vnode.h revision 103926
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 * 3. All advertising materials mentioning features or use of this software
14 *    must display the following acknowledgement:
15 *	This product includes software developed by the University of
16 *	California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 *    may be used to endorse or promote products derived from this software
19 *    without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 *	@(#)vnode.h	8.7 (Berkeley) 2/4/94
34 * $FreeBSD: head/sys/sys/vnode.h 103926 2002-09-25 01:37:25Z jeff $
35 */
36
37#ifndef _SYS_VNODE_H_
38#define	_SYS_VNODE_H_
39
40/*
41 * XXX - compatability until lockmgr() goes away or all the #includes are
42 * updated.
43 */
44#include <sys/lockmgr.h>
45
46#include <sys/queue.h>
47#include <sys/_label.h>
48#include <sys/_lock.h>
49#include <sys/lock.h>
50#include <sys/_mutex.h>
51#include <sys/mutex.h>
52#include <sys/selinfo.h>
53#include <sys/uio.h>
54#include <sys/acl.h>
55#include <sys/ktr.h>
56
57/*
58 * The vnode is the focus of all file activity in UNIX.  There is a
59 * unique vnode allocated for each active file, each current directory,
60 * each mounted-on file, text file, and the root.
61 */
62
63/*
64 * Vnode types.  VNON means no type.
65 */
66enum vtype	{ VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD };
67
68/*
69 * Each underlying filesystem allocates its own private area and hangs
70 * it from v_data.  If non-null, this area is freed in getnewvnode().
71 */
72TAILQ_HEAD(buflists, buf);
73
74typedef	int	vop_t(void *);
75struct namecache;
76
77struct vpollinfo {
78	struct	mtx vpi_lock;		/* lock to protect below */
79	struct	selinfo vpi_selinfo;	/* identity of poller(s) */
80	short	vpi_events;		/* what they are looking for */
81	short	vpi_revents;		/* what has happened */
82};
83
84/*
85 * Reading or writing any of these items requires holding the appropriate lock.
86 *
87 * Lock reference:
88 *	c - namecache mutex
89 *	f - freelist mutex
90 *	i - interlock
91 *	m - mntvnodes mutex
92 *	p - pollinfo lock
93 *	s - spechash mutex
94 *	S - syncer mutex
95 *	u - Only a reference to the vnode is needed to read.
96 *	v - vnode lock
97 *
98 * XXX Not all fields are locked yet and some fields that are marked are not
99 * locked consistently.  This is a work in progress.
100 */
101
102struct vnode {
103	struct	mtx v_interlock;		/* lock for "i" things */
104	u_long	v_iflag;			/* i vnode flags (see below) */
105	int	v_usecount;			/* i ref count of users */
106	long	v_numoutput;			/* i writes in progress */
107	struct thread *v_vxproc;		/* i thread owning VXLOCK */
108	int	v_holdcnt;			/* i page & buffer references */
109	struct	buflists v_cleanblkhd;		/* i SORTED clean blocklist */
110	struct buf	*v_cleanblkroot;	/* i clean buf splay tree  */
111	struct	buflists v_dirtyblkhd;		/* i SORTED dirty blocklist */
112	struct buf	*v_dirtyblkroot;	/* i dirty buf splay tree */
113	u_long	v_vflag;			/* v vnode flags */
114	int	v_writecount;			/* v ref count of writers */
115	struct vm_object *v_object;		/* v Place to store VM object */
116	daddr_t	v_lastw;			/* v last write (write cluster) */
117	daddr_t	v_cstart;			/* v start block of cluster */
118	daddr_t	v_lasta;			/* v last allocation (cluster) */
