xref: /freebsd/sys/sys/vnode.h (revision 5e4947f7e08a90ba50f9e8cea9ae5e128717cb5f)
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 
70 /*
71  * We frequently need to test is something is a device node.
72  */
73 #define VTYPE_ISDEV(vtype)	((vtype) == VCHR || (vtype) == VBLK)
74 
__enum_uint8_decl(vstate)75 __enum_uint8_decl(vstate) {
76 	VSTATE_UNINITIALIZED,
77 	VSTATE_CONSTRUCTED,
78 	VSTATE_DESTROYING,
79 	VSTATE_DEAD,
80 	VLASTSTATE = VSTATE_DEAD,
81 };
82 
83 enum vgetstate {
84 	VGET_NONE,
85 	VGET_HOLDCNT,
86 	VGET_USECOUNT,
87 };
88 
89 /*
90  * Each underlying filesystem allocates its own private area and hangs
91  * it from v_data.  If non-null, this area is freed in getnewvnode().
92  */
93 
94 struct cache_fpl;
95 struct inotify_watch;
96 struct namecache;
97 
98 struct vpollinfo {
99 	struct	mtx vpi_lock;		/* lock to protect below */
100 	TAILQ_HEAD(, inotify_watch) vpi_inotify; /* list of inotify watchers */
101 	struct	selinfo vpi_selinfo;	/* identity of poller(s) */
102 	short	vpi_events;		/* what they are looking for */
103 	short	vpi_revents;		/* what has happened */
104 };
105 
106 /*
107  * Reading or writing any of these items requires holding the appropriate lock.
108  *
109  * Lock reference:
110  *	c - namecache mutex
111  *	i - interlock
112  *	l - mp mnt_listmtx or freelist mutex
113  *	I - updated with atomics, 0->1 and 1->0 transitions with interlock held
114  *	m - mount point interlock
115  *	p - pollinfo lock
116  *	u - Only a reference to the vnode is needed to read.
117  *	v - vnode lock
118  *
119  * Vnodes may be found on many lists.  The general way to deal with operating
120  * on a vnode that is on a list is:
121  *	1) Lock the list and find the vnode.
122  *	2) Lock interlock so that the vnode does not go away.
123  *	3) Unlock the list to avoid lock order reversals.
124  *	4) vget with LK_INTERLOCK and check for ENOENT, or
125  *	5) Check for DOOMED if the vnode lock is not required.
126  *	6) Perform your operation, then vput().
127  */
128 
129 #if defined(_KERNEL) || defined(_KVM_VNODE)
130 
131 struct vnode {
132 	/*
133 	 * Fields which define the identity of the vnode.  These fields are
134 	 * owned by the filesystem (XXX: and vgone() ?)
135 	 */
136 	__enum_uint8(vtype) v_type;		/* u vnode type */
137 	__enum_uint8(vstate) v_state;		/* u vnode state */
138 	short	v_irflag;			/* i frequently read flags */
139 	seqc_t	v_seqc;				/* i modification count */
140 	uint32_t v_nchash;			/* u namecache hash */
141 	u_int	v_hash;
142 	const struct vop_vector *v_op;		/* u vnode operations vector */
143 	void	*v_data;			/* u private data for fs */
144 
145 	/*
146 	 * Filesystem instance stuff
147 	 */
148 	struct	mount *v_mount;			/* u ptr to vfs we are in */
149 	TAILQ_ENTRY(vnode) v_nmntvnodes;	/* m vnodes for mount point */
150 
151 	/*
152 	 * Type specific fields, only one applies to any given vnode.
153 	 */
154 	union {
155 		struct mount	*v_mountedhere;	/* v ptr to mountpoint (VDIR) */
156 		struct unpcb	*v_unpcb;	/* v unix domain net (VSOCK) */
157 		struct cdev	*v_rdev; 	/* v device (VCHR, VBLK) */
158 		struct fifoinfo	*v_fifoinfo;	/* v fifo (VFIFO) */
159 	};
160 
161 	/*
162 	 * vfs_hash: (mount + inode) -> vnode hash.  The hash value
163 	 * itself is grouped with other int fields, to avoid padding.
164 	 */
165 	LIST_ENTRY(vnode)	v_hashlist;
166 
167 	/*
168 	 * VFS_namecache stuff
169 	 */
170 	LIST_HEAD(, namecache) v_cache_src;	/* c Cache entries from us */
171 	TAILQ_HEAD(, namecache) v_cache_dst;	/* c Cache entries to us */
172 	struct namecache *v_cache_dd;		/* c Cache entry for .. vnode */
173 
174 	/*
175 	 * Locking
176 	 */
177 	struct	lock v_lock;			/* u (if fs don't have one) */
178 	struct	mtx v_interlock;		/* lock for "i" things */
179 	struct	lock *v_vnlock;			/* u pointer to vnode lock */
180 
181 	/*
182 	 * The machinery of being a vnode
183 	 */
184 	TAILQ_ENTRY(vnode) v_vnodelist;		/* l vnode lists */
185 	TAILQ_ENTRY(vnode) v_lazylist;		/* l vnode lazy list */
186 	struct bufobj	v_bufobj;		/* * Buffer cache object */
187 
188 	/*
189 	 * Hooks for various subsystems and features.
190 	 */
191 	struct vpollinfo *v_pollinfo;		/* i Poll events, p for *v_pi */
192 	struct label *v_label;			/* MAC label for vnode */
193 	struct lockf *v_lockf;		/* Byte-level advisory lock list */
194 	struct rangelock v_rl;			/* Byte-range lock */
195 
196 	u_int	v_holdcnt;			/* I prevents recycling. */
197 	u_int	v_usecount;			/* I ref count of users */
198 	u_short	v_iflag;			/* i vnode flags (see below) */
199 	u_short	v_vflag;			/* v vnode flags */
200 	u_short	v_mflag;			/* l mnt-specific vnode flags */
201 	short	v_dbatchcpu;			/* i LRU requeue deferral batch */
202 	int	v_writecount;			/* I ref count of writers or
203 						   (negative) text users */
204 	int	v_seqc_users;			/* i modifications pending */
205 };
206 
207 #define VN_ISDEV(vp)		VTYPE_ISDEV((vp)->v_type)
208 
209 #ifndef DEBUG_LOCKS
210 #ifdef _LP64
211 /*
212  * Not crossing 448 bytes fits 9 vnodes per page. If you have to add fields
213  * to the structure and there is nothing which can be done to prevent growth
214  * then so be it. But don't grow it without a good reason.
215  */
216 _Static_assert(sizeof(struct vnode) <= 448, "vnode size crosses 448 bytes");
217 #endif
218 #endif
219 
220 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
221 
222 #define	bo2vnode(bo)	__containerof((bo), struct vnode, v_bufobj)
223 
224 #define v_object	v_bufobj.bo_object
225 
226 #define VN_KNOTE(vp, b, a) do {                    			\
227 	if ((vn_irflag_read(vp) & VIRF_KNOTE) != 0) {			\
228 		KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b),	\
229 		    (a) | KNF_NOKQLOCK);				\
230 	}								\
231 } while (0)
232 #define   VN_KNOTE_LOCKED(vp, b)     VN_KNOTE(vp, b, KNF_LISTLOCKED)
233 #define   VN_KNOTE_UNLOCKED(vp, b)   VN_KNOTE(vp, b, 0)
234 
235 /*
236  * Vnode flags.
