1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1989, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 *
31 * @(#)vnode.h 8.7 (Berkeley) 2/4/94
32 */
33
34 #ifndef _SYS_VNODE_H_
35 #define _SYS_VNODE_H_
36
37 #include <sys/bufobj.h>
38 #include <sys/queue.h>
39 #include <sys/lock.h>
40 #include <sys/lockmgr.h>
41 #include <sys/mutex.h>
42 #include <sys/rangelock.h>
43 #include <sys/selinfo.h>
44 #include <sys/uio.h>
45 #include <sys/acl.h>
46 #include <sys/ktr.h>
47 #include <sys/_seqc.h>
48
49 /*
50 * The vnode is the focus of all file activity in UNIX. There is a
51 * unique vnode allocated for each active file, each current directory,
52 * each mounted-on file, text file, and the root.
53 */
54
55 /*
56 * Vnode types. VNON means no type.
57 */
58 enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD,
59 VMARKER };
60
61 enum vgetstate { VGET_NONE, VGET_HOLDCNT, VGET_USECOUNT };
62 /*
63 * Each underlying filesystem allocates its own private area and hangs
64 * it from v_data. If non-null, this area is freed in getnewvnode().
65 */
66
67 struct namecache;
68 struct cache_fpl;
69
70 struct vpollinfo {
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 */
75 };
76
77 /*
78 * Reading or writing any of these items requires holding the appropriate lock.
79 *
80 * Lock reference:
81 * c - namecache mutex
82 * i - interlock
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
86 * p - pollinfo lock
87 * u - Only a reference to the vnode is needed to read.
88 * v - vnode lock
89 *
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().
98 */
99
100 #if defined(_KERNEL) || defined(_KVM_VNODE)
101
102 struct vnode {
103 /*
104 * Fields which define the identity of the vnode. These fields are
105 * owned by the filesystem (XXX: and vgone() ?)
106 */
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 */
113
114 /*
115 * Filesystem instance stuff
116 */
117 struct mount *v_mount; /* u ptr to vfs we are in */
118 TAILQ_ENTRY(vnode) v_nmntvnodes; /* m vnodes for mount point */
119
120 /*
121 * Type specific fields, only one applies to any given vnode.
122 */
123 union {
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) */
128 };
129
130 /*
131 * vfs_hash: (mount + inode) -> vnode hash. The hash value
132 * itself is grouped with other int fields, to avoid padding.
133 */
134 LIST_ENTRY(vnode) v_hashlist;
135
136 /*
137 * VFS_namecache stuff
138 */
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 */
142
143 /*
144 * Locking
145 */
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 */
149
150 /*
151 * The machinery of being a vnode
152 */
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 */
156
157 /*
158 * Hooks for various subsystems and features.
159 */
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 */
164
165 /*
166 * clustering stuff
167 */
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 */
172
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 */
182 u_int v_hash;
183 };
184
185 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
186
187 #define bo2vnode(bo) __containerof((bo), struct vnode, v_bufobj)
188
189 /* XXX: These are temporary to avoid a source sweep at this time */
190 #define v_object v_bufobj.bo_object
191
192 /*
193 * Userland version of struct vnode, for sysctl.
194 */
195 struct xvnode {
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 */
206 union {
207 void *xvu_socket; /* unpcb, if VSOCK */
208 void *xvu_fifo; /* fifo, if VFIFO */
209 dev_t xvu_rdev; /* maj/min, if VBLK/VCHR */
210 struct {
211 dev_t xvu_dev; /* device, if VDIR/VREG/VLNK */
212 ino_t xvu_ino; /* id, if VDIR/VREG/VLNK */
213 } xv_uns;
214 } xv_un;
215 };
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
221
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))
225
226 #define VN_KNOTE(vp, b, a) \
227 do { \
228 if (!VN_KNLIST_EMPTY(vp)) \
229 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
230 (a) | KNF_NOKQLOCK); \
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 * VV flags are protected by the vnode lock and live in v_vflag
239 *
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.
242 */
243 #define VHOLD_NO_SMR (1<<29) /* Disable vhold_smr */
244 #define VHOLD_ALL_FLAGS (VHOLD_NO_SMR)
245
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 */
249 #define VIRF_MOUNTPOINT 0x0004 /* This vnode is mounted on */
250 #define VIRF_CROSSMP 0x0010 /* Cross-mp vnode, no locking */
251
252 #define VI_TEXT_REF 0x0001 /* Text ref grabbed use ref */
253 #define VI_MOUNT 0x0002 /* Mount in progress */
254 #define VI_DOINGINACT 0x0004 /* VOP_INACTIVE is in progress */
255 #define VI_OWEINACT 0x0008 /* Need to call inactive */
256 #define VI_DEFINACT 0x0010 /* deferred inactive */
257 #define VI_FOPENING 0x0020 /* In open, with opening process having the
258 first right to advlock file */
259
260 #define VV_ROOT 0x0001 /* root of its filesystem */
261 #define VV_ISTTY 0x0002 /* vnode represents a tty */
262 #define VV_NOSYNC 0x0004 /* unlinked, stop syncing */
263 #define VV_ETERNALDEV 0x0008 /* device that is never destroyed */
264 #define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */
265 #define VV_VMSIZEVNLOCK 0x0020 /* object size check requires vnode lock */
266 #define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */
267 #define VV_SYSTEM 0x0080 /* vnode being used by kernel */
268 #define VV_PROCDEP 0x0100 /* vnode is process dependent */
269 #define VV_NOKNOTE 0x0200 /* don't activate knotes on this vnode */
270 #define VV_DELETED 0x0400 /* should be removed */
271 #define VV_MD 0x0800 /* vnode backs the md device */
272 #define VV_FORCEINSMQ 0x1000 /* force the insmntque to succeed */
273 #define VV_READLINK 0x2000 /* fdescfs linux vnode */
274 #define VV_UNREF 0x4000 /* vunref, do not drop lock in inactive() */
275 #define VV_UNLINKED 0x8000 /* unlinked but stil open directory */
276
277 #define VMP_LAZYLIST 0x0001 /* Vnode is on mnt's lazy list */
278
279 /*
280 * Vnode attributes. A field value of VNOVAL represents a field whose value
281 * is unavailable (getattr) or which is not to be changed (setattr).
