ip - Linux IPv4 protocol implementation


#include <sys/socket.h>
#include <netinet/in.h>
#include <netinet/ip.h> /* superset of previous */

tcp_socket = socket(AF_INET, SOCK_STREAM, 0);
udp_socket = socket(AF_INET, SOCK_DGRAM, 0);
raw_socket = socket(AF_INET, SOCK_RAW, protocol);


Linux implements the Internet Protocol, version 4, described in RFC 791 and RFC 1122. ip contains a level 2 multicasting implementation conforming to RFC 1112. It also contains an IP router including a packet filter.

The programming interface is BSD-sockets compatible. For more information on sockets, see socket(7).

An IP socket is created by calling the socket(2) function as socket(AF_INET, socket_type, protocol). Valid socket types are SOCK_STREAM to open a tcp(7) socket, SOCK_DGRAM to open a udp(7) socket, or SOCK_RAW to open a raw(7) socket to access the IP protocol directly. protocol is the IP protocol in the IP header to be received or sent. The only valid values for protocol are 0 and IPPROTO_TCP for TCP sockets, and 0 and IPPROTO_UDP for UDP sockets. For SOCK_RAW you may specify a valid IANA IP protocol defined in RFC 1700 assigned numbers.

When a process wants to receive new incoming packets or connections, it should bind a socket to a local interface address using bind(2). Only one IP socket may be bound to any given local (address, port) pair. When INADDR_ANY is specified in the bind call, the socket will be bound to all local interfaces. When listen(2) or connect(2) are called on an unbound socket, it is automatically bound to a random free port with the local address set to INADDR_ANY.

A TCP local socket address that has been bound is unavailable for some time after closing, unless the SO_REUSEADDR flag has been set. Care should be taken when using this flag as it makes TCP less reliable.

Address Format

An IP socket address is defined as a combination of an IP interface address and a 16-bit port number. The basic IP protocol does not supply port numbers, they are implemented by higher level protocols like udp(7) and tcp(7). On raw sockets sin_port is set to the IP protocol.
 struct sockaddr_in {
     sa_family_t    sin_family; /* address family: AF_INET */
     in_port_t      sin_port;   /* port in network byte order */
     struct in_addr sin_addr;   /* internet address */
 /* Internet address. */
 struct in_addr {
     uint32_t       s_addr;     /* address in network byte order */

sin_family is always set to AF_INET. This is required; in Linux 2.2 most networking functions return EINVAL when this setting is missing. sin_port contains the port in network byte order. The port numbers below 1024 are called privileged ports (or sometimes: reserved ports). Only privileged processes (i.e., those having the CAP_NET_BIND_SERVICE capability) may bind(2) to these sockets. Note that the raw IPv4 protocol as such has no concept of a port, they are only implemented by higher protocols like tcp(7) and udp(7).

sin_addr is the IP host address. The s_addr member of struct in_addr contains the host interface address in network byte order. in_addr should be assigned one of the INADDR_* values (e.g., INADDR_ANY) or set using the inet_aton(3), inet_addr(3), inet_makeaddr(3) library functions or directly with the name resolver (see gethostbyname(3)).

IPv4 addresses are divided into unicast, broadcast and multicast addresses. Unicast addresses specify a single interface of a host, broadcast addresses specify all hosts on a network and multicast addresses address all hosts in a multicast group. Datagrams to broadcast addresses can be only sent or received when the SO_BROADCAST socket flag is set. In the current implementation, connection-oriented sockets are only allowed to use unicast addresses.

Note that the address and the port are always stored in network byte order. In particular, this means that you need to call htons(3) on the number that is assigned to a port. All address/port manipulation functions in the standard library work in network byte order.

There are several special addresses: INADDR_LOOPBACK ( always refers to the local host via the loopback device; INADDR_ANY ( means any address for binding; INADDR_BROADCAST ( means any host and has the same effect on bind as INADDR_ANY for historical reasons.

Socket Options

IP supports some protocol-specific socket options that can be set with setsockopt(2) and read with getsockopt(2). The socket option level for IP is IPPROTO_IP. A boolean integer flag is zero when it is false, otherwise true.
IP_ADD_MEMBERSHIP (since Linux 1.2)
Join a multicast group. Argument is an ip_mreqn structure.
 struct ip_mreqn {
     struct in_addr imr_multiaddr; /* IP multicast group
                                      address */
     struct in_addr imr_address;   /* IP address of local
                                      interface */
     int            imr_ifindex;   /* interface index */

imr_multiaddr contains the address of the multicast group the application wants to join or leave. It must be a valid multicast address (or setsockopt(2) fails with the error EINVAL). imr_address is the address of the local interface with which the system should join the multicast group; if it is equal to INADDR_ANY an appropriate interface is chosen by the system. imr_ifindex is the interface index of the interface that should join/leave the imr_multiaddr group, or 0 to indicate any interface.