119	int	v_clen;				/* v length of current cluster */
120	union {
121		struct mount	*vu_mountedhere;/* v ptr to mounted vfs (VDIR) */
122		struct socket	*vu_socket;	/* v unix ipc (VSOCK) */
123		struct {
124			struct specinfo	*vu_specinfo; /* v device (VCHR, VBLK) */
125			SLIST_ENTRY(vnode) vu_specnext;	/* s device aliases */
126		} vu_spec;
127		struct fifoinfo	*vu_fifoinfo;	/* v fifo (VFIFO) */
128	} v_un;
129	TAILQ_ENTRY(vnode) v_freelist;		/* f vnode freelist */
130	TAILQ_ENTRY(vnode) v_nmntvnodes;	/* m vnodes for mount point */
131	LIST_ENTRY(vnode) v_synclist;		/* S dirty vnode list */
132	enum	vtype v_type;			/* u vnode type */
133	const char *v_tag;			/* u type of underlying data */
134	void	*v_data;			/* u private data for fs */
135	struct	lock v_lock;			/* u used if fs don't have one */
136	struct	lock *v_vnlock;			/* u pointer to vnode lock */
137	vop_t	**v_op;				/* u vnode operations vector */
138	struct	mount *v_mount;			/* u ptr to vfs we are in */
139	LIST_HEAD(, namecache) v_cache_src;	/* c Cache entries from us */
140	TAILQ_HEAD(, namecache) v_cache_dst;	/* c Cache entries to us */
141	u_long	v_id;				/* c capability identifier */
142	struct	vnode *v_dd;			/* c .. vnode */
143	u_long	v_ddid;				/* c .. capability identifier */
144	struct vpollinfo *v_pollinfo;		/* p Poll events */
145	struct label v_label;			/* MAC label for vnode */
146#ifdef	DEBUG_LOCKS
147	const char *filename;			/* Source file doing locking */
148	int line;				/* Line number doing locking */
149#endif
150	udev_t	v_cachedfs;			/* cached fs id */
151	ino_t	v_cachedid;			/* cached file id */
152};
153#define	v_mountedhere	v_un.vu_mountedhere
154#define	v_socket	v_un.vu_socket
155#define	v_rdev		v_un.vu_spec.vu_specinfo
156#define	v_specnext	v_un.vu_spec.vu_specnext
157#define	v_fifoinfo	v_un.vu_fifoinfo
158
159/*
160 * Userland version of struct vnode, for sysctl.
161 */
162struct xvnode {
163	size_t	xv_size;			/* sizeof(struct xvnode) */
164	void	*xv_vnode;			/* address of real vnode */
165	u_long	xv_flag;			/* vnode vflags */
166	int	xv_usecount;			/* reference count of users */
167	int	xv_writecount;			/* reference count of writers */
168	int	xv_holdcnt;			/* page & buffer references */
169	u_long	xv_id;				/* capability identifier */
170	void	*xv_mount;			/* address of parent mount */
171	long	xv_numoutput;			/* num of writes in progress */
172	enum	vtype xv_type;			/* vnode type */
173	union {
174		void	*xvu_socket;		/* socket, if VSOCK */
175		void	*xvu_fifo;		/* fifo, if VFIFO */
176		udev_t	xvu_rdev;		/* maj/min, if VBLK/VCHR */
177		struct {
178			udev_t	xvu_dev;	/* device, if VDIR/VREG/VLNK */
179			ino_t	xvu_ino;	/* id, if VDIR/VREG/VLNK */
180		};
181	} xv_un;
182};
183#define xv_socket	xv_un.xvu_socket
184#define xv_fifo		xv_un.xvu_fifo
185#define xv_rdev		xv_un.xvu_rdev
186#define xv_dev		xv_un.xvu_dev
187#define xv_ino		xv_un.xvu_ino
188
189#define	VN_POLLEVENT(vp, events)				\
190	do {							\
191		if ((vp)->v_pollinfo != NULL && 		\
192		    (vp)->v_pollinfo->vpi_events & (events))	\
193			vn_pollevent((vp), (events));		\
194	} while (0)
195
196#define VN_KNOTE(vp, b)						\
197	do {							\
198		if ((vp)->v_pollinfo != NULL)			\
199			KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b)); \
200	} while (0)
201
202/*
203 * Vnode flags.
