1FCNTL(2) Linux Programmer's Manual FCNTL(2)
2
6 fcntl - manipulate file descriptor
7
9 #include <unistd.h>
10 #include <fcntl.h>
11 int fcntl(int fd, int cmd, ... /* arg */ );
13
15 fcntl() performs one of the operations described below on the open file
16 descriptor fd. The operation is determined by cmd.
17 fcntl() can take an optional third argument. Whether or not this argu‐
19 ment is required is determined by cmd. The required argument type is
20 indicated in parentheses after each cmd name (in most cases, the re‐
21 quired type is int, and we identify the argument using the name arg),
22 or void is specified if the argument is not required.
23 Certain of the operations below are supported only since a particular
25 Linux kernel version. The preferred method of checking whether the
26 host kernel supports a particular operation is to invoke fcntl() with
27 the desired cmd value and then test whether the call failed with EIN‐
28 VAL, indicating that the kernel does not recognize this value.
29 Duplicating a file descriptor
31 F_DUPFD (int)
32 Duplicate the file descriptor fd using the lowest-numbered
33 available file descriptor greater than or equal to arg. This is
34 different from dup2(2), which uses exactly the file descriptor
35 specified.
36 On success, the new file descriptor is returned.
38 See dup(2) for further details.
40 F_DUPFD_CLOEXEC (int; since Linux 2.6.24)
42 As for F_DUPFD, but additionally set the close-on-exec flag for
43 the duplicate file descriptor. Specifying this flag permits a
44 program to avoid an additional fcntl() F_SETFD operation to set
45 the FD_CLOEXEC flag. For an explanation of why this flag is
46 useful, see the description of O_CLOEXEC in open(2).
47 File descriptor flags
49 The following commands manipulate the flags associated with a file de‐
50 scriptor. Currently, only one such flag is defined: FD_CLOEXEC, the
51 close-on-exec flag. If the FD_CLOEXEC bit is set, the file descriptor
52 will automatically be closed during a successful execve(2). (If the
53 execve(2) fails, the file descriptor is left open.) If the FD_CLOEXEC
54 bit is not set, the file descriptor will remain open across an ex‐
55 ecve(2).
56 F_GETFD (void)
58 Return (as the function result) the file descriptor flags; arg
59 is ignored.
60 F_SETFD (int)
62 Set the file descriptor flags to the value specified by arg.
63 In multithreaded programs, using fcntl() F_SETFD to set the close-on-
65 exec flag at the same time as another thread performs a fork(2) plus
66 execve(2) is vulnerable to a race condition that may unintentionally
67 leak the file descriptor to the program executed in the child process.
68 See the discussion of the O_CLOEXEC flag in open(2) for details and a
69 remedy to the problem.
70 File status flags
72 Each open file description has certain associated status flags, ini‐
73 tialized by open(2) and possibly modified by fcntl(). Duplicated file
74 descriptors (made with dup(2), fcntl(F_DUPFD), fork(2), etc.) refer to
75 the same open file description, and thus share the same file status
76 flags.
77 The file status flags and their semantics are described in open(2).
79 F_GETFL (void)
81 Return (as the function result) the file access mode and the
82 file status flags; arg is ignored.
83 F_SETFL (int)
85 Set the file status flags to the value specified by arg. File
86 access mode (O_RDONLY, O_WRONLY, O_RDWR) and file creation flags
87 (i.e., O_CREAT, O_EXCL, O_NOCTTY, O_TRUNC) in arg are ignored.
88 On Linux, this command can change only the O_APPEND, O_ASYNC,
89 O_DIRECT, O_NOATIME, and O_NONBLOCK flags. It is not possible
90 to change the O_DSYNC and O_SYNC flags; see BUGS, below.
91 Advisory record locking
93 Linux implements traditional ("process-associated") UNIX record locks,
94 as standardized by POSIX. For a Linux-specific alternative with better
95 semantics, see the discussion of open file description locks below.
96 F_SETLK, F_SETLKW, and F_GETLK are used to acquire, release, and test
98 for the existence of record locks (also known as byte-range, file-seg‐
99 ment, or file-region locks). The third argument, lock, is a pointer to
100 a structure that has at least the following fields (in unspecified or‐
101 der).
102 struct flock {
104 ...
105 short l_type; /* Type of lock: F_RDLCK,
106 F_WRLCK, F_UNLCK */
107 short l_whence; /* How to interpret l_start:
108 SEEK_SET, SEEK_CUR, SEEK_END */
109 off_t l_start; /* Starting offset for lock */
110 off_t l_len; /* Number of bytes to lock */
111 pid_t l_pid; /* PID of process blocking our lock
112 (set by F_GETLK and F_OFD_GETLK) */
113 ...
114 };
115 The l_whence, l_start, and l_len fields of this structure specify the
117 range of bytes we wish to lock. Bytes past the end of the file may be
118 locked, but not bytes before the start of the file.
119 l_start is the starting offset for the lock, and is interpreted rela‐
121 tive to either: the start of the file (if l_whence is SEEK_SET); the
122 current file offset (if l_whence is SEEK_CUR); or the end of the file
123 (if l_whence is SEEK_END). In the final two cases, l_start can be a
124 negative number provided the offset does not lie before the start of
125 the file.
126 l_len specifies the number of bytes to be locked. If l_len is posi‐
128 tive, then the range to be locked covers bytes l_start up to and in‐
129 cluding l_start+l_len-1. Specifying 0 for l_len has the special mean‐
130 ing: lock all bytes starting at the location specified by l_whence and
131 l_start through to the end of file, no matter how large the file grows.
132 POSIX.1-2001 allows (but does not require) an implementation to support
134 a negative l_len value; if l_len is negative, the interval described by
135 lock covers bytes l_start+l_len up to and including l_start-1. This is
136 supported by Linux since kernel versions 2.4.21 and 2.5.49.
137 The l_type field can be used to place a read (F_RDLCK) or a write
139 (F_WRLCK) lock on a file. Any number of processes may hold a read lock
140 (shared lock) on a file region, but only one process may hold a write
141 lock (exclusive lock). An exclusive lock excludes all other locks,
142 both shared and exclusive. A single process can hold only one type of
143 lock on a file region; if a new lock is applied to an already-locked
144 region, then the existing lock is converted to the new lock type.
