vnode.h revision 234607
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 234607 2012-04-23 14:10:34Z trasz $
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 *	i - interlock
79 *	m - mount point interlock
80 *	p - pollinfo lock
81 *	u - Only a reference to the vnode is needed to read.
82 *	v - vnode lock
83 *
84 * Vnodes may be found on many lists.  The general way to deal with operating
85 * on a vnode that is on a list is:
86 *	1) Lock the list and find the vnode.
87 *	2) Lock interlock so that the vnode does not go away.
88 *	3) Unlock the list to avoid lock order reversals.
89 *	4) vget with LK_INTERLOCK and check for ENOENT, or
90 *	5) Check for DOOMED if the vnode lock is not required.
91 *	6) Perform your operation, then vput().
92 */
93
94#if defined(_KERNEL) || defined(_KVM_VNODE)
95
96struct vnode {
97	/*
98	 * Fields which define the identity of the vnode.  These fields are
99	 * owned by the filesystem (XXX: and vgone() ?)
100	 */
101	enum	vtype v_type;			/* u vnode type */
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.
125	 */
126	LIST_ENTRY(vnode)	v_hashlist;
127	u_int			v_hash;
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	 * clustering stuff
138	 */
139	daddr_t	v_cstart;			/* v start block of cluster */
140	daddr_t	v_lasta;			/* v last allocation  */
141	daddr_t	v_lastw;			/* v last write  */
142	int	v_clen;				/* v length of cur. cluster */
143
144	/*
145	 * Locking
146	 */
147	struct	lock v_lock;			/* u (if fs don't have one) */
148	struct	mtx v_interlock;		/* lock for "i" things */
149	struct	lock *v_vnlock;			/* u pointer to vnode lock */
150	int	v_holdcnt;			/* i prevents recycling. */
151	int	v_usecount;			/* i ref count of users */
152	u_int	v_iflag;			/* i vnode flags (see below) */
153	u_int	v_vflag;			/* v vnode flags */
154	int	v_writecount;			/* v ref count of writers */
155
156	/*
157	 * The machinery of being a vnode
158	 */
159	TAILQ_ENTRY(vnode) v_actfreelist;	/* f vnode active/free lists */
160	struct bufobj	v_bufobj;		/* * Buffer cache object */
161
162	/*
163	 * Hooks for various subsystems and features.
164	 */
165	struct vpollinfo *v_pollinfo;		/* i Poll events, p for *v_pi */
166	struct label *v_label;			/* MAC label for vnode */
167	struct lockf *v_lockf;			/* Byte-level lock list */
168};
169
170#endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
171
172#define	v_mountedhere	v_un.vu_mount
173#define	v_socket	v_un.vu_socket
174#define	v_rdev		v_un.vu_cdev
175#define	v_fifoinfo	v_un.vu_fifoinfo
176
177/* XXX: These are temporary to avoid a source sweep at this time */
178#define v_object	v_bufobj.bo_object
179
180/*
181 * Userland version of struct vnode, for sysctl.
182 */
183struct xvnode {
184	size_t	xv_size;			/* sizeof(struct xvnode) */
185	void	*xv_vnode;			/* address of real vnode */
186	u_long	xv_flag;			/* vnode vflags */
187	int	xv_usecount;			/* reference count of users */
188	int	xv_writecount;			/* reference count of writers */
189	int	xv_holdcnt;			/* page & buffer references */
190	u_long	xv_id;				/* capability identifier */
191	void	*xv_mount;			/* address of parent mount */
192	long	xv_numoutput;			/* num of writes in progress */
193	enum	vtype xv_type;			/* vnode type */
194	union {
195		void	*xvu_socket;		/* socket, if VSOCK */
196		void	*xvu_fifo;		/* fifo, if VFIFO */
197		dev_t	xvu_rdev;		/* maj/min, if VBLK/VCHR */
198		struct {
199			dev_t	xvu_dev;	/* device, if VDIR/VREG/VLNK */
200			ino_t	xvu_ino;	/* id, if VDIR/VREG/VLNK */
201		} xv_uns;
202	} xv_un;
203};
204#define xv_socket	xv_un.xvu_socket
205#define xv_fifo		xv_un.xvu_fifo
206#define xv_rdev		xv_un.xvu_rdev
207#define xv_dev		xv_un.xv_uns.xvu_dev
208#define xv_ino		xv_un.xv_uns.xvu_ino
209
210/* We don't need to lock the knlist */
211#define	VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL ||	\
212	    KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
213
214#define VN_KNOTE(vp, b, a)					\
215	do {							\
216		if (!VN_KNLIST_EMPTY(vp))			\
217			KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
218			    (a) | KNF_NOKQLOCK);		\
219	} while (0)
220#define	VN_KNOTE_LOCKED(vp, b)		VN_KNOTE(vp, b, KNF_LISTLOCKED)
221#define	VN_KNOTE_UNLOCKED(vp, b)	VN_KNOTE(vp, b, 0)
222
223/*
224 * Vnode flags.