237  *	VI flags are protected by interlock and live in v_iflag
238  *	VIRF flags are protected by interlock and live in v_irflag
239  *	VV flags are protected by the vnode lock and live in v_vflag
240  *
241  *	VIRF_DOOMED is doubly protected by the interlock and vnode lock.  Both
242  *	are required for writing but the status may be checked with either.
243  */
244 #define	VHOLD_NO_SMR	(1<<29)	/* Disable vhold_smr */
245 #define VHOLD_ALL_FLAGS (VHOLD_NO_SMR)
246 
247 #define	VIRF_DOOMED	0x0001	/* This vnode is being recycled */
248 #define	VIRF_PGREAD	0x0002	/* Direct reads from the page cache are permitted,
249 				   never cleared once set */
250 #define	VIRF_MOUNTPOINT	0x0004	/* This vnode is mounted on */
251 #define	VIRF_TEXT_REF	0x0008	/* Executable mappings ref the vnode */
252 #define	VIRF_CROSSMP	0x0010	/* Cross-mp vnode, no locking */
253 #define	VIRF_NAMEDDIR	0x0020	/* Named attribute directory */
254 #define	VIRF_NAMEDATTR	0x0040	/* Named attribute */
255 #define	VIRF_INOTIFY	0x0080	/* This vnode is being watched */
256 #define	VIRF_INOTIFY_PARENT 0x0100 /* A parent of this vnode may be being
257 				      watched */
258 #define	VIRF_KNOTE	0x0200	/* Has knlist */
259 
260 #define	VI_UNUSED0	0x0001	/* unused */
261 #define	VI_MOUNT	0x0002	/* Mount in progress */
262 #define	VI_DOINGINACT	0x0004	/* VOP_INACTIVE is in progress */
263 #define	VI_OWEINACT	0x0008	/* Need to call inactive */
264 #define	VI_DEFINACT	0x0010	/* deferred inactive */
265 #define	VI_FOPENING	0x0020	/* In open, with opening process having the
266 				   first right to advlock file */
267 #define	VI_DELAYED_SETSIZE 0x0040	/* Delayed setsize */
268 
269 #define	VV_ROOT		0x0001	/* root of its filesystem */
270 #define	VV_ISTTY	0x0002	/* vnode represents a tty */
271 #define	VV_NOSYNC	0x0004	/* unlinked, stop syncing */
272 #define	VV_ETERNALDEV	0x0008	/* device that is never destroyed */
273 #define	VV_CACHEDLABEL	0x0010	/* Vnode has valid cached MAC label */
274 #define	VV_VMSIZEVNLOCK	0x0020	/* object size check requires vnode lock */
275 #define	VV_COPYONWRITE	0x0040	/* vnode is doing copy-on-write */
276 #define	VV_SYSTEM	0x0080	/* vnode being used by kernel */
277 #define	VV_PROCDEP	0x0100	/* vnode is process dependent */
278 #define	VV_UNLINKED	0x0200	/* unlinked but stil open directory */
279 #define	VV_DELETED	0x0400	/* should be removed */
280 #define	VV_MD		0x0800	/* vnode backs the md device */
281 #define	VV_FORCEINSMQ	0x1000	/* force the insmntque to succeed */
282 #define	VV_READLINK	0x2000	/* fdescfs linux vnode */
283 #define	VV_UNREF	0x4000	/* vunref, do not drop lock in inactive() */
284 #define	VV_CROSSLOCK	0x8000	/* vnode lock is shared w/ root mounted here */
285 
286 #define	VMP_LAZYLIST	0x0001	/* Vnode is on mnt's lazy list */
287 
288 /*
289  * Vnode attributes.  A field value of VNOVAL represents a field whose value
290  * is unavailable (getattr) or which is not to be changed (setattr).
291  */
292 struct vattr {
293 	__enum_uint8(vtype)	va_type;	/* vnode type (for create) */
294 	u_short		va_mode;	/* files access mode and type */
295 	uint16_t	va_bsdflags;	/* same as st_bsdflags from stat(2) */
296 	uid_t		va_uid;		/* owner user id */
297 	gid_t		va_gid;		/* owner group id */
298 	nlink_t		va_nlink;	/* number of references to file */
299 	dev_t		va_fsid;	/* filesystem id */
300 	ino_t		va_fileid;	/* file id */
301 	u_quad_t	va_size;	/* file size in bytes */
302 	long		va_blocksize;	/* blocksize preferred for i/o */
303 	struct timespec	va_atime;	/* time of last access */
304 	struct timespec	va_mtime;	/* time of last modification */
305 	struct timespec	va_ctime;	/* time file changed */
306 	struct timespec	va_birthtime;	/* time file created */
307 	u_long		va_gen;		/* generation number of file */
308 	u_long		va_flags;	/* flags defined for file */
309 	dev_t		va_rdev;	/* device the special file represents */
310 	u_quad_t	va_bytes;	/* bytes of disk space held by file */
311 	u_quad_t	va_filerev;	/* file modification number */
312 	u_int		va_vaflags;	/* operations flags, see below */
313 	long		va_spare;	/* remain quad aligned */
314 };
315 
316 #define VATTR_ISDEV(vap)	VTYPE_ISDEV((vap)->va_type)
317 
318 /*
319  * Flags for va_vaflags.
320  */
321 #define	VA_UTIMES_NULL	0x01		/* utimes argument was NULL */
322 #define	VA_EXCLUSIVE	0x02		/* exclusive create request */
323 #define	VA_SYNC		0x04		/* O_SYNC truncation */
324 
325 /*
326  * Flags for ioflag. (high 16 bits used to ask for read-ahead and
327  * help with write clustering)
328  * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
329  */
330 #define	IO_UNIT		0x0001		/* do I/O as atomic unit */
331 #define	IO_APPEND	0x0002		/* append write to end */
332 #define	IO_NDELAY	0x0004		/* FNDELAY flag set in file table */
333 #define	IO_NODELOCKED	0x0008		/* underlying node already locked */
334 #define	IO_ASYNC	0x0010		/* bawrite rather then bdwrite */
335 #define	IO_VMIO		0x0020		/* data already in VMIO space */
336 #define	IO_INVAL	0x0040		/* invalidate after I/O */
337 #define	IO_SYNC		0x0080		/* do I/O synchronously */
338 #define	IO_DIRECT	0x0100		/* attempt to bypass buffer cache */
339 #define	IO_NOREUSE	0x0200		/* VMIO data won't be reused */
340 #define	IO_EXT		0x0400		/* operate on external attributes */
341 #define	IO_NORMAL	0x0800		/* operate on regular data */
342 #define	IO_NOMACCHECK	0x1000		/* MAC checks unnecessary */
343 #define	IO_BUFLOCKED	0x2000		/* ffs flag; indir buf is locked */
344 #define	IO_RANGELOCKED	0x4000		/* range locked */
345 #define	IO_DATASYNC	0x8000		/* do only data I/O synchronously */
346 
347 #define IO_SEQMAX	0x7F		/* seq heuristic max value */
348 #define IO_SEQSHIFT	16		/* seq heuristic in upper 16 bits */
349 
350 /*
351  * Flags for accmode_t.
352  */
353 #define	VEXEC			000000000100 /* execute/search permission */
354 #define	VWRITE			000000000200 /* write permission */
355 #define	VREAD			000000000400 /* read permission */
356 #define	VADMIN			000000010000 /* being the file owner */
357 #define	VAPPEND			000000040000 /* permission to write/append */
358 /*
359  * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
360  * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
361  * and 0 otherwise.  This never happens with ordinary unix access rights
362  * or POSIX.1e ACLs.  Obviously, VEXPLICIT_DENY must be OR-ed with
363  * some other V* constant.