145 (Such conversions may involve splitting, shrinking, or coalescing with
146 an existing lock if the byte range specified by the new lock does not
147 precisely coincide with the range of the existing lock.)
148 F_SETLK (struct flock *)
150 Acquire a lock (when l_type is F_RDLCK or F_WRLCK) or release a
151 lock (when l_type is F_UNLCK) on the bytes specified by the
152 l_whence, l_start, and l_len fields of lock. If a conflicting
153 lock is held by another process, this call returns -1 and sets
154 errno to EACCES or EAGAIN. (The error returned in this case
155 differs across implementations, so POSIX requires a portable ap‐
156 plication to check for both errors.)
157 F_SETLKW (struct flock *)
159 As for F_SETLK, but if a conflicting lock is held on the file,
160 then wait for that lock to be released. If a signal is caught
161 while waiting, then the call is interrupted and (after the sig‐
162 nal handler has returned) returns immediately (with return value
163 -1 and errno set to EINTR; see signal(7)).
164 F_GETLK (struct flock *)
166 On input to this call, lock describes a lock we would like to
167 place on the file. If the lock could be placed, fcntl() does
168 not actually place it, but returns F_UNLCK in the l_type field
169 of lock and leaves the other fields of the structure unchanged.
170 If one or more incompatible locks would prevent this lock being
172 placed, then fcntl() returns details about one of those locks in
173 the l_type, l_whence, l_start, and l_len fields of lock. If the
174 conflicting lock is a traditional (process-associated) record
175 lock, then the l_pid field is set to the PID of the process
176 holding that lock. If the conflicting lock is an open file de‐
177 scription lock, then l_pid is set to -1. Note that the returned
178 information may already be out of date by the time the caller
179 inspects it.
180 In order to place a read lock, fd must be open for reading. In order
182 to place a write lock, fd must be open for writing. To place both
183 types of lock, open a file read-write.
184 When placing locks with F_SETLKW, the kernel detects deadlocks, whereby
186 two or more processes have their lock requests mutually blocked by
187 locks held by the other processes. For example, suppose process A
188 holds a write lock on byte 100 of a file, and process B holds a write
189 lock on byte 200. If each process then attempts to lock the byte al‐
190 ready locked by the other process using F_SETLKW, then, without dead‐
191 lock detection, both processes would remain blocked indefinitely. When
192 the kernel detects such deadlocks, it causes one of the blocking lock
193 requests to immediately fail with the error EDEADLK; an application
194 that encounters such an error should release some of its locks to allow
195 other applications to proceed before attempting regain the locks that
196 it requires. Circular deadlocks involving more than two processes are
197 also detected. Note, however, that there are limitations to the ker‐
198 nel's deadlock-detection algorithm; see BUGS.
199 As well as being removed by an explicit F_UNLCK, record locks are auto‐
201 matically released when the process terminates.
202 Record locks are not inherited by a child created via fork(2), but are
204 preserved across an execve(2).
205 Because of the buffering performed by the stdio(3) library, the use of
207 record locking with routines in that package should be avoided; use
208 read(2) and write(2) instead.
209 The record locks described above are associated with the process (un‐
211 like the open file description locks described below). This has some
212 unfortunate consequences:
213 * If a process closes any file descriptor referring to a file, then
215 all of the process's locks on that file are released, regardless of
216 the file descriptor(s) on which the locks were obtained. This is
217 bad: it means that a process can lose its locks on a file such as
218 /etc/passwd or /etc/mtab when for some reason a library function de‐
219 cides to open, read, and close the same file.
220 * The threads in a process share locks. In other words, a multi‐
222 threaded program can't use record locking to ensure that threads
223 don't simultaneously access the same region of a file.
224 Open file description locks solve both of these problems.
226 Open file description locks (non-POSIX)
228 Open file description locks are advisory byte-range locks whose opera‐
229 tion is in most respects identical to the traditional record locks de‐
230 scribed above. This lock type is Linux-specific, and available since
231 Linux 3.15. (There is a proposal with the Austin Group to include this
232 lock type in the next revision of POSIX.1.) For an explanation of open
233 file descriptions, see open(2).
234 The principal difference between the two lock types is that whereas
236 traditional record locks are associated with a process, open file de‐
237 scription locks are associated with the open file description on which
238 they are acquired, much like locks acquired with flock(2). Conse‐
239 quently (and unlike traditional advisory record locks), open file de‐
240 scription locks are inherited across fork(2) (and clone(2) with
241 CLONE_FILES), and are only automatically released on the last close of
242 the open file description, instead of being released on any close of
243 the file.
244 Conflicting lock combinations (i.e., a read lock and a write lock or
246 two write locks) where one lock is an open file description lock and
247 the other is a traditional record lock conflict even when they are ac‐
248 quired by the same process on the same file descriptor.
249 Open file description locks placed via the same open file description
251 (i.e., via the same file descriptor, or via a duplicate of the file de‐
252 scriptor created by fork(2), dup(2), fcntl() F_DUPFD, and so on) are
253 always compatible: if a new lock is placed on an already locked region,
254 then the existing lock is converted to the new lock type. (Such con‐
255 versions may result in splitting, shrinking, or coalescing with an ex‐
256 isting lock as discussed above.)
257 On the other hand, open file description locks may conflict with each
259 other when they are acquired via different open file descriptions.
260 Thus, the threads in a multithreaded program can use open file descrip‐
261 tion locks to synchronize access to a file region by having each thread
262 perform its own open(2) on the file and applying locks via the result‐
263 ing file descriptor.
264 As with traditional advisory locks, the third argument to fcntl(),
266 lock, is a pointer to an flock structure. By contrast with traditional
267 record locks, the l_pid field of that structure must be set to zero
268 when using the commands described below.
269 The commands for working with open file description locks are analogous
271 to those used with traditional locks:
272 F_OFD_SETLK (struct flock *)
274 Acquire an open file description lock (when l_type is F_RDLCK or
275 F_WRLCK) or release an open file description lock (when l_type
276 is F_UNLCK) on the bytes specified by the l_whence, l_start, and
277 l_len fields of lock. If a conflicting lock is held by another
278 process, this call returns -1 and sets errno to EAGAIN.
