1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1989, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 */
31
32 #ifndef _SYS_VNODE_H_
33 #define _SYS_VNODE_H_
34
35 #include <sys/bufobj.h>
36 #include <sys/queue.h>
37 #include <sys/lock.h>
38 #include <sys/lockmgr.h>
39 #include <sys/mutex.h>
40 #include <sys/rangelock.h>
41 #include <sys/selinfo.h>
42 #include <sys/uio.h>
43 #include <sys/acl.h>
44 #include <sys/ktr.h>
45 #include <sys/_seqc.h>
46
47 /*
48 * The vnode is the focus of all file activity in UNIX. There is a
49 * unique vnode allocated for each active file, each current directory,
50 * each mounted-on file, text file, and the root.
51 */
52
53 /*
54 * Vnode types. VNON means no type.
55 */
__enum_uint8_decl(vtype)56 __enum_uint8_decl(vtype) {
57 VNON,
58 VREG,
59 VDIR,
60 VBLK,
61 VCHR,
62 VLNK,
63 VSOCK,
64 VFIFO,
65 VBAD,
66 VMARKER,
67 VLASTTYPE = VMARKER,
68 };
69
70 /*
71 * We frequently need to test is something is a device node.
72 */
73 #define VTYPE_ISDEV(vtype) ((vtype) == VCHR || (vtype) == VBLK)
74
__enum_uint8_decl(vstate)75 __enum_uint8_decl(vstate) {
76 VSTATE_UNINITIALIZED,
77 VSTATE_CONSTRUCTED,
78 VSTATE_DESTROYING,
79 VSTATE_DEAD,
80 VLASTSTATE = VSTATE_DEAD,
81 };
82
83 enum vgetstate {
84 VGET_NONE,
85 VGET_HOLDCNT,
86 VGET_USECOUNT,
87 };
88
89 /*
90 * Each underlying filesystem allocates its own private area and hangs
91 * it from v_data. If non-null, this area is freed in getnewvnode().
92 */
93
94 struct cache_fpl;
95 struct inotify_watch;
96 struct namecache;
97
98 struct vpollinfo {
99 struct mtx vpi_lock; /* lock to protect below */
100 TAILQ_HEAD(, inotify_watch) vpi_inotify; /* list of inotify watchers */
101 struct selinfo vpi_selinfo; /* identity of poller(s) */
102 short vpi_events; /* what they are looking for */
103 short vpi_revents; /* what has happened */
104 };
105
106 /*
107 * Reading or writing any of these items requires holding the appropriate lock.
108 *
109 * Lock reference:
110 * c - namecache mutex
111 * i - interlock
112 * l - mp mnt_listmtx or freelist mutex
113 * I - updated with atomics, 0->1 and 1->0 transitions with interlock held
114 * m - mount point interlock
115 * p - pollinfo lock
116 * u - Only a reference to the vnode is needed to read.
117 * v - vnode lock
118 *
119 * Vnodes may be found on many lists. The general way to deal with operating
120 * on a vnode that is on a list is:
121 * 1) Lock the list and find the vnode.
122 * 2) Lock interlock so that the vnode does not go away.
123 * 3) Unlock the list to avoid lock order reversals.
124 * 4) vget with LK_INTERLOCK and check for ENOENT, or
125 * 5) Check for DOOMED if the vnode lock is not required.
126 * 6) Perform your operation, then vput().
127 */
128
129 #if defined(_KERNEL) || defined(_KVM_VNODE)
130
131 struct vnode {
132 /*
133 * Fields which define the identity of the vnode. These fields are
134 * owned by the filesystem (XXX: and vgone() ?)
135 */
136 short v_v2flag; /* v frequently read flag */
137 short v_irflag; /* i frequently read flags */
138 seqc_t v_seqc; /* i modification count */
139 uint32_t v_nchash; /* u namecache hash */
140 u_int v_hash;
141 const struct vop_vector *v_op; /* u vnode operations vector */
142 void *v_data; /* u private data for fs */
143
144 /*
145 * Filesystem instance stuff
146 */
147 struct mount *v_mount; /* u ptr to vfs we are in */
148 TAILQ_ENTRY(vnode) v_nmntvnodes; /* m vnodes for mount point */
149
150 /*
151 * Type specific fields, only one applies to any given vnode.
152 */
153 union {
154 struct mount *v_mountedhere; /* v ptr to mountpoint (VDIR) */
155 struct unpcb *v_unpcb; /* v unix domain net (VSOCK) */
156 struct cdev *v_rdev; /* v device (VCHR, VBLK) */
157 struct fifoinfo *v_fifoinfo; /* v fifo (VFIFO) */
158 };
159
160 /*
161 * vfs_hash: (mount + inode) -> vnode hash. The hash value
162 * itself is grouped with other int fields, to avoid padding.
163 */
164 LIST_ENTRY(vnode) v_hashlist;
165
166 /*
167 * VFS_namecache stuff
168 */
169 LIST_HEAD(, namecache) v_cache_src; /* c Cache entries from us */
170 TAILQ_HEAD(, namecache) v_cache_dst; /* c Cache entries to us */
171 struct namecache *v_cache_dd; /* c Cache entry for .. vnode */
172
173 /*
174 * Locking
175 */
176 struct lock v_lock; /* u (if fs don't have one) */
177 struct mtx v_interlock; /* lock for "i" things */
178 struct lock *v_vnlock; /* u pointer to vnode lock */
179
180 /*
181 * The machinery of being a vnode
182 */
183 TAILQ_ENTRY(vnode) v_vnodelist; /* l vnode lists */
184 TAILQ_ENTRY(vnode) v_lazylist; /* l vnode lazy list */
185 struct bufobj v_bufobj; /* * Buffer cache object */
186
187 /*
188 * Hooks for various subsystems and features.
189 */
190 struct vpollinfo *v_pollinfo; /* i Poll events, p for *v_pi */
191 struct label *v_label; /* MAC label for vnode */
192 struct lockf *v_lockf; /* Byte-level advisory lock list */
193 struct rangelock v_rl; /* Byte-range lock */
194
195 u_int v_holdcnt; /* I prevents recycling. */
196 u_int v_usecount; /* I ref count of users */
197 u_short v_iflag; /* i vnode flags (see below) */
198 u_short v_vflag; /* v vnode flags */
199 u_short v_mflag; /* l mnt-specific vnode flags */
200 short v_dbatchcpu; /* i LRU requeue deferral batch */
201 int v_writecount; /* I ref count of writers or
202 (negative) text users */
203 int v_seqc_users; /* i modifications pending */
204 };
205 /*
206 __enum_uint8(vtype) v_type;
207 __enum_uint8(vstate) v_state;
208 */
209 #define v_type v_rl.resv1 /* u vnode type */
210 #define v_state v_rl.resv2 /* u vnode state */
211
212 #define VN_ISDEV(vp) VTYPE_ISDEV((vp)->v_type)
213
214 #ifndef DEBUG_LOCKS
215 #ifdef _LP64
216 /*
217 * Not crossing 448 bytes fits 9 vnodes per page. If you have to add fields
218 * to the structure and there is nothing which can be done to prevent growth
219 * then so be it. But don't grow it without a good reason.
