2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1989, 1993
5 * The Regents of the University of California. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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.
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
31 * @(#)vnode.h 8.7 (Berkeley) 2/4/94
38 #include <sys/bufobj.h>
39 #include <sys/queue.h>
41 #include <sys/lockmgr.h>
42 #include <sys/mutex.h>
43 #include <sys/rangelock.h>
44 #include <sys/selinfo.h>
48 #include <sys/_seqc.h>
51 * The vnode is the focus of all file activity in UNIX. There is a
52 * unique vnode allocated for each active file, each current directory,
53 * each mounted-on file, text file, and the root.
57 * Vnode types. VNON means no type.
59 enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD,
62 enum vgetstate { VGET_NONE, VGET_HOLDCNT, VGET_USECOUNT };
64 * Each underlying filesystem allocates its own private area and hangs
65 * it from v_data. If non-null, this area is freed in getnewvnode().
71 struct mtx vpi_lock; /* lock to protect below */
72 struct selinfo vpi_selinfo; /* identity of poller(s) */
73 short vpi_events; /* what they are looking for */
74 short vpi_revents; /* what has happened */
78 * Reading or writing any of these items requires holding the appropriate lock.
83 * l - mp mnt_listmtx or freelist mutex
84 * I - updated with atomics, 0->1 and 1->0 transitions with interlock held
85 * m - mount point interlock
87 * u - Only a reference to the vnode is needed to read.
90 * Vnodes may be found on many lists. The general way to deal with operating
91 * on a vnode that is on a list is:
92 * 1) Lock the list and find the vnode.
93 * 2) Lock interlock so that the vnode does not go away.
94 * 3) Unlock the list to avoid lock order reversals.
95 * 4) vget with LK_INTERLOCK and check for ENOENT, or
96 * 5) Check for DOOMED if the vnode lock is not required.
97 * 6) Perform your operation, then vput().
100 #if defined(_KERNEL) || defined(_KVM_VNODE)
104 * Fields which define the identity of the vnode. These fields are
105 * owned by the filesystem (XXX: and vgone() ?)
107 enum vtype v_type:8; /* u vnode type */
108 short v_irflag; /* i frequently read flags */
109 seqc_t v_seqc; /* i modification count */
110 uint32_t v_nchash; /* u namecache hash */
111 struct vop_vector *v_op; /* u vnode operations vector */
112 void *v_data; /* u private data for fs */
115 * Filesystem instance stuff
117 struct mount *v_mount; /* u ptr to vfs we are in */
118 TAILQ_ENTRY(vnode) v_nmntvnodes; /* m vnodes for mount point */
121 * Type specific fields, only one applies to any given vnode.
124 struct mount *v_mountedhere; /* v ptr to mountpoint (VDIR) */
125 struct unpcb *v_unpcb; /* v unix domain net (VSOCK) */
126 struct cdev *v_rdev; /* v device (VCHR, VBLK) */
127 struct fifoinfo *v_fifoinfo; /* v fifo (VFIFO) */
131 * vfs_hash: (mount + inode) -> vnode hash. The hash value
132 * itself is grouped with other int fields, to avoid padding.
134 LIST_ENTRY(vnode) v_hashlist;
137 * VFS_namecache stuff
139 LIST_HEAD(, namecache) v_cache_src; /* c Cache entries from us */
140 TAILQ_HEAD(, namecache) v_cache_dst; /* c Cache entries to us */
141 struct namecache *v_cache_dd; /* c Cache entry for .. vnode */
146 struct lock v_lock; /* u (if fs don't have one) */
147 struct mtx v_interlock; /* lock for "i" things */
148 struct lock *v_vnlock; /* u pointer to vnode lock */
151 * The machinery of being a vnode
153 TAILQ_ENTRY(vnode) v_vnodelist; /* l vnode lists */
154 TAILQ_ENTRY(vnode) v_lazylist; /* l vnode lazy list */
155 struct bufobj v_bufobj; /* * Buffer cache object */
158 * Hooks for various subsystems and features.
160 struct vpollinfo *v_pollinfo; /* i Poll events, p for *v_pi */
161 struct label *v_label; /* MAC label for vnode */
162 struct lockf *v_lockf; /* Byte-level advisory lock list */
163 struct rangelock v_rl; /* Byte-range lock */
168 daddr_t v_cstart; /* v start block of cluster */
169 daddr_t v_lasta; /* v last allocation */
170 daddr_t v_lastw; /* v last write */
171 int v_clen; /* v length of cur. cluster */
173 u_int v_holdcnt; /* I prevents recycling. */
174 u_int v_usecount; /* I ref count of users */
175 u_short v_iflag; /* i vnode flags (see below) */
176 u_short v_vflag; /* v vnode flags */
177 u_short v_mflag; /* l mnt-specific vnode flags */
178 short v_dbatchcpu; /* i LRU requeue deferral batch */
179 int v_writecount; /* I ref count of writers or
180 (negative) text users */
181 int v_seqc_users; /* i modifications pending */
185 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
187 #define bo2vnode(bo) __containerof((bo), struct vnode, v_bufobj)
189 /* XXX: These are temporary to avoid a source sweep at this time */
190 #define v_object v_bufobj.bo_object
193 * Userland version of struct vnode, for sysctl.
