2 .\" This manpage is Copyright (C) 1992 Drew Eckhardt;
3 .\" and Copyright (C) 1993 Michael Haardt, Ian Jackson;
4 .\" and Copyright (C) 1998 Jamie Lokier;
5 .\" and Copyright (C) 2002-2010 Michael Kerrisk.
7 .\" %%%LICENSE_START(VERBATIM)
8 .\" Permission is granted to make and distribute verbatim copies of this
9 .\" manual provided the copyright notice and this permission notice are
10 .\" preserved on all copies.
12 .\" Permission is granted to copy and distribute modified versions of this
13 .\" manual under the conditions for verbatim copying, provided that the
14 .\" entire resulting derived work is distributed under the terms of a
15 .\" permission notice identical to this one.
17 .\" Since the Linux kernel and libraries are constantly changing, this
18 .\" manual page may be incorrect or out-of-date. The author(s) assume no
19 .\" responsibility for errors or omissions, or for damages resulting from
20 .\" the use of the information contained herein. The author(s) may not
21 .\" have taken the same level of care in the production of this manual,
22 .\" which is licensed free of charge, as they might when working
25 .\" Formatted or processed versions of this manual, if unaccompanied by
26 .\" the source, must acknowledge the copyright and authors of this work.
29 .\" Modified 1993-07-24 by Rik Faith <faith@cs.unc.edu>
30 .\" Modified 1995-09-26 by Andries Brouwer <aeb@cwi.nl>
31 .\" and again on 960413 and 980804 and 981223.
32 .\" Modified 1998-12-11 by Jamie Lokier <jamie@imbolc.ucc.ie>
33 .\" Applied correction by Christian Ehrhardt - aeb, 990712
34 .\" Modified 2002-04-23 by Michael Kerrisk <mtk.manpages@gmail.com>
35 .\" Added note on F_SETFL and O_DIRECT
36 .\" Complete rewrite + expansion of material on file locking
37 .\" Incorporated description of F_NOTIFY, drawing on
38 .\" Stephen Rothwell's notes in Documentation/dnotify.txt.
39 .\" Added description of F_SETLEASE and F_GETLEASE
40 .\" Corrected and polished, aeb, 020527.
41 .\" Modified 2004-03-03 by Michael Kerrisk <mtk.manpages@gmail.com>
42 .\" Modified description of file leases: fixed some errors of detail
43 .\" Replaced the term "lease contestant" by "lease breaker"
44 .\" Modified, 27 May 2004, Michael Kerrisk <mtk.manpages@gmail.com>
45 .\" Added notes on capability requirements
46 .\" Modified 2004-12-08, added O_NOATIME after note from Martin Pool
47 .\" 2004-12-10, mtk, noted F_GETOWN bug after suggestion from aeb.
48 .\" 2005-04-08 Jamie Lokier <jamie@shareable.org>, mtk
49 .\" Described behavior of F_SETOWN/F_SETSIG in
50 .\" multithreaded processes, and generally cleaned
51 .\" up the discussion of F_SETOWN.
52 .\" 2005-05-20, Johannes Nicolai <johannes.nicolai@hpi.uni-potsdam.de>,
53 .\" mtk: Noted F_SETOWN bug for socket file descriptor in Linux 2.4
54 .\" and earlier. Added text on permissions required to send signal.
55 .\" 2009-09-30, Michael Kerrisk
56 .\" Note obsolete F_SETOWN behavior with threads.
57 .\" Document F_SETOWN_EX and F_GETOWN_EX
58 .\" 2010-06-17, Michael Kerrisk
59 .\" Document F_SETPIPE_SZ and F_GETPIPE_SZ.
61 .TH FCNTL 2 2014-02-20 "Linux" "Linux Programmer's Manual"
63 fcntl \- manipulate file descriptor
66 .B #include <unistd.h>
69 .BI "int fcntl(int " fd ", int " cmd ", ... /* " arg " */ );"
73 performs one of the operations described below on the open file descriptor
75 The operation is determined by
79 can take an optional third argument.
