xref: /freebsd/sys/sys/vnode.h (revision d653aaec)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1989, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the University nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  *
31  *	@(#)vnode.h	8.7 (Berkeley) 2/4/94
32  * $FreeBSD$
33  */
34 
35 #ifndef _SYS_VNODE_H_
36 #define	_SYS_VNODE_H_
37 
38 #include <sys/bufobj.h>
39 #include <sys/queue.h>
40 #include <sys/lock.h>
41 #include <sys/lockmgr.h>
42 #include <sys/mutex.h>
43 #include <sys/rangelock.h>
44 #include <sys/selinfo.h>
45 #include <sys/uio.h>
46 #include <sys/acl.h>
47 #include <sys/ktr.h>
48 #include <sys/_seqc.h>
49 
50 /*
51  * The vnode is the focus of all file activity in UNIX.  There is a
52  * unique vnode allocated for each active file, each current directory,
53  * each mounted-on file, text file, and the root.
54  */
55 
56 /*
57  * Vnode types.  VNON means no type.
58  */
59 enum vtype	{ VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD,
60 		  VMARKER };
61 
62 enum vgetstate	{ VGET_NONE, VGET_HOLDCNT, VGET_USECOUNT };
63 /*
64  * Each underlying filesystem allocates its own private area and hangs
65  * it from v_data.  If non-null, this area is freed in getnewvnode().
66  */
67 
68 struct namecache;
69 struct cache_fpl;
70 
71 struct vpollinfo {
72 	struct	mtx vpi_lock;		/* lock to protect below */
73 	struct	selinfo vpi_selinfo;	/* identity of poller(s) */
74 	short	vpi_events;		/* what they are looking for */
75 	short	vpi_revents;		/* what has happened */
76 };
77 
78 /*
79  * Reading or writing any of these items requires holding the appropriate lock.
80  *
81  * Lock reference:
82  *	c - namecache mutex
83  *	i - interlock
84  *	l - mp mnt_listmtx or freelist mutex
85  *	I - updated with atomics, 0->1 and 1->0 transitions with interlock held
86  *	m - mount point interlock
87  *	p - pollinfo lock
88  *	u - Only a reference to the vnode is needed to read.
89  *	v - vnode lock
90  *
91  * Vnodes may be found on many lists.  The general way to deal with operating
92  * on a vnode that is on a list is:
93  *	1) Lock the list and find the vnode.
94  *	2) Lock interlock so that the vnode does not go away.
95  *	3) Unlock the list to avoid lock order reversals.
96  *	4) vget with LK_INTERLOCK and check for ENOENT, or
97  *	5) Check for DOOMED if the vnode lock is not required.
98  *	6) Perform your operation, then vput().
99  */
100 
101 #if defined(_KERNEL) || defined(_KVM_VNODE)
102 
103 struct vnode {
104 	/*
105 	 * Fields which define the identity of the vnode.  These fields are
106 	 * owned by the filesystem (XXX: and vgone() ?)
107 	 */
108 	enum	vtype v_type:8;			/* u vnode type */
109 	short	v_irflag;			/* i frequently read flags */
110 	seqc_t	v_seqc;				/* i modification count */
111 	uint32_t v_nchash;			/* u namecache hash */
112 	u_int	v_hash;
113 	struct	vop_vector *v_op;		/* u vnode operations vector */
114 	void	*v_data;			/* u private data for fs */
115 
116 	/*
117 	 * Filesystem instance stuff
118 	 */
119 	struct	mount *v_mount;			/* u ptr to vfs we are in */
120 	TAILQ_ENTRY(vnode) v_nmntvnodes;	/* m vnodes for mount point */
121 
122 	/*
123 	 * Type specific fields, only one applies to any given vnode.
124 	 */
125 	union {
126 		struct mount	*v_mountedhere;	/* v ptr to mountpoint (VDIR) */
127 		struct unpcb	*v_unpcb;	/* v unix domain net (VSOCK) */
128 		struct cdev	*v_rdev; 	/* v device (VCHR, VBLK) */
129 		struct fifoinfo	*v_fifoinfo;	/* v fifo (VFIFO) */
130 	};
131 
132 	/*
133 	 * vfs_hash: (mount + inode) -> vnode hash.  The hash value
134 	 * itself is grouped with other int fields, to avoid padding.
135 	 */
136 	LIST_ENTRY(vnode)	v_hashlist;
137 
138 	/*
139 	 * VFS_namecache stuff
140 	 */
141 	LIST_HEAD(, namecache) v_cache_src;	/* c Cache entries from us */
142 	TAILQ_HEAD(, namecache) v_cache_dst;	/* c Cache entries to us */
143 	struct namecache *v_cache_dd;		/* c Cache entry for .. vnode */
144 
145 	/*
146 	 * Locking
147 	 */
148 	struct	lock v_lock;			/* u (if fs don't have one) */
149 	struct	mtx v_interlock;		/* lock for "i" things */
150 	struct	lock *v_vnlock;			/* u pointer to vnode lock */
151 
152 	/*
153 	 * The machinery of being a vnode
154 	 */
155 	TAILQ_ENTRY(vnode) v_vnodelist;		/* l vnode lists */
156 	TAILQ_ENTRY(vnode) v_lazylist;		/* l vnode lazy list */
157 	struct bufobj	v_bufobj;		/* * Buffer cache object */
158 
159 	/*
160 	 * Hooks for various subsystems and features.
161 	 */
162 	struct vpollinfo *v_pollinfo;		/* i Poll events, p for *v_pi */
163 	struct label *v_label;			/* MAC label for vnode */
164 	struct lockf *v_lockf;		/* Byte-level advisory lock list */
165 	struct rangelock v_rl;			/* Byte-range lock */
166 
167 	u_int	v_holdcnt;			/* I prevents recycling. */
168 	u_int	v_usecount;			/* I ref count of users */
169 	u_short	v_iflag;			/* i vnode flags (see below) */
170 	u_short	v_vflag;			/* v vnode flags */
171 	u_short	v_mflag;			/* l mnt-specific vnode flags */
172 	short	v_dbatchcpu;			/* i LRU requeue deferral batch */
173 	int	v_writecount;			/* I ref count of writers or
174 						   (negative) text users */
175 	int	v_seqc_users;			/* i modifications pending */
176 };
177 
178 #ifndef DEBUG_LOCKS
179 #ifdef _LP64
180 /*
181  * Not crossing 448 bytes fits 9 vnodes per page. If you have to add fields
182  * to the structure and there is nothing which can be done to prevent growth
183  * then so be it. But don't grow it without a good reason.
