net: WireGuard secure network tunnel

WireGuard is a layer 3 secure networking tunnel made specifically for
the kernel, that aims to be much simpler and easier to audit than IPsec.
Extensive documentation and description of the protocol and
considerations, along with formal proofs of the cryptography, are
available at:

  * https://www.wireguard.com/
  * https://www.wireguard.com/papers/wireguard.pdf

This commit implements WireGuard as a simple network device driver,
accessible in the usual RTNL way used by virtual network drivers. It
makes use of the udp_tunnel APIs, GRO, GSO, NAPI, and the usual set of
networking subsystem APIs. It has a somewhat novel multicore queueing
system designed for maximum throughput and minimal latency of encryption
operations, but it is implemented modestly using workqueues and NAPI.
Configuration is done via generic Netlink, and following a review from
the Netlink maintainer a year ago, several high profile userspace tools
have already implemented the API.

This commit also comes with several different tests, both in-kernel
tests and out-of-kernel tests based on network namespaces, taking profit
of the fact that sockets used by WireGuard intentionally stay in the
namespace the WireGuard interface was originally created, exactly like
the semantics of userspace tun devices. See wireguard.com/netns/ for
pictures and examples.

The source code is fairly short, but rather than combining everything
into a single file, WireGuard is developed as cleanly separable files,
making auditing and comprehension easier. Things are laid out as
follows:

  * noise.[ch], cookie.[ch], messages.h: These implement the bulk of the
    cryptographic aspects of the protocol, and are mostly data-only in
    nature, taking in buffers of bytes and spitting out buffers of
    bytes. They also handle reference counting for their various shared
    pieces of data, like keys and key lists.

  * ratelimiter.[ch]: Used as an integral part of cookie.[ch] for
    ratelimiting certain types of cryptographic operations in accordance
    with particular WireGuard semantics.

  * allowedips.[ch], peerlookup.[ch]: The main lookup structures of
    WireGuard, the former being trie-like with particular semantics, an
    integral part of the design of the protocol, and the latter just
    being nice helper functions around the various hashtables we use.

  * device.[ch]: Implementation of functions for the netdevice and for
    rtnl, responsible for maintaining the life of a given interface and
    wiring it up to the rest of WireGuard.

  * peer.[ch]: Each interface has a list of peers, with helper functions
    available here for creation, destruction, and reference counting.

  * socket.[ch]: Implementation of functions related to udp_socket and
    the general set of kernel socket APIs, for sending and receiving
    ciphertext UDP packets, and taking care of WireGuard-specific sticky
    socket routing semantics for the automatic roaming.

  * netlink.[ch]: Userspace API entry point for configuring WireGuard
    peers and devices. The API has been implemented by several userspace
    tools and network management utility, and the WireGuard project
    distributes the basic wg(8) tool.

  * queueing.[ch]: Shared function on the rx and tx path for handling
    the various queues used in the multicore algorithms.

  * send.c: Handles encrypting outgoing packets in parallel on
    multiple cores, before sending them in order on a single core, via
    workqueues and ring buffers. Also handles sending handshake and cookie
    messages as part of the protocol, in parallel.

  * receive.c: Handles decrypting incoming packets in parallel on
    multiple cores, before passing them off in order to be ingested via
    the rest of the networking subsystem with GRO via the typical NAPI
    poll function. Also handles receiving handshake and cookie messages
    as part of the protocol, in parallel.

  * timers.[ch]: Uses the timer wheel to implement protocol particular
    event timeouts, and gives a set of very simple event-driven entry
    point functions for callers.

  * main.c, version.h: Initialization and deinitialization of the module.

  * selftest/*.h: Runtime unit tests for some of the most security
    sensitive functions.

  * tools/testing/selftests/wireguard/netns.sh: Aforementioned testing
    script using network namespaces.

This commit aims to be as self-contained as possible, implementing
WireGuard as a standalone module not needing much special handling or
coordination from the network subsystem. I expect for future
optimizations to the network stack to positively improve WireGuard, and
vice-versa, but for the time being, this exists as intentionally
standalone.