196 size_t xv_size; /* sizeof(struct xvnode) */
197 void *xv_vnode; /* address of real vnode */
198 u_long xv_flag; /* vnode vflags */
199 int xv_usecount; /* reference count of users */
200 int xv_writecount; /* reference count of writers */
201 int xv_holdcnt; /* page & buffer references */
202 u_long xv_id; /* capability identifier */
203 void *xv_mount; /* address of parent mount */
204 long xv_numoutput; /* num of writes in progress */
205 enum vtype xv_type; /* vnode type */
207 void *xvu_socket; /* unpcb, if VSOCK */
208 void *xvu_fifo; /* fifo, if VFIFO */
209 dev_t xvu_rdev; /* maj/min, if VBLK/VCHR */
211 dev_t xvu_dev; /* device, if VDIR/VREG/VLNK */
212 ino_t xvu_ino; /* id, if VDIR/VREG/VLNK */
216 #define xv_socket xv_un.xvu_socket
217 #define xv_fifo xv_un.xvu_fifo
218 #define xv_rdev xv_un.xvu_rdev
219 #define xv_dev xv_un.xv_uns.xvu_dev
220 #define xv_ino xv_un.xv_uns.xvu_ino
222 /* We don't need to lock the knlist */
223 #define VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL || \
224 KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
226 #define VN_KNOTE(vp, b, a) \
228 if (!VN_KNLIST_EMPTY(vp)) \
229 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
230 (a) | KNF_NOKQLOCK); \
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)
237 * VI flags are protected by interlock and live in v_iflag
238 * VV flags are protected by the vnode lock and live in v_vflag
240 * VIRF_DOOMED is doubly protected by the interlock and vnode lock. Both
241 * are required for writing but the status may be checked with either.
243 #define VHOLD_NO_SMR (1<<29) /* Disable vhold_smr */
244 #define VHOLD_ALL_FLAGS (VHOLD_NO_SMR)
246 #define VIRF_DOOMED 0x0001 /* This vnode is being recycled */
247 #define VIRF_PGREAD 0x0002 /* Direct reads from the page cache are permitted,
248 never cleared once set */
250 #define VI_TEXT_REF 0x0001 /* Text ref grabbed use ref */
251 #define VI_MOUNT 0x0002 /* Mount in progress */
252 #define VI_DOINGINACT 0x0004 /* VOP_INACTIVE is in progress */
253 #define VI_OWEINACT 0x0008 /* Need to call inactive */
254 #define VI_DEFINACT 0x0010 /* deferred inactive */
256 #define VV_ROOT 0x0001 /* root of its filesystem */
257 #define VV_ISTTY 0x0002 /* vnode represents a tty */
258 #define VV_NOSYNC 0x0004 /* unlinked, stop syncing */
259 #define VV_ETERNALDEV 0x0008 /* device that is never destroyed */
260 #define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */
261 #define VV_VMSIZEVNLOCK 0x0020 /* object size check requires vnode lock */
262 #define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */
263 #define VV_SYSTEM 0x0080 /* vnode being used by kernel */
264 #define VV_PROCDEP 0x0100 /* vnode is process dependent */
265 #define VV_NOKNOTE 0x0200 /* don't activate knotes on this vnode */
266 #define VV_DELETED 0x0400 /* should be removed */
267 #define VV_MD 0x0800 /* vnode backs the md device */
268 #define VV_FORCEINSMQ 0x1000 /* force the insmntque to succeed */
269 #define VV_READLINK 0x2000 /* fdescfs linux vnode */
271 #define VMP_LAZYLIST 0x0001 /* Vnode is on mnt's lazy list */
274 * Vnode attributes. A field value of VNOVAL represents a field whose value
275 * is unavailable (getattr) or which is not to be changed (setattr).
278 enum vtype va_type; /* vnode type (for create) */
279 u_short va_mode; /* files access mode and type */
281 uid_t va_uid; /* owner user id */
282 gid_t va_gid; /* owner group id */
283 nlink_t va_nlink; /* number of references to file */
284 dev_t va_fsid; /* filesystem id */
285 ino_t va_fileid; /* file id */
286 u_quad_t va_size; /* file size in bytes */
287 long va_blocksize; /* blocksize preferred for i/o */
288 struct timespec va_atime; /* time of last access */
289 struct timespec va_mtime; /* time of last modification */
290 struct timespec va_ctime; /* time file changed */
291 struct timespec va_birthtime; /* time file created */
292 u_long va_gen; /* generation number of file */
293 u_long va_flags; /* flags defined for file */
294 dev_t va_rdev; /* device the special file represents */
295 u_quad_t va_bytes; /* bytes of disk space held by file */
296 u_quad_t va_filerev; /* file modification number */
297 u_int va_vaflags; /* operations flags, see below */
298 long va_spare; /* remain quad aligned */
302 * Flags for va_vaflags.