220 */
221 _Static_assert(sizeof(struct vnode) <= 448, "vnode size crosses 448 bytes");
222 #endif
223 #endif
224
225 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
226
227 #define bo2vnode(bo) __containerof((bo), struct vnode, v_bufobj)
228
229 #define v_object v_bufobj.bo_object
230
231 #define VN_KNOTE(vp, b, a) do { \
232 if ((vp->v_v2flag & V2_KNOTE) != 0) { \
233 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
234 (a) | KNF_NOKQLOCK); \
235 } \
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 * V2 flags are protected by the vnode lock and live in v_v2flag
246 *
247 * VIRF_DOOMED is doubly protected by the interlock and vnode lock. Both
248 * are required for writing but the status may be checked with either.
249 */
250 #define VHOLD_NO_SMR (1<<29) /* Disable vhold_smr */
251 #define VHOLD_ALL_FLAGS (VHOLD_NO_SMR)
252
253 #define VIRF_DOOMED 0x0001 /* This vnode is being recycled */
254 #define VIRF_PGREAD 0x0002 /* Direct reads from the page cache are permitted,
255 never cleared once set */
256 #define VIRF_MOUNTPOINT 0x0004 /* This vnode is mounted on */
257 #define VIRF_TEXT_REF 0x0008 /* Executable mappings ref the vnode */
258 #define VIRF_CROSSMP 0x0010 /* Cross-mp vnode, no locking */
259 #define VIRF_NAMEDDIR 0x0020 /* Named attribute directory */
260 #define VIRF_NAMEDATTR 0x0040 /* Named attribute */
261 #define VIRF_INOTIFY 0x0080 /* This vnode is being watched */
262 #define VIRF_INOTIFY_PARENT 0x0100 /* A parent of this vnode may be being
263 watched */
264
265 #define V2_KNOTE 0x0001 /* Has knlist */
266
267 #define VI_UNUSED0 0x0001 /* unused */
268 #define VI_MOUNT 0x0002 /* Mount in progress */
269 #define VI_DOINGINACT 0x0004 /* VOP_INACTIVE is in progress */
270 #define VI_OWEINACT 0x0008 /* Need to call inactive */
271 #define VI_DEFINACT 0x0010 /* deferred inactive */
272 #define VI_FOPENING 0x0020 /* In open, with opening process having the
273 first right to advlock file */
274 #define VI_DELAYED_SETSIZE 0x0040 /* Delayed setsize */
275
276 #define VV_ROOT 0x0001 /* root of its filesystem */
277 #define VV_ISTTY 0x0002 /* vnode represents a tty */
278 #define VV_NOSYNC 0x0004 /* unlinked, stop syncing */
279 #define VV_ETERNALDEV 0x0008 /* device that is never destroyed */
280 #define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */
281 #define VV_VMSIZEVNLOCK 0x0020 /* object size check requires vnode lock */
282 #define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */
283 #define VV_SYSTEM 0x0080 /* vnode being used by kernel */
284 #define VV_PROCDEP 0x0100 /* vnode is process dependent */
285 #define VV_UNLINKED 0x0200 /* unlinked but stil open directory */
286 #define VV_DELETED 0x0400 /* should be removed */
287 #define VV_MD 0x0800 /* vnode backs the md device */
288 #define VV_FORCEINSMQ 0x1000 /* force the insmntque to succeed */
289 #define VV_READLINK 0x2000 /* fdescfs linux vnode */
290 #define VV_UNREF 0x4000 /* vunref, do not drop lock in inactive() */
291 #define VV_CROSSLOCK 0x8000 /* vnode lock is shared w/ root mounted here */
292
293 #define VMP_LAZYLIST 0x0001 /* Vnode is on mnt's lazy list */
294
295 /*
296 * Vnode attributes. A field value of VNOVAL represents a field whose value
297 * is unavailable (getattr) or which is not to be changed (setattr).
298 */
299 struct vattr {
300 __enum_uint8(vtype) va_type; /* vnode type (for create) */
301 u_short va_mode; /* files access mode and type */
302 uint16_t va_bsdflags; /* same as st_bsdflags from stat(2) */
303 uid_t va_uid; /* owner user id */
304 gid_t va_gid; /* owner group id */
305 nlink_t va_nlink; /* number of references to file */
306 dev_t va_fsid; /* filesystem id */
307 ino_t va_fileid; /* file id */
308 u_quad_t va_size; /* file size in bytes */
309 long va_blocksize; /* blocksize preferred for i/o */
310 struct timespec va_atime; /* time of last access */
311 struct timespec va_mtime; /* time of last modification */
312 struct timespec va_ctime; /* time file changed */
313 struct timespec va_birthtime; /* time file created */
314 u_long va_gen; /* generation number of file */
315 u_long va_flags; /* flags defined for file */
316 dev_t va_rdev; /* device the special file represents */
317 u_quad_t va_bytes; /* bytes of disk space held by file */
318 u_quad_t va_filerev; /* file modification number */
319 u_int va_vaflags; /* operations flags, see below */
320 long va_spare; /* remain quad aligned */
321 };
322
323 #define VATTR_ISDEV(vap) VTYPE_ISDEV((vap)->va_type)
324
325 /*
326 * Flags for va_vaflags.
327 */
328 #define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */
329 #define VA_EXCLUSIVE 0x02 /* exclusive create request */
330 #define VA_SYNC 0x04 /* O_SYNC truncation */
331
332 /*
333 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
334 * help with write clustering)
335 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
336 */
337 #define IO_UNIT 0x0001 /* do I/O as atomic unit */
338 #define IO_APPEND 0x0002 /* append write to end */
339 #define IO_NDELAY 0x0004 /* FNDELAY flag set in file table */
340 #define IO_NODELOCKED 0x0008 /* underlying node already locked */
341 #define IO_ASYNC 0x0010 /* bawrite rather then bdwrite */
342 #define IO_VMIO 0x0020 /* data already in VMIO space */
343 #define IO_INVAL 0x0040 /* invalidate after I/O */
344 #define IO_SYNC 0x0080 /* do I/O synchronously */
345 #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */
346 #define IO_NOREUSE 0x0200 /* VMIO data won't be reused */
347 #define IO_EXT 0x0400 /* operate on external attributes */
348 #define IO_NORMAL 0x0800 /* operate on regular data */
349 #define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */
350 #define IO_BUFLOCKED 0x2000 /* ffs flag; indir buf is locked */
351 #define IO_RANGELOCKED 0x4000 /* range locked */
352 #define IO_DATASYNC 0x8000 /* do only data I/O synchronously */
353
354 #define IO_SEQMAX 0x7F /* seq heuristic max value */
355 #define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */
356
357 /*
358 * Flags for accmode_t.