We introduce a menu option for CONFIG_WIREGUARD, as well as providing a
verbose debug log and self-tests via CONFIG_WIREGUARD_DEBUG.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: David Miller <davem@davemloft.net>
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: linux-crypto@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 642 insertions, 0 deletions

diff --git a/drivers/net/wireguard/netlink.c b/drivers/net/wireguard/netlink.c
new file mode 100644
index 000000000000..0fdbd1c45977
--- /dev/null
+++ b/drivers/net/wireguard/netlink.c
@@ -0,0 +1,642 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2015-2019 Jason A. Donenfeld <Jason@zx2c4.com>. All Rights Reserved.
+ */
+
+#include "netlink.h"
+#include "device.h"
+#include "peer.h"
+#include "socket.h"
+#include "queueing.h"
+#include "messages.h"
+
+#include <uapi/linux/wireguard.h>
+
+#include <linux/if.h>
+#include <net/genetlink.h>
+#include <net/sock.h>
+#include <crypto/algapi.h>
+
+static struct genl_family genl_family;
+
+static const struct nla_policy device_policy[WGDEVICE_A_MAX + 1] = {
+	[WGDEVICE_A_IFINDEX]		= { .type = NLA_U32 },
+	[WGDEVICE_A_IFNAME]		= { .type = NLA_NUL_STRING, .len = IFNAMSIZ - 1 },
+	[WGDEVICE_A_PRIVATE_KEY]	= { .type = NLA_EXACT_LEN, .len = NOISE_PUBLIC_KEY_LEN },
+	[WGDEVICE_A_PUBLIC_KEY]		= { .type = NLA_EXACT_LEN, .len = NOISE_PUBLIC_KEY_LEN },
+	[WGDEVICE_A_FLAGS]		= { .type = NLA_U32 },
+	[WGDEVICE_A_LISTEN_PORT]	= { .type = NLA_U16 },
+	[WGDEVICE_A_FWMARK]		= { .type = NLA_U32 },
+	[WGDEVICE_A_PEERS]		= { .type = NLA_NESTED }
+};
+
+static const struct nla_policy peer_policy[WGPEER_A_MAX + 1] = {
+	[WGPEER_A_PUBLIC_KEY]				= { .type = NLA_EXACT_LEN, .len = NOISE_PUBLIC_KEY_LEN },
+	[WGPEER_A_PRESHARED_KEY]			= { .type = NLA_EXACT_LEN, .len = NOISE_SYMMETRIC_KEY_LEN },
+	[WGPEER_A_FLAGS]				= { .type = NLA_U32 },
+	[WGPEER_A_ENDPOINT]				= { .type = NLA_MIN_LEN, .len = sizeof(struct sockaddr) },
+	[WGPEER_A_PERSISTENT_KEEPALIVE_INTERVAL]	= { .type = NLA_U16 },
+	[WGPEER_A_LAST_HANDSHAKE_TIME]			= { .type = NLA_EXACT_LEN, .len = sizeof(struct __kernel_timespec) },
+	[WGPEER_A_RX_BYTES]				= { .type = NLA_U64 },
+	[WGPEER_A_TX_BYTES]				= { .type = NLA_U64 },
+	[WGPEER_A_ALLOWEDIPS]				= { .type = NLA_NESTED },
+	[WGPEER_A_PROTOCOL_VERSION]			= { .type = NLA_U32 }
+};
+
+static const struct nla_policy allowedip_policy[WGALLOWEDIP_A_MAX + 1] = {
+	[WGALLOWEDIP_A_FAMILY]		= { .type = NLA_U16 },
+	[WGALLOWEDIP_A_IPADDR]		= { .type = NLA_MIN_LEN, .len = sizeof(struct in_addr) },
+	[WGALLOWEDIP_A_CIDR_MASK]	= { .type = NLA_U8 }
+};
+
+static struct wg_device *lookup_interface(struct nlattr **attrs,
+					  struct sk_buff *skb)
+{
+	struct net_device *dev = NULL;
+
+	if (!attrs[WGDEVICE_A_IFINDEX] == !attrs[WGDEVICE_A_IFNAME])
