]>
Commit | Line | Data |
---|---|---|
fea681da MK |
1 | .\" Hey Emacs! This file is -*- nroff -*- source. |
2 | .\" | |
3 | .\" This manpage is Copyright (C) 1992 Drew Eckhardt; | |
4 | .\" 1993 Michael Haardt, Ian Jackson. | |
ddc4d339 | 5 | .\" 2008 Greg Banks |
fea681da MK |
6 | .\" |
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. | |
c13182ef | 15 | .\" |
fea681da MK |
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17 | .\" manual page may be incorrect or out-of-date.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. | |
c13182ef | 23 | .\" |
fea681da MK |
24 | .\" Formatted or processed versions of this manual, if unaccompanied by |
25 | .\" the source, must acknowledge the copyright and authors of this work. | |
26 | .\" | |
27 | .\" Modified 1993-07-21 by Rik Faith <faith@cs.unc.edu> | |
28 | .\" Modified 1994-08-21 by Michael Haardt | |
29 | .\" Modified 1996-04-13 by Andries Brouwer <aeb@cwi.nl> | |
30 | .\" Modified 1996-05-13 by Thomas Koenig | |
31 | .\" Modified 1996-12-20 by Michael Haardt | |
32 | .\" Modified 1999-02-19 by Andries Brouwer <aeb@cwi.nl> | |
33 | .\" Modified 1998-11-28 by Joseph S. Myers <jsm28@hermes.cam.ac.uk> | |
34 | .\" Modified 1999-06-03 by Michael Haardt | |
c11b1abf MK |
35 | .\" Modified 2002-05-07 by Michael Kerrisk <mtk.manpages@gmail.com> |
36 | .\" Modified 2004-06-23 by Michael Kerrisk <mtk.manpages@gmail.com> | |
1c1e15ed MK |
37 | .\" 2004-12-08, mtk, reordered flags list alphabetically |
38 | .\" 2004-12-08, Martin Pool <mbp@sourcefrog.net> (& mtk), added O_NOATIME | |
fe75ec04 | 39 | .\" 2007-09-18, mtk, Added description of O_CLOEXEC + other minor edits |
447bb15e | 40 | .\" 2008-01-03, mtk, with input from Trond Myklebust |
f4b9d6a5 MK |
41 | .\" <trond.myklebust@fys.uio.no> and Timo Sirainen <tss@iki.fi> |
42 | .\" Rewrite description of O_EXCL. | |
ddc4d339 MK |
43 | .\" 2008-01-11, Greg Banks <gnb@melbourne.sgi.com>: add more detail |
44 | .\" on O_DIRECT. | |
d77eb764 | 45 | .\" 2008-02-26, Michael Haardt: Reorganized text for O_CREAT and mode |
fea681da | 46 | .\" |
61b7c1e1 | 47 | .\" FIXME . Apr 08: The next POSIX revision has O_EXEC, O_SEARCH, and |
9f91e36c | 48 | .\" O_TTYINIT. Eventually these may need to be documented. --mtk |
803e1d2f | 49 | .\" FIXME Linux 2.6.33 has O_DSYNC, and a hidden __O_SYNC. |
9f91e36c | 50 | .\" |
e417acb0 | 51 | .TH OPEN 2 2010-09-10 "Linux" "Linux Programmer's Manual" |
fea681da MK |
52 | .SH NAME |
53 | open, creat \- open and possibly create a file or device | |
54 | .SH SYNOPSIS | |
55 | .nf | |
56 | .B #include <sys/types.h> | |
57 | .B #include <sys/stat.h> | |
58 | .B #include <fcntl.h> | |
59 | .sp | |
60 | .BI "int open(const char *" pathname ", int " flags ); | |
61 | .BI "int open(const char *" pathname ", int " flags ", mode_t " mode ); | |
5895e7eb | 62 | |
fea681da MK |
63 | .BI "int creat(const char *" pathname ", mode_t " mode ); |
64 | .fi | |
65 | .SH DESCRIPTION | |
e366dbc4 | 66 | Given a |
0daa9e92 | 67 | .I pathname |
e366dbc4 | 68 | for a file, |
1f6ceb40 | 69 | .BR open () |
2fda57bd | 70 | returns a file descriptor, a small, nonnegative integer |
e366dbc4 MK |
71 | for use in subsequent system calls |
72 | .RB ( read "(2), " write "(2), " lseek "(2), " fcntl "(2), etc.)." | |
73 | The file descriptor returned by a successful call will be | |
2c4bff36 | 74 | the lowest-numbered file descriptor not currently open for the process. |
e366dbc4 | 75 | .PP |
fe75ec04 | 76 | By default, the new file descriptor is set to remain open across an |
e366dbc4 | 77 | .BR execve (2) |
1f6ceb40 MK |
78 | (i.e., the |
79 | .B FD_CLOEXEC | |
80 | file descriptor flag described in | |
81 | .BR fcntl (2) | |
fe75ec04 MK |
82 | is initially disabled; the Linux-specific |
83 | .B O_CLOEXEC | |
84 | flag, described below, can be used to change this default). | |
1f6ceb40 | 85 | The file offset is set to the beginning of the file (see |
c13182ef | 86 | .BR lseek (2)). |
e366dbc4 MK |
87 | .PP |
88 | A call to | |
89 | .BR open () | |
90 | creates a new | |
91 | .IR "open file description" , | |
92 | an entry in the system-wide table of open files. | |
e366dbc4 MK |
93 | This entry records the file offset and the file status flags |
94 | (modifiable via the | |
0bfa087b | 95 | .BR fcntl (2) |
e366dbc4 MK |
96 | .B F_SETFL |
97 | operation). | |
2c4bff36 MK |
