1.\" Copyright (c) 1983, 1991, 1993 2.\" The Regents of the University of California. All rights reserved. 3.\" 4.\" Redistribution and use in source and binary forms, with or without 5.\" modification, are permitted provided that the following conditions 6.\" are met: 7.\" 1. Redistributions of source code must retain the above copyright 8.\" notice, this list of conditions and the following disclaimer. 9.\" 2. Redistributions in binary form must reproduce the above copyright 10.\" notice, this list of conditions and the following disclaimer in the 11.\" documentation and/or other materials provided with the distribution. 12.\" 3. All advertising materials mentioning features or use of this software 13.\" must display the following acknowledgement: 14.\" This product includes software developed by the University of 15.\" California, Berkeley and its contributors. 16.\" 4. Neither the name of the University nor the names of its contributors 17.\" may be used to endorse or promote products derived from this software 18.\" without specific prior written permission. 19.\" 20.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 21.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23.\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 24.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30.\" SUCH DAMAGE. 31.\" 32.\" From: @(#)tcp.4 8.1 (Berkeley) 6/5/93 33.\" $FreeBSD$ 34.\" 35.Dd February 14, 1995 36.Dt TCP 4 37.Os 38.Sh NAME 39.Nm tcp 40.Nd Internet Transmission Control Protocol 41.Sh SYNOPSIS 42.In sys/types.h 43.In sys/socket.h 44.In netinet/in.h 45.Ft int 46.Fn socket AF_INET SOCK_STREAM 0 47.Sh DESCRIPTION 48The 49.Tn TCP 50protocol provides reliable, flow-controlled, two-way 51transmission of data. It is a byte-stream protocol used to 52support the 53.Dv SOCK_STREAM 54abstraction. TCP uses the standard 55Internet address format and, in addition, provides a per-host 56collection of 57.Dq port addresses . 58Thus, each address is composed 59of an Internet address specifying the host and network, with 60a specific 61.Tn TCP 62port on the host identifying the peer entity. 63.Pp 64Sockets utilizing the tcp protocol are either 65.Dq active 66or 67.Dq passive . 68Active sockets initiate connections to passive 69sockets. By default 70.Tn TCP 71sockets are created active; to create a 72passive socket the 73.Xr listen 2 74system call must be used 75after binding the socket with the 76.Xr bind 2 77system call. Only 78passive sockets may use the 79.Xr accept 2 80call to accept incoming connections. Only active sockets may 81use the 82.Xr connect 2 83call to initiate connections. 84.Tn TCP 85also supports a more datagram-like mode, called Transaction 86.Tn TCP , 87which is described in 88.Xr ttcp 4 . 89.Pp 90Passive sockets may 91.Dq underspecify 92their location to match 93incoming connection requests from multiple networks. This 94technique, termed 95.Dq wildcard addressing , 96allows a single 97server to provide service to clients on multiple networks. 98To create a socket which listens on all networks, the Internet 99address 100.Dv INADDR_ANY 101must be bound. The 102.Tn TCP 103port may still be specified 104at this time; if the port is not specified the system will assign one. 105Once a connection has been established the socket's address is 106fixed by the peer entity's location. The address assigned the 107socket is the address associated with the network interface 108through which packets are being transmitted and received. Normally 109this address corresponds to the peer entity's network. 