359 */
360 #define VEXEC 000000000100 /* execute/search permission */
361 #define VWRITE 000000000200 /* write permission */
362 #define VREAD 000000000400 /* read permission */
363 #define VADMIN 000000010000 /* being the file owner */
364 #define VAPPEND 000000040000 /* permission to write/append */
365 /*
366 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
367 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
368 * and 0 otherwise. This never happens with ordinary unix access rights
369 * or POSIX.1e ACLs. Obviously, VEXPLICIT_DENY must be OR-ed with
370 * some other V* constant.
371 */
372 #define VEXPLICIT_DENY 000000100000
373 #define VREAD_NAMED_ATTRS 000000200000 /* not used */
374 #define VWRITE_NAMED_ATTRS 000000400000 /* not used */
375 #define VDELETE_CHILD 000001000000
376 #define VREAD_ATTRIBUTES 000002000000 /* permission to stat(2) */
377 #define VWRITE_ATTRIBUTES 000004000000 /* change {m,c,a}time */
378 #define VDELETE 000010000000
379 #define VREAD_ACL 000020000000 /* read ACL and file mode */
380 #define VWRITE_ACL 000040000000 /* change ACL and/or file mode */
381 #define VWRITE_OWNER 000100000000 /* change file owner */
382 #define VSYNCHRONIZE 000200000000 /* not used */
383 #define VCREAT 000400000000 /* creating new file */
384 #define VVERIFY 001000000000 /* verification required */
385
386 /*
387 * Permissions that were traditionally granted only to the file owner.
388 */
389 #define VADMIN_PERMS (VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
390 VWRITE_OWNER)
391
392 /*
393 * Permissions that were traditionally granted to everyone.
394 */
395 #define VSTAT_PERMS (VREAD_ATTRIBUTES | VREAD_ACL)
396
397 /*
398 * Permissions that allow to change the state of the file in any way.
399 */
400 #define VMODIFY_PERMS (VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
401 VDELETE)
402
403 /*
404 * Token indicating no attribute value yet assigned.
405 */
406 #define VNOVAL (-1)
407
408 /*
409 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
410 */
411 #define VLKTIMEOUT (hz / 20 + 1)
412
413 #ifdef _KERNEL
414
415 #ifdef MALLOC_DECLARE
416 MALLOC_DECLARE(M_VNODE);
417 #endif
418
419 extern u_int ncsizefactor;
420 extern const u_int io_hold_cnt;
421
422 /*
423 * Convert between vnode types and inode formats (since POSIX.1
424 * defines mode word of stat structure in terms of inode formats).
425 */
426 extern __enum_uint8(vtype) iftovt_tab[];
427 extern int vttoif_tab[];
428 #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12])
429 #define VTTOIF(indx) (vttoif_tab[(int)(indx)])
430 #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode))
431
432 /*
433 * Flags to various vnode functions.
434 */
435 #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */
436 #define FORCECLOSE 0x0002 /* vflush: force file closure */
437 #define WRITECLOSE 0x0004 /* vflush: only close writable files */
438 #define EARLYFLUSH 0x0008 /* vflush: early call for ffs_flushfiles */
439 #define V_SAVE 0x0001 /* vinvalbuf: sync file first */
440 #define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */
441 #define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */
442 #define V_CLEANONLY 0x0008 /* vinvalbuf: invalidate only clean bufs */
443 #define V_VMIO 0x0010 /* vinvalbuf: called during pageout */
444 #define V_ALLOWCLEAN 0x0020 /* vinvalbuf: allow clean buffers after flush */
445 #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */
446 #define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */
447 #define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */
448 #define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */
449 #define V_PCATCH 0x0008 /* vn_start_write: make the sleep interruptible */
450 #define V_VALID_FLAGS (V_WAIT | V_NOWAIT | V_XSLEEP | V_PCATCH)
451
452 #define VR_START_WRITE 0x0001 /* vfs_write_resume: start write atomically */
453 #define VR_NO_SUSPCLR 0x0002 /* vfs_write_resume: do not clear suspension */
454
455 #define VS_SKIP_UNMOUNT 0x0001 /* vfs_write_suspend: fail if the
456 filesystem is being unmounted */
457
458 #ifdef DIAGNOSTIC
459 #define VATTR_NULL(vap) vattr_null(vap)
460 #else
461 #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */
462 #endif /* DIAGNOSTIC */
463
464 /*
465 * Global vnode data.
466 */
467 extern struct vnode *rootvnode; /* root (i.e. "/") vnode */
468 extern struct mount *rootdevmp; /* "/dev" mount */
469 extern u_long desiredvnodes; /* number of vnodes desired */
470 extern struct uma_zone *namei_zone;
471 extern struct vattr va_null; /* predefined null vattr structure */
472 extern struct vnode *vp_crossmp;
473
474 extern u_int vn_lock_pair_pause_max;
475
476 #define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock)
477 #define VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
478 #define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock)
479 #define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock)
480 #define VI_MTX(vp) (&(vp)->v_interlock)
481
482 #define VN_LOCK_AREC(vp) lockallowrecurse((vp)->v_vnlock)
483 #define VN_LOCK_DREC(vp) lockdisablerecurse((vp)->v_vnlock)
484 #define VN_LOCK_CANREC(vp) lockcanrecurse((vp)->v_vnlock)
485 #define VN_LOCK_ASHARE(vp) lockallowshare((vp)->v_vnlock)
486 #define VN_LOCK_DSHARE(vp) lockdisableshare((vp)->v_vnlock)
487
488 #endif /* _KERNEL */
489
490 /*
491 * Mods for extensibility.
492 */
493
494 /*
495 * Flags for vdesc_flags:
496 */
497 #define VDESC_MAX_VPS 16
498 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
499 #define VDESC_VP0_WILLRELE 0x0001
500 #define VDESC_VP1_WILLRELE 0x0002
501 #define VDESC_VP2_WILLRELE 0x0004
502 #define VDESC_VP3_WILLRELE 0x0008
503
504 /*
505 * A generic structure.
506 * This can be used by bypass routines to identify generic arguments.