+		return ERR_PTR(-EBADR);
+	if (attrs[WGDEVICE_A_IFINDEX])
+		dev = dev_get_by_index(sock_net(skb->sk),
+				       nla_get_u32(attrs[WGDEVICE_A_IFINDEX]));
+	else if (attrs[WGDEVICE_A_IFNAME])
+		dev = dev_get_by_name(sock_net(skb->sk),
+				      nla_data(attrs[WGDEVICE_A_IFNAME]));
+	if (!dev)
+		return ERR_PTR(-ENODEV);
+	if (!dev->rtnl_link_ops || !dev->rtnl_link_ops->kind ||
+	    strcmp(dev->rtnl_link_ops->kind, KBUILD_MODNAME)) {
+		dev_put(dev);
+		return ERR_PTR(-EOPNOTSUPP);
+	}
+	return netdev_priv(dev);
+}
+
+static int get_allowedips(struct sk_buff *skb, const u8 *ip, u8 cidr,
+			  int family)
+{
+	struct nlattr *allowedip_nest;
+
+	allowedip_nest = nla_nest_start(skb, 0);
+	if (!allowedip_nest)
+		return -EMSGSIZE;
+
+	if (nla_put_u8(skb, WGALLOWEDIP_A_CIDR_MASK, cidr) ||
+	    nla_put_u16(skb, WGALLOWEDIP_A_FAMILY, family) ||
+	    nla_put(skb, WGALLOWEDIP_A_IPADDR, family == AF_INET6 ?
+		    sizeof(struct in6_addr) : sizeof(struct in_addr), ip)) {
+		nla_nest_cancel(skb, allowedip_nest);
+		return -EMSGSIZE;
+	}
+
+	nla_nest_end(skb, allowedip_nest);
+	return 0;
+}
+
+struct dump_ctx {
+	struct wg_device *wg;
+	struct wg_peer *next_peer;
+	u64 allowedips_seq;
+	struct allowedips_node *next_allowedip;
+};
+
+#define DUMP_CTX(cb) ((struct dump_ctx *)(cb)->args)
+
+static int
+get_peer(struct wg_peer *peer, struct sk_buff *skb, struct dump_ctx *ctx)
+{
+
+	struct nlattr *allowedips_nest, *peer_nest = nla_nest_start(skb, 0);
+	struct allowedips_node *allowedips_node = ctx->next_allowedip;
+	bool fail;
+
+	if (!peer_nest)
+		return -EMSGSIZE;
+
+	down_read(&peer->handshake.lock);
+	fail = nla_put(skb, WGPEER_A_PUBLIC_KEY, NOISE_PUBLIC_KEY_LEN,
+		       peer->handshake.remote_static);
+	up_read(&peer->handshake.lock);
+	if (fail)
+		goto err;
+
+	if (!allowedips_node) {
+		const struct __kernel_timespec last_handshake = {
+			.tv_sec = peer->walltime_last_handshake.tv_sec,
+			.tv_nsec = peer->walltime_last_handshake.tv_nsec
+		};
+
+		down_read(&peer->handshake.lock);
+		fail = nla_put(skb, WGPEER_A_PRESHARED_KEY,
+			       NOISE_SYMMETRIC_KEY_LEN,
+			       peer->handshake.preshared_key);
+		up_read(&peer->handshake.lock);
+		if (fail)
+			goto err;
+
+		if (nla_put(skb, WGPEER_A_LAST_HANDSHAKE_TIME,
+			    sizeof(last_handshake), &last_handshake) ||
+		    nla_put_u16(skb, WGPEER_A_PERSISTENT_KEEPALIVE_INTERVAL,
+				peer->persistent_keepalive_interval) ||
+		    nla_put_u64_64bit(skb, WGPEER_A_TX_BYTES, peer->tx_bytes,
+				      WGPEER_A_UNSPEC) ||
+		    nla_put_u64_64bit(skb, WGPEER_A_RX_BYTES, peer->rx_bytes,
+				      WGPEER_A_UNSPEC) ||
+		    nla_put_u32(skb, WGPEER_A_PROTOCOL_VERSION, 1))
+			goto err;
+
+		read_lock_bh(&peer->endpoint_lock);
+		if (peer->endpoint.addr.sa_family == AF_INET)