204 *	VI flags are protected by interlock and live in v_iflag
205 *	VV flags are protected by the vnode lock and live in v_vflag
206 */
207#define	VI_XLOCK	0x0001	/* vnode is locked to change vtype */
208#define	VI_XWANT	0x0002	/* thread is waiting for vnode */
209#define	VI_BWAIT	0x0004	/* waiting for output to complete */
210#define	VI_OLOCK	0x0008	/* vnode is locked waiting for an object */
211#define	VI_OWANT	0x0010	/* a thread is waiting for VOLOCK */
212#define	VI_MOUNT	0x0020	/* Mount in progress */
213#define	VI_AGE		0x0040	/* Insert vnode at head of free list */
214#define	VI_DOOMED	0x0080	/* This vnode is being recycled */
215#define	VI_FREE		0x0100	/* This vnode is on the freelist */
216#define	VI_OBJDIRTY	0x0400	/* object might be dirty */
217/*
218 * XXX VI_ONWORKLST could be replaced with a check for NULL list elements
219 * in v_synclist.
220 */
221#define	VI_ONWORKLST	0x0200	/* On syncer work-list */
222
223#define	VV_ROOT		0x0001	/* root of its filesystem */
224#define	VV_ISTTY	0x0002	/* vnode represents a tty */
225#define	VV_NOSYNC	0x0004	/* unlinked, stop syncing */
226#define	VV_OBJBUF	0x0008	/* Allocate buffers in VM object */
227#define	VV_CACHEDLABEL	0x0010	/* Vnode has valid cached MAC label */
228#define	VV_TEXT		0x0020	/* vnode is a pure text prototype */
229#define	VV_COPYONWRITE	0x0040	/* vnode is doing copy-on-write */
230#define	VV_SYSTEM	0x0080	/* vnode being used by kernel */
231#define	VV_PROCDEP	0x0100	/* vnode is process dependent */
232
233/*
234 * Vnode attributes.  A field value of VNOVAL represents a field whose value
235 * is unavailable (getattr) or which is not to be changed (setattr).
236 */
237struct vattr {
238	enum vtype	va_type;	/* vnode type (for create) */
239	u_short		va_mode;	/* files access mode and type */
240	short		va_nlink;	/* number of references to file */
241	uid_t		va_uid;		/* owner user id */
242	gid_t		va_gid;		/* owner group id */
243	udev_t		va_fsid;	/* filesystem id */
244	long		va_fileid;	/* file id */
245	u_quad_t	va_size;	/* file size in bytes */
246	long		va_blocksize;	/* blocksize preferred for i/o */
247	struct timespec	va_atime;	/* time of last access */
248	struct timespec	va_mtime;	/* time of last modification */
249	struct timespec	va_ctime;	/* time file changed */
250	struct timespec	va_birthtime;	/* time file created */
251	u_long		va_gen;		/* generation number of file */
252	u_long		va_flags;	/* flags defined for file */
253	udev_t		va_rdev;	/* device the special file represents */
254	u_quad_t	va_bytes;	/* bytes of disk space held by file */
255	u_quad_t	va_filerev;	/* file modification number */
256	u_int		va_vaflags;	/* operations flags, see below */
257	long		va_spare;	/* remain quad aligned */
258};
259
260/*
261 * Flags for va_vaflags.
262 */
263#define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
264#define	VA_EXCLUSIVE	0x02		/* exclusive create request */
265
266/*
267 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
268 * help with write clustering)
269 */
270#define	IO_UNIT		0x0001		/* do I/O as atomic unit */
271#define	IO_APPEND	0x0002		/* append write to end */
272#define	IO_SYNC		0x0004		/* do I/O synchronously */
273#define	IO_NODELOCKED	0x0008		/* underlying node already locked */
274#define	IO_NDELAY	0x0010		/* FNDELAY flag set in file table */
275#define	IO_VMIO		0x0020		/* data already in VMIO space */
276#define	IO_INVAL	0x0040		/* invalidate after I/O */
277#define	IO_ASYNC	0x0080		/* bawrite rather then bdwrite */
278#define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
279#define	IO_NOWDRAIN	0x0200		/* do not block on wdrain */
280#define	IO_EXT		0x0400		/* operate on external attributes */
281#define	IO_NORMAL	0x0800		/* operate on regular data */
282#define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
283
284/*
285 *  Modes.  Some values same as Ixxx entries from inode.h for now.