282 */
283 struct vattr {
284 enum vtype va_type; /* vnode type (for create) */
285 u_short va_mode; /* files access mode and type */
286 u_short va_padding0;
287 uid_t va_uid; /* owner user id */
288 gid_t va_gid; /* owner group id */
289 nlink_t va_nlink; /* number of references to file */
290 dev_t va_fsid; /* filesystem id */
291 ino_t va_fileid; /* file id */
292 u_quad_t va_size; /* file size in bytes */
293 long va_blocksize; /* blocksize preferred for i/o */
294 struct timespec va_atime; /* time of last access */
295 struct timespec va_mtime; /* time of last modification */
296 struct timespec va_ctime; /* time file changed */
297 struct timespec va_birthtime; /* time file created */
298 u_long va_gen; /* generation number of file */
299 u_long va_flags; /* flags defined for file */
300 dev_t va_rdev; /* device the special file represents */
301 u_quad_t va_bytes; /* bytes of disk space held by file */
302 u_quad_t va_filerev; /* file modification number */
303 u_int va_vaflags; /* operations flags, see below */
304 long va_spare; /* remain quad aligned */
305 };
306
307 /*
308 * Flags for va_vaflags.
309 */
310 #define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */
311 #define VA_EXCLUSIVE 0x02 /* exclusive create request */
312 #define VA_SYNC 0x04 /* O_SYNC truncation */
313
314 /*
315 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
316 * help with write clustering)
317 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
318 */
319 #define IO_UNIT 0x0001 /* do I/O as atomic unit */
320 #define IO_APPEND 0x0002 /* append write to end */
321 #define IO_NDELAY 0x0004 /* FNDELAY flag set in file table */
322 #define IO_NODELOCKED 0x0008 /* underlying node already locked */
323 #define IO_ASYNC 0x0010 /* bawrite rather then bdwrite */
324 #define IO_VMIO 0x0020 /* data already in VMIO space */
325 #define IO_INVAL 0x0040 /* invalidate after I/O */
326 #define IO_SYNC 0x0080 /* do I/O synchronously */
327 #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */
328 #define IO_NOREUSE 0x0200 /* VMIO data won't be reused */
329 #define IO_EXT 0x0400 /* operate on external attributes */
330 #define IO_NORMAL 0x0800 /* operate on regular data */
331 #define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */
332 #define IO_BUFLOCKED 0x2000 /* ffs flag; indir buf is locked */
333 #define IO_RANGELOCKED 0x4000 /* range locked */
334 #define IO_DATASYNC 0x8000 /* do only data I/O synchronously */
335
336 #define IO_SEQMAX 0x7F /* seq heuristic max value */
337 #define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */
338
339 /*
340 * Flags for accmode_t.
341 */
342 #define VEXEC 000000000100 /* execute/search permission */
343 #define VWRITE 000000000200 /* write permission */
344 #define VREAD 000000000400 /* read permission */
345 #define VADMIN 000000010000 /* being the file owner */
346 #define VAPPEND 000000040000 /* permission to write/append */
347 /*
348 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
349 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
350 * and 0 otherwise. This never happens with ordinary unix access rights
351 * or POSIX.1e ACLs. Obviously, VEXPLICIT_DENY must be OR-ed with
352 * some other V* constant.
353 */
354 #define VEXPLICIT_DENY 000000100000
355 #define VREAD_NAMED_ATTRS 000000200000 /* not used */
356 #define VWRITE_NAMED_ATTRS 000000400000 /* not used */
357 #define VDELETE_CHILD 000001000000
358 #define VREAD_ATTRIBUTES 000002000000 /* permission to stat(2) */
359 #define VWRITE_ATTRIBUTES 000004000000 /* change {m,c,a}time */
360 #define VDELETE 000010000000
361 #define VREAD_ACL 000020000000 /* read ACL and file mode */
362 #define VWRITE_ACL 000040000000 /* change ACL and/or file mode */
363 #define VWRITE_OWNER 000100000000 /* change file owner */
364 #define VSYNCHRONIZE 000200000000 /* not used */
365 #define VCREAT 000400000000 /* creating new file */
366 #define VVERIFY 001000000000 /* verification required */
367
368 /*
369 * Permissions that were traditionally granted only to the file owner.
370 */
371 #define VADMIN_PERMS (VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
372 VWRITE_OWNER)
373
374 /*
375 * Permissions that were traditionally granted to everyone.
376 */
377 #define VSTAT_PERMS (VREAD_ATTRIBUTES | VREAD_ACL)
378
379 /*
380 * Permissions that allow to change the state of the file in any way.
381 */
382 #define VMODIFY_PERMS (VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
383 VDELETE)
384
385 /*
386 * Token indicating no attribute value yet assigned.