The ip_mreqn is available only since Linux 2.2. For compatibility, the old ip_mreq structure (present since Linux 1.2) is still supported. It differs from ip_mreqn only by not including the imr_ifindex field. Only valid as a setsockopt(2).
IP_DROP_MEMBERSHIP (since Linux 1.2)
Leave a multicast group. Argument is an ip_mreqn or ip_mreq structure similar to IP_ADD_MEMBERSHIP.
IP_HDRINCL (since Linux 2.0)
If enabled, the user supplies an IP header in front of the user data. Only valid for SOCK_RAW sockets. See raw(7) for more information. When this flag is enabled the values set by IP_OPTIONS, IP_TTL and IP_TOS are ignored.
IP_MTU (since Linux 2.2)
Retrieve the current known path MTU of the current socket. Only valid when the socket has been connected. Returns an integer. Only valid as a getsockopt(2).
IP_MTU_DISCOVER (since Linux 2.2)
Set or receive the Path MTU Discovery setting for a socket. When enabled, Linux will perform Path MTU Discovery as defined in RFC 1191 on this socket. The don't-fragment flag is set on all outgoing datagrams. The system-wide default is controlled by the /proc/sys/net/ipv4/ip_no_pmtu_disc file for SOCK_STREAM sockets, and disabled on all others. For non-SOCK_STREAM sockets, it is the user's responsibility to packetize the data in MTU sized chunks and to do the retransmits if necessary. The kernel will reject packets that are bigger than the known path MTU if this flag is set (with EMSGSIZE ).
Path MTU discovery flags Meaning
IP_PMTUDISC_WANT Use per-route settings.
IP_PMTUDISC_DONT Never do Path MTU Discovery.
IP_PMTUDISC_DO Always do Path MTU Discovery.
IP_PMTUDISC_PROBE Set DF but ignore Path MTU.

When PMTU discovery is enabled, the kernel automatically keeps track of the path MTU per destination host. When it is connected to a specific peer with connect(2), the currently known path MTU can be retrieved conveniently using the IP_MTU socket option (e.g., after a EMSGSIZE error occurred). It may change over time. For connectionless sockets with many destinations, the new MTU for a given destination can also be accessed using the error queue (see IP_RECVERR). A new error will be queued for every incoming MTU update.

While MTU discovery is in progress, initial packets from datagram sockets may be dropped. Applications using UDP should be aware of this and not take it into account for their packet retransmit strategy.

To bootstrap the path MTU discovery process on unconnected sockets, it is possible to start with a big datagram size (up to 64K-headers bytes long) and let it shrink by updates of the path MTU.

To get an initial estimate of the path MTU, connect a datagram socket to the destination address using connect(2) and retrieve the MTU by calling getsockopt(2) with the IP_MTU option.

It is possible to implement RFC 4821 MTU probing with SOCK_DGRAM or SOCK_RAW sockets by setting a value of IP_PMTUDISC_PROBE (available since Linux 2.6.22). This is also particularly useful for diagnostic tools such as tracepath(8) that wish to deliberately send probe packets larger than the observed Path MTU.

