vnode.h revision 331722
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: stable/11/sys/sys/vnode.h 331722 2018-03-29 02:50:57Z eadler $
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 *	I - updated with atomics, 0->1 and 1->0 transitions with interlock held
81 *	m - mount point interlock
82 *	p - pollinfo lock
83 *	u - Only a reference to the vnode is needed to read.
84 *	v - vnode lock
85 *
86 * Vnodes may be found on many lists.  The general way to deal with operating
87 * on a vnode that is on a list is:
88 *	1) Lock the list and find the vnode.
89 *	2) Lock interlock so that the vnode does not go away.
90 *	3) Unlock the list to avoid lock order reversals.
91 *	4) vget with LK_INTERLOCK and check for ENOENT, or
92 *	5) Check for DOOMED if the vnode lock is not required.
93 *	6) Perform your operation, then vput().
94 */
95
96#if defined(_KERNEL) || defined(_KVM_VNODE)
97
98struct vnode {
99	/*
100	 * Fields which define the identity of the vnode.  These fields are
101	 * owned by the filesystem (XXX: and vgone() ?)
102	 */
103	const char *v_tag;			/* u type of underlying data */
104	struct	vop_vector *v_op;		/* u vnode operations vector */
105	void	*v_data;			/* u private data for fs */
106
107	/*
108	 * Filesystem instance stuff
109	 */
110	struct	mount *v_mount;			/* u ptr to vfs we are in */
111	TAILQ_ENTRY(vnode) v_nmntvnodes;	/* m vnodes for mount point */
112
113	/*
114	 * Type specific fields, only one applies to any given vnode.
115	 * See #defines below for renaming to v_* namespace.
116	 */
117	union {
118		struct mount	*vu_mount;	/* v ptr to mountpoint (VDIR) */
119		struct socket	*vu_socket;	/* v unix domain net (VSOCK) */
120		struct cdev	*vu_cdev; 	/* v device (VCHR, VBLK) */
121		struct fifoinfo	*vu_fifoinfo;	/* v fifo (VFIFO) */
122	} v_un;
123
124	/*
125	 * vfs_hash: (mount + inode) -> vnode hash.  The hash value
126	 * itself is grouped with other int fields, to avoid padding.
127	 */
128	LIST_ENTRY(vnode)	v_hashlist;
129
130	/*
131	 * VFS_namecache stuff
132	 */
133	LIST_HEAD(, namecache) v_cache_src;	/* c Cache entries from us */
134	TAILQ_HEAD(, namecache) v_cache_dst;	/* c Cache entries to us */
135	struct namecache *v_cache_dd;		/* c Cache entry for .. vnode */
136
137	/*
138	 * Locking
139	 */
140	struct	lock v_lock;			/* u (if fs don't have one) */
141	struct	mtx v_interlock;		/* lock for "i" things */
142	struct	lock *v_vnlock;			/* u pointer to vnode lock */
143
144	/*
145	 * The machinery of being a vnode
146	 */
147	TAILQ_ENTRY(vnode) v_actfreelist;	/* f vnode active/free lists */
148	struct bufobj	v_bufobj;		/* * Buffer cache object */
149
150	/*
151	 * Hooks for various subsystems and features.
152	 */
153	struct vpollinfo *v_pollinfo;		/* i Poll events, p for *v_pi */
154	struct label *v_label;			/* MAC label for vnode */
155	struct lockf *v_lockf;		/* Byte-level advisory lock list */
156	struct rangelock v_rl;			/* Byte-range lock */
157
158	/*
159	 * clustering stuff
160	 */
161	daddr_t	v_cstart;			/* v start block of cluster */
162	daddr_t	v_lasta;			/* v last allocation  */
163	daddr_t	v_lastw;			/* v last write  */
164	int	v_clen;				/* v length of cur. cluster */
165
166	u_int	v_holdcnt;			/* I prevents recycling. */
167	u_int	v_usecount;			/* I ref count of users */
168	u_int	v_iflag;			/* i vnode flags (see below) */
169	u_int	v_vflag;			/* v vnode flags */
170	int	v_writecount;			/* v ref count of writers */
171	u_int	v_hash;
172	enum	vtype v_type;			/* u vnode type */
173};
174
175#endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
176
177#define	v_mountedhere	v_un.vu_mount
178#define	v_socket	v_un.vu_socket
179#define	v_rdev		v_un.vu_cdev
180#define	v_fifoinfo	v_un.vu_fifoinfo
181
182/* XXX: These are temporary to avoid a source sweep at this time */
183#define v_object	v_bufobj.bo_object
184
185/*
186 * Userland version of struct vnode, for sysctl.