364  */
365 #define	VEXPLICIT_DENY		000000100000
366 #define	VREAD_NAMED_ATTRS 	000000200000 /* not used */
367 #define	VWRITE_NAMED_ATTRS 	000000400000 /* not used */
368 #define	VDELETE_CHILD	 	000001000000
369 #define	VREAD_ATTRIBUTES 	000002000000 /* permission to stat(2) */
370 #define	VWRITE_ATTRIBUTES 	000004000000 /* change {m,c,a}time */
371 #define	VDELETE		 	000010000000
372 #define	VREAD_ACL	 	000020000000 /* read ACL and file mode */
373 #define	VWRITE_ACL	 	000040000000 /* change ACL and/or file mode */
374 #define	VWRITE_OWNER	 	000100000000 /* change file owner */
375 #define	VSYNCHRONIZE	 	000200000000 /* not used */
376 #define	VCREAT			000400000000 /* creating new file */
377 #define	VVERIFY			001000000000 /* verification required */
378 
379 /*
380  * Permissions that were traditionally granted only to the file owner.
381  */
382 #define VADMIN_PERMS	(VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
383     VWRITE_OWNER)
384 
385 /*
386  * Permissions that were traditionally granted to everyone.
387  */
388 #define VSTAT_PERMS	(VREAD_ATTRIBUTES | VREAD_ACL)
389 
390 /*
391  * Permissions that allow to change the state of the file in any way.
392  */
393 #define VMODIFY_PERMS	(VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
394     VDELETE)
395 
396 /*
397  * Token indicating no attribute value yet assigned.
398  */
399 #define	VNOVAL	(-1)
400 
401 /*
402  * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
403  */
404 #define VLKTIMEOUT	(hz / 20 + 1)
405 
406 #ifdef _KERNEL
407 
408 #ifdef MALLOC_DECLARE
409 MALLOC_DECLARE(M_VNODE);
410 #endif
411 
412 extern u_int ncsizefactor;
413 extern const u_int io_hold_cnt;
414 
415 /*
416  * Convert between vnode types and inode formats (since POSIX.1
417  * defines mode word of stat structure in terms of inode formats).
418  */
419 extern __enum_uint8(vtype)	iftovt_tab[];
420 extern int		vttoif_tab[];
421 #define	IFTOVT(mode)	(iftovt_tab[((mode) & S_IFMT) >> 12])
422 #define	VTTOIF(indx)	(vttoif_tab[(int)(indx)])
423 #define	MAKEIMODE(indx, mode)	(int)(VTTOIF(indx) | (mode))
424 
425 /*
426  * Flags to various vnode functions.
427  */
428 #define	SKIPSYSTEM	0x0001	/* vflush: skip vnodes marked VSYSTEM */
429 #define	FORCECLOSE	0x0002	/* vflush: force file closure */
430 #define	WRITECLOSE	0x0004	/* vflush: only close writable files */
431 #define	EARLYFLUSH	0x0008	/* vflush: early call for ffs_flushfiles */
432 #define	V_SAVE		0x0001	/* vinvalbuf: sync file first */
433 #define	V_ALT		0x0002	/* vinvalbuf: invalidate only alternate bufs */
434 #define	V_NORMAL	0x0004	/* vinvalbuf: invalidate only regular bufs */
435 #define	V_CLEANONLY	0x0008	/* vinvalbuf: invalidate only clean bufs */
436 #define	V_VMIO		0x0010	/* vinvalbuf: called during pageout */
437 #define	V_ALLOWCLEAN	0x0020	/* vinvalbuf: allow clean buffers after flush */
438 #define	REVOKEALL	0x0001	/* vop_revoke: revoke all aliases */
439 #define	V_WAIT		0x0001	/* vn_start_write: sleep for suspend */
440 #define	V_NOWAIT	0x0002	/* vn_start_write: don't sleep for suspend */
441 #define	V_XSLEEP	0x0004	/* vn_start_write: just return after sleep */
442 #define	V_PCATCH	0x0008	/* vn_start_write: make the sleep interruptible */
443 #define	V_VALID_FLAGS (V_WAIT | V_NOWAIT | V_XSLEEP | V_PCATCH)
444 
445 #define	VR_START_WRITE	0x0001	/* vfs_write_resume: start write atomically */
446 #define	VR_NO_SUSPCLR	0x0002	/* vfs_write_resume: do not clear suspension */
447 
448 #define	VS_SKIP_UNMOUNT	0x0001	/* vfs_write_suspend: fail if the
449 				   filesystem is being unmounted */
450 
451 #ifdef DIAGNOSTIC
452 #define	VATTR_NULL(vap)	vattr_null(vap)
453 #else
454 #define	VATTR_NULL(vap)	(*(vap) = va_null)	/* initialize a vattr */
455 #endif /* DIAGNOSTIC */
456 
457 /*
458  * Global vnode data.
459  */
460 extern	struct vnode *rootvnode;	/* root (i.e. "/") vnode */
461 extern	struct mount *rootdevmp;	/* "/dev" mount */
462 extern	u_long desiredvnodes;		/* number of vnodes desired */
463 extern	struct uma_zone *namei_zone;
464 extern	struct vattr va_null;		/* predefined null vattr structure */
465 
466 extern	u_int vn_lock_pair_pause_max;
467 
468 #define	VI_LOCK(vp)	mtx_lock(&(vp)->v_interlock)
469 #define	VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
470 #define	VI_TRYLOCK(vp)	mtx_trylock(&(vp)->v_interlock)
471 #define	VI_UNLOCK(vp)	mtx_unlock(&(vp)->v_interlock)
472 #define	VI_MTX(vp)	(&(vp)->v_interlock)
473 
474 #define	VN_LOCK_AREC(vp)	lockallowrecurse((vp)->v_vnlock)
475 #define	VN_LOCK_ASHARE(vp)	lockallowshare((vp)->v_vnlock)
476 #define	VN_LOCK_DSHARE(vp)	lockdisableshare((vp)->v_vnlock)
477 
478 #endif /* _KERNEL */
479 
480 /*
481  * Mods for extensibility.
482  */
483 
484 /*
485  * Flags for vdesc_flags:
486  */
487 #define	VDESC_MAX_VPS		16
488 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
489 #define	VDESC_VP0_WILLRELE	0x0001
490 #define	VDESC_VP1_WILLRELE	0x0002
491 #define	VDESC_VP2_WILLRELE	0x0004
492 #define	VDESC_VP3_WILLRELE	0x0008
493 
494 /*
495  * A generic structure.
496  * This can be used by bypass routines to identify generic arguments.
497  */
498 struct vop_generic_args {
499 	struct vnodeop_desc *a_desc;
500 	/* other random data follows, presumably */
501 };
502 
503 typedef int vop_bypass_t(struct vop_generic_args *);
504 
505 /*
506  * VDESC_NO_OFFSET is used to identify the end of the offset list
507  * and in places where no such field exists.
508  */
509 #define VDESC_NO_OFFSET -1
510 
511 /*
512  * This structure describes the vnode operation taking place.
513  */
514 struct vnodeop_desc {
515 	char	*vdesc_name;		/* a readable name for debugging */
516 	int	 vdesc_flags;		/* VDESC_* flags */
517 	int	vdesc_vop_offset;
518 	vop_bypass_t	*vdesc_call;	/* Function to call */
519 
520 	/*
521 	 * These ops are used by bypass routines to map and locate arguments.
