xref: /openbsd/share/man/man9/vnsubr.9 (revision d415bd75)
1.\"     $OpenBSD: vnsubr.9,v 1.16 2019/10/06 16:32:56 jmc Exp $
2.\"     $NetBSD: vnsubr.9,v 1.21 2004/05/25 14:54:56 hannken Exp $
3.\"
4.\" Copyright (c) 2001 The NetBSD Foundation, Inc.
5.\" All rights reserved.
6.\"
7.\" This code is derived from software contributed to The NetBSD Foundation
8.\" by Gregory McGarry.
9.\"
10.\" Redistribution and use in source and binary forms, with or without
11.\" modification, are permitted provided that the following conditions
12.\" are met:
13.\" 1. Redistributions of source code must retain the above copyright
14.\"    notice, this list of conditions and the following disclaimer.
15.\" 2. Redistributions in binary form must reproduce the above copyright
16.\"    notice, this list of conditions and the following disclaimer in the
17.\"    documentation and/or other materials provided with the distribution.
18.\"
19.\" THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20.\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21.\" TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22.\" PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23.\" BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24.\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25.\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26.\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27.\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28.\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29.\" POSSIBILITY OF SUCH DAMAGE.
30.\"
31.Dd $Mdocdate: October 6 2019 $
32.Dt VNSUBR 9
33.Os
34.Sh NAME
35.Nm vnsubr ,
36.Nm vn_close ,
37.Nm vn_default_error ,
38.Nm vn_isunder ,
39.Nm vn_lock ,
40.Nm vn_marktext ,
41.Nm vn_rdwr ,
42.Nm vn_open ,
43.Nm vn_stat ,
44.Nm vn_writechk
45.Nd high-level convenience functions for vnode operations
46.Sh SYNOPSIS
47.In sys/param.h
48.In sys/lock.h
49.In sys/vnode.h
50.Ft int
51.Fn vn_close "struct vnode *vp" "int flags" "struct ucred *cred" "struct proc *p"
52.Ft int
53.Fn vn_default_error "void *v"
54.Ft int
55.Fn vn_isunder "struct vnode *dvp" "struct vnode *rvp" "struct proc *p"
56.Ft int
57.Fn vn_lock "struct vnode *vp" "int flags"
58.Ft void
59.Fn vn_marktext "struct vnode *vp"
60.Ft int
61.Fn vn_open "struct nameidata *ndp" "int fmode" "int cmode"
62.Ft int
63.Fo vn_rdwr
64.Fa "enum uio_rw rw" "struct vnode *vp" "caddr_t base"
65.Fa "int len" "off_t offset" "enum uio_seg segflg" "int ioflg"
66.Fa "struct ucred *cred" "size_t *aresid" "struct proc *p"
67.Fc
68.Ft int
69.Fn vn_stat "struct vnode *vp" "struct stat *sb" "struct proc *p"
70.Ft int
71.Fn vn_writechk "struct vnode *vp"
72.Sh DESCRIPTION
73The high-level functions described in this page are convenience
74functions for simplified access to the vnode operations described in
75.Xr VOP_LOOKUP 9 .
76.Bl -tag -width Ds
77.It Fn vn_close "vp" "flags" "cred" "p"
78Common code for a vnode close.
79The argument
80.Fa vp
81is the unlocked vnode of the vnode to close.
82.Fn vn_close
83simply locks the vnode, invokes the vnode operation
84.Fn VOP_CLOSE
85and calls
86.Xr vput 9
87to return the vnode to the freelist or holdlist.
88Note that
89.Fn vn_close
90expects an unlocked, referenced vnode and will dereference the vnode
91prior to returning.
92If the operation is successful, zero is returned;
93otherwise an appropriate error is returned.
94.It Fn vn_default_error "v"
95A generic "default" routine that just returns error.
96It is used by a file system to specify unsupported operations in
97the vnode operations vector.
98.It Fn vn_isunder "dvp" "rvp" "p"
99Common code to check if one directory specified by the vnode
100.Fa rvp
101can be found inside the directory specified by the vnode
102.Fa dvp .
103The argument
104.Fa p
105is the calling process.
106.Fn vn_isunder
107is intended to be used in
108.Xr chroot 2 ,
109.Xr chdir 2 ,
110.Xr fchdir 2 ,
111etc., to ensure that
112.Xr chroot 2
113actually means something.
114If the operation is successful, zero is returned; otherwise 1 is returned.
115.It Fn vn_lock "vp" "flags"
116Acquire the vnode lock.
117Certain file system operations require that
118the vnode lock be held when they are called.
119.Pp
120The
121.Fn vn_lock
122function must not be called when the vnode's reference count is
123zero.
124Instead, the
125.Xr vget 9
126function should be used.
127.Pp
128In addition to the
129.Fa flags
130accepted by
131.Xr VOP_LOCK 9 ,
132the
133.Dv LK_RETRY
134flag may be used.
