1SOCKET(7) Linux Programmer's Manual SOCKET(7)
2
6 socket - Linux socket interface
7
9 #include <sys/socket.h>
10 mysocket = socket(int socket_family, int socket_type, int protocol);
11
14 This manual page describes the Linux networking socket layer user
15 interface. The BSD compatible sockets are the uniform interface between
16 the user process and the network protocol stacks in the kernel. The
17 protocol modules are grouped into protocol families like PF_INET,
18 PF_IPX, PF_PACKET and socket types like SOCK_STREAM or SOCK_DGRAM. See
19 socket(2) for more information on families and types.
20
23 These functions are used by the user process to send or receive packets
24 and to do other socket operations. For more information see their
25 respective manual pages.
26 socket(2) creates a socket, connect(2) connects a socket to a remote
28 socket address, the bind(2) function binds a socket to a local socket
29 address, listen(2) tells the socket that new connections shall be
30 accepted, and accept(2) is used to get a new socket with a new incoming
31 connection. socketpair(2) returns two connected anonymous sockets
32 (only implemented for a few local families like PF_UNIX)
33 send(2), sendto(2), and sendmsg(2) send data over a socket, and
35 recv(2), recvfrom(2), recvmsg(2) receive data from a socket. poll(2)
36 and select(2) wait for arriving data or a readiness to send data. In
37 addition, the standard I/O operations like write(2), writev(2), send‐
38 file(2), read(2), and readv(2) can be used to read and write data.
39 getsockname(2) returns the local socket address and getpeername(2)
41 returns the remote socket address. getsockopt(2) and setsockopt(2) are
42 used to set or get socket layer or protocol options. ioctl(2) can be
43 used to set or read some other options.
44 close(2) is used to close a socket. shutdown(2) closes parts of a full
46 duplex socket connection.
47 Seeking, or calling pread(2) or pwrite(2) with a non-zero position is
49 not supported on sockets.
50 It is possible to do non-blocking I/O on sockets by setting the O_NON‐
52 BLOCK flag on a socket file descriptor using fcntl(2). Then all opera‐
53 tions that would block will (usually) return with EAGAIN (operation
54 should be retried later); connect(2) will return EINPROGRESS error.
55 The user can then wait for various events via poll(2) or select(2).
56 ┌────────────────────────────────────────────────────────────────────┐
58 │ I/O events │
59 ├───────────┬───────────┬────────────────────────────────────────────┤
60 │Event │ Poll flag │ Occurrence │
61 ├───────────┼───────────┼────────────────────────────────────────────┤
62 │Read │ POLLIN │ New data arrived. │
63 ├───────────┼───────────┼────────────────────────────────────────────┤
64 │Read │ POLLIN │ A connection setup has been completed (for │
65 │ │ │ connection-oriented sockets) │
66 ├───────────┼───────────┼────────────────────────────────────────────┤
67 │Read │ POLLHUP │ A disconnection request has been initiated │
68 │ │ │ by the other end. │
69 ├───────────┼───────────┼────────────────────────────────────────────┤
70 │Read │ POLLHUP │ A connection is broken (only for connec‐ │
71 │ │ │ tion-oriented protocols). When the socket │
72 │ │ │ is written SIGPIPE is also sent. │
73 ├───────────┼───────────┼────────────────────────────────────────────┤
74 │Write │ POLLOUT │ Socket has enough send buffer space for │
75 │ │ │ writing new data. │
76 ├───────────┼───────────┼────────────────────────────────────────────┤
77 │Read/Write │ POLLIN| │ An outgoing connect(2) finished. │
78 │ │ POLLOUT │ │
79 ├───────────┼───────────┼────────────────────────────────────────────┤
80 │Read/Write │ POLLERR │ An asynchronous error occurred. │
81 ├───────────┼───────────┼────────────────────────────────────────────┤
82 │Read/Write │ POLLHUP │ The other end has shut down one direction. │
83 ├───────────┼───────────┼────────────────────────────────────────────┤
84 │Exception │ POLLPRI │ Urgent data arrived. SIGURG is sent then. │
85 └───────────┴───────────┴────────────────────────────────────────────┘
86 An alternative to poll() and select() is to let the kernel inform the
88 application about events via a SIGIO signal. For that the FASYNC flag
89 must be set on a socket file descriptor via fcntl(2) and a valid signal
90 handler for SIGIO must be installed via sigaction(2). See the SIGNALS
91 discussion below.
