.\" This manpage is Copyright (C) 1992 Drew Eckhardt;
.\"             and Copyright (C) 1993 Michael Haardt, Ian Jackson.
.\" and Copyright (C) 2006, 2014 Michael Kerrisk
.\"
.\" SPDX-License-Identifier: Linux-man-pages-copyleft
.\"
.\" Modified 1993-07-23 by Rik Faith <faith@cs.unc.edu>
.\" Modified 1994-08-21 by Michael Haardt
.\" Modified 2004-06-23 by Michael Kerrisk <mtk.manpages@gmail.com>
.\" Modified 2005-04-04, as per suggestion by Michael Hardt for rename.2
.\"
.TH link 2 2024-05-02 "Linux man-pages 6.8"
.SH NAME
link, linkat \- make a new name for a file
.SH LIBRARY
Standard C library
.RI ( libc ", " \-lc )
.SH SYNOPSIS
.nf
.B #include <unistd.h>
.P
.BI "int link(const char *" oldpath ", const char *" newpath );
.P
.BR "#include <fcntl.h>           " "/* Definition of " AT_* " constants */"
.B #include <unistd.h>
.P
.BI "int linkat(int " olddirfd ", const char *" oldpath ,
.BI "           int " newdirfd ", const char *" newpath ", int " flags );
.fi
.P
.RS -4
Feature Test Macro Requirements for glibc (see
.BR feature_test_macros (7)):
.RE
.P
.BR linkat ():
.nf
    Since glibc 2.10:
        _POSIX_C_SOURCE >= 200809L
    Before glibc 2.10:
        _ATFILE_SOURCE
.fi
.SH DESCRIPTION
.BR link ()
creates a new link (also known as a hard link) to an existing file.
.P
If
.I newpath
exists, it will
.I not
be overwritten.
.P
This new name may be used exactly as the old one for any operation;
both names refer to the same file (and so have the same permissions
and ownership) and it is impossible to tell which name was the
"original".
.SS linkat()
The
.BR linkat ()
system call operates in exactly the same way as
.BR link (),
except for the differences described here.
.P
If the pathname given in
.I oldpath
is relative, then it is interpreted relative to the directory
referred to by the file descriptor
.I olddirfd
(rather than relative to the current working directory of
the calling process, as is done by
.BR link ()
for a relative pathname).
.P
If
.I oldpath
is relative and
.I olddirfd
is the special value
.BR AT_FDCWD ,
then
.I oldpath
is interpreted relative to the current working
directory of the calling process (like
.BR link ()).
.P
If
.I oldpath
is absolute, then
.I olddirfd
is ignored.
.P
The interpretation of
.I newpath
is as for
.IR oldpath ,
except that a relative pathname is interpreted relative
to the directory referred to by the file descriptor
.IR newdirfd .
.P
The following values can be bitwise ORed in
.IR flags :
.TP
.BR AT_EMPTY_PATH " (since Linux 2.6.39)"
.\" commit 11a7b371b64ef39fc5fb1b6f2218eef7c4d035e3
If
.I oldpath
is an empty string, create a link to the file referenced by
.I olddirfd
(which may have been obtained using the
.BR open (2)
.B O_PATH
flag).
In this case,
.I olddirfd
can refer to any type of file except a directory.
This will generally not work if the file has a link count of zero (files
created with
.B O_TMPFILE
and without
.B O_EXCL
are an exception).
The caller must have the
.B CAP_DAC_READ_SEARCH
capability in order to use this flag.
This flag is Linux-specific; define
.B _GNU_SOURCE
.\" Before glibc 2.16, defining _ATFILE_SOURCE sufficed
to obtain its definition.
.TP
.BR AT_SYMLINK_FOLLOW " (since Linux 2.6.18)"
By default,
.BR linkat (),
does not dereference
.I oldpath
if it is a symbolic link (like
.BR link ()).
The flag
.B AT_SYMLINK_FOLLOW
can be specified in
.I flags
to cause
.I oldpath
to be dereferenced if it is a symbolic link.
If procfs is mounted,
this can be used as an alternative to
.BR AT_EMPTY_PATH ,
like this:
.IP
.in +4n
.EX
linkat(AT_FDCWD, "/proc/self/fd/<fd>", newdirfd,
       newname, AT_SYMLINK_FOLLOW);
.EE
.in
.P
Before Linux 2.6.18, the
.I flags
argument was unused, and had to be specified as 0.
.P
See
.BR openat (2)
for an explanation of the need for
.BR linkat ().
.SH RETURN VALUE
On success, zero is returned.
On error, \-1 is returned, and
.I errno
is set to indicate the error.
.SH ERRORS
.TP
.B EACCES
Write access to the directory containing
.I newpath
is denied, or search permission is denied for one of the directories
in the path prefix of
.I oldpath
or
.IR newpath .
(See also
.BR path_resolution (7).)
.TP
.B EDQUOT
The user's quota of disk blocks on the filesystem has been exhausted.
.TP
.B EEXIST
.I newpath
already exists.
.TP
.B EFAULT
.IR oldpath " or " newpath " points outside your accessible address space."
.TP
.B EIO
An I/O error occurred.
.TP
.B ELOOP
Too many symbolic links were encountered in resolving
.IR oldpath " or " newpath .
.TP
.B EMLINK
The file referred to by
.I oldpath
already has the maximum number of links to it.
