xref: /netbsd-src/share/man/man4/ip.4 (revision 7088981fda502269766c8af1ef1471d7d85b2985)
1.\"	$NetBSD: ip.4,v 1.47 2020/09/08 17:30:44 wiz Exp $
2.\"
3.\" Copyright (c) 1983, 1991, 1993
4.\"	The Regents of the University of California.  All rights reserved.
5.\"
6.\" Redistribution and use in source and binary forms, with or without
7.\" modification, are permitted provided that the following conditions
8.\" are met:
9.\" 1. Redistributions of source code must retain the above copyright
10.\"    notice, this list of conditions and the following disclaimer.
11.\" 2. Redistributions in binary form must reproduce the above copyright
12.\"    notice, this list of conditions and the following disclaimer in the
13.\"    documentation and/or other materials provided with the distribution.
14.\" 3. Neither the name of the University nor the names of its contributors
15.\"    may be used to endorse or promote products derived from this software
16.\"    without specific prior written permission.
17.\"
18.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
19.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
22.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28.\" SUCH DAMAGE.
29.\"
30.\"     @(#)ip.4	8.2 (Berkeley) 11/30/93
31.\"
32.Dd September 8, 2020
33.Dt IP 4
34.Os
35.Sh NAME
36.Nm ip
37.Nd Internet Protocol
38.Sh SYNOPSIS
39.In sys/socket.h
40.In netinet/in.h
41.Ft int
42.Fn socket AF_INET SOCK_RAW proto
43.Sh DESCRIPTION
44IP is the network layer protocol used by the Internet protocol family.
45Options may be set at the
46IP level when using higher-level protocols that are based on
47IP (such as TCP and UDP).
48It may also be accessed through a
49.Dq raw socket
50when developing new protocols, or special-purpose applications.
51.Pp
52There are several IP-level
53.Xr setsockopt 2 Ns / Ns Xr getsockopt 2
54options.
55.Dv IP_OPTIONS
56may be used to provide IP options to be transmitted in the
57IP header of each outgoing packet
58or to examine the header options on incoming packets.
59IP options may be used with any socket type in the Internet family.
60The format of IP options to be sent is that specified by the
61IP protocol specification (RFC 791), with one exception:
62the list of addresses for Source Route options must include the first-hop
63gateway at the beginning of the list of gateways.
64The first-hop gateway address will be extracted from the option list
65and the size adjusted accordingly before use.
66To disable previously specified options, use a zero-length buffer:
67.Bd -literal
68setsockopt(s, IPPROTO_IP, IP_OPTIONS, NULL, 0);
69.Ed
70.Pp
71.Dv IP_TOS
72and
73.Dv IP_TTL
74may be used to set the type-of-service and time-to-live fields in the
75IP header for
76.Dv SOCK_STREAM
77and
78.Dv SOCK_DGRAM
79sockets.
80For example,
81.Bd -literal
82int tos = IPTOS_LOWDELAY;       /* see <netinet/ip.h> */
83setsockopt(s, IPPROTO_IP, IP_TOS, &tos, sizeof(tos));
84
85int ttl = 60;                   /* max = 255 */
86setsockopt(s, IPPROTO_IP, IP_TTL, &ttl, sizeof(ttl));
87.Ed
88.Pp
89.Dv IP_IPSEC_POLICY
90controls IPSec policy for sockets.
91For example,
92.Bd -literal
93const char *policy = "in ipsec ah/transport//require";
94char *buf = ipsec_set_policy(policy, strlen(policy));
95setsockopt(s, IPPROTO_IP, IP_IPSEC_POLICY, buf, ipsec_get_policylen(buf));
96.Ed
97.Pp
98The
99.Dv IP_RECVPKTINFO
100option can be used to turn on receiving of information about the destination
101address of the packet, and the interface index.
102The information is passed in a
103.Vt struct in_pktinfo
104structure, which contains
105.Bd -literal
106	struct in_addr ipi_addr;	/* the source or destination address */
107	unsigned int ipi_ifindex;	/* the interface index */
108.Ed
109.Pp
110and added to the control portion of the message:
111The cmsghdr fields have the following values:
112.Bd -literal
113cmsg_len = CMSG_LEN(sizeof(struct in_pktinfo))
114cmsg_level = IPPROTO_IP
115cmsg_type = IP_PKTINFO
116.Ed
117.Pp
118For
119.Xr sendmsg 2 ,
120the source address or output interface can be specified by adding an
121.Dv IP_PKTINFO
122message to the control part of the message on a
123.Dv SOCK_DGRAM
124or
125.Dv SOCK_RAW
126socket.
127Setting ipi_ifindex will cause the primary address of that interface
128to be used; setting ipi_addr will directly choose that address.