279 F_OFD_SETLKW (struct flock *)
281 As for F_OFD_SETLK, but if a conflicting lock is held on the
282 file, then wait for that lock to be released. If a signal is
283 caught while waiting, then the call is interrupted and (after
284 the signal handler has returned) returns immediately (with re‐
285 turn value -1 and errno set to EINTR; see signal(7)).
286 F_OFD_GETLK (struct flock *)
288 On input to this call, lock describes an open file description
289 lock we would like to place on the file. If the lock could be
290 placed, fcntl() does not actually place it, but returns F_UNLCK
291 in the l_type field of lock and leaves the other fields of the
292 structure unchanged. If one or more incompatible locks would
293 prevent this lock being placed, then details about one of these
294 locks are returned via lock, as described above for F_GETLK.
295 In the current implementation, no deadlock detection is performed for
297 open file description locks. (This contrasts with process-associated
298 record locks, for which the kernel does perform deadlock detection.)
299 Mandatory locking
301 Warning: the Linux implementation of mandatory locking is unreliable.
302 See BUGS below. Because of these bugs, and the fact that the feature
303 is believed to be little used, since Linux 4.5, mandatory locking has
304 been made an optional feature, governed by a configuration option (CON‐
305 FIG_MANDATORY_FILE_LOCKING). This is an initial step toward removing
306 this feature completely.
307 By default, both traditional (process-associated) and open file de‐
309 scription record locks are advisory. Advisory locks are not enforced
310 and are useful only between cooperating processes.
311 Both lock types can also be mandatory. Mandatory locks are enforced
313 for all processes. If a process tries to perform an incompatible ac‐
314 cess (e.g., read(2) or write(2)) on a file region that has an incompat‐
315 ible mandatory lock, then the result depends upon whether the O_NON‐
316 BLOCK flag is enabled for its open file description. If the O_NONBLOCK
317 flag is not enabled, then the system call is blocked until the lock is
318 removed or converted to a mode that is compatible with the access. If
319 the O_NONBLOCK flag is enabled, then the system call fails with the er‐
320 ror EAGAIN.
321 To make use of mandatory locks, mandatory locking must be enabled both
323 on the filesystem that contains the file to be locked, and on the file
324 itself. Mandatory locking is enabled on a filesystem using the "-o
325 mand" option to mount(8), or the MS_MANDLOCK flag for mount(2). Manda‐
326 tory locking is enabled on a file by disabling group execute permission
327 on the file and enabling the set-group-ID permission bit (see chmod(1)
328 and chmod(2)).
329 Mandatory locking is not specified by POSIX. Some other systems also
331 support mandatory locking, although the details of how to enable it
332 vary across systems.
333 Lost locks
335 When an advisory lock is obtained on a networked filesystem such as NFS
336 it is possible that the lock might get lost. This may happen due to
337 administrative action on the server, or due to a network partition
338 (i.e., loss of network connectivity with the server) which lasts long
339 enough for the server to assume that the client is no longer function‐
340 ing.
341 When the filesystem determines that a lock has been lost, future
343 read(2) or write(2) requests may fail with the error EIO. This error
344 will persist until the lock is removed or the file descriptor is
345 closed. Since Linux 3.12, this happens at least for NFSv4 (including
346 all minor versions).
347 Some versions of UNIX send a signal (SIGLOST) in this circumstance.
349 Linux does not define this signal, and does not provide any asynchro‐
350 nous notification of lost locks.
351 Managing signals
353 F_GETOWN, F_SETOWN, F_GETOWN_EX, F_SETOWN_EX, F_GETSIG, and F_SETSIG
354 are used to manage I/O availability signals:
355 F_GETOWN (void)
357 Return (as the function result) the process ID or process group
358 ID currently receiving SIGIO and SIGURG signals for events on
359 file descriptor fd. Process IDs are returned as positive val‐
360 ues; process group IDs are returned as negative values (but see
361 BUGS below). arg is ignored.
362 F_SETOWN (int)
364 Set the process ID or process group ID that will receive SIGIO
365 and SIGURG signals for events on the file descriptor fd. The
366 target process or process group ID is specified in arg. A
367 process ID is specified as a positive value; a process group ID
368 is specified as a negative value. Most commonly, the calling
369 process specifies itself as the owner (that is, arg is specified
370 as getpid(2)).
371 As well as setting the file descriptor owner, one must also en‐
373 able generation of signals on the file descriptor. This is done
374 by using the fcntl() F_SETFL command to set the O_ASYNC file
375 status flag on the file descriptor. Subsequently, a SIGIO sig‐
376 nal is sent whenever input or output becomes possible on the
377 file descriptor. The fcntl() F_SETSIG command can be used to
378 obtain delivery of a signal other than SIGIO.
379 Sending a signal to the owner process (group) specified by F_SE‐
381 TOWN is subject to the same permissions checks as are described
382 for kill(2), where the sending process is the one that employs
383 F_SETOWN (but see BUGS below). If this permission check fails,
384 then the signal is silently discarded. Note: The F_SETOWN oper‐
385 ation records the caller's credentials at the time of the fc‐
386 ntl() call, and it is these saved credentials that are used for
387 the permission checks.
388 If the file descriptor fd refers to a socket, F_SETOWN also se‐
390 lects the recipient of SIGURG signals that are delivered when
391 out-of-band data arrives on that socket. (SIGURG is sent in any
392 situation where select(2) would report the socket as having an
393 "exceptional condition".)
