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 */
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
75__enum_uint8_decl(vstate) {
76 VSTATE_UNINITIALIZED,
77 VSTATE_CONSTRUCTED,
78 VSTATE_DESTROYING,
79 VSTATE_DEAD,
80 VLASTSTATE = VSTATE_DEAD,
81};
82
83enum 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
94struct cache_fpl;
95struct inotify_watch;
96struct namecache;
97
98struct 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
131struct 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/* XXX: These are temporary to avoid a source sweep at this time */
225#define v_object v_bufobj.bo_object
226
227/* We don't need to lock the knlist */
228#define VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL || \
229 KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
230
231#define VN_KNOTE(vp, b, a) \
232 do { \
233 if (!VN_KNLIST_EMPTY(vp)) \
234 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
235 (a) | KNF_NOKQLOCK); \
236 } while (0)
237#define VN_KNOTE_LOCKED(vp, b) VN_KNOTE(vp, b, KNF_LISTLOCKED)
238#define VN_KNOTE_UNLOCKED(vp, b) VN_KNOTE(vp, b, 0)
239
240/*
241 * Vnode flags.
242 * VI flags are protected by interlock and live in v_iflag
243 * VIRF flags are protected by interlock and live in v_irflag
244 * VV flags are protected by the vnode lock and live in v_vflag
245 *
246 * VIRF_DOOMED is doubly protected by the interlock and vnode lock. Both
247 * are required for writing but the status may be checked with either.
248 */
249#define VHOLD_NO_SMR (1<<29) /* Disable vhold_smr */
250#define VHOLD_ALL_FLAGS (VHOLD_NO_SMR)
251
252#define VIRF_DOOMED 0x0001 /* This vnode is being recycled */
253#define VIRF_PGREAD 0x0002 /* Direct reads from the page cache are permitted,
254 never cleared once set */
255#define VIRF_MOUNTPOINT 0x0004 /* This vnode is mounted on */
256#define VIRF_TEXT_REF 0x0008 /* Executable mappings ref the vnode */
257#define VIRF_CROSSMP 0x0010 /* Cross-mp vnode, no locking */
258#define VIRF_NAMEDDIR 0x0020 /* Named attribute directory */
259#define VIRF_NAMEDATTR 0x0040 /* Named attribute */
260#define VIRF_INOTIFY 0x0080 /* This vnode is being watched */
261#define VIRF_INOTIFY_PARENT 0x0100 /* A parent of this vnode may be being
262 watched */
263
264#define VI_UNUSED0 0x0001 /* unused */
265#define VI_MOUNT 0x0002 /* Mount in progress */
266#define VI_DOINGINACT 0x0004 /* VOP_INACTIVE is in progress */
267#define VI_OWEINACT 0x0008 /* Need to call inactive */
268#define VI_DEFINACT 0x0010 /* deferred inactive */
269#define VI_FOPENING 0x0020 /* In open, with opening process having the
270 first right to advlock file */
271
272#define VV_ROOT 0x0001 /* root of its filesystem */
273#define VV_ISTTY 0x0002 /* vnode represents a tty */
274#define VV_NOSYNC 0x0004 /* unlinked, stop syncing */
275#define VV_ETERNALDEV 0x0008 /* device that is never destroyed */
276#define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */
277#define VV_VMSIZEVNLOCK 0x0020 /* object size check requires vnode lock */
278#define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */
279#define VV_SYSTEM 0x0080 /* vnode being used by kernel */
280#define VV_PROCDEP 0x0100 /* vnode is process dependent */
281#define VV_UNLINKED 0x0200 /* unlinked but stil open directory */
282#define VV_DELETED 0x0400 /* should be removed */
283#define VV_MD 0x0800 /* vnode backs the md device */
284#define VV_FORCEINSMQ 0x1000 /* force the insmntque to succeed */
285#define VV_READLINK 0x2000 /* fdescfs linux vnode */
286#define VV_UNREF 0x4000 /* vunref, do not drop lock in inactive() */
287#define VV_CROSSLOCK 0x8000 /* vnode lock is shared w/ root mounted here */
288
289#define VMP_LAZYLIST 0x0001 /* Vnode is on mnt's lazy list */
290
291/*
292 * Vnode attributes. A field value of VNOVAL represents a field whose value
293 * is unavailable (getattr) or which is not to be changed (setattr).
294 */
295struct vattr {
296 __enum_uint8(vtype) va_type; /* vnode type (for create) */
297 u_short va_mode; /* files access mode and type */
298 uint16_t va_bsdflags; /* same as st_bsdflags from stat(2) */
299 uid_t va_uid; /* owner user id */
300 gid_t va_gid; /* owner group id */
301 nlink_t va_nlink; /* number of references to file */
302 dev_t va_fsid; /* filesystem id */
303 ino_t va_fileid; /* file id */
304 u_quad_t va_size; /* file size in bytes */
305 long va_blocksize; /* blocksize preferred for i/o */
306 struct timespec va_atime; /* time of last access */
307 struct timespec va_mtime; /* time of last modification */
308 struct timespec va_ctime; /* time file changed */
309 struct timespec va_birthtime; /* time file created */
310 u_long va_gen; /* generation number of file */
311 u_long va_flags; /* flags defined for file */
312 dev_t va_rdev; /* device the special file represents */
313 u_quad_t va_bytes; /* bytes of disk space held by file */
314 u_quad_t va_filerev; /* file modification number */
315 u_int va_vaflags; /* operations flags, see below */
316 long va_spare; /* remain quad aligned */
317};
318
319#define VATTR_ISDEV(vap) VTYPE_ISDEV((vap)->va_type)
320
321/*
322 * Flags for va_vaflags.