507 */
508 struct vop_generic_args {
509 struct vnodeop_desc *a_desc;
510 /* other random data follows, presumably */
511 };
512
513 typedef int vop_bypass_t(struct vop_generic_args *);
514
515 /*
516 * VDESC_NO_OFFSET is used to identify the end of the offset list
517 * and in places where no such field exists.
518 */
519 #define VDESC_NO_OFFSET -1
520
521 /*
522 * This structure describes the vnode operation taking place.
523 */
524 struct vnodeop_desc {
525 char *vdesc_name; /* a readable name for debugging */
526 int vdesc_flags; /* VDESC_* flags */
527 int vdesc_vop_offset;
528 vop_bypass_t *vdesc_call; /* Function to call */
529
530 /*
531 * These ops are used by bypass routines to map and locate arguments.
532 * Creds and procs are not needed in bypass routines, but sometimes
533 * they are useful to (for example) transport layers.
534 * Nameidata is useful because it has a cred in it.
535 */
536 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */
537 int vdesc_vpp_offset; /* return vpp location */
538 int vdesc_cred_offset; /* cred location, if any */
539 int vdesc_thread_offset; /* thread location, if any */
540 int vdesc_componentname_offset; /* if any */
541 };
542
543 #ifdef _KERNEL
544 /*
545 * A list of all the operation descs.
546 */
547 extern struct vnodeop_desc *vnodeop_descs[];
548
549 #define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field)
550 #define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
551 ((s_type)(((char*)(struct_p)) + (s_offset)))
552
553 #ifdef INVARIANTS
554 /*
555 * Support code to aid in debugging VFS locking problems. Not totally
556 * reliable since if the thread sleeps between changing the lock
557 * state and checking it with the assert, some other thread could
558 * change the state. They are good enough for debugging a single
559 * filesystem using a single-threaded test. Note that the unreliability is
560 * limited to false negatives; efforts were made to ensure that false
561 * positives cannot occur.
562 */
563 void assert_vi_locked(struct vnode *vp, const char *str);
564 void assert_vi_unlocked(struct vnode *vp, const char *str);
565 void assert_vop_elocked(struct vnode *vp, const char *str);
566 void assert_vop_locked(struct vnode *vp, const char *str);
567 void assert_vop_unlocked(struct vnode *vp, const char *str);
568
569 #define ASSERT_VI_LOCKED(vp, str) assert_vi_locked((vp), (str))
570 #define ASSERT_VI_UNLOCKED(vp, str) assert_vi_unlocked((vp), (str))
571 #define ASSERT_VOP_ELOCKED(vp, str) assert_vop_elocked((vp), (str))
572 #define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked((vp), (str))
573 #define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked((vp), (str))
574
575 #define ASSERT_VOP_IN_SEQC(vp) do { \
576 struct vnode *_vp = (vp); \
577 \
578 VNPASS(seqc_in_modify(_vp->v_seqc), _vp); \
579 } while (0)
580
581 #define ASSERT_VOP_NOT_IN_SEQC(vp) do { \
582 struct vnode *_vp = (vp); \
583 \
584 VNPASS(!seqc_in_modify(_vp->v_seqc), _vp); \
585 } while (0)
586
587 #else /* !INVARIANTS */
588
589 #define ASSERT_VI_LOCKED(vp, str) ((void)0)
590 #define ASSERT_VI_UNLOCKED(vp, str) ((void)0)
591 #define ASSERT_VOP_ELOCKED(vp, str) ((void)0)
592 #define ASSERT_VOP_LOCKED(vp, str) ((void)0)
593 #define ASSERT_VOP_UNLOCKED(vp, str) ((void)0)
594
595 #define ASSERT_VOP_IN_SEQC(vp) ((void)0)
596 #define ASSERT_VOP_NOT_IN_SEQC(vp) ((void)0)
597
598 #endif /* INVARIANTS */
599
600 #define DOINGASYNC(vp) \
601 (((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 && \
602 ((curthread->td_pflags & TDP_SYNCIO) == 0))
603
604 /*
605 * VMIO support inline
606 */
607
608 extern int vmiodirenable;
609
610 static __inline int
vn_canvmio(struct vnode * vp)611 vn_canvmio(struct vnode *vp)
612 {
613 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
614 return(TRUE);
615 return(FALSE);
616 }
617
618 /*
619 * Finally, include the default set of vnode operations.
620 */
621 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
622 #include "vnode_if.h"
623
624 /* vn_open_flags */
625 #define VN_OPEN_NOAUDIT 0x00000001
626 #define VN_OPEN_NOCAPCHECK 0x00000002
627 #define VN_OPEN_NAMECACHE 0x00000004
628 #define VN_OPEN_INVFS 0x00000008
629 #define VN_OPEN_WANTIOCTLCAPS 0x00000010
630
631 /* copy_file_range kernel flags */
632 #define COPY_FILE_RANGE_KFLAGS 0xff000000
633 #define COPY_FILE_RANGE_TIMEO1SEC 0x01000000 /* Return after 1sec. */
634
635 /*
636 * Public vnode manipulation functions.