304 #define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */
305 #define VA_EXCLUSIVE 0x02 /* exclusive create request */
306 #define VA_SYNC 0x04 /* O_SYNC truncation */
309 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
310 * help with write clustering)
311 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
313 #define IO_UNIT 0x0001 /* do I/O as atomic unit */
314 #define IO_APPEND 0x0002 /* append write to end */
315 #define IO_NDELAY 0x0004 /* FNDELAY flag set in file table */
316 #define IO_NODELOCKED 0x0008 /* underlying node already locked */
317 #define IO_ASYNC 0x0010 /* bawrite rather then bdwrite */
318 #define IO_VMIO 0x0020 /* data already in VMIO space */
319 #define IO_INVAL 0x0040 /* invalidate after I/O */
320 #define IO_SYNC 0x0080 /* do I/O synchronously */
321 #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */
322 #define IO_NOREUSE 0x0200 /* VMIO data won't be reused */
323 #define IO_EXT 0x0400 /* operate on external attributes */
324 #define IO_NORMAL 0x0800 /* operate on regular data */
325 #define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */
326 #define IO_BUFLOCKED 0x2000 /* ffs flag; indir buf is locked */
327 #define IO_RANGELOCKED 0x4000 /* range locked */
329 #define IO_SEQMAX 0x7F /* seq heuristic max value */
330 #define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */
333 * Flags for accmode_t.
335 #define VEXEC 000000000100 /* execute/search permission */
336 #define VWRITE 000000000200 /* write permission */
337 #define VREAD 000000000400 /* read permission */
338 #define VADMIN 000000010000 /* being the file owner */
339 #define VAPPEND 000000040000 /* permission to write/append */
341 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
342 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
343 * and 0 otherwise. This never happens with ordinary unix access rights
344 * or POSIX.1e ACLs. Obviously, VEXPLICIT_DENY must be OR-ed with
345 * some other V* constant.
347 #define VEXPLICIT_DENY 000000100000
348 #define VREAD_NAMED_ATTRS 000000200000 /* not used */
349 #define VWRITE_NAMED_ATTRS 000000400000 /* not used */
350 #define VDELETE_CHILD 000001000000
351 #define VREAD_ATTRIBUTES 000002000000 /* permission to stat(2) */
352 #define VWRITE_ATTRIBUTES 000004000000 /* change {m,c,a}time */
353 #define VDELETE 000010000000
354 #define VREAD_ACL 000020000000 /* read ACL and file mode */
355 #define VWRITE_ACL 000040000000 /* change ACL and/or file mode */
356 #define VWRITE_OWNER 000100000000 /* change file owner */
357 #define VSYNCHRONIZE 000200000000 /* not used */
358 #define VCREAT 000400000000 /* creating new file */
359 #define VVERIFY 001000000000 /* verification required */
362 * Permissions that were traditionally granted only to the file owner.
364 #define VADMIN_PERMS (VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
368 * Permissions that were traditionally granted to everyone.
370 #define VSTAT_PERMS (VREAD_ATTRIBUTES | VREAD_ACL)
373 * Permissions that allow to change the state of the file in any way.
375 #define VMODIFY_PERMS (VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
379 * Token indicating no attribute value yet assigned.
384 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
386 #define VLKTIMEOUT (hz / 20 + 1)
390 #ifdef MALLOC_DECLARE
391 MALLOC_DECLARE(M_VNODE);
394 extern u_int ncsizefactor;
395 extern const u_int io_hold_cnt;
398 * Convert between vnode types and inode formats (since POSIX.1
399 * defines mode word of stat structure in terms of inode formats).
401 extern enum vtype iftovt_tab[];
402 extern int vttoif_tab[];
403 #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12])
404 #define VTTOIF(indx) (vttoif_tab[(int)(indx)])
405 #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode))
408 * Flags to various vnode functions.