522 	 * Creds and procs are not needed in bypass routines, but sometimes
523 	 * they are useful to (for example) transport layers.
524 	 * Nameidata is useful because it has a cred in it.
525 	 */
526 	int	*vdesc_vp_offsets;	/* list ended by VDESC_NO_OFFSET */
527 	int	vdesc_vpp_offset;	/* return vpp location */
528 	int	vdesc_cred_offset;	/* cred location, if any */
529 	int	vdesc_thread_offset;	/* thread location, if any */
530 	int	vdesc_componentname_offset; /* if any */
531 };
532 
533 #ifdef _KERNEL
534 /*
535  * A list of all the operation descs.
536  */
537 extern struct vnodeop_desc *vnodeop_descs[];
538 
539 #define	VOPARG_OFFSETOF(s_type, field)	__offsetof(s_type, field)
540 #define	VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
541     ((s_type)(((char*)(struct_p)) + (s_offset)))
542 
543 #ifdef INVARIANTS
544 /*
545  * Support code to aid in debugging VFS locking problems.  Not totally
546  * reliable since if the thread sleeps between changing the lock
547  * state and checking it with the assert, some other thread could
548  * change the state.  They are good enough for debugging a single
549  * filesystem using a single-threaded test.  Note that the unreliability is
550  * limited to false negatives; efforts were made to ensure that false
551  * positives cannot occur.
552  */
553 void	assert_vi_locked(struct vnode *vp, const char *str);
554 void	assert_vi_unlocked(struct vnode *vp, const char *str);
555 void	assert_vop_elocked(struct vnode *vp, const char *str);
556 void	assert_vop_locked(struct vnode *vp, const char *str);
557 void	assert_vop_unlocked(struct vnode *vp, const char *str);
558 
559 #define	ASSERT_VI_LOCKED(vp, str)	assert_vi_locked((vp), (str))
560 #define	ASSERT_VI_UNLOCKED(vp, str)	assert_vi_unlocked((vp), (str))
561 #define	ASSERT_VOP_ELOCKED(vp, str)	assert_vop_elocked((vp), (str))
562 #define	ASSERT_VOP_LOCKED(vp, str)	assert_vop_locked((vp), (str))
563 #define	ASSERT_VOP_UNLOCKED(vp, str)	assert_vop_unlocked((vp), (str))
564 
565 #define ASSERT_VOP_IN_SEQC(vp)	do {				\
566 	struct vnode *_vp = (vp);				\
567 								\
568 	VNPASS(seqc_in_modify(_vp->v_seqc), _vp);		\
569 } while (0)
570 
571 #define ASSERT_VOP_NOT_IN_SEQC(vp)	do {			\
572 	struct vnode *_vp = (vp);				\
573 								\
574 	VNPASS(!seqc_in_modify(_vp->v_seqc), _vp);		\
575 } while (0)
576 
577 #else /* !INVARIANTS */
578 
579 #define	ASSERT_VI_LOCKED(vp, str)	((void)0)
580 #define	ASSERT_VI_UNLOCKED(vp, str)	((void)0)
581 #define	ASSERT_VOP_ELOCKED(vp, str)	((void)0)
582 #define	ASSERT_VOP_LOCKED(vp, str)	((void)0)
583 #define	ASSERT_VOP_UNLOCKED(vp, str)	((void)0)
584 
585 #define ASSERT_VOP_IN_SEQC(vp)		((void)0)
586 #define ASSERT_VOP_NOT_IN_SEQC(vp)	((void)0)
587 
588 #endif /* INVARIANTS */
589 
590 #define DOINGASYNC(vp)	   					\
591 	(((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 &&	\
592 	 ((curthread->td_pflags & TDP_SYNCIO) == 0))
593 
594 /*
595  * VMIO support inline
596  */
597 
598 extern int vmiodirenable;
599 
600 static __inline int
vn_canvmio(struct vnode * vp)601 vn_canvmio(struct vnode *vp)
602 {
603       if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
604 		return(TRUE);
605 	return(FALSE);
606 }
607 
608 /*
609  * Finally, include the default set of vnode operations.
610  */
611 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
612 #include "vnode_if.h"
613 
614 /* vn_open_flags */
615 #define	VN_OPEN_NOAUDIT		0x00000001
616 #define	VN_OPEN_NOCAPCHECK	0x00000002
617 #define	VN_OPEN_NAMECACHE	0x00000004
618 #define	VN_OPEN_INVFS		0x00000008
619 #define	VN_OPEN_WANTIOCTLCAPS	0x00000010
620 
621 /* copy_file_range kernel flags */
622 #define	COPY_FILE_RANGE_KFLAGS		0xff000000
623 #define	COPY_FILE_RANGE_TIMEO1SEC	0x01000000	/* Return after 1sec. */
624 
625 /*
626  * Public vnode manipulation functions.
627  */
628 struct componentname;
629 struct file;
630 struct mount;
631 struct nameidata;
632 struct ostat;
633 struct freebsd11_stat;
634 struct thread;
635 struct proc;
636 struct stat;
637 struct nstat;
638 struct ucred;
639 struct uio;
640 struct vattr;
641 struct vfsops;
642 struct vnode;
643 
644 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
645 
646 int	bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
647 	    daddr_t endn);
648 /* cache_* may belong in namei.h. */
649 void	cache_changesize(u_long newhashsize);
650 
651 #define	VFS_CACHE_DROPOLD	0x1
652 
653 void	cache_enter_time_flags(struct vnode *dvp, struct vnode *vp,
654 	    struct componentname *cnp, struct timespec *tsp,
655 	    struct timespec *dtsp, int flags);
656 #define	cache_enter(dvp, vp, cnp)					\
657 	cache_enter_time(dvp, vp, cnp, NULL, NULL)
658 void	cache_enter_time(struct vnode *dvp, struct vnode *vp,
659 	    struct componentname *cnp, struct timespec *tsp,
660 	    struct timespec *dtsp);
661 int	cache_lookup(struct vnode *dvp, struct vnode **vpp,
662 	    struct componentname *cnp, struct timespec *tsp, int *ticksp);
663 void	cache_vnode_init(struct vnode *vp);
664 void	cache_purge(struct vnode *vp);
665 void	cache_purge_vgone(struct vnode *vp);
666 void	cache_purge_negative(struct vnode *vp);
667 void	cache_purgevfs(struct mount *mp);
668 char	*cache_symlink_alloc(size_t size, int flags);
669 void	cache_symlink_free(char *string, size_t size);
670 int	cache_symlink_resolve(struct cache_fpl *fpl, const char *string,
671 	    size_t len);
672 void	cache_vop_inotify(struct vnode *vp, int event, uint32_t cookie);
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 void	cache_vop_vector_register(struct vop_vector *);
677 #ifdef INVARIANTS
678 void	cache_validate(struct vnode *dvp, struct vnode *vp,
679 	    struct componentname *cnp);
680 void	cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops);
681 void	cache_assert_no_entries(struct vnode *vp);
682 #else
683 static inline void
cache_validate(struct vnode * dvp,struct vnode * vp,struct componentname * cnp)684 cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
685 {
686 }
687 
688 static inline void
cache_validate_vop_vector(struct mount * mp,struct vop_vector * vops)689 cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops)
690 {
691 }
692 
693 static inline void
cache_assert_no_entries(struct vnode * vp)694 cache_assert_no_entries(struct vnode *vp)
695 {
696 }
697 #endif
698 void	cache_fast_lookup_enabled_recalc(void);
699 int	change_dir(struct vnode *vp, struct thread *td);
700 void	cvtstat(struct stat *st, struct ostat *ost);