323 */
324#define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */
325#define VA_EXCLUSIVE 0x02 /* exclusive create request */
326#define VA_SYNC 0x04 /* O_SYNC truncation */
327
328/*
329 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
330 * help with write clustering)
331 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
332 */
333#define IO_UNIT 0x0001 /* do I/O as atomic unit */
334#define IO_APPEND 0x0002 /* append write to end */
335#define IO_NDELAY 0x0004 /* FNDELAY flag set in file table */
336#define IO_NODELOCKED 0x0008 /* underlying node already locked */
337#define IO_ASYNC 0x0010 /* bawrite rather then bdwrite */
338#define IO_VMIO 0x0020 /* data already in VMIO space */
339#define IO_INVAL 0x0040 /* invalidate after I/O */
340#define IO_SYNC 0x0080 /* do I/O synchronously */
341#define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */
342#define IO_NOREUSE 0x0200 /* VMIO data won't be reused */
343#define IO_EXT 0x0400 /* operate on external attributes */
344#define IO_NORMAL 0x0800 /* operate on regular data */
345#define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */
346#define IO_BUFLOCKED 0x2000 /* ffs flag; indir buf is locked */
347#define IO_RANGELOCKED 0x4000 /* range locked */
348#define IO_DATASYNC 0x8000 /* do only data I/O synchronously */
349
350#define IO_SEQMAX 0x7F /* seq heuristic max value */
351#define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */
352
353/*
354 * Flags for accmode_t.
355 */
356#define VEXEC 000000000100 /* execute/search permission */
357#define VWRITE 000000000200 /* write permission */
358#define VREAD 000000000400 /* read permission */
359#define VADMIN 000000010000 /* being the file owner */
360#define VAPPEND 000000040000 /* permission to write/append */
361/*
362 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
363 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
364 * and 0 otherwise. This never happens with ordinary unix access rights
365 * or POSIX.1e ACLs. Obviously, VEXPLICIT_DENY must be OR-ed with
366 * some other V* constant.
367 */
368#define VEXPLICIT_DENY 000000100000
369#define VREAD_NAMED_ATTRS 000000200000 /* not used */
370#define VWRITE_NAMED_ATTRS 000000400000 /* not used */
371#define VDELETE_CHILD 000001000000
372#define VREAD_ATTRIBUTES 000002000000 /* permission to stat(2) */
373#define VWRITE_ATTRIBUTES 000004000000 /* change {m,c,a}time */
374#define VDELETE 000010000000
375#define VREAD_ACL 000020000000 /* read ACL and file mode */
376#define VWRITE_ACL 000040000000 /* change ACL and/or file mode */
377#define VWRITE_OWNER 000100000000 /* change file owner */
378#define VSYNCHRONIZE 000200000000 /* not used */
379#define VCREAT 000400000000 /* creating new file */
380#define VVERIFY 001000000000 /* verification required */
381
382/*
383 * Permissions that were traditionally granted only to the file owner.
384 */
385#define VADMIN_PERMS (VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
386 VWRITE_OWNER)
387
388/*
389 * Permissions that were traditionally granted to everyone.
390 */
391#define VSTAT_PERMS (VREAD_ATTRIBUTES | VREAD_ACL)
392
393/*
394 * Permissions that allow to change the state of the file in any way.
395 */
396#define VMODIFY_PERMS (VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
397 VDELETE)
398
399/*
400 * Token indicating no attribute value yet assigned.
401 */
402#define VNOVAL (-1)
403
404/*
405 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
406 */
407#define VLKTIMEOUT (hz / 20 + 1)
408
409#ifdef _KERNEL
410
411#ifdef MALLOC_DECLARE
412MALLOC_DECLARE(M_VNODE);
413#endif
414
415extern u_int ncsizefactor;
416extern const u_int io_hold_cnt;
417
418/*
419 * Convert between vnode types and inode formats (since POSIX.1
420 * defines mode word of stat structure in terms of inode formats).
421 */
422extern __enum_uint8(vtype) iftovt_tab[];
423extern int vttoif_tab[];
424#define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12])
425#define VTTOIF(indx) (vttoif_tab[(int)(indx)])
426#define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode))
427
428/*
429 * Flags to various vnode functions.
430 */
431#define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */
432#define FORCECLOSE 0x0002 /* vflush: force file closure */
433#define WRITECLOSE 0x0004 /* vflush: only close writable files */
434#define EARLYFLUSH 0x0008 /* vflush: early call for ffs_flushfiles */
435#define V_SAVE 0x0001 /* vinvalbuf: sync file first */
436#define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */
437#define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */
438#define V_CLEANONLY 0x0008 /* vinvalbuf: invalidate only clean bufs */
439#define V_VMIO 0x0010 /* vinvalbuf: called during pageout */
440#define V_ALLOWCLEAN 0x0020 /* vinvalbuf: allow clean buffers after flush */
441#define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */
442#define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */
443#define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */
444#define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */
445#define V_PCATCH 0x0008 /* vn_start_write: make the sleep interruptible */
446#define V_VALID_FLAGS (V_WAIT | V_NOWAIT | V_XSLEEP | V_PCATCH)
447
448#define VR_START_WRITE 0x0001 /* vfs_write_resume: start write atomically */
449#define VR_NO_SUSPCLR 0x0002 /* vfs_write_resume: do not clear suspension */
450
451#define VS_SKIP_UNMOUNT 0x0001 /* vfs_write_suspend: fail if the
452 filesystem is being unmounted */
453
454#define VREF(vp) vref(vp)
455
456#ifdef DIAGNOSTIC
457#define VATTR_NULL(vap) vattr_null(vap)
458#else
459#define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */
460#endif /* DIAGNOSTIC */
461
462#define NULLVP ((struct vnode *)NULL)
463
464/*
465 * Global vnode data.