225 *	VI flags are protected by interlock and live in v_iflag
226 *	VV flags are protected by the vnode lock and live in v_vflag
227 *
228 *	VI_DOOMED is doubly protected by the interlock and vnode lock.  Both
229 *	are required for writing but the status may be checked with either.
230 */
231#define	VI_MOUNT	0x0020	/* Mount in progress */
232#define	VI_AGE		0x0040	/* Insert vnode at head of free list */
233#define	VI_DOOMED	0x0080	/* This vnode is being recycled */
234#define	VI_FREE		0x0100	/* This vnode is on the freelist */
235#define	VI_ACTIVE	0x0200	/* This vnode is on the active list */
236#define	VI_DOINGINACT	0x0800	/* VOP_INACTIVE is in progress */
237#define	VI_OWEINACT	0x1000	/* Need to call inactive */
238
239#define	VV_ROOT		0x0001	/* root of its filesystem */
240#define	VV_ISTTY	0x0002	/* vnode represents a tty */
241#define	VV_NOSYNC	0x0004	/* unlinked, stop syncing */
242#define	VV_ETERNALDEV	0x0008	/* device that is never destroyed */
243#define	VV_CACHEDLABEL	0x0010	/* Vnode has valid cached MAC label */
244#define	VV_TEXT		0x0020	/* vnode is a pure text prototype */
245#define	VV_COPYONWRITE	0x0040	/* vnode is doing copy-on-write */
246#define	VV_SYSTEM	0x0080	/* vnode being used by kernel */
247#define	VV_PROCDEP	0x0100	/* vnode is process dependent */
248#define	VV_NOKNOTE	0x0200	/* don't activate knotes on this vnode */
249#define	VV_DELETED	0x0400	/* should be removed */
250#define	VV_MD		0x0800	/* vnode backs the md device */
251#define	VV_FORCEINSMQ	0x1000	/* force the insmntque to succeed */
252
253/*
254 * Vnode attributes.  A field value of VNOVAL represents a field whose value
255 * is unavailable (getattr) or which is not to be changed (setattr).
256 */
257struct vattr {
258	enum vtype	va_type;	/* vnode type (for create) */
259	u_short		va_mode;	/* files access mode and type */
260	short		va_nlink;	/* number of references to file */
261	uid_t		va_uid;		/* owner user id */
262	gid_t		va_gid;		/* owner group id */
263	dev_t		va_fsid;	/* filesystem id */
264	long		va_fileid;	/* file id */
265	u_quad_t	va_size;	/* file size in bytes */
266	long		va_blocksize;	/* blocksize preferred for i/o */
267	struct timespec	va_atime;	/* time of last access */
268	struct timespec	va_mtime;	/* time of last modification */
269	struct timespec	va_ctime;	/* time file changed */
270	struct timespec	va_birthtime;	/* time file created */
271	u_long		va_gen;		/* generation number of file */
272	u_long		va_flags;	/* flags defined for file */
273	dev_t		va_rdev;	/* device the special file represents */
274	u_quad_t	va_bytes;	/* bytes of disk space held by file */
275	u_quad_t	va_filerev;	/* file modification number */
276	u_int		va_vaflags;	/* operations flags, see below */
277	long		va_spare;	/* remain quad aligned */
278};
279
280/*
281 * Flags for va_vaflags.
