xref: /freebsd/sys/sys/vnode.h (revision bc4430dc203ed7b6765fb5df041bf545c50f859b)
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 
32 #ifndef _SYS_VNODE_H_
33 #define	_SYS_VNODE_H_
34 
35 #include <sys/bufobj.h>
36 #include <sys/queue.h>
37 #include <sys/lock.h>
38 #include <sys/lockmgr.h>
39 #include <sys/mutex.h>
40 #include <sys/rangelock.h>
41 #include <sys/selinfo.h>
42 #include <sys/uio.h>
43 #include <sys/acl.h>
44 #include <sys/ktr.h>
45 #include <sys/_seqc.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  */
__enum_uint8_decl(vtype)56 __enum_uint8_decl(vtype) {
57 	VNON,
58 	VREG,
59 	VDIR,
60 	VBLK,
61 	VCHR,
62 	VLNK,
63 	VSOCK,
64 	VFIFO,
65 	VBAD,
66 	VMARKER,
67 	VLASTTYPE = VMARKER,
68 };
69 
__enum_uint8_decl(vstate)70 __enum_uint8_decl(vstate) {
71 	VSTATE_UNINITIALIZED,
72 	VSTATE_CONSTRUCTED,
73 	VSTATE_DESTROYING,
74 	VSTATE_DEAD,
75 	VLASTSTATE = VSTATE_DEAD,
76 };
77 
78 enum vgetstate {
79 	VGET_NONE,
80 	VGET_HOLDCNT,
81 	VGET_USECOUNT,
82 };
83 
84 /*
85  * Each underlying filesystem allocates its own private area and hangs
86  * it from v_data.  If non-null, this area is freed in getnewvnode().
87  */
88 
89 struct cache_fpl;
90 struct inotify_watch;
91 struct namecache;
92 
93 struct vpollinfo {
94 	struct	mtx vpi_lock;		/* lock to protect below */
95 	TAILQ_HEAD(, inotify_watch) vpi_inotify; /* list of inotify watchers */
96 	struct	selinfo vpi_selinfo;	/* identity of poller(s) */
97 	short	vpi_events;		/* what they are looking for */
98 	short	vpi_revents;		/* what has happened */
99 };
100 
101 /*
102  * Reading or writing any of these items requires holding the appropriate lock.
103  *
104  * Lock reference:
105  *	c - namecache mutex
106  *	i - interlock
107  *	l - mp mnt_listmtx or freelist mutex
108  *	I - updated with atomics, 0->1 and 1->0 transitions with interlock held
109  *	m - mount point interlock
110  *	p - pollinfo lock
111  *	u - Only a reference to the vnode is needed to read.
112  *	v - vnode lock
113  *
114  * Vnodes may be found on many lists.  The general way to deal with operating
115  * on a vnode that is on a list is:
116  *	1) Lock the list and find the vnode.
117  *	2) Lock interlock so that the vnode does not go away.
118  *	3) Unlock the list to avoid lock order reversals.
119  *	4) vget with LK_INTERLOCK and check for ENOENT, or
120  *	5) Check for DOOMED if the vnode lock is not required.
121  *	6) Perform your operation, then vput().
122  */
123 
124 #if defined(_KERNEL) || defined(_KVM_VNODE)
125 
126 struct vnode {
127 	/*
128 	 * Fields which define the identity of the vnode.  These fields are
129 	 * owned by the filesystem (XXX: and vgone() ?)
130 	 */
131 	__enum_uint8(vtype) v_type;		/* u vnode type */
132 	__enum_uint8(vstate) v_state;		/* u vnode state */
133 	short	v_irflag;			/* i frequently read flags */
134 	seqc_t	v_seqc;				/* i modification count */
135 	uint32_t v_nchash;			/* u namecache hash */
136 	u_int	v_hash;
137 	const struct vop_vector *v_op;		/* u vnode operations vector */
138 	void	*v_data;			/* u private data for fs */
139 
140 	/*
141 	 * Filesystem instance stuff
142 	 */
143 	struct	mount *v_mount;			/* u ptr to vfs we are in */
144 	TAILQ_ENTRY(vnode) v_nmntvnodes;	/* m vnodes for mount point */
145 
146 	/*
147 	 * Type specific fields, only one applies to any given vnode.
148 	 */
149 	union {
150 		struct mount	*v_mountedhere;	/* v ptr to mountpoint (VDIR) */
151 		struct unpcb	*v_unpcb;	/* v unix domain net (VSOCK) */
152 		struct cdev	*v_rdev; 	/* v device (VCHR, VBLK) */
153 		struct fifoinfo	*v_fifoinfo;	/* v fifo (VFIFO) */
154 	};
155 
156 	/*
157 	 * vfs_hash: (mount + inode) -> vnode hash.  The hash value
158 	 * itself is grouped with other int fields, to avoid padding.
159 	 */
160 	LIST_ENTRY(vnode)	v_hashlist;
161 
162 	/*
163 	 * VFS_namecache stuff
164 	 */
165 	LIST_HEAD(, namecache) v_cache_src;	/* c Cache entries from us */
166 	TAILQ_HEAD(, namecache) v_cache_dst;	/* c Cache entries to us */
167 	struct namecache *v_cache_dd;		/* c Cache entry for .. vnode */
168 
169 	/*
170 	 * Locking
171 	 */
172 	struct	lock v_lock;			/* u (if fs don't have one) */
173 	struct	mtx v_interlock;		/* lock for "i" things */
174 	struct	lock *v_vnlock;			/* u pointer to vnode lock */
175 
176 	/*
177 	 * The machinery of being a vnode
178 	 */
179 	TAILQ_ENTRY(vnode) v_vnodelist;		/* l vnode lists */
180 	TAILQ_ENTRY(vnode) v_lazylist;		/* l vnode lazy list */
181 	struct bufobj	v_bufobj;		/* * Buffer cache object */
182 
183 	/*
184 	 * Hooks for various subsystems and features.
185 	 */
186 	struct vpollinfo *v_pollinfo;		/* i Poll events, p for *v_pi */
187 	struct label *v_label;			/* MAC label for vnode */
188 	struct lockf *v_lockf;		/* Byte-level advisory lock list */
189 	struct rangelock v_rl;			/* Byte-range lock */
190 
191 	u_int	v_holdcnt;			/* I prevents recycling. */
192 	u_int	v_usecount;			/* I ref count of users */
193 	u_short	v_iflag;			/* i vnode flags (see below) */
194 	u_short	v_vflag;			/* v vnode flags */
195 	u_short	v_mflag;			/* l mnt-specific vnode flags */
196 	short	v_dbatchcpu;			/* i LRU requeue deferral batch */
197 	int	v_writecount;			/* I ref count of writers or
198 						   (negative) text users */
199 	int	v_seqc_users;			/* i modifications pending */
200 };
201 
202 #ifndef DEBUG_LOCKS
203 #ifdef _LP64
204 /*
205  * Not crossing 448 bytes fits 9 vnodes per page. If you have to add fields
206  * to the structure and there is nothing which can be done to prevent growth
207  * then so be it. But don't grow it without a good reason.