184  */
185 _Static_assert(sizeof(struct vnode) <= 448, "vnode size crosses 448 bytes");
186 #endif
187 #endif
188 
189 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
190 
191 #define	bo2vnode(bo)	__containerof((bo), struct vnode, v_bufobj)
192 
193 /* XXX: These are temporary to avoid a source sweep at this time */
194 #define v_object	v_bufobj.bo_object
195 
196 /*
197  * Userland version of struct vnode, for sysctl.
198  */
199 struct xvnode {
200 	size_t	xv_size;			/* sizeof(struct xvnode) */
201 	void	*xv_vnode;			/* address of real vnode */
202 	u_long	xv_flag;			/* vnode vflags */
203 	int	xv_usecount;			/* reference count of users */
204 	int	xv_writecount;			/* reference count of writers */
205 	int	xv_holdcnt;			/* page & buffer references */
206 	u_long	xv_id;				/* capability identifier */
207 	void	*xv_mount;			/* address of parent mount */
208 	long	xv_numoutput;			/* num of writes in progress */
209 	enum	vtype xv_type;			/* vnode type */
210 	union {
211 		void	*xvu_socket;		/* unpcb, if VSOCK */
212 		void	*xvu_fifo;		/* fifo, if VFIFO */
213 		dev_t	xvu_rdev;		/* maj/min, if VBLK/VCHR */
214 		struct {
215 			dev_t	xvu_dev;	/* device, if VDIR/VREG/VLNK */
216 			ino_t	xvu_ino;	/* id, if VDIR/VREG/VLNK */
217 		} xv_uns;
218 	} xv_un;
219 };
220 #define xv_socket	xv_un.xvu_socket
221 #define xv_fifo		xv_un.xvu_fifo
222 #define xv_rdev		xv_un.xvu_rdev
223 #define xv_dev		xv_un.xv_uns.xvu_dev
224 #define xv_ino		xv_un.xv_uns.xvu_ino
225 
226 /* We don't need to lock the knlist */
227 #define	VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL ||	\
228 	    KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
229 
230 #define VN_KNOTE(vp, b, a)					\
231 	do {							\
232 		if (!VN_KNLIST_EMPTY(vp))			\
233 			KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
234 			    (a) | KNF_NOKQLOCK);		\
235 	} while (0)
236 #define	VN_KNOTE_LOCKED(vp, b)		VN_KNOTE(vp, b, KNF_LISTLOCKED)
237 #define	VN_KNOTE_UNLOCKED(vp, b)	VN_KNOTE(vp, b, 0)
238 
239 /*
240  * Vnode flags.
241  *	VI flags are protected by interlock and live in v_iflag
242  *	VV flags are protected by the vnode lock and live in v_vflag
243  *
244  *	VIRF_DOOMED is doubly protected by the interlock and vnode lock.  Both
245  *	are required for writing but the status may be checked with either.
246  */
247 #define	VHOLD_NO_SMR	(1<<29)	/* Disable vhold_smr */
248 #define VHOLD_ALL_FLAGS (VHOLD_NO_SMR)
249 
250 #define	VIRF_DOOMED	0x0001	/* This vnode is being recycled */
251 #define	VIRF_PGREAD	0x0002	/* Direct reads from the page cache are permitted,
252 				   never cleared once set */
253 #define	VIRF_MOUNTPOINT	0x0004	/* This vnode is mounted on */
254 #define	VIRF_TEXT_REF	0x0008	/* Executable mappings ref the vnode */
255 
256 #define	VI_UNUSED0	0x0001	/* unused */
257 #define	VI_MOUNT	0x0002	/* Mount in progress */
258 #define	VI_DOINGINACT	0x0004	/* VOP_INACTIVE is in progress */
259 #define	VI_OWEINACT	0x0008	/* Need to call inactive */
260 #define	VI_DEFINACT	0x0010	/* deferred inactive */
261 #define	VI_FOPENING	0x0020	/* In open, with opening process having the
262 				   first right to advlock file */
263 
264 #define	VV_ROOT		0x0001	/* root of its filesystem */
265 #define	VV_ISTTY	0x0002	/* vnode represents a tty */
266 #define	VV_NOSYNC	0x0004	/* unlinked, stop syncing */
267 #define	VV_ETERNALDEV	0x0008	/* device that is never destroyed */
268 #define	VV_CACHEDLABEL	0x0010	/* Vnode has valid cached MAC label */
269 #define	VV_VMSIZEVNLOCK	0x0020	/* object size check requires vnode lock */
270 #define	VV_COPYONWRITE	0x0040	/* vnode is doing copy-on-write */
271 #define	VV_SYSTEM	0x0080	/* vnode being used by kernel */
272 #define	VV_PROCDEP	0x0100	/* vnode is process dependent */
273 #define	VV_UNLINKED	0x0200	/* unlinked but stil open directory */
274 #define	VV_DELETED	0x0400	/* should be removed */
275 #define	VV_MD		0x0800	/* vnode backs the md device */
276 #define	VV_FORCEINSMQ	0x1000	/* force the insmntque to succeed */
277 #define	VV_READLINK	0x2000	/* fdescfs linux vnode */
278 #define	VV_UNREF	0x4000	/* vunref, do not drop lock in inactive() */
279 
280 #define	VMP_LAZYLIST	0x0001	/* Vnode is on mnt's lazy list */
281 
282 /*
283  * Vnode attributes.  A field value of VNOVAL represents a field whose value
284  * is unavailable (getattr) or which is not to be changed (setattr).
285  */
286 struct vattr {
287 	enum vtype	va_type;	/* vnode type (for create) */
288 	u_short		va_mode;	/* files access mode and type */
289 	u_short		va_padding0;
290 	uid_t		va_uid;		/* owner user id */
291 	gid_t		va_gid;		/* owner group id */
292 	nlink_t		va_nlink;	/* number of references to file */
293 	dev_t		va_fsid;	/* filesystem id */
294 	ino_t		va_fileid;	/* file id */
295 	u_quad_t	va_size;	/* file size in bytes */
296 	long		va_blocksize;	/* blocksize preferred for i/o */
297 	struct timespec	va_atime;	/* time of last access */
298 	struct timespec	va_mtime;	/* time of last modification */
299 	struct timespec	va_ctime;	/* time file changed */
300 	struct timespec	va_birthtime;	/* time file created */
301 	u_long		va_gen;		/* generation number of file */
302 	u_long		va_flags;	/* flags defined for file */
303 	dev_t		va_rdev;	/* device the special file represents */
304 	u_quad_t	va_bytes;	/* bytes of disk space held by file */
305 	u_quad_t	va_filerev;	/* file modification number */
306 	u_int		va_vaflags;	/* operations flags, see below */
307 	long		va_spare;	/* remain quad aligned */
308 };
309 
310 /*
311  * Flags for va_vaflags.