286 */
287#define	VEXEC	000100		/* execute/search permission */
288#define	VWRITE	000200		/* write permission */
289#define	VREAD	000400		/* read permission */
290#define	VSVTX	001000		/* save swapped text even after use */
291#define	VSGID	002000		/* set group id on execution */
292#define	VSUID	004000		/* set user id on execution */
293#define	VADMIN	010000		/* permission to administer */
294#define	VSTAT	020000		/* permission to retrieve attrs */
295#define	VAPPEND	040000		/* permission to write/append */
296#define	VALLPERM	(VEXEC | VWRITE | VREAD | VADMIN | VSTAT | VAPPEND)
297
298/*
299 * Token indicating no attribute value yet assigned.
300 */
301#define	VNOVAL	(-1)
302
303/*
304 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
305 */
306#define VLKTIMEOUT	(hz / 20 + 1)
307
308#ifdef _KERNEL
309
310#ifdef MALLOC_DECLARE
311MALLOC_DECLARE(M_VNODE);
312#endif
313
314/*
315 * Convert between vnode types and inode formats (since POSIX.1
316 * defines mode word of stat structure in terms of inode formats).
317 */
318extern enum vtype	iftovt_tab[];
319extern int		vttoif_tab[];
320#define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
321#define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
322#define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
323
324/*
325 * Flags to various vnode functions.
326 */
327#define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
328#define	FORCECLOSE	0x0002	/* vflush: force file closure */
329#define	WRITECLOSE	0x0004	/* vflush: only close writable files */
330#define	DOCLOSE		0x0008	/* vclean: close active files */
331#define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
332#define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
333#define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
334#define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
335#define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
336#define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
337#define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
338
339#define	VREF(vp)	vref(vp)
340
341
342#ifdef DIAGNOSTIC
343#define	VATTR_NULL(vap)	vattr_null(vap)
344#else
345#define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
346#endif /* DIAGNOSTIC */
347
348#define	NULLVP	((struct vnode *)NULL)
349
350#define	VNODEOP_SET(f) \
351	C_SYSINIT(f##init, SI_SUB_VFS, SI_ORDER_SECOND, vfs_add_vnodeops, &f); \
352	C_SYSUNINIT(f##uninit, SI_SUB_VFS, SI_ORDER_SECOND, vfs_rm_vnodeops, &f);
353
354/*
355 * Global vnode data.
356 */
357extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
358extern	int desiredvnodes;		/* number of vnodes desired */
359extern	struct uma_zone *namei_zone;
360extern	int prtactive;			/* nonzero to call vprint() */
361extern	struct vattr va_null;		/* predefined null vattr structure */
362extern	int vfs_ioopt;
363
364/*
365 * Macro/function to check for client cache inconsistency w.r.t. leasing.
366 */
367#define	LEASE_READ	0x1		/* Check lease for readers */
368#define	LEASE_WRITE	0x2		/* Check lease for modifiers */
369
370
371extern void	(*lease_updatetime)(int deltat);
372
373/* Requires interlock */
374#define	VSHOULDFREE(vp)	\
375	(!((vp)->v_iflag & (VI_FREE|VI_DOOMED)) && \
376	 !(vp)->v_holdcnt && !(vp)->v_usecount && \
377	 (!(vp)->v_object || \
378	  !((vp)->v_object->ref_count || (vp)->v_object->resident_page_count)))
379
380/* Requires interlock */
381#define VMIGHTFREE(vp) \
382	(!((vp)->v_iflag & (VI_FREE|VI_DOOMED|VI_XLOCK)) &&	\
383	 LIST_EMPTY(&(vp)->v_cache_src) && !(vp)->v_usecount)
384
385/* Requires interlock */
386#define	VSHOULDBUSY(vp)	\
387	(((vp)->v_iflag & VI_FREE) && \
388	 ((vp)->v_holdcnt || (vp)->v_usecount))
389
390#define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
391#define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
392#define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
393#define	VI_MTX(vp)	(&(vp)->v_interlock)
394
395#endif /* _KERNEL */
396
397
398/*
399 * Mods for extensibility.
400 */
401
402/*
403 * Flags for vdesc_flags:
404 */
405#define	VDESC_MAX_VPS		16
406/* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
407#define	VDESC_VP0_WILLRELE	0x0001
408#define	VDESC_VP1_WILLRELE	0x0002
409#define	VDESC_VP2_WILLRELE	0x0004
410#define	VDESC_VP3_WILLRELE	0x0008
411#define	VDESC_NOMAP_VPP		0x0100
412#define	VDESC_VPP_WILLRELE	0x0200
413
414/*
415 * VDESC_NO_OFFSET is used to identify the end of the offset list
416 * and in places where no such field exists.
