path: root/man2/send.2

.\" Copyright (c) 1983, 1991 The Regents of the University of California.
.\" All rights reserved.
.\"
.\" SPDX-License-Identifier: BSD-4-Clause-UC
.\"
.\" Modified 1993-07-24 by Rik Faith <faith@cs.unc.edu>
.\" Modified 1996-10-22 by Eric S. Raymond <esr@thyrsus.com>
.\" Modified Oct 1998 by Andi Kleen
.\" Modified Oct 2003 by aeb
.\" Modified 2004-07-01 by mtk
.\"
.TH send 2 2023-03-30 "Linux man-pages 6.05.01"
.SH NAME
send, sendto, sendmsg \- send a message on a socket
.SH LIBRARY
Standard C library
.RI ( libc ", " \-lc )
.SH SYNOPSIS
.nf
.B #include <sys/socket.h>
.PP
.BI "ssize_t send(int " sockfd ", const void " buf [. len "], size_t " len \
", int " flags );
.BI "ssize_t sendto(int " sockfd ", const void " buf [. len "], size_t " len \
", int " flags ,
.BI "               const struct sockaddr *" dest_addr ", socklen_t " addrlen );
.BI "ssize_t sendmsg(int " sockfd ", const struct msghdr *" msg \
", int " flags );
.fi
.SH DESCRIPTION
The system calls
.BR send (),
.BR sendto (),
and
.BR sendmsg ()
are used to transmit a message to another socket.
.PP
The
.BR send ()
call may be used only when the socket is in a
.I connected
state (so that the intended recipient is known).
The only difference between
.BR send ()
and
.BR write (2)
is the presence of
.IR flags .
With a zero
.I flags
argument,
.BR send ()
is equivalent to
.BR write (2).
Also, the following call
.PP
.in +4n
.EX
send(sockfd, buf, len, flags);
.EE
.in
.PP
is equivalent to
.PP
.in +4n
.EX
sendto(sockfd, buf, len, flags, NULL, 0);
.EE
.in
.PP
The argument
.I sockfd
is the file descriptor of the sending socket.
.PP
If
.BR sendto ()
is used on a connection-mode
.RB ( SOCK_STREAM ,
.BR SOCK_SEQPACKET )
socket, the arguments
.I dest_addr
and
.I addrlen
are ignored (and the error
.B EISCONN
may be returned when they are
not NULL and 0), and the error
.B ENOTCONN
is returned when the socket was not actually connected.
Otherwise, the address of the target is given by
.I dest_addr
with
.I addrlen
specifying its size.
For
.BR sendmsg (),
the address of the target is given by
.IR msg.msg_name ,
with
.I msg.msg_namelen
specifying its size.
.PP
For
.BR send ()
and
.BR sendto (),
the message is found in
.I buf
and has length
.IR len .
For
.BR sendmsg (),
the message is pointed to by the elements of the array
.IR msg.msg_iov .
The
.BR sendmsg ()
call also allows sending ancillary data (also known as control information).
.PP
If the message is too long to pass atomically through the
underlying protocol, the error
.B EMSGSIZE
is returned, and the message is not transmitted.
.PP
No indication of failure to deliver is implicit in a
.BR send ().
Locally detected errors are indicated by a return value of \-1.
.PP
When the message does not fit into the send buffer of the socket,
.BR send ()
normally blocks, unless the socket has been placed in nonblocking I/O
mode.
In nonblocking mode it would fail with the error
.B EAGAIN
or
.B EWOULDBLOCK
in this case.
The
.BR select (2)
call may be used to determine when it is possible to send more data.
.SS The flags argument
The
.I flags
argument is the bitwise OR
of zero or more of the following flags.
.\" FIXME . ? document MSG_PROXY (which went away in Linux 2.3.15)
.TP
.BR MSG_CONFIRM " (since Linux 2.3.15)"
Tell the link layer that forward progress happened: you got a successful
reply from the other side.
If the link layer doesn't get this
it will regularly reprobe the neighbor (e.g., via a unicast ARP).
Valid only on
.B SOCK_DGRAM
and
.B SOCK_RAW
sockets and currently implemented only for IPv4 and IPv6.
See
.BR arp (7)
for details.
.TP
.B MSG_DONTROUTE
Don't use a gateway to send out the packet, send to hosts only on
directly connected networks.
This is usually used only
by diagnostic or routing programs.
This is defined only for protocol
families that route; packet sockets don't.
.TP
.BR MSG_DONTWAIT " (since Linux 2.2)"
Enables nonblocking operation; if the operation would block,
.B EAGAIN
or
.B EWOULDBLOCK
is returned.