466 */
467extern struct vnode *rootvnode; /* root (i.e. "/") vnode */
468extern struct mount *rootdevmp; /* "/dev" mount */
469extern u_long desiredvnodes; /* number of vnodes desired */
470extern struct uma_zone *namei_zone;
471extern struct vattr va_null; /* predefined null vattr structure */
472
473extern u_int vn_lock_pair_pause_max;
474
475#define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock)
476#define VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
477#define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock)
478#define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock)
479#define VI_MTX(vp) (&(vp)->v_interlock)
480
481#define VN_LOCK_AREC(vp) lockallowrecurse((vp)->v_vnlock)
482#define VN_LOCK_ASHARE(vp) lockallowshare((vp)->v_vnlock)
483#define VN_LOCK_DSHARE(vp) lockdisableshare((vp)->v_vnlock)
484
485#endif /* _KERNEL */
486
487/*
488 * Mods for extensibility.
489 */
490
491/*
492 * Flags for vdesc_flags:
493 */
494#define VDESC_MAX_VPS 16
495/* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
496#define VDESC_VP0_WILLRELE 0x0001
497#define VDESC_VP1_WILLRELE 0x0002
498#define VDESC_VP2_WILLRELE 0x0004
499#define VDESC_VP3_WILLRELE 0x0008
500
501/*
502 * A generic structure.
503 * This can be used by bypass routines to identify generic arguments.
504 */
505struct vop_generic_args {
506 struct vnodeop_desc *a_desc;
507 /* other random data follows, presumably */
508};
509
510typedef int vop_bypass_t(struct vop_generic_args *);
511
512/*
513 * VDESC_NO_OFFSET is used to identify the end of the offset list
514 * and in places where no such field exists.
515 */
516#define VDESC_NO_OFFSET -1
517
518/*
519 * This structure describes the vnode operation taking place.
520 */
521struct vnodeop_desc {
522 char *vdesc_name; /* a readable name for debugging */
523 int vdesc_flags; /* VDESC_* flags */
524 int vdesc_vop_offset;
525 vop_bypass_t *vdesc_call; /* Function to call */
526
527 /*
528 * These ops are used by bypass routines to map and locate arguments.
529 * Creds and procs are not needed in bypass routines, but sometimes
530 * they are useful to (for example) transport layers.
531 * Nameidata is useful because it has a cred in it.
532 */
533 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */
534 int vdesc_vpp_offset; /* return vpp location */
535 int vdesc_cred_offset; /* cred location, if any */
536 int vdesc_thread_offset; /* thread location, if any */
537 int vdesc_componentname_offset; /* if any */
538};
539
540#ifdef _KERNEL
541/*
542 * A list of all the operation descs.
543 */
544extern struct vnodeop_desc *vnodeop_descs[];
545
546#define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field)
547#define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
548 ((s_type)(((char*)(struct_p)) + (s_offset)))
549
550#ifdef INVARIANTS
551/*
552 * Support code to aid in debugging VFS locking problems. Not totally
553 * reliable since if the thread sleeps between changing the lock
554 * state and checking it with the assert, some other thread could
555 * change the state. They are good enough for debugging a single
556 * filesystem using a single-threaded test. Note that the unreliability is
557 * limited to false negatives; efforts were made to ensure that false
558 * positives cannot occur.
559 */
560void assert_vi_locked(struct vnode *vp, const char *str);
561void assert_vi_unlocked(struct vnode *vp, const char *str);
562void assert_vop_elocked(struct vnode *vp, const char *str);
563void assert_vop_locked(struct vnode *vp, const char *str);
564void assert_vop_unlocked(struct vnode *vp, const char *str);
565
566#define ASSERT_VI_LOCKED(vp, str) assert_vi_locked((vp), (str))
567#define ASSERT_VI_UNLOCKED(vp, str) assert_vi_unlocked((vp), (str))
568#define ASSERT_VOP_ELOCKED(vp, str) assert_vop_elocked((vp), (str))
569#define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked((vp), (str))
570#define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked((vp), (str))
571
572#define ASSERT_VOP_IN_SEQC(vp) do { \
573 struct vnode *_vp = (vp); \
574 \
575 VNPASS(seqc_in_modify(_vp->v_seqc), _vp); \
576} while (0)
577
578#define ASSERT_VOP_NOT_IN_SEQC(vp) do { \
579 struct vnode *_vp = (vp); \
580 \
581 VNPASS(!seqc_in_modify(_vp->v_seqc), _vp); \
582} while (0)
583
584#else /* !INVARIANTS */
585
586#define ASSERT_VI_LOCKED(vp, str) ((void)0)
587#define ASSERT_VI_UNLOCKED(vp, str) ((void)0)
588#define ASSERT_VOP_ELOCKED(vp, str) ((void)0)
589#define ASSERT_VOP_LOCKED(vp, str) ((void)0)
590#define ASSERT_VOP_UNLOCKED(vp, str) ((void)0)
591
592#define ASSERT_VOP_IN_SEQC(vp) ((void)0)
593#define ASSERT_VOP_NOT_IN_SEQC(vp) ((void)0)
594
595#endif /* INVARIANTS */
596
597/*
598 * This call works for vnodes in the kernel.
599 */
600#define VCALL(c) ((c)->a_desc->vdesc_call(c))
601
602#define DOINGASYNC(vp) \
603 (((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 && \
604 ((curthread->td_pflags & TDP_SYNCIO) == 0))
605
606/*
607 * VMIO support inline
608 */
609
610extern int vmiodirenable;
611
612static __inline int
613vn_canvmio(struct vnode *vp)
614{
615 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
616 return(TRUE);
617 return(FALSE);
618}
619
620/*
621 * Finally, include the default set of vnode operations.