417 */
418#define VDESC_NO_OFFSET -1
419
420/*
421 * This structure describes the vnode operation taking place.
422 */
423struct vnodeop_desc {
424	int	 vdesc_offset;		/* offset in vector,first for speed */
425	char	*vdesc_name;		/* a readable name for debugging */
426	int	 vdesc_flags;		/* VDESC_* flags */
427
428	/*
429	 * These ops are used by bypass routines to map and locate arguments.
430	 * Creds and procs are not needed in bypass routines, but sometimes
431	 * they are useful to (for example) transport layers.
432	 * Nameidata is useful because it has a cred in it.
433	 */
434	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
435	int	vdesc_vpp_offset;	/* return vpp location */
436	int	vdesc_cred_offset;	/* cred location, if any */
437	int	vdesc_thread_offset;	/* thread location, if any */
438	int	vdesc_componentname_offset; /* if any */
439	/*
440	 * Finally, we've got a list of private data (about each operation)
441	 * for each transport layer.  (Support to manage this list is not
442	 * yet part of BSD.)
443	 */
444	caddr_t	*vdesc_transports;
445};
446
447#ifdef _KERNEL
448/*
449 * A list of all the operation descs.
450 */
451extern struct vnodeop_desc *vnodeop_descs[];
452
453/*
454 * Interlock for scanning list of vnodes attached to a mountpoint
455 */
456extern struct mtx mntvnode_mtx;
457
458/*
459 * This macro is very helpful in defining those offsets in the vdesc struct.
460 *
461 * This is stolen from X11R4.  I ignored all the fancy stuff for
462 * Crays, so if you decide to port this to such a serious machine,
463 * you might want to consult Intrinsic.h's XtOffset{,Of,To}.
464 */
465#define	VOPARG_OFFSET(p_type,field) \
466	((int) (((char *) (&(((p_type)NULL)->field))) - ((char *) NULL)))
467#define	VOPARG_OFFSETOF(s_type,field) \
468	VOPARG_OFFSET(s_type*,field)
469#define	VOPARG_OFFSETTO(S_TYPE,S_OFFSET,STRUCT_P) \
470	((S_TYPE)(((char*)(STRUCT_P))+(S_OFFSET)))
471
472
473/*
474 * This structure is used to configure the new vnodeops vector.
475 */
476struct vnodeopv_entry_desc {
477	struct vnodeop_desc *opve_op;   /* which operation this is */
478	vop_t *opve_impl;		/* code implementing this operation */
479};
480struct vnodeopv_desc {
481			/* ptr to the ptr to the vector where op should go */
482	vop_t ***opv_desc_vector_p;
483	struct vnodeopv_entry_desc *opv_desc_ops;   /* null terminated list */
484};
485
486/*
487 * A generic structure.
488 * This can be used by bypass routines to identify generic arguments.
489 */
490struct vop_generic_args {
491	struct vnodeop_desc *a_desc;
492	/* other random data follows, presumably */
493};
494
495
496#ifdef DEBUG_VFS_LOCKS
497/*
498 * Macros to aid in tracing VFS locking problems.  Not totally
499 * reliable since if the thread sleeps between changing the lock
500 * state and checking it with the assert, some other thread could
501 * change the state.  They are good enough for debugging a single
502 * filesystem using a single-threaded test.  I find that 'cvs co src'
503 * is a pretty good test.
504 */
505
506extern int vfs_badlock_panic;
507extern int vfs_badlock_print;
508extern int vfs_badlock_mutex;
509
510#define	ASSERT_VI_UNLOCKED(vp)						\
511do {									\
512	struct vnode *_vp = (vp);					\
513									\
514	if (vfs_badlock_mutex)						\
515		mtx_assert(VI_MTX(_vp), MA_NOTOWNED);			\
516} while (0)								\
517
518#define	ASSERT_VI_LOCKED(vp)						\
519do {									\
520	struct vnode *_vp = (vp);					\
521									\
522	if (vfs_badlock_mutex)						\
523		mtx_assert(VI_MTX(_vp), MA_OWNED);			\
524} while (0)								\
525
526
527/*
528 * This only exists to supress warnings from unlocked specfs accesses.  It is
529 * no longer ok to have an unlocked VFS.