187 */
188struct xvnode {
189	size_t	xv_size;			/* sizeof(struct xvnode) */
190	void	*xv_vnode;			/* address of real vnode */
191	u_long	xv_flag;			/* vnode vflags */
192	int	xv_usecount;			/* reference count of users */
193	int	xv_writecount;			/* reference count of writers */
194	int	xv_holdcnt;			/* page & buffer references */
195	u_long	xv_id;				/* capability identifier */
196	void	*xv_mount;			/* address of parent mount */
197	long	xv_numoutput;			/* num of writes in progress */
198	enum	vtype xv_type;			/* vnode type */
199	union {
200		void	*xvu_socket;		/* socket, if VSOCK */
201		void	*xvu_fifo;		/* fifo, if VFIFO */
202		dev_t	xvu_rdev;		/* maj/min, if VBLK/VCHR */
203		struct {
204			dev_t	xvu_dev;	/* device, if VDIR/VREG/VLNK */
205			ino_t	xvu_ino;	/* id, if VDIR/VREG/VLNK */
206		} xv_uns;
207	} xv_un;
208};
209#define xv_socket	xv_un.xvu_socket
210#define xv_fifo		xv_un.xvu_fifo
211#define xv_rdev		xv_un.xvu_rdev
212#define xv_dev		xv_un.xv_uns.xvu_dev
213#define xv_ino		xv_un.xv_uns.xvu_ino
214
215/* We don't need to lock the knlist */
216#define	VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL ||	\
217	    KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
218
219#define VN_KNOTE(vp, b, a)					\
220	do {							\
221		if (!VN_KNLIST_EMPTY(vp))			\
222			KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
223			    (a) | KNF_NOKQLOCK);		\
224	} while (0)
225#define	VN_KNOTE_LOCKED(vp, b)		VN_KNOTE(vp, b, KNF_LISTLOCKED)
226#define	VN_KNOTE_UNLOCKED(vp, b)	VN_KNOTE(vp, b, 0)
227
228/*
229 * Vnode flags.
230 *	VI flags are protected by interlock and live in v_iflag
231 *	VV flags are protected by the vnode lock and live in v_vflag
232 *
233 *	VI_DOOMED is doubly protected by the interlock and vnode lock.  Both
234 *	are required for writing but the status may be checked with either.
235 */
236#define	VI_MOUNT	0x0020	/* Mount in progress */
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#define	VV_READLINK	0x2000	/* fdescfs linux vnode */
257
258/*
259 * Vnode attributes.  A field value of VNOVAL represents a field whose value
260 * is unavailable (getattr) or which is not to be changed (setattr).
261 */
262struct vattr {
263	enum vtype	va_type;	/* vnode type (for create) */
264	u_short		va_mode;	/* files access mode and type */
265	short		va_nlink;	/* number of references to file */
266	uid_t		va_uid;		/* owner user id */
267	gid_t		va_gid;		/* owner group id */
268	dev_t		va_fsid;	/* filesystem id */
269	long		va_fileid;	/* file id */
270	u_quad_t	va_size;	/* file size in bytes */
271	long		va_blocksize;	/* blocksize preferred for i/o */
272	struct timespec	va_atime;	/* time of last access */
273	struct timespec	va_mtime;	/* time of last modification */
274	struct timespec	va_ctime;	/* time file changed */
275	struct timespec	va_birthtime;	/* time file created */
276	u_long		va_gen;		/* generation number of file */
277	u_long		va_flags;	/* flags defined for file */
278	dev_t		va_rdev;	/* device the special file represents */
279	u_quad_t	va_bytes;	/* bytes of disk space held by file */
280	u_quad_t	va_filerev;	/* file modification number */
281	u_int		va_vaflags;	/* operations flags, see below */
282	long		va_spare;	/* remain quad aligned */
283};
284
285/*
286 * Flags for va_vaflags.
