1OPEN(2) Linux Programmer's Manual OPEN(2)
2
6 open, creat - open and possibly create a file or device
7
9 #include <sys/types.h>
10 #include <sys/stat.h>
11 #include <fcntl.h>
12 int open(const char *pathname, int flags);
14 int open(const char *pathname, int flags, mode_t mode);
15 int creat(const char *pathname, mode_t mode);
16
18 Given a pathname for a file, open() returns a file descriptor, a small,
19 non-negative integer for use in subsequent system calls (read(2),
20 write(2), lseek(2), fcntl(2), etc.). The file descriptor returned by a
21 successful call will be the lowest-numbered file descriptor not cur‐
22 rently open for the process.
23 The new file descriptor is set to remain open across an execve(2)
25 (i.e., the FD_CLOEXEC file descriptor flag described in fcntl(2) is
26 initially disabled). The file offset is set to the beginning of the
27 file (see lseek(2)).
28 A call to open() creates a new open file description, an entry in the
30 system-wide table of open files. This entry records the file offset
31 and the file status flags (modifiable via the fcntl() F_SETFL opera‐
32 tion). A file descriptor is a reference to one of these entries; this
33 reference is unaffected if pathname is subsequently removed or modified
34 to refer to a different file. The new open file description is ini‐
35 tially not shared with any other process, but sharing may arise via
36 fork(2).
37 The parameter flags must include one of the following access modes:
39 O_RDONLY, O_WRONLY, or O_RDWR. These request opening the file read-
40 only, write-only, or read/write, respectively.
41 In addition, zero or more file creation flags and file status flags can
43 be bitwise-or'd in flags. The file creation flags are O_CREAT, O_EXCL,
44 O_NOCTTY, and O_TRUNC. The file status flags are all of the remaining
45 flags listed below. The distinction between these two groups of flags
46 is that the file status flags can be retrieved and (in some cases) mod‐
47 ified using fcntl(2). The full list of file creation flags and file
48 status flags is as follows:
49 O_APPEND
51 The file is opened in append mode. Before each write(), the file
52 offset is positioned at the end of the file, as if with lseek().
53 O_APPEND may lead to corrupted files on NFS file systems if more
54 than one process appends data to a file at once. This is
55 because NFS does not support appending to a file, so the client
56 kernel has to simulate it, which can't be done without a race
57 condition.
58 O_ASYNC
60 Enable signal-driven I/O: generate a signal (SIGIO by default,
61 but this can be changed via fcntl(2)) when input or output
62 becomes possible on this file descriptor. This feature is only
63 available for terminals, pseudo-terminals, sockets, and (since
64 Linux 2.6) pipes and FIFOs. See fcntl(2) for further details.
65 O_CREAT
67 If the file does not exist it will be created. The owner (user
68 ID) of the file is set to the effective user ID of the process.
69 The group ownership (group ID) is set either to the effective
70 group ID of the process or to the group ID of the parent direc‐
71 tory (depending on filesystem type and mount options, and the
72 mode of the parent directory, see, e.g., the mount options bsd‐
73 groups and sysvgroups of the ext2 filesystem, as described in
74 mount(8)).
75 O_DIRECT
77 Try to minimize cache effects of the I/O to and from this file.
78 In general this will degrade performance, but it is useful in
79 special situations, such as when applications do their own
80 caching. File I/O is done directly to/from user space buffers.
81 The I/O is synchronous, i.e., at the completion of a read(2) or
82 write(2), data is guaranteed to have been transferred. Under
83 Linux 2.4 transfer sizes, and the alignment of user buffer and
84 file offset must all be multiples of the logical block size of
85 the file system. Under Linux 2.6 alignment to 512-byte bound‐
86 aries suffices.
87 A semantically similar (but deprecated) interface for block
89 devices is described in raw(8).
90 O_DIRECTORY
92 If pathname is not a directory, cause the open to fail. This
93 flag is Linux-specific, and was added in kernel version 2.1.126,
94 to avoid denial-of-service problems if opendir(3) is called on a
95 FIFO or tape device, but should not be used outside of the
96 implementation of opendir.
97 O_EXCL When used with O_CREAT, if the file already exists it is an
99 error and the open() will fail. In this context, a symbolic link
100 exists, regardless of where it points to. O_EXCL is broken on
101 NFS file systems; programs which rely on it for performing lock‐
102 ing tasks will contain a race condition. The solution for per‐
103 forming atomic file locking using a lockfile is to create a
104 unique file on the same file system (e.g., incorporating host‐
105 name and pid), use link(2) to make a link to the lockfile. If
106 link() returns 0, the lock is successful. Otherwise, use
107 stat(2) on the unique file to check if its link count has
108 increased to 2, in which case the lock is also successful.
