1 /*-
2 * Copyright (c) 1989, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * @(#)vnode.h 8.7 (Berkeley) 2/4/94
30 * $FreeBSD$
31 */
32
33 #ifndef _SYS_VNODE_H_
34 #define _SYS_VNODE_H_
35
36 #include <sys/bufobj.h>
37 #include <sys/queue.h>
38 #include <sys/lock.h>
39 #include <sys/lockmgr.h>
40 #include <sys/mutex.h>
41 #include <sys/rangelock.h>
42 #include <sys/selinfo.h>
43 #include <sys/uio.h>
44 #include <sys/acl.h>
45 #include <sys/ktr.h>
46
47 /*
48 * The vnode is the focus of all file activity in UNIX. There is a
49 * unique vnode allocated for each active file, each current directory,
50 * each mounted-on file, text file, and the root.
51 */
52
53 /*
54 * Vnode types. VNON means no type.
55 */
56 enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD,
57 VMARKER };
58
59 /*
60 * Each underlying filesystem allocates its own private area and hangs
61 * it from v_data. If non-null, this area is freed in getnewvnode().
62 */
63
64 struct namecache;
65
66 struct vpollinfo {
67 struct mtx vpi_lock; /* lock to protect below */
68 struct selinfo vpi_selinfo; /* identity of poller(s) */
69 short vpi_events; /* what they are looking for */
70 short vpi_revents; /* what has happened */
71 };
72
73 /*
74 * Reading or writing any of these items requires holding the appropriate lock.
75 *
76 * Lock reference:
77 * c - namecache mutex
78 * f - freelist mutex
79 * i - interlock
80 * I - updated with atomics, 0->1 and 1->0 transitions with interlock held
81 * m - mount point interlock
82 * p - pollinfo lock
83 * u - Only a reference to the vnode is needed to read.
84 * v - vnode lock
85 *
86 * Vnodes may be found on many lists. The general way to deal with operating
87 * on a vnode that is on a list is:
88 * 1) Lock the list and find the vnode.
89 * 2) Lock interlock so that the vnode does not go away.
90 * 3) Unlock the list to avoid lock order reversals.
91 * 4) vget with LK_INTERLOCK and check for ENOENT, or
92 * 5) Check for DOOMED if the vnode lock is not required.
93 * 6) Perform your operation, then vput().
94 */
95
96 #if defined(_KERNEL) || defined(_KVM_VNODE)
97
98 struct vnode {
99 /*
100 * Fields which define the identity of the vnode. These fields are
101 * owned by the filesystem (XXX: and vgone() ?)
102 */
103 const char *v_tag; /* u type of underlying data */
104 struct vop_vector *v_op; /* u vnode operations vector */
105 void *v_data; /* u private data for fs */
106
107 /*
108 * Filesystem instance stuff
109 */
110 struct mount *v_mount; /* u ptr to vfs we are in */
111 TAILQ_ENTRY(vnode) v_nmntvnodes; /* m vnodes for mount point */
112
113 /*
114 * Type specific fields, only one applies to any given vnode.
115 * See #defines below for renaming to v_* namespace.
116 */
117 union {
118 struct mount *vu_mount; /* v ptr to mountpoint (VDIR) */
119 struct socket *vu_socket; /* v unix domain net (VSOCK) */
120 struct cdev *vu_cdev; /* v device (VCHR, VBLK) */
121 struct fifoinfo *vu_fifoinfo; /* v fifo (VFIFO) */
122 } v_un;
123
124 /*
125 * vfs_hash: (mount + inode) -> vnode hash. The hash value
126 * itself is grouped with other int fields, to avoid padding.
127 */
128 LIST_ENTRY(vnode) v_hashlist;
129
130 /*
131 * VFS_namecache stuff
132 */
133 LIST_HEAD(, namecache) v_cache_src; /* c Cache entries from us */
134 TAILQ_HEAD(, namecache) v_cache_dst; /* c Cache entries to us */
135 struct namecache *v_cache_dd; /* c Cache entry for .. vnode */
136
137 /*
138 * Locking
139 */
140 struct lock v_lock; /* u (if fs don't have one) */
141 struct mtx v_interlock; /* lock for "i" things */
142 struct lock *v_vnlock; /* u pointer to vnode lock */
143
144 /*
145 * The machinery of being a vnode
146 */
147 TAILQ_ENTRY(vnode) v_actfreelist; /* f vnode active/free lists */
148 struct bufobj v_bufobj; /* * Buffer cache object */
149
150 /*
151 * Hooks for various subsystems and features.
