1access(2) System Calls Manual access(2)
2
6 access, faccessat, faccessat2 - check user's permissions for a file
7
9 Standard C library (libc, -lc)
10
12 #include <unistd.h>
13 int access(const char *pathname, int mode);
15 #include <fcntl.h> /* Definition of AT_* constants */
17 #include <unistd.h>
18 int faccessat(int dirfd, const char *pathname, int mode, int flags);
20 /* But see C library/kernel differences, below */
21 #include <fcntl.h> /* Definition of AT_* constants */
23 #include <sys/syscall.h> /* Definition of SYS_* constants */
24 #include <unistd.h>
25 int syscall(SYS_faccessat2,
27 int dirfd, const char *pathname, int mode, int flags);
28 Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
30 faccessat():
32 Since glibc 2.10:
33 _POSIX_C_SOURCE >= 200809L
34 Before glibc 2.10:
35 _ATFILE_SOURCE
36
38 access() checks whether the calling process can access the file path‐
39 name. If pathname is a symbolic link, it is dereferenced.
40 The mode specifies the accessibility check(s) to be performed, and is
42 either the value F_OK, or a mask consisting of the bitwise OR of one or
43 more of R_OK, W_OK, and X_OK. F_OK tests for the existence of the
44 file. R_OK, W_OK, and X_OK test whether the file exists and grants
45 read, write, and execute permissions, respectively.
46 The check is done using the calling process's real UID and GID, rather
48 than the effective IDs as is done when actually attempting an operation
49 (e.g., open(2)) on the file. Similarly, for the root user, the check
50 uses the set of permitted capabilities rather than the set of effective
51 capabilities; and for non-root users, the check uses an empty set of
52 capabilities.
53 This allows set-user-ID programs and capability-endowed programs to
55 easily determine the invoking user's authority. In other words, ac‐
56 cess() does not answer the "can I read/write/execute this file?" ques‐
57 tion. It answers a slightly different question: "(assuming I'm a se‐
58 tuid binary) can the user who invoked me read/write/execute this
59 file?", which gives set-user-ID programs the possibility to prevent ma‐
60 licious users from causing them to read files which users shouldn't be
61 able to read.
62 If the calling process is privileged (i.e., its real UID is zero), then
64 an X_OK check is successful for a regular file if execute permission is
65 enabled for any of the file owner, group, or other.
66 faccessat()
68 faccessat() operates in exactly the same way as access(), except for
69 the differences described here.
70 If the pathname given in pathname is relative, then it is interpreted
72 relative to the directory referred to by the file descriptor dirfd
73 (rather than relative to the current working directory of the calling
74 process, as is done by access() for a relative pathname).
75 If pathname is relative and dirfd is the special value AT_FDCWD, then
77 pathname is interpreted relative to the current working directory of
78 the calling process (like access()).
79 If pathname is absolute, then dirfd is ignored.
81 flags is constructed by ORing together zero or more of the following
83 values:
84 AT_EACCESS
86 Perform access checks using the effective user and group IDs.
87 By default, faccessat() uses the real IDs (like access()).
88 AT_SYMLINK_NOFOLLOW
90 If pathname is a symbolic link, do not dereference it: instead
91 return information about the link itself.
92 See openat(2) for an explanation of the need for faccessat().
94 faccessat2()
96 The description of faccessat() given above corresponds to POSIX.1 and
97 to the implementation provided by glibc. However, the glibc implemen‐
98 tation was an imperfect emulation (see BUGS) that papered over the fact
99 that the raw Linux faccessat() system call does not have a flags argu‐
100 ment. To allow for a proper implementation, Linux 5.8 added the fac‐
101 cessat2() system call, which supports the flags argument and allows a
102 correct implementation of the faccessat() wrapper function.
103
105 On success (all requested permissions granted, or mode is F_OK and the
106 file exists), zero is returned. On error (at least one bit in mode
107 asked for a permission that is denied, or mode is F_OK and the file
108 does not exist, or some other error occurred), -1 is returned, and er‐
109 rno is set to indicate the error.
110
112 EACCES The requested access would be denied to the file, or search per‐
113 mission is denied for one of the directories in the path prefix
114 of pathname. (See also path_resolution(7).)
115 EBADF (faccessat()) pathname is relative but dirfd is neither AT_FDCWD
117 (faccessat()) nor a valid file descriptor.
118 EFAULT pathname points outside your accessible address space.
120 EINVAL mode was incorrectly specified.
122 EINVAL (faccessat()) Invalid flag specified in flags.
124 EIO An I/O error occurred.
126 ELOOP Too many symbolic links were encountered in resolving pathname.
128 ENAMETOOLONG
130 pathname is too long.
131 ENOENT A component of pathname does not exist or is a dangling symbolic
133 link.