+			fail = nla_put(skb, WGPEER_A_ENDPOINT,
+				       sizeof(peer->endpoint.addr4),
+				       &peer->endpoint.addr4);
+		else if (peer->endpoint.addr.sa_family == AF_INET6)
+			fail = nla_put(skb, WGPEER_A_ENDPOINT,
+				       sizeof(peer->endpoint.addr6),
+				       &peer->endpoint.addr6);
+		read_unlock_bh(&peer->endpoint_lock);
+		if (fail)
+			goto err;
+		allowedips_node =
+			list_first_entry_or_null(&peer->allowedips_list,
+					struct allowedips_node, peer_list);
+	}
+	if (!allowedips_node)
+		goto no_allowedips;
+	if (!ctx->allowedips_seq)
+		ctx->allowedips_seq = peer->device->peer_allowedips.seq;
+	else if (ctx->allowedips_seq != peer->device->peer_allowedips.seq)
+		goto no_allowedips;
+
+	allowedips_nest = nla_nest_start(skb, WGPEER_A_ALLOWEDIPS);
+	if (!allowedips_nest)
+		goto err;
+
+	list_for_each_entry_from(allowedips_node, &peer->allowedips_list,
+				 peer_list) {
+		u8 cidr, ip[16] __aligned(__alignof(u64));
+		int family;
+
+		family = wg_allowedips_read_node(allowedips_node, ip, &cidr);
+		if (get_allowedips(skb, ip, cidr, family)) {
+			nla_nest_end(skb, allowedips_nest);
+			nla_nest_end(skb, peer_nest);
+			ctx->next_allowedip = allowedips_node;
+			return -EMSGSIZE;
+		}
+	}
+	nla_nest_end(skb, allowedips_nest);
+no_allowedips:
+	nla_nest_end(skb, peer_nest);
+	ctx->next_allowedip = NULL;
+	ctx->allowedips_seq = 0;
+	return 0;
+err:
+	nla_nest_cancel(skb, peer_nest);
+	return -EMSGSIZE;
+}
+
+static int wg_get_device_start(struct netlink_callback *cb)
+{
+	struct wg_device *wg;
+
+	wg = lookup_interface(genl_dumpit_info(cb)->attrs, cb->skb);
+	if (IS_ERR(wg))
+		return PTR_ERR(wg);
+	DUMP_CTX(cb)->wg = wg;
+	return 0;
+}
+
+static int wg_get_device_dump(struct sk_buff *skb, struct netlink_callback *cb)
+{
+	struct wg_peer *peer, *next_peer_cursor;
+	struct dump_ctx *ctx = DUMP_CTX(cb);
+	struct wg_device *wg = ctx->wg;
+	struct nlattr *peers_nest;
+	int ret = -EMSGSIZE;
+	bool done = true;
+	void *hdr;
+
+	rtnl_lock();
+	mutex_lock(&wg->device_update_lock);
+	cb->seq = wg->device_update_gen;
+	next_peer_cursor = ctx->next_peer;
+
+	hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
+			  &genl_family, NLM_F_MULTI, WG_CMD_GET_DEVICE);
+	if (!hdr)
+		goto out;
+	genl_dump_check_consistent(cb, hdr);
+
+	if (!ctx->next_peer) {
+		if (nla_put_u16(skb, WGDEVICE_A_LISTEN_PORT,
+				wg->incoming_port) ||
+		    nla_put_u32(skb, WGDEVICE_A_FWMARK, wg->fwmark) ||
+		    nla_put_u32(skb, WGDEVICE_A_IFINDEX, wg->dev->ifindex) ||
+		    nla_put_string(skb, WGDEVICE_A_IFNAME, wg->dev->name))
+			goto out;
+
+		down_read(&wg->static_identity.lock);
+		if (wg->static_identity.has_identity) {
+			if (nla_put(skb, WGDEVICE_A_PRIVATE_KEY,
+				    NOISE_PUBLIC_KEY_LEN,
+				    wg->static_identity.static_private) ||
