IPv6(7P) IPv6(7P)
NAME
IPv6, ipv6, ip6 - Internet Protocol Version 6
SYNOPSIS
DESCRIPTION
IPv6 is the next generation network-layer protocol designed to be the successor to the current Internet Protocol version 4 (IPv4). It pro-
vides the packet delivery service for TCP, UDP and ICMPv6.
IPv6 has significant advantages over IPv4 in terms of increased address space, simplified header format, integrated QoS support and manda-
tory security.
IPv6 also allows optional internet-layer information to be encoded in separate headers called extension headers which are placed between
the IPv6 header and upper layer headers. Extension headers currently supported are hop-by-hop option header, destination option header,
fragment header and routing (type 0) header. An IPv6 packet may carry zero, one, or more extension headers, each identified by the next
header field of the preceding header.
IPv6 has extended the address size from 32 bits to 128 bits and they are textually represented in hex-colon notation as where the x's are
the hexadecimal values of the eight 16-bit pieces of the address. For example
IPv6 has three types of addresses: and
o An is an identifier for a single interface. A packet sent to an unicast address is delivered to the interface identified by that
address.
o An is an identifier for a set of interfaces. A packet sent to an anycast address is delivered to one of the interfaces identi-
fied by that address.
o A is an identifier for a set of interfaces. A packet sent to a multicast address is delivered to all interfaces identified by
that address.
There are no broadcast addresses in IPv6, their function is superseded by multicast addresses.
Every IPv6 address has a associated with it. A scope is a topological span within which the address may be used as an unique identifier
for an interface or set of interfaces.
There are several types of unicast address.
o The loopback address is assigned to the loopback interface. It may be used by a node to send an IPv6 packet to itself.
o Link-local address uniquely identifies interfaces within a single link and it has a fixed prefix of For example,
o Global address uniquely identifies interfaces anywhere in the internet.
o Unique Local IPv6 Unicast address is defined in RFC 4193 and has a fixed prefix of These addresses have a high probability of
global uniqueness, and are intended for local communications within a limited area such as a site.
Note: Site-local address with a fixed prefix of designed originally to uniquely identify interfaces within a single site only is
deprecated by RFC 3879. The IPv6 Address Architecture RFC 4291 removes the special meaning for the prefix and makes this prefix
part of the global unicast pool.
There are 2 special unicast addresses which hold an embedded IPv4 address in the low order 32-bits.
o The first type is termed as IPv4-compatible IPv6 address and is of the form This type of address is used by dual stack
(IPv4/IPv6) nodes to perform automatic IPv6-over-IPv4 tunneling where the IPv4 tunnel endpoint address is determined from the
IPv4 address embedded in the IPv4-compatible destination address of the IPv6 packet being tunneled.
Note: Starting with RFC4291, the IPv4-compatible IPv6 address has been deprecated because the current IPv6 transition mechanisms
no longer use them.
o The second type is termed as IPv4-mapped IPv6 address and is of the form This address facilitates IPv6 applications to interoper-
ate with IPv4 applications. Applications can automatically generate this address using (see getaddrinfo(3N)) when the specified
host has only IPv4 address.
IPv6 Multicast
As with IPv4, an application joins a multicast group on a network interface in order to receive multicast datagrams sent on the network to
which that interface connects.
The application must also bind to the destination port number in order to receive datagrams that are sent to that port number. If the
application binds to the unspecified address (all zeroes), it may receive all datagrams that are sent to the port number. If the applica-
tion binds to a multicast group address, it may receive only datagrams sent to that group and port number. It is not necessary to join a
multicast group in order to send datagrams to it.
For each multicast group that an application joins on a given socket and network interface, there is an associated filter mode and source
list. The filter mode can be "exclude" mode or "include" mode. The source list is a set of IP addresses which will be compared to the
source addresses of received multicast datagrams.
An application uses "exclude" mode when it wants to block reception of multicast datagrams from a specific set of sources, while allowing
multicast datagrams from all other sources.
For groups in "exclude" mode, the source list may be empty, thus allowing reception of multicast datagrams from all sources. An applica-
tion uses "include" mode when it wants to allow reception of multicast datagrams from a specific set of sources, while blocking multicast
datagrams from all other sources.
For groups in "include" mode, the source list needs to contain at least one member in order for the application to stay joined to the
group; removing the last address from an "include" mode source list will cause the application to leave the group.
Multicast socket options are not supported on SOCK_STREAM sockets.
IPv6 Socket Options
New socket options are defined for IPv6 to send and receive extension headers and to exchange other optional information between the kernel
and application. The options are supported at the protocol level. The type of the variable pointed to by the optval parameter is indi-
cated in parenthesis.
