xref: /freebsd-14.2/lib/libc/sys/open.2 (revision 1e078fb6)
1.\" Copyright (c) 1980, 1991, 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.\"     @(#)open.2	8.2 (Berkeley) 11/16/93
29.\" $FreeBSD$
30.\"
31.Dd May 29, 2023
32.Dt OPEN 2
33.Os
34.Sh NAME
35.Nm open , openat
36.Nd open or create a file for reading, writing or executing
37.Sh LIBRARY
38.Lb libc
39.Sh SYNOPSIS
40.In fcntl.h
41.Ft int
42.Fn open "const char *path" "int flags" "..."
43.Ft int
44.Fn openat "int fd" "const char *path" "int flags" "..."
45.Sh DESCRIPTION
46The file name specified by
47.Fa path
48is opened
49for either execution or reading and/or writing as specified by the
50argument
51.Fa flags
52and the file descriptor returned to the calling process.
53The
54.Fa flags
55argument may indicate the file is to be
56created if it does not exist (by specifying the
57.Dv O_CREAT
58flag).
59In this case
60.Fn open
61and
62.Fn openat
63require an additional argument
64.Fa "mode_t mode" ,
65and the file is created with mode
66.Fa mode
67as described in
68.Xr chmod 2
69and modified by the process' umask value (see
70.Xr umask 2 ) .
71.Pp
72The
73.Fn openat
74function is equivalent to the
75.Fn open
76function except in the case where the
77.Fa path
78specifies a relative path.
79For
80.Fn openat
81and relative
82.Fa path ,
83the file to be opened is determined relative to the directory
84associated with the file descriptor
85.Fa fd
86instead of the current working directory.
87The
88.Fa flag
89parameter and the optional fourth parameter correspond exactly to
90the parameters of
91.Fn open .
92If
93.Fn openat
94is passed the special value
95.Dv AT_FDCWD
96in the
97.Fa fd
98parameter, the current working directory is used
99and the behavior is identical to a call to
100.Fn open .
101.Pp
102When
103.Fn openat
104is called with an absolute
105.Fa path ,
106it ignores the
107.Fa fd
108argument.
109.Pp
110In
111.Xr capsicum 4
112capability mode,
113.Fn open
114is not permitted.
115The
116.Fa path
117argument to
118.Fn openat
119must be strictly relative to a file descriptor
120.Fa fd .
121.Fa path
122must not be an absolute path and must not contain ".." components
123which cause the path resolution to escape the directory hierarchy
124starting at
125.Fa fd .
126Additionally, no symbolic link in
127.Fa path
128may target absolute path or contain escaping ".." components.
129.Fa fd
130must not be
131.Dv AT_FDCWD .
132.Pp
133If the
134.Dv vfs.lookup_cap_dotdot
135.Xr sysctl 3
136MIB is set to zero, ".." components in the paths,
137used in capability mode,
138are completely disabled.
139If the
140.Dv vfs.lookup_cap_dotdot_nonlocal
141MIB is set to zero, ".." is not allowed if found on non-local filesystem.
142.Pp
143The flags specified are formed by
144.Em or Ns 'ing
145the following values
146.Pp
147.Bd -literal -offset indent -compact
148O_RDONLY	open for reading only
149O_WRONLY	open for writing only
150O_RDWR		open for reading and writing
151O_EXEC		open for execute only
152O_SEARCH	open for search only, an alias for O_EXEC
153O_NONBLOCK	do not block on open
154O_APPEND	append on each write
155O_CREAT		create file if it does not exist
156O_TRUNC		truncate size to 0
157O_EXCL		error if create and file exists
158O_SHLOCK	atomically obtain a shared lock
159O_EXLOCK	atomically obtain an exclusive lock
160O_DIRECT	eliminate or reduce cache effects
161O_FSYNC		synchronous writes (historical synonym for O_SYNC)
162O_SYNC		synchronous writes
163O_DSYNC		synchronous data writes
164O_NOFOLLOW	do not follow symlinks
165O_NOCTTY	ignored
166O_TTY_INIT	ignored
167O_DIRECTORY	error if file is not a directory
168O_CLOEXEC	set FD_CLOEXEC upon open
169O_VERIFY	verify the contents of the file
170O_RESOLVE_BENEATH	path resolution must not cross the fd directory
171O_PATH		record only the target path in the opened descriptor
172O_EMPTY_PATH	openat, open file referenced by fd if path is empty
173.Ed
174.Pp
175Opening a file with
176.Dv O_APPEND
177set causes each write on the file
178to be appended to the end.