530 */
531
532#define IGNORE_LOCK(vp)	((vp)->v_type == VCHR || (vp)->v_type == VBAD)
533
534#define ASSERT_VOP_LOCKED(vp, str)					\
535do {									\
536	struct vnode *_vp = (vp);					\
537									\
538	if (_vp && !IGNORE_LOCK(_vp) && !VOP_ISLOCKED(_vp, NULL)) {	\
539		if (vfs_badlock_print)					\
540			printf("%s: %p is not locked but should be\n",	\
541			    str, _vp);					\
542		if (vfs_badlock_panic)					\
543			Debugger("Lock violation.\n");			\
544	}								\
545} while (0)
546
547#define ASSERT_VOP_UNLOCKED(vp, str)					\
548do {									\
549	struct vnode *_vp = (vp);					\
550	int lockstate;							\
551									\
552	if (_vp && !IGNORE_LOCK(_vp)) {					\
553		lockstate = VOP_ISLOCKED(_vp, curthread);		\
554		if (lockstate == LK_EXCLUSIVE) {			\
555			if (vfs_badlock_print)				\
556				printf("%s: %p is locked but should not be\n",	\
557				    str, _vp);				\
558			if (vfs_badlock_panic)				\
559				Debugger("Lock Violation.\n");		\
560		}							\
561	}								\
562} while (0)
563
564#define ASSERT_VOP_ELOCKED(vp, str)					\
565do {									\
566	struct vnode *_vp = (vp);					\
567									\
568	if (_vp && !IGNORE_LOCK(_vp) &&					\
569	    VOP_ISLOCKED(_vp, curthread) != LK_EXCLUSIVE) {		\
570		if (vfs_badlock_print)					\
571			printf("%s: %p is not exclusive locked but should be\n",\
572			    str, _vp);					\
573		if (vfs_badlock_panic)					\
574			Debugger("Lock violation.\n");			\
575	}								\
576} while (0)
577
578#define ASSERT_VOP_ELOCKED_OTHER(vp, str)				\
579do {									\
580	struct vnode *_vp = (vp);					\
581									\
582	if (_vp && !IGNORE_LOCK(_vp) &&					\
583	    VOP_ISLOCKED(_vp, curthread) != LK_EXCLOTHER) {		\
584		if (vfs_badlock_print)					\
585			printf("%s: %p is not exclusive locked by another thread\n",\
586			    str, _vp);					\
587		if (vfs_badlock_panic)					\
588			Debugger("Lock violation.\n");			\
589	}								\
590} while (0)
591
592#define ASSERT_VOP_SLOCKED(vp, str)					\
593do {									\
594	struct vnode *_vp = (vp);					\
595									\
596	if (_vp && !IGNORE_LOCK(_vp) &&					\
597	    VOP_ISLOCKED(_vp, NULL) != LK_SHARED) {			\
598		if (vfs_badlock_print)					\
599			printf("%s: %p is not locked shared but should be",\
600		    str, _vp);						\
601		if (vfs_badlock_panic)					\
602			Debugger("Lock violation.\n");			\
603	}								\
604} while (0)
605
606void vop_rename_pre(void *a);
607void vop_strategy_pre(void *a);
608void vop_lookup_pre(void *a);
609void vop_lookup_post(void *a, int rc);
610void vop_lock_pre(void *a);
611void vop_lock_post(void *a, int rc);
612void vop_unlock_pre(void *a);
613void vop_unlock_post(void *a, int rc);
614
615#else
616
617#define ASSERT_VOP_LOCKED(vp, str)
618#define ASSERT_VOP_UNLOCKED(vp, str)
619#define	ASSERT_VI_UNLOCKED(vp)
620#define	ASSERT_VI_LOCKED(vp)
621
622#endif
623
624/*
625 * VOCALL calls an op given an ops vector.  We break it out because BSD's
626 * vclean changes the ops vector and then wants to call ops with the old
627 * vector.