IP_MULTICAST_IF (since Linux 1.2)
Set the local device for a multicast socket. Argument is an ip_mreqn or ip_mreq structure similar to IP_ADD_MEMBERSHIP.
When an invalid socket option is passed, ENOPROTOOPT is returned.
IP_MULTICAST_LOOP (since Linux 1.2)
Set or read a boolean integer argument that determines whether sent multicast packets should be looped back to the local sockets.
IP_MULTICAST_TTL (since Linux 1.2)
Set or read the time-to-live value of outgoing multicast packets for this socket. It is very important for multicast packets to set the smallest TTL possible. The default is 1 which means that multicast packets don't leave the local network unless the user program explicitly requests it. Argument is an integer.
IP_OPTIONS (since Linux 2.0)
Set or get the IP options to be sent with every packet from this socket. The arguments are a pointer to a memory buffer containing the options and the option length. The setsockopt(2) call sets the IP options associated with a socket. The maximum option size for IPv4 is 40 bytes. See RFC 791 for the allowed options. When the initial connection request packet for a SOCK_STREAM socket contains IP options, the IP options will be set automatically to the options from the initial packet with routing headers reversed. Incoming packets are not allowed to change options after the connection is established. The processing of all incoming source routing options is disabled by default and can be enabled by using the accept_source_route /proc interface. Other options like timestamps are still handled. For datagram sockets, IP options can be only set by the local user. Calling getsockopt(2) with IP_OPTIONS puts the current IP options used for sending into the supplied buffer.
IP_PKTINFO (since Linux 2.2)
Pass an IP_PKTINFO ancillary message that contains a pktinfo structure that supplies some information about the incoming packet. This only works for datagram oriented sockets. The argument is a flag that tells the socket whether the IP_PKTINFO message should be passed or not. The message itself can only be sent/retrieved as control message with a packet using recvmsg(2) or sendmsg(2).
 struct in_pktinfo {
     unsigned int   ipi_ifindex;  /* Interface index */
     struct in_addr ipi_spec_dst; /* Local address */
     struct in_addr ipi_addr;     /* Header Destination
                                     address */
ipi_ifindex is the unique index of the interface the packet was received on. ipi_spec_dst is the local address of the packet and ipi_addr is the destination address in the packet header. If IP_PKTINFO is passed to sendmsg(2) and ipi_spec_dst is not zero, then it is used as the local source address for the routing table lookup and for setting up IP source route options. When ipi_ifindex is not zero, the primary local address of the interface specified by the index overwrites ipi_spec_dst for the routing table lookup.
IP_RECVERR (since Linux 2.2)
Enable extended reliable error message passing. When enabled on a datagram socket, all generated errors will be queued in a per-socket error queue. When the user receives an error from a socket operation, the errors can be received by calling recvmsg(2) with the MSG_ERRQUEUE flag set. The sock_extended_err structure describing the error will be passed in an ancillary message with the type IP_RECVERR and the level IPPROTO_IP. This is useful for reliable error handling on unconnected sockets. The received data portion of the error queue contains the error packet.
The IP_RECVERR control message contains a sock_extended_err structure:
 #define SO_EE_ORIGIN_NONE    0
 #define SO_EE_ORIGIN_LOCAL   1
 #define SO_EE_ORIGIN_ICMP    2
 #define SO_EE_ORIGIN_ICMP6   3
 struct sock_extended_err {
     uint32_t ee_errno;   /* error number */
     uint8_t  ee_origin;  /* where the error originated */
     uint8_t  ee_type;    /* type */
     uint8_t  ee_code;    /* code */
     uint8_t  ee_pad;
     uint32_t ee_info;    /* additional information */
     uint32_t ee_data;    /* other data */
     /* More data may follow */
 struct sockaddr *SO_EE_OFFENDER(struct sock_extended_err *);
ee_errno contains the errno number of the queued error. ee_origin is the origin code of where the error originated. The other fields are protocol-specific. The macro SO_EE_OFFENDER returns a pointer to the address of the network object where the error originated from given a pointer to the ancillary message. If this address is not known, the sa_family member of the sockaddr contains AF_UNSPEC and the other fields of the sockaddr are undefined.
IP uses the sock_extended_err structure as follows: ee_origin is set to SO_EE_ORIGIN_ICMP for errors received as an ICMP packet, or SO_EE_ORIGIN_LOCAL for locally generated errors. Unknown values should be ignored. ee_type and ee_code are set from the type and code fields of the ICMP header. ee_info contains the discovered MTU for EMSGSIZE errors. The message also contains the sockaddr_in of the node caused the error, which can be accessed with the SO_EE_OFFENDER macro. The sin_family field of the SO_EE_OFFENDER address is AF_UNSPEC when the source was unknown. When the error originated from the network, all IP options (IP_OPTIONS, IP_TTL, etc.) enabled on the socket and contained in the error packet are passed as control messages. The payload of the packet causing the error is returned as normal payload. Note that TCP has no error queue; MSG_ERRQUEUE is not permitted on SOCK_STREAM sockets. IP_RECVERR is valid for TCP, but all errors are returned by socket function return or SO_ERROR only.
For raw sockets, IP_RECVERR enables passing of all received ICMP errors to the application, otherwise errors are only reported on connected sockets
It sets or retrieves an integer boolean flag. IP_RECVERR defaults to off.
IP_RECVOPTS (since Linux 2.2)
Pass all incoming IP options to the user in a IP_OPTIONS control message. The routing header and other options are already filled in for the local host. Not supported for SOCK_STREAM sockets.
IP_RECVTOS (since Linux 2.2)
If enabled the IP_TOS ancillary message is passed with incoming packets. It contains a byte which specifies the Type of Service/Precedence field of the packet header. Expects a boolean integer flag.
IP_RECVTTL (since Linux 2.2)
When this flag is set, pass a IP_TTL control message with the time to live field of the received packet as a byte. Not supported for SOCK_STREAM sockets.
IP_RETOPTS (since Linux 2.2)
Identical to IP_RECVOPTS, but returns raw unprocessed options with timestamp and route record options not filled in for this hop.
IP_ROUTER_ALERT (since Linux 2.2)
Pass all to-be forwarded packets with the IP Router Alert option set to this socket. Only valid for raw sockets. This is useful, for instance, for user-space RSVP daemons. The tapped packets are not forwarded by the kernel; it is the user's responsibility to send them out again. Socket binding is ignored, such packets are only filtered by protocol. Expects an integer flag.
IP_TOS (since Linux 1.0)
Set or receive the Type-Of-Service (TOS) field that is sent with every IP packet originating from this socket. It is used to prioritize packets on the network. TOS is a byte. There are some standard TOS flags defined: IPTOS_LOWDELAY to minimize delays for interactive traffic, IPTOS_THROUGHPUT to optimize throughput, IPTOS_RELIABILITY to optimize for reliability, IPTOS_MINCOST should be used for "filler data" where slow transmission doesn't matter. At most one of these TOS values can be specified. Other bits are invalid and shall be cleared. Linux sends IPTOS_LOWDELAY datagrams first by default, but the exact behavior depends on the configured queueing discipline. Some high priority levels may require superuser privileges (the CAP_NET_ADMIN capability). The priority can also be set in a protocol independent way by the (SOL_SOCKET, SO_PRIORITY) socket option (see socket(7)).
IP_TTL (since Linux 1.0)
Set or retrieve the current time-to-live field that is used in every packet sent from this socket.

