path: root/man7/netlink.7

'\" t
.\" Don't change the first line, it tells man that tbl is needed.
.\" This man page is Copyright (c) 1998 by Andi Kleen.
.\" %%%LICENSE_START(GPL_NOVERSION_ONELINE)
.\" Subject to the GPL.
.\" %%%LICENSE_END
.\" Based on the original comments from Alexey Kuznetsov
.\" Modified 2005-12-27 by Hasso Tepper <hasso@estpak.ee>
.\" $Id: netlink.7,v 1.8 2000/06/22 13:23:00 ak Exp $
.TH NETLINK  7 2012-08-05 "Linux" "Linux Programmer's Manual"
.SH NAME
netlink \- communication between kernel and user space (AF_NETLINK)
.SH SYNOPSIS
.nf
.B #include <asm/types.h>
.B #include <sys/socket.h>
.B #include <linux/netlink.h>

.BI "netlink_socket = socket(AF_NETLINK, " socket_type ", " netlink_family );
.fi
.SH DESCRIPTION
Netlink is used to transfer information between kernel and
user-space processes.
It consists of a standard sockets-based interface for user space
processes and an internal kernel API for kernel modules.
The internal kernel interface is not documented in this manual page.
There is also an obsolete netlink interface
via netlink character devices; this interface is not documented here
and is only provided for backward compatibility.

Netlink is a datagram-oriented service.
Both
.B SOCK_RAW
and
.B SOCK_DGRAM
are valid values for
.IR socket_type .
However, the netlink protocol does not distinguish between datagram
and raw sockets.

.I netlink_family
selects the kernel module or netlink group to communicate with.
The currently assigned netlink families are:
.TP
.B NETLINK_ROUTE
Receives routing and link updates and may be used to modify the routing
tables (both IPv4 and IPv6), IP addresses, link parameters,
neighbor setups, queueing disciplines, traffic classes and
packet classifiers (see
.BR rtnetlink (7)).
.TP
.B NETLINK_W1
Messages from 1-wire subsystem.
.TP
.B NETLINK_USERSOCK
Reserved for user-mode socket protocols.
.TP
.B NETLINK_FIREWALL
Transport IPv4 packets from netfilter to user space.
Used by
.I ip_queue
kernel module.
.TP
.B NETLINK_INET_DIAG
.\" FIXME More details on NETLINK_INET_DIAG needed.
INET socket monitoring.
.TP
.B NETLINK_NFLOG
Netfilter/iptables ULOG.
.TP
.B NETLINK_XFRM
.\" FIXME More details on NETLINK_XFRM needed.
IPsec.
.TP
.B NETLINK_SELINUX
SELinux event notifications.
.TP
.B NETLINK_ISCSI
.\" FIXME More details on NETLINK_ISCSI needed.
Open-iSCSI.
.TP
.B NETLINK_AUDIT
.\" FIXME More details on NETLINK_AUDIT needed.
Auditing.
.TP
.B NETLINK_FIB_LOOKUP
.\" FIXME More details on NETLINK_FIB_LOOKUP needed.
Access to FIB lookup from user space.
.TP
.B NETLINK_CONNECTOR
Kernel connector.
See
.I Documentation/connector/*
in the Linux kernel source tree for further information.
.TP
.B NETLINK_NETFILTER
.\" FIXME More details on NETLINK_NETFILTER needed.
Netfilter subsystem.
.TP
.B NETLINK_IP6_FW
Transport IPv6 packets from netfilter to user space.
Used by
.I ip6_queue
kernel module.
.TP
.B NETLINK_DNRTMSG
DECnet routing messages.
.TP
.B NETLINK_KOBJECT_UEVENT
.\" FIXME More details on NETLINK_KOBJECT_UEVENT needed.
Kernel messages to user space.
.TP
.B NETLINK_GENERIC
Generic netlink family for simplified netlink usage.
.PP
Netlink messages consist of a byte stream with one or multiple
.I nlmsghdr
headers and associated payload.
The byte stream should only be accessed with the standard
.B NLMSG_*
macros.
See
.BR netlink (3)
for further information.

In multipart messages (multiple
.I nlmsghdr
headers with associated payload in one byte stream) the first and all
following headers have the
.B NLM_F_MULTI
flag set, except for the last header which has the type
.BR NLMSG_DONE .

After each
.I nlmsghdr
the payload follows.

.in +4n
.nf
struct nlmsghdr {
    __u32 nlmsg_len;    /* Length of message including header. */
    __u16 nlmsg_type;   /* Type of message content. */
    __u16 nlmsg_flags;  /* Additional flags. */
    __u32 nlmsg_seq;    /* Sequence number. */
    __u32 nlmsg_pid;    /* Sender port ID. */
};
.fi
.in

.I nlmsg_type
can be one of the standard message types:
.B NLMSG_NOOP
message is to be ignored,
.B NLMSG_ERROR
message signals an error and the payload contains an
.I nlmsgerr
structure,
.B NLMSG_DONE
message terminates a multipart message.