179If
180.Dv O_TRUNC
181is specified and the
182file exists, the file is truncated to zero length.
183If
184.Dv O_EXCL
185is set with
186.Dv O_CREAT
187and the file already
188exists,
189.Fn open
190returns an error.
191This may be used to
192implement a simple exclusive access locking mechanism.
193If
194.Dv O_EXCL
195is set and the last component of the pathname is
196a symbolic link,
197.Fn open
198will fail even if the symbolic
199link points to a non-existent name.
200If the
201.Dv O_NONBLOCK
202flag is specified and the
203.Fn open
204system call would result
205in the process being blocked for some reason (e.g., waiting for
206carrier on a dialup line),
207.Fn open
208returns immediately.
209The descriptor remains in non-blocking mode for subsequent operations.
210.Pp
211If
212.Dv O_SYNC
213is used in the mask, all writes will
214immediately and synchronously be written to disk.
215.Dv O_FSYNC
216is an historical synonym for
217.Dv O_SYNC .
218.Pp
219If
220.Dv O_DSYNC
221is used in the mask, all data and metadata required to read the data will be
222synchronously written to disk, but changes to metadata such as file access and
223modification timestamps may be written later.
224.Pp
225If
226.Dv O_NOFOLLOW
227is used in the mask and the target file passed to
228.Fn open
229is a symbolic link then the
230.Fn open
231will fail.
232.Pp
233When opening a file, a lock with
234.Xr flock 2
235semantics can be obtained by setting
236.Dv O_SHLOCK
237for a shared lock, or
238.Dv O_EXLOCK
239for an exclusive lock.
240If creating a file with
241.Dv O_CREAT ,
242the request for the lock will never fail
243(provided that the underlying file system supports locking).
244.Pp
245.Dv O_DIRECT
246may be used to minimize or eliminate the cache effects of reading and writing.
247The system will attempt to avoid caching the data you read or write.
248If it cannot avoid caching the data,
249it will minimize the impact the data has on the cache.
250Use of this flag can drastically reduce performance if not used with care.
251.Pp
252.Dv O_NOCTTY
253may be used to ensure the OS does not assign this file as the
254controlling terminal when it opens a tty device.
255This is the default on
256.Fx ,
257but is present for
258.Tn POSIX
259compatibility.
260The
261.Fn open
262system call will not assign controlling terminals on
263.Fx .
264.Pp
265.Dv O_TTY_INIT
266may be used to ensure the OS restores the terminal attributes when
267initially opening a TTY.
268This is the default on
269.Fx ,
270but is present for
271.Tn POSIX
272compatibility.
273The initial call to
274.Fn open
275on a TTY will always restore default terminal attributes on
276.Fx .
277.Pp
278.Dv O_DIRECTORY
279may be used to ensure the resulting file descriptor refers to a
280directory.
281This flag can be used to prevent applications with elevated privileges
282from opening files which are even unsafe to open with
283.Dv O_RDONLY ,
284such as device nodes.
285.Pp
286.Dv O_CLOEXEC
287may be used to set
288.Dv FD_CLOEXEC
289flag for the newly returned file descriptor.
290.Pp
291.Dv O_VERIFY
292may be used to indicate to the kernel that the contents of the file should
293be verified before allowing the open to proceed.
294The details of what
295.Dq verified
296means is implementation specific.