287 */
288#define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
289#define	VA_EXCLUSIVE	0x02		/* exclusive create request */
290#define	VA_SYNC		0x04		/* O_SYNC truncation */
291
292/*
293 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
294 * help with write clustering)
295 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
296 */
297#define	IO_UNIT		0x0001		/* do I/O as atomic unit */
298#define	IO_APPEND	0x0002		/* append write to end */
299#define	IO_NDELAY	0x0004		/* FNDELAY flag set in file table */
300#define	IO_NODELOCKED	0x0008		/* underlying node already locked */
301#define	IO_ASYNC	0x0010		/* bawrite rather then bdwrite */
302#define	IO_VMIO		0x0020		/* data already in VMIO space */
303#define	IO_INVAL	0x0040		/* invalidate after I/O */
304#define	IO_SYNC		0x0080		/* do I/O synchronously */
305#define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
306#define	IO_NOREUSE	0x0200		/* VMIO data won't be reused */
307#define	IO_EXT		0x0400		/* operate on external attributes */
308#define	IO_NORMAL	0x0800		/* operate on regular data */
309#define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
310#define	IO_BUFLOCKED	0x2000		/* ffs flag; indir buf is locked */
311#define	IO_RANGELOCKED	0x4000		/* range locked */
312
313#define IO_SEQMAX	0x7F		/* seq heuristic max value */
314#define IO_SEQSHIFT	16		/* seq heuristic in upper 16 bits */
315
316/*
317 * Flags for accmode_t.
318 */
319#define	VEXEC			000000000100 /* execute/search permission */
320#define	VWRITE			000000000200 /* write permission */
321#define	VREAD			000000000400 /* read permission */
322#define	VADMIN			000000010000 /* being the file owner */
323#define	VAPPEND			000000040000 /* permission to write/append */
324/*
325 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
326 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
327 * and 0 otherwise.  This never happens with ordinary unix access rights
328 * or POSIX.1e ACLs.  Obviously, VEXPLICIT_DENY must be OR-ed with
329 * some other V* constant.
330 */
331#define	VEXPLICIT_DENY		000000100000
332#define	VREAD_NAMED_ATTRS 	000000200000 /* not used */
333#define	VWRITE_NAMED_ATTRS 	000000400000 /* not used */
334#define	VDELETE_CHILD	 	000001000000
335#define	VREAD_ATTRIBUTES 	000002000000 /* permission to stat(2) */
336#define	VWRITE_ATTRIBUTES 	000004000000 /* change {m,c,a}time */
337#define	VDELETE		 	000010000000
338#define	VREAD_ACL	 	000020000000 /* read ACL and file mode */
339#define	VWRITE_ACL	 	000040000000 /* change ACL and/or file mode */
340#define	VWRITE_OWNER	 	000100000000 /* change file owner */
341#define	VSYNCHRONIZE	 	000200000000 /* not used */
342#define	VCREAT			000400000000 /* creating new file */
343#define	VVERIFY			001000000000 /* verification required */
344
345/*
346 * Permissions that were traditionally granted only to the file owner.
347 */
348#define VADMIN_PERMS	(VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
349    VWRITE_OWNER)
350
351/*
352 * Permissions that were traditionally granted to everyone.
353 */
354#define VSTAT_PERMS	(VREAD_ATTRIBUTES | VREAD_ACL)
355
356/*
357 * Permissions that allow to change the state of the file in any way.
358 */
359#define VMODIFY_PERMS	(VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
360    VDELETE)
361
362/*
363 * Token indicating no attribute value yet assigned.
364 */
365#define	VNOVAL	(-1)
366
367/*
368 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
369 */
370#define VLKTIMEOUT	(hz / 20 + 1)
371
372#ifdef _KERNEL
373
374#ifdef MALLOC_DECLARE
375MALLOC_DECLARE(M_VNODE);
376#endif
377
378extern u_int ncsizefactor;
379
380/*
381 * Convert between vnode types and inode formats (since POSIX.1
382 * defines mode word of stat structure in terms of inode formats).
383 */
384extern enum vtype	iftovt_tab[];
385extern int		vttoif_tab[];
386#define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
387#define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
388#define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
389
390/*
391 * Flags to various vnode functions.
