE_POST(ap, ret)						\
739	noffset = (ap)->a_uio->uio_offset;				\
740	if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) {	\
741		VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE			\
742		    | (noffset > osize ? NOTE_EXTEND : 0));		\
743	}
744
745#define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
746
747
748void	vput(struct vnode *vp);
749void	vrele(struct vnode *vp);
750void	vref(struct vnode *vp);
751int	vrefcnt(struct vnode *vp);
752void 	v_addpollinfo(struct vnode *vp);
753
754int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
755void vnode_destroy_vobject(struct vnode *vp);
756
757extern struct vop_vector fifo_specops;
758extern struct vop_vector dead_vnodeops;
759extern struct vop_vector default_vnodeops;
760
761#define VOP_PANIC	((void*)(uintptr_t)vop_panic)
762#define VOP_NULL	((void*)(uintptr_t)vop_null)
763#define VOP_EBADF	((void*)(uintptr_t)vop_ebadf)
764#define VOP_ENOTTY	((void*)(uintptr_t)vop_enotty)
765#define VOP_EINVAL	((void*)(uintptr_t)vop_einval)
766#define VOP_ENOENT	((void*)(uintptr_t)vop_enoent)
767#define VOP_EOPNOTSUPP	((void*)(uintptr_t)vop_eopnotsupp)
768
769/* vfs_hash.c */
770typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
771
772int 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);
773int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
774void vfs_hash_rehash(struct vnode *vp, u_int hash);
775void vfs_hash_remove(struct vnode *vp);
776
777int vfs_kqfilter(struct vop_kqfilter_args *);
778void vfs_mark_atime(struct vnode *vp, struct ucred *cred);
779struct dirent;
780int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
781
782int	vfs_unixify_accmode(accmode_t *accmode);
783
784#endif /* _KERNEL */
785
786#endif /* !_SYS_VNODE_H_ */
787