387 */
388 #define VNOVAL (-1)
389
390 /*
391 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
392 */
393 #define VLKTIMEOUT (hz / 20 + 1)
394
395 #ifdef _KERNEL
396
397 #ifdef MALLOC_DECLARE
398 MALLOC_DECLARE(M_VNODE);
399 #endif
400
401 extern u_int ncsizefactor;
402 extern const u_int io_hold_cnt;
403
404 /*
405 * Convert between vnode types and inode formats (since POSIX.1
406 * defines mode word of stat structure in terms of inode formats).
407 */
408 extern enum vtype iftovt_tab[];
409 extern int vttoif_tab[];
410 #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12])
411 #define VTTOIF(indx) (vttoif_tab[(int)(indx)])
412 #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode))
413
414 /*
415 * Flags to various vnode functions.
416 */
417 #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */
418 #define FORCECLOSE 0x0002 /* vflush: force file closure */
419 #define WRITECLOSE 0x0004 /* vflush: only close writable files */
420 #define EARLYFLUSH 0x0008 /* vflush: early call for ffs_flushfiles */
421 #define V_SAVE 0x0001 /* vinvalbuf: sync file first */
422 #define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */
423 #define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */
424 #define V_CLEANONLY 0x0008 /* vinvalbuf: invalidate only clean bufs */
425 #define V_VMIO 0x0010 /* vinvalbuf: called during pageout */
426 #define V_ALLOWCLEAN 0x0020 /* vinvalbuf: allow clean buffers after flush */
427 #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */
428 #define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */
429 #define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */
430 #define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */
431 #define V_MNTREF 0x0010 /* vn_start_write: mp is already ref-ed */
432
433 #define VR_START_WRITE 0x0001 /* vfs_write_resume: start write atomically */
434 #define VR_NO_SUSPCLR 0x0002 /* vfs_write_resume: do not clear suspension */
435
436 #define VS_SKIP_UNMOUNT 0x0001 /* vfs_write_suspend: fail if the
437 filesystem is being unmounted */
438
439 #define VREF(vp) vref(vp)
440
441 #ifdef DIAGNOSTIC
442 #define VATTR_NULL(vap) vattr_null(vap)
443 #else
444 #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */
445 #endif /* DIAGNOSTIC */
446
447 #define NULLVP ((struct vnode *)NULL)
448
449 /*
450 * Global vnode data.
451 */
452 extern struct vnode *rootvnode; /* root (i.e. "/") vnode */
453 extern struct mount *rootdevmp; /* "/dev" mount */
454 extern u_long desiredvnodes; /* number of vnodes desired */
455 extern struct uma_zone *namei_zone;
456 extern struct vattr va_null; /* predefined null vattr structure */
457
458 extern u_int vn_lock_pair_pause_max;
459
460 #define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock)
461 #define VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
462 #define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock)
463 #define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock)
464 #define VI_MTX(vp) (&(vp)->v_interlock)
465
466 #define VN_LOCK_AREC(vp) lockallowrecurse((vp)->v_vnlock)
467 #define VN_LOCK_ASHARE(vp) lockallowshare((vp)->v_vnlock)
468 #define VN_LOCK_DSHARE(vp) lockdisableshare((vp)->v_vnlock)
469
470 #endif /* _KERNEL */
471
472 /*
473 * Mods for extensibility.
474 */
475
476 /*
477 * Flags for vdesc_flags:
478 */
479 #define VDESC_MAX_VPS 16
480 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
481 #define VDESC_VP0_WILLRELE 0x0001
482 #define VDESC_VP1_WILLRELE 0x0002
483 #define VDESC_VP2_WILLRELE 0x0004
484 #define VDESC_VP3_WILLRELE 0x0008
485
486 /*
487 * A generic structure.
488 * This can be used by bypass routines to identify generic arguments.
489 */
490 struct vop_generic_args {
491 struct vnodeop_desc *a_desc;
492 /* other random data follows, presumably */
493 };
494
495 typedef int vop_bypass_t(struct vop_generic_args *);
496
497 /*
498 * VDESC_NO_OFFSET is used to identify the end of the offset list
499 * and in places where no such field exists.
500 */
501 #define VDESC_NO_OFFSET -1
502
503 /*
504 * This structure describes the vnode operation taking place.
505 */
506 struct vnodeop_desc {
507 char *vdesc_name; /* a readable name for debugging */
508 int vdesc_flags; /* VDESC_* flags */
509 int vdesc_vop_offset;
510 vop_bypass_t *vdesc_call; /* Function to call */
511
512 /*
513 * These ops are used by bypass routines to map and locate arguments.
514 * Creds and procs are not needed in bypass routines, but sometimes
515 * they are useful to (for example) transport layers.
516 * Nameidata is useful because it has a cred in it.
517 */
518 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */
519 int vdesc_vpp_offset; /* return vpp location */
520 int vdesc_cred_offset; /* cred location, if any */
521 int vdesc_thread_offset; /* thread location, if any */
522 int vdesc_componentname_offset; /* if any */
523 };
524
525 #ifdef _KERNEL
526 /*
527 * A list of all the operation descs.
528 */
529 extern struct vnodeop_desc *vnodeop_descs[];
530
531 #define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field)
532 #define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
533 ((s_type)(((char*)(struct_p)) + (s_offset)))
534
535 #ifdef DEBUG_VFS_LOCKS
536 /*
537 * Support code to aid in debugging VFS locking problems. Not totally
538 * reliable since if the thread sleeps between changing the lock
539 * state and checking it with the assert, some other thread could
540 * change the state. They are good enough for debugging a single
541 * filesystem using a single-threaded test. Note that the unreliability is
542 * limited to false negatives; efforts were made to ensure that false
543 * positives cannot occur.