622 */
623typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
624#include "vnode_if.h"
625
626/* vn_open_flags */
627#define VN_OPEN_NOAUDIT 0x00000001
628#define VN_OPEN_NOCAPCHECK 0x00000002
629#define VN_OPEN_NAMECACHE 0x00000004
630#define VN_OPEN_INVFS 0x00000008
631#define VN_OPEN_WANTIOCTLCAPS 0x00000010
632
633/* copy_file_range kernel flags */
634#define COPY_FILE_RANGE_KFLAGS 0xff000000
635#define COPY_FILE_RANGE_TIMEO1SEC 0x01000000 /* Return after 1sec. */
636
637/*
638 * Public vnode manipulation functions.
639 */
640struct componentname;
641struct file;
642struct mount;
643struct nameidata;
644struct ostat;
645struct freebsd11_stat;
646struct thread;
647struct proc;
648struct stat;
649struct nstat;
650struct ucred;
651struct uio;
652struct vattr;
653struct vfsops;
654struct vnode;
655
656typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
657
658int bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
659 daddr_t endn);
660/* cache_* may belong in namei.h. */
661void cache_changesize(u_long newhashsize);
662
663#define VFS_CACHE_DROPOLD 0x1
664
665void cache_enter_time_flags(struct vnode *dvp, struct vnode *vp,
666 struct componentname *cnp, struct timespec *tsp,
667 struct timespec *dtsp, int flags);
668#define cache_enter(dvp, vp, cnp) \
669 cache_enter_time(dvp, vp, cnp, NULL, NULL)
670void cache_enter_time(struct vnode *dvp, struct vnode *vp,
671 struct componentname *cnp, struct timespec *tsp,
672 struct timespec *dtsp);
673int cache_lookup(struct vnode *dvp, struct vnode **vpp,
674 struct componentname *cnp, struct timespec *tsp, int *ticksp);
675void cache_vnode_init(struct vnode *vp);
676void cache_purge(struct vnode *vp);
677void cache_purge_vgone(struct vnode *vp);
678void cache_purge_negative(struct vnode *vp);
679void cache_purgevfs(struct mount *mp);
680char *cache_symlink_alloc(size_t size, int flags);
681void cache_symlink_free(char *string, size_t size);
682int cache_symlink_resolve(struct cache_fpl *fpl, const char *string,
683 size_t len);
684void cache_vop_inotify(struct vnode *vp, int event, uint32_t cookie);
685void cache_vop_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp,
686 struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp);
687void cache_vop_rmdir(struct vnode *dvp, struct vnode *vp);
688void cache_vop_vector_register(struct vop_vector *);
689#ifdef INVARIANTS
690void cache_validate(struct vnode *dvp, struct vnode *vp,
691 struct componentname *cnp);
692void cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops);
693void cache_assert_no_entries(struct vnode *vp);
694#else
695static inline void
696cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
697{
698}
699
700static inline void
701cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops)
702{
703}
704
705static inline void
706cache_assert_no_entries(struct vnode *vp)
707{
708}
709#endif
710void cache_fast_lookup_enabled_recalc(void);
711int change_dir(struct vnode *vp, struct thread *td);
712void cvtstat(struct stat *st, struct ostat *ost);
713int freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb);
714int freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost);
715int getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
716 struct vnode **vpp);
717void getnewvnode_reserve(void);
718void getnewvnode_drop_reserve(void);
719int insmntque(struct vnode *vp, struct mount *mp);
720int insmntque1(struct vnode *vp, struct mount *mp);
721u_quad_t init_va_filerev(void);
722int speedup_syncer(void);
723int vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen);
724int vn_getcwd(char *buf, char **retbuf, size_t *buflen);
725int vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf);
726int vn_fullpath_jail(struct vnode *vp, char **retbuf, char **freebuf);
727int vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf);
728int vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp,
729 const char *hdrl_name, size_t hrdl_name_length, char **retbuf,
730 char **freebuf, size_t *buflen);
731struct vnode *
732 vn_dir_dd_ino(struct vnode *vp);
733int vn_commname(struct vnode *vn, char *buf, u_int buflen);
734int vn_path_to_global_path(struct thread *td, struct vnode *vp,
735 char *path, u_int pathlen);
736int vn_path_to_global_path_hardlink(struct thread *td, struct vnode *vp,
737 struct vnode *dvp, char *path, u_int pathlen, const char *leaf_name,
738 size_t leaf_length);
739int vaccess(__enum_uint8(vtype) type, mode_t file_mode, uid_t file_uid,
740 gid_t file_gid, accmode_t accmode, struct ucred *cred);
741int vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid,
742 struct ucred *cred);
743int vaccess_acl_nfs4(__enum_uint8(vtype) type, uid_t file_uid, gid_t file_gid,
744 struct acl *aclp, accmode_t accmode, struct ucred *cred);
745int vaccess_acl_posix1e(__enum_uint8(vtype) type, uid_t file_uid,
746 gid_t file_gid, struct acl *acl, accmode_t accmode,
747 struct ucred *cred);
748void vattr_null(struct vattr *vap);
749void vlazy(struct vnode *);
750void vdrop(struct vnode *);
751void vdropl(struct vnode *);
752int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
753int vget(struct vnode *vp, int flags);
754enum vgetstate vget_prep_smr(struct vnode *vp);
755enum vgetstate vget_prep(struct vnode *vp);
756int vget_finish(struct vnode *vp, int flags, enum vgetstate vs);