+			    nla_put(skb, WGDEVICE_A_PUBLIC_KEY,
+				    NOISE_PUBLIC_KEY_LEN,
+				    wg->static_identity.static_public)) {
+				up_read(&wg->static_identity.lock);
+				goto out;
+			}
+		}
+		up_read(&wg->static_identity.lock);
+	}
+
+	peers_nest = nla_nest_start(skb, WGDEVICE_A_PEERS);
+	if (!peers_nest)
+		goto out;
+	ret = 0;
+	/* If the last cursor was removed via list_del_init in peer_remove, then
+	 * we just treat this the same as there being no more peers left. The
+	 * reason is that seq_nr should indicate to userspace that this isn't a
+	 * coherent dump anyway, so they'll try again.
+	 */
+	if (list_empty(&wg->peer_list) ||
+	    (ctx->next_peer && list_empty(&ctx->next_peer->peer_list))) {
+		nla_nest_cancel(skb, peers_nest);
+		goto out;
+	}
+	lockdep_assert_held(&wg->device_update_lock);
+	peer = list_prepare_entry(ctx->next_peer, &wg->peer_list, peer_list);
+	list_for_each_entry_continue(peer, &wg->peer_list, peer_list) {
+		if (get_peer(peer, skb, ctx)) {
+			done = false;
+			break;
+		}
+		next_peer_cursor = peer;
+	}
+	nla_nest_end(skb, peers_nest);
+
+out:
+	if (!ret && !done && next_peer_cursor)
+		wg_peer_get(next_peer_cursor);
+	wg_peer_put(ctx->next_peer);
+	mutex_unlock(&wg->device_update_lock);
+	rtnl_unlock();
+
+	if (ret) {
+		genlmsg_cancel(skb, hdr);
+		return ret;
+	}
+	genlmsg_end(skb, hdr);
+	if (done) {
+		ctx->next_peer = NULL;
+		return 0;
+	}
+	ctx->next_peer = next_peer_cursor;
+	return skb->len;
+
+	/* At this point, we can't really deal ourselves with safely zeroing out
+	 * the private key material after usage. This will need an additional API
+	 * in the kernel for marking skbs as zero_on_free.
+	 */
+}
+
+static int wg_get_device_done(struct netlink_callback *cb)
+{
+	struct dump_ctx *ctx = DUMP_CTX(cb);
+
+	if (ctx->wg)
+		dev_put(ctx->wg->dev);
+	wg_peer_put(ctx->next_peer);
+	return 0;
+}
+
+static int set_port(struct wg_device *wg, u16 port)
+{
+	struct wg_peer *peer;
+
+	if (wg->incoming_port == port)
+		return 0;
+	list_for_each_entry(peer, &wg->peer_list, peer_list)
+		wg_socket_clear_peer_endpoint_src(peer);
+	if (!netif_running(wg->dev)) {
+		wg->incoming_port = port;
+		return 0;
+	}
+	return wg_socket_init(wg, port);
+}
+
+static int set_allowedip(struct wg_peer *peer, struct nlattr **attrs)
+{
+	int ret = -EINVAL;
+	u16 family;
+	u8 cidr;
+
+	if (!attrs[WGALLOWEDIP_A_FAMILY] || !attrs[WGALLOWEDIP_A_IPADDR] ||
+	    !attrs[WGALLOWEDIP_A_CIDR_MASK])
+		return ret;
+	family = nla_get_u16(attrs[WGALLOWEDIP_A_FAMILY]);
+	cidr = nla_get_u8(attrs[WGALLOWEDIP_A_CIDR_MASK]);
+
+	if (family == AF_INET && cidr <= 32 &&
+	    nla_len(attrs[WGALLOWEDIP_A_IPADDR]) == sizeof(struct in_addr))
+		ret = wg_allowedips_insert_v4(
+			&peer->device->peer_allowedips,