544 */
545 void assert_vi_locked(struct vnode *vp, const char *str);
546 void assert_vi_unlocked(struct vnode *vp, const char *str);
547 void assert_vop_elocked(struct vnode *vp, const char *str);
548 void assert_vop_locked(struct vnode *vp, const char *str);
549 void assert_vop_unlocked(struct vnode *vp, const char *str);
550
551 #define ASSERT_VI_LOCKED(vp, str) assert_vi_locked((vp), (str))
552 #define ASSERT_VI_UNLOCKED(vp, str) assert_vi_unlocked((vp), (str))
553 #define ASSERT_VOP_ELOCKED(vp, str) assert_vop_elocked((vp), (str))
554 #define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked((vp), (str))
555 #define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked((vp), (str))
556
557 #define ASSERT_VOP_IN_SEQC(vp) do { \
558 struct vnode *_vp = (vp); \
559 \
560 VNPASS(seqc_in_modify(_vp->v_seqc), _vp); \
561 } while (0)
562
563 #define ASSERT_VOP_NOT_IN_SEQC(vp) do { \
564 struct vnode *_vp = (vp); \
565 \
566 VNPASS(!seqc_in_modify(_vp->v_seqc), _vp); \
567 } while (0)
568
569 #else /* !DEBUG_VFS_LOCKS */
570
571 #define ASSERT_VI_LOCKED(vp, str) ((void)0)
572 #define ASSERT_VI_UNLOCKED(vp, str) ((void)0)
573 #define ASSERT_VOP_ELOCKED(vp, str) ((void)0)
574 #define ASSERT_VOP_LOCKED(vp, str) ((void)0)
575 #define ASSERT_VOP_UNLOCKED(vp, str) ((void)0)
576
577 #define ASSERT_VOP_IN_SEQC(vp) ((void)0)
578 #define ASSERT_VOP_NOT_IN_SEQC(vp) ((void)0)
579
580 #endif /* DEBUG_VFS_LOCKS */
581
582 /*
583 * This call works for vnodes in the kernel.
584 */
585 #define VCALL(c) ((c)->a_desc->vdesc_call(c))
586
587 #define DOINGASYNC(vp) \
588 (((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 && \
589 ((curthread->td_pflags & TDP_SYNCIO) == 0))
590
591 /*
592 * VMIO support inline
593 */
594
595 extern int vmiodirenable;
596
597 static __inline int
vn_canvmio(struct vnode * vp)598 vn_canvmio(struct vnode *vp)
599 {
600 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
601 return(TRUE);
602 return(FALSE);
603 }
604
605 /*
606 * Finally, include the default set of vnode operations.
607 */
608 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
609 #include "vnode_if.h"
610
611 /* vn_open_flags */
612 #define VN_OPEN_NOAUDIT 0x00000001
613 #define VN_OPEN_NOCAPCHECK 0x00000002
614 #define VN_OPEN_NAMECACHE 0x00000004
615 #define VN_OPEN_INVFS 0x00000008
616 #define VN_OPEN_WANTIOCTLCAPS 0x00000010
617
618 /* copy_file_range kernel flags */
619 #define COPY_FILE_RANGE_KFLAGS 0xff000000
620 #define COPY_FILE_RANGE_TIMEO1SEC 0x01000000 /* Return after 1sec. */
621
622 /*
623 * Public vnode manipulation functions.
624 */
625 struct componentname;
626 struct file;
627 struct mount;
628 struct nameidata;
629 struct ostat;
630 struct freebsd11_stat;
631 struct thread;
632 struct proc;
633 struct stat;
634 struct nstat;
635 struct ucred;
636 struct uio;
637 struct vattr;
638 struct vfsops;
639 struct vnode;
640
641 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
642
643 int bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
644 daddr_t endn);
645 /* cache_* may belong in namei.h. */
646 void cache_changesize(u_long newhashsize);
647
648 #define VFS_CACHE_DROPOLD 0x1
649
650 void cache_enter_time_flags(struct vnode *dvp, struct vnode *vp,
651 struct componentname *cnp, struct timespec *tsp,
652 struct timespec *dtsp, int flags);
653 #define cache_enter(dvp, vp, cnp) \
654 cache_enter_time(dvp, vp, cnp, NULL, NULL)
655 void cache_enter_time(struct vnode *dvp, struct vnode *vp,
656 struct componentname *cnp, struct timespec *tsp,
657 struct timespec *dtsp);
658 int cache_lookup(struct vnode *dvp, struct vnode **vpp,
659 struct componentname *cnp, struct timespec *tsp, int *ticksp);
660 void cache_vnode_init(struct vnode *vp);
661 void cache_purge(struct vnode *vp);
662 void cache_purge_vgone(struct vnode *vp);
663 void cache_purge_negative(struct vnode *vp);
664 void cache_purgevfs(struct mount *mp);
665 char *cache_symlink_alloc(size_t size, int flags);
666 void cache_symlink_free(char *string, size_t size);
667 int cache_symlink_resolve(struct cache_fpl *fpl, const char *string,
668 size_t len);
669 void cache_vop_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp,
670 struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp);
671 void cache_vop_rmdir(struct vnode *dvp, struct vnode *vp);
672 #ifdef INVARIANTS
673 void cache_validate(struct vnode *dvp, struct vnode *vp,
674 struct componentname *cnp);
675 #else
676 static inline void
cache_validate(struct vnode * dvp,struct vnode * vp,struct componentname * cnp)677 cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
678 {
679 }
680 #endif
681 void cache_fast_lookup_enabled_recalc(void);
682 int change_dir(struct vnode *vp, struct thread *td);
683 void cvtstat(struct stat *st, struct ostat *ost);
684 void freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb);
685 int freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost);
686 int getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
687 struct vnode **vpp);
688 void getnewvnode_reserve(void);
689 void getnewvnode_drop_reserve(void);
