xref: /freebsd/lib/libsys/fcntl.2 (revision cc145f5d8425f19fbe07110fed89ecde8d70ce78)
1.\" Copyright (c) 1983, 1993
2.\"	The Regents of the University of California.  All rights reserved.
3.\"
4.\" Redistribution and use in source and binary forms, with or without
5.\" modification, are permitted provided that the following conditions
6.\" are met:
7.\" 1. Redistributions of source code must retain the above copyright
8.\"    notice, this list of conditions and the following disclaimer.
9.\" 2. Redistributions in binary form must reproduce the above copyright
10.\"    notice, this list of conditions and the following disclaimer in the
11.\"    documentation and/or other materials provided with the distribution.
12.\" 3. Neither the name of the University nor the names of its contributors
13.\"    may be used to endorse or promote products derived from this software
14.\"    without specific prior written permission.
15.\"
16.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
17.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
20.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26.\" SUCH DAMAGE.
27.\"
28.Dd June 24, 2025
29.Dt FCNTL 2
30.Os
31.Sh NAME
32.Nm fcntl
33.Nd file control
34.Sh LIBRARY
35.Lb libc
36.Sh SYNOPSIS
37.In fcntl.h
38.Ft int
39.Fn fcntl "int fd" "int cmd" "..."
40.Sh DESCRIPTION
41The
42.Fn fcntl
43system call provides for control over descriptors.
44The argument
45.Fa fd
46is a descriptor to be operated on by
47.Fa cmd
48as described below.
49Depending on the value of
50.Fa cmd ,
51.Fn fcntl
52can take an additional third argument
53.Fa arg .
54Unless otherwise noted below for a specific operation,
55.Fa arg
56has type
57.Vt int .
58.Bl -tag -width F_DUP2FD_CLOEXEC
59.It Dv F_DUPFD
60Return a new descriptor as follows:
61.Pp
62.Bl -bullet -compact -offset 4n
63.It
64Lowest numbered available descriptor greater than or equal to
65.Fa arg .
66.It
67Same object references as the original descriptor.
68.It
69New descriptor shares the same file offset if the object
70was a file.
71.It
72Same access mode (read, write or read/write).
73.It
74Same file status flags (i.e., both file descriptors
75share the same file status flags).
76.It
77The close-on-exec flag
78.Dv FD_CLOEXEC
79associated with the new file descriptor is cleared, so the file descriptor is
80to remain open across
81.Xr execve 2
82system calls.
83.It
84The fork-on-exec flag
85.Dv FD_CLOFORK
86associated with the new file descriptor is cleared, so the file descriptor is
87to remain open across
88.Xr fork 2
89system calls.
90.It
91The
92.Dv FD_RESOLVE_BENEATH
93flag, described below, will be set if it was set on the original
94descriptor.
95.El
96.It Dv F_DUPFD_CLOEXEC
97Like
98.Dv F_DUPFD ,
99but the
100.Dv FD_CLOEXEC
101flag associated with the new file descriptor is set, so the file descriptor
102is closed when
103.Xr execve 2
104system call executes.
105.It Dv F_DUPFD_CLOFORK
106Like
107.Dv F_DUPFD ,
108but the
109.Dv FD_CLOFORK
110flag associated with the new file descriptor is set, so the file descriptor
111is closed when
112.Xr fork 2
113system call executes.
114.It Dv F_DUP2FD
115It is functionally equivalent to
116.Bd -literal -offset indent
117dup2(fd, arg)
118.Ed
119.It Dv F_DUP2FD_CLOEXEC
120Like
121.Dv F_DUP2FD ,
122but the
123.Dv FD_CLOEXEC
124flag associated with the new file descriptor is set.
125.Pp
126The
127.Dv F_DUP2FD
128and
129.Dv F_DUP2FD_CLOEXEC
130constants are not portable, so they should not be used if
131portability is needed.
132Use
133.Fn dup2
134instead of
135.Dv F_DUP2FD .
136.It Dv F_DUP3FD
137Used to implement the
138.Fn dup3
139call.