757void vget_finish_ref(struct vnode *vp, enum vgetstate vs);
758void vget_abort(struct vnode *vp, enum vgetstate vs);
759void vgone(struct vnode *vp);
760void vhold(struct vnode *);
761void vholdnz(struct vnode *);
762bool vhold_smr(struct vnode *);
763int vinactive(struct vnode *vp);
764int vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
765int vtruncbuf(struct vnode *vp, off_t length, int blksize);
766void v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn,
767 int blksize);
768void vunref(struct vnode *);
769void vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
770int vrecycle(struct vnode *vp);
771int vrecyclel(struct vnode *vp);
772int vn_bmap_seekhole_locked(struct vnode *vp, u_long cmd, off_t *off,
773 struct ucred *cred);
774int vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
775 struct ucred *cred);
776int vn_close(struct vnode *vp,
777 int flags, struct ucred *file_cred, struct thread *td);
778int vn_copy_file_range(struct vnode *invp, off_t *inoffp,
779 struct vnode *outvp, off_t *outoffp, size_t *lenp,
780 unsigned int flags, struct ucred *incred, struct ucred *outcred,
781 struct thread *fsize_td);
782int vn_deallocate(struct vnode *vp, off_t *offset, off_t *length, int flags,
783 int ioflg, struct ucred *active_cred, struct ucred *file_cred);
784void vn_finished_write(struct mount *mp);
785void vn_finished_secondary_write(struct mount *mp);
786int vn_fsync_buf(struct vnode *vp, int waitfor);
787int vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp,
788 struct vnode *outvp, off_t *outoffp, size_t *lenp,
789 unsigned int flags, struct ucred *incred, struct ucred *outcred,
790 struct thread *fsize_td);
791int vn_need_pageq_flush(struct vnode *vp);
792bool vn_isdisk_error(struct vnode *vp, int *errp);
793bool vn_isdisk(struct vnode *vp);
794int _vn_lock(struct vnode *vp, int flags, const char *file, int line);
795#define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
796void vn_lock_pair(struct vnode *vp1, bool vp1_locked, int lkflags1,
797 struct vnode *vp2, bool vp2_locked, int lkflags2);
798int vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
799int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
800 u_int vn_open_flags, struct ucred *cred, struct file *fp);
801int vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
802 struct thread *td, struct file *fp);
803void vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
804void vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start,
805 vm_pindex_t end);
806int vn_pollrecord(struct vnode *vp, struct thread *p, int events);
807int vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
808 int len, off_t offset, enum uio_seg segflg, int ioflg,
809 struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
810 struct thread *td);
811int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
812 size_t len, off_t offset, enum uio_seg segflg, int ioflg,
813 struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
814 struct thread *td);
815int vn_read_from_obj(struct vnode *vp, struct uio *uio);
816int vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio,
817 struct thread *td);
818int vn_rlimit_fsizex(const struct vnode *vp, struct uio *uio,
819 off_t maxfsz, ssize_t *resid_adj, struct thread *td);
820void vn_rlimit_fsizex_res(struct uio *uio, ssize_t resid_adj);
821int vn_rlimit_trunc(u_quad_t size, struct thread *td);
822int vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
823int vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
824 int flags);
825int vn_truncate_locked(struct vnode *vp, off_t length, bool sync,
826 struct ucred *cred);
827int vn_writechk(struct vnode *vp);
828int vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
829 const char *attrname, int *buflen, char *buf, struct thread *td);
830int vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
831 const char *attrname, int buflen, char *buf, struct thread *td);
832int vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
833 const char *attrname, struct thread *td);
834int vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
835 struct vnode **rvp);
836int vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
837 void *alloc_arg, int lkflags, struct vnode **rvp);
838int vn_utimes_perm(struct vnode *vp, struct vattr *vap,
839 struct ucred *cred, struct thread *td);
840int vn_cmp(struct file *, struct file *, struct thread *td);
841
842int vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
843int vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
844 struct uio *uio);
845
846void vn_seqc_write_begin_locked(struct vnode *vp);
847void vn_seqc_write_begin(struct vnode *vp);
848void vn_seqc_write_end_locked(struct vnode *vp);
849void vn_seqc_write_end(struct vnode *vp);
850#define vn_seqc_read_any(vp) seqc_read_any(&(vp)->v_seqc)
851#define vn_seqc_read_notmodify(vp) seqc_read_notmodify(&(vp)->v_seqc)
852#define vn_seqc_consistent(vp, seq) seqc_consistent(&(vp)->v_seqc, seq)
853
854#define vn_rangelock_unlock(vp, cookie) \
855 rangelock_unlock(&(vp)->v_rl, (cookie))
856#define vn_rangelock_rlock(vp, start, end) \
857 rangelock_rlock(&(vp)->v_rl, (start), (end))
858#define vn_rangelock_tryrlock(vp, start, end) \
859 rangelock_tryrlock(&(vp)->v_rl, (start), (end))
860#define vn_rangelock_wlock(vp, start, end) \
861 rangelock_wlock(&(vp)->v_rl, (start), (end))