410 #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */
411 #define FORCECLOSE 0x0002 /* vflush: force file closure */
412 #define WRITECLOSE 0x0004 /* vflush: only close writable files */
413 #define EARLYFLUSH 0x0008 /* vflush: early call for ffs_flushfiles */
414 #define V_SAVE 0x0001 /* vinvalbuf: sync file first */
415 #define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */
416 #define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */
417 #define V_CLEANONLY 0x0008 /* vinvalbuf: invalidate only clean bufs */
418 #define V_VMIO 0x0010 /* vinvalbuf: called during pageout */
419 #define V_ALLOWCLEAN 0x0020 /* vinvalbuf: allow clean buffers after flush */
420 #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */
421 #define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */
422 #define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */
423 #define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */
424 #define V_MNTREF 0x0010 /* vn_start_write: mp is already ref-ed */
426 #define VR_START_WRITE 0x0001 /* vfs_write_resume: start write atomically */
427 #define VR_NO_SUSPCLR 0x0002 /* vfs_write_resume: do not clear suspension */
429 #define VS_SKIP_UNMOUNT 0x0001 /* vfs_write_suspend: fail if the
430 filesystem is being unmounted */
432 #define VREF(vp) vref(vp)
435 #define VATTR_NULL(vap) vattr_null(vap)
437 #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */
438 #endif /* DIAGNOSTIC */
440 #define NULLVP ((struct vnode *)NULL)
445 extern struct vnode *rootvnode; /* root (i.e. "/") vnode */
446 extern struct mount *rootdevmp; /* "/dev" mount */
447 extern u_long desiredvnodes; /* number of vnodes desired */
448 extern struct uma_zone *namei_zone;
449 extern struct vattr va_null; /* predefined null vattr structure */
451 #define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock)
452 #define VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
453 #define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock)
454 #define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock)
455 #define VI_MTX(vp) (&(vp)->v_interlock)
457 #define VN_LOCK_AREC(vp) lockallowrecurse((vp)->v_vnlock)
458 #define VN_LOCK_ASHARE(vp) lockallowshare((vp)->v_vnlock)
459 #define VN_LOCK_DSHARE(vp) lockdisableshare((vp)->v_vnlock)
464 * Mods for extensibility.
468 * Flags for vdesc_flags:
470 #define VDESC_MAX_VPS 16
471 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
472 #define VDESC_VP0_WILLRELE 0x0001
473 #define VDESC_VP1_WILLRELE 0x0002
474 #define VDESC_VP2_WILLRELE 0x0004
475 #define VDESC_VP3_WILLRELE 0x0008
478 * A generic structure.
479 * This can be used by bypass routines to identify generic arguments.
481 struct vop_generic_args {
482 struct vnodeop_desc *a_desc;
483 /* other random data follows, presumably */
486 typedef int vop_bypass_t(struct vop_generic_args *);
489 * VDESC_NO_OFFSET is used to identify the end of the offset list
490 * and in places where no such field exists.
492 #define VDESC_NO_OFFSET -1
495 * This structure describes the vnode operation taking place.
497 struct vnodeop_desc {
498 char *vdesc_name; /* a readable name for debugging */
499 int vdesc_flags; /* VDESC_* flags */
500 int vdesc_vop_offset;
501 vop_bypass_t *vdesc_call; /* Function to call */
504 * These ops are used by bypass routines to map and locate arguments.
505 * Creds and procs are not needed in bypass routines, but sometimes
506 * they are useful to (for example) transport layers.
507 * Nameidata is useful because it has a cred in it.
509 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */
510 int vdesc_vpp_offset; /* return vpp location */
511 int vdesc_cred_offset; /* cred location, if any */
512 int vdesc_thread_offset; /* thread location, if any */
513 int vdesc_componentname_offset; /* if any */
518 * A list of all the operation descs.
520 extern struct vnodeop_desc *vnodeop_descs[];
522 #define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field)
523 #define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
524 ((s_type)(((char*)(struct_p)) + (s_offset)))
526 #ifdef DEBUG_VFS_LOCKS
528 * Support code to aid in debugging VFS locking problems. Not totally
529 * reliable since if the thread sleeps between changing the lock
530 * state and checking it with the assert, some other thread could
531 * change the state. They are good enough for debugging a single
532 * filesystem using a single-threaded test. Note that the unreliability is
533 * limited to false negatives; efforts were made to ensure that false
534 * positives cannot occur.
536 void assert_vi_locked(struct vnode *vp, const char *str);
537 void assert_vi_unlocked(struct vnode *vp, const char *str);
538 void assert_vop_elocked(struct vnode *vp, const char *str);
539 void assert_vop_locked(struct vnode *vp, const char *str);
540 void assert_vop_unlocked(struct vnode *vp, const char *str);
542 #define ASSERT_VI_LOCKED(vp, str) assert_vi_locked((vp), (str))
543 #define ASSERT_VI_UNLOCKED(vp, str) assert_vi_unlocked((vp), (str))
544 #define ASSERT_VOP_ELOCKED(vp, str) assert_vop_elocked((vp), (str))
545 #define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked((vp), (str))
546 #define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked((vp), (str))
548 #define ASSERT_VOP_IN_SEQC(vp) do { \
549 struct vnode *_vp = (vp); \
551 VNPASS(seqc_in_modify(_vp->v_seqc), _vp); \
554 #define ASSERT_VOP_NOT_IN_SEQC(vp) do { \
555 struct vnode *_vp = (vp); \
557 VNPASS(!seqc_in_modify(_vp->v_seqc), _vp); \
560 #else /* !DEBUG_VFS_LOCKS */
562 #define ASSERT_VI_LOCKED(vp, str) ((void)0)
563 #define ASSERT_VI_UNLOCKED(vp, str) ((void)0)
564 #define ASSERT_VOP_ELOCKED(vp, str) ((void)0)
565 #define ASSERT_VOP_LOCKED(vp, str) ((void)0)
566 #define ASSERT_VOP_UNLOCKED(vp, str) ((void)0)
568 #define ASSERT_VOP_IN_SEQC(vp) ((void)0)
569 #define ASSERT_VOP_NOT_IN_SEQC(vp) ((void)0)
571 #endif /* DEBUG_VFS_LOCKS */
574 * This call works for vnodes in the kernel.