701 int	freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb);
702 int	freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost);
703 int	getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
704 	    struct vnode **vpp);
705 void	getnewvnode_reserve(void);
706 void	getnewvnode_drop_reserve(void);
707 int	insmntque(struct vnode *vp, struct mount *mp);
708 int	insmntque1(struct vnode *vp, struct mount *mp);
709 u_quad_t init_va_filerev(void);
710 int	speedup_syncer(void);
711 int	vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen);
712 int	vn_getcwd(char *buf, char **retbuf, size_t *buflen);
713 int	vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf);
714 int	vn_fullpath_jail(struct vnode *vp, char **retbuf, char **freebuf);
715 int	vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf);
716 int	vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp,
717 	    const char *hdrl_name, size_t hrdl_name_length, char **retbuf,
718 	    char **freebuf, size_t *buflen);
719 struct vnode *
720 	vn_dir_dd_ino(struct vnode *vp);
721 int	vn_commname(struct vnode *vn, char *buf, u_int buflen);
722 int	vn_path_to_global_path(struct thread *td, struct vnode *vp,
723 	    char *path, u_int pathlen);
724 int	vn_path_to_global_path_hardlink(struct thread *td, struct vnode *vp,
725 	    struct vnode *dvp, char *path, u_int pathlen, const char *leaf_name,
726 	    size_t leaf_length);
727 int	vaccess(__enum_uint8(vtype) type, mode_t file_mode, uid_t file_uid,
728 	    gid_t file_gid, accmode_t accmode, struct ucred *cred);
729 int	vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid,
730 	    struct ucred *cred);
731 int	vaccess_acl_nfs4(__enum_uint8(vtype) type, uid_t file_uid, gid_t file_gid,
732 	    struct acl *aclp, accmode_t accmode, struct ucred *cred);
733 int	vaccess_acl_posix1e(__enum_uint8(vtype) type, uid_t file_uid,
734 	    gid_t file_gid, struct acl *acl, accmode_t accmode,
735 	    struct ucred *cred);
736 void	vattr_null(struct vattr *vap);
737 void	vlazy(struct vnode *);
738 void	vdrop(struct vnode *);
739 void	vdropl(struct vnode *);
740 int	vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
741 int	vget(struct vnode *vp, int flags);
742 enum vgetstate	vget_prep_smr(struct vnode *vp);
743 enum vgetstate	vget_prep(struct vnode *vp);
744 int	vget_finish(struct vnode *vp, int flags, enum vgetstate vs);
745 void	vget_finish_ref(struct vnode *vp, enum vgetstate vs);
746 void	vget_abort(struct vnode *vp, enum vgetstate vs);
747 void	vgone(struct vnode *vp);
748 void	vhold(struct vnode *);
749 void	vholdnz(struct vnode *);
750 bool	vhold_smr(struct vnode *);
751 int	vinactive(struct vnode *vp);
752 int	vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
753 int	vtruncbuf(struct vnode *vp, off_t length, int blksize);
754 void	v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn,
755 	    int blksize);
756 void	vunref(struct vnode *);
757 void	vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
758 int	vrecycle(struct vnode *vp);
759 int	vrecyclel(struct vnode *vp);
760 int	vn_bmap_seekhole_locked(struct vnode *vp, u_long cmd, off_t *off,
761 	    struct ucred *cred);
762 int	vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
763 	    struct ucred *cred);
764 int	vn_close(struct vnode *vp,
765 	    int flags, struct ucred *file_cred, struct thread *td);
766 int	vn_copy_file_range(struct vnode *invp, off_t *inoffp,
767 	    struct vnode *outvp, off_t *outoffp, size_t *lenp,
768 	    unsigned int flags, struct ucred *incred, struct ucred *outcred,
769 	    struct thread *fsize_td);
770 int	vn_deallocate(struct vnode *vp, off_t *offset, off_t *length, int flags,
771 	    int ioflg, struct ucred *active_cred, struct ucred *file_cred);
772 void	vn_finished_write(struct mount *mp);
773 void	vn_finished_secondary_write(struct mount *mp);
774 int	vn_fsync_buf(struct vnode *vp, int waitfor);
775 int	vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp,
776 	    struct vnode *outvp, off_t *outoffp, size_t *lenp,
777 	    unsigned int flags, struct ucred *incred, struct ucred *outcred,
778 	    struct thread *fsize_td);
779 int	vn_need_pageq_flush(struct vnode *vp);
780 bool	vn_isdisk_error(struct vnode *vp, int *errp);
781 bool	vn_isdisk(struct vnode *vp);
782 int	_vn_lock(struct vnode *vp, int flags, const char *file, int line);
783 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
784 void	vn_lock_pair(struct vnode *vp1, bool vp1_locked, int lkflags1,
785 	    struct vnode *vp2, bool vp2_locked, int lkflags2);
786 int	vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
787 int	vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
788 	    u_int vn_open_flags, struct ucred *cred, struct file *fp);
789 int	vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
790 	    struct thread *td, struct file *fp);
791 void	vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
792 void	vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start,
793 	    vm_pindex_t end);
794 int	vn_pollrecord(struct vnode *vp, struct thread *p, int events);
795 int	vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
796 	    int len, off_t offset, enum uio_seg segflg, int ioflg,
797 	    struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
798 	    struct thread *td);
799 int	vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
800 	    size_t len, off_t offset, enum uio_seg segflg, int ioflg,
801 	    struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
802 	    struct thread *td);
803 int	vn_read_from_obj(struct vnode *vp, struct uio *uio);
804 int	vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio,
805 	    struct thread *td);
806 int	vn_rlimit_fsizex(const struct vnode *vp, struct uio *uio,
807 	    off_t maxfsz, ssize_t *resid_adj, struct thread *td);
808 void	vn_rlimit_fsizex_res(struct uio *uio, ssize_t resid_adj);
809 int	vn_rlimit_trunc(u_quad_t size, struct thread *td);
810 int	vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
811 int	vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
812 	    int flags);
813 int	vn_truncate_locked(struct vnode *vp, off_t length, bool sync,
814 	    struct ucred *cred);
815 int	vn_writechk(struct vnode *vp);
816 int	vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
817 	    const char *attrname, int *buflen, char *buf, struct thread *td);
818 int	vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