312  */
313 #define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
314 #define	VA_EXCLUSIVE	0x02		/* exclusive create request */
315 #define	VA_SYNC		0x04		/* O_SYNC truncation */
316 
317 /*
318  * Flags for ioflag. (high 16 bits used to ask for read-ahead and
319  * help with write clustering)
320  * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
321  */
322 #define	IO_UNIT		0x0001		/* do I/O as atomic unit */
323 #define	IO_APPEND	0x0002		/* append write to end */
324 #define	IO_NDELAY	0x0004		/* FNDELAY flag set in file table */
325 #define	IO_NODELOCKED	0x0008		/* underlying node already locked */
326 #define	IO_ASYNC	0x0010		/* bawrite rather then bdwrite */
327 #define	IO_VMIO		0x0020		/* data already in VMIO space */
328 #define	IO_INVAL	0x0040		/* invalidate after I/O */
329 #define	IO_SYNC		0x0080		/* do I/O synchronously */
330 #define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
331 #define	IO_NOREUSE	0x0200		/* VMIO data won't be reused */
332 #define	IO_EXT		0x0400		/* operate on external attributes */
333 #define	IO_NORMAL	0x0800		/* operate on regular data */
334 #define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
335 #define	IO_BUFLOCKED	0x2000		/* ffs flag; indir buf is locked */
336 #define	IO_RANGELOCKED	0x4000		/* range locked */
337 #define	IO_DATASYNC	0x8000		/* do only data I/O synchronously */
338 
339 #define IO_SEQMAX	0x7F		/* seq heuristic max value */
340 #define IO_SEQSHIFT	16		/* seq heuristic in upper 16 bits */
341 
342 /*
343  * Flags for accmode_t.
344  */
345 #define	VEXEC			000000000100 /* execute/search permission */
346 #define	VWRITE			000000000200 /* write permission */
347 #define	VREAD			000000000400 /* read permission */
348 #define	VADMIN			000000010000 /* being the file owner */
349 #define	VAPPEND			000000040000 /* permission to write/append */
350 /*
351  * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
352  * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
353  * and 0 otherwise.  This never happens with ordinary unix access rights
354  * or POSIX.1e ACLs.  Obviously, VEXPLICIT_DENY must be OR-ed with
355  * some other V* constant.
356  */
357 #define	VEXPLICIT_DENY		000000100000
358 #define	VREAD_NAMED_ATTRS 	000000200000 /* not used */
359 #define	VWRITE_NAMED_ATTRS 	000000400000 /* not used */
360 #define	VDELETE_CHILD	 	000001000000
361 #define	VREAD_ATTRIBUTES 	000002000000 /* permission to stat(2) */
362 #define	VWRITE_ATTRIBUTES 	000004000000 /* change {m,c,a}time */
363 #define	VDELETE		 	000010000000
364 #define	VREAD_ACL	 	000020000000 /* read ACL and file mode */
365 #define	VWRITE_ACL	 	000040000000 /* change ACL and/or file mode */
366 #define	VWRITE_OWNER	 	000100000000 /* change file owner */
367 #define	VSYNCHRONIZE	 	000200000000 /* not used */
368 #define	VCREAT			000400000000 /* creating new file */
369 #define	VVERIFY			001000000000 /* verification required */
370 
371 /*
372  * Permissions that were traditionally granted only to the file owner.
373  */
374 #define VADMIN_PERMS	(VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
375     VWRITE_OWNER)
376 
377 /*
378  * Permissions that were traditionally granted to everyone.
379  */
380 #define VSTAT_PERMS	(VREAD_ATTRIBUTES | VREAD_ACL)
381 
382 /*
383  * Permissions that allow to change the state of the file in any way.
384  */
385 #define VMODIFY_PERMS	(VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
386     VDELETE)
387 
388 /*
389  * Token indicating no attribute value yet assigned.
390  */
391 #define	VNOVAL	(-1)
392 
393 /*
394  * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
395  */
396 #define VLKTIMEOUT	(hz / 20 + 1)
397 
398 #ifdef _KERNEL
399 
400 #ifdef MALLOC_DECLARE
401 MALLOC_DECLARE(M_VNODE);
402 #endif
403 
404 extern u_int ncsizefactor;
405 extern const u_int io_hold_cnt;
406 
407 /*
408  * Convert between vnode types and inode formats (since POSIX.1
409  * defines mode word of stat structure in terms of inode formats).
410  */
411 extern enum vtype	iftovt_tab[];
412 extern int		vttoif_tab[];
413 #define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
414 #define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
415 #define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
416 
417 /*
418  * Flags to various vnode functions.
419  */
420 #define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
421 #define	FORCECLOSE	0x0002	/* vflush: force file closure */
422 #define	WRITECLOSE	0x0004	/* vflush: only close writable files */
423 #define	EARLYFLUSH	0x0008	/* vflush: early call for ffs_flushfiles */
424 #define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
425 #define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
426 #define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
427 #define	V_CLEANONLY	0x0008	/* vinvalbuf: invalidate only clean bufs */
428 #define	V_VMIO		0x0010	/* vinvalbuf: called during pageout */
429 #define	V_ALLOWCLEAN	0x0020	/* vinvalbuf: allow clean buffers after flush */
430 #define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
431 #define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
432 #define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
433 #define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
434 #define	V_PCATCH	0x0008	/* vn_start_write: make the sleep interruptible */
435 #define	V_VALID_FLAGS (V_WAIT | V_NOWAIT | V_XSLEEP | V_PCATCH)
436 
437 #define	VR_START_WRITE	0x0001	/* vfs_write_resume: start write atomically */
438 #define	VR_NO_SUSPCLR	0x0002	/* vfs_write_resume: do not clear suspension */
439 
440 #define	VS_SKIP_UNMOUNT	0x0001	/* vfs_write_suspend: fail if the
441 				   filesystem is being unmounted */
442 
443 #define	VREF(vp)	vref(vp)
444 
445 #ifdef DIAGNOSTIC
446 #define	VATTR_NULL(vap)	vattr_null(vap)
447 #else
448 #define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
449 #endif /* DIAGNOSTIC */
450 
451 #define	NULLVP	((struct vnode *)NULL)
452 
453 /*
454  * Global vnode data.