862#define vn_rangelock_trywlock(vp, start, end) \
863 rangelock_trywlock(&(vp)->v_rl, (start), (end))
864
865#define vn_irflag_read(vp) atomic_load_short(&(vp)->v_irflag)
866void vn_irflag_set_locked(struct vnode *vp, short toset);
867void vn_irflag_set(struct vnode *vp, short toset);
868void vn_irflag_set_cond_locked(struct vnode *vp, short toset);
869void vn_irflag_set_cond(struct vnode *vp, short toset);
870void vn_irflag_unset_locked(struct vnode *vp, short tounset);
871void vn_irflag_unset(struct vnode *vp, short tounset);
872
873int vfs_cache_lookup(struct vop_lookup_args *ap);
874int vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp);
875void vfs_timestamp(struct timespec *);
876void vfs_write_resume(struct mount *mp, int flags);
877int vfs_write_suspend(struct mount *mp, int flags);
878int vfs_write_suspend_umnt(struct mount *mp);
879struct vnode *vnlru_alloc_marker(void);
880void vnlru_free_marker(struct vnode *);
881void vnlru_free_vfsops(int, struct vfsops *, struct vnode *);
882int vop_stdbmap(struct vop_bmap_args *);
883int vop_stdfdatasync_buf(struct vop_fdatasync_args *);
884int vop_stdfsync(struct vop_fsync_args *);
885int vop_stdgetwritemount(struct vop_getwritemount_args *);
886int vop_stdgetpages(struct vop_getpages_args *);
887int vop_stdinactive(struct vop_inactive_args *);
888int vop_stdneed_inactive(struct vop_need_inactive_args *);
889int vop_stdinotify(struct vop_inotify_args *);
890int vop_stdinotify_add_watch(struct vop_inotify_add_watch_args *);
891int vop_stdioctl(struct vop_ioctl_args *);
892int vop_stdkqfilter(struct vop_kqfilter_args *);
893int vop_stdlock(struct vop_lock1_args *);
894int vop_stdunlock(struct vop_unlock_args *);
895int vop_stdislocked(struct vop_islocked_args *);
896int vop_lock(struct vop_lock1_args *);
897int vop_unlock(struct vop_unlock_args *);
898int vop_islocked(struct vop_islocked_args *);
899int vop_stdputpages(struct vop_putpages_args *);
900int vop_nopoll(struct vop_poll_args *);
901int vop_stdaccess(struct vop_access_args *ap);
902int vop_stdaccessx(struct vop_accessx_args *ap);
903int vop_stdadvise(struct vop_advise_args *ap);
904int vop_stdadvlock(struct vop_advlock_args *ap);
905int vop_stdadvlockasync(struct vop_advlockasync_args *ap);
906int vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
907int vop_stdallocate(struct vop_allocate_args *ap);
908int vop_stddeallocate(struct vop_deallocate_args *ap);
909int vop_stdset_text(struct vop_set_text_args *ap);
910int vop_stdpathconf(struct vop_pathconf_args *);
911int vop_stdpoll(struct vop_poll_args *);
912int vop_stdvptocnp(struct vop_vptocnp_args *ap);
913int vop_stdvptofh(struct vop_vptofh_args *ap);
914int vop_stdunp_bind(struct vop_unp_bind_args *ap);
915int vop_stdunp_connect(struct vop_unp_connect_args *ap);
916int vop_stdunp_detach(struct vop_unp_detach_args *ap);
917int vop_stdadd_writecount_nomsync(struct vop_add_writecount_args *ap);
918int vop_eopnotsupp(struct vop_generic_args *ap);
919int vop_ebadf(struct vop_generic_args *ap);
920int vop_einval(struct vop_generic_args *ap);
921int vop_enoent(struct vop_generic_args *ap);
922int vop_enotty(struct vop_generic_args *ap);
923int vop_eagain(struct vop_generic_args *ap);
924int vop_null(struct vop_generic_args *ap);
925int vop_panic(struct vop_generic_args *ap);
926int dead_poll(struct vop_poll_args *ap);
927int dead_read(struct vop_read_args *ap);
928int dead_write(struct vop_write_args *ap);
929
930/* These are called from within the actual VOPS. */
931void vop_allocate_post(void *a, int rc);
932void vop_copy_file_range_post(void *ap, int rc);
933void vop_close_post(void *a, int rc);
934void vop_create_pre(void *a);
935void vop_create_post(void *a, int rc);
936void vop_deallocate_post(void *a, int rc);
937void vop_whiteout_pre(void *a);
938void vop_whiteout_post(void *a, int rc);
939void vop_deleteextattr_pre(void *a);
940void vop_deleteextattr_post(void *a, int rc);
941void vop_link_pre(void *a);
942void vop_link_post(void *a, int rc);
943void vop_lookup_post(void *a, int rc);
944void vop_lookup_pre(void *a);
945void vop_mkdir_pre(void *a);
946void vop_mkdir_post(void *a, int rc);
947void vop_mknod_pre(void *a);
948void vop_mknod_post(void *a, int rc);
949void vop_open_post(void *a, int rc);
950void vop_read_post(void *a, int rc);
951void vop_read_pgcache_post(void *ap, int rc);
952void vop_reclaim_post(void *a, int rc);
953void vop_remove_pre(void *a);
954void vop_remove_post(void *a, int rc);
955void vop_rename_post(void *a, int rc);
956void vop_rename_pre(void *a);
957void vop_rmdir_pre(void *a);
958void vop_rmdir_post(void *a, int rc);
959void vop_setattr_pre(void *a);
960void vop_setattr_post(void *a, int rc);
961void vop_setacl_pre(void *a);
962void vop_setacl_post(void *a, int rc);
963void vop_setextattr_pre(void *a);
964void vop_setextattr_post(void *a, int rc);
965void vop_symlink_pre(void *a);
966void vop_symlink_post(void *a, int rc);
967int vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a);
968
969#ifdef INVARIANTS
970void vop_fdatasync_debugpre(void *a);
971void vop_fdatasync_debugpost(void *a, int rc);
972void vop_fplookup_vexec_debugpre(void *a);
973void vop_fplookup_vexec_debugpost(void *a, int rc);
974void vop_fplookup_symlink_debugpre(void *a);
975void vop_fplookup_symlink_debugpost(void *a, int rc);