80 Whether or not this argument is required is determined by
82 The required argument type is indicated in parentheses after each
84 name (in most cases, the required type is
86 and we identify the argument using the name
90 is specified if the argument is not required.
91 .SS Duplicating a file descriptor
93 .BR F_DUPFD " (\fIint\fP)"
94 Find the lowest numbered available file descriptor
95 greater than or equal to
97 and make it be a copy of
99 This is different from
101 which uses exactly the descriptor specified.
103 On success, the new descriptor is returned.
109 .BR F_DUPFD_CLOEXEC " (\fIint\fP; since Linux 2.6.24)"
112 but additionally set the
113 close-on-exec flag for the duplicate descriptor.
114 Specifying this flag permits a program to avoid an additional
120 For an explanation of why this flag is useful,
121 see the description of
125 .SS File descriptor flags
126 The following commands manipulate the flags associated with
128 Currently, only one such flag is defined:
130 the close-on-exec flag.
133 bit is 0, the file descriptor will remain open across an
135 otherwise it will be closed.
137 .BR F_GETFD " (\fIvoid\fP)"
138 Read the file descriptor flags;
142 .BR F_SETFD " (\fIint\fP)"
143 Set the file descriptor flags to the value specified by
145 .SS File status flags
146 Each open file description has certain associated status flags,
151 and possibly modified by
153 Duplicated file descriptors
158 etc.) refer to the same open file description, and thus
159 share the same file status flags.
161 The file status flags and their semantics are described in
164 .BR F_GETFL " (\fIvoid\fP)"
165 Get the file access mode and the file status flags;
169 .BR F_SETFL " (\fIint\fP)"
170 Set the file status flags to the value specified by
173 .RB ( O_RDONLY ", " O_WRONLY ", " O_RDWR )
174 and file creation flags
176 .BR O_CREAT ", " O_EXCL ", " O_NOCTTY ", " O_TRUNC )
180 On Linux this command can change only the
188 It is not possible to change the
192 flags; see BUGS, below.
198 are used to acquire, release, and test for the existence of record
199 locks (also known as file-segment or file-region locks).
202 is a pointer to a structure that has at least the following fields
203 (in unspecified order).
209 short l_type; /* Type of lock: F_RDLCK,
211 short l_whence; /* How to interpret l_start:
212 SEEK_SET, SEEK_CUR, SEEK_END */
213 off_t l_start; /* Starting offset for lock */
214 off_t l_len; /* Number of bytes to lock */
215 pid_t l_pid; /* PID of process blocking our lock
223 .IR l_whence ", " l_start ", and " l_len
224 fields of this structure specify the range of bytes we wish to lock.
225 Bytes past the end of the file may be locked,
226 but not bytes before the start of the file.
229 is the starting offset for the lock, and is interpreted
231 the start of the file (if
235 the current file offset (if
239 or the end of the file (if
243 In the final two cases,
245 can be a negative number provided the
246 offset does not lie before the start of the file.
249 specifies the number of bytes to be locked.
252 is positive, then the range to be locked covers bytes
255 .IR l_start + l_len \-1.
258 has the special meaning: lock all bytes starting at the
259 location specified by
260 .IR l_whence " and " l_start
261 through to the end of file, no matter how large the file grows.
263 POSIX.1-2001 allows (but does not require)
264 an implementation to support a negative
268 is negative, the interval described by
274 This is supported by Linux since kernel versions 2.4.21 and 2.5.49.
278 field can be used to place a read
283 Any number of processes may hold a read lock (shared lock)
284 on a file region, but only one process may hold a write lock
286 An exclusive lock excludes all other locks,
287 both shared and exclusive.
288 A single process can hold only one type of lock on a file region;
289 if a new lock is applied to an already-locked region,
290 then the existing lock is converted to the new lock type.
291 (Such conversions may involve splitting, shrinking, or coalescing with
292 an existing lock if the byte range specified by the new lock does not
293 precisely coincide with the range of the existing lock.)
