netlink(7) Miscellaneous Information Manual netlink(7)

netlink - communication between kernel and user space (AF_NETLINK)

#include <asm/types.h>
#include <sys/socket.h>
#include <linux/netlink.h>
netlink_socket = socket(AF_NETLINK, socket_type, netlink_family);

Netlink is used to transfer information between the 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 provided only for backward compatibility.

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

netlink_family selects the kernel module or netlink group to communicate with. The currently assigned netlink families are:

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 rtnetlink(7)).
Messages from 1-wire subsystem.
Reserved for user-mode socket protocols.
Transport IPv4 packets from netfilter to user space. Used by ip_queue kernel module. After a long period of being declared obsolete (in favor of the more advanced nfnetlink_queue feature), NETLINK_FIREWALL was removed in Linux 3.5.
Query information about sockets of various protocol families from the kernel (see sock_diag(7)).
An obsolete synonym for NETLINK_SOCK_DIAG.
Netfilter/iptables ULOG.
SELinux event notifications.
Access to FIB lookup from user space.
Kernel connector. See Documentation/driver-api/connector.rst (or /Documentation/connector/connector.* in Linux 5.2 and earlier) in the Linux kernel source tree for further information.
Netfilter subsystem.
SCSI Transports.
Infiniband RDMA.
Transport IPv6 packets from netfilter to user space. Used by ip6_queue kernel module.
DECnet routing messages.
Kernel messages to user space.
Generic netlink family for simplified netlink usage.
Netlink interface to request information about ciphers registered with the kernel crypto API as well as allow configuration of the kernel crypto API.

Netlink messages consist of a byte stream with one or multiple nlmsghdr headers and associated payload. The byte stream should be accessed only with the standard NLMSG_* macros. See netlink(3) for further information.

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

After each nlmsghdr the payload follows.

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 */

nlmsg_type can be one of the standard message types: NLMSG_NOOP message is to be ignored, NLMSG_ERROR message signals an error and the payload contains an nlmsgerr structure, NLMSG_DONE message terminates a multipart message. Error messages get the original request appended, unless the user requests to cap the error message, and get extra error data if requested.

struct nlmsgerr {
    int error;        /* Negative errno or 0 for acknowledgements */
    struct nlmsghdr msg;  /* Message header that caused the error */
     * followed by the message contents
     * unless NETLINK_CAP_ACK was set
     * or the ACK indicates success (error == 0).
     * For example Generic Netlink message with attributes.
     * message length is aligned with NLMSG_ALIGN()
     * followed by TLVs defined in enum nlmsgerr_attrs
     * if NETLINK_EXT_ACK was set

A netlink family usually specifies more message types, see the appropriate manual pages for that, for example, rtnetlink(7) for NETLINK_ROUTE.

Standard flag bits in nlmsg_flags
NLM_F_REQUEST Must be set on all request messages.
NLM_F_MULTI The message is part of a multipart message terminated by NLMSG_DONE .
NLM_F_ACK Request for an acknowledgement on success.
NLM_F_ECHO Echo this request.
Additional flag bits for GET requests
NLM_F_ROOT Return the complete table instead of a single entry.
NLM_F_MATCH Return all entries matching criteria passed in message content. Not implemented yet.
NLM_F_ATOMIC Return an atomic snapshot of the table.
NLM_F_DUMP Convenience macro; equivalent to (NLM_F_ROOT|NLM_F_MATCH).

Note that NLM_F_ATOMIC requires the CAP_NET_ADMIN capability or an effective UID of 0.

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.

nlmsg_seq and nlmsg_pid are used to track messages. nlmsg_pid shows the origin of the message. Note that there isn't a 1:1 relationship between nlmsg_pid and the PID of the process if the message originated from a netlink socket. See the ADDRESS FORMATS section for further information.

Both 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 NLM_F_ACK flag. An acknowledgement is an 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 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 ENOBUFS error returned by recvmsg(2)) and resynchronize.

The sockaddr_nl structure describes a netlink client in user space or in the kernel. A sockaddr_nl can be either unicast (only sent to one peer) or sent to netlink multicast groups (nl_groups not equal 0).

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 */

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, nl_pid is usually the PID of the process owning the destination socket. However, nl_pid identifies a netlink socket, not a process. If a process owns several netlink sockets, then nl_pid can be equal to the process ID only for at most one socket. There are two ways to assign nl_pid to a netlink socket. If the application sets nl_pid before calling bind(2), then it is up to the application to make sure that 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 nl_pid to every netlink socket that the process subsequently creates.

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 bind(2) is called on the socket, the nl_groups field in the 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 nl_groups to a bit mask of the groups it wishes to send to when it calls sendmsg(2) or does a connect(2). Only processes with an effective UID of 0 or the CAP_NET_ADMIN capability may send or listen to a netlink multicast group. Since Linux 2.6.13, messages can't be broadcast to multiple groups. 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 NETLINK_KOBJECT_UEVENT, NETLINK_GENERIC, NETLINK_ROUTE, and NETLINK_SELINUX groups allow other users to receive messages. No groups allow other users to send messages.

To set or get a netlink socket option, call getsockopt(2) to read or setsockopt(2) to write the option with the option level argument set to SOL_NETLINK. Unless otherwise noted, optval is a pointer to an int.

Enable nl_pktinfo control messages for received packets to get the extended destination group number.
Join/leave a group specified by optval.
Retrieve all groups a socket is a member of. optval is a pointer to __u32 and optlen is the size of the array. The array is filled with the full membership set of the socket, and the required array size is returned in optlen.
When not set, netlink_broadcast() only reports ESRCH errors and silently ignore ENOBUFS errors.
This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors.
When set, this socket will receive netlink notifications from all network namespaces that have an nsid assigned into the network namespace where the socket has been opened. The nsid is sent to user space via an ancillary data.
The kernel may fail to allocate the necessary room for the acknowledgement message back to user space. This option trims off the payload of the original netlink message. The netlink message header is still included, so the user can guess from the sequence number which message triggered the acknowledgement.

The socket interface to netlink first appeared 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.

It is often better to use netlink via libnetlink or libnl than via the low-level kernel interface.

This manual page is not complete.

The following example creates a NETLINK_ROUTE netlink socket which will listen to the RTMGRP_LINK (network interface create/delete/up/down events) and RTMGRP_IPV4_IFADDR (IPv4 addresses add/delete events) multicast groups.

struct sockaddr_nl sa;
memset(&sa, 0, sizeof(sa));
sa.nl_family = AF_NETLINK;
bind(fd, (struct sockaddr *) &sa, sizeof(sa));

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

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);

And the last example is about reading netlink message.

int len;
/* 8192 to avoid message truncation on platforms with
   page size > 4096 */
struct nlmsghdr buf[8192/sizeof(struct nlmsghdr)];
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)
    if (nh->nlmsg_type == NLMSG_ERROR)
        /* Do some error handling */
    /* Continue with parsing payload */

cmsg(3), netlink(3), capabilities(7), rtnetlink(7), sock_diag(7)

information about libnetlink

information about libnl

RFC 3549 "Linux Netlink as an IP Services Protocol"

2024-05-02 Linux man-pages 6.9.1