392 */
393#define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
394#define	FORCECLOSE	0x0002	/* vflush: force file closure */
395#define	WRITECLOSE	0x0004	/* vflush: only close writable files */
396#define	EARLYFLUSH	0x0008	/* vflush: early call for ffs_flushfiles */
397#define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
398#define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
399#define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
400#define	V_CLEANONLY	0x0008	/* vinvalbuf: invalidate only clean bufs */
401#define	V_VMIO		0x0010	/* vinvalbuf: called during pageout */
402#define	V_ALLOWCLEAN	0x0020	/* vinvalbuf: allow clean buffers after flush */
403#define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
404#define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
405#define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
406#define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
407#define	V_MNTREF	0x0010	/* vn_start_write: mp is already ref-ed */
408
409#define	VR_START_WRITE	0x0001	/* vfs_write_resume: start write atomically */
410#define	VR_NO_SUSPCLR	0x0002	/* vfs_write_resume: do not clear suspension */
411
412#define	VS_SKIP_UNMOUNT	0x0001	/* vfs_write_suspend: fail if the
413				   filesystem is being unmounted */
414
415#define	VREF(vp)	vref(vp)
416
417#ifdef DIAGNOSTIC
418#define	VATTR_NULL(vap)	vattr_null(vap)
419#else
420#define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
421#endif /* DIAGNOSTIC */
422
423#define	NULLVP	((struct vnode *)NULL)
424
425/*
426 * Global vnode data.
427 */
428extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
429extern	struct mount *rootdevmp;	/* "/dev" mount */
430extern	int desiredvnodes;		/* number of vnodes desired */
431extern	struct uma_zone *namei_zone;
432extern	struct vattr va_null;		/* predefined null vattr structure */
433
434#define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
435#define	VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
436#define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
437#define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
438#define	VI_MTX(vp)	(&(vp)->v_interlock)
439
440#define	VN_LOCK_AREC(vp)	lockallowrecurse((vp)->v_vnlock)
441#define	VN_LOCK_ASHARE(vp)	lockallowshare((vp)->v_vnlock)
442#define	VN_LOCK_DSHARE(vp)	lockdisableshare((vp)->v_vnlock)
443
444#endif /* _KERNEL */
445
446/*
447 * Mods for extensibility.
448 */
449
450/*
451 * Flags for vdesc_flags:
452 */
453#define	VDESC_MAX_VPS		16
454/* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
455#define	VDESC_VP0_WILLRELE	0x0001
456#define	VDESC_VP1_WILLRELE	0x0002
457#define	VDESC_VP2_WILLRELE	0x0004
458#define	VDESC_VP3_WILLRELE	0x0008
459#define	VDESC_NOMAP_VPP		0x0100
460#define	VDESC_VPP_WILLRELE	0x0200
461
462/*
463 * A generic structure.
464 * This can be used by bypass routines to identify generic arguments.
465 */
466struct vop_generic_args {
467	struct vnodeop_desc *a_desc;
468	/* other random data follows, presumably */
469};
470
471typedef int vop_bypass_t(struct vop_generic_args *);
472
473/*
474 * VDESC_NO_OFFSET is used to identify the end of the offset list
475 * and in places where no such field exists.
476 */
477#define VDESC_NO_OFFSET -1
478
479/*
480 * This structure describes the vnode operation taking place.
481 */
482struct vnodeop_desc {
483	char	*vdesc_name;		/* a readable name for debugging */
484	int	 vdesc_flags;		/* VDESC_* flags */
485	vop_bypass_t	*vdesc_call;	/* Function to call */
486
487	/*
488	 * These ops are used by bypass routines to map and locate arguments.
489	 * Creds and procs are not needed in bypass routines, but sometimes
490	 * they are useful to (for example) transport layers.
491	 * Nameidata is useful because it has a cred in it.
492	 */
493	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
494	int	vdesc_vpp_offset;	/* return vpp location */
495	int	vdesc_cred_offset;	/* cred location, if any */
496	int	vdesc_thread_offset;	/* thread location, if any */
497	int	vdesc_componentname_offset; /* if any */
498};
499
500#ifdef _KERNEL
501/*
502 * A list of all the operation descs.