295 .BR F_SETLK " (\fIstruct flock *\fP)"
302 or release a lock (when
306 on the bytes specified by the
307 .IR l_whence ", " l_start ", and " l_len
310 If a conflicting lock is held by another process,
311 this call returns \-1 and sets
318 .BR F_SETLKW " (\fIstruct flock *\fP)"
321 but if a conflicting lock is held on the file, then wait for that
323 If a signal is caught while waiting, then the call is interrupted
324 and (after the signal handler has returned)
325 returns immediately (with return value \-1 and
332 .BR F_GETLK " (\fIstruct flock *\fP)"
333 On input to this call,
335 describes a lock we would like to place on the file.
336 If the lock could be placed,
338 does not actually place it, but returns
344 and leaves the other fields of the structure unchanged.
345 If one or more incompatible locks would prevent
346 this lock being placed, then
348 returns details about one of these locks in the
349 .IR l_type ", " l_whence ", " l_start ", and " l_len
354 to be the PID of the process holding that lock.
355 Note that the information returned by
357 may already be out of date by the time the caller inspects it.
359 In order to place a read lock,
361 must be open for reading.
362 In order to place a write lock,
364 must be open for writing.
365 To place both types of lock, open a file read-write.
367 As well as being removed by an explicit
369 record locks are automatically released when the process
370 terminates or if it closes
372 file descriptor referring to a file on which locks are held.
373 .\" (Additional file descriptors referring to the same file
374 .\" may have been obtained by calls to
375 .\" .BR open "(2), " dup "(2), " dup2 "(2), or " fcntl ().)
376 This is bad: it means that a process can lose the locks on
381 when for some reason a library function decides to open, read
384 Record locks are not inherited by a child created via
386 but are preserved across an
389 Because of the buffering performed by the
391 library, the use of record locking with routines in that package
392 should be avoided; use
397 .SS Mandatory locking
399 The above record locks may be either advisory or mandatory,
400 and are advisory by default.
402 Advisory locks are not enforced and are useful only between
403 cooperating processes.
405 Mandatory locks are enforced for all processes.
406 If a process tries to perform an incompatible access (e.g.,
410 on a file region that has an incompatible mandatory lock,
411 then the result depends upon whether the
413 flag is enabled for its open file description.
416 flag is not enabled, then
417 system call is blocked until the lock is removed
418 or converted to a mode that is compatible with the access.
421 flag is enabled, then the system call fails with the error
424 To make use of mandatory locks, mandatory locking must be enabled
425 both on the filesystem that contains the file to be locked,
426 and on the file itself.
427 Mandatory locking is enabled on a filesystem
428 using the "\-o mand" option to
434 Mandatory locking is enabled on a file by disabling
435 group execute permission on the file and enabling the set-group-ID
441 The Linux implementation of mandatory locking is unreliable.
451 are used to manage I/O availability signals:
453 .BR F_GETOWN " (\fIvoid\fP)"
454 Return (as the function result)
455 the process ID or process group currently receiving
459 signals for events on file descriptor
461 Process IDs are returned as positive values;
462 process group IDs are returned as negative values (but see BUGS below).
466 .BR F_SETOWN " (\fIint\fP)"
467 Set the process ID or process group ID that will receive
471 signals for events on file descriptor
475 A process ID is specified as a positive value;
476 a process group ID is specified as a negative value.
477 Most commonly, the calling process specifies itself as the owner
486 status flag on a file descriptor by using the
492 signal is sent whenever input or output becomes possible
493 on that file descriptor.
495 can be used to obtain delivery of a signal other than
497 If this permission check fails, then the signal is
500 Sending a signal to the owner process (group) specified by
502 is subject to the same permissions checks as are described for
504 where the sending process is the one that employs
506 (but see BUGS below).
508 If the file descriptor
515 signals that are delivered when out-of-band
516 data arrives on that socket.
518 is sent in any situation where
520 would report the socket as having an "exceptional condition".)