152 */
153 struct vpollinfo *v_pollinfo; /* i Poll events, p for *v_pi */
154 struct label *v_label; /* MAC label for vnode */
155 struct lockf *v_lockf; /* Byte-level advisory lock list */
156 struct rangelock v_rl; /* Byte-range lock */
157
158 /*
159 * clustering stuff
160 */
161 daddr_t v_cstart; /* v start block of cluster */
162 daddr_t v_lasta; /* v last allocation */
163 daddr_t v_lastw; /* v last write */
164 int v_clen; /* v length of cur. cluster */
165
166 u_int v_holdcnt; /* I prevents recycling. */
167 u_int v_usecount; /* I ref count of users */
168 u_int v_iflag; /* i vnode flags (see below) */
169 u_int v_vflag; /* v vnode flags */
170 int v_writecount; /* v ref count of writers */
171 u_int v_hash;
172 enum vtype v_type; /* u vnode type */
173 };
174
175 #endif /* defined(_KERNEL) || defined(_KVM_VNODE) */
176
177 #define v_mountedhere v_un.vu_mount
178 #define v_socket v_un.vu_socket
179 #define v_rdev v_un.vu_cdev
180 #define v_fifoinfo v_un.vu_fifoinfo
181
182 /* XXX: These are temporary to avoid a source sweep at this time */
183 #define v_object v_bufobj.bo_object
184
185 /*
186 * Userland version of struct vnode, for sysctl.
187 */
188 struct xvnode {
189 size_t xv_size; /* sizeof(struct xvnode) */
190 void *xv_vnode; /* address of real vnode */
191 u_long xv_flag; /* vnode vflags */
192 int xv_usecount; /* reference count of users */
193 int xv_writecount; /* reference count of writers */
194 int xv_holdcnt; /* page & buffer references */
195 u_long xv_id; /* capability identifier */
196 void *xv_mount; /* address of parent mount */
197 long xv_numoutput; /* num of writes in progress */
198 enum vtype xv_type; /* vnode type */
199 union {
200 void *xvu_socket; /* socket, if VSOCK */
201 void *xvu_fifo; /* fifo, if VFIFO */
202 dev_t xvu_rdev; /* maj/min, if VBLK/VCHR */
203 struct {
204 dev_t xvu_dev; /* device, if VDIR/VREG/VLNK */
205 ino_t xvu_ino; /* id, if VDIR/VREG/VLNK */
206 } xv_uns;
207 } xv_un;
208 };
209 #define xv_socket xv_un.xvu_socket
210 #define xv_fifo xv_un.xvu_fifo
211 #define xv_rdev xv_un.xvu_rdev
212 #define xv_dev xv_un.xv_uns.xvu_dev
213 #define xv_ino xv_un.xv_uns.xvu_ino
214
215 /* We don't need to lock the knlist */
216 #define VN_KNLIST_EMPTY(vp) ((vp)->v_pollinfo == NULL || \
217 KNLIST_EMPTY(&(vp)->v_pollinfo->vpi_selinfo.si_note))
218
219 #define VN_KNOTE(vp, b, a) \
220 do { \
221 if (!VN_KNLIST_EMPTY(vp)) \
222 KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b), \
223 (a) | KNF_NOKQLOCK); \
224 } while (0)
225 #define VN_KNOTE_LOCKED(vp, b) VN_KNOTE(vp, b, KNF_LISTLOCKED)
226 #define VN_KNOTE_UNLOCKED(vp, b) VN_KNOTE(vp, b, 0)
227
228 /*
229 * Vnode flags.
230 * VI flags are protected by interlock and live in v_iflag
231 * VV flags are protected by the vnode lock and live in v_vflag
232 *
233 * VI_DOOMED is doubly protected by the interlock and vnode lock. Both
234 * are required for writing but the status may be checked with either.
235 */
236 #define VI_MOUNT 0x0020 /* Mount in progress */
237 #define VI_DOOMED 0x0080 /* This vnode is being recycled */
238 #define VI_FREE 0x0100 /* This vnode is on the freelist */
239 #define VI_ACTIVE 0x0200 /* This vnode is on the active list */
240 #define VI_DOINGINACT 0x0800 /* VOP_INACTIVE is in progress */
241 #define VI_OWEINACT 0x1000 /* Need to call inactive */
242
243 #define VV_ROOT 0x0001 /* root of its filesystem */
244 #define VV_ISTTY 0x0002 /* vnode represents a tty */
245 #define VV_NOSYNC 0x0004 /* unlinked, stop syncing */
246 #define VV_ETERNALDEV 0x0008 /* device that is never destroyed */
247 #define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */
248 #define VV_TEXT 0x0020 /* vnode is a pure text prototype */
249 #define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */
250 #define VV_SYSTEM 0x0080 /* vnode being used by kernel */
251 #define VV_PROCDEP 0x0100 /* vnode is process dependent */
252 #define VV_NOKNOTE 0x0200 /* don't activate knotes on this vnode */
253 #define VV_DELETED 0x0400 /* should be removed */
254 #define VV_MD 0x0800 /* vnode backs the md device */
255 #define VV_FORCEINSMQ 0x1000 /* force the insmntque to succeed */
256
257 /*
258 * Vnode attributes. A field value of VNOVAL represents a field whose value
259 * is unavailable (getattr) or which is not to be changed (setattr).