/proc interfaces

The IP protocol supports a set of /proc interfaces to configure some global parameters. The parameters can be accessed by reading or writing files in the directory /proc/sys/net/ipv4/. Interfaces described as Boolean take an integer value, with a nonzero value ("true") meaning that the corresponding option is enabled, and a zero value ("false") meaning that the option is disabled.
ip_always_defrag (Boolean; since Linux 2.2.13)
[New with kernel 2.2.13; in earlier kernel versions this feature was controlled at compile time by the CONFIG_IP_ALWAYS_DEFRAG option; this option is not present in 2.4.x and later]

When this boolean flag is enabled (not equal 0), incoming fragments (parts of IP packets that arose when some host between origin and destination decided that the packets were too large and cut them into pieces) will be reassembled (defragmented) before being processed, even if they are about to be forwarded.

Only enable if running either a firewall that is the sole link to your network or a transparent proxy; never ever use it for a normal router or host. Otherwise fragmented communication can be disturbed if the fragments travel over different links. Defragmentation also has a large memory and CPU time cost.

This is automagically turned on when masquerading or transparent proxying are configured.

ip_autoconfig (since Linux 2.2 to 2.6.17)
Not documented.
ip_default_ttl (integer; default: 64; since Linux 2.2)
Set the default time-to-live value of outgoing packets. This can be changed per socket with the IP_TTL option.
ip_dynaddr (Boolean; default: disabled; since Linux 2.0.31)
Enable dynamic socket address and masquerading entry rewriting on interface address change. This is useful for dialup interface with changing IP addresses. 0 means no rewriting, 1 turns it on and 2 enables verbose mode.
ip_forward (Boolean; default: disabled; since Linux 1.2)
Enable IP forwarding with a boolean flag. IP forwarding can be also set on a per-interface basis.
ip_local_port_range (since Linux 2.2)
Contains two integers that define the default local port range allocated to sockets. Allocation starts with the first number and ends with the second number. Note that these should not conflict with the ports used by masquerading (although the case is handled). Also arbitrary choices may cause problems with some firewall packet filters that make assumptions about the local ports in use. First number should be at least greater than 1024, or better, greater than 4096, to avoid clashes with well known ports and to minimize firewall problems.
ip_no_pmtu_disc (Boolean; default: disabled; since Linux 2.2)
If enabled, don't do Path MTU Discovery for TCP sockets by default. Path MTU discovery may fail if misconfigured firewalls (that drop all ICMP packets) or misconfigured interfaces (e.g., a point-to-point link where the both ends don't agree on the MTU) are on the path. It is better to fix the broken routers on the path than to turn off Path MTU Discovery globally, because not doing it incurs a high cost to the network.
ip_nonlocal_bind (Boolean; default: disabled; since Linux 2.4)
If set, allows processes to bind(2) to nonlocal IP addresses, which can be quite useful, but may break some applications.
ip6frag_time (integer; default 30)
Time in seconds to keep an IPv6 fragment in memory.
ip6frag_secret_interval (integer; default 600)
Regeneration interval (in seconds) of the hash secret (or lifetime for the hash secret) for IPv6 fragments.
ipfrag_high_thresh (integer), ipfrag_low_thresh (integer)
If the amount of queued IP fragments reaches ipfrag_high_thresh, the queue is pruned down to ipfrag_low_thresh. Contains an integer with the number of bytes.
See arp(7).