208  */
209 _Static_assert(sizeof(struct vnode) <= 448, "vnode size crosses 448 bytes");
210 #endif
211 #endif
212 
213 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
214 
215 #define	bo2vnode(bo)	__containerof((bo), struct vnode, v_bufobj)
216 
217 /* XXX: These are temporary to avoid a source sweep at this time */
218 #define v_object	v_bufobj.bo_object
219 
220 /* We don't need to lock the knlist */
221 #define	VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL ||	\
222 	    KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
223 
224 #define VN_KNOTE(vp, b, a)					\
225 	do {							\
226 		if (!VN_KNLIST_EMPTY(vp))			\
227 			KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
228 			    (a) | KNF_NOKQLOCK);		\
229 	} while (0)
230 #define	VN_KNOTE_LOCKED(vp, b)		VN_KNOTE(vp, b, KNF_LISTLOCKED)
231 #define	VN_KNOTE_UNLOCKED(vp, b)	VN_KNOTE(vp, b, 0)
232 
233 /*
234  * Vnode flags.
235  *	VI flags are protected by interlock and live in v_iflag
236  *	VIRF flags are protected by interlock and live in v_irflag
237  *	VV flags are protected by the vnode lock and live in v_vflag
238  *
239  *	VIRF_DOOMED is doubly protected by the interlock and vnode lock.  Both
240  *	are required for writing but the status may be checked with either.
241  */
242 #define	VHOLD_NO_SMR	(1<<29)	/* Disable vhold_smr */
243 #define VHOLD_ALL_FLAGS (VHOLD_NO_SMR)
244 
245 #define	VIRF_DOOMED	0x0001	/* This vnode is being recycled */
246 #define	VIRF_PGREAD	0x0002	/* Direct reads from the page cache are permitted,
247 				   never cleared once set */
248 #define	VIRF_MOUNTPOINT	0x0004	/* This vnode is mounted on */
249 #define	VIRF_TEXT_REF	0x0008	/* Executable mappings ref the vnode */
250 #define	VIRF_CROSSMP	0x0010	/* Cross-mp vnode, no locking */
251 #define	VIRF_NAMEDDIR	0x0020	/* Named attribute directory */
252 #define	VIRF_NAMEDATTR	0x0040	/* Named attribute */
253 #define	VIRF_INOTIFY	0x0080	/* This vnode is being watched */
254 #define	VIRF_INOTIFY_PARENT 0x0100 /* A parent of this vnode may be being
255 				      watched */
256 
257 #define	VI_UNUSED0	0x0001	/* unused */
258 #define	VI_MOUNT	0x0002	/* Mount in progress */
259 #define	VI_DOINGINACT	0x0004	/* VOP_INACTIVE is in progress */
260 #define	VI_OWEINACT	0x0008	/* Need to call inactive */
261 #define	VI_DEFINACT	0x0010	/* deferred inactive */
262 #define	VI_FOPENING	0x0020	/* In open, with opening process having the
263 				   first right to advlock file */
264 
265 #define	VV_ROOT		0x0001	/* root of its filesystem */
266 #define	VV_ISTTY	0x0002	/* vnode represents a tty */
267 #define	VV_NOSYNC	0x0004	/* unlinked, stop syncing */
268 #define	VV_ETERNALDEV	0x0008	/* device that is never destroyed */
269 #define	VV_CACHEDLABEL	0x0010	/* Vnode has valid cached MAC label */
270 #define	VV_VMSIZEVNLOCK	0x0020	/* object size check requires vnode lock */
271 #define	VV_COPYONWRITE	0x0040	/* vnode is doing copy-on-write */
272 #define	VV_SYSTEM	0x0080	/* vnode being used by kernel */
273 #define	VV_PROCDEP	0x0100	/* vnode is process dependent */
274 #define	VV_UNLINKED	0x0200	/* unlinked but stil open directory */
275 #define	VV_DELETED	0x0400	/* should be removed */
276 #define	VV_MD		0x0800	/* vnode backs the md device */
277 #define	VV_FORCEINSMQ	0x1000	/* force the insmntque to succeed */
278 #define	VV_READLINK	0x2000	/* fdescfs linux vnode */
279 #define	VV_UNREF	0x4000	/* vunref, do not drop lock in inactive() */
280 #define	VV_CROSSLOCK	0x8000	/* vnode lock is shared w/ root mounted here */
281 
282 #define	VMP_LAZYLIST	0x0001	/* Vnode is on mnt's lazy list */
283 
284 /*
285  * Vnode attributes.  A field value of VNOVAL represents a field whose value
286  * is unavailable (getattr) or which is not to be changed (setattr).
287  */
288 struct vattr {
289 	__enum_uint8(vtype)	va_type;	/* vnode type (for create) */
290 	u_short		va_mode;	/* files access mode and type */
291 	uint16_t	va_bsdflags;	/* same as st_bsdflags from stat(2) */
292 	uid_t		va_uid;		/* owner user id */
293 	gid_t		va_gid;		/* owner group id */
294 	nlink_t		va_nlink;	/* number of references to file */
295 	dev_t		va_fsid;	/* filesystem id */
296 	ino_t		va_fileid;	/* file id */
297 	u_quad_t	va_size;	/* file size in bytes */
298 	long		va_blocksize;	/* blocksize preferred for i/o */
299 	struct timespec	va_atime;	/* time of last access */
300 	struct timespec	va_mtime;	/* time of last modification */
301 	struct timespec	va_ctime;	/* time file changed */
302 	struct timespec	va_birthtime;	/* time file created */
303 	u_long		va_gen;		/* generation number of file */
304 	u_long		va_flags;	/* flags defined for file */
305 	dev_t		va_rdev;	/* device the special file represents */
306 	u_quad_t	va_bytes;	/* bytes of disk space held by file */
307 	u_quad_t	va_filerev;	/* file modification number */
308 	u_int		va_vaflags;	/* operations flags, see below */
309 	long		va_spare;	/* remain quad aligned */
310 };
311 
312 /*
313  * Flags for va_vaflags.
314  */
315 #define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
316 #define	VA_EXCLUSIVE	0x02		/* exclusive create request */
317 #define	VA_SYNC		0x04		/* O_SYNC truncation */
318 
319 /*
320  * Flags for ioflag. (high 16 bits used to ask for read-ahead and
321  * help with write clustering)
322  * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
323  */
324 #define	IO_UNIT		0x0001		/* do I/O as atomic unit */
325 #define	IO_APPEND	0x0002		/* append write to end */
326 #define	IO_NDELAY	0x0004		/* FNDELAY flag set in file table */
327 #define	IO_NODELOCKED	0x0008		/* underlying node already locked */
328 #define	IO_ASYNC	0x0010		/* bawrite rather then bdwrite */
329 #define	IO_VMIO		0x0020		/* data already in VMIO space */
330 #define	IO_INVAL	0x0040		/* invalidate after I/O */
331 #define	IO_SYNC		0x0080		/* do I/O synchronously */
332 #define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
333 #define	IO_NOREUSE	0x0200		/* VMIO data won't be reused */
334 #define	IO_EXT		0x0400		/* operate on external attributes */
335 #define	IO_NORMAL	0x0800		/* operate on regular data */
336 #define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
337 #define	IO_BUFLOCKED	0x2000		/* ffs flag; indir buf is locked */
338 #define	IO_RANGELOCKED	0x4000		/* range locked */
339 #define	IO_DATASYNC	0x8000		/* do only data I/O synchronously */
340 
341 #define IO_SEQMAX	0x7F		/* seq heuristic max value */
342 #define IO_SEQSHIFT	16		/* seq heuristic in upper 16 bits */
343 
344 /*
345  * Flags for accmode_t.