819 	    const char *attrname, int buflen, char *buf, struct thread *td);
820 int	vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
821 	    const char *attrname, struct thread *td);
822 int	vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
823 	    struct vnode **rvp);
824 int	vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
825 	    void *alloc_arg, int lkflags, struct vnode **rvp);
826 int	vn_utimes_perm(struct vnode *vp, struct vattr *vap,
827 	    struct ucred *cred, struct thread *td);
828 int	vn_cmp(struct file *, struct file *, struct thread *td);
829 
830 int	vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
831 int	vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
832 	    struct uio *uio);
833 
834 void	vn_seqc_write_begin_locked(struct vnode *vp);
835 void	vn_seqc_write_begin(struct vnode *vp);
836 void	vn_seqc_write_end_locked(struct vnode *vp);
837 void	vn_seqc_write_end(struct vnode *vp);
838 #define	vn_seqc_read_any(vp)		seqc_read_any(&(vp)->v_seqc)
839 #define	vn_seqc_read_notmodify(vp)	seqc_read_notmodify(&(vp)->v_seqc)
840 #define	vn_seqc_consistent(vp, seq)	seqc_consistent(&(vp)->v_seqc, seq)
841 
842 #define	vn_rangelock_unlock(vp, cookie)					\
843 	rangelock_unlock(&(vp)->v_rl, (cookie))
844 #define	vn_rangelock_rlock(vp, start, end)				\
845 	rangelock_rlock(&(vp)->v_rl, (start), (end))
846 #define	vn_rangelock_tryrlock(vp, start, end)				\
847 	rangelock_tryrlock(&(vp)->v_rl, (start), (end))
848 #define	vn_rangelock_wlock(vp, start, end)				\
849 	rangelock_wlock(&(vp)->v_rl, (start), (end))
850 #define	vn_rangelock_trywlock(vp, start, end)				\
851 	rangelock_trywlock(&(vp)->v_rl, (start), (end))
852 
853 #define	vn_irflag_read(vp)	atomic_load_short(&(vp)->v_irflag)
854 void	vn_irflag_set_locked(struct vnode *vp, short toset);
855 void	vn_irflag_set(struct vnode *vp, short toset);
856 void	vn_irflag_set_cond_locked(struct vnode *vp, short toset);
857 void	vn_irflag_set_cond(struct vnode *vp, short toset);
858 void	vn_irflag_unset_locked(struct vnode *vp, short tounset);
859 void	vn_irflag_unset(struct vnode *vp, short tounset);
860 
861 int	vfs_cache_lookup(struct vop_lookup_args *ap);
862 int	vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp);
863 void	vfs_timestamp(struct timespec *);
864 void	vfs_write_resume(struct mount *mp, int flags);
865 int	vfs_write_suspend(struct mount *mp, int flags);
866 int	vfs_write_suspend_umnt(struct mount *mp);
867 struct vnode *vnlru_alloc_marker(void);
868 void	vnlru_free_marker(struct vnode *);
869 void	vnlru_free_vfsops(int, struct vfsops *, struct vnode *);
870 int	vop_stdbmap(struct vop_bmap_args *);
871 int	vop_stdfdatasync_buf(struct vop_fdatasync_args *);
872 int	vop_stdfsync(struct vop_fsync_args *);
873 int	vop_stdgetwritemount(struct vop_getwritemount_args *);
874 int	vop_stdgetpages(struct vop_getpages_args *);
875 int	vop_stdinactive(struct vop_inactive_args *);
876 int	vop_stdneed_inactive(struct vop_need_inactive_args *);
877 int	vop_stdinotify(struct vop_inotify_args *);
878 int	vop_stdinotify_add_watch(struct vop_inotify_add_watch_args *);
879 int	vop_stdioctl(struct vop_ioctl_args *);
880 int	vop_stdkqfilter(struct vop_kqfilter_args *);
881 int	vop_stdlock(struct vop_lock1_args *);
882 int	vop_stdunlock(struct vop_unlock_args *);
883 int	vop_stdislocked(struct vop_islocked_args *);
884 int	vop_lock(struct vop_lock1_args *);
885 int	vop_unlock(struct vop_unlock_args *);
886 int	vop_islocked(struct vop_islocked_args *);
887 int	vop_stdputpages(struct vop_putpages_args *);
888 int	vop_nopoll(struct vop_poll_args *);
889 int	vop_stdaccess(struct vop_access_args *ap);
890 int	vop_stdaccessx(struct vop_accessx_args *ap);
891 int	vop_stdadvise(struct vop_advise_args *ap);
892 int	vop_stdadvlock(struct vop_advlock_args *ap);
893 int	vop_stdadvlockasync(struct vop_advlockasync_args *ap);
894 int	vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
895 int	vop_stdallocate(struct vop_allocate_args *ap);
896 int	vop_stddeallocate(struct vop_deallocate_args *ap);
897 int	vop_stdset_text(struct vop_set_text_args *ap);
898 int	vop_stdpathconf(struct vop_pathconf_args *);
899 int	vop_stdpoll(struct vop_poll_args *);
900 int	vop_stdvptocnp(struct vop_vptocnp_args *ap);
901 int	vop_stdvptofh(struct vop_vptofh_args *ap);
902 int	vop_stdunp_bind(struct vop_unp_bind_args *ap);
903 int	vop_stdunp_connect(struct vop_unp_connect_args *ap);
904 int	vop_stdunp_detach(struct vop_unp_detach_args *ap);
905 int	vop_stdadd_writecount_nomsync(struct vop_add_writecount_args *ap);
906 int	vop_stdcopy_file_range(struct vop_copy_file_range_args *ap);
907 int	vop_eopnotsupp(struct vop_generic_args *ap);
908 int	vop_ebadf(struct vop_generic_args *ap);
909 int	vop_einval(struct vop_generic_args *ap);
910 int	vop_enoent(struct vop_generic_args *ap);
911 int	vop_enotty(struct vop_generic_args *ap);
912 int	vop_eagain(struct vop_generic_args *ap);
913 int	vop_null(struct vop_generic_args *ap);
914 int	vop_panic(struct vop_generic_args *ap);
915 int	dead_poll(struct vop_poll_args *ap);
916 int	dead_read(struct vop_read_args *ap);
917 int	dead_write(struct vop_write_args *ap);
918 
919 /* These are called from within the actual VOPS. */
920 void	vop_allocate_post(void *a, int rc);
921 void	vop_copy_file_range_post(void *ap, int rc);
922 void	vop_close_post(void *a, int rc);
923 void	vop_create_pre(void *a);
924 void	vop_create_post(void *a, int rc);
925 void	vop_deallocate_post(void *a, int rc);
926 void	vop_whiteout_pre(void *a);
927 void	vop_whiteout_post(void *a, int rc);
928 void	vop_deleteextattr_pre(void *a);
929 void	vop_deleteextattr_post(void *a, int rc);
930 void	vop_link_pre(void *a);
931 void	vop_link_post(void *a, int rc);
932 void	vop_lookup_post(void *a, int rc);
933 void	vop_lookup_pre(void *a);
934 void	vop_mkdir_pre(void *a);
935 void	vop_mkdir_post(void *a, int rc);
936 void	vop_mknod_pre(void *a);
937 void	vop_mknod_post(void *a, int rc);
938 void	vop_open_post(void *a, int rc);
939 void	vop_read_post(void *a, int rc);
940 void	vop_read_pgcache_post(void *ap, int rc);
941 void	vop_reclaim_post(void *a, int rc);
942 void	vop_remove_pre(void *a);
943 void	vop_remove_post(void *a, int rc);
944 void	vop_rename_post(void *a, int rc);
945 void	vop_rename_pre(void *a);
946 void	vop_rmdir_pre(void *a);
947 void	vop_rmdir_post(void *a, int rc);
948 void	vop_setattr_pre(void *a);
949 void	vop_setattr_post(void *a, int rc);
950 void	vop_setacl_pre(void *a);