503 */
504extern struct vnodeop_desc *vnodeop_descs[];
505
506#define	VOPARG_OFFSETOF(s_type, field)	__offsetof(s_type, field)
507#define	VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
508    ((s_type)(((char*)(struct_p)) + (s_offset)))
509
510
511#ifdef DEBUG_VFS_LOCKS
512/*
513 * Support code to aid in debugging VFS locking problems.  Not totally
514 * reliable since if the thread sleeps between changing the lock
515 * state and checking it with the assert, some other thread could
516 * change the state.  They are good enough for debugging a single
517 * filesystem using a single-threaded test.  Note that the unreliability is
518 * limited to false negatives; efforts were made to ensure that false
519 * positives cannot occur.
520 */
521void	assert_vi_locked(struct vnode *vp, const char *str);
522void	assert_vi_unlocked(struct vnode *vp, const char *str);
523void	assert_vop_elocked(struct vnode *vp, const char *str);
524#if 0
525void	assert_vop_elocked_other(struct vnode *vp, const char *str);
526#endif
527void	assert_vop_locked(struct vnode *vp, const char *str);
528#if 0
529voi0	assert_vop_slocked(struct vnode *vp, const char *str);
530#endif
531void	assert_vop_unlocked(struct vnode *vp, const char *str);
532
533#define	ASSERT_VI_LOCKED(vp, str)	assert_vi_locked((vp), (str))
534#define	ASSERT_VI_UNLOCKED(vp, str)	assert_vi_unlocked((vp), (str))
535#define	ASSERT_VOP_ELOCKED(vp, str)	assert_vop_elocked((vp), (str))
536#if 0
537#define	ASSERT_VOP_ELOCKED_OTHER(vp, str) assert_vop_locked_other((vp), (str))
538#endif
539#define	ASSERT_VOP_LOCKED(vp, str)	assert_vop_locked((vp), (str))
540#if 0
541#define	ASSERT_VOP_SLOCKED(vp, str)	assert_vop_slocked((vp), (str))
542#endif
543#define	ASSERT_VOP_UNLOCKED(vp, str)	assert_vop_unlocked((vp), (str))
544
545#else /* !DEBUG_VFS_LOCKS */
546
547#define	ASSERT_VI_LOCKED(vp, str)	((void)0)
548#define	ASSERT_VI_UNLOCKED(vp, str)	((void)0)
549#define	ASSERT_VOP_ELOCKED(vp, str)	((void)0)
550#if 0
551#define	ASSERT_VOP_ELOCKED_OTHER(vp, str)
552#endif
553#define	ASSERT_VOP_LOCKED(vp, str)	((void)0)
554#if 0
555#define	ASSERT_VOP_SLOCKED(vp, str)
556#endif
557#define	ASSERT_VOP_UNLOCKED(vp, str)	((void)0)
558#endif /* DEBUG_VFS_LOCKS */
559
560
561/*
562 * This call works for vnodes in the kernel.
563 */
564#define VCALL(c) ((c)->a_desc->vdesc_call(c))
565
566#define DOINGASYNC(vp)	   					\
567	(((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 &&	\
568	 ((curthread->td_pflags & TDP_SYNCIO) == 0))
569
570/*
571 * VMIO support inline
572 */
573
574extern int vmiodirenable;
575
576static __inline int
577vn_canvmio(struct vnode *vp)
578{
579      if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
580		return(TRUE);
581	return(FALSE);
582}
583
584/*
585 * Finally, include the default set of vnode operations.
586 */
587typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
588#include "vnode_if.h"
589
590/* vn_open_flags */
591#define	VN_OPEN_NOAUDIT		0x00000001
592#define	VN_OPEN_NOCAPCHECK	0x00000002
593#define	VN_OPEN_NAMECACHE	0x00000004
594
595/*
596 * Public vnode manipulation functions.