455  */
456 extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
457 extern	struct mount *rootdevmp;	/* "/dev" mount */
458 extern	u_long desiredvnodes;		/* number of vnodes desired */
459 extern	struct uma_zone *namei_zone;
460 extern	struct vattr va_null;		/* predefined null vattr structure */
461 
462 extern	u_int vn_lock_pair_pause_max;
463 
464 #define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
465 #define	VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
466 #define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
467 #define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
468 #define	VI_MTX(vp)	(&(vp)->v_interlock)
469 
470 #define	VN_LOCK_AREC(vp)	lockallowrecurse((vp)->v_vnlock)
471 #define	VN_LOCK_ASHARE(vp)	lockallowshare((vp)->v_vnlock)
472 #define	VN_LOCK_DSHARE(vp)	lockdisableshare((vp)->v_vnlock)
473 
474 #endif /* _KERNEL */
475 
476 /*
477  * Mods for extensibility.
478  */
479 
480 /*
481  * Flags for vdesc_flags:
482  */
483 #define	VDESC_MAX_VPS		16
484 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
485 #define	VDESC_VP0_WILLRELE	0x0001
486 #define	VDESC_VP1_WILLRELE	0x0002
487 #define	VDESC_VP2_WILLRELE	0x0004
488 #define	VDESC_VP3_WILLRELE	0x0008
489 
490 /*
491  * A generic structure.
492  * This can be used by bypass routines to identify generic arguments.
493  */
494 struct vop_generic_args {
495 	struct vnodeop_desc *a_desc;
496 	/* other random data follows, presumably */
497 };
498 
499 typedef int vop_bypass_t(struct vop_generic_args *);
500 
501 /*
502  * VDESC_NO_OFFSET is used to identify the end of the offset list
503  * and in places where no such field exists.
504  */
505 #define VDESC_NO_OFFSET -1
506 
507 /*
508  * This structure describes the vnode operation taking place.
509  */
510 struct vnodeop_desc {
511 	char	*vdesc_name;		/* a readable name for debugging */
512 	int	 vdesc_flags;		/* VDESC_* flags */
513 	int	vdesc_vop_offset;
514 	vop_bypass_t	*vdesc_call;	/* Function to call */
515 
516 	/*
517 	 * These ops are used by bypass routines to map and locate arguments.
518 	 * Creds and procs are not needed in bypass routines, but sometimes
519 	 * they are useful to (for example) transport layers.
520 	 * Nameidata is useful because it has a cred in it.
521 	 */
522 	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
523 	int	vdesc_vpp_offset;	/* return vpp location */
524 	int	vdesc_cred_offset;	/* cred location, if any */
525 	int	vdesc_thread_offset;	/* thread location, if any */
526 	int	vdesc_componentname_offset; /* if any */
527 };
528 
529 #ifdef _KERNEL
530 /*
531  * A list of all the operation descs.
532  */
533 extern struct vnodeop_desc *vnodeop_descs[];
534 
535 #define	VOPARG_OFFSETOF(s_type, field)	__offsetof(s_type, field)
536 #define	VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
537     ((s_type)(((char*)(struct_p)) + (s_offset)))
538 
539 #ifdef DEBUG_VFS_LOCKS
540 /*
541  * Support code to aid in debugging VFS locking problems.  Not totally
542  * reliable since if the thread sleeps between changing the lock
543  * state and checking it with the assert, some other thread could
544  * change the state.  They are good enough for debugging a single
545  * filesystem using a single-threaded test.  Note that the unreliability is
546  * limited to false negatives; efforts were made to ensure that false
547  * positives cannot occur.
548  */
549 void	assert_vi_locked(struct vnode *vp, const char *str);
550 void	assert_vi_unlocked(struct vnode *vp, const char *str);
551 void	assert_vop_elocked(struct vnode *vp, const char *str);
552 void	assert_vop_locked(struct vnode *vp, const char *str);
553 void	assert_vop_unlocked(struct vnode *vp, const char *str);
554 
555 #define	ASSERT_VI_LOCKED(vp, str)	assert_vi_locked((vp), (str))
556 #define	ASSERT_VI_UNLOCKED(vp, str)	assert_vi_unlocked((vp), (str))
557 #define	ASSERT_VOP_ELOCKED(vp, str)	assert_vop_elocked((vp), (str))
558 #define	ASSERT_VOP_LOCKED(vp, str)	assert_vop_locked((vp), (str))
559 #define	ASSERT_VOP_UNLOCKED(vp, str)	assert_vop_unlocked((vp), (str))
560 
561 #define ASSERT_VOP_IN_SEQC(vp)	do {				\
562 	struct vnode *_vp = (vp);				\
563 								\
564 	VNPASS(seqc_in_modify(_vp->v_seqc), _vp);		\
565 } while (0)
566 
567 #define ASSERT_VOP_NOT_IN_SEQC(vp)	do {			\
568 	struct vnode *_vp = (vp);				\
569 								\
570 	VNPASS(!seqc_in_modify(_vp->v_seqc), _vp);		\
571 } while (0)
572 
573 #else /* !DEBUG_VFS_LOCKS */
574 
575 #define	ASSERT_VI_LOCKED(vp, str)	((void)0)
576 #define	ASSERT_VI_UNLOCKED(vp, str)	((void)0)
577 #define	ASSERT_VOP_ELOCKED(vp, str)	((void)0)
578 #define	ASSERT_VOP_LOCKED(vp, str)	((void)0)
579 #define	ASSERT_VOP_UNLOCKED(vp, str)	((void)0)
580 
581 #define ASSERT_VOP_IN_SEQC(vp)		((void)0)
582 #define ASSERT_VOP_NOT_IN_SEQC(vp)	((void)0)
583 
584 #endif /* DEBUG_VFS_LOCKS */
585 
586 /*
587  * This call works for vnodes in the kernel.
588  */
589 #define VCALL(c) ((c)->a_desc->vdesc_call(c))
590 
591 #define DOINGASYNC(vp)	   					\
592 	(((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 &&	\
593 	 ((curthread->td_pflags & TDP_SYNCIO) == 0))
594 
595 /*
596  * VMIO support inline
597  */
598 
599 extern int vmiodirenable;
600 
601 static __inline int
602 vn_canvmio(struct vnode *vp)
603 {
604       if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
605 		return(TRUE);
606 	return(FALSE);
607 }
608 
609 /*
610  * Finally, include the default set of vnode operations.
611  */
612 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
613 #include "vnode_if.h"
614 
615 /* vn_open_flags */
616 #define	VN_OPEN_NOAUDIT		0x00000001
617 #define	VN_OPEN_NOCAPCHECK	0x00000002
618 #define	VN_OPEN_NAMECACHE	0x00000004
619 #define	VN_OPEN_INVFS		0x00000008
620 #define	VN_OPEN_WANTIOCTLCAPS	0x00000010
621 
622 /* copy_file_range kernel flags */
623 #define	COPY_FILE_RANGE_KFLAGS		0xff000000
624 #define	COPY_FILE_RANGE_TIMEO1SEC	0x01000000	/* Return after 1sec. */
625 
626 /*
627  * Public vnode manipulation functions.
