2 * Copyright (c) 1989, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * @(#)vnode.h 8.7 (Berkeley) 2/4/94
40 #include <sys/queue.h>
41 #include <sys/select.h>
44 #include <machine/lock.h>
47 * The vnode is the focus of all file activity in UNIX. There is a
48 * unique vnode allocated for each active file, each current directory,
49 * each mounted-on file, text file, and the root.
53 * Vnode types. VNON means no type.
55 enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD };
59 * These are for the benefit of external programs only (e.g., pstat)
60 * and should NEVER be inspected by the kernel.
63 VT_NON, VT_UFS, VT_NFS, VT_MFS, VT_PC, VT_LFS, VT_LOFS, VT_FDESC,
64 VT_PORTAL, VT_NULL, VT_UMAP, VT_KERNFS, VT_PROCFS, VT_AFS, VT_ISOFS,
65 VT_UNION, VT_MSDOSFS, VT_DEVFS, VT_TFS, VT_VFS, VT_CODA, VT_NTFS
69 * Each underlying filesystem allocates its own private area and hangs
70 * it from v_data. If non-null, this area is freed in getnewvnode().
72 TAILQ_HEAD(buflists, buf);
74 typedef int vop_t __P((void *));
78 * Reading or writing any of these items requires holding the appropriate lock.
79 * v_freelist is locked by the global vnode_free_list simple lock.
80 * v_mntvnodes is locked by the global mntvnodes simple lock.
81 * v_flag, v_usecount, v_holdcount and v_writecount are
82 * locked by the v_interlock simple lock.
83 * v_pollinfo is locked by the lock contained inside it.
86 u_long v_flag; /* vnode flags (see below) */
87 int v_usecount; /* reference count of users */
88 int v_writecount; /* reference count of writers */
89 int v_holdcnt; /* page & buffer references */
90 u_long v_id; /* capability identifier */
91 struct mount *v_mount; /* ptr to vfs we are in */
92 vop_t **v_op; /* vnode operations vector */
93 TAILQ_ENTRY(vnode) v_freelist; /* vnode freelist */
94 LIST_ENTRY(vnode) v_mntvnodes; /* vnodes for mount point */
95 struct buflists v_cleanblkhd; /* clean blocklist head */
96 struct buflists v_dirtyblkhd; /* dirty blocklist head */
97 LIST_ENTRY(vnode) v_synclist; /* vnodes with dirty buffers */
98 long v_numoutput; /* num of writes in progress */
99 enum vtype v_type; /* vnode type */
101 struct mount *vu_mountedhere;/* ptr to mounted vfs (VDIR) */
102 struct socket *vu_socket; /* unix ipc (VSOCK) */
104 struct specinfo *vu_specinfo; /* device (VCHR, VBLK) */
105 SLIST_ENTRY(vnode) vu_specnext;
107 struct fifoinfo *vu_fifoinfo; /* fifo (VFIFO) */
109 struct nqlease *v_lease; /* Soft reference to lease */
110 daddr_t v_lastw; /* last write (write cluster) */
111 daddr_t v_cstart; /* start block of cluster */
112 daddr_t v_lasta; /* last allocation */
113 int v_clen; /* length of current cluster */
114 struct vm_object *v_object; /* Place to store VM object */
115 struct simplelock v_interlock; /* lock on usecount and flag */
116 struct lock *v_vnlock; /* used for non-locking fs's */
117 enum vtagtype v_tag; /* type of underlying data */
118 void *v_data; /* private data for fs */
119 LIST_HEAD(, namecache) v_cache_src; /* Cache entries from us */
120 TAILQ_HEAD(, namecache) v_cache_dst; /* Cache entries to us */
121 struct vnode *v_dd; /* .. vnode */
122 u_long v_ddid; /* .. capability identifier */
124 struct simplelock vpi_lock; /* lock to protect below */
125 struct selinfo vpi_selinfo; /* identity of poller(s) */
126 short vpi_events; /* what they are looking for */
127 short vpi_revents; /* what has happened */
130 const char *filename; /* Source file doing locking */
131 int line; /* Line number doing locking */
134 #define v_mountedhere v_un.vu_mountedhere
135 #define v_socket v_un.vu_socket
136 #define v_rdev v_un.vu_spec.vu_specinfo
137 #define v_specnext v_un.vu_spec.vu_specnext
138 #define v_fifoinfo v_un.vu_fifoinfo
140 #define VN_POLLEVENT(vp, events) \
