man2/rename.2

   1 .\" This manpage is Copyright (C) 1992 Drew Eckhardt;
   2 .\"             and Copyright (C) 1993 Michael Haardt;
   3 .\"             and Copyright (C) 1993,1995 Ian Jackson
   4 .\"             and Copyright (C) 2006, 2014 Michael Kerrisk
   5 .\"
   6 .\" %%%LICENSE_START(VERBATIM)
   7 .\" Permission is granted to make and distribute verbatim copies of this
   8 .\" manual provided the copyright notice and this permission notice are
   9 .\" preserved on all copies.
  10 .\"
  11 .\" Permission is granted to copy and distribute modified versions of this
  12 .\" manual under the conditions for verbatim copying, provided that the
  13 .\" entire resulting derived work is distributed under the terms of a
  14 .\" permission notice identical to this one.
  15 .\"
  16 .\" Since the Linux kernel and libraries are constantly changing, this
  17 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  18 .\" responsibility for errors or omissions, or for damages resulting from
  19 .\" the use of the information contained herein.  The author(s) may not
  20 .\" have taken the same level of care in the production of this manual,
  21 .\" which is licensed free of charge, as they might when working
  22 .\" professionally.
  23 .\"
  24 .\" Formatted or processed versions of this manual, if unaccompanied by
  25 .\" the source, must acknowledge the copyright and authors of this work.
  26 .\" %%%LICENSE_END
  27 .\"
  28 .\" Modified Sat Jul 24 00:35:52 1993 by Rik Faith <faith@cs.unc.edu>
  29 .\" Modified Thu Jun  4 12:21:13 1998 by Andries Brouwer <aeb@cwi.nl>
  30 .\" Modified Thu Mar  3 09:49:35 2005 by Michael Haardt <michael@moria.de>
  31 .\" 2007-03-25, mtk, added various text to DESCRIPTION.
  32 .\"
  33 .TH RENAME 2 2017-09-15 "Linux" "Linux Programmer's Manual"
  34 .SH NAME
  35 rename, renameat, renameat2 \- change the name or location of a file
  36 .SH SYNOPSIS
  37 .nf
  38 .B #include <stdio.h>
  39 .PP
  40 .BI "int rename(const char *" oldpath ", const char *" newpath );
  41
  42 .BR "#include <fcntl.h>           " "/* Definition of AT_* constants */"
  43 .B #include <stdio.h>
  44 .PP
  45 .BI "int renameat(int " olddirfd ", const char *" oldpath ,
  46 .BI "             int " newdirfd ", const char *" newpath );
  47 .PP
  48 .BI "int renameat2(int " olddirfd ", const char *" oldpath ,
  49 .BI "              int " newdirfd ", const char *" newpath \
  50 ", unsigned int " flags );
  51 .fi
  52 .PP
  53 .IR Note :
  54 There is no glibc wrapper for
  55 .BR renameat2 ();
  56 see NOTES.
  57 .PP
  58 .in -4n
  59 Feature Test Macro Requirements for glibc (see
  60 .BR feature_test_macros (7)):
  61 .in
  62 .PP
  63 .BR renameat ():
  64 .PD 0
  65 .ad l
  66 .RS 4
  67 .TP 4
  68 Since glibc 2.10:
  69 _POSIX_C_SOURCE\ >=\ 200809L
  70 .TP
  71 Before glibc 2.10:
  72 _ATFILE_SOURCE
  73 .\" FIXME . need to define FTMs for renameat2(), once it hits glibc
  74 .RE
  75 .ad
  76 .PD
  77 .SH DESCRIPTION
  78 .BR rename ()
  79 renames a file, moving it between directories if required.
  80 Any other hard links to the file (as created using
  81 .BR link (2))
  82 are unaffected.
  83 Open file descriptors for
  84 .I oldpath
  85 are also unaffected.
  86 .PP
  87 Various restrictions determine whether or not the rename operation succeeds:
  88 see ERRORS below.
  89 .PP
  90 If
  91 .I newpath
  92 already exists, it will be atomically replaced, so that there is
  93 no point at which another process attempting to access
  94 .I newpath
  95 will find it missing.
  96 However, there will probably be a window in which both
  97 .I oldpath
  98 and
  99 .I newpath
 100 refer to the file being renamed.
 101 .PP
 102 If
 103 .I oldpath
 104 and
 105 .I newpath
 106 are existing hard links referring to the same file, then
 107 .BR rename ()
 108 does nothing, and returns a success status.