394 The following was true in 2.6.x kernels up to and including ker‐
396 nel 2.6.11:
397 If a nonzero value is given to F_SETSIG in a multi‐
399 threaded process running with a threading library that
400 supports thread groups (e.g., NPTL), then a positive
401 value given to F_SETOWN has a different meaning: instead
402 of being a process ID identifying a whole process, it is
403 a thread ID identifying a specific thread within a
404 process. Consequently, it may be necessary to pass F_SE‐
405 TOWN the result of gettid(2) instead of getpid(2) to get
406 sensible results when F_SETSIG is used. (In current
407 Linux threading implementations, a main thread's thread
408 ID is the same as its process ID. This means that a sin‐
409 gle-threaded program can equally use gettid(2) or get‐
410 pid(2) in this scenario.) Note, however, that the state‐
411 ments in this paragraph do not apply to the SIGURG signal
412 generated for out-of-band data on a socket: this signal
413 is always sent to either a process or a process group,
414 depending on the value given to F_SETOWN.
415 The above behavior was accidentally dropped in Linux 2.6.12, and
417 won't be restored. From Linux 2.6.32 onward, use F_SETOWN_EX to
418 target SIGIO and SIGURG signals at a particular thread.
419 F_GETOWN_EX (struct f_owner_ex *) (since Linux 2.6.32)
421 Return the current file descriptor owner settings as defined by
422 a previous F_SETOWN_EX operation. The information is returned
423 in the structure pointed to by arg, which has the following
424 form:
425 struct f_owner_ex {
427 int type;
428 pid_t pid;
429 };
430 The type field will have one of the values F_OWNER_TID,
432 F_OWNER_PID, or F_OWNER_PGRP. The pid field is a positive inte‐
433 ger representing a thread ID, process ID, or process group ID.
434 See F_SETOWN_EX for more details.
435 F_SETOWN_EX (struct f_owner_ex *) (since Linux 2.6.32)
437 This operation performs a similar task to F_SETOWN. It allows
438 the caller to direct I/O availability signals to a specific
439 thread, process, or process group. The caller specifies the
440 target of signals via arg, which is a pointer to a f_owner_ex
441 structure. The type field has one of the following values,
442 which define how pid is interpreted:
443 F_OWNER_TID
445 Send the signal to the thread whose thread ID (the value
446 returned by a call to clone(2) or gettid(2)) is specified
447 in pid.
448 F_OWNER_PID
450 Send the signal to the process whose ID is specified in
451 pid.
452 F_OWNER_PGRP
454 Send the signal to the process group whose ID is speci‐
455 fied in pid. (Note that, unlike with F_SETOWN, a process
456 group ID is specified as a positive value here.)
457 F_GETSIG (void)
459 Return (as the function result) the signal sent when input or
460 output becomes possible. A value of zero means SIGIO is sent.
461 Any other value (including SIGIO) is the signal sent instead,
462 and in this case additional info is available to the signal han‐
463 dler if installed with SA_SIGINFO. arg is ignored.
464 F_SETSIG (int)
466 Set the signal sent when input or output becomes possible to the
467 value given in arg. A value of zero means to send the default
468 SIGIO signal. Any other value (including SIGIO) is the signal
469 to send instead, and in this case additional info is available
470 to the signal handler if installed with SA_SIGINFO.
471 By using F_SETSIG with a nonzero value, and setting SA_SIGINFO
473 for the signal handler (see sigaction(2)), extra information
474 about I/O events is passed to the handler in a siginfo_t struc‐
475 ture. If the si_code field indicates the source is SI_SIGIO,
476 the si_fd field gives the file descriptor associated with the
477 event. Otherwise, there is no indication which file descriptors
478 are pending, and you should use the usual mechanisms (select(2),
479 poll(2), read(2) with O_NONBLOCK set etc.) to determine which
480 file descriptors are available for I/O.
481 Note that the file descriptor provided in si_fd is the one that
483 was specified during the F_SETSIG operation. This can lead to
484 an unusual corner case. If the file descriptor is duplicated
485 (dup(2) or similar), and the original file descriptor is closed,
486 then I/O events will continue to be generated, but the si_fd
487 field will contain the number of the now closed file descriptor.
488 By selecting a real time signal (value >= SIGRTMIN), multiple
490 I/O events may be queued using the same signal numbers. (Queu‐
491 ing is dependent on available memory.) Extra information is
492 available if SA_SIGINFO is set for the signal handler, as above.
493 Note that Linux imposes a limit on the number of real-time sig‐
495 nals that may be queued to a process (see getrlimit(2) and sig‐
496 nal(7)) and if this limit is reached, then the kernel reverts to
497 delivering SIGIO, and this signal is delivered to the entire
498 process rather than to a specific thread.
499 Using these mechanisms, a program can implement fully asynchronous I/O
501 without using select(2) or poll(2) most of the time.
502 The use of O_ASYNC is specific to BSD and Linux. The only use of
504 F_GETOWN and F_SETOWN specified in POSIX.1 is in conjunction with the
505 use of the SIGURG signal on sockets. (POSIX does not specify the SIGIO
506 signal.) F_GETOWN_EX, F_SETOWN_EX, F_GETSIG, and F_SETSIG are Linux-
507 specific. POSIX has asynchronous I/O and the aio_sigevent structure to
508 achieve similar things; these are also available in Linux as part of
509 the GNU C Library (Glibc).
510 Leases
512 F_SETLEASE and F_GETLEASE (Linux 2.4 onward) are used to establish a
513 new lease, and retrieve the current lease, on the open file description
514 referred to by the file descriptor fd. A file lease provides a mecha‐
515 nism whereby the process holding the lease (the "lease holder") is no‐
516 tified (via delivery of a signal) when a process (the "lease breaker")
517 tries to open(2) or truncate(2) the file referred to by that file de‐
518 scriptor.
519 F_SETLEASE (int)
521 Set or remove a file lease according to which of the following
522 values is specified in the integer arg:
523 F_RDLCK
525 Take out a read lease. This will cause the calling
526 process to be notified when the file is opened for writ‐
527 ing or is truncated. A read lease can be placed only on
528 a file descriptor that is opened read-only.
529 F_WRLCK
531 Take out a write lease. This will cause the caller to be
532 notified when the file is opened for reading or writing
533 or is truncated. A write lease may be placed on a file
534 only if there are no other open file descriptors for the
535 file.