628 */
629#define VOCALL(OPSV,OFF,AP) (( *((OPSV)[(OFF)])) (AP))
630
631/*
632 * This call works for vnodes in the kernel.
633 */
634#define VCALL(VP,OFF,AP) VOCALL((VP)->v_op,(OFF),(AP))
635#define VDESC(OP) (& __CONCAT(OP,_desc))
636#define VOFFSET(OP) (VDESC(OP)->vdesc_offset)
637
638/*
639 * VMIO support inline
640 */
641
642extern int vmiodirenable;
643
644static __inline int
645vn_canvmio(struct vnode *vp)
646{
647      if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
648		return(TRUE);
649	return(FALSE);
650}
651
652/*
653 * Finally, include the default set of vnode operations.
654 */
655#include "vnode_if.h"
656
657/*
658 * Public vnode manipulation functions.
659 */
660struct componentname;
661struct file;
662struct mount;
663struct nameidata;
664struct ostat;
665struct thread;
666struct proc;
667struct stat;
668struct nstat;
669struct ucred;
670struct uio;
671struct vattr;
672struct vnode;
673
674extern int	(*lease_check_hook)(struct vop_lease_args *);
675extern int	(*softdep_fsync_hook)(struct vnode *);
676extern int	(*softdep_process_worklist_hook)(struct mount *);
677
678struct	vnode *addaliasu(struct vnode *vp, udev_t nvp_rdev);
679int	bdevvp(dev_t dev, struct vnode **vpp);
680/* cache_* may belong in namei.h. */
681void	cache_enter(struct vnode *dvp, struct vnode *vp,
682	    struct componentname *cnp);
683int	cache_lookup(struct vnode *dvp, struct vnode **vpp,
684	    struct componentname *cnp);
685void	cache_purge(struct vnode *vp);
686void	cache_purgevfs(struct mount *mp);
687int	cache_leaf_test(struct vnode *vp);
688void	cvtstat(struct stat *st, struct ostat *ost);
689void	cvtnstat(struct stat *sb, struct nstat *nsb);
690int	getnewvnode(const char *tag, struct mount *mp, vop_t **vops,
691	    struct vnode **vpp);
692int	lease_check(struct vop_lease_args *ap);
693int	spec_vnoperate(struct vop_generic_args *);
694int	speedup_syncer(void);
695#define textvp_fullpath(p, rb, rfb) \
696	vn_fullpath(FIRST_THREAD_IN_PROC(p), (p)->p_textvp, rb, rfb)
697int	vn_fullpath(struct thread *td, struct vnode *vn,
698	    char **retbuf, char **freebuf);
699int	vaccess(enum vtype type, mode_t file_mode, uid_t uid, gid_t gid,
700	    mode_t acc_mode, struct ucred *cred, int *privused);
701int	vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
702	    gid_t file_gid, struct acl *acl, mode_t acc_mode,
703	    struct ucred *cred, int *privused);
704void	vattr_null(struct vattr *vap);
705int	vcount(struct vnode *vp);
706void	vdrop(struct vnode *);
707int	vfinddev(dev_t dev, enum vtype type, struct vnode **vpp);
708void	vfs_add_vnodeops(const void *);
709void	vfs_rm_vnodeops(const void *);
710int	vflush(struct mount *mp, int rootrefs, int flags);
711int	vget(struct vnode *vp, int lockflag, struct thread *td);
712void	vgone(struct vnode *vp);
713void	vgonel(struct vnode *vp, struct thread *td);
714void	vhold(struct vnode *);
715int	vinvalbuf(struct vnode *vp, int save, struct ucred *cred,
716	    struct thread *td, int slpflag, int slptimeo);
717int	vtruncbuf(struct vnode *vp, struct ucred *cred, struct thread *td,
718	    off_t length, int blksize);
719void	vprint(char *label, struct vnode *vp);
720int	vrecycle(struct vnode *vp, struct mtx *inter_lkp,
721	    struct thread *td);
722int	vn_close(struct vnode *vp,
723	    int flags, struct ucred *file_cred, struct thread *td);
724void	vn_finished_write(struct mount *mp);
725int	vn_isdisk(struct vnode *vp, int *errp);
726int	vn_lock(struct vnode *vp, int flags, struct thread *td);
727#ifdef	DEBUG_LOCKS
728int	debug_vn_lock(struct vnode *vp, int flags, struct thread *p,
