1RENAME(2) Linux Programmer's Manual RENAME(2)
2
6 rename, renameat, renameat2 - change the name or location of a file
7
9 #include <stdio.h>
10 int rename(const char *oldpath, const char *newpath);
12 #include <fcntl.h> /* Definition of AT_* constants */
14 #include <stdio.h>
15 int renameat(int olddirfd, const char *oldpath,
17 int newdirfd, const char *newpath);
18 int renameat2(int olddirfd, const char *oldpath,
20 int newdirfd, const char *newpath, unsigned int flags);
21 Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
23 renameat():
25 Since glibc 2.10:
26 _POSIX_C_SOURCE >= 200809L
27 Before glibc 2.10:
28 _ATFILE_SOURCE
29 renameat2():
30 _GNU_SOURCE
31
33 rename() renames a file, moving it between directories if required.
34 Any other hard links to the file (as created using link(2)) are unaf‐
35 fected. Open file descriptors for oldpath are also unaffected.
36 Various restrictions determine whether or not the rename operation suc‐
38 ceeds: see ERRORS below.
39 If newpath already exists, it will be atomically replaced, so that
41 there is no point at which another process attempting to access newpath
42 will find it missing. However, there will probably be a window in
43 which both oldpath and newpath refer to the file being renamed.
44 If oldpath and newpath are existing hard links referring to the same
46 file, then rename() does nothing, and returns a success status.
47 If newpath exists but the operation fails for some reason, rename()
49 guarantees to leave an instance of newpath in place.
50 oldpath can specify a directory. In this case, newpath must either not
52 exist, or it must specify an empty directory.
53 If oldpath refers to a symbolic link, the link is renamed; if newpath
55 refers to a symbolic link, the link will be overwritten.
56 renameat()
58 The renameat() system call operates in exactly the same way as
59 rename(), except for the differences described here.
60 If the pathname given in oldpath is relative, then it is interpreted
62 relative to the directory referred to by the file descriptor olddirfd
63 (rather than relative to the current working directory of the calling
64 process, as is done by rename() for a relative pathname).
65 If oldpath is relative and olddirfd is the special value AT_FDCWD, then
67 oldpath is interpreted relative to the current working directory of the
68 calling process (like rename()).
69 If oldpath is absolute, then olddirfd is ignored.
71 The interpretation of newpath is as for oldpath, except that a relative
73 pathname is interpreted relative to the directory referred to by the
74 file descriptor newdirfd.
75 See openat(2) for an explanation of the need for renameat().
77 renameat2()
79 renameat2() has an additional flags argument. A renameat2() call with
80 a zero flags argument is equivalent to renameat().
81 The flags argument is a bit mask consisting of zero or more of the fol‐
83 lowing flags:
84 RENAME_EXCHANGE
86 Atomically exchange oldpath and newpath. Both pathnames must
87 exist but may be of different types (e.g., one could be a non-
88 empty directory and the other a symbolic link).
89 RENAME_NOREPLACE
91 Don't overwrite newpath of the rename. Return an error if new‐
92 path already exists.
93 RENAME_NOREPLACE can't be employed together with
95 RENAME_EXCHANGE.
96 RENAME_NOREPLACE requires support from the underlying filesys‐
98 tem. Support for various filesystems was added as follows:
99 * ext4 (Linux 3.15);
101 * btrfs, shmem, and cifs (Linux 3.17);
103 * xfs (Linux 4.0);
105 * Support for many other filesystems was added in Linux 4.9,
107 including etx2, minix, reiserfs, jfs, vfat, and bpf.
108 RENAME_WHITEOUT (since Linux 3.18)
110 This operation makes sense only for overlay/union filesystem
111 implementations.
112 Specifying RENAME_WHITEOUT creates a "whiteout" object at the
114 source of the rename at the same time as performing the rename.
115 The whole operation is atomic, so that if the rename succeeds
116 then the whiteout will also have been created.
117 A "whiteout" is an object that has special meaning in
119 union/overlay filesystem constructs. In these constructs, mul‐
120 tiple layers exist and only the top one is ever modified. A
121 whiteout on an upper layer will effectively hide a matching file
122 in the lower layer, making it appear as if the file didn't
123 exist.
124 When a file that exists on the lower layer is renamed, the file
126 is first copied up (if not already on the upper layer) and then
127 renamed on the upper, read-write layer. At the same time, the
128 source file needs to be "whiteouted" (so that the version of the
129 source file in the lower layer is rendered invisible). The
130 whole operation needs to be done atomically.
131 When not part of a union/overlay, the whiteout appears as a
133 character device with a {0,0} device number. (Note that other
134 union/overlay implementations may employ different methods for
135 storing whiteout entries; specifically, BSD union mount employs
136 a separate inode type, DT_WHT, which, while supported by some
137 filesystems available in Linux, such as CODA and XFS, is ignored
138 by the kernel's whiteout support code, as of Linux 4.19, at
139 least.)
140 RENAME_WHITEOUT requires the same privileges as creating a
142 device node (i.e., the CAP_MKNOD capability).
143 RENAME_WHITEOUT can't be employed together with RENAME_EXCHANGE.
145 RENAME_WHITEOUT requires support from the underlying filesystem.
147 Among the filesystems that provide that support are tmpfs (since
148 Linux 3.18), ext4 (since Linux 3.18), XFS (since Linux 4.1),
149 f2fs (since Linux 4.2). btrfs (since Linux 4.7), and ubifs
150 (since Linux 4.9).