109 O_LARGEFILE
111 (LFS) Allow files whose sizes cannot be represented in an off_t
112 (but can be represented in an off64_t) to be opened. The
113 _LARGEFILE64_SOURCE macro must be defined in order to obtain
114 this definition. Setting the _FILE_OFFSET_BITS feature test
115 macro to 64 (rather than using O_LARGEFILE) is the preferred
116 method of obtaining method of accessing large files on 32-bit
117 systems (see feature_test_macros(7)).
118 O_NOATIME
120 (Since Linux 2.6.8) Do not update the file last access time
121 (st_atime in the inode) when the file is read(2). This flag is
122 intended for use by indexing or backup programs, where its use
123 can significantly reduce the amount of disk activity. This flag
124 may not be effective on all filesystems. One example is NFS,
125 where the server maintains the access time.
126 O_NOCTTY
128 If pathname refers to a terminal device — see tty(4) — it will
129 not become the process's controlling terminal even if the
130 process does not have one.
131 O_NOFOLLOW
133 If pathname is a symbolic link, then the open fails. This is a
134 FreeBSD extension, which was added to Linux in version 2.1.126.
135 Symbolic links in earlier components of the pathname will still
136 be followed.
137 O_NONBLOCK or O_NDELAY
139 When possible, the file is opened in non-blocking mode. Neither
140 the open() nor any subsequent operations on the file descriptor
141 which is returned will cause the calling process to wait. For
142 the handling of FIFOs (named pipes), see also fifo(7). For a
143 discussion of the effect of O_NONBLOCK in conjunction with
144 mandatory file locks and with file leases, see fcntl(2).
145 O_SYNC The file is opened for synchronous I/O. Any write()s on the
148 resulting file descriptor will block the calling process until
149 the data has been physically written to the underlying hardware.
150 But see RESTRICTIONS below.
151 O_TRUNC
153 If the file already exists and is a regular file and the open
154 mode allows writing (i.e., is O_RDWR or O_WRONLY) it will be
155 truncated to length 0. If the file is a FIFO or terminal device
156 file, the O_TRUNC flag is ignored. Otherwise the effect of
157 O_TRUNC is unspecified.
158 Some of these optional flags can be altered using fcntl() after the
160 file has been opened.
161 The argument mode specifies the permissions to use in case a new file
163 is created. It is modified by the process's umask in the usual way: the
164 permissions of the created file are (mode & ~umask). Note that this
165 mode only applies to future accesses of the newly created file; the
166 open() call that creates a read-only file may well return a read/write
167 file descriptor.
168 The following symbolic constants are provided for mode:
170 S_IRWXU
172 00700 user (file owner) has read, write and execute permission
173 S_IRUSR
175 00400 user has read permission
176 S_IWUSR
178 00200 user has write permission
179 S_IXUSR
181 00100 user has execute permission
182 S_IRWXG
184 00070 group has read, write and execute permission
185 S_IRGRP
187 00040 group has read permission
188 S_IWGRP
190 00020 group has write permission
191 S_IXGRP
193 00010 group has execute permission
194 S_IRWXO
196 00007 others have read, write and execute permission
197 S_IROTH
199 00004 others have read permission
200 S_IWOTH
202 00002 others have write permission
203 S_IXOTH
205 00001 others have execute permission
206 mode must be specified when O_CREAT is in the flags, and is ignored
208 otherwise.
209 creat() is equivalent to open() with flags equal to
211 O_CREAT|O_WRONLY|O_TRUNC.
212
214 open() and creat() return the new file descriptor, or -1 if an error
215 occurred (in which case, errno is set appropriately).
216
218 Note that open() can open device special files, but creat() cannot cre‐
219 ate them; use mknod(2) instead.
220 On NFS file systems with UID mapping enabled, open() may return a file
222 descriptor but e.g. read(2) requests are denied with EACCES. This is
223 because the client performs open() by checking the permissions, but UID
224 mapping is performed by the server upon read and write requests.
225 If the file is newly created, its st_atime, st_ctime, st_mtime fields
227 (respectively, time of last access, time of last status change, and
228 time of last modification; see stat(2)) are set to the current time,
229 and so are the st_ctime and st_mtime fields of the parent directory.
230 Otherwise, if the file is modified because of the O_TRUNC flag, its
231 st_ctime and st_mtime fields are set to the current time.