536 F_UNLCK
538 Remove our lease from the file.
539 Leases are associated with an open file description (see open(2)).
541 This means that duplicate file descriptors (created by, for example,
542 fork(2) or dup(2)) refer to the same lease, and this lease may be modi‐
543 fied or released using any of these descriptors. Furthermore, the
544 lease is released by either an explicit F_UNLCK operation on any of
545 these duplicate file descriptors, or when all such file descriptors
546 have been closed.
547 Leases may be taken out only on regular files. An unprivileged process
549 may take out a lease only on a file whose UID (owner) matches the
550 filesystem UID of the process. A process with the CAP_LEASE capability
551 may take out leases on arbitrary files.
552 F_GETLEASE (void)
554 Indicates what type of lease is associated with the file de‐
555 scriptor fd by returning either F_RDLCK, F_WRLCK, or F_UNLCK,
556 indicating, respectively, a read lease , a write lease, or no
557 lease. arg is ignored.
558 When a process (the "lease breaker") performs an open(2) or truncate(2)
560 that conflicts with a lease established via F_SETLEASE, the system call
561 is blocked by the kernel and the kernel notifies the lease holder by
562 sending it a signal (SIGIO by default). The lease holder should re‐
563 spond to receipt of this signal by doing whatever cleanup is required
564 in preparation for the file to be accessed by another process (e.g.,
565 flushing cached buffers) and then either remove or downgrade its lease.
566 A lease is removed by performing an F_SETLEASE command specifying arg
567 as F_UNLCK. If the lease holder currently holds a write lease on the
568 file, and the lease breaker is opening the file for reading, then it is
569 sufficient for the lease holder to downgrade the lease to a read lease.
570 This is done by performing an F_SETLEASE command specifying arg as
571 F_RDLCK.
572 If the lease holder fails to downgrade or remove the lease within the
574 number of seconds specified in /proc/sys/fs/lease-break-time, then the
575 kernel forcibly removes or downgrades the lease holder's lease.
576 Once a lease break has been initiated, F_GETLEASE returns the target
578 lease type (either F_RDLCK or F_UNLCK, depending on what would be com‐
579 patible with the lease breaker) until the lease holder voluntarily
580 downgrades or removes the lease or the kernel forcibly does so after
581 the lease break timer expires.
582 Once the lease has been voluntarily or forcibly removed or downgraded,
584 and assuming the lease breaker has not unblocked its system call, the
585 kernel permits the lease breaker's system call to proceed.
586 If the lease breaker's blocked open(2) or truncate(2) is interrupted by
588 a signal handler, then the system call fails with the error EINTR, but
589 the other steps still occur as described above. If the lease breaker
590 is killed by a signal while blocked in open(2) or truncate(2), then the
591 other steps still occur as described above. If the lease breaker spec‐
592 ifies the O_NONBLOCK flag when calling open(2), then the call immedi‐
593 ately fails with the error EWOULDBLOCK, but the other steps still occur
594 as described above.
595 The default signal used to notify the lease holder is SIGIO, but this
597 can be changed using the F_SETSIG command to fcntl(). If a F_SETSIG
598 command is performed (even one specifying SIGIO), and the signal han‐
599 dler is established using SA_SIGINFO, then the handler will receive a
600 siginfo_t structure as its second argument, and the si_fd field of this
601 argument will hold the file descriptor of the leased file that has been
602 accessed by another process. (This is useful if the caller holds
603 leases against multiple files.)
604 File and directory change notification (dnotify)
606 F_NOTIFY (int)
607 (Linux 2.4 onward) Provide notification when the directory re‐
608 ferred to by fd or any of the files that it contains is changed.
609 The events to be notified are specified in arg, which is a bit
610 mask specified by ORing together zero or more of the following
611 bits:
612 DN_ACCESS
614 A file was accessed (read(2), pread(2), readv(2), and
615 similar)
616 DN_MODIFY
617 A file was modified (write(2), pwrite(2), writev(2),
618 truncate(2), ftruncate(2), and similar).
619 DN_CREATE
620 A file was created (open(2), creat(2), mknod(2),
621 mkdir(2), link(2), symlink(2), rename(2) into this direc‐
622 tory).
623 DN_DELETE
624 A file was unlinked (unlink(2), rename(2) to another di‐
625 rectory, rmdir(2)).
626 DN_RENAME
627 A file was renamed within this directory (rename(2)).
628 DN_ATTRIB
629 The attributes of a file were changed (chown(2),
630 chmod(2), utime(2), utimensat(2), and similar).
631 (In order to obtain these definitions, the _GNU_SOURCE feature
633 test macro must be defined before including any header files.)
634 Directory notifications are normally "one-shot", and the appli‐
636 cation must reregister to receive further notifications. Alter‐
637 natively, if DN_MULTISHOT is included in arg, then notification
638 will remain in effect until explicitly removed.
639 A series of F_NOTIFY requests is cumulative, with the events in
641 arg being added to the set already monitored. To disable noti‐
642 fication of all events, make an F_NOTIFY call specifying arg as
643 0.
644 Notification occurs via delivery of a signal. The default sig‐
646 nal is SIGIO, but this can be changed using the F_SETSIG command
647 to fcntl(). (Note that SIGIO is one of the nonqueuing standard
648 signals; switching to the use of a real-time signal means that
649 multiple notifications can be queued to the process.) In the
650 latter case, the signal handler receives a siginfo_t structure
651 as its second argument (if the handler was established using
652 SA_SIGINFO) and the si_fd field of this structure contains the
653 file descriptor which generated the notification (useful when
654 establishing notification on multiple directories).
655 Especially when using DN_MULTISHOT, a real time signal should be
657 used for notification, so that multiple notifications can be
658 queued.
659 NOTE: New applications should use the inotify interface (avail‐
661 able since kernel 2.6.13), which provides a much superior inter‐
662 face for obtaining notifications of filesystem events. See ino‐
663 tify(7).
664 Changing the capacity of a pipe
666 F_SETPIPE_SZ (int; since Linux 2.6.35)
667 Change the capacity of the pipe referred to by fd to be at least
668 arg bytes. An unprivileged process can adjust the pipe capacity
669 to any value between the system page size and the limit defined
670 in /proc/sys/fs/pipe-max-size (see proc(5)). Attempts to set
671 the pipe capacity below the page size are silently rounded up to
672 the page size. Attempts by an unprivileged process to set the
673 pipe capacity above the limit in /proc/sys/fs/pipe-max-size
674 yield the error EPERM; a privileged process (CAP_SYS_RESOURCE)
675 can override the limit.