142 if ((vp)->v_pollinfo.vpi_events & (events)) \
143 vn_pollevent((vp), (events)); \
149 #define VROOT 0x00001 /* root of its file system */
150 #define VTEXT 0x00002 /* vnode is a pure text prototype */
151 #define VSYSTEM 0x00004 /* vnode being used by kernel */
152 #define VISTTY 0x00008 /* vnode represents a tty */
153 #define VXLOCK 0x00100 /* vnode is locked to change underlying type */
154 #define VXWANT 0x00200 /* process is waiting for vnode */
155 #define VBWAIT 0x00400 /* waiting for output to complete */
156 /* open for business 0x00800 */
157 /* open for business 0x01000 */
158 #define VOBJBUF 0x02000 /* Allocate buffers in VM object */
159 /* open for business 0x04000 */
160 #define VAGE 0x08000 /* Insert vnode at head of free list */
161 #define VOLOCK 0x10000 /* vnode is locked waiting for an object */
162 #define VOWANT 0x20000 /* a process is waiting for VOLOCK */
163 #define VDOOMED 0x40000 /* This vnode is being recycled */
164 #define VFREE 0x80000 /* This vnode is on the freelist */
165 #define VTBFREE 0x100000 /* This vnode is on the to-be-freelist */
166 #define VONWORKLST 0x200000 /* On syncer work-list */
167 #define VMOUNT 0x400000 /* Mount in progress */
170 * Vnode attributes. A field value of VNOVAL represents a field whose value
171 * is unavailable (getattr) or which is not to be changed (setattr).
174 enum vtype va_type; /* vnode type (for create) */
175 u_short va_mode; /* files access mode and type */
176 short va_nlink; /* number of references to file */
177 uid_t va_uid; /* owner user id */
178 gid_t va_gid; /* owner group id */
179 udev_t va_fsid; /* file system id */
180 long va_fileid; /* file id */
181 u_quad_t va_size; /* file size in bytes */
182 long va_blocksize; /* blocksize preferred for i/o */
183 struct timespec va_atime; /* time of last access */
184 struct timespec va_mtime; /* time of last modification */
185 struct timespec va_ctime; /* time file changed */
186 u_long va_gen; /* generation number of file */
187 u_long va_flags; /* flags defined for file */
188 udev_t va_rdev; /* device the special file represents */
189 u_quad_t va_bytes; /* bytes of disk space held by file */
190 u_quad_t va_filerev; /* file modification number */
191 u_int va_vaflags; /* operations flags, see below */
192 long va_spare; /* remain quad aligned */
196 * Flags for va_vaflags.
198 #define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */
199 #define VA_EXCLUSIVE 0x02 /* exclusive create request */
204 #define IO_UNIT 0x01 /* do I/O as atomic unit */
205 #define IO_APPEND 0x02 /* append write to end */
206 #define IO_SYNC 0x04 /* do I/O synchronously */
207 #define IO_NODELOCKED 0x08 /* underlying node already locked */
208 #define IO_NDELAY 0x10 /* FNDELAY flag set in file table */
209 #define IO_VMIO 0x20 /* data already in VMIO space */
210 #define IO_INVAL 0x40 /* invalidate after I/O */
213 * Modes. Some values same as Ixxx entries from inode.h for now.
215 #define VSUID 04000 /* set user id on execution */
216 #define VSGID 02000 /* set group id on execution */
217 #define VSVTX 01000 /* save swapped text even after use */
218 #define VREAD 00400 /* read, write, execute permissions */
223 * Token indicating no attribute value yet assigned.
229 #ifdef MALLOC_DECLARE
230 MALLOC_DECLARE(M_VNODE);
234 * Convert between vnode types and inode formats (since POSIX.1
235 * defines mode word of stat structure in terms of inode formats).
237 extern enum vtype iftovt_tab[];
238 extern int vttoif_tab[];
239 #define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12])
240 #define VTTOIF(indx) (vttoif_tab[(int)(indx)])
241 #define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode))
244 * Flags to various vnode functions.