151
153 On success, zero is returned. On error, -1 is returned, and errno is
154 set appropriately.
155
157 EACCES Write permission is denied for the directory containing oldpath
158 or newpath, or, search permission is denied for one of the
159 directories in the path prefix of oldpath or newpath, or oldpath
160 is a directory and does not allow write permission (needed to
161 update the .. entry). (See also path_resolution(7).)
162 EBUSY The rename fails because oldpath or newpath is a directory that
164 is in use by some process (perhaps as current working directory,
165 or as root directory, or because it was open for reading) or is
166 in use by the system (for example as mount point), while the
167 system considers this an error. (Note that there is no require‐
168 ment to return EBUSY in such cases—there is nothing wrong with
169 doing the rename anyway—but it is allowed to return EBUSY if the
170 system cannot otherwise handle such situations.)
171 EDQUOT The user's quota of disk blocks on the filesystem has been
173 exhausted.
174 EFAULT oldpath or newpath points outside your accessible address space.
176 EINVAL The new pathname contained a path prefix of the old, or, more
178 generally, an attempt was made to make a directory a subdirec‐
179 tory of itself.
180 EISDIR newpath is an existing directory, but oldpath is not a direc‐
182 tory.
183 ELOOP Too many symbolic links were encountered in resolving oldpath or
185 newpath.
186 EMLINK oldpath already has the maximum number of links to it, or it was
188 a directory and the directory containing newpath has the maximum
189 number of links.
190 ENAMETOOLONG
192 oldpath or newpath was too long.
193 ENOENT The link named by oldpath does not exist; or, a directory compo‐
195 nent in newpath does not exist; or, oldpath or newpath is an
196 empty string.
197 ENOMEM Insufficient kernel memory was available.
199 ENOSPC The device containing the file has no room for the new directory
201 entry.
202 ENOTDIR
204 A component used as a directory in oldpath or newpath is not, in
205 fact, a directory. Or, oldpath is a directory, and newpath
206 exists but is not a directory.
207 ENOTEMPTY or EEXIST
209 newpath is a nonempty directory, that is, contains entries other
210 than "." and "..".
211 EPERM or EACCES
213 The directory containing oldpath has the sticky bit (S_ISVTX)
214 set and the process's effective user ID is neither the user ID
215 of the file to be deleted nor that of the directory containing
216 it, and the process is not privileged (Linux: does not have the
217 CAP_FOWNER capability); or newpath is an existing file and the
218 directory containing it has the sticky bit set and the process's
219 effective user ID is neither the user ID of the file to be
220 replaced nor that of the directory containing it, and the
221 process is not privileged (Linux: does not have the CAP_FOWNER
222 capability); or the filesystem containing pathname does not sup‐
223 port renaming of the type requested.
224 EROFS The file is on a read-only filesystem.
226 EXDEV oldpath and newpath are not on the same mounted filesystem.
228 (Linux permits a filesystem to be mounted at multiple points,
229 but rename() does not work across different mount points, even
230 if the same filesystem is mounted on both.)
231 The following additional errors can occur for renameat() and
233 renameat2():
234 EBADF olddirfd or newdirfd is not a valid file descriptor.
236 ENOTDIR
238 oldpath is relative and olddirfd is a file descriptor referring
239 to a file other than a directory; or similar for newpath and
240 newdirfd
241 The following additional errors can occur for renameat2():
243 EEXIST flags contains RENAME_NOREPLACE and newpath already exists.
245 EINVAL An invalid flag was specified in flags.
247 EINVAL Both RENAME_NOREPLACE and RENAME_EXCHANGE were specified in
249 flags.
250 EINVAL Both RENAME_WHITEOUT and RENAME_EXCHANGE were specified in
252 flags.
253 EINVAL The filesystem does not support one of the flags in flags.
255 ENOENT flags contains RENAME_EXCHANGE and newpath does not exist.
257 EPERM RENAME_WHITEOUT was specified in flags, but the caller does not
259 have the CAP_MKNOD capability.
260
262 renameat() was added to Linux in kernel 2.6.16; library support was
263 added to glibc in version 2.4.
264 renameat2() was added to Linux in kernel 3.15; library support was
266 added in glibc 2.28.
267
269 rename(): 4.3BSD, C89, C99, POSIX.1-2001, POSIX.1-2008.
270 renameat(): POSIX.1-2008.
272 renameat2() is Linux-specific.
274
276 Glibc notes
277 On older kernels where renameat() is unavailable, the glibc wrapper
278 function falls back to the use of rename(). When oldpath and newpath
279 are relative pathnames, glibc constructs pathnames based on the sym‐
280 bolic links in /proc/self/fd that correspond to the olddirfd and
281 newdirfd arguments.
282
284 On NFS filesystems, you can not assume that if the operation failed,
285 the file was not renamed. If the server does the rename operation and
286 then crashes, the retransmitted RPC which will be processed when the
287 server is up again causes a failure. The application is expected to
288 deal with this. See link(2) for a similar problem.
289
291 mv(1), chmod(2), link(2), symlink(2), unlink(2), path_resolution(7),
292 symlink(7)
293
295 This page is part of release 5.04 of the Linux man-pages project. A
296 description of the project, information about reporting bugs, and the
297 latest version of this page, can be found at
298 https://www.kernel.org/doc/man-pages/.
299Linux 2019-03-06 RENAME(2)