297The run-time linker (rtld) uses this flag to ensure shared objects have
298been verified before operating on them.
299.Pp
300.Dv O_RESOLVE_BENEATH
301returns
302.Er ENOTCAPABLE
303if any intermediate component of the specified relative path does not
304reside in the directory hierarchy beneath the starting directory.
305Absolute paths or even the temporal escape from beneath of the starting
306directory is not allowed.
307.Pp
308When
309.Fa fd
310is opened with
311.Dv O_SEARCH ,
312execute permissions are checked at open time.
313The
314.Fa fd
315may not be used for any read operations like
316.Xr getdirentries 2 .
317The primary use for this descriptor will be as the lookup descriptor for the
318.Fn *at
319family of functions.
320.Pp
321.Dv O_PATH
322returns a file descriptor that can be used as a directory file descriptor for
323.Xr openat 2
324and other system calls taking a file descriptor argument, like
325.Xr fstatat 2
326and others.
327The other functionality of the returned file descriptor is limited to
328the descriptor-level operations.
329It can be used for
330.Bl -tag -width readlinkat(2) -offset indent -compact
331.It Xr fcntl 2
332but advisory locking is not allowed
333.It Xr dup 2
334.It Xr close 2
335.It Xr fstat 2
336.It Xr fexecve 2
337.It Dv SCM_RIGHTS
338can be passed over a
339.Xr unix 4
340socket using a
341.Dv SCM_RIGHTS
342message
343.It Xr kqueue 2
344using for
345.Dv EVFILT_VNODE
346.It Xr readlinkat 2
347.It Xr __acl_get_fd 2 , Xr __acl_aclcheck_fd 2
348.El
349But operations like
350.Xr read 2 ,
351.Xr ftruncate 2 ,
352and any other that operate on file and not on file descriptor (except
353.Xr fstat 2 ),
354are not allowed.
355.Pp
356A file descriptor created with the
357.Dv O_PATH
358flag can be opened into normal (operable) file descriptor by
359specifying it as the
360.Fa fd
361argument to
362.Fn openat
363with empty
364.Fa path
365and flag
366.Dv O_EMPTY_PATH .
367Such an open behaves as if the current path of the file referenced by
368.Fa fd
369is passed, except that the path walk permissions are not checked.
370See also the description of
371.Dv AT_EMPTY_PATH
372flag for
373.Xr fstatat 2
374and related syscalls.
375.Pp
376If successful,
377.Fn open
378returns a non-negative integer, termed a file descriptor.
379It returns \-1 on failure.
380The file pointer used to mark the current position within the
381file is set to the beginning of the file.
382.Pp
383If a sleeping open of a device node from
384.Xr devfs 5
385is interrupted by a signal, the call always fails with
386.Er EINTR ,
387even if the
388.Dv SA_RESTART
389flag is set for the signal.
390A sleeping open of a fifo (see
391.Xr mkfifo 2 )
392is restarted as normal.
393.Pp
394When a new file is created it is given the group of the directory
395which contains it.
396.Pp
397Unless
398.Dv O_CLOEXEC
399flag was specified,
400the new descriptor is set to remain open across
401.Xr execve 2
402system calls; see
403.Xr close 2 ,
404.Xr fcntl 2
405and
406.Dv O_CLOEXEC
407description.
408.Pp
409The system imposes a limit on the number of file descriptors
410open simultaneously by one process.
411The
412.Xr getdtablesize 2
413system call returns the current system limit.
414.Sh RETURN VALUES
415If successful,
416.Fn open
417and
418.Fn openat
419return a non-negative integer, termed a file descriptor.
420They return \-1 on failure, and set
421.Va errno
422to indicate the error.
423.Sh ERRORS
424The named file is opened unless:
425.Bl -tag -width Er
426.It Bq Er ENOTDIR
427A component of the path prefix is not a directory.
428.It Bq Er ENAMETOOLONG
429A component of a pathname exceeded 255 characters,
430or an entire path name exceeded 1023 characters.