+			nla_data(attrs[WGALLOWEDIP_A_IPADDR]), cidr, peer,
+			&peer->device->device_update_lock);
+	else if (family == AF_INET6 && cidr <= 128 &&
+		 nla_len(attrs[WGALLOWEDIP_A_IPADDR]) == sizeof(struct in6_addr))
+		ret = wg_allowedips_insert_v6(
+			&peer->device->peer_allowedips,
+			nla_data(attrs[WGALLOWEDIP_A_IPADDR]), cidr, peer,
+			&peer->device->device_update_lock);
+
+	return ret;
+}
+
+static int set_peer(struct wg_device *wg, struct nlattr **attrs)
+{
+	u8 *public_key = NULL, *preshared_key = NULL;
+	struct wg_peer *peer = NULL;
+	u32 flags = 0;
+	int ret;
+
+	ret = -EINVAL;
+	if (attrs[WGPEER_A_PUBLIC_KEY] &&
+	    nla_len(attrs[WGPEER_A_PUBLIC_KEY]) == NOISE_PUBLIC_KEY_LEN)
+		public_key = nla_data(attrs[WGPEER_A_PUBLIC_KEY]);
+	else
+		goto out;
+	if (attrs[WGPEER_A_PRESHARED_KEY] &&
+	    nla_len(attrs[WGPEER_A_PRESHARED_KEY]) == NOISE_SYMMETRIC_KEY_LEN)
+		preshared_key = nla_data(attrs[WGPEER_A_PRESHARED_KEY]);
+
+	if (attrs[WGPEER_A_FLAGS])
+		flags = nla_get_u32(attrs[WGPEER_A_FLAGS]);
+	ret = -EOPNOTSUPP;
+	if (flags & ~__WGPEER_F_ALL)
+		goto out;
+
+	ret = -EPFNOSUPPORT;
+	if (attrs[WGPEER_A_PROTOCOL_VERSION]) {
+		if (nla_get_u32(attrs[WGPEER_A_PROTOCOL_VERSION]) != 1)
+			goto out;
+	}
+
+	peer = wg_pubkey_hashtable_lookup(wg->peer_hashtable,
+					  nla_data(attrs[WGPEER_A_PUBLIC_KEY]));
+	ret = 0;
+	if (!peer) { /* Peer doesn't exist yet. Add a new one. */
+		if (flags & (WGPEER_F_REMOVE_ME | WGPEER_F_UPDATE_ONLY))
+			goto out;
+
+		/* The peer is new, so there aren't allowed IPs to remove. */
+		flags &= ~WGPEER_F_REPLACE_ALLOWEDIPS;
+
+		down_read(&wg->static_identity.lock);
+		if (wg->static_identity.has_identity &&
+		    !memcmp(nla_data(attrs[WGPEER_A_PUBLIC_KEY]),
+			    wg->static_identity.static_public,
+			    NOISE_PUBLIC_KEY_LEN)) {
+			/* We silently ignore peers that have the same public
+			 * key as the device. The reason we do it silently is
+			 * that we'd like for people to be able to reuse the
+			 * same set of API calls across peers.
+			 */
+			up_read(&wg->static_identity.lock);
+			ret = 0;
+			goto out;
+		}
+		up_read(&wg->static_identity.lock);
+
+		peer = wg_peer_create(wg, public_key, preshared_key);
+		if (IS_ERR(peer)) {
+			/* Similar to the above, if the key is invalid, we skip
+			 * it without fanfare, so that services don't need to
+			 * worry about doing key validation themselves.
+			 */
+			ret = PTR_ERR(peer) == -EKEYREJECTED ? 0 : PTR_ERR(peer);
+			peer = NULL;
+			goto out;
+		}
+		/* Take additional reference, as though we've just been
+		 * looked up.
+		 */
+		wg_peer_get(peer);
+	}
+
+	if (flags & WGPEER_F_REMOVE_ME) {
+		wg_peer_remove(peer);
+		goto out;
+	}
+
+	if (preshared_key) {
+		down_write(&peer->handshake.lock);
+		memcpy(&peer->handshake.preshared_key, preshared_key,