521 .\" The following appears to be rubbish. It doesn't seem to
522 .\" be true according to the kernel source, and I can write
523 .\" a program that gets a terminal-generated SIGIO even though
524 .\" it is not the foreground process group of the terminal.
527 .\" If the file descriptor
529 .\" refers to a terminal device, then SIGIO
530 .\" signals are sent to the foreground process group of the terminal.
532 The following was true in 2.6.x kernels up to and including
536 If a nonzero value is given to
538 in a multithreaded process running with a threading library
539 that supports thread groups (e.g., NPTL),
540 then a positive value given to
542 has a different meaning:
543 .\" The relevant place in the (2.6) kernel source is the
544 .\" 'switch' in fs/fcntl.c::send_sigio_to_task() -- MTK, Apr 2005
545 instead of being a process ID identifying a whole process,
546 it is a thread ID identifying a specific thread within a process.
547 Consequently, it may be necessary to pass
553 to get sensible results when
556 (In current Linux threading implementations,
557 a main thread's thread ID is the same as its process ID.
558 This means that a single-threaded program can equally use
563 Note, however, that the statements in this paragraph do not apply
566 signal generated for out-of-band data on a socket:
567 this signal is always sent to either a process or a process group,
568 depending on the value given to
570 .\" send_sigurg()/send_sigurg_to_task() bypasses
571 .\" kill_fasync()/send_sigio()/send_sigio_to_task()
572 .\" to directly call send_group_sig_info()
573 .\" -- MTK, Apr 2005 (kernel 2.6.11)
576 The above behavior was accidentally dropped in Linux 2.6.12,
577 and won't be restored.
578 From Linux 2.6.32 onward, use
584 signals at a particular thread.
586 .BR F_GETOWN_EX " (struct f_owner_ex *) (since Linux 2.6.32)"
587 Return the current file descriptor owner settings
588 as defined by a previous
591 The information is returned in the structure pointed to by
593 which has the following form:
606 field will have one of the values
613 field is a positive integer representing a thread ID, process ID,
619 .BR F_SETOWN_EX " (struct f_owner_ex *) (since Linux 2.6.32)"
620 This operation performs a similar task to
622 It allows the caller to direct I/O availability signals
623 to a specific thread, process, or process group.
624 The caller specifies the target of signals via
626 which is a pointer to a
631 field has one of the following values, which define how
637 Send the signal to the thread whose thread ID
638 (the value returned by a call to
646 Send the signal to the process whose ID
651 Send the signal to the process group whose ID
654 (Note that, unlike with
656 a process group ID is specified as a positive value here.)
659 .BR F_GETSIG " (\fIvoid\fP)"
660 Return (as the function result)
661 the signal sent when input or output becomes possible.
662 A value of zero means
665 Any other value (including
668 signal sent instead, and in this case additional info is available to
669 the signal handler if installed with
674 .BR F_SETSIG " (\fIint\fP)"
675 Set the signal sent when input or output becomes possible
676 to the value given in
678 A value of zero means to send the default
681 Any other value (including
683 is the signal to send instead, and in this case additional info
684 is available to the signal handler if installed with
687 .\" The following was true only up until 2.6.11:
689 .\" Additionally, passing a nonzero value to
691 .\" changes the signal recipient from a whole process to a specific thread
692 .\" within a process.
693 .\" See the description of
695 .\" for more details.
699 with a nonzero value, and setting
704 extra information about I/O events is passed to
710 field indicates the source is
714 field gives the file descriptor associated with the event.
716 there is no indication which file descriptors are pending, and you
717 should use the usual mechanisms
723 set etc.) to determine which file descriptors are available for I/O.
725 By selecting a real time signal (value >=
727 multiple I/O events may be queued using the same signal numbers.
728 (Queuing is dependent on available memory).
729 Extra information is available
732 is set for the signal handler, as above.
734 Note that Linux imposes a limit on the
735 number of real-time signals that may be queued to a
740 and if this limit is reached, then the kernel reverts to
743 and this signal is delivered to the entire
744 process rather than to a specific thread.