98 | A file descriptor is a reference to one of these entries; |
99 | this reference is unaffected if | |
100 | .I pathname | |
101 | is subsequently removed or modified to refer to a different file. | |
e366dbc4 | 102 | The new open file description is initially not shared |
2c4bff36 MK |
103 | with any other process, |
104 | but sharing may arise via | |
105 | .BR fork (2). | |
e366dbc4 | 106 | .PP |
c4bb193f | 107 | The argument |
fea681da | 108 | .I flags |
e366dbc4 MK |
109 | must include one of the following |
110 | .IR "access modes" : | |
c7992edc | 111 | .BR O_RDONLY ", " O_WRONLY ", or " O_RDWR . |
e366dbc4 MK |
112 | These request opening the file read-only, write-only, or read/write, |
113 | respectively. | |
bfe9ba67 MK |
114 | |
115 | In addition, zero or more file creation flags and file status flags | |
c13182ef | 116 | can be |
fea681da | 117 | .RI bitwise- or 'd |
e366dbc4 | 118 | in |
bfe9ba67 | 119 | .IR flags . |
c13182ef MK |
120 | The |
121 | .I file creation flags | |
122 | are | |
bfe9ba67 | 123 | .BR O_CREAT ", " O_EXCL ", " O_NOCTTY ", and " O_TRUNC . |
c13182ef MK |
124 | The |
125 | .I file status flags | |
bfe9ba67 | 126 | are all of the remaining flags listed below. |
93ee8f96 MK |
127 | .\" FIXME . Actually is it true that the "file status flags" are all of the |
128 | .\" remaining flags listed below? SUSv4 divides the flags into: | |
129 | .\" * Access mode | |
130 | .\" * File creation | |
131 | .\" * File status | |
132 | .\" * Other (O_CLOEXEC, O_DIRECTORY, O_NOFOLLOW) | |
133 | .\" though it's not clear what the difference between "other" and | |
9502479b | 134 | .\" "File creation" flags is. (I've raised an Aardvark to see if this |
93ee8f96 | 135 | .\" can be clarified in SUSv4; 10 Oct 2008.) |
bfe9ba67 MK |
136 | The distinction between these two groups of flags is that |
137 | the file status flags can be retrieved and (in some cases) | |
138 | modified using | |
139 | .BR fcntl (2). | |
140 | The full list of file creation flags and file status flags is as follows: | |
fea681da | 141 | .TP |
1c1e15ed | 142 | .B O_APPEND |
c13182ef MK |
143 | The file is opened in append mode. |
144 | Before each | |
0bfa087b | 145 | .BR write (2), |
1e568304 | 146 | the file offset is positioned at the end of the file, |
1c1e15ed | 147 | as if with |
0bfa087b | 148 | .BR lseek (2). |
1c1e15ed MK |
149 | .B O_APPEND |
150 | may lead to corrupted files on NFS file systems if more than one process | |
c13182ef | 151 | appends data to a file at once. |
a4391429 MK |
152 | .\" For more background, see |
153 | .\" http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=453946 | |
154 | .\" http://nfs.sourceforge.net/ | |
c13182ef | 155 | This is because NFS does not support |
1c1e15ed MK |
156 | appending to a file, so the client kernel has to simulate it, which |
157 | can't be done without a race condition. | |
158 | .TP | |
159 | .B O_ASYNC | |
b50582eb | 160 | Enable signal-driven I/O: |
8bd58774 MK |
161 | generate a signal |
162 | .RB ( SIGIO | |
163 | by default, but this can be changed via | |
1c1e15ed MK |
164 | .BR fcntl (2)) |
165 | when input or output becomes possible on this file descriptor. | |
662b2137 | 166 | This feature is only available for terminals, psuedoterminals, |
1f6ceb40 MK |
167 | sockets, and (since Linux 2.6) pipes and FIFOs. |
168 | See | |
1c1e15ed MK |
169 | .BR fcntl (2) |
170 | for further details. | |
fe75ec04 MK |
171 | .TP |
172 | .BR O_CLOEXEC " (Since Linux 2.6.23)" | |
173 | Enable the close-on-exec flag for the new file descriptor. | |
24ec631f | 174 | Specifying this flag permits a program to avoid additional |
fe75ec04 MK |
175 | .BR fcntl (2) |
176 | .B F_SETFD | |
24ec631f | 177 | operations to set the |
0daa9e92 | 178 | .B FD_CLOEXEC |
fe75ec04 MK |
179 | flag. |
180 | Additionally, | |
181 | use of this flag is essential in some multithreaded programs | |
182 | since using a separate | |
183 | .BR fcntl (2) | |
184 | .B F_SETFD | |
185 | operation to set the | |
0daa9e92 | 186 | .B FD_CLOEXEC |
fe75ec04 MK |
187 | flag does not suffice to avoid race conditions |
188 | where one thread opens a file descriptor at the same | |
189 | time as another thread does a | |
190 | .BR fork (2) | |
191 | plus | |
192 | .BR execve (2). | |
193 | .\" This flag fixes only one form of the race condition; | |
194 | .\" The race can also occur with, for example, descriptors | |