140Do not use it.
141.It Dv F_GETFD
142Get the flags associated with the file descriptor
143.Fa fd .
144The following flags are defined:
145.Bl -tag -width FD_RESOLVE_BENEATH
146.It Dv FD_CLOEXEC
147The file will be closed upon execution of
148.Fn exec
149.Fa ( arg
150is ignored).
151Otherwise, the file descriptor will remain open.
152.It Dv FD_CLOFORK
153The file will be closed upon execution of the
154.Fn fork
155family of system calls.
156.It Dv FD_RESOLVE_BENEATH
157All path name lookups relative to that file descriptor
158will behave as if the lookup had
159.Dv O_RESOLVE_BENEATH
160or
161.Dv AT_RESOLVE_BENEATH
162semantics.
163It is not permitted to call
164.Xr fchdir 2
165or
166.Xr fchroot 2
167on such a file descriptor.
168The
169.Dv FD_RESOLVE_BENEATH
170flag is sticky, meaning that it is preserved by
171.Xr dup 2
172and similar operations, and opening a directory with
173.Xr openat 2
174where the directory descriptor has the flag set causes the new directory
175descriptor to also have the flag set.
176.El
177.It Dv F_SETFD
178Set flags associated with
179.Fa fd
180to
181.Fa arg .
182The available flags are
183.Dv FD_CLOEXEC ,
184.Dv FD_CLOFORK
185and
186.Dv FD_RESOLVE_BENEATH .
187The
188.Dv FD_RESOLVE_BENEATH
189flag cannot be cleared once set.
190.It Dv F_GETFL
191Get descriptor status flags, as described below
192.Fa ( arg
193is ignored).
194.It Dv F_SETFL
195Set descriptor status flags to
196.Fa arg .
197.It Dv F_GETOWN
198Get the process ID or process group
199currently receiving
200.Dv SIGIO
201and
202.Dv SIGURG
203signals; process groups are returned
204as negative values
205.Fa ( arg
206is ignored).
207.It Dv F_SETOWN
208Set the process or process group
209to receive
210.Dv SIGIO
211and
212.Dv SIGURG
213signals;
214process groups are specified by supplying
215.Fa arg
216as negative, otherwise
217.Fa arg
218is interpreted as a process ID.
219.It Dv F_READAHEAD
220Set or clear the read ahead amount for sequential access to the third
221argument,
222.Fa arg ,
223which is rounded up to the nearest block size.
224A zero value in
225.Fa arg
226turns off read ahead, a negative value restores the system default.
227.It Dv F_RDAHEAD
228Equivalent to Darwin counterpart which sets read ahead amount of 128KB
229when the third argument,
230.Fa arg
231is non-zero.
232A zero value in
233.Fa arg
234turns off read ahead.
235.It Dv F_ADD_SEALS
236Add seals to the file as described below, if the underlying filesystem supports
237seals.
238.It Dv F_GET_SEALS
239Get seals associated with the file, if the underlying filesystem supports seals.
240.It Dv F_ISUNIONSTACK
241Check if the vnode is part of a union stack (either the "union" flag from
242.Xr mount 2
243or unionfs).
244This is a hack not intended to be used outside of libc.
245.It Dv F_KINFO
246Fills a
247.Vt struct kinfo_file
248for the file referenced by the specified file descriptor.
249The
250.Fa arg
251argument should point to the storage for
252.Vt struct kinfo_file .
253The
254.Va kf_structsize
255member of the passed structure must be initialized with the sizeof of
256.Vt struct kinfo_file ,
257to allow for the interface versioning and evolution.
258.El
259.Pp
260The flags for the
261.Dv F_GETFL
262and
263.Dv F_SETFL
264commands are as follows:
265.Bl -tag -width O_NONBLOCKX
266.It Dv O_NONBLOCK
267Non-blocking I/O; if no data is available to a
268.Xr read 2
269system call, or if a
270.Xr write 2
271operation would block,
272the read or write call returns -1 with the error
273.Er EAGAIN .