246 #define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */
247 #define FORCECLOSE 0x0002 /* vflush: force file closure */
248 #define WRITECLOSE 0x0004 /* vflush: only close writable files */
249 #define DOCLOSE 0x0008 /* vclean: close active files */
250 #define V_SAVE 0x0001 /* vinvalbuf: sync file first */
251 #define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */
253 #define VREF(vp) vref(vp)
257 #define VATTR_NULL(vap) vattr_null(vap)
259 #define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */
260 #endif /* DIAGNOSTIC */
262 #define NULLVP ((struct vnode *)NULL)
264 #define VNODEOP_SET(f) \
265 C_SYSINIT(f##init, SI_SUB_VFS, SI_ORDER_SECOND, vfs_add_vnodeops, &f); \
266 C_SYSUNINIT(f##uninit, SI_SUB_VFS, SI_ORDER_SECOND, vfs_rm_vnodeops, &f);
271 extern struct vnode *rootvnode; /* root (i.e. "/") vnode */
272 extern int desiredvnodes; /* number of vnodes desired */
273 extern time_t syncdelay; /* max time to delay syncing data */
274 extern time_t filedelay; /* time to delay syncing files */
275 extern time_t dirdelay; /* time to delay syncing directories */
276 extern time_t metadelay; /* time to delay syncing metadata */
277 extern struct vm_zone *namei_zone;
278 extern int prtactive; /* nonzero to call vprint() */
279 extern struct vattr va_null; /* predefined null vattr structure */
280 extern int vfs_ioopt;
283 * Macro/function to check for client cache inconsistency w.r.t. leasing.
285 #define LEASE_READ 0x1 /* Check lease for readers */
286 #define LEASE_WRITE 0x2 /* Check lease for modifiers */
289 extern void (*lease_updatetime) __P((int deltat));
291 #define VSHOULDFREE(vp) \
292 (!((vp)->v_flag & (VFREE|VDOOMED)) && \
293 !(vp)->v_holdcnt && !(vp)->v_usecount && \
294 (!(vp)->v_object || \
295 !((vp)->v_object->ref_count || (vp)->v_object->resident_page_count)))
297 #define VSHOULDBUSY(vp) \
298 (((vp)->v_flag & (VFREE|VTBFREE)) && \
299 ((vp)->v_holdcnt || (vp)->v_usecount))
305 * Mods for extensibility.
309 * Flags for vdesc_flags:
311 #define VDESC_MAX_VPS 16
312 /* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
313 #define VDESC_VP0_WILLRELE 0x0001
314 #define VDESC_VP1_WILLRELE 0x0002
315 #define VDESC_VP2_WILLRELE 0x0004
316 #define VDESC_VP3_WILLRELE 0x0008
317 #define VDESC_NOMAP_VPP 0x0100
318 #define VDESC_VPP_WILLRELE 0x0200
321 * VDESC_NO_OFFSET is used to identify the end of the offset list
322 * and in places where no such field exists.
324 #define VDESC_NO_OFFSET -1
327 * This structure describes the vnode operation taking place.
329 struct vnodeop_desc {
330 int vdesc_offset; /* offset in vector--first for speed */
331 char *vdesc_name; /* a readable name for debugging */
332 int vdesc_flags; /* VDESC_* flags */
335 * These ops are used by bypass routines to map and locate arguments.
336 * Creds and procs are not needed in bypass routines, but sometimes
337 * they are useful to (for example) transport layers.
338 * Nameidata is useful because it has a cred in it.
340 int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */
341 int vdesc_vpp_offset; /* return vpp location */
342 int vdesc_cred_offset; /* cred location, if any */
343 int vdesc_proc_offset; /* proc location, if any */
344 int vdesc_componentname_offset; /* if any */
346 * Finally, we've got a list of private data (about each operation)
347 * for each transport layer. (Support to manage this list is not
350 caddr_t *vdesc_transports;
355 * A list of all the operation descs.
357 extern struct vnodeop_desc *vnodeop_descs[];
360 * Interlock for scanning list of vnodes attached to a mountpoint
362 extern struct simplelock mntvnode_slock;
365 * This macro is very helpful in defining those offsets in the vdesc struct.
367 * This is stolen from X11R4. I ignored all the fancy stuff for
368 * Crays, so if you decide to port this to such a serious machine,
369 * you might want to consult Intrinsic.h's XtOffset{,Of,To}.