637 */
638 struct componentname;
639 struct file;
640 struct mount;
641 struct nameidata;
642 struct ostat;
643 struct freebsd11_stat;
644 struct thread;
645 struct proc;
646 struct stat;
647 struct nstat;
648 struct ucred;
649 struct uio;
650 struct vattr;
651 struct vfsops;
652 struct vnode;
653
654 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
655
656 int bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
657 daddr_t endn);
658 /* cache_* may belong in namei.h. */
659 void cache_changesize(u_long newhashsize);
660
661 #define VFS_CACHE_DROPOLD 0x1
662
663 void cache_enter_time_flags(struct vnode *dvp, struct vnode *vp,
664 struct componentname *cnp, struct timespec *tsp,
665 struct timespec *dtsp, int flags);
666 #define cache_enter(dvp, vp, cnp) \
667 cache_enter_time(dvp, vp, cnp, NULL, NULL)
668 void cache_enter_time(struct vnode *dvp, struct vnode *vp,
669 struct componentname *cnp, struct timespec *tsp,
670 struct timespec *dtsp);
671 int cache_lookup(struct vnode *dvp, struct vnode **vpp,
672 struct componentname *cnp, struct timespec *tsp, int *ticksp);
673 void cache_vnode_init(struct vnode *vp);
674 void cache_purge(struct vnode *vp);
675 void cache_purge_vgone(struct vnode *vp);
676 void cache_purge_negative(struct vnode *vp);
677 void cache_purgevfs(struct mount *mp);
678 char *cache_symlink_alloc(size_t size, int flags);
679 void cache_symlink_free(char *string, size_t size);
680 int cache_symlink_resolve(struct cache_fpl *fpl, const char *string,
681 size_t len);
682 void cache_vop_inotify(struct vnode *vp, int event, uint32_t cookie);
683 void cache_vop_rename(struct vnode *fdvp, struct vnode *fvp, struct vnode *tdvp,
684 struct vnode *tvp, struct componentname *fcnp, struct componentname *tcnp);
685 void cache_vop_rmdir(struct vnode *dvp, struct vnode *vp);
686 void cache_vop_vector_register(struct vop_vector *);
687 #ifdef INVARIANTS
688 void cache_validate(struct vnode *dvp, struct vnode *vp,
689 struct componentname *cnp);
690 void cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops);
691 void cache_assert_no_entries(struct vnode *vp);
692 #else
693 static inline void
cache_validate(struct vnode * dvp,struct vnode * vp,struct componentname * cnp)694 cache_validate(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
695 {
696 }
697
698 static inline void
cache_validate_vop_vector(struct mount * mp,struct vop_vector * vops)699 cache_validate_vop_vector(struct mount *mp, struct vop_vector *vops)
700 {
701 }
702
703 static inline void
cache_assert_no_entries(struct vnode * vp)704 cache_assert_no_entries(struct vnode *vp)
705 {
706 }
707 #endif
708 void cache_fast_lookup_enabled_recalc(void);
709 int change_dir(struct vnode *vp, struct thread *td);
710 void cvtstat(struct stat *st, struct ostat *ost);
711 int freebsd11_cvtnstat(struct stat *sb, struct nstat *nsb);
712 int freebsd11_cvtstat(struct stat *st, struct freebsd11_stat *ost);
713 int getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
714 struct vnode **vpp);
715 void getnewvnode_reserve(void);
716 void getnewvnode_drop_reserve(void);
717 int insmntque(struct vnode *vp, struct mount *mp);
718 int insmntque1(struct vnode *vp, struct mount *mp);
719 u_quad_t init_va_filerev(void);
720 int speedup_syncer(void);
721 int vn_vptocnp(struct vnode **vp, char *buf, size_t *buflen);
722 int vn_getcwd(char *buf, char **retbuf, size_t *buflen);
723 int vn_fullpath(struct vnode *vp, char **retbuf, char **freebuf);
724 int vn_fullpath_jail(struct vnode *vp, char **retbuf, char **freebuf);
725 int vn_fullpath_global(struct vnode *vp, char **retbuf, char **freebuf);
726 int vn_fullpath_hardlink(struct vnode *vp, struct vnode *dvp,
727 const char *hdrl_name, size_t hrdl_name_length, char **retbuf,
728 char **freebuf, size_t *buflen);
729 struct vnode *
730 vn_dir_dd_ino(struct vnode *vp);
731 int vn_commname(struct vnode *vn, char *buf, u_int buflen);
732 int vn_path_to_global_path(struct thread *td, struct vnode *vp,
733 char *path, u_int pathlen);
734 int vn_path_to_global_path_hardlink(struct thread *td, struct vnode *vp,
735 struct vnode *dvp, char *path, u_int pathlen, const char *leaf_name,
736 size_t leaf_length);
737 int vaccess(__enum_uint8(vtype) type, mode_t file_mode, uid_t file_uid,
738 gid_t file_gid, accmode_t accmode, struct ucred *cred);
739 int vaccess_vexec_smr(mode_t file_mode, uid_t file_uid, gid_t file_gid,
740 struct ucred *cred);
741 int vaccess_acl_nfs4(__enum_uint8(vtype) type, uid_t file_uid, gid_t file_gid,
742 struct acl *aclp, accmode_t accmode, struct ucred *cred);
743 int vaccess_acl_posix1e(__enum_uint8(vtype) type, uid_t file_uid,
744 gid_t file_gid, struct acl *acl, accmode_t accmode,
745 struct ucred *cred);
746 void vattr_null(struct vattr *vap);
747 void vlazy(struct vnode *);
748 void vdrop(struct vnode *);
749 void vdropl(struct vnode *);
750 int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
751 int vget(struct vnode *vp, int flags);
752 enum vgetstate vget_prep_smr(struct vnode *vp);
753 enum vgetstate vget_prep(struct vnode *vp);
754 int vget_finish(struct vnode *vp, int flags, enum vgetstate vs);
755 void vget_finish_ref(struct vnode *vp, enum vgetstate vs);
756 void vget_abort(struct vnode *vp, enum vgetstate vs);
757 void vgone(struct vnode *vp);
758 void vhold(struct vnode *);
759 void vholdnz(struct vnode *);
760 bool vhold_smr(struct vnode *);
761 int vinactive(struct vnode *vp);
762 int vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
763 int vtruncbuf(struct vnode *vp, off_t length, int blksize);
764 void v_inval_buf_range(struct vnode *vp, daddr_t startlbn, daddr_t endlbn,
765 int blksize);
766 void vunref(struct vnode *);
767 void vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
768 int vrecycle(struct vnode *vp);
769 int vrecyclel(struct vnode *vp);
770 int vn_bmap_seekhole_locked(struct vnode *vp, u_long cmd, off_t *off,
771 struct ucred *cred);
772 int vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
773 struct ucred *cred);
774 int vn_close(struct vnode *vp,
775 int flags, struct ucred *file_cred, struct thread *td);
776 int vn_copy_file_range(struct vnode *invp, off_t *inoffp,
777 struct vnode *outvp, off_t *outoffp, size_t *lenp,