129The
130.Dv IP_PKTINFO
131cmsghdr structure from a received message may be
132used unchanged, in which case the outgoing message will be sent
133from the address the incoming message was received on.
134.Pp
135Setting the
136.Dv IP_PKTINFO
137option on a socket, with the same
138.Vt struct in_pktinfo
139structure, will set the default source address to be used until set
140again, unless explicitly overridden on a per-packet basis, as above.
141.Pp
142The
143.Dv IP_PORTALGO
144can be used to randomize the port selection.
145Valid algorithms are described in
146.Xr rfc6056 7
147and their respective constants are in
148.In netinet/portalgo.h .
149For example,
150.Bd -literal
151int algo = PORTALGO_ALGO_RANDOM_PICK;       /* see <netinet/portalgo.h> */
152setsockopt(s, IPPROTO_IP, IP_PORTALGO, &algo, sizeof(algo));
153.Ed
154.Pp
155The port selection can be also viewed and controlled at a global level for all
156IP sockets using the following
157.Xr sysctl 7
158variables:
159.Dv net.inet.ip.anonportalgo.available
160and
161.Dv net.inet.ip.anonportalgo.selected .
162.Pp
163.Dv IP_PORTRANGE
164controls how ephemeral ports are allocated for
165.Dv SOCK_STREAM
166and
167.Dv SOCK_DGRAM
168sockets.
169For example,
170.Bd -literal
171int range = IP_PORTRANGE_LOW;       /* see <netinet/in.h> */
172setsockopt(s, IPPROTO_IP, IP_PORTRANGE, &range, sizeof(range));
173.Ed
174.Pp
175If the
176.Dv IP_RECVDSTADDR
177option is enabled on a
178.Dv SOCK_DGRAM
179or
180.Dv SOCK_RAW
181socket,
182the
183.Xr recvmsg 2
184call will return the destination IP address for a UDP datagram.
185The msg_control field in the msghdr structure points to a buffer
186that contains a cmsghdr structure followed by the IP address.
187The cmsghdr fields have the following values:
188.Bd -literal
189cmsg_len = CMSG_LEN(sizeof(struct in_addr))
190cmsg_level = IPPROTO_IP
191cmsg_type = IP_RECVDSTADDR
192.Ed
193.Pp
194For
195.Xr sendmsg 2 ,
196the source address can be specified by adding
197.Dv IP_SENDSRCADDR
198to the control part of the message on a
199.Dv SOCK_DGRAM
200or
201.Dv SOCK_RAW
202socket.
203The
204.Dv IP_RECVDSTADDR
205cmsghdr structure from a received message may
206be used unchanged, in which case the outgoing message will be sent
207from the address the incoming message was received on.
208.Pp
209If the
210.Dv IP_RECVIF
211option is enabled on a
212.Dv SOCK_DGRAM
213or
214.Dv SOCK_RAW
215socket,
216the
217.Xr recvmsg 2
218call will return a struct sockaddr_dl corresponding to
219the interface on which the packet was received.
220the msg_control field in the msghdr structure points to a buffer
221that contains a cmsghdr structure followed by the struct sockaddr_dl.
222The cmsghdr fields have the following values:
223.Bd -literal
224cmsg_len = CMSG_LEN(sizeof(struct sockaddr_dl))
225cmsg_level = IPPROTO_IP
226cmsg_type = IP_RECVIF
227.Ed
228.Pp
229If the
230.Dv IP_BINDANY
231option is enabled on a
232.Dv SOCK_STREAM ,
233.Dv SOCK_DGRAM ,
234or a
235.Dv SOCK_RAW
236socket, one can
237.Xr bind 2
238to any address, even one not bound to any available network interface in the
239system.
240This functionality (in conjunction with special firewall rules) can be used for
241implementing a transparent proxy.
242The
243.Dv KAUTH_REQ_NETWORK_BIND_ANYADDR
244privilege is needed to set this option.
245.Pp
246If the
247.Dv IP_RECVTTL
248option is enabled on a
249.Dv SOCK_DGRAM
250socket, the
251.Xr recvmsg 2
252call will return the TTL of the received datagram.
253The msg_control field in the msghdr structure points to a buffer
254that contains a cmsghdr structure followed by the TTL value.
255The cmsghdr fields have the following values:
256.Bd -literal
257cmsg_len = CMSG_LEN(sizeof(uint8_t))
258cmsg_level = IPPROTO_IP
259cmsg_type = IP_RECVTTL
260.Ed
261.Pp
262The
263.Dv IP_MINTTL
264option may be used on
265.Dv SOCK_DGRAM
266or
267.Dv SOCK_STREAM
268sockets to discard packets with a TTL lower than the option value.
269This can be used to implement the
270.Em Generalized TTL Security Mechanism (GTSM)
271according to RFC 3682.