745 .\" See fs/fcntl.c::send_sigio_to_task() (2.4/2.6) sources -- MTK, Apr 05
747 Using these mechanisms, a program can implement fully asynchronous I/O
756 is specific to BSD and Linux.
761 specified in POSIX.1 is in conjunction with the use of the
764 (POSIX does not specify the
773 POSIX has asynchronous I/O and the
775 structure to achieve similar things; these are also available
776 in Linux as part of the GNU C Library (Glibc).
781 (Linux 2.4 onward) are used (respectively) to establish a new lease,
782 and retrieve the current lease, on the open file description
783 referred to by the file descriptor
785 A file lease provides a mechanism whereby the process holding
786 the lease (the "lease holder") is notified (via delivery of a signal)
787 when a process (the "lease breaker") tries to
791 the file referred to by that file descriptor.
793 .BR F_SETLEASE " (\fIint\fP)"
794 Set or remove a file lease according to which of the following
795 values is specified in the integer
800 Take out a read lease.
801 This will cause the calling process to be notified when
802 the file is opened for writing or is truncated.
803 .\" The following became true in kernel 2.6.10:
804 .\" See the man-pages-2.09 Changelog for further info.
805 A read lease can be placed only on a file descriptor that
809 Take out a write lease.
810 This will cause the caller to be notified when
811 the file is opened for reading or writing or is truncated.
812 A write lease may be placed on a file only if there are no
813 other open file descriptors for the file.
816 Remove our lease from the file.
819 Leases are associated with an open file description (see
821 This means that duplicate file descriptors (created by, for example,
825 refer to the same lease, and this lease may be modified
826 or released using any of these descriptors.
827 Furthermore, the lease is released by either an explicit
829 operation on any of these duplicate descriptors, or when all
830 such descriptors have been closed.
832 Leases may be taken out only on regular files.
833 An unprivileged process may take out a lease only on a file whose
834 UID (owner) matches the filesystem UID of the process.
837 capability may take out leases on arbitrary files.
839 .BR F_GETLEASE " (\fIvoid\fP)"
840 Indicates what type of lease is associated with the file descriptor
843 .BR F_RDLCK ", " F_WRLCK ", or " F_UNLCK ,
844 indicating, respectively, a read lease , a write lease, or no lease.
848 When a process (the "lease breaker") performs an
852 that conflicts with a lease established via
854 the system call is blocked by the kernel and
855 the kernel notifies the lease holder by sending it a signal
858 The lease holder should respond to receipt of this signal by doing
859 whatever cleanup is required in preparation for the file to be
860 accessed by another process (e.g., flushing cached buffers) and
861 then either remove or downgrade its lease.
862 A lease is removed by performing an
868 If the lease holder currently holds a write lease on the file,
869 and the lease breaker is opening the file for reading,
870 then it is sufficient for the lease holder to downgrade
871 the lease to a read lease.
872 This is done by performing an
879 If the lease holder fails to downgrade or remove the lease within
880 the number of seconds specified in
881 .IR /proc/sys/fs/lease-break-time ,
882 then the kernel forcibly removes or downgrades the lease holder's lease.
884 Once a lease break has been initiated,
886 returns the target lease type (either
890 depending on what would be compatible with the lease breaker)
891 until the lease holder voluntarily downgrades or removes the lease or
892 the kernel forcibly does so after the lease break timer expires.
894 Once the lease has been voluntarily or forcibly removed or downgraded,
895 and assuming the lease breaker has not unblocked its system call,
896 the kernel permits the lease breaker's system call to proceed.
898 If the lease breaker's blocked
902 is interrupted by a signal handler,
903 then the system call fails with the error
905 but the other steps still occur as described above.
906 If the lease breaker is killed by a signal while blocked in
910 then the other steps still occur as described above.
911 If the lease breaker specifies the
915 then the call immediately fails with the error
917 but the other steps still occur as described above.
919 The default signal used to notify the lease holder is
921 but this can be changed using the
927 command is performed (even one specifying
930 handler is established using
932 then the handler will receive a
934 structure as its second argument, and the
936 field of this argument will hold the descriptor of the leased file
937 that has been accessed by another process.