976void vop_fsync_debugpre(void *a);
977void vop_fsync_debugpost(void *a, int rc);
978void vop_strategy_debugpre(void *a);
979void vop_lock_debugpre(void *a);
980void vop_lock_debugpost(void *a, int rc);
981void vop_unlock_debugpre(void *a);
982void vop_need_inactive_debugpre(void *a);
983void vop_need_inactive_debugpost(void *a, int rc);
984void vop_mkdir_debugpost(void *a, int rc);
985#else
986#define vop_fdatasync_debugpre(x) do { } while (0)
987#define vop_fdatasync_debugpost(x, y) do { } while (0)
988#define vop_fplookup_vexec_debugpre(x) do { } while (0)
989#define vop_fplookup_vexec_debugpost(x, y) do { } while (0)
990#define vop_fplookup_symlink_debugpre(x) do { } while (0)
991#define vop_fplookup_symlink_debugpost(x, y) do { } while (0)
992#define vop_fsync_debugpre(x) do { } while (0)
993#define vop_fsync_debugpost(x, y) do { } while (0)
994#define vop_strategy_debugpre(x) do { } while (0)
995#define vop_lock_debugpre(x) do { } while (0)
996#define vop_lock_debugpost(x, y) do { } while (0)
997#define vop_unlock_debugpre(x) do { } while (0)
998#define vop_need_inactive_debugpre(x) do { } while (0)
999#define vop_need_inactive_debugpost(x, y) do { } while (0)
1000#define vop_mkdir_debugpost(x, y) do { } while (0)
1001#endif
1002
1003void vop_rename_fail(struct vop_rename_args *ap);
1004
1005#define vop_stat_helper_pre(ap) ({ \
1006 struct vop_stat_args *_ap = (ap); \
1007 int _error; \
1008 AUDIT_ARG_VNODE1(ap->a_vp); \
1009 _error = mac_vnode_check_stat(_ap->a_active_cred, _ap->a_file_cred, _ap->a_vp);\
1010 if (__predict_true(_error == 0)) { \
1011 ap->a_sb->st_padding1 = 0; \
1012 bzero(_ap->a_sb->st_spare, sizeof(_ap->a_sb->st_spare)); \
1013 ap->a_sb->st_filerev = 0; \
1014 ap->a_sb->st_bsdflags = 0; \
1015 } \
1016 _error; \
1017})
1018
1019#define vop_stat_helper_post(ap, error) ({ \
1020 struct vop_stat_args *_ap = (ap); \
1021 int _error = (error); \
1022 if (priv_check_cred_vfs_generation(_ap->a_active_cred)) \
1023 _ap->a_sb->st_gen = 0; \
1024 _error; \
1025})
1026
1027#ifdef INVARIANTS
1028#define vop_readdir_pre_assert(ap) \
1029 ssize_t nresid, oresid; \
1030 \
1031 oresid = (ap)->a_uio->uio_resid;
1032
1033#define vop_readdir_post_assert(ap, ret) \
1034 nresid = (ap)->a_uio->uio_resid; \
1035 if ((ret) == 0 && (ap)->a_eofflag != NULL) { \
1036 VNASSERT(oresid == 0 || nresid != oresid || \
1037 *(ap)->a_eofflag == 1, \
1038 (ap)->a_vp, ("VOP_READDIR: eofflag not set")); \
1039 }
1040#else
1041#define vop_readdir_pre_assert(ap)
1042#define vop_readdir_post_assert(ap, ret)
1043#endif
1044
1045#define vop_readdir_pre(ap) do { \
1046 vop_readdir_pre_assert(ap)
1047
1048#define vop_readdir_post(ap, ret) \
1049 vop_readdir_post_assert(ap, ret); \
1050 if ((ret) == 0) { \
1051 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_READ); \
1052 INOTIFY((ap)->a_vp, IN_ACCESS); \
1053 } \
1054} while (0)
1055
1056#define vop_write_pre(ap) \
1057 struct vattr va; \
1058 int error; \
1059 off_t osize, ooffset, noffset; \
1060 \
1061 osize = ooffset = noffset = 0; \
1062 if (!VN_KNLIST_EMPTY((ap)->a_vp)) { \
1063 error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred); \
1064 if (error) \
1065 return (error); \
1066 ooffset = (ap)->a_uio->uio_offset; \
1067 osize = (off_t)va.va_size; \
1068 }
1069
1070#define vop_write_post(ap, ret) \
1071 noffset = (ap)->a_uio->uio_offset; \
1072 if (noffset > ooffset) { \
1073 if (!VN_KNLIST_EMPTY((ap)->a_vp)) { \
1074 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE | \
1075 (noffset > osize ? NOTE_EXTEND : 0)); \
1076 } \
1077 INOTIFY((ap)->a_vp, IN_MODIFY); \
1078 }
1079
1080#define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
1081
1082#ifdef INVARIANTS
1083#define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) \
1084do { \
1085 int error_; \
1086 \
1087 error_ = VOP_ADD_WRITECOUNT((vp), (cnt)); \
1088 VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d", \
1089 error_)); \
1090} while (0)
1091#define VOP_SET_TEXT_CHECKED(vp) \
1092do { \
1093 int error_; \
1094 \
1095 error_ = VOP_SET_TEXT((vp)); \
1096 VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d", \
1097 error_)); \
1098} while (0)
1099#define VOP_UNSET_TEXT_CHECKED(vp) \
1100do { \
1101 int error_; \
1102 \
1103 error_ = VOP_UNSET_TEXT((vp)); \
1104 VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d", \
1105 error_)); \
1106} while (0)
1107#else
1108#define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) VOP_ADD_WRITECOUNT((vp), (cnt))
1109#define VOP_SET_TEXT_CHECKED(vp) VOP_SET_TEXT((vp))
1110#define VOP_UNSET_TEXT_CHECKED(vp) VOP_UNSET_TEXT((vp))
1111#endif
1112
1113#define VN_IS_DOOMED(vp) __predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0)
1114
1115void vput(struct vnode *vp);
1116void vrele(struct vnode *vp);
1117void vref(struct vnode *vp);
1118void vrefact(struct vnode *vp);
1119void v_addpollinfo(struct vnode *vp);
1120static __inline int
1121vrefcnt(struct vnode *vp)
1122{
1123
1124 return (vp->v_usecount);
1125}
1126
1127#define vholdl(vp) do { \
1128 ASSERT_VI_LOCKED(vp, __func__); \
1129 vhold(vp); \
1130} while (0)
1131
1132#define vrefl(vp) do { \
1133 ASSERT_VI_LOCKED(vp, __func__); \
1134 vref(vp); \
1135} while (0)
1136
1137/*
1138 * The caller doesn't know the file size and vnode_create_vobject() should
1139 * determine the size on its own.