431.It Bq Er ENOENT
432.Dv O_CREAT
433is not set and the named file does not exist.
434.It Bq Er ENOENT
435A component of the path name that must exist does not exist.
436.It Bq Er EACCES
437Search permission is denied for a component of the path prefix.
438.It Bq Er EACCES
439The required permissions (for reading and/or writing)
440are denied for the given flags.
441.It Bq Er EACCES
442.Dv O_TRUNC
443is specified and write permission is denied.
444.It Bq Er EACCES
445.Dv O_CREAT
446is specified,
447the file does not exist,
448and the directory in which it is to be created
449does not permit writing.
450.It Bq Er EPERM
451.Dv O_CREAT
452is specified, the file does not exist, and the directory in which it is to be
453created has its immutable flag set, see the
454.Xr chflags 2
455manual page for more information.
456.It Bq Er EPERM
457The named file has its immutable flag set and the file is to be modified.
458.It Bq Er EPERM
459The named file has its append-only flag set, the file is to be modified, and
460.Dv O_TRUNC
461is specified or
462.Dv O_APPEND
463is not specified.
464.It Bq Er ELOOP
465Too many symbolic links were encountered in translating the pathname.
466.It Bq Er EISDIR
467The named file is a directory, and the arguments specify
468it is to be modified.
469.It Bq Er EISDIR
470The named file is a directory, and the flags specified
471.Dv O_CREAT
472without
473.Dv O_DIRECTORY .
474.It Bq Er EROFS
475The named file resides on a read-only file system,
476and the file is to be modified.
477.It Bq Er EROFS
478.Dv O_CREAT
479is specified and the named file would reside on a read-only file system.
480.It Bq Er EMFILE
481The process has already reached its limit for open file descriptors.
482.It Bq Er ENFILE
483The system file table is full.
484.It Bq Er EMLINK
485.Dv O_NOFOLLOW
486was specified and the target is a symbolic link.
487.It Bq Er ENXIO
488The named file is a character special or block
489special file, and the device associated with this special file
490does not exist.
491.It Bq Er ENXIO
492.Dv O_NONBLOCK
493is set, the named file is a fifo,
494.Dv O_WRONLY
495is set, and no process has the file open for reading.
496.It Bq Er EINTR
497The
498.Fn open
499operation was interrupted by a signal.
500.It Bq Er EOPNOTSUPP
501.Dv O_SHLOCK
502or
503.Dv O_EXLOCK
504is specified but the underlying file system does not support locking.
505.It Bq Er EOPNOTSUPP
506The named file is a special file mounted through a file system that
507does not support access to it (e.g.\& NFS).
508.It Bq Er EWOULDBLOCK
509.Dv O_NONBLOCK
510and one of
511.Dv O_SHLOCK
512or
513.Dv O_EXLOCK
514is specified and the file is locked.
515.It Bq Er ENOSPC
516.Dv O_CREAT
517is specified,
518the file does not exist,
519and the directory in which the entry for the new file is being placed
520cannot be extended because there is no space left on the file
521system containing the directory.
522.It Bq Er ENOSPC
523.Dv O_CREAT
524is specified,
525the file does not exist,
526and there are no free inodes on the file system on which the
527file is being created.
528.It Bq Er EDQUOT
529.Dv O_CREAT
530is specified,
531the file does not exist,
532and the directory in which the entry for the new file
533is being placed cannot be extended because the
534user's quota of disk blocks on the file system
535containing the directory has been exhausted.
536.It Bq Er EDQUOT
537.Dv O_CREAT
538is specified,
539the file does not exist,
540and the user's quota of inodes on the file system on
541which the file is being created has been exhausted.
542.It Bq Er EIO
543An I/O error occurred while making the directory entry or
544allocating the inode for
545.Dv O_CREAT .
546.It Bq Er EINTEGRITY
547Corrupted data was detected while reading from the file system.
548.It Bq Er ETXTBSY
549The file is a pure procedure (shared text) file that is being
550executed and the
551.Fn open
552system call requests write access.