690 int insmntque1(struct vnode *vp, struct mount *mp,
691 void (*dtr)(struct vnode *, void *), void *dtr_arg);
692 int insmntque(struct vnode *vp, struct mount *mp);
693 u_quad_t init_va_filerev(void);
694 int speedup_syncer(void);
695 int vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen);
696 int vn_getcwd(char *buf, char **retbuf, size_t *buflen);
697 int vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf);
698 int vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf);
699 int vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp,
700 const char *hdrl_name, size_t hrdl_name_length, char **retbuf,
701 char **freebuf, size_t *buflen);
702 struct vnode *
703 vn_dir_dd_ino(struct vnode *vp);
704 int vn_commname(struct vnode *vn, char *buf, u_int buflen);
705 int vn_path_to_global_path(struct thread *td, struct vnode *vp,
706 char *path, u_int pathlen);
707 int vn_path_to_global_path_hardlink(struct thread *td, struct vnode *vp,
708 struct vnode *dvp, char *path, u_int pathlen, const char *leaf_name,
709 size_t leaf_length);
710 int vaccess(enum vtype type, mode_t file_mode, uid_t file_uid,
711 gid_t file_gid, accmode_t accmode, struct ucred *cred);
712 int vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid,
713 struct ucred *cred);
714 int vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid,
715 struct acl *aclp, accmode_t accmode, struct ucred *cred);
716 int vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
717 gid_t file_gid, struct acl *acl, accmode_t accmode,
718 struct ucred *cred);
719 void vattr_null(struct vattr *vap);
720 void vlazy(struct vnode *);
721 void vdrop(struct vnode *);
722 void vdropl(struct vnode *);
723 int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
724 int vget(struct vnode *vp, int flags);
725 enum vgetstate vget_prep_smr(struct vnode *vp);
726 enum vgetstate vget_prep(struct vnode *vp);
727 int vget_finish(struct vnode *vp, int flags, enum vgetstate vs);
728 void vget_finish_ref(struct vnode *vp, enum vgetstate vs);
729 void vget_abort(struct vnode *vp, enum vgetstate vs);
730 void vgone(struct vnode *vp);
731 void vhold(struct vnode *);
732 void vholdnz(struct vnode *);
733 bool vhold_smr(struct vnode *);
734 int vinactive(struct vnode *vp);
735 int vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
736 int vtruncbuf(struct vnode *vp, off_t length, int blksize);
737 void v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn,
738 int blksize);
739 void vunref(struct vnode *);
740 void vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
741 int vrecycle(struct vnode *vp);
742 int vrecyclel(struct vnode *vp);
743 int vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
744 struct ucred *cred);
745 int vn_close(struct vnode *vp,
746 int flags, struct ucred *file_cred, struct thread *td);
747 int vn_copy_file_range(struct vnode *invp, off_t *inoffp,
748 struct vnode *outvp, off_t *outoffp, size_t *lenp,
749 unsigned int flags, struct ucred *incred, struct ucred *outcred,
750 struct thread *fsize_td);
751 void vn_finished_write(struct mount *mp);
752 void vn_finished_secondary_write(struct mount *mp);
753 int vn_fsync_buf(struct vnode *vp, int waitfor);
754 int vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp,
755 struct vnode *outvp, off_t *outoffp, size_t *lenp,
756 unsigned int flags, struct ucred *incred, struct ucred *outcred,
757 struct thread *fsize_td);
758 int vn_need_pageq_flush(struct vnode *vp);
759 bool vn_isdisk_error(struct vnode *vp, int *errp);
760 bool vn_isdisk(struct vnode *vp);
761 int _vn_lock(struct vnode *vp, int flags, const char *file, int line);
762 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
763 void vn_lock_pair(struct vnode *vp1, bool vp1_locked, int lkflags1,
764 struct vnode *vp2, bool vp2_locked, int lkflags2);
765 int vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
766 int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
767 u_int vn_open_flags, struct ucred *cred, struct file *fp);
768 int vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
769 struct thread *td, struct file *fp);
770 void vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
771 void vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start,
772 vm_pindex_t end);
773 int vn_pollrecord(struct vnode *vp, struct thread *p, int events);
774 int vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
775 int len, off_t offset, enum uio_seg segflg, int ioflg,
776 struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
777 struct thread *td);
778 int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
779 size_t len, off_t offset, enum uio_seg segflg, int ioflg,
780 struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
781 struct thread *td);
782 int vn_read_from_obj(struct vnode *vp, struct uio *uio);
783 int vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio,
784 struct thread *td);
785 int vn_rlimit_fsizex(const struct vnode *vp, struct uio *uio,
786 off_t maxfsz, ssize_t *resid_adj, struct thread *td);
787 void vn_rlimit_fsizex_res(struct uio *uio, ssize_t resid_adj);