 109 .PP
 110 If
 111 .I newpath
 112 exists but the operation fails for some reason,
 113 .BR rename ()
 114 guarantees to leave an instance of
 115 .I newpath
 116 in place.
 117 .PP
 118 .I oldpath
 119 can specify a directory.
 120 In this case,
 121 .I newpath
 122 must either not exist, or it must specify an empty directory.
 123 .PP
 124 If
 125 .I oldpath
 126 refers to a symbolic link, the link is renamed; if
 127 .I newpath
 128 refers to a symbolic link, the link will be overwritten.
 129 .SS renameat()
 130 The
 131 .BR renameat ()
 132 system call operates in exactly the same way as
 133 .BR rename (),
 134 except for the differences described here.
 135 .PP
 136 If the pathname given in
 137 .I oldpath
 138 is relative, then it is interpreted relative to the directory
 139 referred to by the file descriptor
 140 .I olddirfd
 141 (rather than relative to the current working directory of
 142 the calling process, as is done by
 143 .BR rename ()
 144 for a relative pathname).
 145 .PP
 146 If
 147 .I oldpath
 148 is relative and
 149 .I olddirfd
 150 is the special value
 151 .BR AT_FDCWD ,
 152 then
 153 .I oldpath
 154 is interpreted relative to the current working
 155 directory of the calling process (like
 156 .BR rename ()).
 157 .PP
 158 If
 159 .I oldpath
 160 is absolute, then
 161 .I olddirfd
 162 is ignored.
 163 .PP
 164 The interpretation of
 165 .I newpath
 166 is as for
 167 .IR oldpath ,
 168 except that a relative pathname is interpreted relative
 169 to the directory referred to by the file descriptor
 170 .IR newdirfd .
 171 .PP
 172 See
 173 .BR openat (2)
 174 for an explanation of the need for
 175 .BR renameat ().
 176 .SS renameat2()
 177 .BR renameat2 ()
 178 has an additional
 179 .I flags
 180 argument.
 181 A
 182 .BR renameat2 ()
 183 call with a zero
 184 .I flags
 185 argument is equivalent to
 186 .BR renameat ().
 187 .PP
 188 The
 189 .I flags
 190 argument is a bit mask consisting of zero or more of the following flags:
 191 .TP
 192 .B RENAME_EXCHANGE
 193 Atomically exchange
 194 .IR oldpath
 195 and
 196 .IR newpath .
 197 Both pathnames must exist
 198 but may be of different types (e.g., one could be a non-empty directory
 199 and the other a symbolic link).
 200 .TP
 201 .B RENAME_NOREPLACE
 202 Don't overwrite
 203 .IR newpath
 204 of the rename.
 205 Return an error if
 206 .IR newpath
 207 already exists.
 208 .IP
 209 .B RENAME_NOREPLACE
 210 can't be employed together with
 211 .BR RENAME_EXCHANGE .
 212 .TP
 213 .BR RENAME_WHITEOUT " (since Linux 3.18)"
 214 .\" commit 0d7a855526dd672e114aff2ac22b60fc6f155b08
 215 .\" commit 787fb6bc9682ec7c05fb5d9561b57100fbc1cc41
 216 This operation makes sense only for overlay/union
 217 filesystem implementations.
 218 .IP
 219 Specifying
 220 .B RENAME_WHITEOUT
 221 creates a "whiteout" object at the source of
 222 the rename at the same time as performing the rename.
 223 The whole operation is atomic,
 224 so that if the rename succeeds then the whiteout will also have been created.
 225 .IP
 226 A "whiteout" is an object that has special meaning in union/overlay
 227 filesystem constructs.
 228 In these constructs,
 229 multiple layers exist and only the top one is ever modified.
 230 A whiteout on an upper layer will effectively hide a
 231 matching file in the lower layer,
 232 making it appear as if the file didn't exist.
 233 .IP
 234 When a file that exists on the lower layer is renamed,
 235 the file is first copied up (if not already on the upper layer)
 236 and then renamed on the upper, read-write layer.
 237 At the same time, the source file needs to be "whiteouted"
 238 (so that the version of the source file in the lower layer
 239 is rendered invisible).
 240 The whole operation needs to be done atomically.
 241 .IP
 242 When not part of a union/overlay,
 243 the whiteout appears as a character device with a {0,0} device number.
 244 .IP
 245 .B RENAME_WHITEOUT
 246 requires the same privileges as creating a device node (i.e., the
 247 .BR CAP_MKNOD
 248 capability).