676 When allocating the buffer for the pipe, the kernel may use a
678 capacity larger than arg, if that is convenient for the imple‐
679 mentation. (In the current implementation, the allocation is
680 the next higher power-of-two page-size multiple of the requested
681 size.) The actual capacity (in bytes) that is set is returned
682 as the function result.
683 Attempting to set the pipe capacity smaller than the amount of
685 buffer space currently used to store data produces the error
686 EBUSY.
687 Note that because of the way the pages of the pipe buffer are
689 employed when data is written to the pipe, the number of bytes
690 that can be written may be less than the nominal size, depending
691 on the size of the writes.
692 F_GETPIPE_SZ (void; since Linux 2.6.35)
694 Return (as the function result) the capacity of the pipe re‐
695 ferred to by fd.
696 File Sealing
698 File seals limit the set of allowed operations on a given file. For
699 each seal that is set on a file, a specific set of operations will fail
700 with EPERM on this file from now on. The file is said to be sealed.
701 The default set of seals depends on the type of the underlying file and
702 filesystem. For an overview of file sealing, a discussion of its pur‐
703 pose, and some code examples, see memfd_create(2).
704 Currently, file seals can be applied only to a file descriptor returned
706 by memfd_create(2) (if the MFD_ALLOW_SEALING was employed). On other
707 filesystems, all fcntl() operations that operate on seals will return
708 EINVAL.
709 Seals are a property of an inode. Thus, all open file descriptors re‐
711 ferring to the same inode share the same set of seals. Furthermore,
712 seals can never be removed, only added.
713 F_ADD_SEALS (int; since Linux 3.17)
715 Add the seals given in the bit-mask argument arg to the set of
716 seals of the inode referred to by the file descriptor fd. Seals
717 cannot be removed again. Once this call succeeds, the seals are
718 enforced by the kernel immediately. If the current set of seals
719 includes F_SEAL_SEAL (see below), then this call will be re‐
720 jected with EPERM. Adding a seal that is already set is a no-
721 op, in case F_SEAL_SEAL is not set already. In order to place a
722 seal, the file descriptor fd must be writable.
723 F_GET_SEALS (void; since Linux 3.17)
725 Return (as the function result) the current set of seals of the
726 inode referred to by fd. If no seals are set, 0 is returned.
727 If the file does not support sealing, -1 is returned and errno
728 is set to EINVAL.
729 The following seals are available:
731 F_SEAL_SEAL
733 If this seal is set, any further call to fcntl() with
734 F_ADD_SEALS fails with the error EPERM. Therefore, this seal
735 prevents any modifications to the set of seals itself. If the
736 initial set of seals of a file includes F_SEAL_SEAL, then this
737 effectively causes the set of seals to be constant and locked.
738 F_SEAL_SHRINK
740 If this seal is set, the file in question cannot be reduced in
741 size. This affects open(2) with the O_TRUNC flag as well as
742 truncate(2) and ftruncate(2). Those calls fail with EPERM if
743 you try to shrink the file in question. Increasing the file
744 size is still possible.
745 F_SEAL_GROW
747 If this seal is set, the size of the file in question cannot be
748 increased. This affects write(2) beyond the end of the file,
749 truncate(2), ftruncate(2), and fallocate(2). These calls fail
750 with EPERM if you use them to increase the file size. If you
751 keep the size or shrink it, those calls still work as expected.
752 F_SEAL_WRITE
754 If this seal is set, you cannot modify the contents of the file.
755 Note that shrinking or growing the size of the file is still
756 possible and allowed. Thus, this seal is normally used in com‐
757 bination with one of the other seals. This seal affects
758 write(2) and fallocate(2) (only in combination with the FAL‐
759 LOC_FL_PUNCH_HOLE flag). Those calls fail with EPERM if this
760 seal is set. Furthermore, trying to create new shared, writable
761 memory-mappings via mmap(2) will also fail with EPERM.
762 Using the F_ADD_SEALS operation to set the F_SEAL_WRITE seal
764 fails with EBUSY if any writable, shared mapping exists. Such
765 mappings must be unmapped before you can add this seal. Fur‐
766 thermore, if there are any asynchronous I/O operations (io_sub‐
767 mit(2)) pending on the file, all outstanding writes will be dis‐
768 carded.
769 F_SEAL_FUTURE_WRITE (since Linux 5.1)
771 The effect of this seal is similar to F_SEAL_WRITE, but the con‐
772 tents of the file can still be modified via shared writable map‐
773 pings that were created prior to the seal being set. Any at‐
774 tempt to create a new writable mapping on the file via mmap(2)
775 will fail with EPERM. Likewise, an attempt to write to the file
776 via write(2) will fail with EPERM.
777 Using this seal, one process can create a memory buffer that it
779 can continue to modify while sharing that buffer on a "read-
780 only" basis with other processes.
781 File read/write hints
783 Write lifetime hints can be used to inform the kernel about the rela‐
784 tive expected lifetime of writes on a given inode or via a particular
785 open file description. (See open(2) for an explanation of open file
786 descriptions.) In this context, the term "write lifetime" means the
787 expected time the data will live on media, before being overwritten or
788 erased.
789 An application may use the different hint values specified below to
791 separate writes into different write classes, so that multiple users or
792 applications running on a single storage back-end can aggregate their
793 I/O patterns in a consistent manner. However, there are no functional
794 semantics implied by these flags, and different I/O classes can use the
795 write lifetime hints in arbitrary ways, so long as the hints are used
796 consistently.
797 The following operations can be applied to the file descriptor, fd:
799 F_GET_RW_HINT (uint64_t *; since Linux 4.13)
801 Returns the value of the read/write hint associated with the un‐
802 derlying inode referred to by fd.
803 F_SET_RW_HINT (uint64_t *; since Linux 4.13)
805 Sets the read/write hint value associated with the underlying
806 inode referred to by fd. This hint persists until either it is
807 explicitly modified or the underlying filesystem is unmounted.
