1ACCESS(2) Linux Programmer's Manual ACCESS(2)
2
6 access, faccessat, faccessat2 - check user's permissions for a file
7
9 #include <unistd.h>
10 int access(const char *pathname, int mode);
12 #include <fcntl.h> /* Definition of AT_* constants */
14 #include <unistd.h>
15 int faccessat(int dirfd, const char *pathname, int mode, int flags);
17 /* But see C library/kernel differences, below */
18 #include <fcntl.h> /* Definition of AT_* constants */
20 #include <sys/syscall.h> /* Definition of SYS_* constants */
21 #include <unistd.h>
22 int syscall(SYS_faccessat2,
24 int dirfd, const char *pathname, int mode, int flags);
25 Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
27 faccessat():
29 Since glibc 2.10:
30 _POSIX_C_SOURCE >= 200809L
31 Before glibc 2.10:
32 _ATFILE_SOURCE
33
35 access() checks whether the calling process can access the file path‐
36 name. If pathname is a symbolic link, it is dereferenced.
37 The mode specifies the accessibility check(s) to be performed, and is
39 either the value F_OK, or a mask consisting of the bitwise OR of one or
40 more of R_OK, W_OK, and X_OK. F_OK tests for the existence of the
41 file. R_OK, W_OK, and X_OK test whether the file exists and grants
42 read, write, and execute permissions, respectively.
43 The check is done using the calling process's real UID and GID, rather
45 than the effective IDs as is done when actually attempting an operation
46 (e.g., open(2)) on the file. Similarly, for the root user, the check
47 uses the set of permitted capabilities rather than the set of effective
48 capabilities; and for non-root users, the check uses an empty set of
49 capabilities.
50 This allows set-user-ID programs and capability-endowed programs to
52 easily determine the invoking user's authority. In other words, ac‐
53 cess() does not answer the "can I read/write/execute this file?" ques‐
54 tion. It answers a slightly different question: "(assuming I'm a se‐
55 tuid binary) can the user who invoked me read/write/execute this
56 file?", which gives set-user-ID programs the possibility to prevent ma‐
57 licious users from causing them to read files which users shouldn't be
58 able to read.
59 If the calling process is privileged (i.e., its real UID is zero), then
61 an X_OK check is successful for a regular file if execute permission is
62 enabled for any of the file owner, group, or other.
63 faccessat()
65 faccessat() operates in exactly the same way as access(), except for
66 the differences described here.
67 If the pathname given in pathname is relative, then it is interpreted
69 relative to the directory referred to by the file descriptor dirfd
70 (rather than relative to the current working directory of the calling
71 process, as is done by access() for a relative pathname).
72 If pathname is relative and dirfd is the special value AT_FDCWD, then
74 pathname is interpreted relative to the current working directory of
75 the calling process (like access()).
76 If pathname is absolute, then dirfd is ignored.
78 flags is constructed by ORing together zero or more of the following
80 values:
81 AT_EACCESS
83 Perform access checks using the effective user and group IDs.
84 By default, faccessat() uses the real IDs (like access()).
85 AT_SYMLINK_NOFOLLOW
87 If pathname is a symbolic link, do not dereference it: instead
88 return information about the link itself.
89 See openat(2) for an explanation of the need for faccessat().
91 faccessat2()
93 The description of faccessat() given above corresponds to POSIX.1 and
94 to the implementation provided by glibc. However, the glibc implemen‐
95 tation was an imperfect emulation (see BUGS) that papered over the fact
96 that the raw Linux faccessat() system call does not have a flags argu‐
97 ment. To allow for a proper implementation, Linux 5.8 added the fac‐
98 cessat2() system call, which supports the flags argument and allows a
99 correct implementation of the faccessat() wrapper function.
100
102 On success (all requested permissions granted, or mode is F_OK and the
103 file exists), zero is returned. On error (at least one bit in mode
104 asked for a permission that is denied, or mode is F_OK and the file
105 does not exist, or some other error occurred), -1 is returned, and er‐
106 rno is set to indicate the error.
107
109 access() and faccessat() shall fail if:
110 EACCES The requested access would be denied to the file, or search per‐
112 mission is denied for one of the directories in the path prefix
113 of pathname. (See also path_resolution(7).)
114 ELOOP Too many symbolic links were encountered in resolving pathname.
116 ENAMETOOLONG
118 pathname is too long.
119 ENOENT A component of pathname does not exist or is a dangling symbolic
121 link.
122 ENOTDIR
124 A component used as a directory in pathname is not, in fact, a
125 directory.
126 EROFS Write permission was requested for a file on a read-only
128 filesystem.
129 access() and faccessat() may fail if:
131 EFAULT pathname points outside your accessible address space.
133 EINVAL mode was incorrectly specified.
135 EIO An I/O error occurred.
137 ENOMEM Insufficient kernel memory was available.
139 ETXTBSY
141 Write access was requested to an executable which is being exe‐
142 cuted.