260 */
261 struct vattr {
262 enum vtype va_type; /* vnode type (for create) */
263 u_short va_mode; /* files access mode and type */
264 short va_nlink; /* number of references to file */
265 uid_t va_uid; /* owner user id */
266 gid_t va_gid; /* owner group id */
267 dev_t va_fsid; /* filesystem id */
268 long va_fileid; /* file id */
269 u_quad_t va_size; /* file size in bytes */
270 long va_blocksize; /* blocksize preferred for i/o */
271 struct timespec va_atime; /* time of last access */
272 struct timespec va_mtime; /* time of last modification */
273 struct timespec va_ctime; /* time file changed */
274 struct timespec va_birthtime; /* time file created */
275 u_long va_gen; /* generation number of file */
276 u_long va_flags; /* flags defined for file */
277 dev_t va_rdev; /* device the special file represents */
278 u_quad_t va_bytes; /* bytes of disk space held by file */
279 u_quad_t va_filerev; /* file modification number */
280 u_int va_vaflags; /* operations flags, see below */
281 long va_spare; /* remain quad aligned */
282 };
283
284 /*
285 * Flags for va_vaflags.
286 */
287 #define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */
288 #define VA_EXCLUSIVE 0x02 /* exclusive create request */
289
290 /*
291 * Flags for ioflag. (high 16 bits used to ask for read-ahead and
292 * help with write clustering)
293 * NB: IO_NDELAY and IO_DIRECT are linked to fcntl.h
294 */
295 #define IO_UNIT 0x0001 /* do I/O as atomic unit */
296 #define IO_APPEND 0x0002 /* append write to end */
297 #define IO_NDELAY 0x0004 /* FNDELAY flag set in file table */
298 #define IO_NODELOCKED 0x0008 /* underlying node already locked */
299 #define IO_ASYNC 0x0010 /* bawrite rather then bdwrite */
300 #define IO_VMIO 0x0020 /* data already in VMIO space */
301 #define IO_INVAL 0x0040 /* invalidate after I/O */
302 #define IO_SYNC 0x0080 /* do I/O synchronously */
303 #define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */
304 #define IO_EXT 0x0400 /* operate on external attributes */
305 #define IO_NORMAL 0x0800 /* operate on regular data */
306 #define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */
307 #define IO_BUFLOCKED 0x2000 /* ffs flag; indir buf is locked */
308 #define IO_RANGELOCKED 0x4000 /* range locked */
309
310 #define IO_SEQMAX 0x7F /* seq heuristic max value */
311 #define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */
312
313 /*
314 * Flags for accmode_t.
315 */
316 #define VEXEC 000000000100 /* execute/search permission */
317 #define VWRITE 000000000200 /* write permission */
318 #define VREAD 000000000400 /* read permission */
319 #define VADMIN 000000010000 /* being the file owner */
320 #define VAPPEND 000000040000 /* permission to write/append */
321 /*
322 * VEXPLICIT_DENY makes VOP_ACCESSX(9) return EPERM or EACCES only
323 * if permission was denied explicitly, by a "deny" rule in NFSv4 ACL,
324 * and 0 otherwise. This never happens with ordinary unix access rights
325 * or POSIX.1e ACLs. Obviously, VEXPLICIT_DENY must be OR-ed with
326 * some other V* constant.
327 */
328 #define VEXPLICIT_DENY 000000100000
329 #define VREAD_NAMED_ATTRS 000000200000 /* not used */
330 #define VWRITE_NAMED_ATTRS 000000400000 /* not used */
331 #define VDELETE_CHILD 000001000000
332 #define VREAD_ATTRIBUTES 000002000000 /* permission to stat(2) */
333 #define VWRITE_ATTRIBUTES 000004000000 /* change {m,c,a}time */
334 #define VDELETE 000010000000
335 #define VREAD_ACL 000020000000 /* read ACL and file mode */
336 #define VWRITE_ACL 000040000000 /* change ACL and/or file mode */
337 #define VWRITE_OWNER 000100000000 /* change file owner */
338 #define VSYNCHRONIZE 000200000000 /* not used */
339 #define VCREAT 000400000000 /* creating new file */
340 #define VVERIFY 001000000000 /* verification required */
341
342 /*
343 * Permissions that were traditionally granted only to the file owner.
344 */
345 #define VADMIN_PERMS (VADMIN | VWRITE_ATTRIBUTES | VWRITE_ACL | \
346 VWRITE_OWNER)
347
348 /*
349 * Permissions that were traditionally granted to everyone.
350 */
351 #define VSTAT_PERMS (VREAD_ATTRIBUTES | VREAD_ACL)
352
353 /*
354 * Permissions that allow to change the state of the file in any way.