282 */
283#define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
284#define	VA_EXCLUSIVE	0x02		/* exclusive create request */
285
286/*
287 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
288 * help with write clustering)
289 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
290 */
291#define	IO_UNIT		0x0001		/* do I/O as atomic unit */
292#define	IO_APPEND	0x0002		/* append write to end */
293#define	IO_NDELAY	0x0004		/* FNDELAY flag set in file table */
294#define	IO_NODELOCKED	0x0008		/* underlying node already locked */
295#define	IO_ASYNC	0x0010		/* bawrite rather then bdwrite */
296#define	IO_VMIO		0x0020		/* data already in VMIO space */
297#define	IO_INVAL	0x0040		/* invalidate after I/O */
298#define	IO_SYNC		0x0080		/* do I/O synchronously */
299#define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
300#define	IO_EXT		0x0400		/* operate on external attributes */
301#define	IO_NORMAL	0x0800		/* operate on regular data */
302#define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
303#define	IO_BUFLOCKED	0x2000		/* ffs flag; indir buf is locked */
304
305#define IO_SEQMAX	0x7F		/* seq heuristic max value */
306#define IO_SEQSHIFT	16		/* seq heuristic in upper 16 bits */
307
308/*
309 * Flags for accmode_t.
310 */
311#define	VEXEC			000000000100 /* execute/search permission */
312#define	VWRITE			000000000200 /* write permission */
313#define	VREAD			000000000400 /* read permission */
314#define	VADMIN			000000010000 /* being the file owner */
315#define	VAPPEND			000000040000 /* permission to write/append */
316/*
317 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
318 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
319 * and 0 otherwise.  This never happens with ordinary unix access rights
320 * or POSIX.1e ACLs.  Obviously, VEXPLICIT_DENY must be OR-ed with
321 * some other V* constant.
322 */
323#define	VEXPLICIT_DENY		000000100000
324#define	VREAD_NAMED_ATTRS 	000000200000 /* not used */
325#define	VWRITE_NAMED_ATTRS 	000000400000 /* not used */
326#define	VDELETE_CHILD	 	000001000000
327#define	VREAD_ATTRIBUTES 	000002000000 /* permission to stat(2) */
328#define	VWRITE_ATTRIBUTES 	000004000000 /* change {m,c,a}time */
329#define	VDELETE		 	000010000000
330#define	VREAD_ACL	 	000020000000 /* read ACL and file mode */
331#define	VWRITE_ACL	 	000040000000 /* change ACL and/or file mode */
332#define	VWRITE_OWNER	 	000100000000 /* change file owner */
333#define	VSYNCHRONIZE	 	000200000000 /* not used */
334
335/*
336 * Permissions that were traditionally granted only to the file owner.
337 */
338#define VADMIN_PERMS	(VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
339    VWRITE_OWNER)
340
341/*
342 * Permissions that were traditionally granted to everyone.
343 */
344#define VSTAT_PERMS	(VREAD_ATTRIBUTES | VREAD_ACL)
345
346/*
347 * Permissions that allow to change the state of the file in any way.
348 */
349#define VMODIFY_PERMS	(VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
350    VDELETE)
351
352/*
353 * Token indicating no attribute value yet assigned.
354 */
355#define	VNOVAL	(-1)
356
357/*
358 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
359 */
360#define VLKTIMEOUT	(hz / 20 + 1)
361
362#ifdef _KERNEL
363
364#ifdef MALLOC_DECLARE
365MALLOC_DECLARE(M_VNODE);
366#endif
367
368/*
369 * Convert between vnode types and inode formats (since POSIX.1
370 * defines mode word of stat structure in terms of inode formats).
371 */
372extern enum vtype	iftovt_tab[];
373extern int		vttoif_tab[];
374#define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
375#define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
376#define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
377
378/*
379 * Flags to various vnode functions.
380 */
381#define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
382#define	FORCECLOSE	0x0002	/* vflush: force file closure */
383#define	WRITECLOSE	0x0004	/* vflush: only close writable files */
384#define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
385#define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
386#define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
387#define	V_CLEANONLY	0x0008	/* vinvalbuf: invalidate only clean bufs */
388#define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
389#define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
390#define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
391#define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
392
393#define	VREF(vp)	vref(vp)
394
395#ifdef DIAGNOSTIC
396#define	VATTR_NULL(vap)	vattr_null(vap)
397#else
398#define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
399#endif /* DIAGNOSTIC */
400
401#define	NULLVP	((struct vnode *)NULL)
402
403/*
404 * Global vnode data.
