Go to main content

man pages section 4: Device and Network Interfaces

Exit Print View

Updated: Thursday, June 13, 2019
 
 

TCP (4P)

Name

tcp, TCP - Internet Transmission Control Protocol

Synopsis

#include <sys/socket.h>
#include <netinet/in.h>
s = socket(AF_INET, SOCK_STREAM, 0);
s = socket(AF_INET6, SOCK_STREAM, 0);
t = t_open("/dev/tcp", O_RDWR);
t = t_open("/dev/tcp6", O_RDWR);

Description

TCP is the virtual circuit protocol of the Internet protocol family. It provides reliable, flow-controlled, in order, two-way transmission of data. It is a byte-stream protocol layered above the Internet Protocol (IP), or the Internet Protocol Version 6 (IPv6), the Internet protocol family's internetwork datagram delivery protocol.

Programs can access TCP using the socket interface as a SOCK_STREAM socket type, or using the Transport Level Interface (TLI) where it supports the connection-oriented (T_COTS_ORD) service type.

TCP uses IP's host-level addressing and adds its own per-host collection of “port addresses.” The endpoints of a TCP connection are identified by the combination of an IP or IPv6 address and a TCP port number. Although other protocols, such as the User Datagram Protocol (UDP), can use the same host and port address format, the port space of these protocols is distinct. See inet(4P) and inet6(4P) for details on the common aspects of addressing in the Internet protocol family.

Sockets utilizing TCP are either “active” or “passive.” Active sockets initiate connections to passive sockets. Both types of sockets must have their local IP or IPv6 address and TCP port number bound with the bind(3C) system call after the socket is created. By default, TCP sockets are active. A passive socket is created by calling the listen(3C) system call after binding the socket with bind(). This establishes a queueing parameter for the passive socket. After this, connections to the passive socket can be received with the accept(3C) system call. Active sockets use the connect(3C) call after binding to initiate connections.

By using the special value INADDR_ANY with IP, or the unspecified address (all zeroes) with IPv6, the local IP address can be left unspecified in the bind() call by either active or passive TCP sockets. This feature is usually used if the local address is either unknown or irrelevant. If left unspecified, the local IP or IPv6 address is bound at connection time to the address of the network interface used to service the connection.

No two TCP sockets can be bound to the same port unless the bound IP addresses are different. This behavior can be changed by using the SO_REUSEPORT option. If both the binding and existing bound sockets have this option enabled, and the user IDs of both sockets (at bind() calling time) are the same, then such bind() is allowed. But only one of the two sockets can become a listener socket.

When comparing addresses at bind() time, IPv4 INADDR_ANY and IPv6 unspecified addresses compare as equal to any IPv4 or IPv6 address. For example, if a socket is bound to INADDR_ANY or unspecified address and port X, no other socket can bind to port X, regardless of the binding address. This special consideration of INADDR_ANY and unspecified address can be changed using the socket option SO_REUSEADDR . If SO_REUSEADDR is set on a socket doing a bind, IPv4 INADDR_ANY and IPv6 unspecified address do not compare as equal to any IP address. This means that as long as the two sockets are not both bound to INADDR_ANY/unspecified address or the same IP address, the two sockets can be bound to the same port.

If an application does not want to allow another socket using the SO_REUSEADDR/SO_REUSEPORT option to bind to a port its socket is bound to, the application can set the socket level option SO_EXCLBIND on a socket. The option values of 0 and 1 mean enabling and disabling the option respectively. Once this option is enabled on a socket, no other socket can be bound to the same port.

Once a connection has been established, data can be exchanged using the read(2) and write(2) system calls.

Under most circumstances, TCP sends data when it is presented. When outstanding data has not yet been acknowledged, TCP gathers small amounts of output to be sent in a single packet once an acknowledgement has been received. For a small number of clients, such as window systems that send a stream of mouse events which receive no replies, this packetization can cause significant delays. To circumvent this problem, TCP provides a socket-level boolean option, TCP_NODELAY. TCP_NODELAY is defined in <netinet/tcp.h>, and is set with setsockopt(3C) and tested with getsockopt(3C). The option level for the setsockopt() call is the protocol number for TCP, available from getprotobyname(3C).

For some applications, it can be desirable for TCP not to send out data unless a full TCP segment can be sent. To enable this behavior, an application can use the TCP_CORK socket option. When TCP_CORK is set with a non-zero value, TCP sends out a full TCP segment only. When TCP_CORK is set to zero after it has been enabled, all buffered data is sent out (as permitted by the peer's receive window and the current congestion window). TCP_CORK is defined in <netinet/tcp.h>, and is set with setsockopt(3C) and tested with getsockopt(3C). The option level for the setsockopt() call is the protocol number for TCP, available from getprotobyname(3C).