143 The following additional errors can occur for faccessat():
145 EBADF dirfd is not a valid file descriptor.
147 EINVAL Invalid flag specified in flags.
149 ENOTDIR
151 pathname is relative and dirfd is a file descriptor referring to
152 a file other than a directory.
153
155 faccessat() was added to Linux in kernel 2.6.16; library support was
156 added to glibc in version 2.4.
157 faccessat2() was added to Linux in version 5.8.
159
161 access(): SVr4, 4.3BSD, POSIX.1-2001, POSIX.1-2008.
162 faccessat(): POSIX.1-2008.
164 faccessat2(): Linux-specific.
166
168 Warning: Using these calls to check if a user is authorized to, for ex‐
169 ample, open a file before actually doing so using open(2) creates a se‐
170 curity hole, because the user might exploit the short time interval be‐
171 tween checking and opening the file to manipulate it. For this reason,
172 the use of this system call should be avoided. (In the example just
173 described, a safer alternative would be to temporarily switch the
174 process's effective user ID to the real ID and then call open(2).)
175 access() always dereferences symbolic links. If you need to check the
177 permissions on a symbolic link, use faccessat() with the flag AT_SYM‐
178 LINK_NOFOLLOW.
179 These calls return an error if any of the access types in mode is de‐
181 nied, even if some of the other access types in mode are permitted.
182 If the calling process has appropriate privileges (i.e., is superuser),
184 POSIX.1-2001 permits an implementation to indicate success for an X_OK
185 check even if none of the execute file permission bits are set. Linux
186 does not do this.
187 A file is accessible only if the permissions on each of the directories
189 in the path prefix of pathname grant search (i.e., execute) access. If
190 any directory is inaccessible, then the access() call fails, regardless
191 of the permissions on the file itself.
192 Only access bits are checked, not the file type or contents. There‐
194 fore, if a directory is found to be writable, it probably means that
195 files can be created in the directory, and not that the directory can
196 be written as a file. Similarly, a DOS file may be reported as exe‐
197 cutable, but the execve(2) call will still fail.
198 These calls may not work correctly on NFSv2 filesystems with UID map‐
200 ping enabled, because UID mapping is done on the server and hidden from
201 the client, which checks permissions. (NFS versions 3 and higher per‐
202 form the check on the server.) Similar problems can occur to FUSE
203 mounts.
204 C library/kernel differences
206 The raw faccessat() system call takes only the first three arguments.
207 The AT_EACCESS and AT_SYMLINK_NOFOLLOW flags are actually implemented
208 within the glibc wrapper function for faccessat(). If either of these
209 flags is specified, then the wrapper function employs fstatat(2) to de‐
210 termine access permissions, but see BUGS.
211 Glibc notes
213 On older kernels where faccessat() is unavailable (and when the AT_EAC‐
214 CESS and AT_SYMLINK_NOFOLLOW flags are not specified), the glibc wrap‐
215 per function falls back to the use of access(). When pathname is a
216 relative pathname, glibc constructs a pathname based on the symbolic
217 link in /proc/self/fd that corresponds to the dirfd argument.
218
220 Because the Linux kernel's faccessat() system call does not support a
221 flags argument, the glibc faccessat() wrapper function provided in
222 glibc 2.32 and earlier emulates the required functionality using a com‐
223 bination of the faccessat() system call and fstatat(2). However, this
224 emulation does not take ACLs into account. Starting with glibc 2.33,
225 the wrapper function avoids this bug by making use of the faccessat2()
226 system call where it is provided by the underlying kernel.
227 In kernel 2.4 (and earlier) there is some strangeness in the handling
229 of X_OK tests for superuser. If all categories of execute permission
230 are disabled for a nondirectory file, then the only access() test that
231 returns -1 is when mode is specified as just X_OK; if R_OK or W_OK is
232 also specified in mode, then access() returns 0 for such files. Early
233 2.6 kernels (up to and including 2.6.3) also behaved in the same way as
234 kernel 2.4.
235 In kernels before 2.6.20, these calls ignored the effect of the
237 MS_NOEXEC flag if it was used to mount(2) the underlying filesystem.
238 Since kernel 2.6.20, the MS_NOEXEC flag is honored.
239
241 chmod(2), chown(2), open(2), setgid(2), setuid(2), stat(2), euidac‐
242 cess(3), credentials(7), path_resolution(7), symlink(7)
243
245 This page is part of release 5.12 of the Linux man-pages project. A
246 description of the project, information about reporting bugs, and the
247 latest version of this page, can be found at
248 https://www.kernel.org/doc/man-pages/.
249Linux 2021-03-22 ACCESS(2)