134 ENOMEM Insufficient kernel memory was available.
136 ENOTDIR
138 A component used as a directory in pathname is not, in fact, a
139 directory.
140 ENOTDIR
142 (faccessat()) pathname is relative and dirfd is a file descrip‐
143 tor referring to a file other than a directory.
144 EPERM Write permission was requested to a file that has the immutable
146 flag set. See also ioctl_iflags(2).
147 EROFS Write permission was requested for a file on a read-only
149 filesystem.
150 ETXTBSY
152 Write access was requested to an executable which is being exe‐
153 cuted.
154
156 If the calling process has appropriate privileges (i.e., is superuser),
157 POSIX.1-2001 permits an implementation to indicate success for an X_OK
158 check even if none of the execute file permission bits are set. Linux
159 does not do this.
160 C library/kernel differences
162 The raw faccessat() system call takes only the first three arguments.
163 The AT_EACCESS and AT_SYMLINK_NOFOLLOW flags are actually implemented
164 within the glibc wrapper function for faccessat(). If either of these
165 flags is specified, then the wrapper function employs fstatat(2) to de‐
166 termine access permissions, but see BUGS.
167 glibc notes
169 On older kernels where faccessat() is unavailable (and when the AT_EAC‐
170 CESS and AT_SYMLINK_NOFOLLOW flags are not specified), the glibc wrap‐
171 per function falls back to the use of access(). When pathname is a
172 relative pathname, glibc constructs a pathname based on the symbolic
173 link in /proc/self/fd that corresponds to the dirfd argument.
174
176 access()
177 faccessat()
178 POSIX.1-2008.
179 faccessat2()
181 Linux.
182
184 access()
185 SVr4, 4.3BSD, POSIX.1-2001.
186 faccessat()
188 Linux 2.6.16, glibc 2.4.
189 faccessat2()
191 Linux 5.8.
192
194 Warning: Using these calls to check if a user is authorized to, for ex‐
195 ample, open a file before actually doing so using open(2) creates a se‐
196 curity hole, because the user might exploit the short time interval be‐
197 tween checking and opening the file to manipulate it. For this reason,
198 the use of this system call should be avoided. (In the example just
199 described, a safer alternative would be to temporarily switch the
200 process's effective user ID to the real ID and then call open(2).)
201 access() always dereferences symbolic links. If you need to check the
203 permissions on a symbolic link, use faccessat() with the flag AT_SYM‐
204 LINK_NOFOLLOW.
205 These calls return an error if any of the access types in mode is de‐
207 nied, even if some of the other access types in mode are permitted.
208 A file is accessible only if the permissions on each of the directories
210 in the path prefix of pathname grant search (i.e., execute) access. If
211 any directory is inaccessible, then the access() call fails, regardless
212 of the permissions on the file itself.
213 Only access bits are checked, not the file type or contents. There‐
215 fore, if a directory is found to be writable, it probably means that
216 files can be created in the directory, and not that the directory can
217 be written as a file. Similarly, a DOS file may be reported as exe‐
218 cutable, but the execve(2) call will still fail.
219 These calls may not work correctly on NFSv2 filesystems with UID map‐
221 ping enabled, because UID mapping is done on the server and hidden from
222 the client, which checks permissions. (NFS versions 3 and higher per‐
223 form the check on the server.) Similar problems can occur to FUSE
224 mounts.
225
227 Because the Linux kernel's faccessat() system call does not support a
228 flags argument, the glibc faccessat() wrapper function provided in
229 glibc 2.32 and earlier emulates the required functionality using a com‐
230 bination of the faccessat() system call and fstatat(2). However, this
231 emulation does not take ACLs into account. Starting with glibc 2.33,
232 the wrapper function avoids this bug by making use of the faccessat2()
233 system call where it is provided by the underlying kernel.
234 In Linux 2.4 (and earlier) there is some strangeness in the handling of
236 X_OK tests for superuser. If all categories of execute permission are
237 disabled for a nondirectory file, then the only access() test that re‐
238 turns -1 is when mode is specified as just X_OK; if R_OK or W_OK is
239 also specified in mode, then access() returns 0 for such files. Early
240 Linux 2.6 (up to and including Linux 2.6.3) also behaved in the same
241 way as Linux 2.4.
242 Before Linux 2.6.20, these calls ignored the effect of the MS_NOEXEC
244 flag if it was used to mount(2) the underlying filesystem. Since Linux
245 2.6.20, the MS_NOEXEC flag is honored.
246
248 chmod(2), chown(2), open(2), setgid(2), setuid(2), stat(2), euidac‐
249 cess(3), credentials(7), path_resolution(7), symlink(7)
250Linux man-pages 6.04 2023-03-30 access(2)