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: src/share/man/man4/tcp.4,v 1.11.2.14 2002/12/29 16:35:38 schweikh Exp $ 34.\" $DragonFly: src/share/man/man4/tcp.4,v 1.9 2008/10/17 11:30:24 swildner Exp $ 35.\" 36.Dd February 14, 1995 37.Dt TCP 4 38.Os 39.Sh NAME 40.Nm tcp 41.Nd Internet Transmission Control Protocol 42.Sh SYNOPSIS 43.In sys/types.h 44.In sys/socket.h 45.In netinet/in.h 46.Ft int 47.Fn socket AF_INET SOCK_STREAM 0 48.Sh DESCRIPTION 49The 50.Tn TCP 51protocol provides reliable, flow-controlled, two-way 52transmission of data. It is a byte-stream protocol used to 53support the 54.Dv SOCK_STREAM 55abstraction. TCP uses the standard 56Internet address format and, in addition, provides a per-host 57collection of 58.Dq port addresses . 59Thus, each address is composed 60of an Internet address specifying the host and network, with 61a specific 62.Tn TCP 63port on the host identifying the peer entity. 64.Pp 65Sockets utilizing the tcp protocol are either 66.Dq active 67or 68.Dq passive . 69Active sockets initiate connections to passive 70sockets. By default 71.Tn TCP 72sockets are created active; to create a 73passive socket the 74.Xr listen 2 75system call must be used 76after binding the socket with the 77.Xr bind 2 78system call. Only 79passive sockets may use the 80.Xr accept 2 81call to accept incoming connections. Only active sockets may 82use the 83.Xr connect 2 84call to initiate connections. 85.Pp 86Passive sockets may 87.Dq underspecify 88their location to match 89incoming connection requests from multiple networks. This 90technique, termed 91.Dq wildcard addressing , 92allows a single 93server to provide service to clients on multiple networks. 94To create a socket which listens on all networks, the Internet 95address 96.Dv INADDR_ANY 97must be bound. The 98.Tn TCP 99port may still be specified 100at this time; if the port is not specified the system will assign one. 101Once a connection has been established the socket's address is 102fixed by the peer entity's location. The address assigned the 103socket is the address associated with the network interface 104through which packets are being transmitted and received. Normally 105this address corresponds to the peer entity's network. 106.Pp 107.Tn TCP 108supports a number of socket options which can be set with 109.Xr setsockopt 2 110and tested with 111.Xr getsockopt 2 : 112.Bl -tag -width TCP_NODELAYx 113.It Dv TCP_NODELAY 114Under most circumstances, 115.Tn TCP 116sends data when it is presented; 117when outstanding data has not yet been acknowledged, it gathers 118small amounts of output to be sent in a single packet once 119an acknowledgement is received. 120For a small number of clients, such as window systems 121that send a stream of mouse events which receive no replies, 122this packetization may cause significant delays. 123The boolean option 124.Dv TCP_NODELAY 125defeats this algorithm. 126.It Dv TCP_MAXSEG 127By default, a sender\- and receiver-TCP 128will negotiate among themselves to determine the maximum segment size 129to be used for each connection. The 130.Dv TCP_MAXSEG 131option allows the user to determine the result of this negotiation, 132and to reduce it if desired. 133.It Dv TCP_NOOPT 134.Tn TCP 135usually sends a number of options in each packet, corresponding to 136various 137.Tn TCP 138extensions which are provided in this implementation. The boolean 139option 140.Dv TCP_NOOPT 141is provided to disable 142.Tn TCP 143option use on a per-connection basis. 144.It Dv TCP_NOPUSH 145By convention, the sender-TCP 146will set the 147.Dq push 148bit and begin transmission immediately (if permitted) at the end of 149every user call to 150.Xr write 2 151or 152.Xr writev 2 . 153When the 154.Dv TCP_NOPUSH 155option is set to a non-zero value, 156.Tn TCP 157will delay sending any data at all until either the socket is closed, 158or the internal send buffer is filled. 159.It Dv TCP_SIGNATURE_ENABLE 160This option enables the use of MD5 digests (also known as TCP-MD5) 161on writes to the specified socket. 162In the current release, only outgoing traffic is digested; 163digests on incoming traffic are not verified. 