.in +4n
.nf
struct nlmsgerr {
    int error;        /* Negative errno or 0 for acknowledgements */
    struct nlmsghdr msg;  /* Message header that caused the error */
};
.fi
.in

A netlink family usually specifies more message types, see the
appropriate manual pages for that, for example,
.BR rtnetlink (7)
for
.BR NETLINK_ROUTE .
.TS
tab(:);
l s
lB l.
Standard flag bits in \fInlmsg_flags\fP
_
NLM_F_REQUEST:Must be set on all request messages.
NLM_F_MULTI:T{
The message is part of a multipart message terminated by
.BR NLMSG_DONE .
T}
NLM_F_ACK:Request for an acknowledgment on success.
NLM_F_ECHO:Echo this request.
.TE
.ad
.sp 1
.\" No right adjustment for text blocks in tables
.na
.TS
tab(:);
l s
lB l.
Additional flag bits for GET requests
_
NLM_F_ROOT:Return the complete table instead of a single entry.
NLM_F_MATCH:T{
Return all entries matching criteria passed in message content.
Not implemented yet.
T}
.\" FIXME NLM_F_ATOMIC is not used any more?
NLM_F_ATOMIC:Return an atomic snapshot of the table.
NLM_F_DUMP:T{
Convenience macro; equivalent to (NLM_F_ROOT|NLM_F_MATCH).
T}
.TE
.ad
.sp 1
Note that
.B NLM_F_ATOMIC
requires the
.B CAP_NET_ADMIN
capability or an effective UID of 0.
.na
.TS
tab(:);
l s
lB l.
Additional flag bits for NEW requests
_
NLM_F_REPLACE:Replace existing matching object.
NLM_F_EXCL:Don't replace if the object already exists.
NLM_F_CREATE:Create object if it doesn't already exist.
NLM_F_APPEND:Add to the end of the object list.
.TE
.ad
.sp 1
.I nlmsg_seq
and
.I nlmsg_pid
are used to track messages.
.I nlmsg_pid
shows the origin of the message.
Note that there isn't a 1:1 relationship between
.I nlmsg_pid
and the PID of the process if the message originated from a netlink
socket.
See the
.B ADDRESS FORMATS
section for further information.

Both
.I nlmsg_seq
and
.I nlmsg_pid
.\" FIXME Explain more about nlmsg_seq and nlmsg_pid.
are opaque to netlink core.

Netlink is not a reliable protocol.
It tries its best to deliver a message to its destination(s),
but may drop messages when an out-of-memory condition or
other error occurs.
For reliable transfer the sender can request an
acknowledgement from the receiver by setting the
.B NLM_F_ACK
flag.
An acknowledgment is an
.B NLMSG_ERROR
packet with the error field set to 0.
The application must generate acknowledgements for
received messages itself.
The kernel tries to send an
.B NLMSG_ERROR
message for every failed packet.
A user process should follow this convention too.

However, reliable transmissions from kernel to user are impossible
in any case.
The kernel can't send a netlink message if the socket buffer is full:
the message will be dropped and the kernel and the user-space process will
no longer have the same view of kernel state.
It is up to the application to detect when this happens (via the
.B ENOBUFS
error returned by
.BR recvmsg (2))
and resynchronize.
.SS Address formats
The
.I sockaddr_nl
structure describes a netlink client in user space or in the kernel.
A
.I sockaddr_nl
can be either unicast (only sent to one peer) or sent to
netlink multicast groups
.RI ( nl_groups
not equal 0).

.in +4n
.nf
struct sockaddr_nl {
    sa_family_t     nl_family;  /* AF_NETLINK */
    unsigned short  nl_pad;     /* Zero. */
    pid_t           nl_pid;     /* Port ID. */
    __u32           nl_groups;  /* Multicast groups mask. */
};
.fi
.in

.I nl_pid
is the unicast address of netlink socket.
It's always 0 if the destination is in the kernel.
For a user-space process,
.I nl_pid
is usually the PID of the process owning the destination socket.
However,
.I nl_pid
identifies a netlink socket, not a process.
If a process owns several netlink
sockets, then
.I nl_pid
can only be equal to the process ID for at most one socket.
There are two ways to assign
.I nl_pid
to a netlink socket.
If the application sets
.I nl_pid
before calling
.BR bind (2),
then it is up to the application to make sure that
.I nl_pid
is unique.
If the application sets it to 0, the kernel takes care of assigning it.
The kernel assigns the process ID to the first netlink socket the process
opens and assigns a unique
.I nl_pid
to every netlink socket that the process subsequently creates.