405 */
406extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
407extern	int async_io_version;		/* 0 or POSIX version of AIO i'face */
408extern	int desiredvnodes;		/* number of vnodes desired */
409extern	struct uma_zone *namei_zone;
410extern	struct vattr va_null;		/* predefined null vattr structure */
411
412#define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
413#define	VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
414#define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
415#define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
416#define	VI_MTX(vp)	(&(vp)->v_interlock)
417
418#define	VN_LOCK_AREC(vp)	lockallowrecurse((vp)->v_vnlock)
419#define	VN_LOCK_ASHARE(vp)	lockallowshare((vp)->v_vnlock)
420
421#endif /* _KERNEL */
422
423/*
424 * Mods for extensibility.
425 */
426
427/*
428 * Flags for vdesc_flags:
429 */
430#define	VDESC_MAX_VPS		16
431/* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
432#define	VDESC_VP0_WILLRELE	0x0001
433#define	VDESC_VP1_WILLRELE	0x0002
434#define	VDESC_VP2_WILLRELE	0x0004
435#define	VDESC_VP3_WILLRELE	0x0008
436#define	VDESC_NOMAP_VPP		0x0100
437#define	VDESC_VPP_WILLRELE	0x0200
438
439/*
440 * A generic structure.
441 * This can be used by bypass routines to identify generic arguments.
442 */
443struct vop_generic_args {
444	struct vnodeop_desc *a_desc;
445	/* other random data follows, presumably */
446};
447
448typedef int vop_bypass_t(struct vop_generic_args *);
449
450/*
451 * VDESC_NO_OFFSET is used to identify the end of the offset list
452 * and in places where no such field exists.
453 */
454#define VDESC_NO_OFFSET -1
455
456/*
457 * This structure describes the vnode operation taking place.
458 */
459struct vnodeop_desc {
460	char	*vdesc_name;		/* a readable name for debugging */
461	int	 vdesc_flags;		/* VDESC_* flags */
462	vop_bypass_t	*vdesc_call;	/* Function to call */
463
464	/*
465	 * These ops are used by bypass routines to map and locate arguments.
466	 * Creds and procs are not needed in bypass routines, but sometimes
467	 * they are useful to (for example) transport layers.
468	 * Nameidata is useful because it has a cred in it.
469	 */
470	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
471	int	vdesc_vpp_offset;	/* return vpp location */
472	int	vdesc_cred_offset;	/* cred location, if any */
473	int	vdesc_thread_offset;	/* thread location, if any */
474	int	vdesc_componentname_offset; /* if any */
475};
476
477#ifdef _KERNEL
478/*
479 * A list of all the operation descs.
480 */
481extern struct vnodeop_desc *vnodeop_descs[];
482
483#define	VOPARG_OFFSETOF(s_type, field)	__offsetof(s_type, field)
484#define	VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
485    ((s_type)(((char*)(struct_p)) + (s_offset)))
486
487
488#ifdef DEBUG_VFS_LOCKS
489/*
490 * Support code to aid in debugging VFS locking problems.  Not totally
491 * reliable since if the thread sleeps between changing the lock
492 * state and checking it with the assert, some other thread could
493 * change the state.  They are good enough for debugging a single
494 * filesystem using a single-threaded test.