355 */
356 #define VMODIFY_PERMS (VWRITE | VAPPEND | VADMIN_PERMS | VDELETE_CHILD | \
357 VDELETE)
358
359 /*
360 * Token indicating no attribute value yet assigned.
361 */
362 #define VNOVAL (-1)
363
364 /*
365 * LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
366 */
367 #define VLKTIMEOUT (hz / 20 + 1)
368
369 #ifdef _KERNEL
370
371 #ifdef MALLOC_DECLARE
372 MALLOC_DECLARE(M_VNODE);
373 #endif
374
375 /*
376 * Convert between vnode types and inode formats (since POSIX.1
377 * defines mode word of stat structure in terms of inode formats).
378 */
379 extern enum vtype iftovt_tab[];
380 extern int vttoif_tab[];
381 #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12])
382 #define VTTOIF(indx) (vttoif_tab[(int)(indx)])
383 #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode))
384
385 /*
386 * Flags to various vnode functions.
387 */
388 #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */
389 #define FORCECLOSE 0x0002 /* vflush: force file closure */
390 #define WRITECLOSE 0x0004 /* vflush: only close writable files */
391 #define EARLYFLUSH 0x0008 /* vflush: early call for ffs_flushfiles */
392 #define V_SAVE 0x0001 /* vinvalbuf: sync file first */
393 #define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */
394 #define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */
395 #define V_CLEANONLY 0x0008 /* vinvalbuf: invalidate only clean bufs */
396 #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */
397 #define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */
398 #define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */
399 #define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */
400 #define V_MNTREF 0x0010 /* vn_start_write: mp is already ref-ed */
401
402 #define VR_START_WRITE 0x0001 /* vfs_write_resume: start write atomically */
403 #define VR_NO_SUSPCLR 0x0002 /* vfs_write_resume: do not clear suspension */
404
405 #define VS_SKIP_UNMOUNT 0x0001 /* vfs_write_suspend: fail if the
406 filesystem is being unmounted */
407
408 #define VREF(vp) vref(vp)
409
410 #ifdef DIAGNOSTIC
411 #define VATTR_NULL(vap) vattr_null(vap)
412 #else
413 #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */
414 #endif /* DIAGNOSTIC */
415
416 #define NULLVP ((struct vnode *)NULL)
417
418 /*
419 * Global vnode data.
420 */
421 extern struct vnode *rootvnode; /* root (i.e. "/") vnode */
422 extern struct mount *rootdevmp; /* "/dev" mount */
423 extern int async_io_version; /* 0 or POSIX version of AIO i'face */
424 extern int desiredvnodes; /* number of vnodes desired */
425 extern struct uma_zone *namei_zone;
426 extern struct vattr va_null; /* predefined null vattr structure */
427
428 #define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock)
429 #define VI_LOCK_FLAGS(vp, flags) mtx_lock_flags(&(vp)->v_interlock, (flags))
430 #define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock)
431 #define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock)
432 #define VI_MTX(vp) (&(vp)->v_interlock)
433
434 #define VN_LOCK_AREC(vp) lockallowrecurse((vp)->v_vnlock)
435 #define VN_LOCK_ASHARE(vp) lockallowshare((vp)->v_vnlock)
436 #define VN_LOCK_DSHARE(vp) lockdisableshare((vp)->v_vnlock)
437
438 #endif /* _KERNEL */
439
440 /*
441 * Mods for extensibility.
442 */
443
444 /*
445 * Flags for vdesc_flags:
446 */
447 #define VDESC_MAX_VPS 16
448 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
449 #define VDESC_VP0_WILLRELE 0x0001
450 #define VDESC_VP1_WILLRELE 0x0002
451 #define VDESC_VP2_WILLRELE 0x0004
452 #define VDESC_VP3_WILLRELE 0x0008
453 #define VDESC_NOMAP_VPP 0x0100
454 #define VDESC_VPP_WILLRELE 0x0200
455
456 /*
457 * A generic structure.
458 * This can be used by bypass routines to identify generic arguments.
459 */
460 struct vop_generic_args {
461 struct vnodeop_desc *a_desc;
462 /* other random data follows, presumably */
463 };
464
465 typedef int vop_bypass_t(struct vop_generic_args *);
466
467 /*
468 * VDESC_NO_OFFSET is used to identify the end of the offset list
469 * and in places where no such field exists.
470 */
471 #define VDESC_NO_OFFSET -1
472
473 /*
474 * This structure describes the vnode operation taking place.
475 */
476 struct vnodeop_desc {
477 char *vdesc_name; /* a readable name for debugging */
478 int vdesc_flags; /* VDESC_* flags */
479 vop_bypass_t *vdesc_call; /* Function to call */
480
481 /*
482 * These ops are used by bypass routines to map and locate arguments.
483 * Creds and procs are not needed in bypass routines, but sometimes
484 * they are useful to (for example) transport layers.