Sets or gets the hop limit used in outgoing unicast packets. When this option is set using (see setsockopt(2)), the new option
value specified is used as the hop limit for all subsequent unicast packets sent via that socket. Valid values are in the range
0-255 (both inclusive) and the default value is 64. For example,
int hoplimit = 50;
setsockopt(s, IPPROTO_IPV6, IPV6_UNICAST_HOPS,
&hoplimit, sizeof(hoplimit));
This option can be used with (see getsockopt(2)) to determine the hop limit value the system will use for subsequent unicast
packets sent via that socket.
Sets or gets the hop limit used in outgoing multicast packets.
When this option is set, the new option value specified is used as the hop limit for all subsequent multicast packets sent
via that socket. Valid values are in the range 0-255 (both inclusive) and the default value is 1.
Sets the interface to use for outgoing multicast
packets. The option value is the index of the selected outgoing interface. For example,
unsigned int index;
index = if_nametoindex("lan0");
setsockopt(s, IPPROTO_IPV6, IPV6_MULTICAST_IF,
&index, sizeof(index));
Enables or disables loopback in the IP
layer for multicast datagrams sent through this socket. The value of the variable pointed to by optval is zero (disable) or
non-zero (enable). Default: enabled.
Joins a multicast group on a specified local interface.
The group is joined in "exclude" mode, with an empty source list. is defined in as:
struct ipv6_mreq {
struct in6_addr ipv6mr_multiaddr;
uint32_t ipv6mr_interface;
};
The field holds the interface index of a local interface, or zero. If the interface index is specified as 0 then the default
multicast interface is used. The field holds an IPv6 multicast address or an IPv4 multicast address in IPv4-mapped IPv6
address format.
Joins a multicast group on a specified local interface.
The group is joined in "exclude" mode, with an empty source list. is defined in as:
struct group_req {
uint32_t gr_interface;
struct sockaddr_storage gr_group;
};
The field holds the interface index of a local interface, or zero. If the interface index is specified as 0 then the default
multicast interface is used. The field holds a containing an IPv6 multicast address or an IPv4 multicast address in
IPv4-mapped IPv6 address format.
Leaves a multicast group on a specified local interface.
The multicast address of the group to leave and the interface index are specified using The interface index should match the
index used while joining the group. If the interface index is specified as 0 then the default multicast interface is used.
Leaves a multicast group on a specified local interface.
The multicast address of the group to leave and the interface index are specified using The interface index should match the
index used while joining the group. If the interface index is specified as 0 then the default multicast interface is used.
Adds a source address to the list of blocked sources for a multicast group.
The group must already be joined, and must be in "exclude" mode. is defined in as:
struct group_source_req {
uint32_t gsr_interface;
struct sockaddr_storage gsr_group;
struct sockaddr_storage gsr_source;
};
The field holds the interface index of a local interface, or zero. If the interface index is specified as 0 then the default
multicast interface is used. The field holds a containing an IPv6 multicast address or an IPv4 multicast address in
IPv4-mapped IPv6 address format. The field holds a containing an IPv6 source address or an IPv4 source address in
IPv4-mapped IPv6 address format. The multicast address and the source address must both be native IPv6 addresses, or must
both be IPv4-mapped IPv6 addresses.
Removes a source address from the list of blocked sources for a multicast group.
The group must already be joined, and must be in "exclude" mode. The multicast address, interface index, and source address
are specified using The interface index should match the index used while joining the group. If the interface index is spec-
ified as 0 then the default multicast interface is used.
Adds a source address to the list of allowed sources
for a multicast group in "include" mode, joining the group in "include" mode if not already joined. The multicast address,
interface index, and source address are specified using If the interface index is specified as 0 then the default multicast
interface is used.
Removes a source address from the list of allowed sources
for a multicast group in "include" mode, leaving the group if the last source is being removed. The multicast address,
interface index, and source address are specified using If the interface index is specified as 0 then the default multicast
interface is used.
When this option is set, kernel computes the
checksum for outbound packets and verifies checksum on inbound packets. The option value is the byte offset of the checksum
location in the user data. This option is not valid for since checksum computation is mandatory for The default value is -1
(checksums not computed nor verified for protocols other than
When this option is enabled,
(destination IPv6 address and the arriving interface index) is returned as ancillary data by (See recvmsg(2).) The informa-
tion is returned in structure and it is defined in as:
struct in6_pktinfo {
struct in6_addr ipi6_addr;
uint32_t ipi6_ifindex;
};
By default this option is disabled.
When this option is enabled, inbound packet's
hoplimit is returned as ancillary data by For example,
int on = 1;
setsockopt(s, IPPROTO_IPV6, IPV6_RECVHOPLIMIT,
&on, sizeof(on));
By default this option is disabled.
When this option is enabled, inbound packet's
traffic class is returned as ancillary data by For example,
int on = 1;
setsockopt(s, IPPROTO_IPV6, IPV6_RECVTCLASS,
&on, sizeof(on));
By default this option is disabled.
When this option is enabled, the inbound packet's
destination options (when present) is returned as ancillary data by By default this option is disabled.