346  */
347 #define	VEXEC			000000000100 /* execute/search permission */
348 #define	VWRITE			000000000200 /* write permission */
349 #define	VREAD			000000000400 /* read permission */
350 #define	VADMIN			000000010000 /* being the file owner */
351 #define	VAPPEND			000000040000 /* permission to write/append */
352 /*
353  * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
354  * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
355  * and 0 otherwise.  This never happens with ordinary unix access rights
356  * or POSIX.1e ACLs.  Obviously, VEXPLICIT_DENY must be OR-ed with
357  * some other V* constant.
358  */
359 #define	VEXPLICIT_DENY		000000100000
360 #define	VREAD_NAMED_ATTRS 	000000200000 /* not used */
361 #define	VWRITE_NAMED_ATTRS 	000000400000 /* not used */
362 #define	VDELETE_CHILD	 	000001000000
363 #define	VREAD_ATTRIBUTES 	000002000000 /* permission to stat(2) */
364 #define	VWRITE_ATTRIBUTES 	000004000000 /* change {m,c,a}time */
365 #define	VDELETE		 	000010000000
366 #define	VREAD_ACL	 	000020000000 /* read ACL and file mode */
367 #define	VWRITE_ACL	 	000040000000 /* change ACL and/or file mode */
368 #define	VWRITE_OWNER	 	000100000000 /* change file owner */
369 #define	VSYNCHRONIZE	 	000200000000 /* not used */
370 #define	VCREAT			000400000000 /* creating new file */
371 #define	VVERIFY			001000000000 /* verification required */
372 
373 /*
374  * Permissions that were traditionally granted only to the file owner.
375  */
376 #define VADMIN_PERMS	(VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
377     VWRITE_OWNER)
378 
379 /*
380  * Permissions that were traditionally granted to everyone.
381  */
382 #define VSTAT_PERMS	(VREAD_ATTRIBUTES | VREAD_ACL)
383 
384 /*
385  * Permissions that allow to change the state of the file in any way.
386  */
387 #define VMODIFY_PERMS	(VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
388     VDELETE)
389 
390 /*
391  * Token indicating no attribute value yet assigned.
392  */
393 #define	VNOVAL	(-1)
394 
395 /*
396  * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
397  */
398 #define VLKTIMEOUT	(hz / 20 + 1)
399 
400 #ifdef _KERNEL
401 
402 #ifdef MALLOC_DECLARE
403 MALLOC_DECLARE(M_VNODE);
404 #endif
405 
406 extern u_int ncsizefactor;
407 extern const u_int io_hold_cnt;
408 
409 /*
410  * Convert between vnode types and inode formats (since POSIX.1
411  * defines mode word of stat structure in terms of inode formats).
412  */
413 extern __enum_uint8(vtype)	iftovt_tab[];
414 extern int		vttoif_tab[];
415 #define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
416 #define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
417 #define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
418 
419 /*
420  * Flags to various vnode functions.
421  */
422 #define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
423 #define	FORCECLOSE	0x0002	/* vflush: force file closure */
424 #define	WRITECLOSE	0x0004	/* vflush: only close writable files */
425 #define	EARLYFLUSH	0x0008	/* vflush: early call for ffs_flushfiles */
426 #define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
427 #define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
428 #define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
429 #define	V_CLEANONLY	0x0008	/* vinvalbuf: invalidate only clean bufs */
430 #define	V_VMIO		0x0010	/* vinvalbuf: called during pageout */
431 #define	V_ALLOWCLEAN	0x0020	/* vinvalbuf: allow clean buffers after flush */
432 #define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
433 #define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
434 #define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
435 #define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
436 #define	V_PCATCH	0x0008	/* vn_start_write: make the sleep interruptible */
437 #define	V_VALID_FLAGS (V_WAIT | V_NOWAIT | V_XSLEEP | V_PCATCH)
438 
439 #define	VR_START_WRITE	0x0001	/* vfs_write_resume: start write atomically */
440 #define	VR_NO_SUSPCLR	0x0002	/* vfs_write_resume: do not clear suspension */
441 
442 #define	VS_SKIP_UNMOUNT	0x0001	/* vfs_write_suspend: fail if the
443 				   filesystem is being unmounted */
444 
445 #define	VREF(vp)	vref(vp)
446 
447 #ifdef DIAGNOSTIC
448 #define	VATTR_NULL(vap)	vattr_null(vap)
449 #else
450 #define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
451 #endif /* DIAGNOSTIC */
452 
453 #define	NULLVP	((struct vnode *)NULL)
454 
455 /*
456  * Global vnode data.
457  */
458 extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
459 extern	struct mount *rootdevmp;	/* "/dev" mount */
460 extern	u_long desiredvnodes;		/* number of vnodes desired */
461 extern	struct uma_zone *namei_zone;
462 extern	struct vattr va_null;		/* predefined null vattr structure */
463 
464 extern	u_int vn_lock_pair_pause_max;
465 
466 #define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
467 #define	VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
468 #define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
469 #define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
470 #define	VI_MTX(vp)	(&(vp)->v_interlock)
471 
472 #define	VN_LOCK_AREC(vp)	lockallowrecurse((vp)->v_vnlock)
473 #define	VN_LOCK_ASHARE(vp)	lockallowshare((vp)->v_vnlock)
474 #define	VN_LOCK_DSHARE(vp)	lockdisableshare((vp)->v_vnlock)
475 
476 #endif /* _KERNEL */
477 
478 /*
479  * Mods for extensibility.
480  */
481 
482 /*
483  * Flags for vdesc_flags:
484  */
485 #define	VDESC_MAX_VPS		16
486 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
487 #define	VDESC_VP0_WILLRELE	0x0001
488 #define	VDESC_VP1_WILLRELE	0x0002
489 #define	VDESC_VP2_WILLRELE	0x0004
490 #define	VDESC_VP3_WILLRELE	0x0008
491 
492 /*
493  * A generic structure.
494  * This can be used by bypass routines to identify generic arguments.
495  */
496 struct vop_generic_args {
497 	struct vnodeop_desc *a_desc;
498 	/* other random data follows, presumably */
499 };
500 
501 typedef int vop_bypass_t(struct vop_generic_args *);
502 
503 /*
504  * VDESC_NO_OFFSET is used to identify the end of the offset list
505  * and in places where no such field exists.
