NAME

socket — create an endpoint for communication

LIBRARY

Standard C Library (libc, -lc)

SYNOPSIS

#include <sys/socket.h>

int
socket(int domain, int type, int protocol);

DESCRIPTION

The socket() system call creates an endpoint for communication and returns a descriptor.

The domain argument specifies a communications domain within which communication will take place; this selects the protocol family which should be used. These families are defined in the include file <sys/socket.h>. The currently understood formats are:

PF_LOCAL	Host-internal protocols (alias for PF_UNIX),
PF_UNIX		Host-internal protocols,
PF_INET		Internet version 4 protocols,
PF_INET6	Internet version 6 protocols,
PF_DIVERT	Firewall packet diversion/re-injection,
PF_ROUTE	Internal routing protocol,
PF_KEY		Internal key-management function,
PF_NETGRAPH	Netgraph sockets,
PF_NETLINK	Netlink protocols,
PF_BLUETOOTH	Bluetooth protocols,
PF_INET_SDP	OFED socket direct protocol (IPv4),
AF_HYPERV	HyperV sockets

Each protocol family is connected to an address family, which has the same name except that the prefix is “AF_” in place of “PF_”. Other protocol families may be also defined, beginning with “PF_”, with corresponding address families.

The socket has the indicated type, which specifies the semantics of communication. Currently defined types are:

SOCK_STREAM	Stream socket,
SOCK_DGRAM	Datagram socket,
SOCK_RAW	Raw-protocol interface,
SOCK_SEQPACKET	Sequenced packet stream

Additionally, the following flags are allowed in the type argument:

SOCK_CLOEXEC	Set close-on-exec on the new descriptor,
SOCK_CLOFORK	Set close-on-fork on the new descriptor,
SOCK_NONBLOCK	Set non-blocking mode on the new socket

The protocol argument specifies a particular protocol to be used with the socket. Normally only a single protocol exists to support a particular socket type within a given protocol family. However, it is possible that many protocols may exist, in which case a particular protocol must be specified in this manner. The protocol number to use is particular to the “communication domain” in which communication is to take place; see protocols(5). The protocol argument may be set to zero (0) to request the default implementation of a socket type for the protocol, if any.

STREAM SOCKET TYPE

The SOCK_STREAM socket type provides reliable, sequenced, full-duplex octet streams between the socket and a peer to which the socket is connected. A socket of type SOCK_STREAM needs to be in a connected state before any data can be sent or received. A connection to another socket is created with a connect(2) system call. (Some protocol families, such as the Internet family, support the notion of an “implied connect”, which permits data to be sent piggybacked onto a connect operation by using the sendto(2) system call.) Once connected, data may be sent using send(2), sendto(2), sendmsg(2) and write(2) system calls. Data may be received using recv(2), recvfrom(2), recvmsg(2), and read(2) system calls. Record boundaries are not maintained; data sent on a stream socket using output operations of one size can be received using input operations of smaller or larger sizes without loss of data. Data may be buffered; successful return from an output function does not imply that the data has been delivered to the peer or even transmitted from the local system. For certain protocols out-of-band data may also be transmitted as described in send(2) and received as described in recv(2).

If data cannot be successfully transmitted within a given time then the connection is considered broken, and subsequent operations shall fail with a protocol specific error code. A SIGPIPE signal is raised if a thread attempts to send data on a broken stream (one that is no longer connected). The signal can be suppressed by the MSG_NOSIGNAL flag with distinct send(2), sendto(2), and sendmsg(2) system calls or by the SO_NOSIGPIPE socket option set on the socket with setsockopt(2).

The SOCK_STREAM socket is supported by the following protocol families: PF_INET, PF_INET6, PF_UNIX, PF_BLUETOOTH, PF_HYPERV, and PF_INET_SDP. Out-of-band data transmission mechanism is supported for stream sockets of PF_INET and PF_INET6 protocol families.

DATAGRAM SOCKET TYPE

The SOCK_DGRAM socket type supports connectionless data transfer which is not necessarily acknowledged or reliable. Datagrams can be sent to the address specified (possibly multicast or broadcast) in each output operation, and incoming datagrams can be received from multiple sources. The source address of each datagram is available when receiving the datagram with recvfrom(2) or recvmsg(2). An application can also pre-specify a peer address with sendto(2) or sendmsg(2), in which case calls to output functions that do not specify a peer address shall send to the pre-specified peer. If a peer has been specified, only datagrams from that peer shall be received. A datagram shall be sent in a single output operation, and needs to be received in a single input operation. The maximum size of a datagram is protocol-specific. Output datagrams may be buffered within the system; thus, a successful return from an output function does not guarantee that a datagram is actually sent or received.