272To discard all packets with a TTL lower than 255:
273.Bd -literal -offset indent
274int minttl = 255;
275setsockopt(s, IPPROTO_IP, IP_MINTTL, &minttl, sizeof(minttl));
276.Ed
277.Ss MULTICAST OPTIONS
278IP multicasting is supported only on
279.Dv AF_INET
280sockets of type
281.Dv SOCK_DGRAM
282and
283.Dv SOCK_RAW ,
284and only on networks where the interface driver supports multicasting.
285.Pp
286The
287.Dv IP_MULTICAST_TTL
288option changes the time-to-live (TTL) for outgoing multicast datagrams
289in order to control the scope of the multicasts:
290.Bd -literal
291u_char ttl;	/* range: 0 to 255, default = 1 */
292setsockopt(s, IPPROTO_IP, IP_MULTICAST_TTL, &ttl, sizeof(ttl));
293.Ed
294.Pp
295Datagrams with a TTL of 1 are not forwarded beyond the local network.
296Multicast datagrams with a TTL of 0 will not be transmitted on any network,
297but may be delivered locally if the sending host belongs to the destination
298group and if multicast loopback has not been disabled on the sending socket
299(see below).
300Multicast datagrams with TTL greater than 1 may be forwarded
301to other networks if a multicast router is attached to the local network.
302.Pp
303For hosts with multiple interfaces, each multicast transmission is
304sent from the primary network interface.
305The
306.Dv IP_MULTICAST_IF
307option overrides the default for
308subsequent transmissions from a given socket:
309.Bd -literal
310struct in_addr addr;
311setsockopt(s, IPPROTO_IP, IP_MULTICAST_IF, &addr, sizeof(addr));
312.Ed
313.Pp
314where "addr" is the local IP address of the desired interface or
315.Dv INADDR_ANY
316to specify the default interface.
317An interface's local IP address and multicast capability can
318be obtained via the
319.Dv SIOCGIFCONF
320and
321.Dv SIOCGIFFLAGS
322ioctls.
323An application may also specify an alternative to the default network interface
324by index:
325.Bd -literal
326struct uint32_t idx = htonl(ifindex);
327setsockopt(s, IPPROTO_IP, IP_MULTICAST_IF, &idx, sizeof(idx));
328.Ed
329.Pp
330where "ifindex" is an interface index as returned by
331.Xr if_nametoindex 3 .
332.Pp
333Normal applications should not need to use
334.Dv IP_MULTICAST_IF .
335.Pp
336If a multicast datagram is sent to a group to which the sending host itself
337belongs (on the outgoing interface), a copy of the datagram is, by default,
338looped back by the IP layer for local delivery.
339The
340.Dv IP_MULTICAST_LOOP
341option gives the sender explicit control
342over whether or not subsequent datagrams are looped back:
343.Bd -literal
344u_char loop;	/* 0 = disable, 1 = enable (default) */
345setsockopt(s, IPPROTO_IP, IP_MULTICAST_LOOP, &loop, sizeof(loop));
346.Ed
347.Pp
348This option
349improves performance for applications that may have no more than one
350instance on a single host (such as a router demon), by eliminating
351the overhead of receiving their own transmissions.
352It should generally not be used by applications for which there
353may be more than one instance on a single host (such as a conferencing
354program) or for which the sender does not belong to the destination
355group (such as a time querying program).
356.Pp
357A multicast datagram sent with an initial TTL greater than 1 may be delivered
358to the sending host on a different interface from that on which it was sent,
359if the host belongs to the destination group on that other interface.
360The loopback control option has no effect on such delivery.
361.Pp
362A host must become a member of a multicast group before it can receive
363datagrams sent to the group.
364To join a multicast group, use the
365.Dv IP_ADD_MEMBERSHIP
366option:
367.Bd -literal
368struct ip_mreq mreq;
369setsockopt(s, IPPROTO_IP, IP_ADD_MEMBERSHIP, &mreq, sizeof(mreq));
370.Ed
371.Pp
372where
373.Fa mreq
374is the following structure:
375.Bd -literal
376struct ip_mreq {
377    struct in_addr imr_multiaddr; /* multicast group to join */
378    struct in_addr imr_interface; /* interface to join on */
379}
380.Ed
381.Pp
382.Dv imr_interface
383should be
384.Dv INADDR_ANY
385to choose the default multicast interface, or the IP
386address of a particular multicast-capable interface if
387the host is multihomed.
388Membership is associated with a single interface;
389programs running on multihomed hosts may need to
390join the same group on more than one interface.
391Up to
392.Dv IP_MAX_MEMBERSHIPS
393(currently 20) memberships may be added on a single socket.