506  */
507 #define VDESC_NO_OFFSET -1
508 
509 /*
510  * This structure describes the vnode operation taking place.
511  */
512 struct vnodeop_desc {
513 	char	*vdesc_name;		/* a readable name for debugging */
514 	int	 vdesc_flags;		/* VDESC_* flags */
515 	int	vdesc_vop_offset;
516 	vop_bypass_t	*vdesc_call;	/* Function to call */
517 
518 	/*
519 	 * These ops are used by bypass routines to map and locate arguments.
520 	 * Creds and procs are not needed in bypass routines, but sometimes
521 	 * they are useful to (for example) transport layers.
522 	 * Nameidata is useful because it has a cred in it.
523 	 */
524 	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
525 	int	vdesc_vpp_offset;	/* return vpp location */
526 	int	vdesc_cred_offset;	/* cred location, if any */
527 	int	vdesc_thread_offset;	/* thread location, if any */
528 	int	vdesc_componentname_offset; /* if any */
529 };
530 
531 #ifdef _KERNEL
532 /*
533  * A list of all the operation descs.
534  */
535 extern struct vnodeop_desc *vnodeop_descs[];
536 
537 #define	VOPARG_OFFSETOF(s_type, field)	__offsetof(s_type, field)
538 #define	VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
539     ((s_type)(((char*)(struct_p)) + (s_offset)))
540 
541 #ifdef DEBUG_VFS_LOCKS
542 /*
543  * Support code to aid in debugging VFS locking problems.  Not totally
544  * reliable since if the thread sleeps between changing the lock
545  * state and checking it with the assert, some other thread could
546  * change the state.  They are good enough for debugging a single
547  * filesystem using a single-threaded test.  Note that the unreliability is
548  * limited to false negatives; efforts were made to ensure that false
549  * positives cannot occur.
550  */
551 void	assert_vi_locked(struct vnode *vp, const char *str);
552 void	assert_vi_unlocked(struct vnode *vp, const char *str);
553 void	assert_vop_elocked(struct vnode *vp, const char *str);
554 void	assert_vop_locked(struct vnode *vp, const char *str);
555 void	assert_vop_unlocked(struct vnode *vp, const char *str);
556 
557 #define	ASSERT_VI_LOCKED(vp, str)	assert_vi_locked((vp), (str))
558 #define	ASSERT_VI_UNLOCKED(vp, str)	assert_vi_unlocked((vp), (str))
559 #define	ASSERT_VOP_ELOCKED(vp, str)	assert_vop_elocked((vp), (str))
560 #define	ASSERT_VOP_LOCKED(vp, str)	assert_vop_locked((vp), (str))
561 #define	ASSERT_VOP_UNLOCKED(vp, str)	assert_vop_unlocked((vp), (str))
562 
563 #define ASSERT_VOP_IN_SEQC(vp)	do {				\
564 	struct vnode *_vp = (vp);				\
565 								\
566 	VNPASS(seqc_in_modify(_vp->v_seqc), _vp);		\
567 } while (0)
568 
569 #define ASSERT_VOP_NOT_IN_SEQC(vp)	do {			\
570 	struct vnode *_vp = (vp);				\
571 								\
572 	VNPASS(!seqc_in_modify(_vp->v_seqc), _vp);		\
573 } while (0)
574 
575 #else /* !DEBUG_VFS_LOCKS */
576 
577 #define	ASSERT_VI_LOCKED(vp, str)	((void)0)
578 #define	ASSERT_VI_UNLOCKED(vp, str)	((void)0)
579 #define	ASSERT_VOP_ELOCKED(vp, str)	((void)0)
580 #define	ASSERT_VOP_LOCKED(vp, str)	((void)0)
581 #define	ASSERT_VOP_UNLOCKED(vp, str)	((void)0)
582 
583 #define ASSERT_VOP_IN_SEQC(vp)		((void)0)
584 #define ASSERT_VOP_NOT_IN_SEQC(vp)	((void)0)
585 
586 #endif /* DEBUG_VFS_LOCKS */
587 
588 /*
589  * This call works for vnodes in the kernel.
590  */
591 #define VCALL(c) ((c)->a_desc->vdesc_call(c))
592 
593 #define DOINGASYNC(vp)	   					\
594 	(((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 &&	\
595 	 ((curthread->td_pflags & TDP_SYNCIO) == 0))
596 
597 /*
598  * VMIO support inline
599  */
600 
601 extern int vmiodirenable;
602 
603 static __inline int
vn_canvmio(struct vnode * vp)604 vn_canvmio(struct vnode *vp)
605 {
606       if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
607 		return(TRUE);
608 	return(FALSE);
609 }
610 
611 /*
612  * Finally, include the default set of vnode operations.
613  */
614 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
615 #include "vnode_if.h"
616 
617 /* vn_open_flags */
618 #define	VN_OPEN_NOAUDIT		0x00000001
619 #define	VN_OPEN_NOCAPCHECK	0x00000002
620 #define	VN_OPEN_NAMECACHE	0x00000004
621 #define	VN_OPEN_INVFS		0x00000008
622 #define	VN_OPEN_WANTIOCTLCAPS	0x00000010
623 
624 /* copy_file_range kernel flags */
625 #define	COPY_FILE_RANGE_KFLAGS		0xff000000
626 #define	COPY_FILE_RANGE_TIMEO1SEC	0x01000000	/* Return after 1sec. */
627 
628 /*
629  * Public vnode manipulation functions.