495 */
496void	assert_vi_locked(struct vnode *vp, const char *str);
497void	assert_vi_unlocked(struct vnode *vp, const char *str);
498void	assert_vop_elocked(struct vnode *vp, const char *str);
499#if 0
500void	assert_vop_elocked_other(struct vnode *vp, const char *str);
501#endif
502void	assert_vop_locked(struct vnode *vp, const char *str);
503#if 0
504voi0	assert_vop_slocked(struct vnode *vp, const char *str);
505#endif
506void	assert_vop_unlocked(struct vnode *vp, const char *str);
507
508#define	ASSERT_VI_LOCKED(vp, str)	assert_vi_locked((vp), (str))
509#define	ASSERT_VI_UNLOCKED(vp, str)	assert_vi_unlocked((vp), (str))
510#define	ASSERT_VOP_ELOCKED(vp, str)	assert_vop_elocked((vp), (str))
511#if 0
512#define	ASSERT_VOP_ELOCKED_OTHER(vp, str) assert_vop_locked_other((vp), (str))
513#endif
514#define	ASSERT_VOP_LOCKED(vp, str)	assert_vop_locked((vp), (str))
515#if 0
516#define	ASSERT_VOP_SLOCKED(vp, str)	assert_vop_slocked((vp), (str))
517#endif
518#define	ASSERT_VOP_UNLOCKED(vp, str)	assert_vop_unlocked((vp), (str))
519
520#else /* !DEBUG_VFS_LOCKS */
521
522#define	ASSERT_VI_LOCKED(vp, str)	((void)0)
523#define	ASSERT_VI_UNLOCKED(vp, str)	((void)0)
524#define	ASSERT_VOP_ELOCKED(vp, str)	((void)0)
525#if 0
526#define	ASSERT_VOP_ELOCKED_OTHER(vp, str)
527#endif
528#define	ASSERT_VOP_LOCKED(vp, str)	((void)0)
529#if 0
530#define	ASSERT_VOP_SLOCKED(vp, str)
531#endif
532#define	ASSERT_VOP_UNLOCKED(vp, str)	((void)0)
533#endif /* DEBUG_VFS_LOCKS */
534
535
536/*
537 * This call works for vnodes in the kernel.
538 */
539#define VCALL(c) ((c)->a_desc->vdesc_call(c))
540
541#define DOINGASYNC(vp)	   					\
542	(((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 &&	\
543	 ((curthread->td_pflags & TDP_SYNCIO) == 0))
544
545/*
546 * VMIO support inline
547 */
548
549extern int vmiodirenable;
550
551static __inline int
552vn_canvmio(struct vnode *vp)
553{
554      if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
555		return(TRUE);
556	return(FALSE);
557}
558
559/*
560 * Finally, include the default set of vnode operations.
561 */
562#include "vnode_if.h"
563
564/* vn_open_flags */
565#define	VN_OPEN_NOAUDIT		0x00000001
566
567/*
568 * Public vnode manipulation functions.
569 */
570struct componentname;
571struct file;
572struct mount;
573struct nameidata;
574struct ostat;
575struct thread;
576struct proc;
577struct stat;
578struct nstat;
579struct ucred;
580struct uio;
581struct vattr;
582struct vnode;
583
584/* cache_* may belong in namei.h. */
585#define	cache_enter(dvp, vp, cnp)					\
586	cache_enter_time(dvp, vp, cnp, NULL, NULL)
587void	cache_enter_time(struct vnode *dvp, struct vnode *vp,
588	    struct componentname *cnp, struct timespec *tsp,
589	    struct timespec *dtsp);
590int	cache_lookup(struct vnode *dvp, struct vnode **vpp,
591	    struct componentname *cnp, struct timespec *tsp, int *ticksp);
592void	cache_purge(struct vnode *vp);
593void	cache_purge_negative(struct vnode *vp);
594void	cache_purgevfs(struct mount *mp);
595int	change_dir(struct vnode *vp, struct thread *td);
596int	change_root(struct vnode *vp, struct thread *td);
597void	cvtstat(struct stat *st, struct ostat *ost);
598void	cvtnstat(struct stat *sb, struct nstat *nsb);
599int	getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
600	    struct vnode **vpp);
601int	insmntque1(struct vnode *vp, struct mount *mp,
602	    void (*dtr)(struct vnode *, void *), void *dtr_arg);
603int	insmntque(struct vnode *vp, struct mount *mp);
604u_quad_t init_va_filerev(void);
605int	speedup_syncer(void);
606int	vn_vptocnp(struct vnode **vp, struct ucred *cred, char *buf,
607	    u_int *buflen);
608#define textvp_fullpath(p, rb, rfb) \
609	vn_fullpath(FIRST_THREAD_IN_PROC(p), (p)->p_textvp, rb, rfb)
610int	vn_fullpath(struct thread *td, struct vnode *vn,
611	    char **retbuf, char **freebuf);
612int	vn_fullpath_global(struct thread *td, struct vnode *vn,
613	    char **retbuf, char **freebuf);
614int	vn_commname(struct vnode *vn, char *buf, u_int buflen);
615int	vn_path_to_global_path(struct thread *td, struct vnode *vp,
616	    char *path, u_int pathlen);
617int	vaccess(enum vtype type, mode_t file_mode, uid_t file_uid,
618	    gid_t file_gid, accmode_t accmode, struct ucred *cred,
