1 .\" This manpage is Copyright (C) 1992 Drew Eckhardt;
2 .\" and Copyright (C) 1993 Michael Haardt, Ian Jackson.
3 .\" and Copyright (C) 2004, 2006, 2007, 2014 Michael Kerrisk <mtk.manpages@gmail.com>
5 .\" %%%LICENSE_START(VERBATIM)
6 .\" Permission is granted to make and distribute verbatim copies of this
7 .\" manual provided the copyright notice and this permission notice are
8 .\" preserved on all copies.
10 .\" Permission is granted to copy and distribute modified versions of this
11 .\" manual under the conditions for verbatim copying, provided that the
12 .\" entire resulting derived work is distributed under the terms of a
13 .\" permission notice identical to this one.
15 .\" Since the Linux kernel and libraries are constantly changing, this
16 .\" manual page may be incorrect or out-of-date. The author(s) assume no
17 .\" responsibility for errors or omissions, or for damages resulting from
18 .\" the use of the information contained herein. The author(s) may not
19 .\" have taken the same level of care in the production of this manual,
20 .\" which is licensed free of charge, as they might when working
23 .\" Formatted or processed versions of this manual, if unaccompanied by
24 .\" the source, must acknowledge the copyright and authors of this work.
27 .\" Modified 1993-07-21 Rik Faith (faith@cs.unc.edu)
28 .\" Modified 1994-08-21 by Michael Chastain (mec@shell.portal.com):
29 .\" Removed note about old kernel (pre-1.1.44) using wrong id on path.
30 .\" Modified 1996-03-18 by Martin Schulze (joey@infodrom.north.de):
31 .\" Stated more clearly how it behaves with symbolic links.
32 .\" Added correction due to Nick Duffek (nsd@bbc.com), aeb, 960426
33 .\" Modified 1996-09-07 by Michael Haardt:
34 .\" Restrictions for NFS
35 .\" Modified 1997-09-09 by Joseph S. Myers <jsm28@cam.ac.uk>
36 .\" Modified 1998-01-13 by Michael Haardt:
37 .\" Using access is often insecure
38 .\" Modified 2001-10-16 by aeb
39 .\" Modified 2002-04-23 by Roger Luethi <rl@hellgate.ch>
40 .\" Modified 2004-06-23 by Michael Kerrisk
41 .\" 2007-06-10, mtk, various parts rewritten, and added BUGS section.
43 .TH ACCESS 2 2021-03-22 "Linux" "Linux Programmer's Manual"
45 access, faccessat, faccessat2 \- check user's permissions for a file
48 .B #include <unistd.h>
50 .BI "int access(const char *" pathname ", int " mode );
52 .BR "#include <fcntl.h>" " /* Definition of " AT_* " constants */"
53 .B #include <unistd.h>
55 .BI "int faccessat(int " dirfd ", const char *" pathname ", int " \
56 mode ", int " flags );
57 /* But see C library/kernel differences, below */
59 .BR "#include <fcntl.h>" " /* Definition of " AT_* " constants */"
60 .BR "#include <sys/syscall.h>" " /* Definition of " SYS_* " constants */"
61 .B #include <unistd.h>
63 .BI "int syscall(SYS_faccessat2,"
64 .BI " int " dirfd ", const char *" pathname ", int " mode \
69 Feature Test Macro Requirements for glibc (see
70 .BR feature_test_macros (7)):
76 _POSIX_C_SOURCE >= 200809L
82 checks whether the calling process can access the file
86 is a symbolic link, it is dereferenced.
90 specifies the accessibility check(s) to be performed,
91 and is either the value
93 .\" F_OK is defined as 0 on every system that I know of.
94 or a mask consisting of the bitwise OR of one or more of
95 .BR R_OK ", " W_OK ", and " X_OK .
97 tests for the existence of the file.
98 .BR R_OK ", " W_OK ", and " X_OK
99 test whether the file exists and grants read, write, and
100 execute permissions, respectively.
102 The check is done using the calling process's
104 UID and GID, rather than the effective IDs as is done when
105 actually attempting an operation (e.g.,
108 Similarly, for the root user, the check uses the set of
109 permitted capabilities rather than the set of effective
110 capabilities; and for non-root users, the check uses an empty set
113 This allows set-user-ID programs and capability-endowed programs
114 to easily determine the invoking user's authority.
117 does not answer the "can I read/write/execute this file?" question.
118 It answers a slightly different question:
119 "(assuming I'm a setuid binary) can
120 .I the user who invoked me
121 read/write/execute this file?",
122 which gives set-user-ID programs the possibility to
123 prevent malicious users from causing them to read files
124 which users shouldn't be able to read.
126 If the calling process is privileged (i.e., its real UID is zero),
129 check is successful for a regular file if execute permission
130 is enabled for any of the file owner, group, or other.
133 operates in exactly the same way as
135 except for the differences described here.
137 If the pathname given in
139 is relative, then it is interpreted relative to the directory
140 referred to by the file descriptor
142 (rather than relative to the current working directory of
143 the calling process, as is done by
145 for a relative pathname).