630  */
631 struct componentname;
632 struct file;
633 struct mount;
634 struct nameidata;
635 struct ostat;
636 struct freebsd11_stat;
637 struct thread;
638 struct proc;
639 struct stat;
640 struct nstat;
641 struct ucred;
642 struct uio;
643 struct vattr;
644 struct vfsops;
645 struct vnode;
646 
647 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
648 
649 int	bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
650 	    daddr_t endn);
651 /* cache_* may belong in namei.h. */
652 void	cache_changesize(u_long newhashsize);
653 
654 #define	VFS_CACHE_DROPOLD	0x1
655 
656 void	cache_enter_time_flags(struct vnode *dvp, struct vnode *vp,
657 	    struct componentname *cnp, struct timespec *tsp,
658 	    struct timespec *dtsp, int flags);
659 #define	cache_enter(dvp, vp, cnp)					\
660 	cache_enter_time(dvp, vp, cnp, NULL, NULL)
661 void	cache_enter_time(struct vnode *dvp, struct vnode *vp,
662 	    struct componentname *cnp, struct timespec *tsp,
663 	    struct timespec *dtsp);
664 int	cache_lookup(struct vnode *dvp, struct vnode **vpp,
665 	    struct componentname *cnp, struct timespec *tsp, int *ticksp);
666 void	cache_vnode_init(struct vnode *vp);
667 void	cache_purge(struct vnode *vp);
668 void	cache_purge_vgone(struct vnode *vp);
669 void	cache_purge_negative(struct vnode *vp);
670 void	cache_purgevfs(struct mount *mp);
671 char	*cache_symlink_alloc(size_t size, int flags);
672 void	cache_symlink_free(char *string, size_t size);
673 int	cache_symlink_resolve(struct cache_fpl *fpl, const char *string,
674 	    size_t len);
675 void	cache_vop_inotify(struct vnode *vp, int event, uint32_t cookie);
676 void	cache_vop_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp,
677     struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp);
678 void	cache_vop_rmdir(struct vnode *dvp, struct vnode *vp);
679 void	cache_vop_vector_register(struct vop_vector *);
680 #ifdef INVARIANTS
681 void	cache_validate(struct vnode *dvp, struct vnode *vp,
682 	    struct componentname *cnp);
683 void	cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops);
684 void	cache_assert_no_entries(struct vnode *vp);
685 #else
686 static inline void
cache_validate(struct vnode * dvp,struct vnode * vp,struct componentname * cnp)687 cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
688 {
689 }
690 
691 static inline void
cache_validate_vop_vector(struct mount * mp,struct vop_vector * vops)692 cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops)
693 {
694 }
695 
696 static inline void
cache_assert_no_entries(struct vnode * vp)697 cache_assert_no_entries(struct vnode *vp)
698 {
699 }
700 #endif
701 void	cache_fast_lookup_enabled_recalc(void);
702 int	change_dir(struct vnode *vp, struct thread *td);
703 void	cvtstat(struct stat *st, struct ostat *ost);
704 int	freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb);
705 int	freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost);
706 int	getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
707 	    struct vnode **vpp);
708 void	getnewvnode_reserve(void);
709 void	getnewvnode_drop_reserve(void);
710 int	insmntque(struct vnode *vp, struct mount *mp);
711 int	insmntque1(struct vnode *vp, struct mount *mp);
712 u_quad_t init_va_filerev(void);
713 int	speedup_syncer(void);
714 int	vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen);
715 int	vn_getcwd(char *buf, char **retbuf, size_t *buflen);
716 int	vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf);
717 int	vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf);
718 int	vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp,
719 	    const char *hdrl_name, size_t hrdl_name_length, char **retbuf,
720 	    char **freebuf, size_t *buflen);
721 struct vnode *
722 	vn_dir_dd_ino(struct vnode *vp);
723 int	vn_commname(struct vnode *vn, char *buf, u_int buflen);
724 int	vn_path_to_global_path(struct thread *td, struct vnode *vp,
725 	    char *path, u_int pathlen);
726 int	vn_path_to_global_path_hardlink(struct thread *td, struct vnode *vp,
727 	    struct vnode *dvp, char *path, u_int pathlen, const char *leaf_name,
728 	    size_t leaf_length);
729 int	vaccess(__enum_uint8(vtype) type, mode_t file_mode, uid_t file_uid,
730 	    gid_t file_gid, accmode_t accmode, struct ucred *cred);
731 int	vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid,
732 	    struct ucred *cred);
733 int	vaccess_acl_nfs4(__enum_uint8(vtype) type, uid_t file_uid, gid_t file_gid,
734 	    struct acl *aclp, accmode_t accmode, struct ucred *cred);
735 int	vaccess_acl_posix1e(__enum_uint8(vtype) type, uid_t file_uid,
736 	    gid_t file_gid, struct acl *acl, accmode_t accmode,
737 	    struct ucred *cred);
738 void	vattr_null(struct vattr *vap);
739 void	vlazy(struct vnode *);
740 void	vdrop(struct vnode *);
741 void	vdropl(struct vnode *);
742 int	vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
743 int	vget(struct vnode *vp, int flags);
744 enum vgetstate	vget_prep_smr(struct vnode *vp);
745 enum vgetstate	vget_prep(struct vnode *vp);
746 int	vget_finish(struct vnode *vp, int flags, enum vgetstate vs);
747 void	vget_finish_ref(struct vnode *vp, enum vgetstate vs);
748 void	vget_abort(struct vnode *vp, enum vgetstate vs);
749 void	vgone(struct vnode *vp);
750 void	vhold(struct vnode *);
751 void	vholdnz(struct vnode *);
752 bool	vhold_smr(struct vnode *);
753 int	vinactive(struct vnode *vp);
754 int	vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
755 int	vtruncbuf(struct vnode *vp, off_t length, int blksize);
756 void	v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn,
757 	    int blksize);
758 void	vunref(struct vnode *);
759 void	vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
760 int	vrecycle(struct vnode *vp);
761 int	vrecyclel(struct vnode *vp);
762 int	vn_bmap_seekhole_locked(struct vnode *vp, u_long cmd, off_t *off,
763 	    struct ucred *cred);
764 int	vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
765 	    struct ucred *cred);
766 int	vn_close(struct vnode *vp,
767 	    int flags, struct ucred *file_cred, struct thread *td);
768 int	vn_copy_file_range(struct vnode *invp, off_t *inoffp,
769 	    struct vnode *outvp, off_t *outoffp, size_t *lenp,
770 	    unsigned int flags, struct ucred *incred, struct ucred *outcred,
771 	    struct thread *fsize_td);
772 int	vn_deallocate(struct vnode *vp, off_t *offset, off_t *length, int flags,
773 	    int ioflg, struct ucred *active_cred, struct ucred *file_cred);
774 void	vn_finished_write(struct mount *mp);
775 void	vn_finished_secondary_write(struct mount *mp);
776 int	vn_fsync_buf(struct vnode *vp, int waitfor);
777 int	vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp,
778 	    struct vnode *outvp, off_t *outoffp, size_t *lenp,
779 	    unsigned int flags, struct ucred *incred, struct ucred *outcred,
780 	    struct thread *fsize_td);
781 int	vn_need_pageq_flush(struct vnode *vp);
782 bool	vn_isdisk_error(struct vnode *vp, int *errp);
783 bool	vn_isdisk(struct vnode *vp);
784 int	_vn_lock(struct vnode *vp, int flags, const char *file, int line);