For example, on an
.BR ext4 (5)
filesystem that does not employ the
.I dir_index
feature, the limit on the number of hard links to a file is 65,000; on
.BR btrfs (5),
the limit is 65,535 links.
.TP
.B ENAMETOOLONG
.IR oldpath " or " newpath " was too long."
.TP
.B ENOENT
A directory component in
.IR oldpath " or " newpath
does not exist or is a dangling symbolic link.
.TP
.B ENOMEM
Insufficient kernel memory was available.
.TP
.B ENOSPC
The device containing the file has no room for the new directory
entry.
.TP
.B ENOTDIR
A component used as a directory in
.IR oldpath " or " newpath
is not, in fact, a directory.
.TP
.B EPERM
.I oldpath
is a directory.
.TP
.B EPERM
The filesystem containing
.IR oldpath " and " newpath
does not support the creation of hard links.
.TP
.BR EPERM " (since Linux 3.6)"
The caller does not have permission to create a hard link to this file
(see the description of
.I /proc/sys/fs/protected_hardlinks
in
.BR proc (5)).
.TP
.B EPERM
.I oldpath
is marked immutable or append-only.
(See
.BR ioctl_iflags (2).)
.TP
.B EROFS
The file is on a read-only filesystem.
.TP
.B EXDEV
.IR oldpath " and " newpath
are not on the same mounted filesystem.
(Linux permits a filesystem to be mounted at multiple points, but
.BR link ()
does not work across different mounts,
even if the same filesystem is mounted on both.)
.P
The following additional errors can occur for
.BR linkat ():
.TP
.B EBADF
.I oldpath
.RI ( newpath )
is relative but
.I olddirfd
.RI ( newdirfd )
is neither
.B AT_FDCWD
nor a valid file descriptor.
.TP
.B EINVAL
An invalid flag value was specified in
.IR flags .
.TP
.B ENOENT
.B AT_EMPTY_PATH
was specified in
.IR flags ,
but the caller did not have the
.B CAP_DAC_READ_SEARCH
capability.
.TP
.B ENOENT
An attempt was made to link to the
.I /proc/self/fd/NN
file corresponding to a file descriptor created with
.IP
.in +4n
.EX
open(path, O_TMPFILE | O_EXCL, mode);
.EE
.in
.IP
See
.BR open (2).
.TP
.B ENOENT
An attempt was made to link to a
.I /proc/self/fd/NN
file corresponding to a file that has been deleted.
.TP
.B ENOENT
.I oldpath
is a relative pathname and
.I olddirfd
refers to a directory that has been deleted,
or
.I newpath
is a relative pathname and
.I newdirfd
refers to a directory that has been deleted.
.TP
.B ENOTDIR
.I oldpath
is relative and
.I olddirfd
is a file descriptor referring to a file other than a directory;
or similar for
.I newpath
and
.I newdirfd
.TP
.B EPERM
.B AT_EMPTY_PATH
was specified in
.IR flags ,
.I oldpath
is an empty string, and
.I olddirfd
refers to a directory.
.SH VERSIONS
POSIX.1-2001 says that
.BR link ()
should dereference
.I oldpath
if it is a symbolic link.
However, since Linux 2.0,
.\" more precisely: since Linux 1.3.56
Linux does not do so: if
.I oldpath
is a symbolic link, then
.I newpath
is created as a (hard) link to the same symbolic link file
(i.e.,
.I newpath
becomes a symbolic link to the same file that
.I oldpath
refers to).
Some other implementations behave in the same manner as Linux.
.\" For example, the default Solaris compilation environment
.\" behaves like Linux, and contributors to a March 2005
.\" thread in the Austin mailing list reported that some
.\" other (System V) implementations did/do the same -- MTK, Apr 05
POSIX.1-2008 changes the specification of
.BR link (),
making it implementation-dependent whether or not
.I oldpath
is dereferenced if it is a symbolic link.
For precise control over the treatment of symbolic links when
creating a link, use
.BR linkat ().
.SS glibc
On older kernels where
.BR linkat ()
is unavailable, the glibc wrapper function falls back to the use of
.BR link (),
unless the
.B AT_SYMLINK_FOLLOW
is specified.
When
.I oldpath
and
.I newpath
are relative pathnames,
glibc constructs pathnames based on the symbolic links in
.I /proc/self/fd
that correspond to the
.I olddirfd
and
.I newdirfd
arguments.
.SH STANDARDS
.TP
.BR link ()
POSIX.1-2008.
.SH HISTORY
.TP
.BR link ()
SVr4, 4.3BSD, POSIX.1-2001 (but see VERSIONS).
.\" SVr4 documents additional ENOLINK and
.\" EMULTIHOP error conditions; POSIX.1 does not document ELOOP.
.\" X/OPEN does not document EFAULT, ENOMEM or EIO.
.TP
.BR linkat ()
POSIX.1-2008.
Linux 2.6.16,
glibc 2.4.
.SH NOTES
Hard links, as created by
.BR link (),
cannot span filesystems.
Use
.BR symlink (2)
if this is required.
.SH BUGS
On NFS filesystems, the return code may be wrong in case the NFS server
performs the link creation and dies before it can say so.
Use
.BR stat (2)
to find out if the link got created.
.SH SEE ALSO
.BR ln (1),
.BR open (2),
.BR rename (2),
.BR stat (2),
.BR symlink (2),
.BR unlink (2),
.BR path_resolution (7),
.BR symlink (7)