597 */
598struct componentname;
599struct file;
600struct mount;
601struct nameidata;
602struct ostat;
603struct thread;
604struct proc;
605struct stat;
606struct nstat;
607struct ucred;
608struct uio;
609struct vattr;
610struct vfsops;
611struct vnode;
612
613typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
614
615int	bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
616	    daddr_t endn);
617/* cache_* may belong in namei.h. */
618void	cache_changesize(int newhashsize);
619#define	cache_enter(dvp, vp, cnp)					\
620	cache_enter_time(dvp, vp, cnp, NULL, NULL)
621void	cache_enter_time(struct vnode *dvp, struct vnode *vp,
622	    struct componentname *cnp, struct timespec *tsp,
623	    struct timespec *dtsp);
624int	cache_lookup(struct vnode *dvp, struct vnode **vpp,
625	    struct componentname *cnp, struct timespec *tsp, int *ticksp);
626void	cache_purge(struct vnode *vp);
627void	cache_purge_negative(struct vnode *vp);
628void	cache_purgevfs(struct mount *mp, bool force);
629int	change_dir(struct vnode *vp, struct thread *td);
630void	cvtstat(struct stat *st, struct ostat *ost);
631void	cvtnstat(struct stat *sb, struct nstat *nsb);
632int	getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
633	    struct vnode **vpp);
634void	getnewvnode_reserve(u_int count);
635void	getnewvnode_drop_reserve(void);
636int	insmntque1(struct vnode *vp, struct mount *mp,
637	    void (*dtr)(struct vnode *, void *), void *dtr_arg);
638int	insmntque(struct vnode *vp, struct mount *mp);
639u_quad_t init_va_filerev(void);
640int	speedup_syncer(void);
641int	vn_vptocnp(struct vnode **vp, struct ucred *cred, char *buf,
642	    u_int *buflen);
643#define textvp_fullpath(p, rb, rfb) \
644	vn_fullpath(FIRST_THREAD_IN_PROC(p), (p)->p_textvp, rb, rfb)
645int	vn_fullpath(struct thread *td, struct vnode *vn,
646	    char **retbuf, char **freebuf);
647int	vn_fullpath_global(struct thread *td, struct vnode *vn,
648	    char **retbuf, char **freebuf);
649struct vnode *
650	vn_dir_dd_ino(struct vnode *vp);
651int	vn_commname(struct vnode *vn, char *buf, u_int buflen);
652int	vn_path_to_global_path(struct thread *td, struct vnode *vp,
653	    char *path, u_int pathlen);
654int	vaccess(enum vtype type, mode_t file_mode, uid_t file_uid,
655	    gid_t file_gid, accmode_t accmode, struct ucred *cred,
656	    int *privused);
657int	vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid,
658	    struct acl *aclp, accmode_t accmode, struct ucred *cred,
659	    int *privused);
660int	vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
661	    gid_t file_gid, struct acl *acl, accmode_t accmode,
662	    struct ucred *cred, int *privused);
663void	vattr_null(struct vattr *vap);
664int	vcount(struct vnode *vp);
665#define	vdrop(vp)	_vdrop((vp), 0)
666#define	vdropl(vp)	_vdrop((vp), 1)
667void	_vdrop(struct vnode *, bool);
668int	vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
669int	vget(struct vnode *vp, int lockflag, struct thread *td);
670void	vgone(struct vnode *vp);
671#define	vhold(vp)	_vhold((vp), 0)
672#define	vholdl(vp)	_vhold((vp), 1)
673void	_vhold(struct vnode *, bool);
674void	vinactive(struct vnode *, struct thread *);
675int	vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
676int	vtruncbuf(struct vnode *vp, struct ucred *cred, off_t length,
677	    int blksize);
678void	vunref(struct vnode *);
679void	vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
680int	vrecycle(struct vnode *vp);
681int	vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
682	    struct ucred *cred);
683int	vn_close(struct vnode *vp,
684	    int flags, struct ucred *file_cred, struct thread *td);
685void	vn_finished_write(struct mount *mp);
686void	vn_finished_secondary_write(struct mount *mp);
687int	vn_isdisk(struct vnode *vp, int *errp);
688int	_vn_lock(struct vnode *vp, int flags, char *file, int line);
689#define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
690int	vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
691int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
692	    u_int vn_open_flags, struct ucred *cred, struct file *fp);
693int	vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
694	    struct thread *td, struct file *fp);
695void	vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
696int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