628  */
629 struct componentname;
630 struct file;
631 struct mount;
632 struct nameidata;
633 struct ostat;
634 struct freebsd11_stat;
635 struct thread;
636 struct proc;
637 struct stat;
638 struct nstat;
639 struct ucred;
640 struct uio;
641 struct vattr;
642 struct vfsops;
643 struct vnode;
644 
645 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
646 
647 int	bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
648 	    daddr_t endn);
649 /* cache_* may belong in namei.h. */
650 void	cache_changesize(u_long newhashsize);
651 
652 #define	VFS_CACHE_DROPOLD	0x1
653 
654 void	cache_enter_time_flags(struct vnode *dvp, struct vnode *vp,
655 	    struct componentname *cnp, struct timespec *tsp,
656 	    struct timespec *dtsp, int flags);
657 #define	cache_enter(dvp, vp, cnp)					\
658 	cache_enter_time(dvp, vp, cnp, NULL, NULL)
659 void	cache_enter_time(struct vnode *dvp, struct vnode *vp,
660 	    struct componentname *cnp, struct timespec *tsp,
661 	    struct timespec *dtsp);
662 int	cache_lookup(struct vnode *dvp, struct vnode **vpp,
663 	    struct componentname *cnp, struct timespec *tsp, int *ticksp);
664 void	cache_vnode_init(struct vnode *vp);
665 void	cache_purge(struct vnode *vp);
666 void	cache_purge_vgone(struct vnode *vp);
667 void	cache_purge_negative(struct vnode *vp);
668 void	cache_purgevfs(struct mount *mp);
669 char	*cache_symlink_alloc(size_t size, int flags);
670 void	cache_symlink_free(char *string, size_t size);
671 int	cache_symlink_resolve(struct cache_fpl *fpl, const char *string,
672 	    size_t len);
673 void	cache_vop_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp,
674     struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp);
675 void	cache_vop_rmdir(struct vnode *dvp, struct vnode *vp);
676 #ifdef INVARIANTS
677 void	cache_validate(struct vnode *dvp, struct vnode *vp,
678 	    struct componentname *cnp);
679 void	cache_assert_no_entries(struct vnode *vp);
680 #else
681 static inline void
682 cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
683 {
684 }
685 
686 static inline void
687 cache_assert_no_entries(struct vnode *vp)
688 {
689 }
690 #endif
691 void	cache_fast_lookup_enabled_recalc(void);
692 int	change_dir(struct vnode *vp, struct thread *td);
693 void	cvtstat(struct stat *st, struct ostat *ost);
694 int	freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb);
695 int	freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost);
696 int	getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
697 	    struct vnode **vpp);
698 void	getnewvnode_reserve(void);
699 void	getnewvnode_drop_reserve(void);
700 int	insmntque(struct vnode *vp, struct mount *mp);
701 int	insmntque1(struct vnode *vp, struct mount *mp);
702 u_quad_t init_va_filerev(void);
703 int	speedup_syncer(void);
704 int	vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen);
705 int	vn_getcwd(char *buf, char **retbuf, size_t *buflen);
706 int	vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf);
707 int	vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf);
708 int	vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp,
709 	    const char *hdrl_name, size_t hrdl_name_length, char **retbuf,
710 	    char **freebuf, size_t *buflen);
711 struct vnode *
712 	vn_dir_dd_ino(struct vnode *vp);
713 int	vn_commname(struct vnode *vn, char *buf, u_int buflen);
714 int	vn_path_to_global_path(struct thread *td, struct vnode *vp,
715 	    char *path, u_int pathlen);
716 int	vaccess(enum vtype type, mode_t file_mode, uid_t file_uid,
717 	    gid_t file_gid, accmode_t accmode, struct ucred *cred);
718 int	vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid,
719 	    struct ucred *cred);
720 int	vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid,
721 	    struct acl *aclp, accmode_t accmode, struct ucred *cred);
722 int	vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
723 	    gid_t file_gid, struct acl *acl, accmode_t accmode,
724 	    struct ucred *cred);
725 void	vattr_null(struct vattr *vap);
726 void	vlazy(struct vnode *);
727 void	vdrop(struct vnode *);
728 void	vdropl(struct vnode *);
729 int	vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
730 int	vget(struct vnode *vp, int flags);
731 enum vgetstate	vget_prep_smr(struct vnode *vp);
732 enum vgetstate	vget_prep(struct vnode *vp);
733 int	vget_finish(struct vnode *vp, int flags, enum vgetstate vs);
734 void	vget_finish_ref(struct vnode *vp, enum vgetstate vs);
735 void	vget_abort(struct vnode *vp, enum vgetstate vs);
736 void	vgone(struct vnode *vp);
737 void	vhold(struct vnode *);
738 void	vholdnz(struct vnode *);
739 bool	vhold_smr(struct vnode *);
740 int	vinactive(struct vnode *vp);
741 int	vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
742 int	vtruncbuf(struct vnode *vp, off_t length, int blksize);
743 void	v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn,
744 	    int blksize);
745 void	vunref(struct vnode *);
746 void	vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
747 int	vrecycle(struct vnode *vp);
748 int	vrecyclel(struct vnode *vp);
749 int	vn_bmap_seekhole_locked(struct vnode *vp, u_long cmd, off_t *off,
750 	    struct ucred *cred);
751 int	vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
752 	    struct ucred *cred);
753 int	vn_close(struct vnode *vp,
754 	    int flags, struct ucred *file_cred, struct thread *td);
755 int	vn_copy_file_range(struct vnode *invp, off_t *inoffp,
756 	    struct vnode *outvp, off_t *outoffp, size_t *lenp,
757 	    unsigned int flags, struct ucred *incred, struct ucred *outcred,
758 	    struct thread *fsize_td);
759 int	vn_deallocate(struct vnode *vp, off_t *offset, off_t *length, int flags,
760 	    int ioflg, struct ucred *active_cred, struct ucred *file_cred);
761 void	vn_finished_write(struct mount *mp);
762 void	vn_finished_secondary_write(struct mount *mp);
763 int	vn_fsync_buf(struct vnode *vp, int waitfor);