274.It Dv O_APPEND
275Force each write to append at the end of file;
276corresponds to the
277.Dv O_APPEND
278flag of
279.Xr open 2 .
280.It Dv O_DIRECT
281Minimize or eliminate the cache effects of reading and writing.
282The system
283will attempt to avoid caching the data you read or write.
284If it cannot
285avoid caching the data, it will minimize the impact the data has on the cache.
286Use of this flag can drastically reduce performance if not used with care.
287.It Dv O_ASYNC
288Enable the
289.Dv SIGIO
290signal to be sent to the process group
291when I/O is possible, e.g.,
292upon availability of data to be read.
293.It Dv O_SYNC
294Enable synchronous writes.
295Corresponds to the
296.Dv O_SYNC
297flag of
298.Xr open 2 .
299.Dv O_FSYNC
300is an historical synonym for
301.Dv O_SYNC .
302.It Dv O_DSYNC
303Enable synchronous data writes.
304Corresponds to the
305.Dv O_DSYNC
306flag of
307.Xr open 2 .
308.El
309.Pp
310The seals that may be applied with
311.Dv F_ADD_SEALS
312are as follows:
313.Bl -tag -width F_SEAL_SHRINK
314.It Dv F_SEAL_SEAL
315Prevent any further seals from being applied to the file.
316.It Dv F_SEAL_SHRINK
317Prevent the file from being shrunk with
318.Xr ftruncate 2 .
319.It Dv F_SEAL_GROW
320Prevent the file from being enlarged with
321.Xr ftruncate 2 .
322.It Dv F_SEAL_WRITE
323Prevent any further
324.Xr write 2
325calls to the file.
326Any writes in progress will finish before
327.Fn fcntl
328returns.
329If any writeable mappings exist, F_ADD_SEALS will fail and return
330.Dv EBUSY .
331.El
332.Pp
333Seals are on a per-inode basis and require support by the underlying filesystem.
334If the underlying filesystem does not support seals,
335.Dv F_ADD_SEALS
336and
337.Dv F_GET_SEALS
338will fail and return
339.Dv EINVAL .
340.Pp
341Several operations are available for doing advisory file locking;
342they all operate on the following structure:
343.Bd -literal
344struct flock {
345	off_t	l_start;	/* starting offset */
346	off_t	l_len;		/* len = 0 means until end of file */
347	pid_t	l_pid;		/* lock owner */
348	short	l_type;		/* lock type: read/write, etc. */
349	short	l_whence;	/* type of l_start */
350	int	l_sysid;	/* remote system id or zero for local */
351};
352.Ed
353These advisory file locking operations take a pointer to
354.Vt struct flock
355as the third argument
356.Fa arg .
357The commands available for advisory record locking are as follows:
358.Bl -tag -width F_SETLKWX
359.It Dv F_GETLK
360Get the first lock that blocks the lock description pointed to by the
361third argument,
362.Fa arg ,
363taken as a pointer to a
364.Fa "struct flock"
365(see above).
366The information retrieved overwrites the information passed to
367.Fn fcntl
368in the
369.Fa flock
370structure.
371If no lock is found that would prevent this lock from being created,
372the structure is left unchanged by this system call except for the
373lock type which is set to
374.Dv F_UNLCK .
375.It Dv F_SETLK
376Set or clear a file segment lock according to the lock description
377pointed to by the third argument,
378.Fa arg ,
379taken as a pointer to a
380.Fa "struct flock"
381(see above).
382.Dv F_SETLK
383is used to establish shared (or read) locks
384.Pq Dv F_RDLCK
385or exclusive (or write) locks,
386.Pq Dv F_WRLCK ,
387as well as remove either type of lock
388.Pq Dv F_UNLCK .
389If a shared or exclusive lock cannot be set,
390.Fn fcntl
391returns immediately with
392.Er EAGAIN .
393.It Dv F_SETLKW
394This command is the same as
395.Dv F_SETLK
396except that if a shared or exclusive lock is blocked by other locks,
397the process waits until the request can be satisfied.