697int	vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
698	    int len, off_t offset, enum uio_seg segflg, int ioflg,
699	    struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
700	    struct thread *td);
701int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
702	    size_t len, off_t offset, enum uio_seg segflg, int ioflg,
703	    struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
704	    struct thread *td);
705int	vn_rlimit_fsize(const struct vnode *vn, const struct uio *uio,
706	    struct thread *td);
707int	vn_stat(struct vnode *vp, struct stat *sb, struct ucred *active_cred,
708	    struct ucred *file_cred, struct thread *td);
709int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
710int	vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
711	    int flags);
712int	vn_writechk(struct vnode *vp);
713int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
714	    const char *attrname, int *buflen, char *buf, struct thread *td);
715int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
716	    const char *attrname, int buflen, char *buf, struct thread *td);
717int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
718	    const char *attrname, struct thread *td);
719int	vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
720	    struct vnode **rvp);
721int	vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
722	    void *alloc_arg, int lkflags, struct vnode **rvp);
723int	vn_utimes_perm(struct vnode *vp, struct vattr *vap,
724	    struct ucred *cred, struct thread *td);
725
726int	vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
727int	vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
728	    struct uio *uio);
729
730#define	vn_rangelock_unlock(vp, cookie)					\
731	rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp))
732#define	vn_rangelock_unlock_range(vp, cookie, start, end)		\
733	rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), 	\
734	    VI_MTX(vp))
735#define	vn_rangelock_rlock(vp, start, end)				\
736	rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
737#define	vn_rangelock_wlock(vp, start, end)				\
738	rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
739
740int	vfs_cache_lookup(struct vop_lookup_args *ap);
741void	vfs_timestamp(struct timespec *);
742void	vfs_write_resume(struct mount *mp, int flags);
743int	vfs_write_suspend(struct mount *mp, int flags);
744int	vfs_write_suspend_umnt(struct mount *mp);
745void	vnlru_free(int, struct vfsops *);
746int	vop_stdbmap(struct vop_bmap_args *);
747int	vop_stdfdatasync_buf(struct vop_fdatasync_args *);
748int	vop_stdfsync(struct vop_fsync_args *);
749int	vop_stdgetwritemount(struct vop_getwritemount_args *);
750int	vop_stdgetpages(struct vop_getpages_args *);
751int	vop_stdinactive(struct vop_inactive_args *);
752int	vop_stdislocked(struct vop_islocked_args *);
753int	vop_stdkqfilter(struct vop_kqfilter_args *);
754int	vop_stdlock(struct vop_lock1_args *);
755int	vop_stdputpages(struct vop_putpages_args *);
756int	vop_stdunlock(struct vop_unlock_args *);
757int	vop_nopoll(struct vop_poll_args *);
758int	vop_stdaccess(struct vop_access_args *ap);
759int	vop_stdaccessx(struct vop_accessx_args *ap);
760int	vop_stdadvise(struct vop_advise_args *ap);
761int	vop_stdadvlock(struct vop_advlock_args *ap);
762int	vop_stdadvlockasync(struct vop_advlockasync_args *ap);
763int	vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
764int	vop_stdallocate(struct vop_allocate_args *ap);
765int	vop_stdpathconf(struct vop_pathconf_args *);
766int	vop_stdpoll(struct vop_poll_args *);
767int	vop_stdvptocnp(struct vop_vptocnp_args *ap);
768int	vop_stdvptofh(struct vop_vptofh_args *ap);
769int	vop_stdunp_bind(struct vop_unp_bind_args *ap);
770int	vop_stdunp_connect(struct vop_unp_connect_args *ap);
771int	vop_stdunp_detach(struct vop_unp_detach_args *ap);
772int	vop_eopnotsupp(struct vop_generic_args *ap);
773int	vop_ebadf(struct vop_generic_args *ap);
774int	vop_einval(struct vop_generic_args *ap);
775int	vop_enoent(struct vop_generic_args *ap);
776int	vop_enotty(struct vop_generic_args *ap);
777int	vop_null(struct vop_generic_args *ap);
778int	vop_panic(struct vop_generic_args *ap);
779int	dead_poll(struct vop_poll_args *ap);
780int	dead_read(struct vop_read_args *ap);
781int	dead_write(struct vop_write_args *ap);
782
783/* These are called from within the actual VOPS. */