371 #define VOPARG_OFFSET(p_type,field) \
372 ((int) (((char *) (&(((p_type)NULL)->field))) - ((char *) NULL)))
373 #define VOPARG_OFFSETOF(s_type,field) \
374 VOPARG_OFFSET(s_type*,field)
375 #define VOPARG_OFFSETTO(S_TYPE,S_OFFSET,STRUCT_P) \
376 ((S_TYPE)(((char*)(STRUCT_P))+(S_OFFSET)))
380 * This structure is used to configure the new vnodeops vector.
382 struct vnodeopv_entry_desc {
383 struct vnodeop_desc *opve_op; /* which operation this is */
384 vop_t *opve_impl; /* code implementing this operation */
386 struct vnodeopv_desc {
387 /* ptr to the ptr to the vector where op should go */
388 vop_t ***opv_desc_vector_p;
389 struct vnodeopv_entry_desc *opv_desc_ops; /* null terminated list */
393 * A generic structure.
394 * This can be used by bypass routines to identify generic arguments.
396 struct vop_generic_args {
397 struct vnodeop_desc *a_desc;
398 /* other random data follows, presumably */
402 #ifdef DEBUG_VFS_LOCKS
404 * Macros to aid in tracing VFS locking problems. Not totally
405 * reliable since if the process sleeps between changing the lock
406 * state and checking it with the assert, some other process could
407 * change the state. They are good enough for debugging a single
408 * filesystem using a single-threaded test. I find that 'cvs co src'
409 * is a pretty good test.
413 * [dfr] Kludge until I get around to fixing all the vfs locking.
415 #define IS_LOCKING_VFS(vp) ((vp)->v_tag == VT_UFS \
416 || (vp)->v_tag == VT_MFS \
417 || (vp)->v_tag == VT_NFS \
418 || (vp)->v_tag == VT_LFS \
419 || (vp)->v_tag == VT_ISOFS \
420 || (vp)->v_tag == VT_MSDOSFS \
421 || (vp)->v_tag == VT_DEVFS)
423 #define ASSERT_VOP_LOCKED(vp, str) \
424 if ((vp) && IS_LOCKING_VFS(vp) && !VOP_ISLOCKED(vp)) { \
425 panic("%s: %p is not locked but should be", str, vp); \
428 #define ASSERT_VOP_UNLOCKED(vp, str) \
429 if ((vp) && IS_LOCKING_VFS(vp) && VOP_ISLOCKED(vp)) { \
430 panic("%s: %p is locked but shouldn't be", str, vp); \
435 #define ASSERT_VOP_LOCKED(vp, str)
436 #define ASSERT_VOP_UNLOCKED(vp, str)
441 * VOCALL calls an op given an ops vector. We break it out because BSD's
442 * vclean changes the ops vector and then wants to call ops with the old
445 #define VOCALL(OPSV,OFF,AP) (( *((OPSV)[(OFF)])) (AP))
448 * This call works for vnodes in the kernel.
450 #define VCALL(VP,OFF,AP) VOCALL((VP)->v_op,(OFF),(AP))
451 #define VDESC(OP) (& __CONCAT(OP,_desc))
452 #define VOFFSET(OP) (VDESC(OP)->vdesc_offset)
455 * VMIO support inline
458 extern int vmiodirenable;
461 vn_canvmio(struct vnode *vp)
463 if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
469 * Finally, include the default set of vnode operations.
471 #include "vnode_if.h"
474 * Public vnode manipulation functions.