1140 */
1141#define VNODE_NO_SIZE ((off_t)-1)
1142
1143int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
1144int vnode_create_disk_vobject(struct vnode *vp, off_t size, struct thread *td);
1145void vnode_destroy_vobject(struct vnode *vp);
1146
1147extern struct vop_vector fifo_specops;
1148extern struct vop_vector dead_vnodeops;
1149extern struct vop_vector default_vnodeops;
1150
1151#define VOP_PANIC ((void*)(uintptr_t)vop_panic)
1152#define VOP_NULL ((void*)(uintptr_t)vop_null)
1153#define VOP_EBADF ((void*)(uintptr_t)vop_ebadf)
1154#define VOP_ENOTTY ((void*)(uintptr_t)vop_enotty)
1155#define VOP_EINVAL ((void*)(uintptr_t)vop_einval)
1156#define VOP_ENOENT ((void*)(uintptr_t)vop_enoent)
1157#define VOP_EOPNOTSUPP ((void*)(uintptr_t)vop_eopnotsupp)
1158#define VOP_EAGAIN ((void*)(uintptr_t)vop_eagain)
1159
1160/* fifo_vnops.c */
1161int fifo_printinfo(struct vnode *);
1162
1163/* vfs_hash.c */
1164typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
1165
1166void vfs_hash_changesize(u_long newhashsize);
1167int vfs_hash_get(const struct mount *mp, u_int hash, int flags,
1168 struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1169u_int vfs_hash_index(struct vnode *vp);
1170int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
1171 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1172void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
1173 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1174void vfs_hash_rehash(struct vnode *vp, u_int hash);
1175void vfs_hash_remove(struct vnode *vp);
1176
1177int vfs_kqfilter(struct vop_kqfilter_args *);
1178struct dirent;
1179int vn_dir_next_dirent(struct vnode *vp, struct thread *td,
1180 char *dirbuf, size_t dirbuflen,
1181 struct dirent **dpp, size_t *len, off_t *off, int *eofflag);
1182int vn_dir_check_empty(struct vnode *vp);
1183int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
1184
1185int vfs_unixify_accmode(accmode_t *accmode);
1186
1187void vfs_unp_reclaim(struct vnode *vp);
1188
1189int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
1190int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
1191 gid_t gid);
1192int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
1193 struct thread *td);
1194int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
1195 struct thread *td);
1196int vn_getsize_locked(struct vnode *vp, off_t *size, struct ucred *active_cred);
1197int vn_getsize(struct vnode *vp, off_t *size, struct ucred *active_cred);
1198
1199void vn_fsid(struct vnode *vp, struct vattr *va);
1200
1201int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp);
1202int vn_lktype_write(struct mount *mp, struct vnode *vp);
1203
1204#ifdef INVARIANTS
1205void vn_set_state_validate(struct vnode *vp, __enum_uint8(vstate) state);
1206#endif
1207
1208static inline void
1209vn_set_state(struct vnode *vp, __enum_uint8(vstate) state)
1210{
1211#ifdef INVARIANTS
1212 vn_set_state_validate(vp, state);
1213#endif
1214 vp->v_state = state;
1215}
1216
1217static inline __enum_uint8(vstate)
1218vn_get_state(struct vnode *vp)
1219{
1220 return (vp->v_state);
1221}
1222
1223#define VOP_UNLOCK_FLAGS(vp, flags) ({ \
1224 struct vnode *_vp = (vp); \
1225 int _flags = (flags); \
1226 int _error; \
1227 \
1228 if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0) \
1229 panic("%s: unsupported flags %x\n", __func__, flags); \
1230 _error = VOP_UNLOCK(_vp); \
1231 if (_flags & LK_INTERLOCK) \
1232 VI_UNLOCK(_vp); \
1233 _error; \
1234})
1235
1236#include <sys/kernel.h>
1237
1238#define VFS_VOP_VECTOR_REGISTER(vnodeops) \
1239 SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \
1240 vfs_vector_op_register, &vnodeops)
1241
1242#define VFS_SMR_DECLARE \
1243 extern smr_t vfs_smr
1244
1245#define VFS_SMR() vfs_smr
1246#define vfs_smr_enter() smr_enter(VFS_SMR())
1247#define vfs_smr_exit() smr_exit(VFS_SMR())
1248#define vfs_smr_synchronize() smr_synchronize(VFS_SMR())
1249#define vfs_smr_entered_load(ptr) smr_entered_load((ptr), VFS_SMR())
1250#define VFS_SMR_ENTERED() SMR_ENTERED(VFS_SMR())
1251#define VFS_SMR_ASSERT_ENTERED() SMR_ASSERT_ENTERED(VFS_SMR())
1252#define VFS_SMR_ASSERT_NOT_ENTERED() SMR_ASSERT_NOT_ENTERED(VFS_SMR())
1253#define VFS_SMR_ZONE_SET(zone) uma_zone_set_smr((zone), VFS_SMR())
1254
1255#define vn_load_v_data_smr(vp) ({ \
1256 struct vnode *_vp = (vp); \
1257 \
1258 VFS_SMR_ASSERT_ENTERED(); \
1259 atomic_load_consume_ptr(&(_vp)->v_data);\
1260})
1261
1262#endif /* _KERNEL */
1263
1264#endif /* !_SYS_VNODE_H_ */