195 | .\" returned by accept(), pipe(), etc. | |
1c1e15ed | 196 | .TP |
fea681da MK |
197 | .B O_CREAT |
198 | If the file does not exist it will be created. | |
199 | The owner (user ID) of the file is set to the effective user ID | |
c13182ef MK |
200 | of the process. |
201 | The group ownership (group ID) is set either to | |
fea681da | 202 | the effective group ID of the process or to the group ID of the |
24d01c53 | 203 | parent directory (depending on file system type and mount options, |
8b39ad66 | 204 | and the mode of the parent directory, see the mount options |
fea681da MK |
205 | .I bsdgroups |
206 | and | |
207 | .I sysvgroups | |
8b39ad66 | 208 | described in |
fea681da | 209 | .BR mount (8)). |
8b39ad66 MK |
210 | .\" As at 2.6.25, bsdgroups is supported by ext2, ext3, ext4, and |
211 | .\" XFS (since 2.6.14). | |
4e698277 MK |
212 | .RS |
213 | .PP | |
214 | .I mode | |
215 | specifies the permissions to use in case a new file is created. | |
216 | This argument must be supplied when | |
217 | .B O_CREAT | |
218 | is specified in | |
219 | .IR flags ; | |
220 | if | |
221 | .B O_CREAT | |
222 | is not specified, then | |
223 | .I mode | |
224 | is ignored. | |
225 | The effective permissions are modified by | |
226 | the process's | |
227 | .I umask | |
228 | in the usual way: The permissions of the created file are | |
84a275c4 | 229 | .IR "(mode\ &\ ~umask)" . |
4e698277 MK |
230 | Note that this mode only applies to future accesses of the |
231 | newly created file; the | |
232 | .BR open () | |
233 | call that creates a read-only file may well return a read/write | |
234 | file descriptor. | |
235 | .PP | |
236 | The following symbolic constants are provided for | |
237 | .IR mode : | |
238 | .TP 9 | |
239 | .B S_IRWXU | |
240 | 00700 user (file owner) has read, write and execute permission | |
241 | .TP | |
242 | .B S_IRUSR | |
243 | 00400 user has read permission | |
244 | .TP | |
245 | .B S_IWUSR | |
246 | 00200 user has write permission | |
247 | .TP | |
248 | .B S_IXUSR | |
249 | 00100 user has execute permission | |
250 | .TP | |
251 | .B S_IRWXG | |
252 | 00070 group has read, write and execute permission | |
253 | .TP | |
254 | .B S_IRGRP | |
255 | 00040 group has read permission | |
256 | .TP | |
257 | .B S_IWGRP | |
258 | 00020 group has write permission | |
259 | .TP | |
260 | .B S_IXGRP | |
261 | 00010 group has execute permission | |
262 | .TP | |
263 | .B S_IRWXO | |
264 | 00007 others have read, write and execute permission | |
265 | .TP | |
266 | .B S_IROTH | |
267 | 00004 others have read permission | |
268 | .TP | |
269 | .B S_IWOTH | |
270 | 00002 others have write permission | |
271 | .TP | |
272 | .B S_IXOTH | |
273 | 00001 others have execute permission | |
274 | .RE | |
fea681da | 275 | .TP |
ddc4d339 | 276 | .BR O_DIRECT " (Since Linux 2.4.10)" |
1c1e15ed MK |
277 | Try to minimize cache effects of the I/O to and from this file. |
278 | In general this will degrade performance, but it is useful in | |
279 | special situations, such as when applications do their own caching. | |
280 | File I/O is done directly to/from user space buffers. | |
015221ef CH |
281 | The |
282 | .B O_DIRECT | |
283 | flag on its own makes at an effort to transfer data synchronously, | |
284 | but does not give the guarantees of the | |
285 | .B O_SYNC | |
286 | that data and necessary metadata are transferred. | |
287 | To guarantee synchronous I/O the | |
288 | .B O_SYNC | |
289 | must be used in addition to | |
290 | .BR O_DIRECT . | |
c734b9f2 | 291 | See |
ddc4d339 MK |
292 | .B NOTES |
293 | below for further discussion. | |
9b54d4fa | 294 | .sp |
c13182ef | 295 | A semantically similar (but deprecated) interface for block devices |
9b54d4fa | 296 | is described in |
1c1e15ed MK |
297 | .BR raw (8). |
298 | .TP | |
299 | .B O_DIRECTORY | |
a8d55537 | 300 | If \fIpathname\fP is not a directory, cause the open to fail. |
9f8d688a MK |
301 | .\" But see the following and its replies: |
302 | .\" http://marc.theaimsgroup.com/?t=112748702800001&r=1&w=2 | |
303 | .\" [PATCH] open: O_DIRECTORY and O_CREAT together should fail | |
304 | .\" O_DIRECTORY | O_CREAT causes O_DIRECTORY to be ignored. | |
8382f16d | 305 | This flag is Linux-specific, and was added in kernel version 2.1.126, to |
60a90ecd MK |
306 | avoid denial-of-service problems if |
307 | .BR opendir (3) | |