619	    int *privused);
620int	vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid,
621	    struct acl *aclp, accmode_t accmode, struct ucred *cred,
622	    int *privused);
623int	vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
624	    gid_t file_gid, struct acl *acl, accmode_t accmode,
625	    struct ucred *cred, int *privused);
626void	vattr_null(struct vattr *vap);
627int	vcount(struct vnode *vp);
628void	vdrop(struct vnode *);
629void	vdropl(struct vnode *);
630int	vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
631int	vget(struct vnode *vp, int lockflag, struct thread *td);
632void	vgone(struct vnode *vp);
633void	vhold(struct vnode *);
634void	vholdl(struct vnode *);
635void	vinactive(struct vnode *, struct thread *);
636int	vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
637int	vtruncbuf(struct vnode *vp, struct ucred *cred, off_t length,
638	    int blksize);
639void	vunref(struct vnode *);
640void	vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
641#define vprint(label, vp) vn_printf((vp), "%s\n", (label))
642int	vrecycle(struct vnode *vp);
643int	vn_close(struct vnode *vp,
644	    int flags, struct ucred *file_cred, struct thread *td);
645void	vn_finished_write(struct mount *mp);
646void	vn_finished_secondary_write(struct mount *mp);
647int	vn_isdisk(struct vnode *vp, int *errp);
648int	_vn_lock(struct vnode *vp, int flags, char *file, int line);
649#define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
650int	vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
651int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
652	    u_int vn_open_flags, struct ucred *cred, struct file *fp);
653void	vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
654int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
655int	vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
656	    int len, off_t offset, enum uio_seg segflg, int ioflg,
657	    struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
658	    struct thread *td);
659int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
660	    size_t len, off_t offset, enum uio_seg segflg, int ioflg,
661	    struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
662	    struct thread *td);
663int	vn_rlimit_fsize(const struct vnode *vn, const struct uio *uio,
664	    const struct thread *td);
665int	vn_stat(struct vnode *vp, struct stat *sb, struct ucred *active_cred,
666	    struct ucred *file_cred, struct thread *td);
667int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
668int	vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
669	    int flags);
670int	vn_writechk(struct vnode *vp);
671int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
672	    const char *attrname, int *buflen, char *buf, struct thread *td);
673int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
674	    const char *attrname, int buflen, char *buf, struct thread *td);
675int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
676	    const char *attrname, struct thread *td);
677int	vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
678	    struct vnode **rvp);
679
680
681int	vfs_cache_lookup(struct vop_lookup_args *ap);
682void	vfs_timestamp(struct timespec *);
683void	vfs_write_resume(struct mount *mp);
684int	vfs_write_suspend(struct mount *mp);
685int	vop_stdbmap(struct vop_bmap_args *);
686int	vop_stdfsync(struct vop_fsync_args *);
687int	vop_stdgetwritemount(struct vop_getwritemount_args *);
688int	vop_stdgetpages(struct vop_getpages_args *);
689int	vop_stdinactive(struct vop_inactive_args *);
690int	vop_stdislocked(struct vop_islocked_args *);
691int	vop_stdkqfilter(struct vop_kqfilter_args *);
692int	vop_stdlock(struct vop_lock1_args *);
693int	vop_stdputpages(struct vop_putpages_args *);
694int	vop_stdunlock(struct vop_unlock_args *);
695int	vop_nopoll(struct vop_poll_args *);
696int	vop_stdaccess(struct vop_access_args *ap);
697int	vop_stdaccessx(struct vop_accessx_args *ap);
698int	vop_stdadvise(struct vop_advise_args *ap);
699int	vop_stdadvlock(struct vop_advlock_args *ap);
700int	vop_stdadvlockasync(struct vop_advlockasync_args *ap);
701int	vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
702int	vop_stdallocate(struct vop_allocate_args *ap);