808 F_GET_FILE_RW_HINT (uint64_t *; since Linux 4.13)
810 Returns the value of the read/write hint associated with the
811 open file description referred to by fd.
812 F_SET_FILE_RW_HINT (uint64_t *; since Linux 4.13)
814 Sets the read/write hint value associated with the open file de‐
815 scription referred to by fd.
816 If an open file description has not been assigned a read/write hint,
818 then it shall use the value assigned to the inode, if any.
819 The following read/write hints are valid since Linux 4.13:
821 RWH_WRITE_LIFE_NOT_SET
823 No specific hint has been set. This is the default value.
824 RWH_WRITE_LIFE_NONE
826 No specific write lifetime is associated with this file or in‐
827 ode.
828 RWH_WRITE_LIFE_SHORT
830 Data written to this inode or via this open file description is
831 expected to have a short lifetime.
832 RWH_WRITE_LIFE_MEDIUM
834 Data written to this inode or via this open file description is
835 expected to have a lifetime longer than data written with
836 RWH_WRITE_LIFE_SHORT.
837 RWH_WRITE_LIFE_LONG
839 Data written to this inode or via this open file description is
840 expected to have a lifetime longer than data written with
841 RWH_WRITE_LIFE_MEDIUM.
842 RWH_WRITE_LIFE_EXTREME
844 Data written to this inode or via this open file description is
845 expected to have a lifetime longer than data written with
846 RWH_WRITE_LIFE_LONG.
847 All the write-specific hints are relative to each other, and no indi‐
849 vidual absolute meaning should be attributed to them.
850
852 For a successful call, the return value depends on the operation:
853 F_DUPFD
855 The new file descriptor.
856 F_GETFD
858 Value of file descriptor flags.
859 F_GETFL
861 Value of file status flags.
862 F_GETLEASE
864 Type of lease held on file descriptor.
865 F_GETOWN
867 Value of file descriptor owner.
868 F_GETSIG
870 Value of signal sent when read or write becomes possible, or
871 zero for traditional SIGIO behavior.
872 F_GETPIPE_SZ, F_SETPIPE_SZ
874 The pipe capacity.
875 F_GET_SEALS
877 A bit mask identifying the seals that have been set for the in‐
878 ode referred to by fd.
879 All other commands
881 Zero.
882 On error, -1 is returned, and errno is set appropriately.
884
886 EACCES or EAGAIN
887 Operation is prohibited by locks held by other processes.
888 EAGAIN The operation is prohibited because the file has been memory-
890 mapped by another process.
891 EBADF fd is not an open file descriptor
893 EBADF cmd is F_SETLK or F_SETLKW and the file descriptor open mode
895 doesn't match with the type of lock requested.
896 EBUSY cmd is F_SETPIPE_SZ and the new pipe capacity specified in arg
898 is smaller than the amount of buffer space currently used to
899 store data in the pipe.
900 EBUSY cmd is F_ADD_SEALS, arg includes F_SEAL_WRITE, and there exists
902 a writable, shared mapping on the file referred to by fd.
903 EDEADLK
905 It was detected that the specified F_SETLKW command would cause
906 a deadlock.
907 EFAULT lock is outside your accessible address space.
909 EINTR cmd is F_SETLKW or F_OFD_SETLKW and the operation was inter‐
911 rupted by a signal; see signal(7).
912 EINTR cmd is F_GETLK, F_SETLK, F_OFD_GETLK, or F_OFD_SETLK, and the
914 operation was interrupted by a signal before the lock was
915 checked or acquired. Most likely when locking a remote file
916 (e.g., locking over NFS), but can sometimes happen locally.
917 EINVAL The value specified in cmd is not recognized by this kernel.
919 EINVAL cmd is F_ADD_SEALS and arg includes an unrecognized sealing bit.
921 EINVAL cmd is F_ADD_SEALS or F_GET_SEALS and the filesystem containing
923 the inode referred to by fd does not support sealing.
924 EINVAL cmd is F_DUPFD and arg is negative or is greater than the maxi‐
926 mum allowable value (see the discussion of RLIMIT_NOFILE in
927 getrlimit(2)).
928 EINVAL cmd is F_SETSIG and arg is not an allowable signal number.
930 EINVAL cmd is F_OFD_SETLK, F_OFD_SETLKW, or F_OFD_GETLK, and l_pid was
932 not specified as zero.
933 EMFILE cmd is F_DUPFD and the per-process limit on the number of open
935 file descriptors has been reached.
936 ENOLCK Too many segment locks open, lock table is full, or a remote
938 locking protocol failed (e.g., locking over NFS).
939 ENOTDIR
941 F_NOTIFY was specified in cmd, but fd does not refer to a direc‐
942 tory.
943 EPERM cmd is F_SETPIPE_SZ and the soft or hard user pipe limit has
945 been reached; see pipe(7).
946 EPERM Attempted to clear the O_APPEND flag on a file that has the ap‐
948 pend-only attribute set.
949 EPERM cmd was F_ADD_SEALS, but fd was not open for writing or the cur‐
951 rent set of seals on the file already includes F_SEAL_SEAL.
952
954 SVr4, 4.3BSD, POSIX.1-2001. Only the operations F_DUPFD, F_GETFD,
955 F_SETFD, F_GETFL, F_SETFL, F_GETLK, F_SETLK, and F_SETLKW are specified
956 in POSIX.1-2001.
957 F_GETOWN and F_SETOWN are specified in POSIX.1-2001. (To get their
959 definitions, define either _XOPEN_SOURCE with the value 500 or greater,
960 or _POSIX_C_SOURCE with the value 200809L or greater.)
961 F_DUPFD_CLOEXEC is specified in POSIX.1-2008. (To get this definition,
963 define _POSIX_C_SOURCE with the value 200809L or greater, or
964 _XOPEN_SOURCE with the value 700 or greater.)
965 F_GETOWN_EX, F_SETOWN_EX, F_SETPIPE_SZ, F_GETPIPE_SZ, F_GETSIG, F_SET‐
967 SIG, F_NOTIFY, F_GETLEASE, and F_SETLEASE are Linux-specific. (Define
968 the _GNU_SOURCE macro to obtain these definitions.)