The TCP_MAXSEG socket option can be used to determine the TCP maximum segment size (MSS) that will be used by a socket. When set after a connection has been completed, the TCP_MAXSEG socket option will limit the sizes of the segments being sent. Only reductions in the segment size are allowed; this should never increase the size of segments.

When set before a connection is begun (by either the listen(3C) or the connect(3C) call) the option will additionally specify the size sent in the outgoing TCP MSS option on the SYN segment; this can be useful in dealing with networks that impose unusual restrictions on packet size. If the user-specified value is larger than the value that would have been sent otherwise, the smaller value is used.

When retrieved before a connection is completed, the TCP_MAXSEG socket option will return the default segment size that will be used if the peer does not send an MSS option (536 for IPv4, 1220 for IPv6). When retrieved after the connection is completed, the value returned is the current maximum segment size used by the stack. This may vary over time, due to Path MTU Discovery, but will never exceed any user-specified TCP_MAXSEG value.

TCP_MAXSEG is defined in netinet/tcp.h, and is set with setsockopt(3C) and retrieved with getsockopt(3C). The option level for the setsockopt() call is the protocol number for TCP, available from getprotobyname(3C). The option value is an int.

The TCP_MD5SIG socket option can be set with setsockopt(3C) to sign a TCP segment using MD5 hash as described in RFC 2385. Setting this option has an adverse effect on performance. This option is turned off by default. Setting TCP_MD5SIG option when no SAs have been configured is not supported, currently a connection request will proceed without MD5 option but an incoming connection will not be accepted. TCP_MD5SIG option is applied only if set before initiating or accepting a connection.

Another socket level option, SO_RCVBUF, can be used to control the window that TCP advertises to the peer. IP level options can also be used with TCP. See ip(4P) and ip6(4P).

TCP provides an urgent data mechanism, which can be invoked using the out-of-band provisions of send(3C). The caller can mark one byte as “urgent” with the MSG_OOB flag to send(3C). This sets an “urgent pointer” pointing to this byte in the TCP stream. The receiver on the other side of the stream is notified of the urgent data by a SIGURG signal. The SIOCATMARK ioctl(2) request returns a value indicating whether the stream is at the urgent mark. Because the system never returns data across the urgent mark in a single read(2) call, it is possible to advance to the urgent data in a simple loop which reads data, testing the socket with the SIOCATMARK ioctl() request, until it reaches the mark.

Incoming connection requests that include an IP source route option are noted, and the reverse source route is used in responding.

A checksum over all data helps TCP implement reliability. Using a window-based flow control mechanism that makes use of positive acknowledgements, sequence numbers, and a retransmission strategy, TCP can usually recover when datagrams are damaged, delayed, duplicated or delivered out of order by the underlying communication medium.

If the local TCP receives no acknowledgements from its peer for a period of time, (for example, if the remote machine crashes), the connection is closed and an error is returned.

The TCP level socket options, TCP_CONN_ABORT_THRESHOLD and TCP_ABORT_THRESHOLD can be used to change and retrieve this period of time. The option value is uint32_t and the unit is millisecond. TCP_CONN_ABORT_THRESHOLD and TCP_ABORT_THRESHOLD control respectively this period before and after a connection is established. If the application does not want TCP to time out, it can use the option value 0.

During this period, TCP tries to retransmit the unacknowledged data multiple times, each after a timeout. And the timeout interval is exponentially backed off. The TCP level socket options, TCP_RTO_INITIAL, TCP_RTO_MIN, and TCP_RTO_MAX can be used to control the timeout interval. TCP_RTO_INITIAL controls the initial retransmission timeout period. TCP_RTO_MIN and TCP_RTO_MAX control the minimum and maximum timeout period respectively. The option value is an uint32_t and the unit is millisecond.

The default values of the above options, TCP_CONN_ABORT_THRESHOLD, TCP_ABORT_THRESHOLD, TCP_RTO_MIN, TCP_RTO_MAX, and TCP_RTO_INITIAL are appropriate for most situations. An application should only alter their values in special circumstances and when it has detailed knowledge of the network environment.

TCP follows the congestion control algorithm described in RFC 2581, and also supports the initial congestion window (cwnd) changes in RFC 3390. The initial cwnd calculation can be overridden by the socket option TCP_INIT_CWND. An application can use this option to set the initial cwnd to a specified number of TCP segments. This applies to the cases when the connection first starts and restarts after an idle period. The process must have the PRIV_SYS_NET_CONFIG privilege if it wants to specify a number greater than that calculated by RFC 3390.

