xref: /freebsd/share/man/man9/namei.9 (revision 3c6e15bceeab4470243c60c9a4b5b9cafca9abaa)
1.\"
2.\" Copyright (c) 1998, 1999 Eivind Eklund
3.\" Copyright (c) 2003 Hiten M. Pandya
4.\" Copyright (c) 2005 Robert N. M. Watson
5.\"
6.\" All rights reserved.
7.\"
8.\" This program is free software.
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 DEVELOPERS ``AS IS'' AND ANY EXPRESS OR
20.\" IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
21.\" OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
22.\" IN NO EVENT SHALL THE DEVELOPERS BE LIABLE FOR ANY DIRECT, INDIRECT,
23.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
24.\" NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25.\" DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26.\" THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27.\" (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
28.\" THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
29.\"
30.\"
31.\" If you integrate this manpage in another OS, I'd appreciate a note
32.\"	- eivind@FreeBSD.org
33.\"
34.\" $FreeBSD$
35.\"
36.Dd September 21, 2005
37.Os
38.Dt NAMEI 9
39.Sh NAME
40.Nm namei ,
41.Nm NDINIT ,
42.Nm NDFREE ,
43.Nm NDHASGIANT
44.Nd pathname translation and lookup operations
45.Sh SYNOPSIS
46.In sys/param.h
47.In sys/proc.h
48.In sys/namei.h
49.Ft int
50.Fn namei "struct nameidata *ndp"
51.Ft void
52.Fo NDINIT
53.Fa "struct nameidata *ndp" "u_long op" "u_long flags"
54.Fa "enum uio_seg segflg" "const char *namep" "struct thread *td"
55.Fc
56.Ft void
57.Fn NDFREE "struct nameidata *ndp" "const uint flags"
58.Ft int
59.Fn NDHASGIANT "struct nameidata *ndp"
60.Sh DESCRIPTION
61The
62.Nm
63facility allows the client to perform pathname translation and lookup
64operations.
65The
66.Nm
67functions will increment the reference count for the vnode in question.
68The reference count has to be decremented after use of the vnode, by
69using either
70.Xr vrele 9
71or
72.Xr vput 9 ,
73depending on whether the
74.Dv LOCKLEAF
75flag was specified or not.
76If the
77.Va Giant
78lock is required,
79.Nm
80will acquire it if the caller indicates it is
81.Dv MPSAFE ,
82in which case the caller must later release
83.Va Giant
84based on the results of
85.Fn NDHASGIANT .
86.Pp
87The
88.Fn NDINIT
89function is used to initialize
90.Nm
91components.
92It takes the following arguments:
93.Bl -tag -width ".Fa segflg"
94.It Fa ndp
95The
96.Vt "struct nameidata"
97to initialize.
98.It Fa op
99The operation which
100.Fn namei
101will perform.
102The following operations are valid:
103.Dv LOOKUP , CREATE , DELETE ,
104and
105.Dv RENAME .
106The latter three are just setup for those
107effects; just calling
108.Fn namei
109will not result in
110.Fn VOP_RENAME
111being called.
112.It Fa flags
113Operation flags.
114Several of these can be effective at the same time.
115.It Fa segflg
116UIO segment indicator.
117This indicates if the name of the object is in userspace
118.Pq Dv UIO_USERSPACE
119or in the kernel address space
120.Pq Dv UIO_SYSSPACE .
121.It Fa namep
122Pointer to the component's pathname buffer
123(the file or directory name that will be looked up).
124.It Fa td
125The thread context to use for
126.Nm
127operations and locks.
128.El
129.Sh NAMEI OPERATION FLAGS
130The
131.Fn namei
132function takes the following set of
133.Dq "operation flags"
134that influence its operation:
135.Bl -tag -width ".Dv WANTPARENT"
136.It Dv LOCKLEAF
137Lock vnode on return.
138This is a full lock of the vnode; the
139.Xr VOP_UNLOCK 9
140should be used
141to release the lock (or
142.Xr vput 9
143which is equivalent to calling
144.Xr VOP_UNLOCK 9
145followed by
146.Xr vrele 9 ,
147all in one).
148.It Dv LOCKPARENT
149This flag lets the
150.Fn namei
151function return the parent (directory) vnode,
152.Va ni_dvp
153in locked state, unless it is identical to
154.Va ni_vp ,
155in which case
156.Va ni_dvp
157is not locked per se (but may be locked due to
158.Dv LOCKLEAF ) .
159If a lock is enforced, it should be released using
160.Xr vput 9
161or
162.Xr VOP_UNLOCK 9
163and
164.Xr vrele 9 .
165.It Dv WANTPARENT
166This flag allows the
167.Fn namei
168function to return the parent (directory) vnode in an unlocked state.
169The parent vnode must be released separately by using
170.Xr vrele 9 .
171.It Dv MPSAFE
172With this flag set,
173.Fn namei
174will conditionally acquire
175.Va Giant
176if it is required by a traversed file system.
177MPSAFE callers should pass the results of
178.Fn NDHASGIANT
179to
180.Xr VFS_UNLOCK_GIANT
181in order to conditionally release
182.Va Giant
183if necessary.