164The current default behavior for the system is to respond to a system 165advertising this option with TCP-MD5; this may change. 166.Pp 167One common use for this in a DragonFlyBSD router deployment is to enable 168based routers to interwork with Cisco equipment at peering points. 169Support for this feature conforms to RFC 2385. 170Only IPv4 (AF_INET) sessions are supported. 171.Pp 172In order for this option to function correctly, it is necessary for the 173administrator to add a tcp-md5 key entry to the system's security 174associations database (SADB) using the 175.Xr setkey 8 176utility. 177This entry must have an SPI of 0x1000 and can therefore only be specified 178on a per-host basis at this time. 179.Pp 180If an SADB entry cannot be found for the destination, the outgoing traffic 181will have an invalid digest option prepended, and the following error message 182will be visible on the system console: 183.Em "tcpsignature_compute: SADB lookup failed for %d.%d.%d.%d" . 184.It Dv TCP_KEEPINIT 185If a 186.Tn TCP 187connection cannot be established within a period of time, 188.Tn TCP 189will time out the connection attempt. 190The 191.Dv TCP_KEEPINIT 192option specifies the number of milliseconds to wait 193before the connection attempt times out. 194The default value for 195.Dv TCP_KEEPINIT 196is tcp.keepinit milliseconds. 197For the accepted sockets, the 198.Dv TCP_KEEPINIT 199option value is inherited from the listening socket. 200.It Dv TCP_KEEPIDLE 201When the 202.Dv SO_KEEPALIVE 203option is enabled, 204.Tn TCP 205sends a keepalive probe to the remote system of a connection 206that has been idle for a period of time. 207The 208.Dv TCP_KEEPIDLE 209specifies the number of milliseconds before 210.Tn TCP 211will send the initial keepalive probe. 212The default value for 213.Dv TCP_KEEPIDLE 214is tcp.keepidle milliseconds. 215For the accepted sockets, 216the 217.Dv TCP_KEEPIDLE 218option value is inherited from the listening socket. 219.It Dv TCP_KEEPINTVL 220When the 221.Dv SO_KEEPALIVE 222option is enabled, 223.Tn TCP 224sends a keepalive probe to the remote system of a connection 225that has been idle for a period of time. 226The 227.Dv TCP_KEEPINTVL 228option specifies the number of milliseconds to wait 229before retransmitting a keepalive probe. 230The default value for 231.Dv TCP_KEEPINTVL 232is tcp.keepintvl milliseconds. 233For the accepted sockets, 234the 235.Dv TCP_KEEPINTVL 236option value is inherited from the listening socket. 237.It Dv TCP_KEEPCNT 238When the 239.Dv SO_KEEPALIVE 240option is enabled, 241.Tn TCP 242sends a keepalive probe to the remote system of a connection 243that has been idle for a period of time. 244The 245.Dv TCP_KEEPCNT 246option specifies the maximum number of keepalive 247probes to be sent before dropping the connection. 248The default value for 249.Dv TCP_KEEPCNT 250is tcp.keepcnt milliseconds. 251For the accepted sockets, 252the 253.Dv TCP_KEEPCNT 254option value is inherited from the listening socket. 255.El 256.Pp 257The option level for the 258.Xr setsockopt 2 259call is the protocol number for 260.Tn TCP , 261available from 262.Xr getprotobyname 3 , 263or 264.Dv IPPROTO_TCP . 265All options are declared in 266.In netinet/tcp.h . 267.Pp 268Options at the 269.Tn IP 270transport level may be used with 271.Tn TCP ; 272see 273.Xr ip 4 . 274Incoming connection requests that are source-routed are noted, 275and the reverse source route is used in responding. 276.Sh MIB VARIABLES 277The 278.Nm 279protocol implements a number of variables in the 280.Li net.inet 281branch of the 282.Xr sysctl 3 283MIB. 284.Bl -tag -width TCPCTL_DO_RFC1644 285.It Dv TCPCTL_DO_RFC1323 286.Pq tcp.rfc1323 287Implement the window scaling and timestamp options of RFC 1323 288(default true). 289.It Dv TCPCTL_MSSDFLT 290.Pq tcp.mssdflt 291The default value used for the maximum segment size 292.Pq Dq MSS 293when no advice to the contrary is received from MSS negotiation. 294.It Dv TCPCTL_SENDSPACE 295.Pq tcp.sendspace 296Maximum TCP send window. 297.It Dv TCPCTL_RECVSPACE 298.Pq tcp.recvspace 299Maximum TCP receive window. 