308 | is called on a | |
1c1e15ed | 309 | FIFO or tape device, but should not be used outside of the |
9e370fba MK |
310 | implementation of |
311 | .BR opendir (3). | |
1c1e15ed | 312 | .TP |
fea681da | 313 | .B O_EXCL |
f4b9d6a5 MK |
314 | Ensure that this call creates the file: |
315 | if this flag is specified in conjunction with | |
fea681da | 316 | .BR O_CREAT , |
f4b9d6a5 MK |
317 | and |
318 | .I pathname | |
319 | already exists, then | |
1c1e15ed | 320 | .BR open () |
c13182ef | 321 | will fail. |
f4b9d6a5 MK |
322 | The behavior of |
323 | .B O_EXCL | |
324 | is undefined if | |
325 | .B O_CREAT | |
326 | is not specified. | |
327 | ||
328 | When these two flags are specified, symbolic links are not followed: | |
329 | .\" POSIX.1-2001 explicitly requires this behavior. | |
330 | if | |
331 | .I pathname | |
332 | is a symbolic link, then | |
333 | .BR open () | |
334 | fails regardless of where the symbolic link points to. | |
335 | ||
efe08656 | 336 | On NFS, |
f4b9d6a5 | 337 | .B O_EXCL |
efe08656 MK |
338 | is only supported when using NFSv3 or later on kernel 2.6 or later. |
339 | In NFS environments where | |
fea681da | 340 | .B O_EXCL |
f4b9d6a5 MK |
341 | support is not provided, programs that rely on it |
342 | for performing locking tasks will contain a race condition. | |
343 | Portable programs that want to perform atomic file locking using a lockfile, | |
344 | and need to avoid reliance on NFS support for | |
345 | .BR O_EXCL , | |
346 | can create a unique file on | |
347 | the same file system (e.g., incorporating hostname and PID), and use | |
fea681da | 348 | .BR link (2) |
c13182ef | 349 | to make a link to the lockfile. |
60a90ecd MK |
350 | If |
351 | .BR link (2) | |
f4b9d6a5 | 352 | returns 0, the lock is successful. |
c13182ef | 353 | Otherwise, use |
fea681da MK |
354 | .BR stat (2) |
355 | on the unique file to check if its link count has increased to 2, | |
356 | in which case the lock is also successful. | |
357 | .TP | |
1c1e15ed MK |
358 | .B O_LARGEFILE |
359 | (LFS) | |
360 | Allow files whose sizes cannot be represented in an | |
8478ee02 | 361 | .I off_t |
1c1e15ed | 362 | (but can be represented in an |
8478ee02 | 363 | .IR off64_t ) |
1c1e15ed | 364 | to be opened. |
c13182ef | 365 | The |
bcdd964e | 366 | .B _LARGEFILE64_SOURCE |
e417acb0 MK |
367 | macro must be defined |
368 | (before including | |
369 | .I any | |
370 | header files) | |
371 | in order to obtain this definition. | |
c13182ef | 372 | Setting the |
bcdd964e | 373 | .B _FILE_OFFSET_BITS |
9f3d8b28 MK |
374 | feature test macro to 64 (rather than using |
375 | .BR O_LARGEFILE ) | |
376 | is the preferred method of obtaining | |
377 | method of accessing large files on 32-bit systems (see | |
2dcbf4f7 | 378 | .BR feature_test_macros (7)). |
1c1e15ed | 379 | .TP |
fe75ec04 | 380 | .BR O_NOATIME " (Since Linux 2.6.8)" |
310b7919 MK |
381 | Do not update the file last access time (st_atime in the inode) |
382 | when the file is | |
1c1e15ed MK |
383 | .BR read (2). |
384 | This flag is intended for use by indexing or backup programs, | |
385 | where its use can significantly reduce the amount of disk activity. | |
24d01c53 | 386 | This flag may not be effective on all file systems. |
1c1e15ed | 387 | One example is NFS, where the server maintains the access time. |
0e1ad98c | 388 | .\" The O_NOATIME flag also affects the treatment of st_atime |
92057f4d | 389 | .\" by mmap() and readdir(2), MTK, Dec 04. |
1c1e15ed | 390 | .TP |
fea681da MK |
391 | .B O_NOCTTY |
392 | If | |
393 | .I pathname | |
394 | refers to a terminal device \(em see | |
395 | .BR tty (4) | |
396 | \(em it will not become the process's controlling terminal even if the | |
397 | process does not have one. | |
398 | .TP | |
1c1e15ed | 399 | .B O_NOFOLLOW |
a8d55537 | 400 | If \fIpathname\fP is a symbolic link, then the open fails. |
c13182ef | 401 | This is a FreeBSD extension, which was added to Linux in version 2.1.126. |
1c1e15ed | 402 | Symbolic links in earlier components of the pathname will still be |
e366dbc4 MK |
403 | followed. |
404 | .\" The headers from glibc 2.0.100 and later include a | |
405 | .\" definition of this flag; \fIkernels before 2.1.126 will ignore it if | |
a8d55537 | 406 | .\" used\fP. |
fea681da MK |
407 | .TP |
408 | .BR O_NONBLOCK " or " O_NDELAY | |