398If a signal that is to be caught is received while
399.Fn fcntl
400is waiting for a region, the
401.Fn fcntl
402will be interrupted if the signal handler has not specified the
403.Dv SA_RESTART
404(see
405.Xr sigaction 2 ) .
406.El
407.Pp
408When a shared lock has been set on a segment of a file,
409other processes can set shared locks on that segment
410or a portion of it.
411A shared lock prevents any other process from setting an exclusive
412lock on any portion of the protected area.
413A request for a shared lock fails if the file descriptor was not
414opened with read access.
415.Pp
416An exclusive lock prevents any other process from setting a shared lock or
417an exclusive lock on any portion of the protected area.
418A request for an exclusive lock fails if the file was not
419opened with write access.
420.Pp
421The value of
422.Fa l_whence
423is
424.Dv SEEK_SET ,
425.Dv SEEK_CUR ,
426or
427.Dv SEEK_END
428to indicate that the relative offset,
429.Fa l_start
430bytes, will be measured from the start of the file,
431current position, or end of the file, respectively.
432The value of
433.Fa l_len
434is the number of consecutive bytes to be locked.
435If
436.Fa l_len
437is negative,
438.Fa l_start
439means end edge of the region.
440The
441.Fa l_pid
442and
443.Fa l_sysid
444fields are only used with
445.Dv F_GETLK
446to return the process ID of the process holding a blocking lock and
447the system ID of the system that owns that process.
448Locks created by the local system will have a system ID of zero.
449After a successful
450.Dv F_GETLK
451request, the value of
452.Fa l_whence
453is
454.Dv SEEK_SET .
455.Pp
456Locks may start and extend beyond the current end of a file,
457but may not start or extend before the beginning of the file.
458A lock is set to extend to the largest possible value of the
459file offset for that file if
460.Fa l_len
461is set to zero.
462If
463.Fa l_whence
464and
465.Fa l_start
466point to the beginning of the file, and
467.Fa l_len
468is zero, the entire file is locked.
469If an application wishes only to do entire file locking, the
470.Xr flock 2
471system call is much more efficient.
472.Pp
473There is at most one type of lock set for each byte in the file.
474Before a successful return from an
475.Dv F_SETLK
476or an
477.Dv F_SETLKW
478request when the calling process has previously existing locks
479on bytes in the region specified by the request,
480the previous lock type for each byte in the specified
481region is replaced by the new lock type.
482As specified above under the descriptions
483of shared locks and exclusive locks, an
484.Dv F_SETLK
485or an
486.Dv F_SETLKW
487request fails or blocks respectively when another process has existing
488locks on bytes in the specified region and the type of any of those
489locks conflicts with the type specified in the request.
490.Pp
491The queuing for
492.Dv F_SETLKW
493requests on local files is fair;
494that is, while the thread is blocked,
495subsequent requests conflicting with its requests will not be granted,
496even if these requests do not conflict with existing locks.
497.Pp
498This interface follows the completely stupid semantics of System V and
499.St -p1003.1-88
500that require that all locks associated with a file for a given process are
501removed when
502.Em any
503file descriptor for that file is closed by that process.
504This semantic means that applications must be aware of any files that
505a subroutine library may access.
506For example if an application for updating the password file locks the
507password file database while making the update, and then calls
508.Xr getpwnam 3
509to retrieve a record,
510the lock will be lost because
511.Xr getpwnam 3
512opens, reads, and closes the password database.
513The database close will release all locks that the process has
514associated with the database, even if the library routine never
515requested a lock on the database.
516Another minor semantic problem with this interface is that
517locks are not inherited by a child process created using the
518.Xr fork 2
519system call.
520The
521.Xr flock 2
522interface has much more rational last close semantics and
523allows locks to be inherited by child processes.
524The
525.Xr flock 2
526system call is recommended for applications that want to ensure the integrity
527of their locks when using library routines or wish to pass locks
528to their children.
529.Pp
530The
531.Fn fcntl ,
532.Xr flock 2 ,
533and
534.Xr lockf 3
535locks are compatible.
536Processes using different locking interfaces can cooperate
537over the same file safely.