184.It Dv NOCACHE
185Avoid
186.Fn namei
187creating this entry in the namecache if it is not
188already present.
189Normally,
190.Fn namei
191will add entries to the name cache
192if they are not already there.
193.It Dv FOLLOW
194With this flag,
195.Fn namei
196will follow the symbolic link if the last part
197of the path supplied is a symbolic link (i.e., it will return a vnode
198for whatever the link points at, instead for the link itself).
199.It Dv NOOBJ
200Do not call
201.Fn vfs_object_create
202for the returned vnode, even though it meets required criteria for VM support.
203.It Dv NOFOLLOW
204Do not follow symbolic links (pseudo).
205This flag is not looked for by the actual code, which looks for
206.Dv FOLLOW .
207.Dv NOFOLLOW
208is used to indicate to the source code reader that symlinks
209are intentionally not followed.
210.It Dv SAVENAME
211Do not free the pathname buffer at the end of the
212.Fn namei
213invocation; instead, free it later in
214.Fn NDFREE
215so that the caller may access the pathname buffer.
216See below for details.
217.It Dv SAVESTART
218Retain an additional reference to the parent directory; do not free
219the pathname buffer.
220See below for details.
221.El
222.Sh ALLOCATED ELEMENTS
223The
224.Vt nameidata
225structure is composed of the following fields:
226.Bl -tag -width ".Va ni_cnd.cn_pnbuf"
227.It Va ni_startdir
228In the normal case, this is either the current directory or the root.
229It is the current directory if the name passed in does not start with
230.Ql /
231and we have not gone through any symlinks with an absolute path, and
232the root otherwise.
233.Pp
234In this case, it is only used by
235.Fn lookup ,
236and should not be
237considered valid after a call to
238.Fn namei .
239If
240.Dv SAVESTART
241is set, this is set to the same as
242.Va ni_dvp ,
243with an extra
244.Xr vref 9 .
245To block
246.Fn NDFREE
247from releasing
248.Va ni_startdir ,
249the
250.Dv NDF_NO_STARTDIR_RELE
251can be set.
252.It Va ni_dvp
253Vnode pointer to directory of the object on which lookup is performed.
254This is available on successful return if
255.Dv LOCKPARENT
256or
257.Dv WANTPARENT
258is set.
259It is locked if
260.Dv LOCKPARENT
261is set.
262Freeing this in
263.Fn NDFREE
264can be inhibited by
265.Dv NDF_NO_DVP_RELE , NDF_NO_DVP_PUT ,
266or
267.Dv NDF_NO_DVP_UNLOCK
268(with the obvious effects).
269.It Va ni_vp
270Vnode pointer to the resulting object,
271.Dv NULL
272otherwise.
273The
274.Va v_usecount
275field of this vnode is incremented.
276If
277.Dv LOCKLEAF
278is set, it is also locked.
279.Pp
280Freeing this in
281.Fn NDFREE
282can be inhibited by
283.Dv NDF_NO_VP_RELE , NDF_NO_VP_PUT ,
284or
285.Dv NDF_NO_VP_UNLOCK
286(with the obvious effects).
287.It Va ni_cnd.cn_pnbuf
288The pathname buffer contains the location of the file or directory
289that will be used by the
290.Nm
291operations.
292It is managed by the
293.Xr uma 9
294zone allocation interface.
295If the
296.Dv SAVESTART
297or
298.Dv SAVENAME
299flag is set, then the pathname buffer is available
300after calling the
301.Fn namei
302function.
303.Pp
304To only deallocate resources used by the pathname buffer,
305.Va ni_cnd.cn_pnbuf ,
306then
307.Dv NDF_ONLY_PNBUF
308flag can be passed to the
309.Fn NDFREE
310function.
311To keep the pathname buffer intact,
312the
313.Dv NDF_NO_FREE_PNBUF
314flag can be passed to the
315.Fn NDFREE
316function.
317.El
318.Sh FILES
319.Bl -tag
320.It Pa src/sys/kern/vfs_lookup.c
321.El
322.Sh SEE ALSO
323.Xr uio 9 ,
324.Xr uma 9 ,
325.Xr VFS 9 ,
326.Xr VFS_UNLOCK_GIANT 9 ,
327.Xr vnode 9 ,
328.Xr vput 9 ,
329.Xr vref 9
330.Sh AUTHORS
331.An -nosplit
332This manual page was written by
333.An Eivind Eklund Aq eivind@FreeBSD.org
334and later significantly revised by
335.An Hiten M. Pandya Aq hmp@FreeBSD.org .
336.Sh BUGS
337The
338.Dv LOCKPARENT
339flag does not always result in the parent vnode being locked.
340This results in complications when the
341.Dv LOCKPARENT
342is used.
343In order to solve this for the cases where both
344.Dv LOCKPARENT
345and
346.Dv LOCKLEAF
347are used, it is necessary to resort to recursive locking.
348.Pp
349Non-MPSAFE file systems exist, requiring callers to conditionally unlock
350.Va Giant .
351