778 unsigned int flags, struct ucred *incred, struct ucred *outcred,
779 struct thread *fsize_td);
780 int vn_deallocate(struct vnode *vp, off_t *offset, off_t *length, int flags,
781 int ioflg, struct ucred *active_cred, struct ucred *file_cred);
782 void vn_finished_write(struct mount *mp);
783 void vn_finished_secondary_write(struct mount *mp);
784 int vn_fsync_buf(struct vnode *vp, int waitfor);
785 int vn_generic_copy_file_range(struct vnode *invp, off_t *inoffp,
786 struct vnode *outvp, off_t *outoffp, size_t *lenp,
787 unsigned int flags, struct ucred *incred, struct ucred *outcred,
788 struct thread *fsize_td);
789 int vn_need_pageq_flush(struct vnode *vp);
790 bool vn_isdisk_error(struct vnode *vp, int *errp);
791 bool vn_isdisk(struct vnode *vp);
792 int _vn_lock(struct vnode *vp, int flags, const char *file, int line);
793 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
794 int vn_lock_pair(struct vnode *vp1, bool vp1_locked, int lkflags1,
795 struct vnode *vp2, bool vp2_locked, int lkflags2);
796 int vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
797 int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
798 u_int vn_open_flags, struct ucred *cred, struct file *fp);
799 int vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
800 struct thread *td, struct file *fp);
801 void vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
802 void vn_pages_remove_valid(struct vnode *vp, vm_pindex_t start,
803 vm_pindex_t end);
804 int vn_pollrecord(struct vnode *vp, struct thread *p, int events);
805 int vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
806 int len, off_t offset, enum uio_seg segflg, int ioflg,
807 struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
808 struct thread *td);
809 int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
810 size_t len, off_t offset, enum uio_seg segflg, int ioflg,
811 struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
812 struct thread *td);
813 int vn_read_from_obj(struct vnode *vp, struct uio *uio);
814 int vn_rlimit_fsize(const struct vnode *vp, const struct uio *uio,
815 struct thread *td);
816 int vn_rlimit_fsizex(const struct vnode *vp, struct uio *uio,
817 off_t maxfsz, ssize_t *resid_adj, struct thread *td);
818 void vn_rlimit_fsizex_res(struct uio *uio, ssize_t resid_adj);
819 int vn_rlimit_trunc(u_quad_t size, struct thread *td);
820 int vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
821 int vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
822 int flags);
823 int vn_truncate_locked(struct vnode *vp, off_t length, bool sync,
824 struct ucred *cred);
825 int vn_writechk(struct vnode *vp);
826 int vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
827 const char *attrname, int *buflen, char *buf, struct thread *td);
828 int vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
829 const char *attrname, int buflen, char *buf, struct thread *td);
830 int vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
831 const char *attrname, struct thread *td);
832 int vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
833 struct vnode **rvp);
834 int vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
835 void *alloc_arg, int lkflags, struct vnode **rvp);
836 int vn_utimes_perm(struct vnode *vp, struct vattr *vap,
837 struct ucred *cred, struct thread *td);
838 int vn_cmp(struct file *, struct file *, struct thread *td);
839
840 int vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
841 int vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
842 struct uio *uio);
843
844 void vn_seqc_write_begin_locked(struct vnode *vp);
845 void vn_seqc_write_begin(struct vnode *vp);
846 void vn_seqc_write_end_locked(struct vnode *vp);
847 void vn_seqc_write_end(struct vnode *vp);
848 #define vn_seqc_read_any(vp) seqc_read_any(&(vp)->v_seqc)
849 #define vn_seqc_read_notmodify(vp) seqc_read_notmodify(&(vp)->v_seqc)
850 #define vn_seqc_consistent(vp, seq) seqc_consistent(&(vp)->v_seqc, seq)
851
852 #define vn_rangelock_unlock(vp, cookie) \
853 rangelock_unlock(&(vp)->v_rl, (cookie))
854 #define vn_rangelock_rlock(vp, start, end) \
855 rangelock_rlock(&(vp)->v_rl, (start), (end))
856 #define vn_rangelock_tryrlock(vp, start, end) \
857 rangelock_tryrlock(&(vp)->v_rl, (start), (end))
858 #define vn_rangelock_wlock(vp, start, end) \
859 rangelock_wlock(&(vp)->v_rl, (start), (end))
860 #define vn_rangelock_trywlock(vp, start, end) \
861 rangelock_trywlock(&(vp)->v_rl, (start), (end))
862
863 #define vn_irflag_read(vp) atomic_load_short(&(vp)->v_irflag)
864 void vn_irflag_set_locked(struct vnode *vp, short toset);
865 void vn_irflag_set(struct vnode *vp, short toset);
866 void vn_irflag_set_cond_locked(struct vnode *vp, short toset);
867 void vn_irflag_set_cond(struct vnode *vp, short toset);
868 void vn_irflag_unset_locked(struct vnode *vp, short tounset);
869 void vn_irflag_unset(struct vnode *vp, short tounset);
870
871 int vfs_cache_lookup(struct vop_lookup_args *ap);
872 int vfs_cache_root(struct mount *mp, int flags, struct vnode **vpp);
873 void vfs_timestamp(struct timespec *);
874 void vfs_write_resume(struct mount *mp, int flags);
875 int vfs_write_suspend(struct mount *mp, int flags);
876 int vfs_write_suspend_umnt(struct mount *mp);
877 struct vnode *vnlru_alloc_marker(void);
878 void vnlru_free_marker(struct vnode *);
879 void vnlru_free_vfsops(int, struct vfsops *, struct vnode *);
880 int vop_stdbmap(struct vop_bmap_args *);
881 int vop_stdfdatasync_buf(struct vop_fdatasync_args *);
882 int vop_stdfsync(struct vop_fsync_args *);
883 int vop_stdgetwritemount(struct vop_getwritemount_args *);
884 int vop_stdgetpages(struct vop_getpages_args *);
885 int vop_stdinactive(struct vop_inactive_args *);
886 int vop_stdneed_inactive(struct vop_need_inactive_args *);
887 int vop_stdinotify(struct vop_inotify_args *);
888 int vop_stdinotify_add_watch(struct vop_inotify_add_watch_args *);
889 int vop_stdioctl(struct vop_ioctl_args *);
890 int vop_stdkqfilter(struct vop_kqfilter_args *);
891 int vop_stdlock(struct vop_lock1_args *);
892 int vop_stdunlock(struct vop_unlock_args *);
893 int vop_stdislocked(struct vop_islocked_args *);
894 int vop_lock(struct vop_lock1_args *);