 249 .IP
 250 .B RENAME_WHITEOUT
 251 can't be employed together with
 252 .BR RENAME_EXCHANGE .
 253 .IP
 254 .B RENAME_WHITEOUT
 255 requires support from the underlying filesystem.
 256 Among the filesystems that provide that support are
 257 shmem (since Linux 3.18),
 258 .\" shmem: commit 46fdb794e3f52ef18b859ebc92f0a9d7db21c5df
 259 ext4 (since Linux 3.18),
 260 .\" ext4: commit cd808deced431b66b5fa4e5c193cb7ec0059eaff
 261 and XFS (since Linux 4.1).
 262 .\" XFS: commit 7dcf5c3e4527cfa2807567b00387cf2ed5e07f00
 263 .SH RETURN VALUE
 264 On success, zero is returned.
 265 On error, \-1 is returned, and
 266 .I errno
 267 is set appropriately.
 268 .SH ERRORS
 269 .TP
 270 .B EACCES
 271 Write permission is denied for the directory containing
 272 .I oldpath
 273 or
 274 .IR newpath ,
 275 or, search permission is denied for one of the directories
 276 in the path prefix of
 277 .I oldpath
 278 or
 279 .IR newpath ,
 280 or
 281 .I oldpath
 282 is a directory and does not allow write permission (needed to update
 283 the
 284 .I ..
 285 entry).
 286 (See also
 287 .BR path_resolution (7).)
 288 .TP
 289 .B EBUSY
 290 The rename fails because
 291 .IR oldpath " or " newpath
 292 is a directory that is in use by some process (perhaps as
 293 current working directory, or as root directory, or because
 294 it was open for reading) or is in use by the system
 295 (for example as mount point), while the system considers
 296 this an error.
 297 (Note that there is no requirement to return
 298 .B EBUSY
 299 in such
 300 cases\(emthere is nothing wrong with doing the rename anyway\(embut
 301 it is allowed to return
 302 .B EBUSY
 303 if the system cannot otherwise
 304 handle such situations.)
 305 .TP
 306 .B EDQUOT
 307 The user's quota of disk blocks on the filesystem has been exhausted.
 308 .TP
 309 .B EFAULT
 310 .IR oldpath " or " newpath " points outside your accessible address space."
 311 .TP
 312 .B EINVAL
 313 The new pathname contained a path prefix of the old, or, more generally,
 314 an attempt was made to make a directory a subdirectory of itself.
 315 .TP
 316 .B EISDIR
 317 .I newpath
 318 is an existing directory, but
 319 .I oldpath
 320 is not a directory.
 321 .TP
 322 .B ELOOP
 323 Too many symbolic links were encountered in resolving
 324 .IR oldpath " or " newpath .
 325 .TP
 326 .B EMLINK
 327 .I oldpath
 328 already has the maximum number of links to it, or
 329 it was a directory and the directory containing
 330 .I newpath
 331 has the maximum number of links.
 332 .TP
 333 .B ENAMETOOLONG
 334 .IR oldpath " or " newpath " was too long."
 335 .TP
 336 .B ENOENT
 337 The link named by
 338 .I oldpath
 339 does not exist;
 340 or, a directory component in
 341 .I newpath
 342 does not exist;
 343 or,
 344 .I oldpath
 345 or
 346 .I newpath
 347 is an empty string.
 348 .TP
 349 .B ENOMEM
 350 Insufficient kernel memory was available.
 351 .TP
 352 .B ENOSPC
 353 The device containing the file has no room for the new directory
 354 entry.
 355 .TP
 356 .B ENOTDIR
 357 A component used as a directory in
 358 .IR oldpath " or " newpath
 359 is not, in fact, a directory.
 360 Or,
 361 .I oldpath
 362 is a directory, and
 363 .I newpath
 364 exists but is not a directory.
 365 .TP
 366 .BR ENOTEMPTY " or " EEXIST
 367 .I newpath
 368 is a nonempty directory, that is, contains entries other than "." and "..".
 369 .TP
 370 .BR EPERM " or " EACCES
 371 The directory containing
 372 .I oldpath
 373 has the sticky bit
 374 .RB ( S_ISVTX )
 375 set and the process's effective user ID is neither
 376 the user ID of the file to be deleted nor that of the directory
 377 containing it, and the process is not privileged
 378 (Linux: does not have the
 379 .B CAP_FOWNER
 380 capability);
 381 or
 382 .I newpath
 383 is an existing file and the directory containing it has the sticky bit set
 384 and the process's effective user ID is neither the user ID of the file
 385 to be replaced nor that of the directory containing it,
 386 and the process is not privileged
 387 (Linux: does not have the
 388 .B CAP_FOWNER
 389 capability);
 390 or the filesystem containing
 391 .I pathname
 392 does not support renaming of the type requested.