576 #define VCALL(c) ((c)->a_desc->vdesc_call(c))
578 #define DOINGASYNC(vp) \
579 (((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 && \
580 ((curthread->td_pflags & TDP_SYNCIO) == 0))
583 * VMIO support inline
586 extern int vmiodirenable;
589 vn_canvmio(struct vnode *vp)
591 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
597 * Finally, include the default set of vnode operations.
599 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
600 #include "vnode_if.h"
603 #define VN_OPEN_NOAUDIT 0x00000001
604 #define VN_OPEN_NOCAPCHECK 0x00000002
605 #define VN_OPEN_NAMECACHE 0x00000004
606 #define VN_OPEN_INVFS 0x00000008
609 * Public vnode manipulation functions.
611 struct componentname;
616 struct freebsd11_stat;
627 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
629 int bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
631 /* cache_* may belong in namei.h. */
632 void cache_changesize(u_long newhashsize);
633 #define cache_enter(dvp, vp, cnp) \
634 cache_enter_time(dvp, vp, cnp, NULL, NULL)
635 void cache_enter_time(struct vnode *dvp, struct vnode *vp,
636 struct componentname *cnp, struct timespec *tsp,
637 struct timespec *dtsp);
638 int cache_lookup(struct vnode *dvp, struct vnode **vpp,
639 struct componentname *cnp, struct timespec *tsp, int *ticksp);
640 void cache_vnode_init(struct vnode *vp);
641 void cache_purge(struct vnode *vp);
642 void cache_purge_vgone(struct vnode *vp);
643 void cache_purge_negative(struct vnode *vp);
644 void cache_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp,
645 struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp);
646 void cache_purgevfs(struct mount *mp);
648 void cache_validate(struct vnode *dvp, struct vnode *vp,
649 struct componentname *cnp);
652 cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
656 int change_dir(struct vnode *vp, struct thread *td);
657 void cvtstat(struct stat *st, struct ostat *ost);
658 void freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb);
659 int freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost);
660 int getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
662 void getnewvnode_reserve(void);
663 void getnewvnode_drop_reserve(void);
664 int insmntque1(struct vnode *vp, struct mount *mp,
665 void (*dtr)(struct vnode *, void *), void *dtr_arg);
666 int insmntque(struct vnode *vp, struct mount *mp);
667 u_quad_t init_va_filerev(void);
668 int speedup_syncer(void);
669 int vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen);
670 int vn_getcwd(char *buf, char **retbuf, size_t *buflen);
671 int vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf);
672 int vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf);
674 vn_dir_dd_ino(struct vnode *vp);
675 int vn_commname(struct vnode *vn, char *buf, u_int buflen);
676 int vn_path_to_global_path(struct thread *td, struct vnode *vp,
677 char *path, u_int pathlen);
678 int vaccess(enum vtype type, mode_t file_mode, uid_t file_uid,
679 gid_t file_gid, accmode_t accmode, struct ucred *cred);
680 int vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid,
682 int vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid,
683 struct acl *aclp, accmode_t accmode, struct ucred *cred);
684 int vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
685 gid_t file_gid, struct acl *acl, accmode_t accmode,
687 void vattr_null(struct vattr *vap);
688 void vlazy(struct vnode *);
689 void vdrop(struct vnode *);
690 void vdropl(struct vnode *);
691 int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
692 int vget(struct vnode *vp, int flags);
693 enum vgetstate vget_prep_smr(struct vnode *vp);
694 enum vgetstate vget_prep(struct vnode *vp);
695 int vget_finish(struct vnode *vp, int flags, enum vgetstate vs);
696 void vget_finish_ref(struct vnode *vp, enum vgetstate vs);
697 void vget_abort(struct vnode *vp, enum vgetstate vs);
698 void vgone(struct vnode *vp);
699 void vhold(struct vnode *);
700 void vholdnz(struct vnode *);
701 bool vhold_smr(struct vnode *);
702 void vinactive(struct vnode *vp);
703 int vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
704 int vtruncbuf(struct vnode *vp, off_t length, int blksize);
705 void v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn,
707 void vunref(struct vnode *);
708 void vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
709 int vrecycle(struct vnode *vp);
710 int vrecyclel(struct vnode *vp);
711 int vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
713 int vn_close(struct vnode *vp,
714 int flags, struct ucred *file_cred, struct thread *td);
715 int vn_copy_file_range(struct vnode *invp, off_t *inoffp,
716 struct vnode *outvp, off_t *outoffp, size_t *lenp,