785 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
786 void	vn_lock_pair(struct vnode *vp1, bool vp1_locked, int lkflags1,
787 	    struct vnode *vp2, bool vp2_locked, int lkflags2);
788 int	vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
789 int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
790 	    u_int vn_open_flags, struct ucred *cred, struct file *fp);
791 int	vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
792 	    struct thread *td, struct file *fp);
793 void	vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
794 void	vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start,
795 	    vm_pindex_t end);
796 int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
797 int	vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
798 	    int len, off_t offset, enum uio_seg segflg, int ioflg,
799 	    struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
800 	    struct thread *td);
801 int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
802 	    size_t len, off_t offset, enum uio_seg segflg, int ioflg,
803 	    struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
804 	    struct thread *td);
805 int	vn_read_from_obj(struct vnode *vp, struct uio *uio);
806 int	vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio,
807 	    struct thread *td);
808 int	vn_rlimit_fsizex(const struct vnode *vp, struct uio *uio,
809 	    off_t maxfsz, ssize_t *resid_adj, struct thread *td);
810 void	vn_rlimit_fsizex_res(struct uio *uio, ssize_t resid_adj);
811 int	vn_rlimit_trunc(u_quad_t size, struct thread *td);
812 int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
813 int	vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
814 	    int flags);
815 int	vn_truncate_locked(struct vnode *vp, off_t length, bool sync,
816 	    struct ucred *cred);
817 int	vn_writechk(struct vnode *vp);
818 int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
819 	    const char *attrname, int *buflen, char *buf, struct thread *td);
820 int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
821 	    const char *attrname, int buflen, char *buf, struct thread *td);
822 int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
823 	    const char *attrname, struct thread *td);
824 int	vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
825 	    struct vnode **rvp);
826 int	vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
827 	    void *alloc_arg, int lkflags, struct vnode **rvp);
828 int	vn_utimes_perm(struct vnode *vp, struct vattr *vap,
829 	    struct ucred *cred, struct thread *td);
830 int	vn_cmp(struct file *, struct file *, struct thread *td);
831 
832 int	vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
833 int	vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
834 	    struct uio *uio);
835 
836 void	vn_seqc_write_begin_locked(struct vnode *vp);
837 void	vn_seqc_write_begin(struct vnode *vp);
838 void	vn_seqc_write_end_locked(struct vnode *vp);
839 void	vn_seqc_write_end(struct vnode *vp);
840 #define	vn_seqc_read_any(vp)		seqc_read_any(&(vp)->v_seqc)
841 #define	vn_seqc_read_notmodify(vp)	seqc_read_notmodify(&(vp)->v_seqc)
842 #define	vn_seqc_consistent(vp, seq)	seqc_consistent(&(vp)->v_seqc, seq)
843 
844 #define	vn_rangelock_unlock(vp, cookie)					\
845 	rangelock_unlock(&(vp)->v_rl, (cookie))
846 #define	vn_rangelock_rlock(vp, start, end)				\
847 	rangelock_rlock(&(vp)->v_rl, (start), (end))
848 #define	vn_rangelock_tryrlock(vp, start, end)				\
849 	rangelock_tryrlock(&(vp)->v_rl, (start), (end))
850 #define	vn_rangelock_wlock(vp, start, end)				\
851 	rangelock_wlock(&(vp)->v_rl, (start), (end))
852 #define	vn_rangelock_trywlock(vp, start, end)				\
853 	rangelock_trywlock(&(vp)->v_rl, (start), (end))
854 
855 #define	vn_irflag_read(vp)	atomic_load_short(&(vp)->v_irflag)
856 void	vn_irflag_set_locked(struct vnode *vp, short toset);
857 void	vn_irflag_set(struct vnode *vp, short toset);
858 void	vn_irflag_set_cond_locked(struct vnode *vp, short toset);
859 void	vn_irflag_set_cond(struct vnode *vp, short toset);
860 void	vn_irflag_unset_locked(struct vnode *vp, short tounset);
861 void	vn_irflag_unset(struct vnode *vp, short tounset);
862 
863 int	vfs_cache_lookup(struct vop_lookup_args *ap);
864 int	vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp);
865 void	vfs_timestamp(struct timespec *);
866 void	vfs_write_resume(struct mount *mp, int flags);
867 int	vfs_write_suspend(struct mount *mp, int flags);
868 int	vfs_write_suspend_umnt(struct mount *mp);
869 struct vnode *vnlru_alloc_marker(void);
870 void	vnlru_free_marker(struct vnode *);
871 void	vnlru_free_vfsops(int, struct vfsops *, struct vnode *);
872 int	vop_stdbmap(struct vop_bmap_args *);
873 int	vop_stdfdatasync_buf(struct vop_fdatasync_args *);
874 int	vop_stdfsync(struct vop_fsync_args *);
875 int	vop_stdgetwritemount(struct vop_getwritemount_args *);
876 int	vop_stdgetpages(struct vop_getpages_args *);
877 int	vop_stdinactive(struct vop_inactive_args *);
878 int	vop_stdneed_inactive(struct vop_need_inactive_args *);
879 int	vop_stdinotify(struct vop_inotify_args *);
880 int	vop_stdinotify_add_watch(struct vop_inotify_add_watch_args *);
881 int	vop_stdioctl(struct vop_ioctl_args *);
882 int	vop_stdkqfilter(struct vop_kqfilter_args *);
883 int	vop_stdlock(struct vop_lock1_args *);
884 int	vop_stdunlock(struct vop_unlock_args *);
885 int	vop_stdislocked(struct vop_islocked_args *);
886 int	vop_lock(struct vop_lock1_args *);
887 int	vop_unlock(struct vop_unlock_args *);
888 int	vop_islocked(struct vop_islocked_args *);
889 int	vop_stdputpages(struct vop_putpages_args *);
890 int	vop_nopoll(struct vop_poll_args *);
891 int	vop_stdaccess(struct vop_access_args *ap);
892 int	vop_stdaccessx(struct vop_accessx_args *ap);
893 int	vop_stdadvise(struct vop_advise_args *ap);
894 int	vop_stdadvlock(struct vop_advlock_args *ap);
895 int	vop_stdadvlockasync(struct vop_advlockasync_args *ap);
896 int	vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
897 int	vop_stdallocate(struct vop_allocate_args *ap);
898 int	vop_stddeallocate(struct vop_deallocate_args *ap);
899 int	vop_stdset_text(struct vop_set_text_args *ap);
900 int	vop_stdpathconf(struct vop_pathconf_args *);
901 int	vop_stdpoll(struct vop_poll_args *);
902 int	vop_stdvptocnp(struct vop_vptocnp_args *ap);
903 int	vop_stdvptofh(struct vop_vptofh_args *ap);
904 int	vop_stdunp_bind(struct vop_unp_bind_args *ap);
905 int	vop_stdunp_connect(struct vop_unp_connect_args *ap);
906 int	vop_stdunp_detach(struct vop_unp_detach_args *ap);
907 int	vop_stdadd_writecount_nomsync(struct vop_add_writecount_args *ap);
908 int	vop_eopnotsupp(struct vop_generic_args *ap);
909 int	vop_ebadf(struct vop_generic_args *ap);
910 int	vop_einval(struct vop_generic_args *ap);
911 int	vop_enoent(struct vop_generic_args *ap);
912 int	vop_enotty(struct vop_generic_args *ap);
913 int	vop_eagain(struct vop_generic_args *ap);
914 int	vop_null(struct vop_generic_args *ap);
915 int	vop_panic(struct vop_generic_args *ap);