764 int	vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp,
765 	    struct vnode *outvp, off_t *outoffp, size_t *lenp,
766 	    unsigned int flags, struct ucred *incred, struct ucred *outcred,
767 	    struct thread *fsize_td);
768 int	vn_need_pageq_flush(struct vnode *vp);
769 bool	vn_isdisk_error(struct vnode *vp, int *errp);
770 bool	vn_isdisk(struct vnode *vp);
771 int	_vn_lock(struct vnode *vp, int flags, const char *file, int line);
772 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
773 void	vn_lock_pair(struct vnode *vp1, bool vp1_locked, struct vnode *vp2,
774 	    bool vp2_locked);
775 int	vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
776 int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
777 	    u_int vn_open_flags, struct ucred *cred, struct file *fp);
778 int	vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
779 	    struct thread *td, struct file *fp);
780 void	vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
781 void	vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start,
782 	    vm_pindex_t end);
783 int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
784 int	vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
785 	    int len, off_t offset, enum uio_seg segflg, int ioflg,
786 	    struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
787 	    struct thread *td);
788 int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
789 	    size_t len, off_t offset, enum uio_seg segflg, int ioflg,
790 	    struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
791 	    struct thread *td);
792 int	vn_read_from_obj(struct vnode *vp, struct uio *uio);
793 int	vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio,
794 	    struct thread *td);
795 int	vn_rlimit_fsizex(const struct vnode *vp, struct uio *uio,
796 	    off_t maxfsz, ssize_t *resid_adj, struct thread *td);
797 void	vn_rlimit_fsizex_res(struct uio *uio, ssize_t resid_adj);
798 int	vn_rlimit_trunc(u_quad_t size, struct thread *td);
799 int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
800 int	vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
801 	    int flags);
802 int	vn_truncate_locked(struct vnode *vp, off_t length, bool sync,
803 	    struct ucred *cred);
804 int	vn_writechk(struct vnode *vp);
805 int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
806 	    const char *attrname, int *buflen, char *buf, struct thread *td);
807 int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
808 	    const char *attrname, int buflen, char *buf, struct thread *td);
809 int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
810 	    const char *attrname, struct thread *td);
811 int	vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
812 	    struct vnode **rvp);
813 int	vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
814 	    void *alloc_arg, int lkflags, struct vnode **rvp);
815 int	vn_utimes_perm(struct vnode *vp, struct vattr *vap,
816 	    struct ucred *cred, struct thread *td);
817 
818 int	vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
819 int	vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
820 	    struct uio *uio);
821 
822 void	vn_seqc_write_begin_locked(struct vnode *vp);
823 void	vn_seqc_write_begin(struct vnode *vp);
824 void	vn_seqc_write_end_locked(struct vnode *vp);
825 void	vn_seqc_write_end(struct vnode *vp);
826 #define	vn_seqc_read_any(vp)		seqc_read_any(&(vp)->v_seqc)
827 #define	vn_seqc_read_notmodify(vp)	seqc_read_notmodify(&(vp)->v_seqc)
828 #define	vn_seqc_consistent(vp, seq)	seqc_consistent(&(vp)->v_seqc, seq)
829 
830 #define	vn_rangelock_unlock(vp, cookie)					\
831 	rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp))
832 #define	vn_rangelock_unlock_range(vp, cookie, start, end)		\
833 	rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), 	\
834 	    VI_MTX(vp))
835 #define	vn_rangelock_rlock(vp, start, end)				\
836 	rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
837 #define	vn_rangelock_tryrlock(vp, start, end)				\
838 	rangelock_tryrlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
839 #define	vn_rangelock_wlock(vp, start, end)				\
840 	rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
841 #define	vn_rangelock_trywlock(vp, start, end)				\
842 	rangelock_trywlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
843 
844 #define	vn_irflag_read(vp)	atomic_load_short(&(vp)->v_irflag)
845 void	vn_irflag_set_locked(struct vnode *vp, short toset);
846 void	vn_irflag_set(struct vnode *vp, short toset);
847 void	vn_irflag_set_cond_locked(struct vnode *vp, short toset);
848 void	vn_irflag_set_cond(struct vnode *vp, short toset);
849 void	vn_irflag_unset_locked(struct vnode *vp, short tounset);
850 void	vn_irflag_unset(struct vnode *vp, short tounset);
851 
852 int	vfs_cache_lookup(struct vop_lookup_args *ap);
853 int	vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp);
854 void	vfs_timestamp(struct timespec *);
855 void	vfs_write_resume(struct mount *mp, int flags);
856 int	vfs_write_suspend(struct mount *mp, int flags);
857 int	vfs_write_suspend_umnt(struct mount *mp);
858 struct vnode *vnlru_alloc_marker(void);
859 void	vnlru_free_marker(struct vnode *);
860 void	vnlru_free_vfsops(int, struct vfsops *, struct vnode *);
861 int	vop_stdbmap(struct vop_bmap_args *);
862 int	vop_stdfdatasync_buf(struct vop_fdatasync_args *);
863 int	vop_stdfsync(struct vop_fsync_args *);
864 int	vop_stdgetwritemount(struct vop_getwritemount_args *);
865 int	vop_stdgetpages(struct vop_getpages_args *);
866 int	vop_stdinactive(struct vop_inactive_args *);
867 int	vop_stdioctl(struct vop_ioctl_args *);
868 int	vop_stdneed_inactive(struct vop_need_inactive_args *);
869 int	vop_stdkqfilter(struct vop_kqfilter_args *);
870 int	vop_stdlock(struct vop_lock1_args *);
871 int	vop_stdunlock(struct vop_unlock_args *);
872 int	vop_stdislocked(struct vop_islocked_args *);
873 int	vop_lock(struct vop_lock1_args *);
874 int	vop_unlock(struct vop_unlock_args *);