703int	vop_stdpathconf(struct vop_pathconf_args *);
704int	vop_stdpoll(struct vop_poll_args *);
705int	vop_stdvptocnp(struct vop_vptocnp_args *ap);
706int	vop_stdvptofh(struct vop_vptofh_args *ap);
707int	vop_stdunp_bind(struct vop_unp_bind_args *ap);
708int	vop_stdunp_connect(struct vop_unp_connect_args *ap);
709int	vop_stdunp_detach(struct vop_unp_detach_args *ap);
710int	vop_eopnotsupp(struct vop_generic_args *ap);
711int	vop_ebadf(struct vop_generic_args *ap);
712int	vop_einval(struct vop_generic_args *ap);
713int	vop_enoent(struct vop_generic_args *ap);
714int	vop_enotty(struct vop_generic_args *ap);
715int	vop_null(struct vop_generic_args *ap);
716int	vop_panic(struct vop_generic_args *ap);
717
718/* These are called from within the actual VOPS. */
719void	vop_create_post(void *a, int rc);
720void	vop_deleteextattr_post(void *a, int rc);
721void	vop_link_post(void *a, int rc);
722void	vop_lock_pre(void *a);
723void	vop_lock_post(void *a, int rc);
724void	vop_lookup_post(void *a, int rc);
725void	vop_lookup_pre(void *a);
726void	vop_mkdir_post(void *a, int rc);
727void	vop_mknod_post(void *a, int rc);
728void	vop_remove_post(void *a, int rc);
729void	vop_rename_post(void *a, int rc);
730void	vop_rename_pre(void *a);
731void	vop_rmdir_post(void *a, int rc);
732void	vop_setattr_post(void *a, int rc);
733void	vop_setextattr_post(void *a, int rc);
734void	vop_strategy_pre(void *a);
735void	vop_symlink_post(void *a, int rc);
736void	vop_unlock_post(void *a, int rc);
737void	vop_unlock_pre(void *a);
738
739void	vop_rename_fail(struct vop_rename_args *ap);
740
741#define	VOP_WRITE_PRE(ap)						\
742	struct vattr va;						\
743	int error, osize, ooffset, noffset;				\
744									\
745	osize = ooffset = noffset = 0;					\
746	if (!VN_KNLIST_EMPTY((ap)->a_vp)) {				\
747		error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred);	\
748		if (error)						\
749			return (error);					\
750		ooffset = (ap)->a_uio->uio_offset;			\
751		osize = va.va_size;					\
752	}
753
754#define VOP_WRITE_POST(ap, ret)						\
755	noffset = (ap)->a_uio->uio_offset;				\
756	if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) {	\
757		VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE			\
758		    | (noffset > osize ? NOTE_EXTEND : 0));		\
759	}
760
761#define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
762
763
764void	vput(struct vnode *vp);
765void	vrele(struct vnode *vp);
766void	vref(struct vnode *vp);
767int	vrefcnt(struct vnode *vp);
768void 	v_addpollinfo(struct vnode *vp);
769
770int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
771void vnode_destroy_vobject(struct vnode *vp);
772
773extern struct vop_vector fifo_specops;
774extern struct vop_vector dead_vnodeops;
775extern struct vop_vector default_vnodeops;
776
777#define VOP_PANIC	((void*)(uintptr_t)vop_panic)
778#define VOP_NULL	((void*)(uintptr_t)vop_null)
779#define VOP_EBADF	((void*)(uintptr_t)vop_ebadf)
780#define VOP_ENOTTY	((void*)(uintptr_t)vop_enotty)
781#define VOP_EINVAL	((void*)(uintptr_t)vop_einval)
782#define VOP_ENOENT	((void*)(uintptr_t)vop_enoent)
783#define VOP_EOPNOTSUPP	((void*)(uintptr_t)vop_eopnotsupp)
784
785/* fifo_vnops.c */
786int	fifo_printinfo(struct vnode *);
787
788/* vfs_hash.c */
789typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
790
791int 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);
792int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
793void vfs_hash_rehash(struct vnode *vp, u_int hash);
794void vfs_hash_remove(struct vnode *vp);
795
796int vfs_kqfilter(struct vop_kqfilter_args *);
797void vfs_mark_atime(struct vnode *vp, struct ucred *cred);
798struct dirent;
799int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
800
801int vfs_unixify_accmode(accmode_t *accmode);
802
803void vfs_unp_reclaim(struct vnode *vp);
804
805int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
806int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
807    gid_t gid);
808int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
809    struct thread *td);
810int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
811    struct thread *td);
812
813#endif /* _KERNEL */
814
815#endif /* !_SYS_VNODE_H_ */
816