476 struct componentname;
488 struct vop_bwrite_args;
490 extern int (*lease_check_hook) __P((struct vop_lease_args *));
492 void addalias __P((struct vnode *vp, dev_t nvp_rdev));
493 void addaliasu __P((struct vnode *vp, udev_t nvp_rdev));
494 int bdevvp __P((dev_t dev, struct vnode **vpp));
495 /* cache_* may belong in namei.h. */
496 void cache_enter __P((struct vnode *dvp, struct vnode *vp,
497 struct componentname *cnp));
498 int cache_lookup __P((struct vnode *dvp, struct vnode **vpp,
499 struct componentname *cnp));
500 void cache_purge __P((struct vnode *vp));
501 void cache_purgevfs __P((struct mount *mp));
502 void cvtstat __P((struct stat *st, struct ostat *ost));
503 void cvtnstat __P((struct stat *sb, struct nstat *nsb));
504 int getnewvnode __P((enum vtagtype tag,
505 struct mount *mp, vop_t **vops, struct vnode **vpp));
506 int lease_check __P((struct vop_lease_args *ap));
507 int spec_vnoperate __P((struct vop_generic_args *));
508 int speedup_syncer __P((void));
509 void vattr_null __P((struct vattr *vap));
510 int vcount __P((struct vnode *vp));
511 void vdrop __P((struct vnode *));
512 int vfinddev __P((dev_t dev, enum vtype type, struct vnode **vpp));
513 void vfs_add_vnodeops __P((const void *));
514 void vfs_rm_vnodeops __P((const void *));
515 int vflush __P((struct mount *mp, struct vnode *skipvp, int flags));
516 int vget __P((struct vnode *vp, int lockflag, struct proc *p));
517 void vgone __P((struct vnode *vp));
518 void vhold __P((struct vnode *));
519 int vinvalbuf __P((struct vnode *vp, int save, struct ucred *cred,
521 struct proc *p, int slpflag, int slptimeo));
522 int vtruncbuf __P((struct vnode *vp, struct ucred *cred, struct proc *p,
523 off_t length, int blksize));
524 void vprint __P((char *label, struct vnode *vp));
525 int vrecycle __P((struct vnode *vp, struct simplelock *inter_lkp,
527 int vn_close __P((struct vnode *vp,
528 int flags, struct ucred *cred, struct proc *p));
529 int vn_isdisk __P((struct vnode *vp));
530 int vn_lock __P((struct vnode *vp, int flags, struct proc *p));
532 int debug_vn_lock __P((struct vnode *vp, int flags, struct proc *p,
533 const char *filename, int line));
534 #define vn_lock(vp,flags,p) debug_vn_lock(vp,flags,p,__FILE__,__LINE__)
536 int vn_open __P((struct nameidata *ndp, int fmode, int cmode));
537 void vn_pollevent __P((struct vnode *vp, int events));
538 void vn_pollgone __P((struct vnode *vp));
539 int vn_pollrecord __P((struct vnode *vp, struct proc *p, int events));
540 int vn_rdwr __P((enum uio_rw rw, struct vnode *vp, caddr_t base,
541 int len, off_t offset, enum uio_seg segflg, int ioflg,
542 struct ucred *cred, int *aresid, struct proc *p));
543 int vn_stat __P((struct vnode *vp, struct stat *sb, struct proc *p));
544 dev_t vn_todev __P((struct vnode *vp));
545 int vfs_cache_lookup __P((struct vop_lookup_args *ap));
546 int vfs_object_create __P((struct vnode *vp, struct proc *p,
547 struct ucred *cred));
548 void vfs_timestamp __P((struct timespec *));
549 int vn_writechk __P((struct vnode *vp));
550 int vop_stdbwrite __P((struct vop_bwrite_args *ap));
551 int vop_stdislocked __P((struct vop_islocked_args *));
552 int vop_stdlock __P((struct vop_lock_args *));
553 int vop_stdunlock __P((struct vop_unlock_args *));
554 int vop_noislocked __P((struct vop_islocked_args *));
555 int vop_nolock __P((struct vop_lock_args *));
556 int vop_nopoll __P((struct vop_poll_args *));
557 int vop_nounlock __P((struct vop_unlock_args *));
558 int vop_stdpathconf __P((struct vop_pathconf_args *));
559 int vop_stdpoll __P((struct vop_poll_args *));
560 int vop_revoke __P((struct vop_revoke_args *));
561 int vop_sharedlock __P((struct vop_lock_args *));
562 int vop_eopnotsupp __P((struct vop_generic_args *ap));
563 int vop_ebadf __P((struct vop_generic_args *ap));
564 int vop_einval __P((struct vop_generic_args *ap));
565 int vop_enotty __P((struct vop_generic_args *ap));
566 int vop_defaultop __P((struct vop_generic_args *ap));
567 int vop_null __P((struct vop_generic_args *ap));
568 int vop_panic __P((struct vop_generic_args *ap));
570 void vput __P((struct vnode *vp));
571 void vrele __P((struct vnode *vp));
572 void vref __P((struct vnode *vp));
573 void vbusy __P((struct vnode *vp));
575 extern vop_t **default_vnodeop_p;
576 extern vop_t **spec_vnodeop_p;
578 extern TAILQ_HEAD(tobefreelist, vnode)
579 vnode_tobefree_list; /* vnode free list */
583 #endif /* !_SYS_VNODE_H_ */