729	    const char *filename, int line);
730#define vn_lock(vp,flags,p) debug_vn_lock(vp,flags,p,__FILE__,__LINE__)
731#endif
732int	vn_open(struct nameidata *ndp, int *flagp, int cmode);
733int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
734	    struct ucred *cred);
735void	vn_pollevent(struct vnode *vp, int events);
736void	vn_pollgone(struct vnode *vp);
737int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
738int	vn_rdwr(enum uio_rw rw, struct vnode *vp, caddr_t base,
739	    int len, off_t offset, enum uio_seg segflg, int ioflg,
740	    struct ucred *active_cred, struct ucred *file_cred, int *aresid,
741	    struct thread *td);
742int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, caddr_t base,
743	    int len, off_t offset, enum uio_seg segflg, int ioflg,
744	    struct ucred *active_cred, struct ucred *file_cred, int *aresid,
745	    struct thread *td);
746int	vn_stat(struct vnode *vp, struct stat *sb, struct ucred *active_cred,
747	    struct ucred *file_cred, struct thread *td);
748int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
749dev_t	vn_todev(struct vnode *vp);
750int	vn_write_suspend_wait(struct vnode *vp, struct mount *mp,
751	    int flags);
752int	vn_writechk(struct vnode *vp);
753int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
754	    const char *attrname, int *buflen, char *buf, struct thread *td);
755int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
756	    const char *attrname, int buflen, char *buf, struct thread *td);
757int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
758	    const char *attrname, struct thread *td);
759int	vfs_cache_lookup(struct vop_lookup_args *ap);
760int	vfs_object_create(struct vnode *vp, struct thread *td,
761	    struct ucred *cred);
762void	vfs_timestamp(struct timespec *);
763void	vfs_write_resume(struct mount *mp);
764void	vfs_write_suspend(struct mount *mp);
765int	vop_stdbmap(struct vop_bmap_args *);
766int	vop_stdgetwritemount(struct vop_getwritemount_args *);
767int	vop_stdgetpages(struct vop_getpages_args *);
768int	vop_stdinactive(struct vop_inactive_args *);
769int	vop_stdislocked(struct vop_islocked_args *);
770int	vop_stdlock(struct vop_lock_args *);
771int	vop_stdputpages(struct vop_putpages_args *);
772int	vop_stdunlock(struct vop_unlock_args *);
773int	vop_noislocked(struct vop_islocked_args *);
774int	vop_nolock(struct vop_lock_args *);
775int	vop_nopoll(struct vop_poll_args *);
776int	vop_nounlock(struct vop_unlock_args *);
777int	vop_stdpathconf(struct vop_pathconf_args *);
778int	vop_stdpoll(struct vop_poll_args *);
779int	vop_revoke(struct vop_revoke_args *);
780int	vop_sharedlock(struct vop_lock_args *);
781int	vop_eopnotsupp(struct vop_generic_args *ap);
782int	vop_ebadf(struct vop_generic_args *ap);
783int	vop_einval(struct vop_generic_args *ap);
784int	vop_enotty(struct vop_generic_args *ap);
785int	vop_defaultop(struct vop_generic_args *ap);
786int	vop_null(struct vop_generic_args *ap);
787int	vop_panic(struct vop_generic_args *ap);
788int	vop_stdcreatevobject(struct vop_createvobject_args *ap);
789int	vop_stddestroyvobject(struct vop_destroyvobject_args *ap);
790int	vop_stdgetvobject(struct vop_getvobject_args *ap);
791
792void	vfree(struct vnode *);
793void	vput(struct vnode *vp);
794void	vrele(struct vnode *vp);
795void	vref(struct vnode *vp);
796void	vbusy(struct vnode *vp);
797void 	v_addpollinfo(struct vnode *vp);
798
799extern	vop_t **default_vnodeop_p;
800extern	vop_t **spec_vnodeop_p;
801extern	vop_t **dead_vnodeop_p;
802
803#endif /* _KERNEL */
804
805#endif /* !_SYS_VNODE_H_ */
806