92
94 These socket options can be set by using setsockopt(2) and read with
95 getsockopt(2) with the socket level set to SOL_SOCKET for all sockets:
96 SO_ACCEPTCONN
98 Returns a value indicating whether or not this socket has been
99 marked to accept connections with listen(). The value 0 indi‐
100 cates that this is not a listening socket, the value 1 indicates
101 that this is a listening socket. Can only be read with getsock‐
102 opt().
103 SO_BINDTODEVICE
105 Bind this socket to a particular device like “eth0”, as speci‐
106 fied in the passed interface name. If the name is an empty
107 string or the option length is zero, the socket device binding
108 is removed. The passed option is a variable-length null termi‐
109 nated interface name string with the maximum size of IFNAMSIZ.
110 If a socket is bound to an interface, only packets received from
111 that particular interface are processed by the socket. Note that
112 this only works for some socket types, particularly AF_INET
113 sockets. It is not supported for packet sockets (use normal
114 bind(8) there).
115 SO_BROADCAST
117 Set or get the broadcast flag. When enabled, datagram sockets
118 receive packets sent to a broadcast address and they are allowed
119 to send packets to a broadcast address. This option has no
120 effect on stream-oriented sockets.
121 SO_BSDCOMPAT
123 Enable BSD bug-to-bug compatibility. This is used by the UDP
124 protocol module in Linux 2.0 and 2.2. If enabled ICMP errors
125 received for a UDP socket will not be passed to the user pro‐
126 gram. In later kernel versions, support for this option has
127 been phased out: Linux 2.4 silently ignores it, and Linux 2.6
128 generates a kernel warning (printk()) if a program uses this
129 option. Linux 2.0 also enabled BSD bug-to-bug compatibility
130 options (random header changing, skipping of the broadcast flag)
131 for raw sockets with this option, but that was removed in Linux
132 2.2.
133 SO_DEBUG
135 Enable socket debugging. Only allowed for processes with the
136 CAP_NET_ADMIN capability or an effective user ID of 0.
137 SO_ERROR
139 Get and clear the pending socket error. Only valid as a getsock‐
140 opt(). Expects an integer.
141 SO_DONTROUTE
143 Don't send via a gateway, only send to directly connected hosts.
144 The same effect can be achieved by setting the MSG_DONTROUTE
145 flag on a socket send(2) operation. Expects an integer boolean
146 flag.
147 SO_KEEPALIVE
149 Enable sending of keep-alive messages on connection-oriented
150 sockets. Expects an integer boolean flag.
151 SO_LINGER
153 Sets or gets the SO_LINGER option. The argument is a linger
154 structure.
155 struct linger {
157 int l_onoff; /* linger active */
158 int l_linger; /* how many seconds to linger for */
159 };
160 When enabled, a close(2) or shutdown(2) will not return until
162 all queued messages for the socket have been successfully sent
163 or the linger timeout has been reached. Otherwise, the call
164 returns immediately and the closing is done in the background.
165 When the socket is closed as part of exit(2), it always lingers
166 in the background.
167 SO_OOBINLINE
169 If this option is enabled, out-of-band data is directly placed
170 into the receive data stream. Otherwise out-of-band data is
171 only passed when the MSG_OOB flag is set during receiving.
172 SO_PASSCRED
174 Enable or disable the receiving of the SCM_CREDENTIALS control
175 message. For more information see unix(7).
176 SO_PEERCRED
178 Return the credentials of the foreign process connected to this
179 socket. This is only possible for connected PF_UNIX stream
180 sockets and PF_UNIX stream and datagram socket pairs created
181 using socketpair(2); see unix(7). The returned credentials are
182 those that were in effect at the time of the call to connect(2)
183 or socketpair(2). Argument is a ucred structure. Only valid as
184 a getsockopt().
185 SO_PRIORITY
187 Set the protocol-defined priority for all packets to be sent on
188 this socket. Linux uses this value to order the networking
189 queues: packets with a higher priority may be processed first
190 depending on the selected device queueing discipline. For
191 ip(7), this also sets the IP type-of-service (TOS) field for
192 outgoing packets. Setting a priority outside the range 0 to 6
193 requires the CAP_NET_ADMIN capability.