717 unsigned int flags, struct ucred *incred, struct ucred *outcred,
718 struct thread *fsize_td);
719 void vn_finished_write(struct mount *mp);
720 void vn_finished_secondary_write(struct mount *mp);
721 int vn_fsync_buf(struct vnode *vp, int waitfor);
722 int vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp,
723 struct vnode *outvp, off_t *outoffp, size_t *lenp,
724 unsigned int flags, struct ucred *incred, struct ucred *outcred,
725 struct thread *fsize_td);
726 int vn_need_pageq_flush(struct vnode *vp);
727 bool vn_isdisk_error(struct vnode *vp, int *errp);
728 bool vn_isdisk(struct vnode *vp);
729 int _vn_lock(struct vnode *vp, int flags, const char *file, int line);
730 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
731 int vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
732 int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
733 u_int vn_open_flags, struct ucred *cred, struct file *fp);
734 int vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
735 struct thread *td, struct file *fp);
736 void vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
737 int vn_pollrecord(struct vnode *vp, struct thread *p, int events);
738 int vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
739 int len, off_t offset, enum uio_seg segflg, int ioflg,
740 struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
742 int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
743 size_t len, off_t offset, enum uio_seg segflg, int ioflg,
744 struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
746 int vn_read_from_obj(struct vnode *vp, struct uio *uio);
747 int vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio,
749 int vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
750 int vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
752 int vn_truncate_locked(struct vnode *vp, off_t length, bool sync,
754 int vn_writechk(struct vnode *vp);
755 int vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
756 const char *attrname, int *buflen, char *buf, struct thread *td);
757 int vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
758 const char *attrname, int buflen, char *buf, struct thread *td);
759 int vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
760 const char *attrname, struct thread *td);
761 int vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
763 int vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
764 void *alloc_arg, int lkflags, struct vnode **rvp);
765 int vn_utimes_perm(struct vnode *vp, struct vattr *vap,
766 struct ucred *cred, struct thread *td);
768 int vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
769 int vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
772 void vn_seqc_write_begin_unheld_locked(struct vnode *vp);
773 void vn_seqc_write_begin_unheld(struct vnode *vp);
774 void vn_seqc_write_begin_locked(struct vnode *vp);
775 void vn_seqc_write_begin(struct vnode *vp);
776 void vn_seqc_write_end_locked(struct vnode *vp);
777 void vn_seqc_write_end(struct vnode *vp);
778 #define vn_seqc_read_any(vp) seqc_read_any(&(vp)->v_seqc)
779 #define vn_seqc_consistent(vp, seq) seqc_consistent(&(vp)->v_seqc, seq)
781 #define vn_rangelock_unlock(vp, cookie) \
782 rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp))
783 #define vn_rangelock_unlock_range(vp, cookie, start, end) \
784 rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), \
786 #define vn_rangelock_rlock(vp, start, end) \
787 rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
788 #define vn_rangelock_tryrlock(vp, start, end) \
789 rangelock_tryrlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
790 #define vn_rangelock_wlock(vp, start, end) \
791 rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
792 #define vn_rangelock_trywlock(vp, start, end) \
793 rangelock_trywlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
795 int vfs_cache_lookup(struct vop_lookup_args *ap);
796 int vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp);
797 void vfs_timestamp(struct timespec *);
798 void vfs_write_resume(struct mount *mp, int flags);
799 int vfs_write_suspend(struct mount *mp, int flags);
800 int vfs_write_suspend_umnt(struct mount *mp);
801 void vnlru_free(int, struct vfsops *);
802 int vop_stdbmap(struct vop_bmap_args *);
803 int vop_stdfdatasync_buf(struct vop_fdatasync_args *);
804 int vop_stdfsync(struct vop_fsync_args *);
805 int vop_stdgetwritemount(struct vop_getwritemount_args *);
806 int vop_stdgetpages(struct vop_getpages_args *);
807 int vop_stdinactive(struct vop_inactive_args *);
808 int vop_stdioctl(struct vop_ioctl_args *);
809 int vop_stdneed_inactive(struct vop_need_inactive_args *);
810 int vop_stdkqfilter(struct vop_kqfilter_args *);
811 int vop_stdlock(struct vop_lock1_args *);