ff40dbb3 | 409 | When possible, the file is opened in nonblocking mode. |
c13182ef | 410 | Neither the |
1c1e15ed | 411 | .BR open () |
fea681da MK |
412 | nor any subsequent operations on the file descriptor which is |
413 | returned will cause the calling process to wait. | |
414 | For the handling of FIFOs (named pipes), see also | |
af5b2ef2 | 415 | .BR fifo (7). |
db28bfac | 416 | For a discussion of the effect of |
0daa9e92 | 417 | .B O_NONBLOCK |
db28bfac MK |
418 | in conjunction with mandatory file locks and with file leases, see |
419 | .BR fcntl (2). | |
fea681da MK |
420 | .TP |
421 | .B O_SYNC | |
c13182ef MK |
422 | The file is opened for synchronous I/O. |
423 | Any | |
0bfa087b | 424 | .BR write (2)s |
fea681da MK |
425 | on the resulting file descriptor will block the calling process until |
426 | the data has been physically written to the underlying hardware. | |
b07cd0a9 | 427 | .IR "But see NOTES below" . |
fea681da | 428 | .TP |
1c1e15ed MK |
429 | .B O_TRUNC |
430 | If the file already exists and is a regular file and the open mode allows | |
682edefb MK |
431 | writing (i.e., is |
432 | .B O_RDWR | |
433 | or | |
434 | .BR O_WRONLY ) | |
435 | it will be truncated to length 0. | |
436 | If the file is a FIFO or terminal device file, the | |
437 | .B O_TRUNC | |
c13182ef | 438 | flag is ignored. |
682edefb MK |
439 | Otherwise the effect of |
440 | .B O_TRUNC | |
441 | is unspecified. | |
fea681da MK |
442 | .PP |
443 | Some of these optional flags can be altered using | |
0bfa087b | 444 | .BR fcntl (2) |
fea681da MK |
445 | after the file has been opened. |
446 | ||
1c1e15ed | 447 | .BR creat () |
fea681da | 448 | is equivalent to |
1c1e15ed | 449 | .BR open () |
fea681da MK |
450 | with |
451 | .I flags | |
452 | equal to | |
453 | .BR O_CREAT|O_WRONLY|O_TRUNC . | |
454 | .SH "RETURN VALUE" | |
c13182ef MK |
455 | .BR open () |
456 | and | |
e1d6264d | 457 | .BR creat () |
1c1e15ed MK |
458 | return the new file descriptor, or \-1 if an error occurred |
459 | (in which case, | |
fea681da MK |
460 | .I errno |
461 | is set appropriately). | |
fea681da MK |
462 | .SH ERRORS |
463 | .TP | |
464 | .B EACCES | |
465 | The requested access to the file is not allowed, or search permission | |
466 | is denied for one of the directories in the path prefix of | |
467 | .IR pathname , | |
468 | or the file did not exist yet and write access to the parent directory | |
469 | is not allowed. | |
470 | (See also | |
ad7cc990 | 471 | .BR path_resolution (7).) |
fea681da MK |
472 | .TP |
473 | .B EEXIST | |
474 | .I pathname | |
475 | already exists and | |
476 | .BR O_CREAT " and " O_EXCL | |
477 | were used. | |
478 | .TP | |
479 | .B EFAULT | |
0daa9e92 | 480 | .I pathname |
e1d6264d | 481 | points outside your accessible address space. |
fea681da | 482 | .TP |
9f5773f7 | 483 | .B EFBIG |
7c7fb552 MK |
484 | See |
485 | .BR EOVERFLOW . | |
9f5773f7 | 486 | .TP |
e51412ea MK |
487 | .B EINTR |
488 | While blocked waiting to complete an open of a slow device | |
489 | (e.g., a FIFO; see | |
490 | .BR fifo (7)), | |
491 | the call was interrupted by a signal handler; see | |
492 | .BR signal (7). | |
493 | .TP | |
fea681da MK |
494 | .B EISDIR |
495 | .I pathname | |
496 | refers to a directory and the access requested involved writing | |
497 | (that is, | |
498 | .B O_WRONLY | |
499 | or | |
500 | .B O_RDWR | |
501 | is set). | |
502 | .TP | |
503 | .B ELOOP | |
504 | Too many symbolic links were encountered in resolving | |
505 | .IR pathname , | |
a8d55537 | 506 | or \fBO_NOFOLLOW\fP was specified but |
fea681da MK |
507 | .I pathname |
508 | was a symbolic link. | |
509 | .TP | |
510 | .B EMFILE | |
511 | The process already has the maximum number of files open. | |
512 | .TP | |
513 | .B ENAMETOOLONG | |
0daa9e92 | 514 | .I pathname |
e1d6264d | 515 | was too long. |
fea681da MK |
516 | .TP |
517 | .B ENFILE | |
518 | The system limit on the total number of open files has been reached. | |
519 | .TP | |
520 | .B ENODEV | |
521 | .I pathname | |
522 | refers to a device special file and no corresponding device exists. | |
682edefb MK |
523 | (This is a Linux kernel bug; in this situation |
524 | .B ENXIO | |
525 | must be returned.) | |
fea681da MK |
526 | .TP |
527 | .B ENOENT | |
682edefb MK |
528 | .B O_CREAT |
529 | is not set and the named file does not exist. | |
fea681da MK |
530 | Or, a directory component in |