916 int	dead_poll(struct vop_poll_args *ap);
917 int	dead_read(struct vop_read_args *ap);
918 int	dead_write(struct vop_write_args *ap);
919 
920 /* These are called from within the actual VOPS. */
921 void	vop_allocate_post(void *a, int rc);
922 void	vop_copy_file_range_post(void *ap, int rc);
923 void	vop_close_post(void *a, int rc);
924 void	vop_create_pre(void *a);
925 void	vop_create_post(void *a, int rc);
926 void	vop_deallocate_post(void *a, int rc);
927 void	vop_whiteout_pre(void *a);
928 void	vop_whiteout_post(void *a, int rc);
929 void	vop_deleteextattr_pre(void *a);
930 void	vop_deleteextattr_post(void *a, int rc);
931 void	vop_link_pre(void *a);
932 void	vop_link_post(void *a, int rc);
933 void	vop_lookup_post(void *a, int rc);
934 void	vop_lookup_pre(void *a);
935 void	vop_mkdir_pre(void *a);
936 void	vop_mkdir_post(void *a, int rc);
937 void	vop_mknod_pre(void *a);
938 void	vop_mknod_post(void *a, int rc);
939 void	vop_open_post(void *a, int rc);
940 void	vop_read_post(void *a, int rc);
941 void	vop_read_pgcache_post(void *ap, int rc);
942 void	vop_readdir_post(void *a, int rc);
943 void	vop_reclaim_post(void *a, int rc);
944 void	vop_remove_pre(void *a);
945 void	vop_remove_post(void *a, int rc);
946 void	vop_rename_post(void *a, int rc);
947 void	vop_rename_pre(void *a);
948 void	vop_rmdir_pre(void *a);
949 void	vop_rmdir_post(void *a, int rc);
950 void	vop_setattr_pre(void *a);
951 void	vop_setattr_post(void *a, int rc);
952 void	vop_setacl_pre(void *a);
953 void	vop_setacl_post(void *a, int rc);
954 void	vop_setextattr_pre(void *a);
955 void	vop_setextattr_post(void *a, int rc);
956 void	vop_symlink_pre(void *a);
957 void	vop_symlink_post(void *a, int rc);
958 int	vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a);
959 
960 #ifdef DEBUG_VFS_LOCKS
961 void	vop_fdatasync_debugpre(void *a);
962 void	vop_fdatasync_debugpost(void *a, int rc);
963 void	vop_fplookup_vexec_debugpre(void *a);
964 void	vop_fplookup_vexec_debugpost(void *a, int rc);
965 void	vop_fplookup_symlink_debugpre(void *a);
966 void	vop_fplookup_symlink_debugpost(void *a, int rc);
967 void	vop_fsync_debugpre(void *a);
968 void	vop_fsync_debugpost(void *a, int rc);
969 void	vop_strategy_debugpre(void *a);
970 void	vop_lock_debugpre(void *a);
971 void	vop_lock_debugpost(void *a, int rc);
972 void	vop_unlock_debugpre(void *a);
973 void	vop_need_inactive_debugpre(void *a);
974 void	vop_need_inactive_debugpost(void *a, int rc);
975 void	vop_mkdir_debugpost(void *a, int rc);
976 #else
977 #define	vop_fdatasync_debugpre(x)		do { } while (0)
978 #define	vop_fdatasync_debugpost(x, y)		do { } while (0)
979 #define	vop_fplookup_vexec_debugpre(x)		do { } while (0)
980 #define	vop_fplookup_vexec_debugpost(x, y)	do { } while (0)
981 #define	vop_fplookup_symlink_debugpre(x)	do { } while (0)
982 #define	vop_fplookup_symlink_debugpost(x, y)	do { } while (0)
983 #define	vop_fsync_debugpre(x)			do { } while (0)
984 #define	vop_fsync_debugpost(x, y)		do { } while (0)
985 #define	vop_strategy_debugpre(x)		do { } while (0)
986 #define	vop_lock_debugpre(x)			do { } while (0)
987 #define	vop_lock_debugpost(x, y)		do { } while (0)
988 #define	vop_unlock_debugpre(x)			do { } while (0)
989 #define	vop_need_inactive_debugpre(x)		do { } while (0)
990 #define	vop_need_inactive_debugpost(x, y)	do { } while (0)
991 #define	vop_mkdir_debugpost(x, y)		do { } while (0)
992 #endif
993 
994 void	vop_rename_fail(struct vop_rename_args *ap);
995 
996 #define	vop_stat_helper_pre(ap)	({						\
997 	struct vop_stat_args *_ap = (ap);					\
998 	int _error;								\
999 	AUDIT_ARG_VNODE1(ap->a_vp);						\
1000 	_error = mac_vnode_check_stat(_ap->a_active_cred, _ap->a_file_cred, _ap->a_vp);\
1001 	if (__predict_true(_error == 0)) {					\
1002 		ap->a_sb->st_padding1 = 0;					\
1003 		bzero(_ap->a_sb->st_spare, sizeof(_ap->a_sb->st_spare));	\
1004 		ap->a_sb->st_filerev = 0;					\
1005 		ap->a_sb->st_bsdflags = 0;					\
1006 	}									\
1007 	_error;									\
1008 })
1009 
1010 #define	vop_stat_helper_post(ap, error)	({					\
1011 	struct vop_stat_args *_ap = (ap);					\
1012 	int _error = (error);							\
1013 	if (priv_check_cred_vfs_generation(_ap->a_active_cred))			\
1014 		_ap->a_sb->st_gen = 0;						\
1015 	_error;									\
1016 })
1017 
1018 #define	VOP_WRITE_PRE(ap)						\
1019 	struct vattr va;						\
1020 	int error;							\
1021 	off_t osize, ooffset, noffset;					\
1022 									\
1023 	osize = ooffset = noffset = 0;					\
1024 	if (!VN_KNLIST_EMPTY((ap)->a_vp)) {				\
1025 		error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred);	\
1026 		if (error)						\
1027 			return (error);					\
1028 		ooffset = (ap)->a_uio->uio_offset;			\
1029 		osize = (off_t)va.va_size;				\
1030 	}
1031 
1032 #define VOP_WRITE_POST(ap, ret)						\
1033 	noffset = (ap)->a_uio->uio_offset;				\
1034 	if (noffset > ooffset) {					\
1035 		if (VN_KNLIST_EMPTY((ap)->a_vp)) {			\
1036 			VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE |	\
1037 			    (noffset > osize ? NOTE_EXTEND : 0));	\
1038 		}							\
1039 		INOTIFY((ap)->a_vp, IN_MODIFY);				\
1040 	}
1041 
1042 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
1043 
1044 #ifdef INVARIANTS
1045 #define	VOP_ADD_WRITECOUNT_CHECKED(vp, cnt)				\
1046 do {									\
1047 	int error_;							\
1048 									\
1049 	error_ = VOP_ADD_WRITECOUNT((vp), (cnt));			\
1050 	VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d",	\
1051 	    error_));							\
1052 } while (0)
1053 #define	VOP_SET_TEXT_CHECKED(vp)					\
1054 do {									\
1055 	int error_;							\
1056 									\
1057 	error_ = VOP_SET_TEXT((vp));					\
1058 	VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d",	\
1059 	    error_));							\
1060 } while (0)
1061 #define	VOP_UNSET_TEXT_CHECKED(vp)					\
1062 do {									\
1063 	int error_;							\
1064 									\
1065 	error_ = VOP_UNSET_TEXT((vp));					\
1066 	VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d",	\
1067 	    error_));							\
1068 } while (0)
1069 #else
1070 #define	VOP_ADD_WRITECOUNT_CHECKED(vp, cnt)	VOP_ADD_WRITECOUNT((vp), (cnt))
1071 #define	VOP_SET_TEXT_CHECKED(vp)		VOP_SET_TEXT((vp))
1072 #define	VOP_UNSET_TEXT_CHECKED(vp)		VOP_UNSET_TEXT((vp))
1073 #endif
1074 
1075 #define	VN_IS_DOOMED(vp)	__predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0)
1076 
1077 void	vput(struct vnode *vp);
1078 void	vrele(struct vnode *vp);
1079 void	vref(struct vnode *vp);
1080 void	vrefact(struct vnode *vp);
1081 void 	v_addpollinfo(struct vnode *vp);
1082 static __inline int
vrefcnt(struct vnode * vp)1083 vrefcnt(struct vnode *vp)
1084 {
1085 
1086 	return (vp->v_usecount);
1087 }
1088 
1089 #define	vholdl(vp)	do {						\
1090 	ASSERT_VI_LOCKED(vp, __func__);					\
1091 	vhold(vp);							\
1092 } while (0)
1093 
1094 #define	vrefl(vp)	do {						\
1095 	ASSERT_VI_LOCKED(vp, __func__);					\
1096 	vref(vp);							\
1097 } while (0)
1098 
1099 /*
1100  * The caller doesn't know the file size and vnode_create_vobject() should
1101  * determine the size on its own.