The SOCK_DGRAM socket is supported by the following protocol families: PF_INET, PF_INET6, PF_UNIX, PF_NETGRAPH, and PF_NETLINK.

SEQUENCED PACKET SOCKET TYPE

The SOCK_SEQPACKET socket type is similar to the SOCK_STREAM type, and is also connection-oriented. The only difference between these types is that record boundaries are maintained using the SOCK_SEQPACKET type. A record can be sent using one or more output operations and received using one or more input operations, but a single operation never transfers parts of more than one record. Record boundaries are set by the sender with the MSG_EOR flag of send(2) or sendmsg(2) functions. There is no possibility to set a record boundary with write(2). Record boundaries are visible to the receiver via the MSG_EOR flag in the received message flags returned by the recvmsg(2) function. It is protocol-specific whether a maximum record size is imposed.

The SOCK_SEQPACKET socket is supported by the following protocol families: PF_INET, PF_INET6, and PF_UNIX.

RAW SOCKET TYPE

The SOCK_RAW socket type provides access to internal network protocols and interfaces. It is a datagram socket in its nature, thus has the same semantics of read and write operations. The SOCK_RAW type is available only to the super-user and is described in ip(4) and ip6(4).

NON-BLOCKING MODE

A socket can be created in non-blocking mode with the help of SOCK_NONBLOCK flag. Alternatively, the non-blocking mode on a socket can be turned on and off with the help of the O_NONBLOCK flag of the fcntl(2) system call.

When a non-blocking socket has not enough data in its receive buffer to fulfill the application supplied buffer, then data receiving system calls like recv(2), recvfrom(2), recvmsg(2) and read(2) will not block waiting for the data but immediately return. Return value will indicate amount of bytes read into the supplied buffer. The errno will be set to EAGAIN (has same value as EWOULDBLOCK).

If application tries to send more data on a non-blocking socket than the socket send buffer can accomodate with send(2), sendto(2), sendmsg(2) or write(2) system calls partial data will be sent. Return value will indicate amount of bytes sent. The errno will be set to EAGAIN. Note that sockets of SOCK_DGRAM type are unreliable, thus for these sockets sending operations will never fail with EAGAIN in non-blocking mode neither will block in blocking mode.

OTHER OPERATIONS ON SOCKETS

Since socket descriptors are file descriptors, many generic file operations performed by fcntl(2), apply. Socket descriptors can be used with all event engines, such as kevent(2), select(2) and poll(2).

An fcntl(2) system call can be used to specify a process group to receive a SIGURG signal when the out-of-band data arrives. It may also enable non-blocking I/O and asynchronous notification of I/O events via SIGIO.

The operation of sockets is controlled by socket level options. These options are defined in the file <sys/socket.h>. The setsockopt(2) and getsockopt(2) system calls are used to set and get options, respectively.

Connection associated with a socket can be terminated by close(2) system call. One direction of communication can be disabled with shutdown(2).

RETURN VALUES

A -1 is returned if an error occurs, otherwise the return value is a descriptor referencing the socket.

ERRORS

The socket() system call fails if:

[EACCES]: Permission to create a socket of the specified type and/or protocol is denied.
[EAFNOSUPPORT]: The address family (domain) is not supported or the specified domain is not supported by this protocol family.
[EMFILE]: The per-process descriptor table is full.
[ENFILE]: The system file table is full.
[ENOBUFS]: Insufficient buffer space is available. The socket cannot be created until sufficient resources are freed.
[EPERM]: User has insufficient privileges to carry out the requested operation.
[EPROTONOSUPPORT]: The protocol type or the specified protocol is not supported within this domain.
[EPROTOTYPE]: The socket type is not supported by the protocol.

STANDARDS

The socket() function conforms to IEEE Std 1003.1-2008 (“POSIX.1”). The POSIX standard specifies only the AF_INET, AF_INET6, and AF_UNIX constants for address families, and requires the use of AF_* constants for the domain argument of socket(). The SOCK_CLOEXEC and SOCK_CLOFORK flags are expected to conform to IEEE Std 1003.1-2024 (“POSIX.1”). POSIX standard. The SOCK_RDM type, the PF_* constants, and other address families are FreeBSD extensions.

HISTORY

The socket() system call appeared in 4.2BSD.

The SOCK_CLOFORK flag appeared in FreeBSD 15.0.