951 void	vop_setacl_post(void *a, int rc);
952 void	vop_setextattr_pre(void *a);
953 void	vop_setextattr_post(void *a, int rc);
954 void	vop_symlink_pre(void *a);
955 void	vop_symlink_post(void *a, int rc);
956 int	vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a);
957 
958 #ifdef INVARIANTS
959 void	vop_fdatasync_debugpre(void *a);
960 void	vop_fdatasync_debugpost(void *a, int rc);
961 void	vop_fplookup_vexec_debugpre(void *a);
962 void	vop_fplookup_vexec_debugpost(void *a, int rc);
963 void	vop_fplookup_symlink_debugpre(void *a);
964 void	vop_fplookup_symlink_debugpost(void *a, int rc);
965 void	vop_fsync_debugpre(void *a);
966 void	vop_fsync_debugpost(void *a, int rc);
967 void	vop_strategy_debugpre(void *a);
968 void	vop_lock_debugpre(void *a);
969 void	vop_lock_debugpost(void *a, int rc);
970 void	vop_unlock_debugpre(void *a);
971 void	vop_need_inactive_debugpre(void *a);
972 void	vop_need_inactive_debugpost(void *a, int rc);
973 void	vop_mkdir_debugpost(void *a, int rc);
974 #else
975 #define	vop_fdatasync_debugpre(x)		do { } while (0)
976 #define	vop_fdatasync_debugpost(x, y)		do { } while (0)
977 #define	vop_fplookup_vexec_debugpre(x)		do { } while (0)
978 #define	vop_fplookup_vexec_debugpost(x, y)	do { } while (0)
979 #define	vop_fplookup_symlink_debugpre(x)	do { } while (0)
980 #define	vop_fplookup_symlink_debugpost(x, y)	do { } while (0)
981 #define	vop_fsync_debugpre(x)			do { } while (0)
982 #define	vop_fsync_debugpost(x, y)		do { } while (0)
983 #define	vop_strategy_debugpre(x)		do { } while (0)
984 #define	vop_lock_debugpre(x)			do { } while (0)
985 #define	vop_lock_debugpost(x, y)		do { } while (0)
986 #define	vop_unlock_debugpre(x)			do { } while (0)
987 #define	vop_need_inactive_debugpre(x)		do { } while (0)
988 #define	vop_need_inactive_debugpost(x, y)	do { } while (0)
989 #define	vop_mkdir_debugpost(x, y)		do { } while (0)
990 #endif
991 
992 void	vop_rename_fail(struct vop_rename_args *ap);
993 
994 #define	vop_stat_helper_pre(ap)	({						\
995 	struct vop_stat_args *_ap = (ap);					\
996 	int _error;								\
997 	AUDIT_ARG_VNODE1(ap->a_vp);						\
998 	_error = mac_vnode_check_stat(_ap->a_active_cred, _ap->a_file_cred, _ap->a_vp);\
999 	if (__predict_true(_error == 0)) {					\
1000 		ap->a_sb->st_padding1 = 0;					\
1001 		bzero(_ap->a_sb->st_spare, sizeof(_ap->a_sb->st_spare));	\
1002 		ap->a_sb->st_filerev = 0;					\
1003 		ap->a_sb->st_bsdflags = 0;					\
1004 	}									\
1005 	_error;									\
1006 })
1007 
1008 #define	vop_stat_helper_post(ap, error)	({					\
1009 	struct vop_stat_args *_ap = (ap);					\
1010 	int _error = (error);							\
1011 	if (priv_check_cred_vfs_generation(_ap->a_active_cred))			\
1012 		_ap->a_sb->st_gen = 0;						\
1013 	_error;									\
1014 })
1015 
1016 #ifdef INVARIANTS
1017 #define	vop_readdir_pre_assert(ap)					\
1018 	ssize_t nresid, oresid;						\
1019 									\
1020 	oresid = (ap)->a_uio->uio_resid;
1021 
1022 #define	vop_readdir_post_assert(ap, ret)				\
1023 	nresid = (ap)->a_uio->uio_resid;				\
1024 	if ((ret) == 0 && (ap)->a_eofflag != NULL) {			\
1025 		VNASSERT(oresid == 0 || nresid != oresid ||		\
1026 		    *(ap)->a_eofflag == 1,				\
1027 		    (ap)->a_vp, ("VOP_READDIR: eofflag not set"));	\
1028 	}
1029 #else
1030 #define	vop_readdir_pre_assert(ap)
1031 #define	vop_readdir_post_assert(ap, ret)
1032 #endif
1033 
1034 #define	vop_readdir_pre(ap) do {					\
1035 	vop_readdir_pre_assert(ap)
1036 
1037 #define vop_readdir_post(ap, ret)					\
1038 	vop_readdir_post_assert(ap, ret);				\
1039 	if ((ret) == 0) {						\
1040 		VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_READ);		\
1041 		INOTIFY((ap)->a_vp, IN_ACCESS);				\
1042 	}								\
1043 } while (0)
1044 
1045 #define	vop_write_pre(ap)						\
1046 	struct vattr va;						\
1047 	int error;							\
1048 	off_t osize, ooffset, noffset;					\
1049 									\
1050 	osize = ooffset = noffset = 0;					\
1051 	if ((vn_irflag_read((ap)->a_vp) & VIRF_KNOTE) != 0) {		\
1052 		error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred);	\
1053 		if (error)						\
1054 			return (error);					\
1055 		ooffset = (ap)->a_uio->uio_offset;			\
1056 		osize = (off_t)va.va_size;				\
1057 	}
1058 
1059 #define vop_write_post(ap, ret)						\
1060 	noffset = (ap)->a_uio->uio_offset;				\
1061 	if (noffset > ooffset) {					\
1062 		VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE |		\
1063 		    (noffset > osize ? NOTE_EXTEND : 0));		\
1064 		INOTIFY((ap)->a_vp, IN_MODIFY);				\
1065 	}
1066 
1067 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
1068 
1069 #ifdef INVARIANTS
1070 #define	VOP_ADD_WRITECOUNT_CHECKED(vp, cnt)				\
1071 do {									\
1072 	int error_;							\
1073 									\
1074 	error_ = VOP_ADD_WRITECOUNT((vp), (cnt));			\
1075 	VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d",	\
1076 	    error_));							\
1077 } while (0)
1078 #define	VOP_SET_TEXT_CHECKED(vp)					\
1079 do {									\
1080 	int error_;							\
1081 									\
1082 	error_ = VOP_SET_TEXT((vp));					\
1083 	VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d",	\
1084 	    error_));							\
1085 } while (0)
1086 #define	VOP_UNSET_TEXT_CHECKED(vp)					\
1087 do {									\
1088 	int error_;							\
1089 									\
1090 	error_ = VOP_UNSET_TEXT((vp));					\
1091 	VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d",	\
1092 	    error_));							\
1093 } while (0)
1094 #else
1095 #define	VOP_ADD_WRITECOUNT_CHECKED(vp, cnt)	VOP_ADD_WRITECOUNT((vp), (cnt))
1096 #define	VOP_SET_TEXT_CHECKED(vp)		VOP_SET_TEXT((vp))
1097 #define	VOP_UNSET_TEXT_CHECKED(vp)		VOP_UNSET_TEXT((vp))
1098 #endif
1099 
1100 #define	VN_IS_DOOMED(vp)	__predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0)
1101 
1102 void	vput(struct vnode *vp);
1103 void	vrele(struct vnode *vp);
1104 void	vref(struct vnode *vp);
1105 void	vrefact(struct vnode *vp);
1106 void 	v_addpollinfo(struct vnode *vp);
1107 static __inline int
vrefcnt(struct vnode * vp)1108 vrefcnt(struct vnode *vp)
1109 {
1110 
1111 	return (vp->v_usecount);
1112 }
1113 
1114 #define	vholdl(vp)	do {						\
1115 	ASSERT_VI_LOCKED(vp, __func__);					\
1116 	vhold(vp);							\
1117 } while (0)
1118 
1119 #define	vrefl(vp)	do {						\
1120 	ASSERT_VI_LOCKED(vp, __func__);					\
1121 	vref(vp);							\
1122 } while (0)
1123 
1124 /*
1125  * The caller doesn't know the file size and vnode_create_vobject() should
1126  * determine the size on its own.