784void	vop_close_post(void *a, int rc);
785void	vop_create_post(void *a, int rc);
786void	vop_deleteextattr_post(void *a, int rc);
787void	vop_link_post(void *a, int rc);
788void	vop_lookup_post(void *a, int rc);
789void	vop_lookup_pre(void *a);
790void	vop_mkdir_post(void *a, int rc);
791void	vop_mknod_post(void *a, int rc);
792void	vop_open_post(void *a, int rc);
793void	vop_read_post(void *a, int rc);
794void	vop_readdir_post(void *a, int rc);
795void	vop_reclaim_post(void *a, int rc);
796void	vop_remove_post(void *a, int rc);
797void	vop_rename_post(void *a, int rc);
798void	vop_rename_pre(void *a);
799void	vop_rmdir_post(void *a, int rc);
800void	vop_setattr_post(void *a, int rc);
801void	vop_setextattr_post(void *a, int rc);
802void	vop_symlink_post(void *a, int rc);
803
804#ifdef DEBUG_VFS_LOCKS
805void	vop_strategy_pre(void *a);
806void	vop_lock_pre(void *a);
807void	vop_lock_post(void *a, int rc);
808void	vop_unlock_post(void *a, int rc);
809void	vop_unlock_pre(void *a);
810#else
811#define	vop_strategy_pre(x)	do { } while (0)
812#define	vop_lock_pre(x)		do { } while (0)
813#define	vop_lock_post(x, y)	do { } while (0)
814#define	vop_unlock_post(x, y)	do { } while (0)
815#define	vop_unlock_pre(x)	do { } while (0)
816#endif
817
818void	vop_rename_fail(struct vop_rename_args *ap);
819
820#define	VOP_WRITE_PRE(ap)						\
821	struct vattr va;						\
822	int error;							\
823	off_t osize, ooffset, noffset;					\
824									\
825	osize = ooffset = noffset = 0;					\
826	if (!VN_KNLIST_EMPTY((ap)->a_vp)) {				\
827		error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred);	\
828		if (error)						\
829			return (error);					\
830		ooffset = (ap)->a_uio->uio_offset;			\
831		osize = (off_t)va.va_size;				\
832	}
833
834#define VOP_WRITE_POST(ap, ret)						\
835	noffset = (ap)->a_uio->uio_offset;				\
836	if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) {	\
837		VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE			\
838		    | (noffset > osize ? NOTE_EXTEND : 0));		\
839	}
840
841#define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
842
843
844void	vput(struct vnode *vp);
845void	vrele(struct vnode *vp);
846void	vref(struct vnode *vp);
847void	vrefl(struct vnode *vp);
848void	vrefact(struct vnode *vp);
849int	vrefcnt(struct vnode *vp);
850void 	v_addpollinfo(struct vnode *vp);
851
852int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
853void vnode_destroy_vobject(struct vnode *vp);
854
855extern struct vop_vector fifo_specops;
856extern struct vop_vector dead_vnodeops;
857extern struct vop_vector default_vnodeops;
858
859#define VOP_PANIC	((void*)(uintptr_t)vop_panic)
860#define VOP_NULL	((void*)(uintptr_t)vop_null)
861#define VOP_EBADF	((void*)(uintptr_t)vop_ebadf)
862#define VOP_ENOTTY	((void*)(uintptr_t)vop_enotty)
863#define VOP_EINVAL	((void*)(uintptr_t)vop_einval)
864#define VOP_ENOENT	((void*)(uintptr_t)vop_enoent)
865#define VOP_EOPNOTSUPP	((void*)(uintptr_t)vop_eopnotsupp)
866
867/* fifo_vnops.c */
868int	fifo_printinfo(struct vnode *);
869
870/* vfs_hash.c */
871typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
872
873void vfs_hash_changesize(int newhashsize);
874int vfs_hash_get(const struct mount *mp, u_int hash, int flags,
875    struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
876u_int vfs_hash_index(struct vnode *vp);
877int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
878    struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
879void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
880    struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
881void vfs_hash_rehash(struct vnode *vp, u_int hash);
882void vfs_hash_remove(struct vnode *vp);
883
884int vfs_kqfilter(struct vop_kqfilter_args *);
885void vfs_mark_atime(struct vnode *vp, struct ucred *cred);
886struct dirent;
887int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
888
889int vfs_unixify_accmode(accmode_t *accmode);
890
891void vfs_unp_reclaim(struct vnode *vp);
892
893int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
894int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
895    gid_t gid);
896int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
897    struct thread *td);
898int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
899    struct thread *td);
900
901#endif /* _KERNEL */
902
903#endif /* !_SYS_VNODE_H_ */
904