788 int vn_rlimit_trunc(u_quad_t size, struct thread *td);
789 int vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
790 int vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
791 int flags);
792 int vn_truncate_locked(struct vnode *vp, off_t length, bool sync,
793 struct ucred *cred);
794 int vn_writechk(struct vnode *vp);
795 int vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
796 const char *attrname, int *buflen, char *buf, struct thread *td);
797 int vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
798 const char *attrname, int buflen, char *buf, struct thread *td);
799 int vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
800 const char *attrname, struct thread *td);
801 int vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
802 struct vnode **rvp);
803 int vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
804 void *alloc_arg, int lkflags, struct vnode **rvp);
805 int vn_utimes_perm(struct vnode *vp, struct vattr *vap,
806 struct ucred *cred, struct thread *td);
807 int vn_cmp(struct file *, struct file *, struct thread *td);
808
809 int vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
810 int vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
811 struct uio *uio);
812
813 void vn_seqc_write_begin_locked(struct vnode *vp);
814 void vn_seqc_write_begin(struct vnode *vp);
815 void vn_seqc_write_end_locked(struct vnode *vp);
816 void vn_seqc_write_end(struct vnode *vp);
817 #define vn_seqc_read_any(vp) seqc_read_any(&(vp)->v_seqc)
818 #define vn_seqc_read_notmodify(vp) seqc_read_notmodify(&(vp)->v_seqc)
819 #define vn_seqc_consistent(vp, seq) seqc_consistent(&(vp)->v_seqc, seq)
820
821 #define vn_rangelock_unlock(vp, cookie) \
822 rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp))
823 #define vn_rangelock_unlock_range(vp, cookie, start, end) \
824 rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), \
825 VI_MTX(vp))
826 #define vn_rangelock_rlock(vp, start, end) \
827 rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
828 #define vn_rangelock_tryrlock(vp, start, end) \
829 rangelock_tryrlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
830 #define vn_rangelock_wlock(vp, start, end) \
831 rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
832 #define vn_rangelock_trywlock(vp, start, end) \
833 rangelock_trywlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
834
835 #define vn_irflag_read(vp) atomic_load_short(&(vp)->v_irflag)
836 void vn_irflag_set_locked(struct vnode *vp, short toset);
837 void vn_irflag_set(struct vnode *vp, short toset);
838 void vn_irflag_set_cond_locked(struct vnode *vp, short toset);
839 void vn_irflag_set_cond(struct vnode *vp, short toset);
840 void vn_irflag_unset_locked(struct vnode *vp, short tounset);
841 void vn_irflag_unset(struct vnode *vp, short tounset);
842
843 int vfs_cache_lookup(struct vop_lookup_args *ap);
844 int vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp);
845 void vfs_timestamp(struct timespec *);
846 void vfs_write_resume(struct mount *mp, int flags);
847 int vfs_write_suspend(struct mount *mp, int flags);
848 int vfs_write_suspend_umnt(struct mount *mp);
849 struct vnode *vnlru_alloc_marker(void);
850 void vnlru_free_marker(struct vnode *);
851 void vnlru_free(int, struct vfsops *);
852 void vnlru_free_vfsops(int, struct vfsops *, struct vnode *);
853 int vop_stdbmap(struct vop_bmap_args *);
854 int vop_stdfdatasync_buf(struct vop_fdatasync_args *);
855 int vop_stdfsync(struct vop_fsync_args *);
856 int vop_stdgetwritemount(struct vop_getwritemount_args *);
857 int vop_stdgetpages(struct vop_getpages_args *);
858 int vop_stdinactive(struct vop_inactive_args *);
859 int vop_stdioctl(struct vop_ioctl_args *);
860 int vop_stdneed_inactive(struct vop_need_inactive_args *);
861 int vop_stdkqfilter(struct vop_kqfilter_args *);
862 int vop_stdlock(struct vop_lock1_args *);
863 int vop_stdunlock(struct vop_unlock_args *);
864 int vop_stdislocked(struct vop_islocked_args *);
865 int vop_lock(struct vop_lock1_args *);
866 int vop_unlock(struct vop_unlock_args *);
867 int vop_islocked(struct vop_islocked_args *);
868 int vop_stdputpages(struct vop_putpages_args *);
869 int vop_nopoll(struct vop_poll_args *);
870 int vop_stdaccess(struct vop_access_args *ap);
871 int vop_stdaccessx(struct vop_accessx_args *ap);
872 int vop_stdadvise(struct vop_advise_args *ap);
873 int vop_stdadvlock(struct vop_advlock_args *ap);
874 int vop_stdadvlockasync(struct vop_advlockasync_args *ap);
875 int vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
876 int vop_stdallocate(struct vop_allocate_args *ap);
877 int vop_stdset_text(struct vop_set_text_args *ap);
878 int vop_stdpathconf(struct vop_pathconf_args *);
879 int vop_stdpoll(struct vop_poll_args *);
880 int vop_stdvptocnp(struct vop_vptocnp_args *ap);
881 int vop_stdvptofh(struct vop_vptofh_args *ap);
882 int vop_stdunp_bind(struct vop_unp_bind_args *ap);
883 int vop_stdunp_connect(struct vop_unp_connect_args *ap);
884 int vop_stdunp_detach(struct vop_unp_detach_args *ap);
885 int vop_eopnotsupp(struct vop_generic_args *ap);