300.It tcp.log_in_vain 301Log any connection attempts to ports where there is not a socket 302accepting connections. 303The value of 1 limits the logging to SYN (connection establishment) 304packets only. 305That of 2 results in any TCP packets to closed ports being logged. 306Any value unlisted above disables the logging 307(default is 0, i.e., the logging is disabled). 308.It tcp.msl 309The Maximum Segment Lifetime for a packet. 310.It tcp.keepinit 311Timeout for new, non-established TCP connections. 312.It tcp.keepidle 313Amount of time the connection should be idle before keepalive 314probes (if enabled) are sent. 315.It tcp.keepintvl 316The interval between keepalive probes sent to remote machines. 317After 318tcp.keepcnt 319(default 8) probes are sent, with no response, the connection is dropped. 320.It tcp.keepcnt 321The maximum number of keepalive probes to be sent 322before dropping the connection. 323.It tcp.always_keepalive 324Assume that 325.Dv SO_KEEPALIVE 326is set on all 327.Tn TCP 328connections, the kernel will 329periodically send a packet to the remote host to verify the connection 330is still up. 331.It tcp.icmp_may_rst 332Certain 333.Tn ICMP 334unreachable messages may abort connections in 335.Tn SYN-SENT 336state. 337.It tcp.do_tcpdrain 338Flush packets in the 339.Tn TCP 340reassembly queue if the system is low on mbufs. 341.It tcp.blackhole 342If enabled, disable sending of RST when a connection is attempted 343to a port where there is not a socket accepting connections. 344See 345.Xr blackhole 4 . 346.It tcp.delayed_ack 347Delay ACK to try and piggyback it onto a data packet. 348.It tcp.delacktime 349Maximum amount of time before a delayed ACK is sent. 350.It tcp.newreno 351Enable TCP NewReno Fast Recovery algorithm, 352as described in RFC 2582. 353.It tcp.path_mtu_discovery 354Enables Path MTU Discovery. PMTU Discovery is helpful for avoiding 355IP fragmentation when tranferring lots of data to the same client. 356For web servers, where most of the connections are short and to 357different clients, PMTU Discovery actually hurts performance due 358to unnecessary retransmissions. Turn this on only if most of your 359TCP connections are long transfers or are repeatedly to the same 360set of clients. 361.It tcp.tcbhashsize 362Size of the 363.Tn TCP 364control-block hashtable 365(read-only). 366This may be tuned using the kernel option 367.Dv TCBHASHSIZE 368or by setting 369.Va net.inet.tcp.tcbhashsize 370in the 371.Xr loader 8 . 372.It tcp.pcbcount 373Number of active process control blocks 374(read-only). 375.It tcp.syncookies 376Determines whether or not syn cookies should be generated for 377outbound syn-ack packets. Syn cookies are a great help during 378syn flood attacks, and are enabled by default. 379.It tcp.isn_reseed_interval 380The interval (in seconds) specifying how often the secret data used in 381RFC 1948 initial sequence number calculations should be reseeded. 382By default, this variable is set to zero, indicating that 383no reseeding will occur. 384Reseeding should not be necessary, and will break 385.Dv TIME_WAIT 386recycling for a few minutes. 387.It tcp.inet.tcp.rexmit_{min,slop} 388Adjust the retransmit timer calculation for TCP. The slop is 389typically added to the raw calculation to take into account 390occasional variances that the SRTT (smoothed round trip time) 391is unable to accommodate, while the minimum specifies an 392absolute minimum. While a number of TCP RFCs suggest a 1 393second minimum these RFCs tend to focus on streaming behavior 394and fail to deal with the fact that a 1 second minimum has severe 395detrimental effects over lossy interactive connections, such 396as a 802.11b wireless link, and over very fast but lossy 397connections for those cases not covered by the fast retransmit 398code. For this reason we suggest changing the slop to 200ms and 399setting the minimum to something out of the way, like 20ms, 400which gives you an effective minimum of 200ms (similar to Linux). 