895 int vop_unlock(struct vop_unlock_args *);
896 int vop_islocked(struct vop_islocked_args *);
897 int vop_stdputpages(struct vop_putpages_args *);
898 int vop_nopoll(struct vop_poll_args *);
899 int vop_stdaccess(struct vop_access_args *ap);
900 int vop_stdaccessx(struct vop_accessx_args *ap);
901 int vop_stdadvise(struct vop_advise_args *ap);
902 int vop_stdadvlock(struct vop_advlock_args *ap);
903 int vop_stdadvlockasync(struct vop_advlockasync_args *ap);
904 int vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
905 int vop_stdallocate(struct vop_allocate_args *ap);
906 int vop_stddeallocate(struct vop_deallocate_args *ap);
907 int vop_stdset_text(struct vop_set_text_args *ap);
908 int vop_stdpathconf(struct vop_pathconf_args *);
909 int vop_stdpoll(struct vop_poll_args *);
910 int vop_stdvptocnp(struct vop_vptocnp_args *ap);
911 int vop_stdvptofh(struct vop_vptofh_args *ap);
912 int vop_stdvput_pair(struct vop_vput_pair_args *ap);
913 int vop_stdunp_bind(struct vop_unp_bind_args *ap);
914 int vop_stdunp_connect(struct vop_unp_connect_args *ap);
915 int vop_stdunp_detach(struct vop_unp_detach_args *ap);
916 int vop_stdadd_writecount_nomsync(struct vop_add_writecount_args *ap);
917 int vop_stdcopy_file_range(struct vop_copy_file_range_args *ap);
918 int vop_eopnotsupp(struct vop_generic_args *ap);
919 int vop_ebadf(struct vop_generic_args *ap);
920 int vop_einval(struct vop_generic_args *ap);
921 int vop_enoent(struct vop_generic_args *ap);
922 int vop_enotty(struct vop_generic_args *ap);
923 int vop_eagain(struct vop_generic_args *ap);
924 int vop_null(struct vop_generic_args *ap);
925 int vop_panic(struct vop_generic_args *ap);
926 int dead_poll(struct vop_poll_args *ap);
927 int dead_read(struct vop_read_args *ap);
928 int dead_write(struct vop_write_args *ap);
929
930 /* These are called from within the actual VOPS. */
931 void vop_allocate_post(void *a, int rc);
932 void vop_copy_file_range_post(void *ap, int rc);
933 void vop_close_post(void *a, int rc);
934 void vop_create_pre(void *a);
935 void vop_create_post(void *a, int rc);
936 void vop_deallocate_post(void *a, int rc);
937 void vop_whiteout_pre(void *a);
938 void vop_whiteout_post(void *a, int rc);
939 void vop_deleteextattr_pre(void *a);
940 void vop_deleteextattr_post(void *a, int rc);
941 void vop_link_pre(void *a);
942 void vop_link_post(void *a, int rc);
943 void vop_lookup_post(void *a, int rc);
944 void vop_lookup_pre(void *a);
945 void vop_mkdir_pre(void *a);
946 void vop_mkdir_post(void *a, int rc);
947 void vop_mknod_pre(void *a);
948 void vop_mknod_post(void *a, int rc);
949 void vop_open_post(void *a, int rc);
950 void vop_read_post(void *a, int rc);
951 void vop_read_pgcache_post(void *ap, int rc);
952 void vop_reclaim_post(void *a, int rc);
953 void vop_remove_pre(void *a);
954 void vop_remove_post(void *a, int rc);
955 void vop_rename_post(void *a, int rc);
956 void vop_rename_pre(void *a);
957 void vop_rmdir_pre(void *a);
958 void vop_rmdir_post(void *a, int rc);
959 void vop_setattr_pre(void *a);
960 void vop_setattr_post(void *a, int rc);
961 void vop_setacl_pre(void *a);
962 void vop_setacl_post(void *a, int rc);
963 void vop_setextattr_pre(void *a);
964 void vop_setextattr_post(void *a, int rc);
965 void vop_symlink_pre(void *a);
966 void vop_symlink_post(void *a, int rc);
967 int vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a);
968
969 #ifdef INVARIANTS
970 void vop_fdatasync_debugpre(void *a);
971 void vop_fdatasync_debugpost(void *a, int rc);
972 void vop_fplookup_vexec_debugpre(void *a);
973 void vop_fplookup_vexec_debugpost(void *a, int rc);
974 void vop_fplookup_symlink_debugpre(void *a);
975 void vop_fplookup_symlink_debugpost(void *a, int rc);
976 void vop_fsync_debugpre(void *a);
977 void vop_fsync_debugpost(void *a, int rc);
978 void vop_strategy_debugpre(void *a);
979 void vop_lock_debugpre(void *a);
980 void vop_lock_debugpost(void *a, int rc);
981 void vop_unlock_debugpre(void *a);
982 void vop_need_inactive_debugpre(void *a);
983 void vop_need_inactive_debugpost(void *a, int rc);
984 void 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
1003 void 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 (((ap)->a_vp->v_v2flag & V2_KNOTE) != 0) { \
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 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE | \
1074 (noffset > osize ? NOTE_EXTEND : 0)); \
1075 INOTIFY((ap)->a_vp, IN_MODIFY); \
1076 }
1077
1078 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
1079
1080 #ifdef INVARIANTS
1081 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) \
1082 do { \
1083 int error_; \
1084 \
1085 error_ = VOP_ADD_WRITECOUNT((vp), (cnt)); \
1086 VNASSERT(error_ == 0, (vp), ("VOP_ADD_WRITECOUNT returned %d", \
1087 error_)); \
1088 } while (0)
1089 #define VOP_SET_TEXT_CHECKED(vp) \
1090 do { \
1091 int error_; \
1092 \
1093 error_ = VOP_SET_TEXT((vp)); \
1094 VNASSERT(error_ == 0, (vp), ("VOP_SET_TEXT returned %d", \
1095 error_)); \
1096 } while (0)
1097 #define VOP_UNSET_TEXT_CHECKED(vp) \
1098 do { \
1099 int error_; \
1100 \
1101 error_ = VOP_UNSET_TEXT((vp)); \
1102 VNASSERT(error_ == 0, (vp), ("VOP_UNSET_TEXT returned %d", \
1103 error_)); \
1104 } while (0)
1105 #else
1106 #define VOP_ADD_WRITECOUNT_CHECKED(vp, cnt) VOP_ADD_WRITECOUNT((vp), (cnt))
1107 #define VOP_SET_TEXT_CHECKED(vp) VOP_SET_TEXT((vp))
1108 #define VOP_UNSET_TEXT_CHECKED(vp) VOP_UNSET_TEXT((vp))
1109 #endif
1110
1111 #define VN_IS_DOOMED(vp) __predict_false((vn_irflag_read(vp) & VIRF_DOOMED) != 0)
1112
1113 void vput(struct vnode *vp);
1114 void vrele(struct vnode *vp);
1115 void vref(struct vnode *vp);
1116 void vrefact(struct vnode *vp);
1117 void v_addpollinfo(struct vnode *vp);
1118 static __inline int
vrefcnt(struct vnode * vp)1119 vrefcnt(struct vnode *vp)
1120 {
1121
1122 return (vp->v_usecount);
1123 }
1124
1125 #define vholdl(vp) do { \
1126 ASSERT_VI_LOCKED(vp, __func__); \
1127 vhold(vp); \
1128 } while (0)
1129
1130 #define vrefl(vp) do { \
1131 ASSERT_VI_LOCKED(vp, __func__); \
1132 vref(vp); \
1133 } while (0)
1134
1135 /*
1136 * The caller doesn't know the file size and vnode_create_vobject() should
1137 * determine the size on its own.