When this option is enabled, the inbound packet's
hop-by-hop options (when present) is returned as ancillary data by By default this option is disabled.
When this option is enabled, the inbound packet's
routing options (when present) is returned as ancillary data by By default this option is disabled.
When this option is enabled, the inbound packet's
destination options appearing before a routing header (when present) is returned as ancillary data by By default this option
is disabled.
When this option is enabled, path MTU information
will be delivered as ancillary data in the form of to when either the application sends packets that are too big for the path
MTU, or when those packets are sent from a socket on which the option is enabled and the packet size is larger than the MTU
of the outgoing interface. (See recvmsg(2).) The structure is defined in as:
struct ip6_mtuinfo {
struct sockaddr_in6 ip6m_addr;
uint32_t ip6m_mtu;
};
In this situation, will return zero but there will be a with set to and will indicate that is bytes long. This option is not
supported on SOCK_STREAM sockets. By default this option is disabled.
This get-only option retrieves the current path MTU
value for the destination of a connected socket. This option cannot be used with
When this option is enabled on a socket, the socket can
be used to send and receive IPv6 packets only. By default this option is disabled.
The next ten socket options can be used with both and as option name in ancillary data to (See sendmsg(2).)
Used to set the source address and interface index for outgoing packets.
Used to set the hop limit for outbound packets. This hop
limit is valid for only a single output operation. To set hop limit for all unicast or multicast IPv6 packets use or options
respectively.
Used to set the traffic class for outbound packets. If it is specified
via ancillary data, this traffic class is valid for only a single output operation. To set traffic class for all outbound
IPv6 packets, should be used. Valid values range from -1 to 255, inclusive. If the value is equal to -1, the default value(0) is used. If the value is between 0 and 255 inclusive, the specified value is used.
Used to set the next hop address. The node
identified by this address must be a neighbor of the sending host. When this address is the same as the destination IPv6
address then this is equivalent to socket option.
Used to specify the routing header for outgoing
packets. Only Type 0 routing header is currently supported.
Used to specify one or more destination options
to be sent in subsequent IPv6 packets.
Used to specify one or more hop-by-hop options
to be sent in subsequent IPv6 packets.
Used to specify one or more destination options
preceding a routing header. This option will be silently ignored when sending packets unless a routing header is also speci-
fied.
Used to control the behavior of path MTU discovery.
specifies that path MTU discovery will be disabled and packets
will be sent using the minimum MTU.
specifies that path MTU discovery will be performed.
specifies
that path MTU discovery will be performed for unicast destinations but not for multicast destinations, and packets to
multicast destinations will be sent using the minimum MTU.
Values other than -1, 0, and 1 are not valid.
Used to disable the automatic insertion of fragment headers.
This option is not supported on SOCK_STREAM sockets. By default this option is disabled.
IPv6 uses the enhanced version of ICMP called ICMPv6 to report errors encountered in processing packets and for diagnostic purposes (like
ping). ICMPv6 is an integral part of IPv6 and has a next header value of 58.
While some of the options and the associated structures are defined in applications are not required to include this header file explic-
itly, it is automatically included by
ERRORS
One of the following errors may be returned when a socket operation fails.
The specified multicast group has been joined
already.
The specified IPv6 address is not a local interface
address or there is no route for the specified multicast address or the specified multicast group has not
been joined.
The address family is not supported.
The parameter 'level' is not
or optval is the NULL address, or the specified multicast address is not valid, or the specified source
address is not valid, or the specified interface index is not valid, or the option is not valid for the cur-
rent filter mode, or the specified hop limit is not in the range 0 <= x<= 255, or the specified traffic class
is not in the range -1 <= x <= 255.
Insufficient memory is available for internal system data structures, or
the number of source addresses exceeds the maximum number of sources allowed, as determined by the tunable
parameters and (for IPv4), or and (for IPv6).
The parameter optname is not a valid socket option for
the level.
The option is not supported for the socket type.
AUTHOR
The socket interfaces to IP were developed by the University of California, Berkeley.
SEE ALSO
ndd(1M), bind(2), getsockopt(2), recv(2), recvmsg(2), send(2), setsockopt(2), socket(2), getipv4sourcefilter(3N), getsourcefilter(3N),
if_nameindex(3N), inet6_opt_init(3N), inet6_rth_space(3N), setipv4sourcefilter(3N), setsourcefilter(3N), inet(7F), ndp(7P).
RFC 2460 Internet Protocol Version 6.
RFC 3493 Basic Socket Interface Extensions for IPv6.
RFC 3542 Advanced Sockets Application Program Interface (API) for IPv6.
RFC 3678 Socket Interface Extensions for Multicast Source Filters.
RFC 3879 Deprecating Site Local Addresses.
RFC 4193 Unique Local IPv6 Unicast Addresses.
RFC 4291 IPv6 Addressing Architecture.
IPv6(7P)