812 int vop_stdunlock(struct vop_unlock_args *);
813 int vop_stdislocked(struct vop_islocked_args *);
814 int vop_lock(struct vop_lock1_args *);
815 int vop_unlock(struct vop_unlock_args *);
816 int vop_islocked(struct vop_islocked_args *);
817 int vop_stdputpages(struct vop_putpages_args *);
818 int vop_nopoll(struct vop_poll_args *);
819 int vop_stdaccess(struct vop_access_args *ap);
820 int vop_stdaccessx(struct vop_accessx_args *ap);
821 int vop_stdadvise(struct vop_advise_args *ap);
822 int vop_stdadvlock(struct vop_advlock_args *ap);
823 int vop_stdadvlockasync(struct vop_advlockasync_args *ap);
824 int vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
825 int vop_stdallocate(struct vop_allocate_args *ap);
826 int vop_stdset_text(struct vop_set_text_args *ap);
827 int vop_stdpathconf(struct vop_pathconf_args *);
828 int vop_stdpoll(struct vop_poll_args *);
829 int vop_stdvptocnp(struct vop_vptocnp_args *ap);
830 int vop_stdvptofh(struct vop_vptofh_args *ap);
831 int vop_stdunp_bind(struct vop_unp_bind_args *ap);
832 int vop_stdunp_connect(struct vop_unp_connect_args *ap);
833 int vop_stdunp_detach(struct vop_unp_detach_args *ap);
834 int vop_eopnotsupp(struct vop_generic_args *ap);
835 int vop_ebadf(struct vop_generic_args *ap);
836 int vop_einval(struct vop_generic_args *ap);
837 int vop_enoent(struct vop_generic_args *ap);
838 int vop_enotty(struct vop_generic_args *ap);
839 int vop_eagain(struct vop_generic_args *ap);
840 int vop_null(struct vop_generic_args *ap);
841 int vop_panic(struct vop_generic_args *ap);
842 int dead_poll(struct vop_poll_args *ap);
843 int dead_read(struct vop_read_args *ap);
844 int dead_write(struct vop_write_args *ap);
846 /* These are called from within the actual VOPS. */
847 void vop_close_post(void *a, int rc);
848 void vop_create_pre(void *a);
849 void vop_create_post(void *a, int rc);
850 void vop_whiteout_pre(void *a);
851 void vop_whiteout_post(void *a, int rc);
852 void vop_deleteextattr_pre(void *a);
853 void vop_deleteextattr_post(void *a, int rc);
854 void vop_link_pre(void *a);
855 void vop_link_post(void *a, int rc);
856 void vop_lookup_post(void *a, int rc);
857 void vop_lookup_pre(void *a);
858 void vop_mkdir_pre(void *a);
859 void vop_mkdir_post(void *a, int rc);
860 void vop_mknod_pre(void *a);
861 void vop_mknod_post(void *a, int rc);
862 void vop_open_post(void *a, int rc);
863 void vop_read_post(void *a, int rc);
864 void vop_read_pgcache_post(void *ap, int rc);
865 void vop_readdir_post(void *a, int rc);
866 void vop_reclaim_post(void *a, int rc);
867 void vop_remove_pre(void *a);
868 void vop_remove_post(void *a, int rc);
869 void vop_rename_post(void *a, int rc);
870 void vop_rename_pre(void *a);
871 void vop_rmdir_pre(void *a);
872 void vop_rmdir_post(void *a, int rc);
873 void vop_setattr_pre(void *a);
874 void vop_setattr_post(void *a, int rc);
875 void vop_setacl_pre(void *a);
876 void vop_setacl_post(void *a, int rc);
877 void vop_setextattr_pre(void *a);
878 void vop_setextattr_post(void *a, int rc);
879 void vop_symlink_pre(void *a);
880 void vop_symlink_post(void *a, int rc);
881 int vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a);
883 #ifdef DEBUG_VFS_LOCKS
884 void vop_fplookup_vexec_debugpre(void *a);
885 void vop_fplookup_vexec_debugpost(void *a, int rc);
886 void vop_strategy_debugpre(void *a);
887 void vop_lock_debugpre(void *a);
888 void vop_lock_debugpost(void *a, int rc);
889 void vop_unlock_debugpre(void *a);
890 void vop_need_inactive_debugpre(void *a);
891 void vop_need_inactive_debugpost(void *a, int rc);
892 void vop_mkdir_debugpost(void *a, int rc);
894 #define vop_fplookup_vexec_debugpre(x) do { } while (0)
895 #define vop_fplookup_vexec_debugpost(x, y) do { } while (0)
896 #define vop_strategy_debugpre(x) do { } while (0)
897 #define vop_lock_debugpre(x) do { } while (0)
898 #define vop_lock_debugpost(x, y) do { } while (0)
899 #define vop_unlock_debugpre(x) do { } while (0)
900 #define vop_need_inactive_debugpre(x) do { } while (0)
901 #define vop_need_inactive_debugpost(x, y) do { } while (0)
902 #define vop_mkdir_debugpost(x, y) do { } while (0)
905 void vop_rename_fail(struct vop_rename_args *ap);
907 #define vop_stat_helper_pre(ap) ({ \
909 AUDIT_ARG_VNODE1(ap->a_vp); \
910 _error = mac_vnode_check_stat(ap->a_active_cred, ap->a_file_cred, ap->a_vp);\
911 if (__predict_true(_error == 0)) \
912 bzero(ap->a_sb, sizeof(*ap->a_sb)); \
916 #define vop_stat_helper_post(ap, error) ({ \
917 int _error = (error); \
918 if (priv_check_cred_vfs_generation(ap->a_td->td_ucred)) \
919 ap->a_sb->st_gen = 0; \
923 #define VOP_WRITE_PRE(ap) \
926 off_t osize, ooffset, noffset; \
928 osize = ooffset = noffset = 0; \