135.Dv LK_RETRY
136causes
137.Fn vn_lock
138to return the vnode even if it has been reclaimed.
139It must not be used with
140.Dv LK_NOWAIT .
141.Pp
142The
143.Fn vn_lock
144function can sleep.
145.It Fn vn_marktext "vp"
146Common code to mark the vnode
147.Fa vp
148as being the text of a running process.
149.It Fn vn_open "ndp" "fmode" "cmode"
150Common code for vnode open operations.
151.Pp
152The pathname is described in the
153.Vt nameidata
154structure pointed to by the
155.Fa ndp
156argument.
157When initializing the nameidata structure for
158.Fn vn_open
159with
160.Xr NDINIT 9 ,
161the
162.Fa op
163must be specified as 0, and the
164.Fa flags
165may only be 0, or
166.Dv KERNELPATH .
167The lookup mode and flags are set internally
168by
169.Fn vn_open .
170.Pp
171The arguments
172.Fa fmode
173and
174.Fa cmode
175specify the
176.Xr open 2
177file mode and the access permissions for creation.
178.Fn vn_open
179checks permissions and invokes the
180.Xr VOP_OPEN 9
181or
182.Xr VOP_CREATE 9
183vnode operations.
184If the operation is successful, zero is returned;
185otherwise an appropriate error code is returned.
186.It Xo
187.Fo vn_rdwr
188.Fa "rw" "vp" "base" "len" "offset"
189.Fa "segflg" "ioflg" "cred" "aresid" "p"
190.Fc
191.Xc
192Common code to package up an I/O request on a vnode into a
193.Vt uio
194and then perform the I/O.
195The argument
196.Fa rw
197specifies whether the I/O is a read
198.Pq Dv UIO_READ
199or write
200.Pq Dv UIO_WRITE
201operation.
202The unlocked vnode is specified by
203.Fa vp .
204The arguments
205.Fa p
206and
207.Fa cred
208are the calling process and its credentials.
209The remaining arguments specify the
210.Vt uio
211parameters.
212For further information on these parameters, see
213.Xr uiomove 9 .
214.It Fn vn_stat "vp" "sb" "p"
215Common code for a vnode stat operation.
216The vnode is specified by the argument
217.Fa vp ,
218and
219.Fa sb
220is the buffer in which to store the stat information.
221The argument
222.Fa p
223is the calling process.
224.Fn vn_stat
225basically calls the vnode operation
226.Xr VOP_GETATTR 9
227and transfers the contents of a
228.Vt vattr
229structure into a
230.Vt struct stat .
231If the operation is successful, zero is returned; otherwise an
232appropriate error code is returned.
233.It Fn vn_writechk "vp"
234Common code to check for write permission on the vnode
235.Fa vp .
236A vnode is read-only if it is in use as a process's text image.
237If the vnode is read-only,
238.Er ETXTBSY
239is returned; otherwise zero is
240returned to indicate that the vnode can be written to.
241.El
242.Sh ERRORS
243.Bl -tag -width Er
244.It Bq Er ETXTBSY
245Cannot write to a vnode since it is a process's text image.
246.It Bq Er ENOENT
247The vnode has been reclaimed and is dead.
248This error is only returned if the
249.Dv LK_RETRY
250flag is not passed to
251.Fn vn_lock .
252.It Bq Er EBUSY
253The
254.Dv LK_NOWAIT
255flag was set and
256.Fn vn_lock
257would have slept.
258.El
259.Sh CODE REFERENCES
260This section describes places within the
261.Ox
262source tree where actual code implementing or using the vnode
263framework can be found.
264All pathnames are relative to
265.Pa /usr/src .
266.Pp
267The high-level convenience functions are implemented within the files
268.Pa sys/kern/vfs_vnops.c
269and
270.Pa sys/sys/vnode.h .
271.Sh SEE ALSO
272.Xr file 9 ,
273.Xr namei 9 ,
274.Xr vfs 9 ,
275.Xr vnode 9 ,
276.Xr VOP_LOOKUP 9
277.Sh BUGS
278The locking discipline is bizarre.
279Many vnode operations are passed locked vnodes on entry but release
280the lock before they exit.
281Discussions with Kirk McKusick indicate that locking
282discipline evolved out of the pre-VFS way of doing inode locking.
283In addition, the current locking discipline may actually save
284lines of code, especially if the number of file systems is fewer
285than the number of call sites.
286However, the VFS interface would
287require less wizardry if the locking discipline were simpler.
288.Pp
289The locking discipline is used in some places to attempt to make a
290series of operations atomic (e.g., permissions check +
291operation).
292This does not work for non-local file systems that do not
293support locking (e.g., NFS).
294.Pp
295Are vnode locks even necessary?
296The security checks can be moved into the individual file systems.
297Each file system can have the responsibility of ensuring that vnode
298operations are suitably atomic.
299.Pp
300The
301.Dv LK_NOWAIT
302flag does prevent the caller from sleeping.
303.Pp
304The locking discipline as it relates to shared locks has yet to be defined.
305