938 (This is useful if the caller holds leases against multiple files).
939 .SS File and directory change notification (dnotify)
941 .BR F_NOTIFY " (\fIint\fP)"
943 Provide notification when the directory referred to by
945 or any of the files that it contains is changed.
946 The events to be notified are specified in
948 which is a bit mask specified by ORing together zero or more of
955 A file was accessed (read, pread, readv)
958 A file was modified (write, pwrite, writev, truncate, ftruncate).
961 A file was created (open, creat, mknod, mkdir, link, symlink, rename).
964 A file was unlinked (unlink, rename to another directory, rmdir).
967 A file was renamed within this directory (rename).
970 The attributes of a file were changed (chown, chmod, utime[s]).
974 (In order to obtain these definitions, the
976 feature test macro must be defined before including
980 Directory notifications are normally "one-shot", and the application
981 must reregister to receive further notifications.
986 then notification will remain in effect until explicitly removed.
988 .\" The following does seem a poor API-design choice...
991 requests is cumulative, with the events in
993 being added to the set already monitored.
994 To disable notification of all events, make an
1000 Notification occurs via delivery of a signal.
1001 The default signal is
1003 but this can be changed using the
1007 In the latter case, the signal handler receives a
1009 structure as its second argument (if the handler was
1014 field of this structure contains the file descriptor which
1015 generated the notification (useful when establishing notification
1016 on multiple directories).
1018 Especially when using
1020 a real time signal should be used for notification,
1021 so that multiple notifications can be queued.
1024 New applications should use the
1026 interface (available since kernel 2.6.13),
1027 which provides a much superior interface for obtaining notifications of
1031 .SS Changing the capacity of a pipe
1033 .BR F_SETPIPE_SZ " (\fIint\fP; since Linux 2.6.35)"
1034 Change the capacity of the pipe referred to by
1039 An unprivileged process can adjust the pipe capacity to any value
1040 between the system page size and the limit defined in
1041 .IR /proc/sys/fs/pipe-max-size
1044 Attempts to set the pipe capacity below the page size are silently
1045 rounded up to the page size.
1046 Attempts by an unprivileged process to set the pipe capacity above the limit in
1047 .IR /proc/sys/fs/pipe-max-size
1050 a privileged process
1051 .RB ( CAP_SYS_RESOURCE )
1052 can override the limit.
1053 When allocating the buffer for the pipe,
1054 the kernel may use a capacity larger than
1056 if that is convenient for the implementation.
1059 operation returns the actual size used.
1060 Attempting to set the pipe capacity smaller than the amount
1061 of buffer space currently used to store data produces the error
1064 .BR F_GETPIPE_SZ " (\fIvoid\fP; since Linux 2.6.35)"
1065 Return (as the function result) the capacity of the pipe referred to by
1068 For a successful call, the return value depends on the operation:
1074 Value of file descriptor flags.
1077 Value of file status flags.
1080 Type of lease held on file descriptor.
1083 Value of descriptor owner.
1086 Value of signal sent when read or write becomes possible, or zero
1097 On error, \-1 is returned, and
1099 is set appropriately.
1102 .BR EACCES " or " EAGAIN
1103 Operation is prohibited by locks held by other processes.
1106 The operation is prohibited because the file has been memory-mapped by
1111 is not an open file descriptor, or the command was
1115 and the file descriptor open mode doesn't match with the
1116 type of lock requested.
1119 It was detected that the specified
1121 command would cause a deadlock.
1125 is outside your accessible address space.
1130 the command was interrupted by a signal; see
1133 .BR F_GETLK " and " F_SETLK ,
1134 the command was interrupted by a signal before the lock was checked or
1136 Most likely when locking a remote file (e.g., locking over
1137 NFS), but can sometimes happen locally.
1143 is negative or is greater than the maximum allowable value.
1147 is not an allowable signal number.
1152 the process already has the maximum number of file descriptors open.