+		       NOISE_SYMMETRIC_KEY_LEN);
+		up_write(&peer->handshake.lock);
+	}
+
+	if (attrs[WGPEER_A_ENDPOINT]) {
+		struct sockaddr *addr = nla_data(attrs[WGPEER_A_ENDPOINT]);
+		size_t len = nla_len(attrs[WGPEER_A_ENDPOINT]);
+
+		if ((len == sizeof(struct sockaddr_in) &&
+		     addr->sa_family == AF_INET) ||
+		    (len == sizeof(struct sockaddr_in6) &&
+		     addr->sa_family == AF_INET6)) {
+			struct endpoint endpoint = { { { 0 } } };
+
+			memcpy(&endpoint.addr, addr, len);
+			wg_socket_set_peer_endpoint(peer, &endpoint);
+		}
+	}
+
+	if (flags & WGPEER_F_REPLACE_ALLOWEDIPS)
+		wg_allowedips_remove_by_peer(&wg->peer_allowedips, peer,
+					     &wg->device_update_lock);
+
+	if (attrs[WGPEER_A_ALLOWEDIPS]) {
+		struct nlattr *attr, *allowedip[WGALLOWEDIP_A_MAX + 1];
+		int rem;
+
+		nla_for_each_nested(attr, attrs[WGPEER_A_ALLOWEDIPS], rem) {
+			ret = nla_parse_nested(allowedip, WGALLOWEDIP_A_MAX,
+					       attr, allowedip_policy, NULL);
+			if (ret < 0)
+				goto out;
+			ret = set_allowedip(peer, allowedip);
+			if (ret < 0)
+				goto out;
+		}
+	}
+
+	if (attrs[WGPEER_A_PERSISTENT_KEEPALIVE_INTERVAL]) {
+		const u16 persistent_keepalive_interval = nla_get_u16(
+				attrs[WGPEER_A_PERSISTENT_KEEPALIVE_INTERVAL]);
+		const bool send_keepalive =
+			!peer->persistent_keepalive_interval &&
+			persistent_keepalive_interval &&
+			netif_running(wg->dev);
+
+		peer->persistent_keepalive_interval = persistent_keepalive_interval;
+		if (send_keepalive)
+			wg_packet_send_keepalive(peer);
+	}
+
+	if (netif_running(wg->dev))
+		wg_packet_send_staged_packets(peer);
+
+out:
+	wg_peer_put(peer);
+	if (attrs[WGPEER_A_PRESHARED_KEY])
+		memzero_explicit(nla_data(attrs[WGPEER_A_PRESHARED_KEY]),
+				 nla_len(attrs[WGPEER_A_PRESHARED_KEY]));
+	return ret;
+}
+
+static int wg_set_device(struct sk_buff *skb, struct genl_info *info)
+{
+	struct wg_device *wg = lookup_interface(info->attrs, skb);
+	u32 flags = 0;
+	int ret;
+
+	if (IS_ERR(wg)) {
+		ret = PTR_ERR(wg);
+		goto out_nodev;
+	}
+
+	rtnl_lock();
+	mutex_lock(&wg->device_update_lock);
+
+	if (info->attrs[WGDEVICE_A_FLAGS])
+		flags = nla_get_u32(info->attrs[WGDEVICE_A_FLAGS]);
+	ret = -EOPNOTSUPP;
+	if (flags & ~__WGDEVICE_F_ALL)
+		goto out;
+
+	ret = -EPERM;
+	if ((info->attrs[WGDEVICE_A_LISTEN_PORT] ||
+	     info->attrs[WGDEVICE_A_FWMARK]) &&
+	    !ns_capable(wg->creating_net->user_ns, CAP_NET_ADMIN))
+		goto out;
+
+	++wg->device_update_gen;
+
+	if (info->attrs[WGDEVICE_A_FWMARK]) {
+		struct wg_peer *peer;
+
+		wg->fwmark = nla_get_u32(info->attrs[WGDEVICE_A_FWMARK]);
+		list_for_each_entry(peer, &wg->peer_list, peer_list)
+			wg_socket_clear_peer_endpoint_src(peer);
+	}
+
+	if (info->attrs[WGDEVICE_A_LISTEN_PORT]) {
+		ret = set_port(wg,