 393 .TP
 394 .B EROFS
 395 The file is on a read-only filesystem.
 396 .TP
 397 .B EXDEV
 398 .IR oldpath " and " newpath
 399 are not on the same mounted filesystem.
 400 (Linux permits a filesystem to be mounted at multiple points, but
 401 .BR rename ()
 402 does not work across different mount points,
 403 even if the same filesystem is mounted on both.)
 404 .PP
 405 The following additional errors can occur for
 406 .BR renameat ()
 407 and
 408 .BR renameat2 ():
 409 .TP
 410 .B EBADF
 411 .I olddirfd
 412 or
 413 .I newdirfd
 414 is not a valid file descriptor.
 415 .TP
 416 .B ENOTDIR
 417 .I oldpath
 418 is relative and
 419 .I olddirfd
 420 is a file descriptor referring to a file other than a directory;
 421 or similar for
 422 .I newpath
 423 and
 424 .I newdirfd
 425 .PP
 426 The following additional errors can occur for
 427 .BR renameat2 ():
 428 .TP
 429 .B EEXIST
 430 .I flags
 431 contains
 432 .B RENAME_NOREPLACE
 433 and
 434 .I newpath
 435 already exists.
 436 .TP
 437 .B EINVAL
 438 An invalid flag was specified in
 439 .IR flags .
 440 .TP
 441 .B EINVAL
 442 Both
 443 .B RENAME_NOREPLACE
 444 and
 445 .B RENAME_EXCHANGE
 446 were specified in
 447 .IR flags .
 448 .TP
 449 .B EINVAL
 450 Both
 451 .B RENAME_WHITEOUT
 452 and
 453 .B RENAME_EXCHANGE
 454 were specified in
 455 .IR flags .
 456 .TP
 457 .B EINVAL
 458 The filesystem does not support one of the flags in
 459 .IR flags .
 460 .TP
 461 .B ENOENT
 462 .I flags
 463 contains
 464 .B RENAME_EXCHANGE
 465 and
 466 .IR newpath
 467 does not exist.
 468 .TP
 469 .B EPERM
 470 .B RENAME_WHITEOUT
 471 was specified in
 472 .IR flags ,
 473 but the caller does not have the
 474 .B CAP_MKNOD
 475 capability.
 476 .SH VERSIONS
 477 .BR renameat ()
 478 was added to Linux in kernel 2.6.16;
 479 library support was added to glibc in version 2.4.
 480 .PP
 481 .BR renameat2 ()
 482 was added to Linux in kernel 3.15; library support was added in glibc 2.28.
 483 .SH CONFORMING TO
 484 .BR rename ():
 485 4.3BSD, C89, C99, POSIX.1-2001, POSIX.1-2008.
 486 .PP
 487 .BR renameat ():
 488 POSIX.1-2008.
 489 .PP
 490 .BR renameat2 ()
 491 is Linux-specific.
 492 .SH NOTES
 493 Glibc does not provide a wrapper for the
 494 .BR renameat2 ()
 495 system call; call it using
 496 .BR syscall (2).
 497 .\"
 498 .SS Glibc notes
 499 On older kernels where
 500 .BR renameat ()
 501 is unavailable, the glibc wrapper function falls back to the use of
 502 .BR rename ().
 503 When
 504 .I oldpath
 505 and
 506 .I newpath
 507 are relative pathnames,
 508 glibc constructs pathnames based on the symbolic links in
 509 .IR /proc/self/fd
 510 that correspond to the
 511 .I olddirfd
 512 and
 513 .IR newdirfd
 514 arguments.
 515 .SH BUGS
 516 On NFS filesystems, you can not assume that if the operation
 517 failed, the file was not renamed.
 518 If the server does the rename operation
 519 and then crashes, the retransmitted RPC which will be processed when the
 520 server is up again causes a failure.
 521 The application is expected to
 522 deal with this.
 523 See
 524 .BR link (2)
 525 for a similar problem.
 526 .SH SEE ALSO
 527 .BR mv (1),
 528 .BR chmod (2),
 529 .BR link (2),
 530 .BR symlink (2),
 531 .BR unlink (2),
 532 .BR path_resolution (7),
 533 .BR symlink (7)