538However, only one of such interfaces should be used within
539the same process.
540If a file is locked by a process through
541.Xr flock 2 ,
542any record within the file will be seen as locked
543from the viewpoint of another process using
544.Fn fcntl
545or
546.Xr lockf 3 ,
547and vice versa.
548Note that
549.Fn fcntl F_GETLK
550returns \-1 in
551.Fa l_pid
552if the process holding a blocking lock previously locked the
553file descriptor by
554.Xr flock 2 .
555.Pp
556All locks associated with a file for a given process are
557removed when the process terminates.
558.Pp
559All locks obtained before a call to
560.Xr execve 2
561remain in effect until the new program releases them.
562If the new program does not know about the locks, they will not be
563released until the program exits.
564.Pp
565A potential for deadlock occurs if a process controlling a locked region
566is put to sleep by attempting to lock the locked region of another process.
567This implementation detects that sleeping until a locked region is unlocked
568would cause a deadlock and fails with an
569.Er EDEADLK
570error.
571.Sh RETURN VALUES
572Upon successful completion, the value returned depends on
573.Fa cmd
574as follows:
575.Bl -tag -width F_GETOWNX -offset indent
576.It Dv F_DUPFD
577A new file descriptor.
578.It Dv F_DUP2FD
579A file descriptor equal to
580.Fa arg .
581.It Dv F_GETFD
582Value of flags.
583.It Dv F_GETFL
584Value of flags.
585.It Dv F_GETOWN
586Value of file descriptor owner.
587.It other
588Value other than -1.
589.El
590.Pp
591Otherwise, a value of -1 is returned and
592.Va errno
593is set to indicate the error.
594.Sh ERRORS
595The
596.Fn fcntl
597system call will fail if:
598.Bl -tag -width Er
599.It Bq Er EAGAIN
600The argument
601.Fa cmd
602is
603.Dv F_SETLK ,
604the type of lock
605.Pq Fa l_type
606is a shared lock
607.Pq Dv F_RDLCK
608or exclusive lock
609.Pq Dv F_WRLCK ,
610and the segment of a file to be locked is already
611exclusive-locked by another process;
612or the type is an exclusive lock and some portion of the
613segment of a file to be locked is already shared-locked or
614exclusive-locked by another process.
615.It Bq Er EBADF
616The
617.Fa fd
618argument
619is not a valid open file descriptor.
620.Pp
621The argument
622.Fa cmd
623is
624.Dv F_DUP2FD ,
625and
626.Fa arg
627is not a valid file descriptor.
628.Pp
629The argument
630.Fa cmd
631is
632.Dv F_SETLK
633or
634.Dv F_SETLKW ,
635the type of lock
636.Pq Fa l_type
637is a shared lock
638.Pq Dv F_RDLCK ,
639and
640.Fa fd
641is not a valid file descriptor open for reading.
642.Pp
643The argument
644.Fa cmd
645is
646.Dv F_SETLK
647or
648.Dv F_SETLKW ,
649the type of lock
650.Pq Fa l_type
651is an exclusive lock
652.Pq Dv F_WRLCK ,
653and
654.Fa fd
655is not a valid file descriptor open for writing.
656.It Bq Er EBUSY
657The argument
658.Fa cmd
659is
660.Dv F_ADD_SEALS ,
661attempting to set
662.Dv F_SEAL_WRITE ,
663and writeable mappings of the file exist.
664.It Bq Er EDEADLK
665The argument
666.Fa cmd
667is
668.Dv F_SETLKW ,
669and a deadlock condition was detected.
670.It Bq Er EINTR
671The argument
672.Fa cmd
673is
674.Dv F_SETLKW ,
675and the system call was interrupted by a signal.
676.It Bq Er EINVAL
677The
678.Fa cmd
679argument
680is
681.Dv F_DUPFD
682and
683.Fa arg
684is negative or greater than the maximum allowable number
685(see
686.Xr getdtablesize 2 ) .