1102  */
1103 #define	VNODE_NO_SIZE	((off_t)-1)
1104 
1105 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
1106 int vnode_create_disk_vobject(struct vnode *vp, off_t size, struct thread *td);
1107 void vnode_destroy_vobject(struct vnode *vp);
1108 
1109 extern struct vop_vector fifo_specops;
1110 extern struct vop_vector dead_vnodeops;
1111 extern struct vop_vector default_vnodeops;
1112 
1113 #define VOP_PANIC	((void*)(uintptr_t)vop_panic)
1114 #define VOP_NULL	((void*)(uintptr_t)vop_null)
1115 #define VOP_EBADF	((void*)(uintptr_t)vop_ebadf)
1116 #define VOP_ENOTTY	((void*)(uintptr_t)vop_enotty)
1117 #define VOP_EINVAL	((void*)(uintptr_t)vop_einval)
1118 #define VOP_ENOENT	((void*)(uintptr_t)vop_enoent)
1119 #define VOP_EOPNOTSUPP	((void*)(uintptr_t)vop_eopnotsupp)
1120 #define VOP_EAGAIN	((void*)(uintptr_t)vop_eagain)
1121 
1122 /* fifo_vnops.c */
1123 int	fifo_printinfo(struct vnode *);
1124 
1125 /* vfs_hash.c */
1126 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
1127 
1128 void vfs_hash_changesize(u_long newhashsize);
1129 int vfs_hash_get(const struct mount *mp, u_int hash, int flags,
1130     struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1131 u_int vfs_hash_index(struct vnode *vp);
1132 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
1133     struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1134 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
1135     struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1136 void vfs_hash_rehash(struct vnode *vp, u_int hash);
1137 void vfs_hash_remove(struct vnode *vp);
1138 
1139 int vfs_kqfilter(struct vop_kqfilter_args *);
1140 struct dirent;
1141 int vn_dir_next_dirent(struct vnode *vp, struct thread *td,
1142     char *dirbuf, size_t dirbuflen,
1143     struct dirent **dpp, size_t *len, off_t *off, int *eofflag);
1144 int vn_dir_check_empty(struct vnode *vp);
1145 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
1146 
1147 int vfs_unixify_accmode(accmode_t *accmode);
1148 
1149 void vfs_unp_reclaim(struct vnode *vp);
1150 
1151 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
1152 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
1153     gid_t gid);
1154 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
1155     struct thread *td);
1156 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
1157     struct thread *td);
1158 int vn_getsize_locked(struct vnode *vp, off_t *size, struct ucred *active_cred);
1159 int vn_getsize(struct vnode *vp, off_t *size, struct ucred *active_cred);
1160 
1161 void vn_fsid(struct vnode *vp, struct vattr *va);
1162 
1163 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp);
1164 int vn_lktype_write(struct mount *mp, struct vnode *vp);
1165 
1166 #ifdef INVARIANTS
1167 void vn_set_state_validate(struct vnode *vp, __enum_uint8(vstate) state);
1168 #endif
1169 
1170 static inline void
vn_set_state(struct vnode * vp,__enum_uint8 (vstate)state)1171 vn_set_state(struct vnode *vp, __enum_uint8(vstate) state)
1172 {
1173 #ifdef INVARIANTS
1174 	vn_set_state_validate(vp, state);
1175 #endif
1176 	vp->v_state = state;
1177 }
1178 
1179 static inline __enum_uint8(vstate)
vn_get_state(struct vnode * vp)1180 vn_get_state(struct vnode *vp)
1181 {
1182 	return (vp->v_state);
1183 }
1184 
1185 #define VOP_UNLOCK_FLAGS(vp, flags)	({				\
1186 	struct vnode *_vp = (vp);					\
1187 	int _flags = (flags);						\
1188 	int _error;							\
1189 									\
1190         if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0)		\
1191                 panic("%s: unsupported flags %x\n", __func__, flags);	\
1192         _error = VOP_UNLOCK(_vp);					\
1193         if (_flags & LK_INTERLOCK)					\
1194                 VI_UNLOCK(_vp);						\
1195         _error;								\
1196 })
1197 
1198 #include <sys/kernel.h>
1199 
1200 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \
1201 	SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \
1202 	    vfs_vector_op_register, &vnodeops)
1203 
1204 #define VFS_SMR_DECLARE				\
1205 	extern smr_t vfs_smr
1206 
1207 #define VFS_SMR()	vfs_smr
1208 #define vfs_smr_enter()	smr_enter(VFS_SMR())
1209 #define vfs_smr_exit()	smr_exit(VFS_SMR())
1210 #define vfs_smr_synchronize()	smr_synchronize(VFS_SMR())
1211 #define vfs_smr_entered_load(ptr)	smr_entered_load((ptr), VFS_SMR())
1212 #define VFS_SMR_ENTERED()		SMR_ENTERED(VFS_SMR())
1213 #define VFS_SMR_ASSERT_ENTERED()	SMR_ASSERT_ENTERED(VFS_SMR())
1214 #define VFS_SMR_ASSERT_NOT_ENTERED()	SMR_ASSERT_NOT_ENTERED(VFS_SMR())
1215 #define VFS_SMR_ZONE_SET(zone)	uma_zone_set_smr((zone), VFS_SMR())
1216 
1217 #define vn_load_v_data_smr(vp)	({		\
1218 	struct vnode *_vp = (vp);		\
1219 						\
1220 	VFS_SMR_ASSERT_ENTERED();		\
1221 	atomic_load_consume_ptr(&(_vp)->v_data);\
1222 })
1223 
1224 #endif /* _KERNEL */
1225 
1226 #endif /* !_SYS_VNODE_H_ */
1227