The TCP_INFO option can be used to collect various information about the current state of a TCP socket, such as connection state, windows sizes, and so forth. The data structure used as an argument is struct tcp_info.

The TCP_CONGESTION option can be used to get or set a socket's congestion control algorithm. Its argument is a pointer to a null-terminated string.

Oracle Solaris OS supports TCP Extensions for High Performance (RFC 1323) which includes the window scale and time stamp options, and Protection Against Wrap Around Sequence Numbers (PAWS). Oracle Solaris OS also supports Selective Acknowledgment (SACK) capabilities (RFC 2018) and Explicit Congestion Notification (ECN) mechanism (RFC 3168).

Turn on the window scale option in one of the following ways:

  • An application can set SO_SNDBUF or SO_RCVBUF size in the setsockopt() option to be larger than 64K. This must be done before the program calls listen() or connect(), because the window scale option is negotiated when the connection is established. Once the connection has been made, it is too late to increase the send or receive window beyond the default TCP limit of 64K.

  • For all applications, use ndd(8) to modify the configuration parameter tcp_wscale_always. If tcp_wscale_always is set to 1, the window scale option is always set when connecting to a remote system. If tcp_wscale_always is 0, the window scale option is set only if the user has requested a send or receive window larger than 64K. The default value of tcp_wscale_always is 1.

  • Regardless of the value of tcp_wscale_always, the window scale option is always included in a connect acknowledgement if the connecting system has used the option.

Turn on SACK capabilities in the following way:

  • Use ndd to modify the configuration parameter tcp_sack_permitted. If tcp_sack_permitted is set to 0, TCP does not accept SACK or send out SACK information. If tcp_sack_permitted is set to 1, TCP does not initiate a connection with SACK permitted option in the SYN segment, but does respond with SACK permitted option in the SYN|ACK segment if an incoming connection request has the SACK permitted option. This means that TCP only accepts SACK information if the other side of the connection also accepts SACK information. If tcp_sack_permitted is set to 2, it both initiates and accepts connections with SACK information. The default for tcp_sack_permitted is 2 (active enabled).

Turn on TCP ECN mechanism in the following way:

  • Use ndd to modify the configuration parameter tcp_ecn_permitted. If tcp_ecn_permitted is set to 0, TCP does not negotiate with a peer that supports ECN mechanism. If tcp_ecn_permitted is set to 1 when initiating a connection, TCP does not tell a peer that it supports ECN mechanism. However, it tells a peer that it supports ECN mechanism when accepting a new incoming connection request if the peer indicates that it supports ECN mechanism in the SYN segment. If tcp_ecn_permitted is set to 2, in addition to negotiating with a peer on ECN mechanism when accepting connections, TCP indicates in the outgoing SYN segment that it supports ECN mechanism when TCP makes active outgoing connections. The default for tcp_ecn_permitted is 1.

Turn on the time stamp option in the following way:

  • Use ndd to modify the configuration parameter tcp_tstamp_always. If tcp_tstamp_always is 1, the time stamp option is always be set when connecting to a remote machine. If tcp_tstamp_always is 0, the timestamp option is not be set when connecting to a remote system. The default for tcp_tstamp_always is 0.

  • Regardless of the value of tcp_tstamp_always, the time stamp option is always included in a connect acknowledgement (and all succeeding packets) if the connecting system has used the time stamp option.

Use the following procedure to turn on the time stamp option only when the window scale option is in effect:

  • Use ndd to modify the configuration parameter tcp_tstamp_if_wscale. Setting tcp_tstamp_if_wscale to 1 causes the time stamp option to be set when connecting to a remote system, if the window scale option has been set. If tcp_tstamp_if_wscale is 0, the time stamp option is not set when connecting to a remote system. The default for tcp_tstamp_if_wscale is 1.

Protection Against Wrap Around Sequence Numbers (PAWS) is always used when the time stamp option is set.

Oracle Solaris OS also supports multiple methods of generating initial sequence numbers. One of these methods is the improved technique suggested in RFC 1948. We HIGHLY recommend that you set sequence number generation parameters as close to boot time as possible. This prevents sequence number problems on connections that use the same connection-ID as ones that used a different sequence number generation. The svc:/network/initial:default service configures the initial sequence number generation. The service reads the value contained in the configuration file /etc/default/inetinit to determine which method to use.

The /etc/default/inetinit file is an unstable interface, and can change in future releases.