This provides similar behavior to setting the
.B O_NONBLOCK
flag (via the
.BR fcntl (2)
.B F_SETFL
operation), but differs in that
.B MSG_DONTWAIT
is a per-call option, whereas
.B O_NONBLOCK
is a setting on the open file description (see
.BR open (2)),
which will affect all threads in the calling process
and as well as other processes that hold file descriptors
referring to the same open file description.
.TP
.BR MSG_EOR " (since Linux 2.2)"
Terminates a record (when this notion is supported, as for sockets of type
.BR SOCK_SEQPACKET ).
.TP
.BR MSG_MORE " (since Linux 2.4.4)"
The caller has more data to send.
This flag is used with TCP sockets to obtain the same effect
as the
.B TCP_CORK
socket option (see
.BR tcp (7)),
with the difference that this flag can be set on a per-call basis.
.IP
Since Linux 2.6, this flag is also supported for UDP sockets, and informs
the kernel to package all of the data sent in calls with this flag set
into a single datagram which is transmitted only when a call is performed
that does not specify this flag.
(See also the
.B UDP_CORK
socket option described in
.BR udp (7).)
.TP
.BR MSG_NOSIGNAL " (since Linux 2.2)"
Don't generate a
.B SIGPIPE
signal if the peer on a stream-oriented socket has closed the connection.
The
.B EPIPE
error is still returned.
This provides similar behavior to using
.BR sigaction (2)
to ignore
.BR SIGPIPE ,
but, whereas
.B MSG_NOSIGNAL
is a per-call feature,
ignoring
.B SIGPIPE
sets a process attribute that affects all threads in the process.
.TP
.B MSG_OOB
Sends
.I out-of-band
data on sockets that support this notion (e.g., of type
.BR SOCK_STREAM );
the underlying protocol must also support
.I out-of-band
data.
.TP
.BR MSG_FASTOPEN " (since Linux 3.7)"
Attempts TCP Fast Open (RFC7413) and sends data in the SYN like a
combination of
.BR connect (2)
and
.BR write (2),
by performing an implicit
.BR connect (2)
operation.
It blocks until the data is buffered and the handshake has completed.
For a non-blocking socket,
it returns the number of bytes buffered and sent in the SYN packet.
If the cookie is not available locally,
it returns
.BR EINPROGRESS ,
and sends a SYN with a Fast Open cookie request automatically.
The caller needs to write the data again when the socket is connected.
On errors,
it sets the same
.I errno
as
.BR connect (2)
if the handshake fails.
This flag requires enabling TCP Fast Open client support on sysctl
.IR net.ipv4.tcp_fastopen .
.IP
Refer to
.B TCP_FASTOPEN_CONNECT
socket option in
.BR tcp (7)
for an alternative approach.
.SS sendmsg()
The definition of the
.I msghdr
structure employed by
.BR sendmsg ()
is as follows:
.PP
.in +4n
.EX
struct msghdr {
    void         *msg_name;       /* Optional address */
    socklen_t     msg_namelen;    /* Size of address */
    struct iovec *msg_iov;        /* Scatter/gather array */
    size_t        msg_iovlen;     /* # elements in msg_iov */
    void         *msg_control;    /* Ancillary data, see below */
    size_t        msg_controllen; /* Ancillary data buffer len */
    int           msg_flags;      /* Flags (unused) */
};
.EE
.in
.PP
The
.I msg_name
field is used on an unconnected socket to specify the target
address for a datagram.
It points to a buffer containing the address; the
.I msg_namelen
field should be set to the size of the address.
For a connected socket, these fields should be specified as NULL and 0,
respectively.
.PP
The
.I msg_iov
and
.I msg_iovlen
fields specify scatter-gather locations, as for
.BR writev (2).
.PP
You may send control information (ancillary data) using the
.I msg_control
and
.I msg_controllen
members.
The maximum control buffer length the kernel can process is limited
per socket by the value in
.IR /proc/sys/net/core/optmem_max ;
see
.BR socket (7).
For further information on the use of ancillary data in various
socket domains, see
.BR unix (7)
and
.BR ip (7).
.PP
The
.I msg_flags
field is ignored.
.\" Still to be documented:
.\"  Send file descriptors and user credentials using the
.\"  msg_control* fields.
.SH RETURN VALUE
On success, these calls return the number of bytes sent.
On error, \-1 is returned, and
.I errno
is set to indicate the error.
.SH ERRORS
These are some standard errors generated by the socket layer.