531 | .I pathname | |
532 | does not exist or is a dangling symbolic link. | |
533 | .TP | |
534 | .B ENOMEM | |
535 | Insufficient kernel memory was available. | |
536 | .TP | |
537 | .B ENOSPC | |
538 | .I pathname | |
539 | was to be created but the device containing | |
540 | .I pathname | |
541 | has no room for the new file. | |
542 | .TP | |
543 | .B ENOTDIR | |
544 | A component used as a directory in | |
545 | .I pathname | |
a8d55537 | 546 | is not, in fact, a directory, or \fBO_DIRECTORY\fP was specified and |
fea681da MK |
547 | .I pathname |
548 | was not a directory. | |
549 | .TP | |
550 | .B ENXIO | |
682edefb MK |
551 | .BR O_NONBLOCK " | " O_WRONLY |
552 | is set, the named file is a FIFO and | |
fea681da MK |
553 | no process has the file open for reading. |
554 | Or, the file is a device special file and no corresponding device exists. | |
555 | .TP | |
7c7fb552 MK |
556 | .B EOVERFLOW |
557 | .I pathname | |
558 | refers to a regular file that is too large to be opened. | |
559 | The usual scenario here is that an application compiled | |
560 | on a 32-bit platform without | |
5e4dc269 | 561 | .I -D_FILE_OFFSET_BITS=64 |
7c7fb552 MK |
562 | tried to open a file whose size exceeds |
563 | .I (2<<31)-1 | |
564 | bits; | |
565 | see also | |
566 | .B O_LARGEFILE | |
567 | above. | |
568 | This is the error specified by POSIX.1-2001; | |
569 | in kernels before 2.6.24, Linux gave the error | |
570 | .B EFBIG | |
571 | for this case. | |
572 | .\" See http://bugzilla.kernel.org/show_bug.cgi?id=7253 | |
573 | .\" "Open of a large file on 32-bit fails with EFBIG, should be EOVERFLOW" | |
574 | .\" Reported 2006-10-03 | |
575 | .TP | |
1c1e15ed MK |
576 | .B EPERM |
577 | The | |
578 | .B O_NOATIME | |
579 | flag was specified, but the effective user ID of the caller | |
580 | .\" Strictly speaking, it's the file system UID... (MTK) | |
581 | did not match the owner of the file and the caller was not privileged | |
582 | .RB ( CAP_FOWNER ). | |
583 | .TP | |
fea681da MK |
584 | .B EROFS |
585 | .I pathname | |
24d01c53 | 586 | refers to a file on a read-only file system and write access was |
fea681da MK |
587 | requested. |
588 | .TP | |
589 | .B ETXTBSY | |
590 | .I pathname | |
591 | refers to an executable image which is currently being executed and | |
592 | write access was requested. | |
d3952311 MK |
593 | .TP |
594 | .B EWOULDBLOCK | |
595 | The | |
596 | .B O_NONBLOCK | |
597 | flag was specified, and an incompatible lease was held on the file | |
598 | (see | |
599 | .BR fcntl (2)). | |
fea681da | 600 | .SH "CONFORMING TO" |
97c1eac8 | 601 | SVr4, 4.3BSD, POSIX.1-2001. |
fea681da | 602 | The |
fe75ec04 | 603 | .BR O_DIRECTORY , |
1c1e15ed | 604 | .BR O_NOATIME , |
fea681da | 605 | and |
0daa9e92 | 606 | .B O_NOFOLLOW |
9f91e36c | 607 | flags are Linux-specific, and one may need to define |
61b7c1e1 | 608 | .B _GNU_SOURCE |
e417acb0 MK |
609 | (before including |
610 | .I any | |
611 | header files) | |
61b7c1e1 | 612 | to obtain their definitions. |
9f91e36c MK |
613 | |
614 | The | |
615 | .BR O_CLOEXEC | |
616 | flag is not specified in POSIX.1-2001, | |
7c5f0513 | 617 | but is specified in POSIX.1-2008. |
9f91e36c | 618 | |
0daa9e92 | 619 | .B O_DIRECT |
fb7339df | 620 | is not specified in POSIX; one has to define |
fe75ec04 | 621 | .B _GNU_SOURCE |
e417acb0 MK |
622 | (before including |
623 | .I any | |
624 | header files) | |
fe75ec04 | 625 | to get its definition. |
a1d5f77c | 626 | .SH NOTES |
988db661 | 627 | Under Linux, the |
a1d5f77c MK |
628 | .B O_NONBLOCK |
629 | flag indicates that one wants to open | |
630 | but does not necessarily have the intention to read or write. | |
631 | This is typically used to open devices in order to get a file descriptor | |
632 | for use with | |
633 | .BR ioctl (2). | |
c734b9f2 MK |
634 | |
635 | Unlike the other values that can be specified in | |
636 | .IR flags , | |
637 | the | |
638 | .I "access mode" | |
639 | values | |
640 | .BR O_RDONLY ", " O_WRONLY ", and " O_RDWR , | |
641 | do not specify individual bits. | |
642 | Rather, they define the low order two bits of | |
643 | .IR flags , | |
644 | and are defined respectively as 0, 1, and 2. | |
645 | In other words, the combination | |
646 | .B "O_RDONLY | O_WRONLY" | |
647 | is a logical error, and certainly does not have the same meaning as | |
648 | .BR O_RDWR . | |