TCP can be configured to report some information on connections that terminate by means of an RST packet. By default, no logging is done. If the ndd(8) parameter tcp_trace is set to 1, then trace data is collected for all new connections established after that time.

The trace data consists of the TCP headers and IP source and destination addresses of the last few packets sent in each direction before RST occurred. Those packets are logged in a series of strlog(9F) calls. This trace facility has a very low overhead, and so is superior to such utilities as snoop(8) for non-intrusive debugging for connections terminating by means of an RST.

Oracle Solaris OS supports the keep-alive mechanism described in RFC 1122. It is enabled using the socket option SO_KEEPALIVE. When enabled, there are two keep-alive machanisms.

  • By default, the first keep-alive probe is sent out after a TCP is idle for two hours. If the peer does not respond to the probe within eight minutes, the TCP connection is aborted. You can alter the interval for sending out the first probe using the socket option TCP_KEEPALIVE_THRESHOLD in milliseconds or TCP_KEEPIDLE in seconds.

    The system default is controlled by the TCP ndd parameter tcp_keepalive_interval. The minimum value is ten seconds. The maximum is ten days, while the default is two hours. If you receive no response to the probe, you can use the TCP_KEEPALIVE_ABORT_THRESHOLD socket option to change the time threshold for aborting a TCP connection.

    The option value is an unsigned integer in milliseconds. The value zero indicates that TCP should never time out and abort the connection when probing. The system default is controlled by the TCP ndd parameter tcp_keepalive_abort_interval. The default is eight minutes.

  • The second implementation is activated if socket option TCP_KEEPINTVL and/or TCP_KEEPCNT are set. The time between each consequent probes is set by TCP_KEEPINTVL in seconds. The minimum value is ten seconds. The maximum is ten days, while the default is two hours. The TCP connection will be aborted after certain amount of probes, which is set by TCP_KEEPCNT, without receiving response.

After an application closes a TCP connection, TCP enters the shutdown sequence. But if the peer does not respond (it crashes), the connection is stuck in this state (FIN-WAIT-2). To prevent this, Oracle Solaris OS starts a timer when TCP enters this state. If the timer fires and the shutdown sequence has not completed, the connection is freed. The socket option TCP_LINGER2 can be used to change and retrieve this timeout period. The option value is an int and the unit is second. The option value cannot be set higher than the system default value, which is controlled by the TCP private parameter tcp_fin_wait_2_flush_interval. The default value is appropriate for most situations. An application should only change the value in some special circumstances and when it has detailed knowledge of the network environment.

See Also

svcs(1), ioctl(2), read(2), write(2), accept(3C), bind(3C), connect(3C), getprotobyname(3C), getsockopt(3C), listen(3C), send(3C), inet(4P), inet6(4P), ip(4P), ip6(4P), smf(7), ndd(8), svcadm(8)

Ramakrishnan, K., Floyd, S., Black, D., RFC 3168, The Addition of Explicit Congestion Notification (ECN) to IP, September 2001.

Mathis, M. and Mahdavi, J. Pittsburgh Supercomputing Center; Floyd, S. Lawrence Berkeley National Laboratory; Romanow, A. Sun Microsystems, Inc. RFC 2018, TCP Selective Acknowledgment Options, October 1996.

Bellovin, S., RFC 1948, Defending Against Sequence Number Attacks, May 1996.

Jacobson, V., Braden, R., and Borman, D., RFC 1323, TCP Extensions for High Performance, May 1992.

Postel, Jon, RFC 793, Transmission Control Protocol - DARPA Internet Program Protocol Specification, Network Information Center, SRI International, Menlo Park, CA., September 1981.

Diagnostics

A socket operation can fail if:

EISCONN

A connect() operation was attempted on a socket on which a connect() operation had already been performed.

ETIMEDOUT

A connection was dropped due to excessive retransmissions.

ECONNRESET

The remote peer forced the connection to be closed (usually because the remote machine has lost state information about the connection due to a crash).

ECONNREFUSED

The remote peer actively refused connection establishment (usually because no process is listening to the port).

EADDRINUSE

A bind() operation was attempted on a socket with a network address/port pair that has already been bound to another socket.

EADDRNOTAVAIL

A bind() operation was attempted on a socket with a network address for which no network interface exists.

EACCES

A bind() operation was attempted with a “reserved” port number and the effective user ID of the process was not the privileged user.

ENOBUFS

The system ran out of memory for internal data structures.

Notes

The tcp service is managed by the service management facility, smf(7), under the service identifier:

svc:/network/initial:default

Administrative actions on this service, such as enabling, disabling, or requesting restart, can be performed using svcadm(8). The service's status can be queried using the svcs(1) command.