553.It Bq Er EFAULT
554The
555.Fa path
556argument
557points outside the process's allocated address space.
558.It Bq Er EEXIST
559.Dv O_CREAT
560and
561.Dv O_EXCL
562were specified and the file exists.
563.It Bq Er EOPNOTSUPP
564An attempt was made to open a socket (not currently implemented).
565.It Bq Er EINVAL
566An attempt was made to open a descriptor with an illegal combination
567of
568.Dv O_RDONLY ,
569.Dv O_WRONLY ,
570or
571.Dv O_RDWR ,
572and
573.Dv O_EXEC
574or
575.Dv O_SEARCH .
576.It Bq Er EBADF
577The
578.Fa path
579argument does not specify an absolute path and the
580.Fa fd
581argument is
582neither
583.Dv AT_FDCWD
584nor a valid file descriptor open for searching.
585.It Bq Er ENOTDIR
586The
587.Fa path
588argument is not an absolute path and
589.Fa fd
590is neither
591.Dv AT_FDCWD
592nor a file descriptor associated with a directory.
593.It Bq Er ENOTDIR
594.Dv O_DIRECTORY
595is specified and the file is not a directory.
596.It Bq Er ECAPMODE
597.Dv AT_FDCWD
598is specified and the process is in capability mode.
599.It Bq Er ECAPMODE
600.Fn open
601was called and the process is in capability mode.
602.It Bq Er ENOTCAPABLE
603.Fa path
604is an absolute path and the process is in capability mode.
605.It Bq Er ENOTCAPABLE
606.Fa path
607is an absolute path and
608.Dv O_RESOLVE_BENEATH
609is specified.
610.It Bq Er ENOTCAPABLE
611.Fa path
612contains a ".." component leading to a directory outside
613of the directory hierarchy specified by
614.Fa fd
615and the process is in capability mode.
616.It Bq Er ENOTCAPABLE
617.Fa path
618contains a ".." component leading to a directory outside
619of the directory hierarchy specified by
620.Fa fd
621and
622.Dv O_RESOLVE_BENEATH
623is specified.
624.It Bq Er ENOTCAPABLE
625.Fa path
626contains a ".." component, the
627.Dv vfs.lookup_cap_dotdot
628.Xr sysctl 3
629is set, and the process is in capability mode.
630.El
631.Sh SEE ALSO
632.Xr chmod 2 ,
633.Xr close 2 ,
634.Xr dup 2 ,
635.Xr fexecve 2 ,
636.Xr fhopen 2 ,
637.Xr getdtablesize 2 ,
638.Xr getfh 2 ,
639.Xr lgetfh 2 ,
640.Xr lseek 2 ,
641.Xr read 2 ,
642.Xr umask 2 ,
643.Xr write 2 ,
644.Xr fopen 3 ,
645.Xr capsicum 4
646.Sh STANDARDS
647These functions are specified by
648.St -p1003.1-2008 .
649.Fx
650sets
651.Va errno
652to
653.Er EMLINK instead of
654.Er ELOOP
655as specified by
656.Tn POSIX
657when
658.Dv O_NOFOLLOW
659is set in flags and the final component of pathname is a symbolic link
660to distinguish it from the case of too many symbolic link traversals
661in one of its non-final components.
662.Sh HISTORY
663The
664.Fn open
665function appeared in
666.At v1 .
667The
668.Fn openat
669function was introduced in
670.Fx 8.0 .
671.Dv O_DSYNC
672appeared in 13.0.
673.Sh BUGS
674The Open Group Extended API Set 2 specification requires that the test
675for whether
676.Fa fd
677is searchable is based on whether
678.Fa fd
679is open for searching, not whether the underlying directory currently
680permits searches.
681The present implementation of the
682.Fa openat
683checks the current permissions of directory instead.
684.Pp
685The
686.Fa mode
687argument is variadic and may result in different calling conventions
688than might otherwise be expected.
689