485 * Nameidata is useful because it has a cred in it.
486 */
487 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */
488 int vdesc_vpp_offset; /* return vpp location */
489 int vdesc_cred_offset; /* cred location, if any */
490 int vdesc_thread_offset; /* thread location, if any */
491 int vdesc_componentname_offset; /* if any */
492 };
493
494 #ifdef _KERNEL
495 /*
496 * A list of all the operation descs.
497 */
498 extern struct vnodeop_desc *vnodeop_descs[];
499
500 #define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field)
501 #define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
502 ((s_type)(((char*)(struct_p)) + (s_offset)))
503
504
505 #ifdef DEBUG_VFS_LOCKS
506 /*
507 * Support code to aid in debugging VFS locking problems. Not totally
508 * reliable since if the thread sleeps between changing the lock
509 * state and checking it with the assert, some other thread could
510 * change the state. They are good enough for debugging a single
511 * filesystem using a single-threaded test. Note that the unreliability is
512 * limited to false negatives; efforts were made to ensure that false
513 * positives cannot occur.
514 */
515 void assert_vi_locked(struct vnode *vp, const char *str);
516 void assert_vi_unlocked(struct vnode *vp, const char *str);
517 void assert_vop_elocked(struct vnode *vp, const char *str);
518 #if 0
519 void assert_vop_elocked_other(struct vnode *vp, const char *str);
520 #endif
521 void assert_vop_locked(struct vnode *vp, const char *str);
522 #if 0
523 voi0 assert_vop_slocked(struct vnode *vp, const char *str);
524 #endif
525 void assert_vop_unlocked(struct vnode *vp, const char *str);
526
527 #define ASSERT_VI_LOCKED(vp, str) assert_vi_locked((vp), (str))
528 #define ASSERT_VI_UNLOCKED(vp, str) assert_vi_unlocked((vp), (str))
529 #define ASSERT_VOP_ELOCKED(vp, str) assert_vop_elocked((vp), (str))
530 #if 0
531 #define ASSERT_VOP_ELOCKED_OTHER(vp, str) assert_vop_locked_other((vp), (str))
532 #endif
533 #define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked((vp), (str))
534 #if 0
535 #define ASSERT_VOP_SLOCKED(vp, str) assert_vop_slocked((vp), (str))
536 #endif
537 #define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked((vp), (str))
538
539 #else /* !DEBUG_VFS_LOCKS */
540
541 #define ASSERT_VI_LOCKED(vp, str) ((void)0)
542 #define ASSERT_VI_UNLOCKED(vp, str) ((void)0)
543 #define ASSERT_VOP_ELOCKED(vp, str) ((void)0)
544 #if 0
545 #define ASSERT_VOP_ELOCKED_OTHER(vp, str)
546 #endif
547 #define ASSERT_VOP_LOCKED(vp, str) ((void)0)
548 #if 0
549 #define ASSERT_VOP_SLOCKED(vp, str)
550 #endif
551 #define ASSERT_VOP_UNLOCKED(vp, str) ((void)0)
552 #endif /* DEBUG_VFS_LOCKS */
553
554
555 /*
556 * This call works for vnodes in the kernel.
557 */
558 #define VCALL(c) ((c)->a_desc->vdesc_call(c))
559
560 #define DOINGASYNC(vp) \
561 (((vp)->v_mount->mnt_kern_flag & MNTK_ASYNC) != 0 && \
562 ((curthread->td_pflags & TDP_SYNCIO) == 0))
563
564 /*
565 * VMIO support inline
566 */
567
568 extern int vmiodirenable;
569
570 static __inline int
vn_canvmio(struct vnode * vp)571 vn_canvmio(struct vnode *vp)
572 {
573 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
574 return(TRUE);
575 return(FALSE);
576 }
577
578 /*
579 * Finally, include the default set of vnode operations.
580 */
581 typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int);
582 #include "vnode_if.h"
583
584 /* vn_open_flags */
585 #define VN_OPEN_NOAUDIT 0x00000001
586 #define VN_OPEN_NOCAPCHECK 0x00000002
587 #define VN_OPEN_NAMECACHE 0x00000004
588
589 /*
590 * Public vnode manipulation functions.