c8f2dd47 | 649 | Linux reserves the special, nonstandard access mode 3 (binary 11) in |
c734b9f2 MK |
650 | .I flags |
651 | to mean: | |
652 | check for read and write permission on the file and return a descriptor | |
653 | that can't be used for reading or writing. | |
c8f2dd47 | 654 | This nonstandard access mode is used by some Linux drivers to return a |
c734b9f2 MK |
655 | descriptor that is only to be used for device-specific |
656 | .BR ioctl (2) | |
657 | operations. | |
658 | .\" See for example util-linux's disk-utils/setfdprm.c | |
659 | .\" For some background on access mode 3, see | |
660 | .\" http://thread.gmane.org/gmane.linux.kernel/653123 | |
661 | .\" "[RFC] correct flags to f_mode conversion in __dentry_open" | |
662 | .\" LKML, 12 Mar 2008 | |
fea681da MK |
663 | .LP |
664 | The (undefined) effect of | |
665 | .B O_RDONLY | O_TRUNC | |
c13182ef | 666 | varies among implementations. |
bcdd964e | 667 | On many systems the file is actually truncated. |
fea681da MK |
668 | .\" Linux 2.0, 2.5: truncate |
669 | .\" Solaris 5.7, 5.8: truncate | |
670 | .\" Irix 6.5: truncate | |
671 | .\" Tru64 5.1B: truncate | |
672 | .\" HP-UX 11.22: truncate | |
673 | .\" FreeBSD 4.7: truncate | |
a1d5f77c MK |
674 | .PP |
675 | There are many infelicities in the protocol underlying NFS, affecting | |
676 | amongst others | |
677 | .BR O_SYNC " and " O_NDELAY . | |
678 | ||
d9bfdb9c | 679 | POSIX provides for three different variants of synchronized I/O, |
015221ef CH |
680 | corresponding to the flags |
681 | .BR O_SYNC , | |
682 | .BR O_DSYNC , | |
683 | and | |
684 | .BR O_RSYNC . | |
685 | Currently (2.6.31), Linux only implements | |
686 | .BR O_SYNC , | |
687 | but glibc maps | |
688 | .B O_DSYNC | |
689 | and | |
690 | .B O_RSYNC | |
691 | to the same numerical value as | |
0a598d26 | 692 | .BR O_SYNC . |
245dec52 | 693 | Most Linux file systems don't actually implement the POSIX |
015221ef CH |
694 | .B O_SYNC |
695 | semantics, which require all metadata updates of a write | |
696 | to be on disk on returning to userspace, but only the | |
697 | .B O_DSYNC | |
698 | semantics, which require only actual file data and metadata necessary | |
699 | to retrieve it to be on disk by the time the system call returns. | |
a1d5f77c MK |
700 | |
701 | Note that | |
702 | .BR open () | |
703 | can open device special files, but | |
704 | .BR creat () | |
705 | cannot create them; use | |
706 | .BR mknod (2) | |
707 | instead. | |
708 | .LP | |
709 | On NFS file systems with UID mapping enabled, | |
710 | .BR open () | |
711 | may | |
75b94dc3 | 712 | return a file descriptor but, for example, |
a1d5f77c MK |
713 | .BR read (2) |
714 | requests are denied | |
715 | with \fBEACCES\fP. | |
716 | This is because the client performs | |
717 | .BR open () | |
718 | by checking the | |
719 | permissions, but UID mapping is performed by the server upon | |
720 | read and write requests. | |
721 | ||
722 | If the file is newly created, its | |
988db661 | 723 | .IR st_atime , |
a1d5f77c MK |
724 | .IR st_ctime , |
725 | .I st_mtime | |
726 | fields | |
727 | (respectively, time of last access, time of last status change, and | |
728 | time of last modification; see | |
729 | .BR stat (2)) | |
730 | are set | |
731 | to the current time, and so are the | |
732 | .I st_ctime | |
988db661 | 733 | and |
a1d5f77c MK |
734 | .I st_mtime |
735 | fields of the | |
736 | parent directory. | |
988db661 | 737 | Otherwise, if the file is modified because of the |
a1d5f77c MK |
738 | .B O_TRUNC |
739 | flag, its st_ctime and st_mtime fields are set to the current time. | |
ddc4d339 MK |
740 | .SS O_DIRECT |
741 | .LP | |
742 | The | |
743 | .B O_DIRECT | |
744 | flag may impose alignment restrictions on the length and address | |
745 | of userspace buffers and the file offset of I/Os. | |
746 | In Linux alignment | |
24d01c53 | 747 | restrictions vary by file system and kernel version and might be |
ddc4d339 | 748 | absent entirely. |
24d01c53 | 749 | However there is currently no file system\-independent |
ddc4d339 | 750 | interface for an application to discover these restrictions for a given |
24d01c53 MK |
751 | file or file system. |
752 | Some file systems provide their own interfaces | |
ddc4d339 MK |
753 | for doing so, for example the |
754 | .B XFS_IOC_DIOINFO | |
755 | operation in | |