1155 Too many segment locks open, lock table is full, or a remote locking
1156 protocol failed (e.g., locking over NFS).
1159 Attempted to clear the
1161 flag on a file that has the append-only attribute set.
1163 SVr4, 4.3BSD, POSIX.1-2001.
1174 are specified in POSIX.1-2001.
1179 are specified in POSIX.1-2001.
1180 (To get their definitions, define
1184 with the value 500 or greater, or define
1186 with the value 200809L or greater.)
1189 is specified in POSIX.1-2008.
1190 (To get this definition, define
1192 with the value 200809L or greater, or
1194 with the value 700 or greater.)
1209 macro to obtain these definitions.)
1211 .\" SVr4 documents additional EIO, ENOLINK and EOVERFLOW error conditions.
1215 system call was not designed to handle large file offsets
1221 system call was added in Linux 2.4.
1222 The newer system call employs a different structure for file locking,
1224 and corresponding commands,
1229 However, these details can be ignored by applications using glibc, whose
1231 wrapper function transparently employs the more recent system call
1232 where it is available.
1234 The errors returned by
1236 are different from those returned by
1239 Since kernel 2.0, there is no interaction between the types of lock
1245 Several systems have more fields in
1247 such as, for example,
1249 .\" e.g., Solaris 8 documents this field in fcntl(2), and Irix 6.5
1250 .\" documents it in fcntl(5). mtk, May 2007
1253 alone is not going to be very useful if the process holding the lock
1254 may live on a different machine.
1257 It is not possible to use
1259 to change the state of the
1264 .\" FIXME . According to POSIX.1-2001, O_SYNC should also be modifiable
1265 .\" via fcntl(2), but currently Linux does not permit this
1266 .\" See http://bugzilla.kernel.org/show_bug.cgi?id=5994
1267 Attempts to change the state of these flags are silently ignored.
1269 A limitation of the Linux system call conventions on some
1270 architectures (notably i386) means that if a (negative)
1271 process group ID to be returned by
1273 falls in the range \-1 to \-4095, then the return value is wrongly
1274 interpreted by glibc as an error in the system call;
1275 .\" glibc source: sysdeps/unix/sysv/linux/i386/sysdep.h
1276 that is, the return value of
1280 will contain the (positive) process group ID.
1283 operation avoids this problem.
1284 .\" mtk, Dec 04: some limited testing on alpha and ia64 seems to
1285 .\" indicate that ANY negative PGID value will cause F_GETOWN
1286 .\" to misinterpret the return as an error. Some other architectures
1287 .\" seem to have the same range check as i386.
1288 Since glibc version 2.11, glibc makes the kernel
1290 problem invisible by implementing
1295 In Linux 2.4 and earlier, there is bug that can occur
1296 when an unprivileged process uses
1298 to specify the owner
1299 of a socket file descriptor
1300 as a process (group) other than the caller.
1307 even when the owner process (group) is one that the caller
1308 has permission to send signals to.
1309 Despite this error return, the file descriptor owner is set,
1310 and signals will be sent to the owner.
1311 .SS Mandatory locking
1312 The implementation of mandatory locking in all known versions of Linux
1313 is subject to race conditions which render it unreliable:
1314 .\" http://marc.info/?l=linux-kernel&m=119013491707153&w=2
1317 call that overlaps with a lock may modify data after the mandatory lock is
1321 call that overlaps with a lock may detect changes to data that were made
1322 only after a write lock was acquired.
1323 Similar races exist between mandatory locks and
1325 It is therefore inadvisable to rely on mandatory locking.
1332 .BR capabilities (7),
1333 .BR feature_test_macros (7)
1336 .IR mandatory-locking.txt ,
1339 in the Linux kernel source directory
1340 .IR Documentation/filesystems/
1341 (on older kernels, these files are directly under the
1344 .I mandatory-locking.txt
1348 This page is part of release 3.64 of the Linux
1351 A description of the project,
1352 and information about reporting bugs,
1354 \%http://www.kernel.org/doc/man\-pages/.