929 if (!VN_KNLIST_EMPTY((ap)->a_vp)) { \
930 error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred); \
933 ooffset = (ap)->a_uio->uio_offset; \
934 osize = (off_t)va.va_size; \
937 #define VOP_WRITE_POST(ap, ret) \
938 noffset = (ap)->a_uio->uio_offset; \
939 if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) { \
940 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE \
941 | (noffset > osize ? NOTE_EXTEND : 0)); \
944 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
947 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) \
951 error_ = VOP_ADD_WRITECOUNT((vp), (cnt)); \
952 VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d", \
955 #define VOP_SET_TEXT_CHECKED(vp) \
959 error_ = VOP_SET_TEXT((vp)); \
960 VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d", \
963 #define VOP_UNSET_TEXT_CHECKED(vp) \
967 error_ = VOP_UNSET_TEXT((vp)); \
968 VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d", \
972 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) VOP_ADD_WRITECOUNT((vp), (cnt))
973 #define VOP_SET_TEXT_CHECKED(vp) VOP_SET_TEXT((vp))
974 #define VOP_UNSET_TEXT_CHECKED(vp) VOP_UNSET_TEXT((vp))
977 #define VN_IS_DOOMED(vp) __predict_false((vp)->v_irflag & VIRF_DOOMED)
979 void vput(struct vnode *vp);
980 void vrele(struct vnode *vp);
981 void vref(struct vnode *vp);
982 void vrefact(struct vnode *vp);
983 void v_addpollinfo(struct vnode *vp);
985 vrefcnt(struct vnode *vp)
988 return (vp->v_usecount);
991 #define vholdl(vp) do { \
992 ASSERT_VI_LOCKED(vp, __func__); \
996 #define vrefl(vp) do { \
997 ASSERT_VI_LOCKED(vp, __func__); \
1001 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
1002 void vnode_destroy_vobject(struct vnode *vp);
1004 extern struct vop_vector fifo_specops;
1005 extern struct vop_vector dead_vnodeops;
1006 extern struct vop_vector default_vnodeops;
1008 #define VOP_PANIC ((void*)(uintptr_t)vop_panic)
1009 #define VOP_NULL ((void*)(uintptr_t)vop_null)
1010 #define VOP_EBADF ((void*)(uintptr_t)vop_ebadf)
1011 #define VOP_ENOTTY ((void*)(uintptr_t)vop_enotty)
1012 #define VOP_EINVAL ((void*)(uintptr_t)vop_einval)
1013 #define VOP_ENOENT ((void*)(uintptr_t)vop_enoent)
1014 #define VOP_EOPNOTSUPP ((void*)(uintptr_t)vop_eopnotsupp)
1015 #define VOP_EAGAIN ((void*)(uintptr_t)vop_eagain)
1018 int fifo_printinfo(struct vnode *);
1021 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
1023 void vfs_hash_changesize(u_long newhashsize);
1024 int vfs_hash_get(const struct mount *mp, u_int hash, int flags,
1025 struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1026 u_int vfs_hash_index(struct vnode *vp);
1027 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
1028 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1029 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
1030 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1031 void vfs_hash_rehash(struct vnode *vp, u_int hash);
1032 void vfs_hash_remove(struct vnode *vp);
1034 int vfs_kqfilter(struct vop_kqfilter_args *);
1036 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
1037 int vfs_emptydir(struct vnode *vp);
1039 int vfs_unixify_accmode(accmode_t *accmode);
1041 void vfs_unp_reclaim(struct vnode *vp);
1043 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
1044 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
1046 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
1048 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
1051 void vn_fsid(struct vnode *vp, struct vattr *va);
1053 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp);
1055 #define VOP_UNLOCK_FLAGS(vp, flags) ({ \
1056 struct vnode *_vp = (vp); \
1057 int _flags = (flags); \
1060 if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0) \
1061 panic("%s: unsupported flags %x\n", __func__, flags); \
1062 _error = VOP_UNLOCK(_vp); \
1063 if (_flags & LK_INTERLOCK) \
1068 #include <sys/kernel.h>
1070 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \
1071 SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \
1072 vfs_vector_op_register, &vnodeops)
1074 #define VFS_SMR_DECLARE \
1075 extern smr_t vfs_smr
1077 #define VFS_SMR() vfs_smr
1078 #define vfs_smr_enter() smr_enter(VFS_SMR())
1079 #define vfs_smr_exit() smr_exit(VFS_SMR())
1080 #define vfs_smr_entered_load(ptr) smr_entered_load((ptr), VFS_SMR())
1081 #define VFS_SMR_ASSERT_ENTERED() SMR_ASSERT_ENTERED(VFS_SMR())
1082 #define VFS_SMR_ASSERT_NOT_ENTERED() SMR_ASSERT_NOT_ENTERED(VFS_SMR())
1083 #define VFS_SMR_ZONE_SET(zone) uma_zone_set_smr((zone), VFS_SMR())
1085 #define vn_load_v_data_smr(vp) ({ \
1086 struct vnode *_vp = (vp); \
1088 VFS_SMR_ASSERT_ENTERED(); \
1089 atomic_load_ptr(&(_vp)->v_data); \
1092 #endif /* _KERNEL */
1094 #endif /* !_SYS_VNODE_H_ */