394.Pp
395To drop a membership, use:
396.Bd -literal
397struct ip_mreq mreq;
398setsockopt(s, IPPROTO_IP, IP_DROP_MEMBERSHIP, &mreq, sizeof(mreq));
399.Ed
400.Pp
401where
402.Fa mreq
403contains the same values as used to add the membership.
404Memberships are dropped when the socket is closed or the process exits.
405.\"-----------------------
406.Ss RAW IP SOCKETS
407Raw IP sockets are connectionless, and are normally used with the
408.Xr sendto 2
409and
410.Xr recvfrom 2
411calls, though the
412.Xr connect 2
413call may also be used to fix the destination for future
414packets (in which case the
415.Xr read 2
416or
417.Xr recv 2
418and
419.Xr write 2
420or
421.Xr send 2
422system calls may be used).
423.Pp
424If
425.Fa proto
426is 0, the default protocol
427.Dv IPPROTO_RAW
428is used for outgoing packets, and only incoming packets destined
429for that protocol are received.
430If
431.Fa proto
432is non-zero, that protocol number will be used on outgoing packets
433and to filter incoming packets.
434.Pp
435Outgoing packets automatically have an IP
436header prepended to them (based on the destination address and the
437protocol number the socket is created with), unless the
438.Dv IP_HDRINCL
439option has been set.
440Incoming packets are received with IP header and options intact.
441.Pp
442.Dv IP_HDRINCL
443indicates the complete IP header is included with the data and may
444be used only with the
445.Dv SOCK_RAW
446type.
447.Bd -literal
448#include <netinet/ip.h>
449
450int hincl = 1;                  /* 1 = on, 0 = off */
451setsockopt(s, IPPROTO_IP, IP_HDRINCL, &hincl, sizeof(hincl));
452.Ed
453.Pp
454Unlike previous
455.Bx
456releases, the program must set all
457the fields of the IP header, including the following:
458.Bd -literal
459ip->ip_v = IPVERSION;
460ip->ip_hl = hlen >> 2;
461ip->ip_id = 0;  /* 0 means kernel set appropriate value */
462ip->ip_off = offset;
463.Ed
464.Pp
465If the header source address is set to
466.Dv INADDR_ANY ,
467the kernel will choose an appropriate address.
468.Sh DIAGNOSTICS
469A socket operation may fail with one of the following errors returned:
470.Bl -tag -width [EADDRNOTAVAIL]
471.It Bq Er EACCES
472when an attempt is made to create a raw IP socket by a non-privileged process.
473.It Bq Er EADDRNOTAVAIL
474when an attempt is made to create a socket with a network address
475for which no network interface exists.
476.It Bq Er EISCONN
477when trying to establish a connection on a socket which already
478has one, or when trying to send a datagram with the destination
479address specified and the socket is already connected;
480.It Bq Er ENOBUFS
481when the system runs out of memory for an internal data structure;
482.It Bq Er ENOTCONN
483when trying to send a datagram, but no destination address is
484specified, and the socket hasn't been connected;
485.El
486.Pp
487The following errors specific to IP may occur when setting or getting
488IP options:
489.Bl -tag -width EADDRNOTAVAILxx
490.It Bq Er EINVAL
491An unknown socket option name was given; or
492the IP option field was improperly formed; an option field was
493shorter than the minimum value or longer than the option buffer provided.
494.El
495.Sh COMPATIBILITY
496The
497.Dv IP_RECVPKTINFO
498option is used because it is directly compatible with Solaris, AIX, etc.,
499and the
500.Dv IP_PKTINFO
501option is intended to be used in their manner, to set the default source
502address for outgoing packets on a
503.Dv SOCK_DGRAM
504or
505.Dv SOCK_RAW
506socket.
507For compatibility with Linux, however, if you attempt to set the
508.Dv IP_PKTINFO
509option, using an integer parameter as a boolean value, this will
510transparently manipulate the
511.Dv IP_RECVPKTINFO
512option instead.
513Source code compatibility with both environments is thus maintained.
514.Sh SEE ALSO
515.Xr getsockopt 2 ,
516.Xr recv 2 ,
517.Xr send 2 ,
518.Xr CMSG_DATA 3 ,
519.Xr ipsec_set_policy 3 ,
520.Xr icmp 4 ,
521.Xr inet 4 ,
522.Xr intro 4
523.Rs
524.%R RFC
525.%N 791
526.%D September 1981
527.%T "Internet Protocol"
528.Re
529.Rs
530.%R RFC
531.%N 1112
532.%D August 1989
533.%T "Host Extensions for IP Multicasting"
534.Re
535.Rs
536.%R RFC
537.%N 1122
538.%D October 1989
539.%T "Requirements for Internet Hosts \(em Communication Layers"
540.Re
541.Sh HISTORY
542The
543.Nm
544protocol appeared in
545.Bx 4.2 .
546