687.Pp
688The argument
689.Fa cmd
690is
691.Dv F_GETLK ,
692.Dv F_SETLK
693or
694.Dv F_SETLKW
695and the data to which
696.Fa arg
697points is not valid.
698.Pp
699The argument
700.Fa cmd
701is
702.Dv F_ADD_SEALS
703or
704.Dv F_GET_SEALS ,
705and the underlying filesystem does not support sealing.
706.Pp
707The argument
708.Fa cmd
709is invalid.
710.It Bq Er EMFILE
711The argument
712.Fa cmd
713is
714.Dv F_DUPFD
715and the maximum number of file descriptors permitted for the
716process are already in use,
717or no file descriptors greater than or equal to
718.Fa arg
719are available.
720.It Bq Er ENOTTY
721The
722.Fa fd
723argument is not a valid file descriptor for the requested operation.
724This may be the case if
725.Fa fd
726is a device node, or a descriptor returned by
727.Xr kqueue 2 .
728.It Bq Er ENOLCK
729The argument
730.Fa cmd
731is
732.Dv F_SETLK
733or
734.Dv F_SETLKW ,
735and satisfying the lock or unlock request would result in the
736number of locked regions in the system exceeding a system-imposed limit.
737.It Bq Er EOPNOTSUPP
738The argument
739.Fa cmd
740is
741.Dv F_GETLK ,
742.Dv F_SETLK
743or
744.Dv F_SETLKW
745and
746.Fa fd
747refers to a file for which locking is not supported.
748.It Bq Er EOVERFLOW
749The argument
750.Fa cmd
751is
752.Dv F_GETLK ,
753.Dv F_SETLK
754or
755.Dv F_SETLKW
756and an
757.Fa off_t
758calculation overflowed.
759.It Bq Er EPERM
760The
761.Fa cmd
762argument
763is
764.Dv F_SETOWN
765and
766the process ID or process group given as an argument is in a
767different session than the caller.
768.Pp
769The
770.Fa cmd
771argument
772is
773.Dv F_ADD_SEALS
774and the
775.Dv F_SEAL_SEAL
776seal has already been set.
777.It Bq Er ESRCH
778The
779.Fa cmd
780argument
781is
782.Dv F_SETOWN
783and
784the process ID given as argument is not in use.
785.El
786.Pp
787In addition, if
788.Fa fd
789refers to a descriptor open on a terminal device (as opposed to a
790descriptor open on a socket), a
791.Fa cmd
792of
793.Dv F_SETOWN
794can fail for the same reasons as in
795.Xr tcsetpgrp 3 ,
796and a
797.Fa cmd
798of
799.Dv F_GETOWN
800for the reasons as stated in
801.Xr tcgetpgrp 3 .
802.Sh SEE ALSO
803.Xr close 2 ,
804.Xr dup2 2 ,
805.Xr execve 2 ,
806.Xr flock 2 ,
807.Xr getdtablesize 2 ,
808.Xr open 2 ,
809.Xr sigaction 2 ,
810.Xr lockf 3 ,
811.Xr tcgetpgrp 3 ,
812.Xr tcsetpgrp 3
813.Sh STANDARDS
814The
815.Dv F_DUP2FD
816and
817.Dv F_DUP3FD
818constants are not portable.
819They are provided for compatibility with AIX and Solaris.
820.Pp
821Per
822.St -susv4 ,
823a call with
824.Dv F_SETLKW
825should fail with
826.Bq Er EINTR
827after any caught signal
828and should continue waiting during thread suspension such as a stop signal.
829However, in this implementation a call with
830.Dv F_SETLKW
831is restarted after catching a signal with a
832.Dv SA_RESTART
833handler or a thread suspension such as a stop signal.
834.Sh HISTORY
835The
836.Fn fcntl
837system call appeared in
838.Bx 4.2 .
839.Pp
840The
841.Dv F_DUP2FD
842constant first appeared in
843.Fx 7.1 .
844.Pp
845The
846.Dv F_DUPFD_CLOFORK
847and
848.Dv F_DUP3FD
849flags appeared in
850.Fx 15.0 .
851