194 SO_RCVBUF
196 Sets or gets the maximum socket receive buffer in bytes. The
197 kernel doubles this value (to allow space for bookkeeping over‐
198 head) when it is set using setsockopt(), and this doubled value
199 is returned by getsockopt(). The default value is set by the
200 rmem_default sysctl and the maximum allowed value is set by the
201 rmem_max sysctl. The minimum (doubled) value for this option is
202 256.
203 SO_RCVBUFFORCE (since Linux 2.6.14)
205 Using this socket option, a privileged (CAP_NET_ADMIN) process
206 can perform the same task as SO_RCVBUF, but the rmem_max limit
207 can be overridden.
208 SO_RCVLOWAT and SO_SNDLOWAT
210 Specify the minimum number of bytes in the buffer until the
211 socket layer will pass the data to the protocol (SO_SNDLOWAT) or
212 the user on receiving (SO_RCVLOWAT). These two values are ini‐
213 tialised to 1. SO_SNDLOWAT is not changeable on Linux (setsock‐
214 opt fails with the error ENOPROTOOPT). SO_RCVLOWAT is change‐
215 able only since Linux 2.4. The select(2) and poll(2) system
216 calls currently do not respect the SO_RCVLOWAT setting on Linux,
217 and mark a socket readable when even a single byte of data is
218 available. A subsequent read from the socket will block until
219 SO_RCVLOWAT bytes are available.
220 SO_RCVTIMEO and SO_SNDTIMEO
222 Specify the receiving or sending timeouts until reporting an
223 error. The parameter is a struct timeval. If an input or out‐
224 put function blocks for this period of time, and data has been
225 sent or received, the return value of that function will be the
226 amount of data transferred; if no data has been transferred and
227 the timeout has been reached then -1 is returned with errno set
228 to EAGAIN or EWOULDBLOCK just as if the socket was specified to
229 be nonblocking. If the timeout is set to zero (the default)
230 then the operation will never timeout.
231 SO_REUSEADDR
233 Indicates that the rules used in validating addresses supplied
234 in a bind(2) call should allow reuse of local addresses. For
235 PF_INET sockets this means that a socket may bind, except when
236 there is an active listening socket bound to the address. When
237 the listening socket is bound to INADDR_ANY with a specific port
238 then it is not possible to bind to this port for any local
239 address.
240 SO_SNDBUF
242 Sets or gets the maximum socket send buffer in bytes. The ker‐
243 nel doubles this value (to allow space for bookkeeping overhead)
244 when it is set using setsockopt(), and this doubled value is
245 returned by getsockopt(). The default value is set by the
246 wmem_default sysctl and the maximum allowed value is set by the
247 wmem_max sysctl. The minimum (doubled) value for this option is
248 2048.
249 SO_SNDBUFFORCE (since Linux 2.6.14)
251 Using this socket option, a privileged (CAP_NET_ADMIN) process
252 can perform the same task as SO_SNDBUF, but the wmem_max limit
253 can be overridden.
254 SO_TIMESTAMP
256 Enable or disable the receiving of the SO_TIMESTAMP control mes‐
257 sage. The timestamp control message is sent with level
258 SOL_SOCKET and the cmsg_data field is a struct timeval indicat‐
259 ing the reception time of the last packet passed to the user in
260 this call. See cmsg(3) for details on control messages.
261 SO_TYPE
263 Gets the socket type as an integer (like SOCK_STREAM). Can only
264 be read with getsockopt().
265
267 When writing onto a connection-oriented socket that has been shut down
268 (by the local or the remote end) SIGPIPE is sent to the writing process
269 and EPIPE is returned. The signal is not sent when the write call
270 specified the MSG_NOSIGNAL flag.
271 When requested with the FIOSETOWN fcntl() or SIOCSPGRP ioctl(), SIGIO
273 is sent when an I/O event occurs. It is possible to use poll(2) or
274 select(2) in the signal handler to find out which socket the event
275 occurred on. An alternative (in Linux 2.2) is to set a realtime signal
276 using the F_SETSIG fcntl(); the handler of the real time signal will be
277 called with the file descriptor in the si_fd field of its siginfo_t.