591 */
592 struct componentname;
593 struct file;
594 struct mount;
595 struct nameidata;
596 struct ostat;
597 struct thread;
598 struct proc;
599 struct stat;
600 struct nstat;
601 struct ucred;
602 struct uio;
603 struct vattr;
604 struct vnode;
605
606 typedef int (*vn_get_ino_t)(struct mount *, void *, int, struct vnode **);
607
608 int bnoreuselist(struct bufv *bufv, struct bufobj *bo, daddr_t startn,
609 daddr_t endn);
610 /* cache_* may belong in namei.h. */
611 void cache_changesize(int newhashsize);
612 #define cache_enter(dvp, vp, cnp) \
613 cache_enter_time(dvp, vp, cnp, NULL, NULL)
614 void cache_enter_time(struct vnode *dvp, struct vnode *vp,
615 struct componentname *cnp, struct timespec *tsp,
616 struct timespec *dtsp);
617 int cache_lookup(struct vnode *dvp, struct vnode **vpp,
618 struct componentname *cnp, struct timespec *tsp, int *ticksp);
619 void cache_purge(struct vnode *vp);
620 void cache_purge_negative(struct vnode *vp);
621 void cache_purgevfs(struct mount *mp);
622 int change_dir(struct vnode *vp, struct thread *td);
623 void cvtstat(struct stat *st, struct ostat *ost);
624 void cvtnstat(struct stat *sb, struct nstat *nsb);
625 int getnewvnode(const char *tag, struct mount *mp, struct vop_vector *vops,
626 struct vnode **vpp);
627 void getnewvnode_reserve(u_int count);
628 void getnewvnode_drop_reserve(void);
629 int insmntque1(struct vnode *vp, struct mount *mp,
630 void (*dtr)(struct vnode *, void *), void *dtr_arg);
631 int insmntque(struct vnode *vp, struct mount *mp);
632 u_quad_t init_va_filerev(void);
633 int speedup_syncer(void);
634 int vn_vptocnp(struct vnode **vp, struct ucred *cred, char *buf,
635 u_int *buflen);
636 #define textvp_fullpath(p, rb, rfb) \
637 vn_fullpath(FIRST_THREAD_IN_PROC(p), (p)->p_textvp, rb, rfb)
638 int vn_fullpath(struct thread *td, struct vnode *vn,
639 char **retbuf, char **freebuf);
640 int vn_fullpath_global(struct thread *td, struct vnode *vn,
641 char **retbuf, char **freebuf);
642 struct vnode *
643 vn_dir_dd_ino(struct vnode *vp);
644 int vn_commname(struct vnode *vn, char *buf, u_int buflen);
645 int vn_path_to_global_path(struct thread *td, struct vnode *vp,
646 char *path, u_int pathlen);
647 int vaccess(enum vtype type, mode_t file_mode, uid_t file_uid,
648 gid_t file_gid, accmode_t accmode, struct ucred *cred,
649 int *privused);
650 int vaccess_acl_nfs4(enum vtype type, uid_t file_uid, gid_t file_gid,
651 struct acl *aclp, accmode_t accmode, struct ucred *cred,
652 int *privused);
653 int vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
654 gid_t file_gid, struct acl *acl, accmode_t accmode,
655 struct ucred *cred, int *privused);
656 void vattr_null(struct vattr *vap);
657 int vcount(struct vnode *vp);
658 #define vdrop(vp) _vdrop((vp), 0)
659 #define vdropl(vp) _vdrop((vp), 1)
660 void _vdrop(struct vnode *, bool);
661 int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td);
662 int vget(struct vnode *vp, int lockflag, struct thread *td);
663 void vgone(struct vnode *vp);
664 #define vhold(vp) _vhold((vp), 0)
665 #define vholdl(vp) _vhold((vp), 1)
666 void _vhold(struct vnode *, bool);
667 void vinactive(struct vnode *, struct thread *);
668 int vinvalbuf(struct vnode *vp, int save, int slpflag, int slptimeo);
669 int vtruncbuf(struct vnode *vp, struct ucred *cred, off_t length,
670 int blksize);
671 void vunref(struct vnode *);
672 void vn_printf(struct vnode *vp, const char *fmt, ...) __printflike(2,3);
673 #define vprint(label, vp) vn_printf((vp), "%s\n", (label))
674 int vrecycle(struct vnode *vp);
675 int vn_bmap_seekhole(struct vnode *vp, u_long cmd, off_t *off,
676 struct ucred *cred);
677 int vn_close(struct vnode *vp,
678 int flags, struct ucred *file_cred, struct thread *td);
679 void vn_finished_write(struct mount *mp);
680 void vn_finished_secondary_write(struct mount *mp);
681 int vn_isdisk(struct vnode *vp, int *errp);
682 int _vn_lock(struct vnode *vp, int flags, char *file, int line);
683 #define vn_lock(vp, flags) _vn_lock(vp, flags, __FILE__, __LINE__)
684 int vn_open(struct nameidata *ndp, int *flagp, int cmode, struct file *fp);
685 int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