Additional errors
may be generated and returned from the underlying protocol modules;
see their respective manual pages.
.TP
.B EACCES
(For UNIX domain sockets, which are identified by pathname)
Write permission is denied on the destination socket file,
or search permission is denied for one of the directories
the path prefix.
(See
.BR path_resolution (7).)
.IP
(For UDP sockets) An attempt was made to send to a
network/broadcast address as though it was a unicast address.
.TP
.BR EAGAIN " or " EWOULDBLOCK
.\" Actually EAGAIN on Linux
The socket is marked nonblocking and the requested operation
would block.
POSIX.1-2001 allows either error to be returned for this case,
and does not require these constants to have the same value,
so a portable application should check for both possibilities.
.TP
.B EAGAIN
(Internet domain datagram sockets)
The socket referred to by
.I sockfd
had not previously been bound to an address and,
upon attempting to bind it to an ephemeral port,
it was determined that all port numbers in the ephemeral port range
are currently in use.
See the discussion of
.I /proc/sys/net/ipv4/ip_local_port_range
in
.BR ip (7).
.TP
.B EALREADY
Another Fast Open is in progress.
.TP
.B EBADF
.I sockfd
is not a valid open file descriptor.
.TP
.B ECONNRESET
Connection reset by peer.
.TP
.B EDESTADDRREQ
The socket is not connection-mode, and no peer address is set.
.TP
.B EFAULT
An invalid user space address was specified for an argument.
.TP
.B EINTR
A signal occurred before any data was transmitted; see
.BR signal (7).
.TP
.B EINVAL
Invalid argument passed.
.TP
.B EISCONN
The connection-mode socket was connected already but a
recipient was specified.
(Now either this error is returned, or the recipient specification
is ignored.)
.TP
.B EMSGSIZE
The socket type
.\" (e.g., SOCK_DGRAM )
requires that message be sent atomically, and the size
of the message to be sent made this impossible.
.TP
.B ENOBUFS
The output queue for a network interface was full.
This generally indicates that the interface has stopped sending,
but may be caused by transient congestion.
(Normally, this does not occur in Linux.
Packets are just silently dropped
when a device queue overflows.)
.TP
.B ENOMEM
No memory available.
.TP
.B ENOTCONN
The socket is not connected, and no target has been given.
.TP
.B ENOTSOCK
The file descriptor
.I sockfd
does not refer to a socket.
.TP
.B EOPNOTSUPP
Some bit in the
.I flags
argument is inappropriate for the socket type.
.TP
.B EPIPE
The local end has been shut down on a connection oriented socket.
In this case, the process
will also receive a
.B SIGPIPE
unless
.B MSG_NOSIGNAL
is set.
.SH VERSIONS
According to POSIX.1-2001, the
.I msg_controllen
field of the
.I msghdr
structure should be typed as
.IR socklen_t ,
and the
.I msg_iovlen
field should be typed as
.IR int ,
but glibc currently types both as
.IR size_t .
.\" glibc bug for msg_controllen raised 12 Mar 2006
.\" http://sourceware.org/bugzilla/show_bug.cgi?id=2448
.\" The problem is an underlying kernel issue: the size of the
.\" __kernel_size_t type used to type these fields varies
.\" across architectures, but socklen_t is always 32 bits,
.\" as (at least with GCC) is int.
.SH STANDARDS
POSIX.1-2008.
.PP
.B MSG_CONFIRM
is a Linux extension.
.SH HISTORY
4.4BSD, SVr4, POSIX.1-2001.
(first appeared in 4.2BSD).
.PP
POSIX.1-2001 describes only the
.B MSG_OOB
and
.B MSG_EOR
flags.
POSIX.1-2008 adds a specification of
.BR MSG_NOSIGNAL .
.SH NOTES
See
.BR sendmmsg (2)
for information about a Linux-specific system call
that can be used to transmit multiple datagrams in a single call.
.SH BUGS
Linux may return
.B EPIPE
instead of
.BR ENOTCONN .
.SH EXAMPLES
An example of the use of
.BR sendto ()
is shown in
.BR getaddrinfo (3).
.SH SEE ALSO
.BR fcntl (2),
.BR getsockopt (2),
.BR recv (2),
.BR select (2),
.BR sendfile (2),
.BR sendmmsg (2),
.BR shutdown (2),
.BR socket (2),
.BR write (2),
.BR cmsg (3),
.BR ip (7),
.BR ipv6 (7),
.BR socket (7),
.BR tcp (7),
.BR udp (7),
.BR unix (7)