+			nla_get_u16(info->attrs[WGDEVICE_A_LISTEN_PORT]));
+		if (ret)
+			goto out;
+	}
+
+	if (flags & WGDEVICE_F_REPLACE_PEERS)
+		wg_peer_remove_all(wg);
+
+	if (info->attrs[WGDEVICE_A_PRIVATE_KEY] &&
+	    nla_len(info->attrs[WGDEVICE_A_PRIVATE_KEY]) ==
+		    NOISE_PUBLIC_KEY_LEN) {
+		u8 *private_key = nla_data(info->attrs[WGDEVICE_A_PRIVATE_KEY]);
+		u8 public_key[NOISE_PUBLIC_KEY_LEN];
+		struct wg_peer *peer, *temp;
+
+		if (!crypto_memneq(wg->static_identity.static_private,
+				   private_key, NOISE_PUBLIC_KEY_LEN))
+			goto skip_set_private_key;
+
+		/* We remove before setting, to prevent race, which means doing
+		 * two 25519-genpub ops.
+		 */
+		if (curve25519_generate_public(public_key, private_key)) {
+			peer = wg_pubkey_hashtable_lookup(wg->peer_hashtable,
+							  public_key);
+			if (peer) {
+				wg_peer_put(peer);
+				wg_peer_remove(peer);
+			}
+		}
+
+		down_write(&wg->static_identity.lock);
+		wg_noise_set_static_identity_private_key(&wg->static_identity,
+							 private_key);
+		list_for_each_entry_safe(peer, temp, &wg->peer_list,
+					 peer_list) {
+			if (wg_noise_precompute_static_static(peer))
+				wg_noise_expire_current_peer_keypairs(peer);
+			else
+				wg_peer_remove(peer);
+		}
+		wg_cookie_checker_precompute_device_keys(&wg->cookie_checker);
+		up_write(&wg->static_identity.lock);
+	}
+skip_set_private_key:
+
+	if (info->attrs[WGDEVICE_A_PEERS]) {
+		struct nlattr *attr, *peer[WGPEER_A_MAX + 1];
+		int rem;
+
+		nla_for_each_nested(attr, info->attrs[WGDEVICE_A_PEERS], rem) {
+			ret = nla_parse_nested(peer, WGPEER_A_MAX, attr,
+					       peer_policy, NULL);
+			if (ret < 0)
+				goto out;
+			ret = set_peer(wg, peer);
+			if (ret < 0)
+				goto out;
+		}
+	}
+	ret = 0;
+
+out:
+	mutex_unlock(&wg->device_update_lock);
+	rtnl_unlock();
+	dev_put(wg->dev);
+out_nodev:
+	if (info->attrs[WGDEVICE_A_PRIVATE_KEY])
+		memzero_explicit(nla_data(info->attrs[WGDEVICE_A_PRIVATE_KEY]),
+				 nla_len(info->attrs[WGDEVICE_A_PRIVATE_KEY]));
+	return ret;
+}
+
+static const struct genl_ops genl_ops[] = {
+	{
+		.cmd = WG_CMD_GET_DEVICE,
+		.start = wg_get_device_start,
+		.dumpit = wg_get_device_dump,
+		.done = wg_get_device_done,
+		.flags = GENL_UNS_ADMIN_PERM
+	}, {
+		.cmd = WG_CMD_SET_DEVICE,
+		.doit = wg_set_device,
+		.flags = GENL_UNS_ADMIN_PERM
+	}
+};
+
+static struct genl_family genl_family __ro_after_init = {
+	.ops = genl_ops,
+	.n_ops = ARRAY_SIZE(genl_ops),
+	.name = WG_GENL_NAME,
+	.version = WG_GENL_VERSION,
+	.maxattr = WGDEVICE_A_MAX,
+	.module = THIS_MODULE,
+	.policy = device_policy,
+	.netnsok = true
+};
+
+int __init wg_genetlink_init(void)
+{
+	return genl_register_family(&genl_family);
+}
+
+void __exit wg_genetlink_uninit(void)
+{
+	genl_unregister_family(&genl_family);
+}