686 u_int vn_open_flags, struct ucred *cred, struct file *fp);
687 int vn_open_vnode(struct vnode *vp, int fmode, struct ucred *cred,
688 struct thread *td, struct file *fp);
689 void vn_pages_remove(struct vnode *vp, vm_pindex_t start, vm_pindex_t end);
690 int vn_pollrecord(struct vnode *vp, struct thread *p, int events);
691 int vn_rdwr(enum uio_rw rw, struct vnode *vp, void *base,
692 int len, off_t offset, enum uio_seg segflg, int ioflg,
693 struct ucred *active_cred, struct ucred *file_cred, ssize_t *aresid,
694 struct thread *td);
695 int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, void *base,
696 size_t len, off_t offset, enum uio_seg segflg, int ioflg,
697 struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
698 struct thread *td);
699 int vn_rlimit_fsize(const struct vnode *vn, const struct uio *uio,
700 struct thread *td);
701 int vn_stat(struct vnode *vp, struct stat *sb, struct ucred *active_cred,
702 struct ucred *file_cred, struct thread *td);
703 int vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
704 int vn_start_secondary_write(struct vnode *vp, struct mount **mpp,
705 int flags);
706 int vn_writechk(struct vnode *vp);
707 int vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
708 const char *attrname, int *buflen, char *buf, struct thread *td);
709 int vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
710 const char *attrname, int buflen, char *buf, struct thread *td);
711 int vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
712 const char *attrname, struct thread *td);
713 int vn_vget_ino(struct vnode *vp, ino_t ino, int lkflags,
714 struct vnode **rvp);
715 int vn_vget_ino_gen(struct vnode *vp, vn_get_ino_t alloc,
716 void *alloc_arg, int lkflags, struct vnode **rvp);
717 int vn_utimes_perm(struct vnode *vp, struct vattr *vap,
718 struct ucred *cred, struct thread *td);
719
720 int vn_io_fault_uiomove(char *data, int xfersize, struct uio *uio);
721 int vn_io_fault_pgmove(vm_page_t ma[], vm_offset_t offset, int xfersize,
722 struct uio *uio);
723
724 #define vn_rangelock_unlock(vp, cookie) \
725 rangelock_unlock(&(vp)->v_rl, (cookie), VI_MTX(vp))
726 #define vn_rangelock_unlock_range(vp, cookie, start, end) \
727 rangelock_unlock_range(&(vp)->v_rl, (cookie), (start), (end), \
728 VI_MTX(vp))
729 #define vn_rangelock_rlock(vp, start, end) \
730 rangelock_rlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
731 #define vn_rangelock_wlock(vp, start, end) \
732 rangelock_wlock(&(vp)->v_rl, (start), (end), VI_MTX(vp))
733
734 int vfs_cache_lookup(struct vop_lookup_args *ap);
735 void vfs_timestamp(struct timespec *);
736 void vfs_write_resume(struct mount *mp, int flags);
737 int vfs_write_suspend(struct mount *mp, int flags);
738 int vfs_write_suspend_umnt(struct mount *mp);
739 int vop_stdbmap(struct vop_bmap_args *);
740 int vop_stdfsync(struct vop_fsync_args *);
741 int vop_stdgetwritemount(struct vop_getwritemount_args *);
742 int vop_stdgetpages(struct vop_getpages_args *);
743 int vop_stdinactive(struct vop_inactive_args *);
744 int vop_stdislocked(struct vop_islocked_args *);
745 int vop_stdkqfilter(struct vop_kqfilter_args *);
746 int vop_stdlock(struct vop_lock1_args *);
747 int vop_stdputpages(struct vop_putpages_args *);
748 int vop_stdunlock(struct vop_unlock_args *);
749 int vop_nopoll(struct vop_poll_args *);
750 int vop_stdaccess(struct vop_access_args *ap);
751 int vop_stdaccessx(struct vop_accessx_args *ap);
752 int vop_stdadvise(struct vop_advise_args *ap);
753 int vop_stdadvlock(struct vop_advlock_args *ap);
754 int vop_stdadvlockasync(struct vop_advlockasync_args *ap);
755 int vop_stdadvlockpurge(struct vop_advlockpurge_args *ap);
756 int vop_stdallocate(struct vop_allocate_args *ap);
757 int vop_stdpathconf(struct vop_pathconf_args *);
758 int vop_stdpoll(struct vop_poll_args *);
759 int vop_stdvptocnp(struct vop_vptocnp_args *ap);
760 int vop_stdvptofh(struct vop_vptofh_args *ap);
761 int vop_stdunp_bind(struct vop_unp_bind_args *ap);
762 int vop_stdunp_connect(struct vop_unp_connect_args *ap);
763 int vop_stdunp_detach(struct vop_unp_detach_args *ap);
764 int vop_eopnotsupp(struct vop_generic_args *ap);
765 int vop_ebadf(struct vop_generic_args *ap);