.I nl_groups
is a bit mask with every bit representing a netlink group number.
Each netlink family has a set of 32 multicast groups.
When
.BR bind (2)
is called on the socket, the
.I nl_groups
field in the
.I sockaddr_nl
should be set to a bit mask of the groups which it wishes to listen to.
The default value for this field is zero which means that no multicasts
will be received.
A socket may multicast messages to any of the multicast groups by setting
.I nl_groups
to a bit mask of the groups it wishes to send to when it calls
.BR sendmsg (2)
or does a
.BR connect (2).
Only processes with an effective UID of 0 or the
.B CAP_NET_ADMIN
capability may send or listen to a netlink multicast group.
Any replies to a message received for a multicast group should be
sent back to the sending PID and the multicast group.
Some Linux kernel subsystems may additionally allow other users
to send and/or receive messages.
As at Linux 3.0, the
.BR NETLINK_KOBJECT_UEVENT ,
.BR NETLINK_GENERIC ,
.BR NETLINK_ROUTE ,
and
.BR NETLINK_SELINUX
groups allow other users to receive messages.
No groups allow other users to send messages.
.SH VERSIONS
The socket interface to netlink is a new feature of Linux 2.2.

Linux 2.0 supported a more primitive device-based netlink interface
(which is still available as a compatibility option).
This obsolete interface is not described here.

NETLINK_SELINUX appeared in Linux 2.6.4.

NETLINK_AUDIT appeared in Linux 2.6.6.

NETLINK_KOBJECT_UEVENT appeared in Linux 2.6.10.

NETLINK_W1 and NETLINK_FIB_LOOKUP appeared in Linux 2.6.13.

NETLINK_INET_DIAG, NETLINK_CONNECTOR and NETLINK_NETFILTER appeared in
Linux 2.6.14.

NETLINK_GENERIC and NETLINK_ISCSI appeared in Linux 2.6.15.
.SH NOTES
It is often better to use netlink via
.I libnetlink
or
.I libnl
than via the low-level kernel interface.
.SH BUGS
This manual page is not complete.
.SH EXAMPLE
The following example creates a
.B NETLINK_ROUTE
netlink socket which will listen to the
.B RTMGRP_LINK
(network interface create/delete/up/down events) and
.B RTMGRP_IPV4_IFADDR
(IPv4 addresses add/delete events) multicast groups.

.in +4n
.nf
struct sockaddr_nl sa;

memset(&sa, 0, sizeof(sa));
sa.nl_family = AF_NETLINK;
sa.nl_groups = RTMGRP_LINK | RTMGRP_IPV4_IFADDR;

fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE);
bind(fd, (struct sockaddr *) &sa, sizeof(sa));
.fi
.in

The next example demonstrates how to send a netlink message to the
kernel (pid 0).
Note that application must take care of message sequence numbers
in order to reliably track acknowledgements.

.in +4n
.nf
struct nlmsghdr *nh;    /* The nlmsghdr with payload to send. */
struct sockaddr_nl sa;
struct iovec iov = { nh, nh\->nlmsg_len };
struct msghdr msg;

msg = { &sa, sizeof(sa), &iov, 1, NULL, 0, 0 };
memset(&sa, 0, sizeof(sa));
sa.nl_family = AF_NETLINK;
nh\->nlmsg_pid = 0;
nh\->nlmsg_seq = ++sequence_number;
/* Request an ack from kernel by setting NLM_F_ACK. */
nh\->nlmsg_flags |= NLM_F_ACK;

sendmsg(fd, &msg, 0);
.fi
.in

And the last example is about reading netlink message.

.in +4n
.nf
int len;
char buf[4096];
struct iovec iov = { buf, sizeof(buf) };
struct sockaddr_nl sa;
struct msghdr msg;
struct nlmsghdr *nh;

msg = { &sa, sizeof(sa), &iov, 1, NULL, 0, 0 };
len = recvmsg(fd, &msg, 0);

for (nh = (struct nlmsghdr *) buf; NLMSG_OK (nh, len);
     nh = NLMSG_NEXT (nh, len)) {
    /* The end of multipart message. */
    if (nh\->nlmsg_type == NLMSG_DONE)
        return;

    if (nh\->nlmsg_type == NLMSG_ERROR)
        /* Do some error handling. */
    ...

    /* Continue with parsing payload. */
    ...
}
.fi
.in
.SH SEE ALSO
.BR cmsg (3),
.BR netlink (3),
.BR capabilities (7),
.BR rtnetlink (7)

.UR ftp://ftp.inr.ac.ru\:/ip-routing\:/iproute2*
information about libnetlink
.UE

.UR http://people.suug.ch\:/~tgr\:/libnl/
information about libnl
.UE

RFC 3549 "Linux Netlink as an IP Services Protocol"