401.It tcp.inflight_enable 402Enable 403.Tn TCP 404bandwidth delay product limiting. An attempt will be made to calculate 405the bandwidth delay product for each individual TCP connection and limit 406the amount of inflight data being transmitted to avoid building up 407unnecessary packets in the network. This option is recommended if you 408are serving a lot of data over connections with high bandwidth-delay 409products, such as modems, GigE links, and fast long-haul WANs, and/or 410you have configured your machine to accommodate large TCP windows. In such 411situations, without this option, you may experience high interactive 412latencies or packet loss due to the overloading of intermediate routers 413and switches. Note that bandwidth delay product limiting only affects 414the transmit side of a TCP connection. 415.It tcp.inflight_debug 416Enable debugging for the bandwidth delay product algorithm. This may 417default to on (1) so if you enable the algorithm you should probably also 418disable debugging by setting this variable to 0. 419.It tcp.inflight_min 420This puts an lower bound on the bandwidth delay product window, in bytes. 421A value of 1024 is typically used for debugging. 6000-16000 is more typical 422in a production installation. Setting this value too low may result in 423slow ramp-up times for bursty connections. Setting this value too high 424effectively disables the algorithm. 425.It tcp.inflight_max 426This puts an upper bound on the bandwidth delay product window, in bytes. 427This value should not generally be modified but may be used to set a 428global per-connection limit on queued data, potentially allowing you to 429intentionally set a less than optimum limit to smooth data flow over a 430network while still being able to specify huge internal TCP buffers. 431.It tcp.inflight_stab 432The bandwidth delay product algorithm requires a slightly larger window 433than it otherwise calculates for stability. This parameter determines the 434extra window in maximal packets / 10. The default value of 20 represents 4352 maximal packets. Reducing this value is not recommended but you may 436come across a situation with very slow links where the ping time 437reduction of the default inflight code is not sufficient. If this case 438occurs you should first try reducing tcp.inflight_min and, if that does not 439work, reduce both tcp.inflight_min and tcp.inflight_stab, trying values of 44015, 10, or 5 for the latter. Never use a value less than 5. Reducing 441tcp.inflight_stab can lead to upwards of a 20% underutilization of the link 442as well as reducing the algorithm's ability to adapt to changing 443situations and should only be done as a last resort. 444.El 445.Sh ERRORS 446A socket operation may fail with one of the following errors returned: 447.Bl -tag -width Er 448.It Bq Er EISCONN 449when trying to establish a connection on a socket which 450already has one; 451.It Bq Er ENOBUFS 452when the system runs out of memory for 453an internal data structure; 454.It Bq Er ETIMEDOUT 455when a connection was dropped 456due to excessive retransmissions; 457.It Bq Er ECONNRESET 458when the remote peer 459forces the connection to be closed; 460.It Bq Er ECONNREFUSED 461when the remote 462peer actively refuses connection establishment (usually because 463no process is listening to the port); 464.It Bq Er EADDRINUSE 465when an attempt 466is made to create a socket with a port which has already been 467allocated; 468.It Bq Er EADDRNOTAVAIL 469when an attempt is made to create a 470socket with a network address for which no network interface 471exists. 472.It Bq Er EAFNOSUPPORT 473when an attempt is made to bind or connect a socket to a multicast 474address. 475.El 476.Sh SEE ALSO 477.Xr getsockopt 2 , 478.Xr socket 2 , 479.Xr sysctl 3 , 480.Xr blackhole 4 , 481.Xr inet 4 , 482.Xr intro 4 , 483.Xr ip 4 , 484.Xr setkey 8 485.Rs 486.%A V. Jacobson 487.%A R. Braden 488.%A D. Borman 489.%T "TCP Extensions for High Performance" 490.%O RFC 1323 491.Re 492.Rs 493.%A "A. Heffernan" 494.%T "Protection of BGP Sessions via the TCP MD5 Signature Option" 495.%O "RFC 2385" 496.Re 497.Sh HISTORY 498The 499.Nm 500protocol appeared in 501.Bx 4.2 . 502The RFC 1323 extensions for window scaling and timestamps were added 503in 504.Bx 4.4 . 505