155 is interpreted relative to the current working
156 directory of the calling process (like
166 is constructed by ORing together zero or more of the following values:
169 Perform access checks using the effective user and group IDs.
172 uses the real IDs (like
175 .B AT_SYMLINK_NOFOLLOW
178 is a symbolic link, do not dereference it:
179 instead return information about the link itself.
183 for an explanation of the need for
189 given above corresponds to POSIX.1 and
190 to the implementation provided by glibc.
191 However, the glibc implementation was an imperfect emulation (see BUGS)
192 that papered over the fact that the raw Linux
194 system call does not have a
197 To allow for a proper implementation, Linux 5.8 added the
199 system call, which supports the
201 argument and allows a correct implementation of the
205 On success (all requested permissions granted, or
209 and the file exists), zero is returned.
210 On error (at least one bit in
212 asked for a permission that is denied, or
216 and the file does not exist, or some other error occurred),
219 is set to indicate the error.
227 The requested access would be denied to the file, or search permission
228 is denied for one of the directories in the path prefix of
231 .BR path_resolution (7).)
234 Too many symbolic links were encountered in resolving
244 does not exist or is a dangling symbolic link.
247 A component used as a directory in
249 is not, in fact, a directory.
252 Write permission was requested for a file on a read-only filesystem.
261 points outside your accessible address space.
265 was incorrectly specified.
268 An I/O error occurred.
271 Insufficient kernel memory was available.
274 Write access was requested to an executable which is being
277 The following additional errors can occur for
282 is not a valid file descriptor.
285 Invalid flag specified in
292 is a file descriptor referring to a file other than a directory.
295 was added to Linux in kernel 2.6.16;
296 library support was added to glibc in version 2.4.
299 was added to Linux in version 5.8.
302 SVr4, 4.3BSD, POSIX.1-2001, POSIX.1-2008.
311 Using these calls to check if a user is authorized to, for example,
312 open a file before actually doing so using
314 creates a security hole, because the user might exploit the short time
315 interval between checking and opening the file to manipulate it.
316 .BR "For this reason, the use of this system call should be avoided" .
317 (In the example just described,
318 a safer alternative would be to temporarily switch the process's
319 effective user ID to the real ID and then call
323 always dereferences symbolic links.
324 If you need to check the permissions on a symbolic link, use
327 .BR AT_SYMLINK_NOFOLLOW .
329 These calls return an error if any of the access types in
331 is denied, even if some of the other access types in
335 If the calling process has appropriate privileges (i.e., is superuser),
336 POSIX.1-2001 permits an implementation to indicate success for an
338 check even if none of the execute file permission bits are set.
339 .\" HPU-UX 11 and Tru64 5.1 do this.
340 Linux does not do this.
342 A file is accessible only if the permissions on each of the
343 directories in the path prefix of
345 grant search (i.e., execute) access.
346 If any directory is inaccessible, then the
348 call fails, regardless of the permissions on the file itself.
350 Only access bits are checked, not the file type or contents.
351 Therefore, if a directory is found to be writable,
352 it probably means that files can be created in the directory,
353 and not that the directory can be written as a file.
354 Similarly, a DOS file may be reported as executable, but the
356 call will still fail.
359 may not work correctly on NFSv2 filesystems with UID mapping enabled,
360 because UID mapping is done on the server and hidden from the client,
361 which checks permissions. (NFS versions 3 and higher perform the check on
363 Similar problems can occur to FUSE mounts.
366 .SS C library/kernel differences
369 system call takes only the first three arguments.
373 .B AT_SYMLINK_NOFOLLOW
374 flags are actually implemented within the glibc wrapper function for
376 If either of these flags is specified, then the wrapper function employs
378 to determine access permissions, but see BUGS.
381 On older kernels where
383 is unavailable (and when the
386 .B AT_SYMLINK_NOFOLLOW
387 flags are not specified),
388 the glibc wrapper function falls back to the use of
392 is a relative pathname,
393 glibc constructs a pathname based on the symbolic link in
395 that corresponds to the
399 Because the Linux kernel's
401 system call does not support a
405 wrapper function provided in glibc 2.32 and earlier
406 emulates the required functionality using
411 However, this emulation does not take ACLs into account.
412 Starting with glibc 2.33, the wrapper function avoids this bug
415 system call where it is provided by the underlying kernel.
417 In kernel 2.4 (and earlier) there is some strangeness in the handling of
420 If all categories of execute permission are disabled
421 for a nondirectory file, then the only
423 test that returns \-1 is when
435 returns 0 for such files.
436 .\" This behavior appears to have been an implementation accident.
437 Early 2.6 kernels (up to and including 2.6.3)
438 also behaved in the same way as kernel 2.4.
440 In kernels before 2.6.20,
441 these calls ignored the effect of the
443 flag if it was used to
445 the underlying filesystem.
446 Since kernel 2.6.20, the
458 .BR path_resolution (7),