1138 */
1139 #define VNODE_NO_SIZE ((off_t)-1)
1140
1141 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
1142 int vnode_create_disk_vobject(struct vnode *vp, off_t size, struct thread *td);
1143 void vnode_destroy_vobject(struct vnode *vp);
1144
1145 extern struct vop_vector fifo_specops;
1146 extern struct vop_vector dead_vnodeops;
1147 extern struct vop_vector default_vnodeops;
1148
1149 #define VOP_PANIC ((void*)(uintptr_t)vop_panic)
1150 #define VOP_NULL ((void*)(uintptr_t)vop_null)
1151 #define VOP_EBADF ((void*)(uintptr_t)vop_ebadf)
1152 #define VOP_ENOTTY ((void*)(uintptr_t)vop_enotty)
1153 #define VOP_EINVAL ((void*)(uintptr_t)vop_einval)
1154 #define VOP_ENOENT ((void*)(uintptr_t)vop_enoent)
1155 #define VOP_EOPNOTSUPP ((void*)(uintptr_t)vop_eopnotsupp)
1156 #define VOP_EAGAIN ((void*)(uintptr_t)vop_eagain)
1157
1158 /* fifo_vnops.c */
1159 int fifo_printinfo(struct vnode *);
1160
1161 /* vfs_hash.c */
1162 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
1163
1164 void vfs_hash_changesize(u_long newhashsize);
1165 int vfs_hash_get(const struct mount *mp, u_int hash, int flags,
1166 struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1167 u_int vfs_hash_index(struct vnode *vp);
1168 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
1169 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1170 void vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
1171 struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
1172 void vfs_hash_rehash(struct vnode *vp, u_int hash);
1173 void vfs_hash_remove(struct vnode *vp);
1174
1175 int vfs_kqfilter(struct vop_kqfilter_args *);
1176 struct dirent;
1177 int vn_dir_next_dirent(struct vnode *vp, struct thread *td,
1178 char *dirbuf, size_t dirbuflen,
1179 struct dirent **dpp, size_t *len, off_t *off, int *eofflag);
1180 int vn_dir_check_empty(struct vnode *vp);
1181 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
1182
1183 int vfs_unixify_accmode(accmode_t *accmode);
1184
1185 void vfs_unp_reclaim(struct vnode *vp);
1186
1187 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
1188 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
1189 gid_t gid);
1190 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
1191 struct thread *td);
1192 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
1193 struct thread *td);
1194 int vn_getsize_locked(struct vnode *vp, off_t *size, struct ucred *active_cred);
1195 int vn_getsize(struct vnode *vp, off_t *size, struct ucred *active_cred);
1196
1197 void vn_fsid(struct vnode *vp, struct vattr *va);
1198
1199 int vn_dir_check_exec(struct vnode *vp, struct componentname *cnp);
1200 int vn_lktype_write(struct mount *mp, struct vnode *vp);
1201
1202 #ifdef INVARIANTS
1203 void vn_set_state_validate(struct vnode *vp, __enum_uint8(vstate) state);
1204 #endif
1205
1206 static inline void
vn_set_state(struct vnode * vp,__enum_uint8 (vstate)state)1207 vn_set_state(struct vnode *vp, __enum_uint8(vstate) state)
1208 {
1209 #ifdef INVARIANTS
1210 vn_set_state_validate(vp, state);
1211 #endif
1212 vp->v_state = state;
1213 }
1214
1215 static inline __enum_uint8(vstate)
vn_get_state(struct vnode * vp)1216 vn_get_state(struct vnode *vp)
1217 {
1218 return (vp->v_state);
1219 }
1220
1221 #define VOP_UNLOCK_FLAGS(vp, flags) ({ \
1222 struct vnode *_vp = (vp); \
1223 int _flags = (flags); \
1224 int _error; \
1225 \
1226 if ((_flags & ~(LK_INTERLOCK | LK_RELEASE)) != 0) \
1227 panic("%s: unsupported flags %x\n", __func__, flags); \
1228 _error = VOP_UNLOCK(_vp); \
1229 if (_flags & LK_INTERLOCK) \
1230 VI_UNLOCK(_vp); \
1231 _error; \
1232 })
1233
1234 #include <sys/kernel.h>
1235
1236 #define VFS_VOP_VECTOR_REGISTER(vnodeops) \
1237 SYSINIT(vfs_vector_##vnodeops##_f, SI_SUB_VFS, SI_ORDER_ANY, \
1238 vfs_vector_op_register, &vnodeops)
1239
1240 #define VFS_SMR_DECLARE \
1241 extern smr_t vfs_smr
1242
1243 #define VFS_SMR() vfs_smr
1244 #define vfs_smr_enter() smr_enter(VFS_SMR())
1245 #define vfs_smr_exit() smr_exit(VFS_SMR())
1246 #define vfs_smr_synchronize() smr_synchronize(VFS_SMR())
1247 #define vfs_smr_entered_load(ptr) smr_entered_load((ptr), VFS_SMR())
1248 #define VFS_SMR_ENTERED() SMR_ENTERED(VFS_SMR())
1249 #define VFS_SMR_ASSERT_ENTERED() SMR_ASSERT_ENTERED(VFS_SMR())
1250 #define VFS_SMR_ASSERT_NOT_ENTERED() SMR_ASSERT_NOT_ENTERED(VFS_SMR())
1251 #define VFS_SMR_ZONE_SET(zone) uma_zone_set_smr((zone), VFS_SMR())
1252
1253 #define vn_load_v_data_smr(vp) ({ \
1254 struct vnode *_vp = (vp); \
1255 \
1256 VFS_SMR_ASSERT_ENTERED(); \
1257 atomic_load_consume_ptr(&(_vp)->v_data);\
1258 })
1259
1260 static inline void
vn_delayed_setsize_locked(struct vnode * vp)1261 vn_delayed_setsize_locked(struct vnode *vp)
1262 {
1263 ASSERT_VI_LOCKED(vp, "delayed_setsize");
1264 vp->v_iflag |= VI_DELAYED_SETSIZE;
1265 }
1266
1267 static inline void
vn_delayed_setsize(struct vnode * vp)1268 vn_delayed_setsize(struct vnode *vp)
1269 {
1270 VI_LOCK(vp);
1271 vn_delayed_setsize_locked(vp);
1272 VI_UNLOCK(vp);
1273 }
1274
1275 static inline void
vn_clear_delayed_setsize_locked(struct vnode * vp)1276 vn_clear_delayed_setsize_locked(struct vnode *vp)
1277 {
1278 ASSERT_VI_LOCKED(vp, "delayed_setsize");
1279 vp->v_iflag &= ~VI_DELAYED_SETSIZE;
1280 }
1281
1282 static inline void
vn_clear_delayed_setsize(struct vnode * vp)1283 vn_clear_delayed_setsize(struct vnode *vp)
1284 {
1285 VI_LOCK(vp);
1286 vn_clear_delayed_setsize_locked(vp);
1287 VI_UNLOCK(vp);
1288 }
1289
1290 #endif /* _KERNEL */
1291
1292 #endif /* !_SYS_VNODE_H_ */
1293