969 F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK are Linux-specific (and one
971 must define _GNU_SOURCE to obtain their definitions), but work is being
972 done to have them included in the next version of POSIX.1.
973 F_ADD_SEALS and F_GET_SEALS are Linux-specific.
975
977 The errors returned by dup2(2) are different from those returned by
978 F_DUPFD.
979 File locking
981 The original Linux fcntl() system call was not designed to handle large
982 file offsets (in the flock structure). Consequently, an fcntl64() sys‐
983 tem call was added in Linux 2.4. The newer system call employs a dif‐
984 ferent structure for file locking, flock64, and corresponding commands,
985 F_GETLK64, F_SETLK64, and F_SETLKW64. However, these details can be
986 ignored by applications using glibc, whose fcntl() wrapper function
987 transparently employs the more recent system call where it is avail‐
988 able.
989 Record locks
991 Since kernel 2.0, there is no interaction between the types of lock
992 placed by flock(2) and fcntl().
993 Several systems have more fields in struct flock such as, for example,
995 l_sysid (to identify the machine where the lock is held). Clearly,
996 l_pid alone is not going to be very useful if the process holding the
997 lock may live on a different machine; on Linux, while present on some
998 architectures (such as MIPS32), this field is not used.
999 The original Linux fcntl() system call was not designed to handle large
1001 file offsets (in the flock structure). Consequently, an fcntl64() sys‐
1002 tem call was added in Linux 2.4. The newer system call employs a dif‐
1003 ferent structure for file locking, flock64, and corresponding commands,
1004 F_GETLK64, F_SETLK64, and F_SETLKW64. However, these details can be
1005 ignored by applications using glibc, whose fcntl() wrapper function
1006 transparently employs the more recent system call where it is avail‐
1007 able.
1008 Record locking and NFS
1010 Before Linux 3.12, if an NFSv4 client loses contact with the server for
1011 a period of time (defined as more than 90 seconds with no communica‐
1012 tion), it might lose and regain a lock without ever being aware of the
1013 fact. (The period of time after which contact is assumed lost is known
1014 as the NFSv4 leasetime. On a Linux NFS server, this can be determined
1015 by looking at /proc/fs/nfsd/nfsv4leasetime, which expresses the period
1016 in seconds. The default value for this file is 90.) This scenario po‐
1017 tentially risks data corruption, since another process might acquire a
1018 lock in the intervening period and perform file I/O.
1019 Since Linux 3.12, if an NFSv4 client loses contact with the server, any
1021 I/O to the file by a process which "thinks" it holds a lock will fail
1022 until that process closes and reopens the file. A kernel parameter,
1023 nfs.recover_lost_locks, can be set to 1 to obtain the pre-3.12 behav‐
1024 ior, whereby the client will attempt to recover lost locks when contact
1025 is reestablished with the server. Because of the attendant risk of
1026 data corruption, this parameter defaults to 0 (disabled).
1027
1029 F_SETFL
1030 It is not possible to use F_SETFL to change the state of the O_DSYNC
1031 and O_SYNC flags. Attempts to change the state of these flags are
1032 silently ignored.
1033 F_GETOWN
1035 A limitation of the Linux system call conventions on some architectures
1036 (notably i386) means that if a (negative) process group ID to be re‐
1037 turned by F_GETOWN falls in the range -1 to -4095, then the return
1038 value is wrongly interpreted by glibc as an error in the system call;
1039 that is, the return value of fcntl() will be -1, and errno will contain
1040 the (positive) process group ID. The Linux-specific F_GETOWN_EX opera‐
1041 tion avoids this problem. Since glibc version 2.11, glibc makes the
1042 kernel F_GETOWN problem invisible by implementing F_GETOWN using
1043 F_GETOWN_EX.
1044 F_SETOWN
1046 In Linux 2.4 and earlier, there is bug that can occur when an unprivi‐
1047 leged process uses F_SETOWN to specify the owner of a socket file de‐
1048 scriptor as a process (group) other than the caller. In this case, fc‐
1049 ntl() can return -1 with errno set to EPERM, even when the owner
1050 process (group) is one that the caller has permission to send signals
1051 to. Despite this error return, the file descriptor owner is set, and
1052 signals will be sent to the owner.
1053 Deadlock detection
1055 The deadlock-detection algorithm employed by the kernel when dealing
1056 with F_SETLKW requests can yield both false negatives (failures to de‐
1057 tect deadlocks, leaving a set of deadlocked processes blocked indefi‐
1058 nitely) and false positives (EDEADLK errors when there is no deadlock).
1059 For example, the kernel limits the lock depth of its dependency search
1060 to 10 steps, meaning that circular deadlock chains that exceed that
1061 size will not be detected. In addition, the kernel may falsely indi‐
1062 cate a deadlock when two or more processes created using the clone(2)
1063 CLONE_FILES flag place locks that appear (to the kernel) to conflict.
1064 Mandatory locking
1066 The Linux implementation of mandatory locking is subject to race condi‐
1067 tions which render it unreliable: a write(2) call that overlaps with a
1068 lock may modify data after the mandatory lock is acquired; a read(2)
1069 call that overlaps with a lock may detect changes to data that were
1070 made only after a write lock was acquired. Similar races exist between
1071 mandatory locks and mmap(2). It is therefore inadvisable to rely on
1072 mandatory locking.
1073
1075 dup2(2), flock(2), open(2), socket(2), lockf(3), capabilities(7), fea‐
1076 ture_test_macros(7), lslocks(8)
1077 locks.txt, mandatory-locking.txt, and dnotify.txt in the Linux kernel
1079 source directory Documentation/filesystems/ (on older kernels, these
1080 files are directly under the Documentation/ directory, and mandatory-
1081 locking.txt is called mandatory.txt)
1082
1084 This page is part of release 5.10 of the Linux man-pages project. A
1085 description of the project, information about reporting bugs, and the
1086 latest version of this page, can be found at
1087 https://www.kernel.org/doc/man-pages/.
1088Linux 2020-12-21 FCNTL(2)