756 | .BR xfsctl (3). | |
757 | .LP | |
85c2bdba MK |
758 | Under Linux 2.4, transfer sizes, and the alignment of the user buffer |
759 | and the file offset must all be multiples of the logical block size | |
ddc4d339 MK |
760 | of the file system. |
761 | Under Linux 2.6, alignment to 512-byte boundaries | |
762 | suffices. | |
763 | .LP | |
764 | The | |
765 | .B O_DIRECT | |
766 | flag was introduced in SGI IRIX, where it has alignment | |
767 | restrictions similar to those of Linux 2.4. | |
768 | IRIX has also a | |
769 | .BR fcntl (2) | |
770 | call to query appropriate alignments, and sizes. | |
771 | FreeBSD 4.x introduced | |
772 | a flag of the same name, but without alignment restrictions. | |
773 | .LP | |
774 | .B O_DIRECT | |
775 | support was added under Linux in kernel version 2.4.10. | |
776 | Older Linux kernels simply ignore this flag. | |
24d01c53 | 777 | Some file systems may not implement the flag and |
ddc4d339 MK |
778 | .BR open () |
779 | will fail with | |
780 | .B EINVAL | |
781 | if it is used. | |
782 | .LP | |
783 | Applications should avoid mixing | |
784 | .B O_DIRECT | |
785 | and normal I/O to the same file, | |
786 | and especially to overlapping byte regions in the same file. | |
24d01c53 | 787 | Even when the file system correctly handles the coherency issues in |
ddc4d339 MK |
788 | this situation, overall I/O throughput is likely to be slower than |
789 | using either mode alone. | |
790 | Likewise, applications should avoid mixing | |
791 | .BR mmap (2) | |
792 | of files with direct I/O to the same files. | |
793 | .LP | |
794 | The behaviour of | |
795 | .B O_DIRECT | |
24d01c53 | 796 | with NFS will differ from local file systems. |
ddc4d339 MK |
797 | Older kernels, or |
798 | kernels configured in certain ways, may not support this combination. | |
799 | The NFS protocol does not support passing the flag to the server, so | |
800 | .B O_DIRECT | |
801 | I/O will only bypass the page cache on the client; the server may | |
802 | still cache the I/O. | |
803 | The client asks the server to make the I/O | |
804 | synchronous to preserve the synchronous semantics of | |
805 | .BR O_DIRECT . | |
806 | Some servers will perform poorly under these circumstances, especially | |
807 | if the I/O size is small. | |
808 | Some servers may also be configured to | |
809 | lie to clients about the I/O having reached stable storage; this | |
810 | will avoid the performance penalty at some risk to data integrity | |
811 | in the event of server power failure. | |
812 | The Linux NFS client places no alignment restrictions on | |
813 | .B O_DIRECT | |
814 | I/O. | |
815 | .PP | |
816 | In summary, | |
817 | .B O_DIRECT | |
818 | is a potentially powerful tool that should be used with caution. | |
819 | It is recommended that applications treat use of | |
820 | .B O_DIRECT | |
821 | as a performance option which is disabled by default. | |
822 | .PP | |
823 | .RS | |
fea681da MK |
824 | "The thing that has always disturbed me about O_DIRECT is that the whole |
825 | interface is just stupid, and was probably designed by a deranged monkey | |
e9496f74 | 826 | on some serious mind-controlling substances." \(em Linus |
ddc4d339 MK |
827 | .RE |
828 | .SH BUGS | |
b50582eb MK |
829 | Currently, it is not possible to enable signal-driven |
830 | I/O by specifying | |
831 | .B O_ASYNC | |
c13182ef | 832 | when calling |
b50582eb MK |
833 | .BR open (); |
834 | use | |
835 | .BR fcntl (2) | |
836 | to enable this flag. | |
0e1ad98c | 837 | .\" FIXME . Check bugzilla report on open(O_ASYNC) |
92057f4d | 838 | .\" See http://bugzilla.kernel.org/show_bug.cgi?id=5993 |
fea681da | 839 | .SH "SEE ALSO" |
a3bf8022 MK |
840 | .BR chmod (2), |
841 | .BR chown (2), | |
fea681da | 842 | .BR close (2), |
e366dbc4 | 843 | .BR dup (2), |
fea681da MK |
844 | .BR fcntl (2), |
845 | .BR link (2), | |
1f6ceb40 | 846 | .BR lseek (2), |
fea681da | 847 | .BR mknod (2), |
e366dbc4 | 848 | .BR mmap (2), |
f0c34053 | 849 | .BR mount (2), |
28c54d45 | 850 | .BR openat (2), |
fea681da MK |
851 | .BR read (2), |
852 | .BR socket (2), | |
853 | .BR stat (2), | |
854 | .BR umask (2), | |
855 | .BR unlink (2), | |
856 | .BR write (2), | |
857 | .BR fopen (3), | |
ad7cc990 | 858 | .BR feature_test_macros (7), |
f0c34053 | 859 | .BR fifo (7), |
a9cfde1d MK |
860 | .BR path_resolution (7), |
861 | .BR symlink (7) |