1127  */
1128 #define	VNODE_NO_SIZE	((off_t)-1)
1129 
1130 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
1131 int vnode_create_disk_vobject(struct vnode *vp, off_t size, struct thread *td);
1132 void vnode_destroy_vobject(struct vnode *vp);
1133 
1134 extern struct vop_vector fifo_specops;
1135 extern struct vop_vector dead_vnodeops;
1136 extern struct vop_vector default_vnodeops;
1137 
1138 #define VOP_PANIC	((void*)(uintptr_t)vop_panic)
1139 #define VOP_NULL	((void*)(uintptr_t)vop_null)
1140 #define VOP_EBADF	((void*)(uintptr_t)vop_ebadf)
1141 #define VOP_ENOTTY	((void*)(uintptr_t)vop_enotty)
1142 #define VOP_EINVAL	((void*)(uintptr_t)vop_einval)
1143 #define VOP_ENOENT	((void*)(uintptr_t)vop_enoent)
1144 #define VOP_EOPNOTSUPP	((void*)(uintptr_t)vop_eopnotsupp)
1145 #define VOP_EAGAIN	((void*)(uintptr_t)vop_eagain)
1146 
1147 /* fifo_vnops.c */
1148 int	fifo_printinfo(struct vnode *);
1149 
1150 /* vfs_hash.c */
1151 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
1152 
1153 void vfs_hash_changesize(u_long newhashsize);
1154 int vfs_hash_get(const struct mount *mp, u_int hash, int flags,
1155     struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1156 u_int vfs_hash_index(struct vnode *vp);
1157 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
1158     struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1159 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
1160     struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1161 void vfs_hash_rehash(struct vnode *vp, u_int hash);
1162 void vfs_hash_remove(struct vnode *vp);
1163 
1164 int vfs_kqfilter(struct vop_kqfilter_args *);
1165 struct dirent;
1166 int vn_dir_next_dirent(struct vnode *vp, struct thread *td,
1167     char *dirbuf, size_t dirbuflen,
1168     struct dirent **dpp, size_t *len, off_t *off, int *eofflag);
1169 int vn_dir_check_empty(struct vnode *vp);
1170 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
1171 
1172 int vfs_unixify_accmode(accmode_t *accmode);
1173 
1174 void vfs_unp_reclaim(struct vnode *vp);
1175 
1176 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
1177 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
1178     gid_t gid);
1179 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
1180     struct thread *td);
1181 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
1182     struct thread *td);
1183 int vn_getsize_locked(struct vnode *vp, off_t *size, struct ucred *active_cred);
1184 int vn_getsize(struct vnode *vp, off_t *size, struct ucred *active_cred);
1185 
1186 void vn_fsid(struct vnode *vp, struct vattr *va);
1187 
1188 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp);
1189 int vn_lktype_write(struct mount *mp, struct vnode *vp);
1190 
1191 #ifdef INVARIANTS
1192 void vn_set_state_validate(struct vnode *vp, __enum_uint8(vstate) state);
1193 #endif
1194 
1195 static inline void
vn_set_state(struct vnode * vp,__enum_uint8 (vstate)state)1196 vn_set_state(struct vnode *vp, __enum_uint8(vstate) state)
1197 {
1198 #ifdef INVARIANTS
1199 	vn_set_state_validate(vp, state);
1200 #endif
1201 	vp->v_state = state;
1202 }
1203 
1204 static inline __enum_uint8(vstate)
vn_get_state(struct vnode * vp)1205 vn_get_state(struct vnode *vp)
1206 {
1207 	return (vp->v_state);
1208 }
1209 
1210 #define VOP_UNLOCK_FLAGS(vp, flags)	({				\
1211 	struct vnode *_vp = (vp);					\
1212 	int _flags = (flags);						\
1213 	int _error;							\
1214 									\
1215         if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0)		\
1216                 panic("%s: unsupported flags %x\n", __func__, flags);	\
1217         _error = VOP_UNLOCK(_vp);					\
1218         if (_flags & LK_INTERLOCK)					\
1219                 VI_UNLOCK(_vp);						\
1220         _error;								\
1221 })
1222 
1223 #include <sys/kernel.h>
1224 
1225 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \
1226 	SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \
1227 	    vfs_vector_op_register, &vnodeops)
1228 
1229 #define VFS_SMR_DECLARE				\
1230 	extern smr_t vfs_smr
1231 
1232 #define VFS_SMR()	vfs_smr
1233 #define vfs_smr_enter()	smr_enter(VFS_SMR())
1234 #define vfs_smr_exit()	smr_exit(VFS_SMR())
1235 #define vfs_smr_synchronize()	smr_synchronize(VFS_SMR())
1236 #define vfs_smr_entered_load(ptr)	smr_entered_load((ptr), VFS_SMR())
1237 #define VFS_SMR_ENTERED()		SMR_ENTERED(VFS_SMR())
1238 #define VFS_SMR_ASSERT_ENTERED()	SMR_ASSERT_ENTERED(VFS_SMR())
1239 #define VFS_SMR_ASSERT_NOT_ENTERED()	SMR_ASSERT_NOT_ENTERED(VFS_SMR())
1240 #define VFS_SMR_ZONE_SET(zone)	uma_zone_set_smr((zone), VFS_SMR())
1241 
1242 #define vn_load_v_data_smr(vp)	({		\
1243 	struct vnode *_vp = (vp);		\
1244 						\
1245 	VFS_SMR_ASSERT_ENTERED();		\
1246 	atomic_load_consume_ptr(&(_vp)->v_data);\
1247 })
1248 
1249 static inline void
vn_delayed_setsize_locked(struct vnode * vp)1250 vn_delayed_setsize_locked(struct vnode *vp)
1251 {
1252 	ASSERT_VI_LOCKED(vp, "delayed_setsize");
1253 	vp->v_iflag |= VI_DELAYED_SETSIZE;
1254 }
1255 
1256 static inline void
vn_delayed_setsize(struct vnode * vp)1257 vn_delayed_setsize(struct vnode *vp)
1258 {
1259 	VI_LOCK(vp);
1260 	vn_delayed_setsize_locked(vp);
1261 	VI_UNLOCK(vp);
1262 }
1263 
1264 static inline void
vn_clear_delayed_setsize_locked(struct vnode * vp)1265 vn_clear_delayed_setsize_locked(struct vnode *vp)
1266 {
1267 	ASSERT_VI_LOCKED(vp, "delayed_setsize");
1268 	vp->v_iflag &= ~VI_DELAYED_SETSIZE;
1269 }
1270 
1271 static inline void
vn_clear_delayed_setsize(struct vnode * vp)1272 vn_clear_delayed_setsize(struct vnode *vp)
1273 {
1274 	VI_LOCK(vp);
1275 	vn_clear_delayed_setsize_locked(vp);
1276 	VI_UNLOCK(vp);
1277 }
1278 
1279 #endif /* _KERNEL */
1280 
1281 #endif /* !_SYS_VNODE_H_ */
1282