766 int vop_einval(struct vop_generic_args *ap);
767 int vop_enoent(struct vop_generic_args *ap);
768 int vop_enotty(struct vop_generic_args *ap);
769 int vop_null(struct vop_generic_args *ap);
770 int vop_panic(struct vop_generic_args *ap);
771 int dead_poll(struct vop_poll_args *ap);
772 int dead_read(struct vop_read_args *ap);
773 int dead_write(struct vop_write_args *ap);
774
775 /* These are called from within the actual VOPS. */
776 void vop_create_post(void *a, int rc);
777 void vop_deleteextattr_post(void *a, int rc);
778 void vop_link_post(void *a, int rc);
779 void vop_lock_pre(void *a);
780 void vop_lock_post(void *a, int rc);
781 void vop_lookup_post(void *a, int rc);
782 void vop_lookup_pre(void *a);
783 void vop_mkdir_post(void *a, int rc);
784 void vop_mknod_post(void *a, int rc);
785 void vop_reclaim_post(void *a, int rc);
786 void vop_remove_post(void *a, int rc);
787 void vop_rename_post(void *a, int rc);
788 void vop_rename_pre(void *a);
789 void vop_rmdir_post(void *a, int rc);
790 void vop_setattr_post(void *a, int rc);
791 void vop_setextattr_post(void *a, int rc);
792 void vop_strategy_pre(void *a);
793 void vop_symlink_post(void *a, int rc);
794 void vop_unlock_post(void *a, int rc);
795 void vop_unlock_pre(void *a);
796
797 void vop_rename_fail(struct vop_rename_args *ap);
798
799 #define VOP_WRITE_PRE(ap) \
800 struct vattr va; \
801 int error; \
802 off_t osize, ooffset, noffset; \
803 \
804 osize = ooffset = noffset = 0; \
805 if (!VN_KNLIST_EMPTY((ap)->a_vp)) { \
806 error = VOP_GETATTR((ap)->a_vp, &va, (ap)->a_cred); \
807 if (error) \
808 return (error); \
809 ooffset = (ap)->a_uio->uio_offset; \
810 osize = (off_t)va.va_size; \
811 }
812
813 #define VOP_WRITE_POST(ap, ret) \
814 noffset = (ap)->a_uio->uio_offset; \
815 if (noffset > ooffset && !VN_KNLIST_EMPTY((ap)->a_vp)) { \
816 VFS_KNOTE_LOCKED((ap)->a_vp, NOTE_WRITE \
817 | (noffset > osize ? NOTE_EXTEND : 0)); \
818 }
819
820 #define VOP_LOCK(vp, flags) VOP_LOCK1(vp, flags, __FILE__, __LINE__)
821
822
823 void vput(struct vnode *vp);
824 void vrele(struct vnode *vp);
825 void vref(struct vnode *vp);
826 int vrefcnt(struct vnode *vp);
827 void v_addpollinfo(struct vnode *vp);
828
829 int vnode_create_vobject(struct vnode *vp, off_t size, struct thread *td);
830 void vnode_destroy_vobject(struct vnode *vp);
831
832 extern struct vop_vector fifo_specops;
833 extern struct vop_vector dead_vnodeops;
834 extern struct vop_vector default_vnodeops;
835
836 #define VOP_PANIC ((void*)(uintptr_t)vop_panic)
837 #define VOP_NULL ((void*)(uintptr_t)vop_null)
838 #define VOP_EBADF ((void*)(uintptr_t)vop_ebadf)
839 #define VOP_ENOTTY ((void*)(uintptr_t)vop_enotty)
840 #define VOP_EINVAL ((void*)(uintptr_t)vop_einval)
841 #define VOP_ENOENT ((void*)(uintptr_t)vop_enoent)
842 #define VOP_EOPNOTSUPP ((void*)(uintptr_t)vop_eopnotsupp)
843
844 /* fifo_vnops.c */
845 int fifo_printinfo(struct vnode *);
846
847 /* vfs_hash.c */
848 typedef int vfs_hash_cmp_t(struct vnode *vp, void *arg);
849
850 void vfs_hash_changesize(int newhashsize);
851 int vfs_hash_get(const struct mount *mp, u_int hash, int flags, struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
852 u_int vfs_hash_index(struct vnode *vp);
853 int vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td, struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg);
854 void vfs_hash_rehash(struct vnode *vp, u_int hash);
855 void vfs_hash_remove(struct vnode *vp);
856
857 int vfs_kqfilter(struct vop_kqfilter_args *);
858 void vfs_mark_atime(struct vnode *vp, struct ucred *cred);
859 struct dirent;
860 int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off);
861
862 int vfs_unixify_accmode(accmode_t *accmode);
863
864 void vfs_unp_reclaim(struct vnode *vp);
865
866 int setfmode(struct thread *td, struct ucred *cred, struct vnode *vp, int mode);
867 int setfown(struct thread *td, struct ucred *cred, struct vnode *vp, uid_t uid,
868 gid_t gid);
869 int vn_chmod(struct file *fp, mode_t mode, struct ucred *active_cred,
870 struct thread *td);
871 int vn_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred,
872 struct thread *td);
873
874 #endif /* _KERNEL */
875
876 #endif /* !_SYS_VNODE_H_ */
877