278 See fcntl(2) for more information.
279 Under some circumstances (e.g. multiple processes accessing a single
281 socket), the condition that caused the SIGIO may have already disap‐
282 peared when the process reacts to the signal. If this happens, the
283 process should wait again because Linux will resend the signal later.
284
286 The core socket networking sysctls can be accessed using the
287 /proc/sys/net/core/* files or with the sysctl(2) interface.
288 rmem_default
290 contains the default setting in bytes of the socket receive buf‐
291 fer.
292 rmem_max
294 contains the maximum socket receive buffer size in bytes which a
295 user may set by using the SO_RCVBUF socket option.
296 wmem_default
298 contains the default setting in bytes of the socket send buffer.
299 wmem_max
301 contains the maximum socket send buffer size in bytes which a
302 user may set by using the SO_SNDBUF socket option.
303 message_cost and message_burst
305 configure the token bucket filter used to load limit warning
306 messages caused by external network events.
307 netdev_max_backlog
309 Maximum number of packets in the global input queue.
310 optmem_max
312 Maximum length of ancillary data and user control data like the
313 iovecs per socket.
314
316 These operations can be accessed using ioctl(2):
317 error = ioctl(ip_socket, ioctl_type, &value_result);
319 SIOCGSTAMP
322 Return a struct timeval with the receive timestamp of the last
323 packet passed to the user. This is useful for accurate round
324 trip time measurements. See setitimer(2) for a description of
325 struct timeval. This ioctl should only be used if the socket
326 option SO_TIMESTAMP is not set on the socket. Otherwise, it
327 returns the timestamp of the last packet that was received while
328 SO_TIMESTAMP was not set, or it fails if no such packet has been
329 received, (i.e., ioctl() returns -1 with errno set to ENOENT).
330 SIOCSPGRP
332 Set the process or process group to send SIGIO or SIGURG signals
333 to when an asynchronous I/O operation has finished or urgent
334 data is available. The argument is a pointer to a pid_t. If
335 the argument is positive, send the signals to that process. If
336 the argument is negative, send the signals to the process group
337 with the ID of the absolute value of the argument. The process
338 may only choose itself or its own process group to receive sig‐
339 nals unless it has the CAP_KILL capability or an effective UID
340 of 0.
341 FIOASYNC
343 Change the O_ASYNC flag to enable or disable asynchronous I/O
344 mode of the socket. Asynchronous I/O mode means that the SIGIO
345 signal or the signal set with F_SETSIG is raised when a new I/O
346 event occurs.
347 Argument is an integer boolean flag.
349 SIOCGPGRP
351 Get the current process or process group that receives SIGIO or
352 SIGURG signals, or 0 when none is set.
353 Valid fcntl() operations:
355 FIOGETOWN
357 The same as the SIOCGPGRP ioctl().
358 FIOSETOWN
360 The same as the SIOCSPGRP ioctl().
361
363 Linux assumes that half of the send/receive buffer is used for internal
364 kernel structures; thus the sysctls are twice what can be observed on
365 the wire.
366 Linux will only allow port re-use with the SO_REUSEADDR option when
368 this option was set both in the previous program that performed a
369 bind() to the port and in the program that wants to re-use the port.
370 This differs from some implementations (e.g., FreeBSD) where only the
371 later program needs to set the SO_REUSEADDR option. Typically this
372 difference is invisible, since, for example, a server program is
373 designed to always set this option.
374
376 The CONFIG_FILTER socket options SO_ATTACH_FILTER and SO_DETACH_FILTER
377 are not documented. The suggested interface to use them is via the
378 libpcap library.
379
381 SO_BINDTODEVICE was introduced in Linux 2.0.30. SO_PASSCRED is new in
382 Linux 2.2. The sysctls are new in Linux 2.2. SO_RCVTIMEO and SO_SND‐
383 TIMEO are supported since Linux 2.3.41. Earlier, timeouts were fixed to
384 a protocol specific setting, and could not be read or written.
385
387 getsockopt(2), setsockopt(2), socket(2), capabilities(7), ddp(7),
388 ip(7), packet(7)
389Linux 2.6.6 2004-05-27 SOCKET(7)