886 int vop_ebadf(struct vop_generic_args *ap);
887 int vop_einval(struct vop_generic_args *ap);
888 int vop_enoent(struct vop_generic_args *ap);
889 int vop_enotty(struct vop_generic_args *ap);
890 int vop_eagain(struct vop_generic_args *ap);
891 int vop_null(struct vop_generic_args *ap);
892 int vop_panic(struct vop_generic_args *ap);
893 int dead_poll(struct vop_poll_args *ap);
894 int dead_read(struct vop_read_args *ap);
895 int dead_write(struct vop_write_args *ap);
896
897 /* These are called from within the actual VOPS. */
898 void vop_close_post(void *a, int rc);
899 void vop_create_pre(void *a);
900 void vop_create_post(void *a, int rc);
901 void vop_whiteout_pre(void *a);
902 void vop_whiteout_post(void *a, int rc);
903 void vop_deleteextattr_pre(void *a);
904 void vop_deleteextattr_post(void *a, int rc);
905 void vop_link_pre(void *a);
906 void vop_link_post(void *a, int rc);
907 void vop_lookup_post(void *a, int rc);
908 void vop_lookup_pre(void *a);
909 void vop_mkdir_pre(void *a);
910 void vop_mkdir_post(void *a, int rc);
911 void vop_mknod_pre(void *a);
912 void vop_mknod_post(void *a, int rc);
913 void vop_open_post(void *a, int rc);
914 void vop_read_post(void *a, int rc);
915 void vop_read_pgcache_post(void *ap, int rc);
916 void vop_readdir_post(void *a, int rc);
917 void vop_reclaim_post(void *a, int rc);
918 void vop_remove_pre(void *a);
919 void vop_remove_post(void *a, int rc);
920 void vop_rename_post(void *a, int rc);
921 void vop_rename_pre(void *a);
922 void vop_rmdir_pre(void *a);
923 void vop_rmdir_post(void *a, int rc);
924 void vop_setattr_pre(void *a);
925 void vop_setattr_post(void *a, int rc);
926 void vop_setacl_pre(void *a);
927 void vop_setacl_post(void *a, int rc);
928 void vop_setextattr_pre(void *a);
929 void vop_setextattr_post(void *a, int rc);
930 void vop_symlink_pre(void *a);
931 void vop_symlink_post(void *a, int rc);
932 int vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a);
933
934 #ifdef DEBUG_VFS_LOCKS
935 void vop_fdatasync_debugpre(void *a);
936 void vop_fdatasync_debugpost(void *a, int rc);
937 void vop_fplookup_vexec_debugpre(void *a);
938 void vop_fplookup_vexec_debugpost(void *a, int rc);
939 void vop_fplookup_symlink_debugpre(void *a);
940 void vop_fplookup_symlink_debugpost(void *a, int rc);
941 void vop_fsync_debugpre(void *a);
942 void vop_fsync_debugpost(void *a, int rc);
943 void vop_strategy_debugpre(void *a);
944 void vop_lock_debugpre(void *a);
945 void vop_lock_debugpost(void *a, int rc);
946 void vop_unlock_debugpre(void *a);
947 void vop_need_inactive_debugpre(void *a);
948 void vop_need_inactive_debugpost(void *a, int rc);
949 void vop_mkdir_debugpost(void *a, int rc);
950 #else
951 #define vop_fdatasync_debugpre(x) do { } while (0)
952 #define vop_fdatasync_debugpost(x, y) do { } while (0)
953 #define vop_fplookup_vexec_debugpre(x) do { } while (0)
954 #define vop_fplookup_vexec_debugpost(x, y) do { } while (0)
955 #define vop_fplookup_symlink_debugpre(x) do { } while (0)
956 #define vop_fplookup_symlink_debugpost(x, y) do { } while (0)
957 #define vop_fsync_debugpre(x) do { } while (0)
958 #define vop_fsync_debugpost(x, y) do { } while (0)
959 #define vop_strategy_debugpre(x) do { } while (0)
960 #define vop_lock_debugpre(x) do { } while (0)
961 #define vop_lock_debugpost(x, y) do { } while (0)
962 #define vop_unlock_debugpre(x) do { } while (0)
963 #define vop_need_inactive_debugpre(x) do { } while (0)
964 #define vop_need_inactive_debugpost(x, y) do { } while (0)
965 #define vop_mkdir_debugpost(x, y) do { } while (0)
966 #endif
967
968 void vop_rename_fail(struct vop_rename_args *ap);
969
970 #define vop_stat_helper_pre(ap) ({ \
971 int _error; \
972 AUDIT_ARG_VNODE1(ap->a_vp); \
973 _error = mac_vnode_check_stat(ap->a_active_cred, ap->a_file_cred, ap->a_vp);\
974 if (__predict_true(_error == 0)) \
975 bzero(ap->a_sb, sizeof(*ap->a_sb)); \
976 _error; \
977 })
978
979 #define vop_stat_helper_post(ap, error) ({ \
980 int _error = (error); \
981 if (priv_check_cred_vfs_generation(ap->a_td->td_ucred)) \
982 ap->a_sb->st_gen = 0; \
983 _error; \
984 })
985
986 #define VOP_WRITE_PRE(ap) \
987 struct vattr va; \
988 int error; \
989 off_t osize, ooffset, noffset; \
990 \
991 osize = ooffset = noffset = 0; \
992 if (!VN_KNLIST_EMPTY((ap)->a_vp)) { \
993 error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred); \
994 if (error) \
995 return (error); \
996 ooffset = (ap)->a_uio->uio_offset; \
997 osize = (off_t)va.va_size; \
998 }
999
1000 #define VOP_WRITE_POST(ap, ret) \
1001 noffset = (ap)->a_uio->uio_offset; \
1002 if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) { \
1003 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE \
1004 | (noffset > osize ? NOTE_EXTEND : 0)); \
1005 }
1006
1007 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
1008
1009 #ifdef INVARIANTS
1010 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) \
1011 do { \
1012 int error_; \
1013 \
1014 error_ = VOP_ADD_WRITECOUNT((vp), (cnt)); \
1015 VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d", \