875 int	vop_islocked(struct vop_islocked_args *);
876 int	vop_stdputpages(struct vop_putpages_args *);
877 int	vop_nopoll(struct vop_poll_args *);
878 int	vop_stdaccess(struct vop_access_args *ap);
879 int	vop_stdaccessx(struct vop_accessx_args *ap);
880 int	vop_stdadvise(struct vop_advise_args *ap);
881 int	vop_stdadvlock(struct vop_advlock_args *ap);
882 int	vop_stdadvlockasync(struct vop_advlockasync_args *ap);
883 int	vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
884 int	vop_stdallocate(struct vop_allocate_args *ap);
885 int	vop_stddeallocate(struct vop_deallocate_args *ap);
886 int	vop_stdset_text(struct vop_set_text_args *ap);
887 int	vop_stdpathconf(struct vop_pathconf_args *);
888 int	vop_stdpoll(struct vop_poll_args *);
889 int	vop_stdvptocnp(struct vop_vptocnp_args *ap);
890 int	vop_stdvptofh(struct vop_vptofh_args *ap);
891 int	vop_stdunp_bind(struct vop_unp_bind_args *ap);
892 int	vop_stdunp_connect(struct vop_unp_connect_args *ap);
893 int	vop_stdunp_detach(struct vop_unp_detach_args *ap);
894 int	vop_stdadd_writecount_nomsync(struct vop_add_writecount_args *ap);
895 int	vop_eopnotsupp(struct vop_generic_args *ap);
896 int	vop_ebadf(struct vop_generic_args *ap);
897 int	vop_einval(struct vop_generic_args *ap);
898 int	vop_enoent(struct vop_generic_args *ap);
899 int	vop_enotty(struct vop_generic_args *ap);
900 int	vop_eagain(struct vop_generic_args *ap);
901 int	vop_null(struct vop_generic_args *ap);
902 int	vop_panic(struct vop_generic_args *ap);
903 int	dead_poll(struct vop_poll_args *ap);
904 int	dead_read(struct vop_read_args *ap);
905 int	dead_write(struct vop_write_args *ap);
906 
907 /* These are called from within the actual VOPS. */
908 void	vop_close_post(void *a, int rc);
909 void	vop_create_pre(void *a);
910 void	vop_create_post(void *a, int rc);
911 void	vop_whiteout_pre(void *a);
912 void	vop_whiteout_post(void *a, int rc);
913 void	vop_deleteextattr_pre(void *a);
914 void	vop_deleteextattr_post(void *a, int rc);
915 void	vop_link_pre(void *a);
916 void	vop_link_post(void *a, int rc);
917 void	vop_lookup_post(void *a, int rc);
918 void	vop_lookup_pre(void *a);
919 void	vop_mkdir_pre(void *a);
920 void	vop_mkdir_post(void *a, int rc);
921 void	vop_mknod_pre(void *a);
922 void	vop_mknod_post(void *a, int rc);
923 void	vop_open_post(void *a, int rc);
924 void	vop_read_post(void *a, int rc);
925 void	vop_read_pgcache_post(void *ap, int rc);
926 void	vop_readdir_post(void *a, int rc);
927 void	vop_reclaim_post(void *a, int rc);
928 void	vop_remove_pre(void *a);
929 void	vop_remove_post(void *a, int rc);
930 void	vop_rename_post(void *a, int rc);
931 void	vop_rename_pre(void *a);
932 void	vop_rmdir_pre(void *a);
933 void	vop_rmdir_post(void *a, int rc);
934 void	vop_setattr_pre(void *a);
935 void	vop_setattr_post(void *a, int rc);
936 void	vop_setacl_pre(void *a);
937 void	vop_setacl_post(void *a, int rc);
938 void	vop_setextattr_pre(void *a);
939 void	vop_setextattr_post(void *a, int rc);
940 void	vop_symlink_pre(void *a);
941 void	vop_symlink_post(void *a, int rc);
942 int	vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a);
943 
944 #ifdef DEBUG_VFS_LOCKS
945 void	vop_fdatasync_debugpre(void *a);
946 void	vop_fdatasync_debugpost(void *a, int rc);
947 void	vop_fplookup_vexec_debugpre(void *a);
948 void	vop_fplookup_vexec_debugpost(void *a, int rc);
949 void	vop_fplookup_symlink_debugpre(void *a);
950 void	vop_fplookup_symlink_debugpost(void *a, int rc);
951 void	vop_fsync_debugpre(void *a);
952 void	vop_fsync_debugpost(void *a, int rc);
953 void	vop_strategy_debugpre(void *a);
954 void	vop_lock_debugpre(void *a);
955 void	vop_lock_debugpost(void *a, int rc);
956 void	vop_unlock_debugpre(void *a);
957 void	vop_need_inactive_debugpre(void *a);
958 void	vop_need_inactive_debugpost(void *a, int rc);
959 void	vop_mkdir_debugpost(void *a, int rc);
960 #else
961 #define	vop_fdatasync_debugpre(x)		do { } while (0)
962 #define	vop_fdatasync_debugpost(x, y)		do { } while (0)
963 #define	vop_fplookup_vexec_debugpre(x)		do { } while (0)
964 #define	vop_fplookup_vexec_debugpost(x, y)	do { } while (0)
965 #define	vop_fplookup_symlink_debugpre(x)	do { } while (0)
966 #define	vop_fplookup_symlink_debugpost(x, y)	do { } while (0)
967 #define	vop_fsync_debugpre(x)			do { } while (0)
968 #define	vop_fsync_debugpost(x, y)		do { } while (0)
969 #define	vop_strategy_debugpre(x)		do { } while (0)
970 #define	vop_lock_debugpre(x)			do { } while (0)
971 #define	vop_lock_debugpost(x, y)		do { } while (0)
972 #define	vop_unlock_debugpre(x)			do { } while (0)
973 #define	vop_need_inactive_debugpre(x)		do { } while (0)
974 #define	vop_need_inactive_debugpost(x, y)	do { } while (0)
975 #define	vop_mkdir_debugpost(x, y)		do { } while (0)
976 #endif
977 
978 void	vop_rename_fail(struct vop_rename_args *ap);
979 
980 #define	vop_stat_helper_pre(ap)	({						\
981 	struct vop_stat_args *_ap = (ap);					\
982 	int _error;								\
983 	AUDIT_ARG_VNODE1(ap->a_vp);						\
984 	_error = mac_vnode_check_stat(_ap->a_active_cred, _ap->a_file_cred, _ap->a_vp);\
985 	if (__predict_true(_error == 0)) {					\
986 		ap->a_sb->st_padding0 = 0;					\
987 		ap->a_sb->st_padding1 = 0;					\
988 		bzero(_ap->a_sb->st_spare, sizeof(_ap->a_sb->st_spare));	\
989 	}									\
990 	_error;									\
991 })
992 
993 #define	vop_stat_helper_post(ap, error)	({					\
994 	struct vop_stat_args *_ap = (ap);					\
995 	int _error = (error);							\
996 	if (priv_check_cred_vfs_generation(_ap->a_active_cred))			\
997 		_ap->a_sb->st_gen = 0;						\
998 	_error;									\
999 })
1000 
1001 #define	VOP_WRITE_PRE(ap)						\
1002 	struct vattr va;						\
1003 	int error;							\
1004 	off_t osize, ooffset, noffset;					\
1005 									\
1006 	osize = ooffset = noffset = 0;					\
1007 	if (!VN_KNLIST_EMPTY((ap)->a_vp)) {				\
1008 		error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred);	\
1009 		if (error)						\
1010 			return (error);					\