232
234 EACCES The requested access to the file is not allowed, or search per‐
235 mission is denied for one of the directories in the path prefix
236 of pathname, or the file did not exist yet and write access to
237 the parent directory is not allowed. (See also path_resolu‐
238 tion(2).)
239 EEXIST pathname already exists and O_CREAT and O_EXCL were used.
241 EFAULT pathname points outside your accessible address space.
243 EFBIG pathname refers to a regular file, too large to be opened; see
245 O_LARGEFILE above. (POSIX.1-2001 specifies the error EOVERFLOW
246 for this case.)
247 EISDIR pathname refers to a directory and the access requested involved
249 writing (that is, O_WRONLY or O_RDWR is set).
250 ELOOP Too many symbolic links were encountered in resolving pathname,
252 or O_NOFOLLOW was specified but pathname was a symbolic link.
253 EMFILE The process already has the maximum number of files open.
255 ENAMETOOLONG
257 pathname was too long.
258 ENFILE The system limit on the total number of open files has been
260 reached.
261 ENODEV pathname refers to a device special file and no corresponding
263 device exists. (This is a Linux kernel bug; in this situation
264 ENXIO must be returned.)
265 ENOENT O_CREAT is not set and the named file does not exist. Or, a
267 directory component in pathname does not exist or is a dangling
268 symbolic link.
269 ENOMEM Insufficient kernel memory was available.
271 ENOSPC pathname was to be created but the device containing pathname
273 has no room for the new file.
274 ENOTDIR
276 A component used as a directory in pathname is not, in fact, a
277 directory, or O_DIRECTORY was specified and pathname was not a
278 directory.
279 ENXIO O_NONBLOCK | O_WRONLY is set, the named file is a FIFO and no
281 process has the file open for reading. Or, the file is a device
282 special file and no corresponding device exists.
283 EPERM The O_NOATIME flag was specified, but the effective user ID of
285 the caller did not match the owner of the file and the caller
286 was not privileged (CAP_FOWNER).
287 EROFS pathname refers to a file on a read-only filesystem and write
289 access was requested.
290 ETXTBSY
292 pathname refers to an executable image which is currently being
293 executed and write access was requested.
294 EWOULDBLOCK
296 The O_NONBLOCK flag was specified, and an incompatible lease was
297 held on the file (see fcntl(2)).
298
300 Under Linux, the O_NONBLOCK flag indicates that one wants to open but
301 does not necessarily have the intention to read or write. This is typ‐
302 ically used to open devices in order to get a file descriptor for use
303 with ioctl(2).
304
306 SVr4, 4.3BSD, POSIX.1-2001. The O_NOATIME, O_NOFOLLOW, and O_DIRECTORY
307 flags are Linux specific. One may have to define the _GNU_SOURCE macro
308 to get their definitions.
309 The (undefined) effect of O_RDONLY | O_TRUNC varies among implementa‐
311 tions. On many systems the file is actually truncated.
312 The O_DIRECT flag was introduced in SGI IRIX, where it has alignment
314 restrictions similar to those of Linux 2.4. IRIX has also a fcntl(2)
315 call to query appropriate alignments, and sizes. FreeBSD 4.x intro‐
316 duced a flag of same name, but without alignment restrictions. Support
317 was added under Linux in kernel version 2.4.10. Older Linux kernels
318 simply ignore this flag. One may have to define the _GNU_SOURCE macro
319 to get its definition.
320
322 "The thing that has always disturbed me about O_DIRECT is that the
323 whole interface is just stupid, and was probably designed by a deranged
324 monkey on some serious mind-controlling substances." — Linus
325 Currently, it is not possible to enable signal-driven I/O by specifying
327 O_ASYNC when calling open(); use fcntl(2) to enable this flag.
328
330 There are many infelicities in the protocol underlying NFS, affecting
331 amongst others O_SYNC and O_NDELAY.
332 POSIX provides for three different variants of synchronised I/O, corre‐
334 sponding to the flags O_SYNC, O_DSYNC and O_RSYNC. Currently (2.1.130)
335 these are all synonymous under Linux.
336
338 close(2), dup(2), fcntl(2), link(2), lseek(2), mknod(2), mount(2),
339 mmap(2), openat(2), path_resolution(2), read(2), socket(2), stat(2),
340 umask(2), unlink(2), write(2), fopen(3), fifo(7), fea‐
341 ture_test_macros(7)
342Linux 2.6.12 2005-06-22 OPEN(2)