All ioctls described in socket(7) apply to ip.

Ioctls to configure generic device parameters are described in netdevice(7).


The user tried to execute an operation without the necessary permissions. These include: sending a packet to a broadcast address without having the SO_BROADCAST flag set; sending a packet via a prohibit route; modifying firewall settings without superuser privileges (the CAP_NET_ADMIN capability); binding to a privileged port without superuser privileges (the CAP_NET_BIND_SERVICE capability).
Tried to bind to an address already in use.
A nonexistent interface was requested or the requested source address was not local.
Operation on a nonblocking socket would block.
An connection operation on a nonblocking socket is already in progress.
A connection was closed during an accept(2).
No valid routing table entry matches the destination address. This error can be caused by a ICMP message from a remote router or for the local routing table.
Invalid argument passed. For send operations this can be caused by sending to a blackhole route.
connect(2) was called on an already connected socket.
Datagram is bigger than an MTU on the path and it cannot be fragmented.
Not enough free memory. This often means that the memory allocation is limited by the socket buffer limits, not by the system memory, but this is not 100% consistent.
SIOCGSTAMP was called on a socket where no packet arrived.
A kernel subsystem was not configured.
Invalid socket option passed.
The operation is only defined on a connected socket, but the socket wasn't connected.
User doesn't have permission to set high priority, change configuration, or send signals to the requested process or group.
The connection was unexpectedly closed or shut down by the other end.
The socket is not configured or an unknown socket type was requested.

Other errors may be generated by the overlaying protocols; see tcp(7), raw(7), udp(7) and socket(7).


IP_MTU, IP_MTU_DISCOVER, IP_PKTINFO, IP_RECVERR and IP_ROUTER_ALERT are Linux-specific and should not be used in programs intended to be portable. Be very careful with the SO_BROADCAST option - it is not privileged in Linux. It is easy to overload the network with careless broadcasts. For new application protocols it is better to use a multicast group instead of broadcasting. Broadcasting is discouraged.

Some other BSD sockets implementations provide IP_RCVDSTADDR and IP_RECVIF socket options to get the destination address and the interface of received datagrams. Linux has the more general IP_PKTINFO for the same task.

Some BSD sockets implementations also provide an IP_RECVTTL option, but an ancillary message with type IP_RECVTTL is passed with the incoming packet. This is different from the IP_TTL option used in Linux.

Using SOL_IP socket options level isn't portable, BSD-based stacks use IPPROTO_IP level.


For compatibility with Linux 2.0, the obsolete socket(AF_INET, SOCK_PACKET, protocol) syntax is still supported to open a packet(7) socket. This is deprecated and should be replaced by socket(AF_PACKET, SOCK_RAW, protocol) instead. The main difference is the new sockaddr_ll address structure for generic link layer information instead of the old sockaddr_pkt.


There are too many inconsistent error values.

The ioctls to configure IP-specific interface options and ARP tables are not described.

Some versions of glibc forget to declare in_pktinfo. Workaround currently is to copy it into your program from this man page.

Receiving the original destination address with MSG_ERRQUEUE in msg_name by recvmsg(2) does not work in some 2.2 kernels.


recvmsg(2), sendmsg(2), byteorder(3), ipfw(4), capabilities(7), netlink(7), raw(7), socket(7), tcp(7), udp(7)

RFC 791 for the original IP specification.
RFC 1122 for the IPv4 host requirements.
RFC 1812 for the IPv4 router requirements.


This page is part of release 3.25 of the Linux man-pages project. A description of the project, and information about reporting bugs, can be found at http://www.kernel.org/doc/man-pages/.