1011 		ooffset = (ap)->a_uio->uio_offset;			\
1012 		osize = (off_t)va.va_size;				\
1013 	}
1014 
1015 #define VOP_WRITE_POST(ap, ret)						\
1016 	noffset = (ap)->a_uio->uio_offset;				\
1017 	if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) {	\
1018 		VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE			\
1019 		    | (noffset > osize ? NOTE_EXTEND : 0));		\
1020 	}
1021 
1022 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
1023 
1024 #ifdef INVARIANTS
1025 #define	VOP_ADD_WRITECOUNT_CHECKED(vp, cnt)				\
1026 do {									\
1027 	int error_;							\
1028 									\
1029 	error_ = VOP_ADD_WRITECOUNT((vp), (cnt));			\
1030 	VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d",	\
1031 	    error_));							\
1032 } while (0)
1033 #define	VOP_SET_TEXT_CHECKED(vp)					\
1034 do {									\
1035 	int error_;							\
1036 									\
1037 	error_ = VOP_SET_TEXT((vp));					\
1038 	VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d",	\
1039 	    error_));							\
1040 } while (0)
1041 #define	VOP_UNSET_TEXT_CHECKED(vp)					\
1042 do {									\
1043 	int error_;							\
1044 									\
1045 	error_ = VOP_UNSET_TEXT((vp));					\
1046 	VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d",	\
1047 	    error_));							\
1048 } while (0)
1049 #else
1050 #define	VOP_ADD_WRITECOUNT_CHECKED(vp, cnt)	VOP_ADD_WRITECOUNT((vp), (cnt))
1051 #define	VOP_SET_TEXT_CHECKED(vp)		VOP_SET_TEXT((vp))
1052 #define	VOP_UNSET_TEXT_CHECKED(vp)		VOP_UNSET_TEXT((vp))
1053 #endif
1054 
1055 #define	VN_IS_DOOMED(vp)	__predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0)
1056 
1057 void	vput(struct vnode *vp);
1058 void	vrele(struct vnode *vp);
1059 void	vref(struct vnode *vp);
1060 void	vrefact(struct vnode *vp);
1061 void 	v_addpollinfo(struct vnode *vp);
1062 static __inline int
1063 vrefcnt(struct vnode *vp)
1064 {
1065 
1066 	return (vp->v_usecount);
1067 }
1068 
1069 #define	vholdl(vp)	do {						\
1070 	ASSERT_VI_LOCKED(vp, __func__);					\
1071 	vhold(vp);							\
1072 } while (0)
1073 
1074 #define	vrefl(vp)	do {						\
1075 	ASSERT_VI_LOCKED(vp, __func__);					\
1076 	vref(vp);							\
1077 } while (0)
1078 
1079 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
1080 void vnode_destroy_vobject(struct vnode *vp);
1081 
1082 extern struct vop_vector fifo_specops;
1083 extern struct vop_vector dead_vnodeops;
1084 extern struct vop_vector default_vnodeops;
1085 
1086 #define VOP_PANIC	((void*)(uintptr_t)vop_panic)
1087 #define VOP_NULL	((void*)(uintptr_t)vop_null)
1088 #define VOP_EBADF	((void*)(uintptr_t)vop_ebadf)
1089 #define VOP_ENOTTY	((void*)(uintptr_t)vop_enotty)
1090 #define VOP_EINVAL	((void*)(uintptr_t)vop_einval)
1091 #define VOP_ENOENT	((void*)(uintptr_t)vop_enoent)
1092 #define VOP_EOPNOTSUPP	((void*)(uintptr_t)vop_eopnotsupp)
1093 #define VOP_EAGAIN	((void*)(uintptr_t)vop_eagain)
1094 
1095 /* fifo_vnops.c */
1096 int	fifo_printinfo(struct vnode *);
1097 
1098 /* vfs_hash.c */
1099 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
1100 
1101 void vfs_hash_changesize(u_long newhashsize);
1102 int vfs_hash_get(const struct mount *mp, u_int hash, int flags,
1103     struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1104 u_int vfs_hash_index(struct vnode *vp);
1105 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
1106     struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1107 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
1108     struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1109 void vfs_hash_rehash(struct vnode *vp, u_int hash);
1110 void vfs_hash_remove(struct vnode *vp);
1111 
1112 int vfs_kqfilter(struct vop_kqfilter_args *);
1113 struct dirent;
1114 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
1115 int vfs_emptydir(struct vnode *vp);
1116 
1117 int vfs_unixify_accmode(accmode_t *accmode);
1118 
1119 void vfs_unp_reclaim(struct vnode *vp);
1120 
1121 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
1122 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
1123     gid_t gid);
1124 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
1125     struct thread *td);
1126 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
1127     struct thread *td);
1128 
1129 void vn_fsid(struct vnode *vp, struct vattr *va);
1130 
1131 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp);
1132 int vn_lktype_write(struct mount *mp, struct vnode *vp);
1133 
1134 #define VOP_UNLOCK_FLAGS(vp, flags)	({				\
1135 	struct vnode *_vp = (vp);					\
1136 	int _flags = (flags);						\
1137 	int _error;							\
1138 									\
1139         if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0)		\
1140                 panic("%s: unsupported flags %x\n", __func__, flags);	\
1141         _error = VOP_UNLOCK(_vp);					\
1142         if (_flags & LK_INTERLOCK)					\
1143                 VI_UNLOCK(_vp);						\
1144         _error;								\
1145 })
1146 
1147 #include <sys/kernel.h>
1148 
1149 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \
1150 	SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \
1151 	    vfs_vector_op_register, &vnodeops)
1152 
1153 #define VFS_SMR_DECLARE				\
1154 	extern smr_t vfs_smr
1155 
1156 #define VFS_SMR()	vfs_smr
1157 #define vfs_smr_enter()	smr_enter(VFS_SMR())
1158 #define vfs_smr_exit()	smr_exit(VFS_SMR())
1159 #define vfs_smr_synchronize()	smr_synchronize(VFS_SMR())
1160 #define vfs_smr_entered_load(ptr)	smr_entered_load((ptr), VFS_SMR())
1161 #define VFS_SMR_ASSERT_ENTERED()	SMR_ASSERT_ENTERED(VFS_SMR())
1162 #define VFS_SMR_ASSERT_NOT_ENTERED()	SMR_ASSERT_NOT_ENTERED(VFS_SMR())
1163 #define VFS_SMR_ZONE_SET(zone)	uma_zone_set_smr((zone), VFS_SMR())
1164 
1165 #define vn_load_v_data_smr(vp)	({		\
1166 	struct vnode *_vp = (vp);		\
1167 						\
1168 	VFS_SMR_ASSERT_ENTERED();		\
1169 	atomic_load_consume_ptr(&(_vp)->v_data);\
1170 })
1171 
1172 #endif /* _KERNEL */
1173 
1174 #endif /* !_SYS_VNODE_H_ */
1175