110.Pp 111.Tn TCP 112supports a number of socket options which can be set with 113.Xr setsockopt 2 114and tested with 115.Xr getsockopt 2 : 116.Bl -tag -width TCP_NODELAYx 117.It Dv TCP_NODELAY 118Under most circumstances, 119.Tn TCP 120sends data when it is presented; 121when outstanding data has not yet been acknowledged, it gathers 122small amounts of output to be sent in a single packet once 123an acknowledgement is received. 124For a small number of clients, such as window systems 125that send a stream of mouse events which receive no replies, 126this packetization may cause significant delays. 127The boolean option 128.Dv TCP_NODELAY 129defeats this algorithm. 130.It Dv TCP_MAXSEG 131By default, a sender\- and receiver-TCP 132will negotiate among themselves to determine the maximum segment size 133to be used for each connection. The 134.Dv TCP_MAXSEG 135option allows the user to determine the result of this negotiation, 136and to reduce it if desired. 137.It Dv TCP_NOOPT 138.Tn TCP 139usually sends a number of options in each packet, corresponding to 140various 141.Tn TCP 142extensions which are provided in this implementation. The boolean 143option 144.Dv TCP_NOOPT 145is provided to disable 146.Tn TCP 147option use on a per-connection basis. 148.It Dv TCP_NOPUSH 149By convention, the sender-TCP 150will set the 151.Dq push 152bit and begin transmission immediately (if permitted) at the end of 153every user call to 154.Xr write 2 155or 156.Xr writev 2 . 157The 158.Dv TCP_NOPUSH 159option is provided to allow servers to easily make use of Transaction 160TCP (see 161.Xr ttcp 4 ) . 162When the option is set to a non-zero value, 163.Tn TCP 164will delay sending any data at all until either the socket is closed, 165or the internal send buffer is filled. 166.El 167.Pp 168The option level for the 169.Xr setsockopt 2 170call is the protocol number for 171.Tn TCP , 172available from 173.Xr getprotobyname 3 , 174or 175.Dv IPPROTO_TCP . 176All options are declared in 177.Aq Pa netinet/tcp.h . 178.Pp 179Options at the 180.Tn IP 181transport level may be used with 182.Tn TCP ; 183see 184.Xr ip 4 . 185Incoming connection requests that are source-routed are noted, 186and the reverse source route is used in responding. 187.Sh MIB VARIABLES 188The 189.Nm 190protocol implements a number of variables in the 191.Li net.inet 192branch of the 193.Xr sysctl 3 194MIB. 195.Bl -tag -width TCPCTL_DO_RFC1644 196.It Dv TCPCTL_DO_RFC1323 197.Pq tcp.rfc1323 198Implement the window scaling and timestamp options of RFC 1323 199(default true). 200.It Dv TCPCTL_DO_RFC1644 201.Pq tcp.rfc1644 202Implement Transaction 203.Tn TCP , 204as described in RFC 1644. 205.It Dv TCPCTL_MSSDFLT 206.Pq tcp.mssdflt 207The default value used for the maximum segment size 208.Pq Dq MSS 209when no advice to the contrary is received from MSS negotiation. 210.It Dv TCPCTL_SENDSPACE 211.Pq tcp.sendspace 212Maximum TCP send window. 213.It Dv TCPCTL_RECVSPACE 214.Pq tcp.recvspace 215Maximum TCP receive window. 216.It tcp.log_in_vain 217Log any connection attempts to ports where there is not a socket 218accepting connections. 219.It tcp.slowstart_flightsize 220The number of packets allowed to be in-flight during the 221.Tn TCP 222slow-start phase on a non-local network. 223.It tcp.local_slowstart_flightsize 224The number of packets allowed to be in-flight during the 225.Tn TCP 226slow-start phase to local machines in the same subnet. 227.It tcp.msl 228The Maximum Segment Lifetime, in milliseconds, for a packet. 229.It tcp.keepinit 230Timeout, in milliseconds, for new, non-established TCP connections. 231.It tcp.keepidle 232Amount of time, in milliseconds, that the connection must be idle 233before keepalive probes (if enabled) are sent. 234.It tcp.keepintvl 235The interval, in milliseconds, between keepalive probes sent to remote 236machines. 237After 238.Dv TCPTV_KEEPCNT 239(default 8) probes are sent, with no response, the connection is dropped. 