1016 error_)); \
1017 } while (0)
1018 #define VOP_SET_TEXT_CHECKED(vp) \
1019 do { \
1020 int error_; \
1021 \
1022 error_ = VOP_SET_TEXT((vp)); \
1023 VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d", \
1024 error_)); \
1025 } while (0)
1026 #define VOP_UNSET_TEXT_CHECKED(vp) \
1027 do { \
1028 int error_; \
1029 \
1030 error_ = VOP_UNSET_TEXT((vp)); \
1031 VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d", \
1032 error_)); \
1033 } while (0)
1034 #else
1035 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) VOP_ADD_WRITECOUNT((vp), (cnt))
1036 #define VOP_SET_TEXT_CHECKED(vp) VOP_SET_TEXT((vp))
1037 #define VOP_UNSET_TEXT_CHECKED(vp) VOP_UNSET_TEXT((vp))
1038 #endif
1039
1040 #define VN_IS_DOOMED(vp) __predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0)
1041
1042 void vput(struct vnode *vp);
1043 void vrele(struct vnode *vp);
1044 void vref(struct vnode *vp);
1045 void vrefact(struct vnode *vp);
1046 void v_addpollinfo(struct vnode *vp);
1047 static __inline int
vrefcnt(struct vnode * vp)1048 vrefcnt(struct vnode *vp)
1049 {
1050
1051 return (vp->v_usecount);
1052 }
1053
1054 #define vholdl(vp) do { \
1055 ASSERT_VI_LOCKED(vp, __func__); \
1056 vhold(vp); \
1057 } while (0)
1058
1059 #define vrefl(vp) do { \
1060 ASSERT_VI_LOCKED(vp, __func__); \
1061 vref(vp); \
1062 } while (0)
1063
1064 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
1065 void vnode_destroy_vobject(struct vnode *vp);
1066
1067 extern struct vop_vector fifo_specops;
1068 extern struct vop_vector dead_vnodeops;
1069 extern struct vop_vector default_vnodeops;
1070
1071 #define VOP_PANIC ((void*)(uintptr_t)vop_panic)
1072 #define VOP_NULL ((void*)(uintptr_t)vop_null)
1073 #define VOP_EBADF ((void*)(uintptr_t)vop_ebadf)
1074 #define VOP_ENOTTY ((void*)(uintptr_t)vop_enotty)
1075 #define VOP_EINVAL ((void*)(uintptr_t)vop_einval)
1076 #define VOP_ENOENT ((void*)(uintptr_t)vop_enoent)
1077 #define VOP_EOPNOTSUPP ((void*)(uintptr_t)vop_eopnotsupp)
1078 #define VOP_EAGAIN ((void*)(uintptr_t)vop_eagain)
1079
1080 /* fifo_vnops.c */
1081 int fifo_printinfo(struct vnode *);
1082
1083 /* vfs_hash.c */
1084 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
1085
1086 void vfs_hash_changesize(u_long newhashsize);
1087 int vfs_hash_get(const struct mount *mp, u_int hash, int flags,
1088 struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1089 u_int vfs_hash_index(struct vnode *vp);
1090 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
1091 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1092 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
1093 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1094 void vfs_hash_rehash(struct vnode *vp, u_int hash);
1095 void vfs_hash_remove(struct vnode *vp);
1096
1097 int vfs_kqfilter(struct vop_kqfilter_args *);
1098 struct dirent;
1099 int vn_dir_next_dirent(struct vnode *vp, struct thread *td,
1100 char *dirbuf, size_t dirbuflen,
1101 struct dirent **dpp, size_t *len, off_t *off, int *eofflag);
1102 int vn_dir_check_empty(struct vnode *vp);
1103 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
1104
1105 int vfs_unixify_accmode(accmode_t *accmode);
1106
1107 void vfs_unp_reclaim(struct vnode *vp);
1108
1109 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
1110 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
1111 gid_t gid);
1112 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
1113 struct thread *td);
1114 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
1115 struct thread *td);
1116
1117 void vn_fsid(struct vnode *vp, struct vattr *va);
1118
1119 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp);
1120 int vn_lktype_write(struct mount *mp, struct vnode *vp);
1121
1122 #define VOP_UNLOCK_FLAGS(vp, flags) ({ \
1123 struct vnode *_vp = (vp); \
1124 int _flags = (flags); \
1125 int _error; \
1126 \
1127 if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0) \
1128 panic("%s: unsupported flags %x\n", __func__, flags); \
1129 _error = VOP_UNLOCK(_vp); \
1130 if (_flags & LK_INTERLOCK) \
1131 VI_UNLOCK(_vp); \
1132 _error; \
1133 })
1134
1135 #include <sys/kernel.h>
1136
1137 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \
1138 SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \
1139 vfs_vector_op_register, &vnodeops)
1140
1141 #define VFS_SMR_DECLARE \
1142 extern smr_t vfs_smr
1143
1144 #define VFS_SMR() vfs_smr
1145 #define vfs_smr_enter() smr_enter(VFS_SMR())
1146 #define vfs_smr_exit() smr_exit(VFS_SMR())
1147 #define vfs_smr_synchronize() smr_synchronize(VFS_SMR())
1148 #define vfs_smr_entered_load(ptr) smr_entered_load((ptr), VFS_SMR())
1149 #define VFS_SMR_ASSERT_ENTERED() SMR_ASSERT_ENTERED(VFS_SMR())
1150 #define VFS_SMR_ASSERT_NOT_ENTERED() SMR_ASSERT_NOT_ENTERED(VFS_SMR())
1151 #define VFS_SMR_ZONE_SET(zone) uma_zone_set_smr((zone), VFS_SMR())
1152
1153 #define vn_load_v_data_smr(vp) ({ \
1154 struct vnode *_vp = (vp); \
1155 \
1156 VFS_SMR_ASSERT_ENTERED(); \
1157 atomic_load_consume_ptr(&(_vp)->v_data);\
1158 })
1159
1160 #endif /* _KERNEL */
1161
1162 #endif /* !_SYS_VNODE_H_ */
1163