240.It tcp.always_keepalive 241Assume that 242.Dv SO_KEEPALIVE 243is set on all 244.Tn TCP 245connections, the kernel will 246periodically send a packet to the remote host to verify the connection 247is still up. 248.It tcp.icmp_may_rst 249Certain 250.Tn ICMP 251unreachable messages may abort connections in 252.Tn SYN-SENT 253state. 254.It tcp.do_tcpdrain 255Flush packets in the 256.Tn TCP 257reassembly queue if the system is low on mbufs. 258.\" 259.\" This option should go away and become the default. 260.\" 261.It tcp.tcp_lq_overflow 262If a connection is dropped due to a listen queue overflow, delete the 263cloned route associated with the connection if it does not have any 264prior information. 265.It tcp.blackhole 266If enabled, disable sending of RST when a connection is attempted 267to a port where there is not a socket accepting connections. 268See 269.Xr blackhole 4 . 270.It tcp.delayed_ack 271Delay ACK to try and piggyback it onto a data packet. 272.It tcp.delacktime 273Maximum amount of time, in milliseconds, before a delayed ACK is sent. 274.It tcp.newreno 275Enable TCP NewReno Fast Recovery algorithm, 276as described in RFC 2582. 277.It tcp.path_mtu_discovery 278Enable Path MTU Discovery 279.It tcp.tcbhashsize 280Size of the 281.Tn TCP 282control-block hashtable 283(read-only). 284This may be tuned using the kernel option 285.Dv TCBHASHSIZE 286or by setting 287.Va net.inet.tcp.tcbhashsize 288in the 289.Xr loader 8 . 290.It tcp.pcbcount 291Number of active process control blocks 292(read-only). 293.It tcp.strict_rfc1948 294Enable strict RFC 1948 (Defending Against Sequence Number Attacks) 295compliance. 296Setting this variable to a non-zero value will disable reseeding and 297will disable the use of randomized initial sequence numbers in favor 298of MD5-generated initial sequence numbers. 299.It tcp.isn_reseed_interval 300The interval (in seconds) specifying how often the secret data used in 301RFC 1948 initial sequence number calculations should be reseeded. 302By default, this variable is set to zero, indicating that 303no reseeding will occur. 304Reseeding should not be necessary, and will break 305.Dv TIME_WAIT 306recycling for a few minutes. 307.El 308.Sh ERRORS 309A socket operation may fail with one of the following errors returned: 310.Bl -tag -width Er 311.It Bq Er EISCONN 312when trying to establish a connection on a socket which 313already has one; 314.It Bq Er ENOBUFS 315when the system runs out of memory for 316an internal data structure; 317.It Bq Er ETIMEDOUT 318when a connection was dropped 319due to excessive retransmissions; 320.It Bq Er ECONNRESET 321when the remote peer 322forces the connection to be closed; 323.It Bq Er ECONNREFUSED 324when the remote 325peer actively refuses connection establishment (usually because 326no process is listening to the port); 327.It Bq Er EADDRINUSE 328when an attempt 329is made to create a socket with a port which has already been 330allocated; 331.It Bq Er EADDRNOTAVAIL 332when an attempt is made to create a 333socket with a network address for which no network interface 334exists. 335.It Bq Er EAFNOSUPPORT 336when an attempt is made to bind or connect a socket to a multicast 337address. 338.El 339.Sh SEE ALSO 340.Xr getsockopt 2 , 341.Xr socket 2 , 342.Xr sysctl 3 , 343.Xr blackhole 4 , 344.Xr inet 4 , 345.Xr intro 4 , 346.Xr ip 4 , 347.Xr ttcp 4 348.Rs 349.%A V. Jacobson 350.%A R. Braden 351.%A D. Borman 352.%T "TCP Extensions for High Performance" 353.%O RFC 1323 354.Re 355.Rs 356.%A R. Braden 357.%T "T/TCP \- TCP Extensions for Transactions" 358.%O RFC 1644 359.Re 360.Sh HISTORY 361The 362.Nm 363protocol appeared in 364.Bx 4.2 . 365The RFC 1323 extensions for window scaling and timestamps were added 366in 367.Bx 4.4 . 368