2 * Copyright (c) 2011 Nicira, Inc.
3 * Copyright (c) 2012 Cisco Systems, Inc.
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of version 2 of the GNU General Public
7 * License as published by the Free Software Foundation.
9 * This program is distributed in the hope that it will be useful, but
10 * WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * General Public License for more details.
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
20 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
22 #include <linux/version.h>
23 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,26)
27 #include <linux/net.h>
28 #include <linux/rculist.h>
29 #include <linux/udp.h>
39 #define VXLAN_FLAGS 0x08000000 /* struct vxlanhdr.vx_flags required value. */
42 * struct vxlanhdr - VXLAN header
43 * @vx_flags: Must have the exact value %VXLAN_FLAGS.
44 * @vx_vni: VXLAN Network Identifier (VNI) in top 24 bits, low 8 bits zeroed.
51 #define VXLAN_HLEN (sizeof(struct udphdr) + sizeof(struct vxlanhdr))
53 static inline int vxlan_hdr_len(const struct ovs_key_ipv4_tunnel *tun_key)
59 * struct vxlan_port - Keeps track of open UDP ports
60 * @list: list element.
61 * @vport: vport for the tunnel.
62 * @socket: The socket created for this port number.
65 struct list_head list;
67 struct socket *vxlan_rcv_socket;
71 static LIST_HEAD(vxlan_ports);
73 static struct vxlan_port *vxlan_find_port(struct net *net, __be16 port)
75 struct vxlan_port *vxlan_port;
77 list_for_each_entry_rcu(vxlan_port, &vxlan_ports, list) {
78 struct tnl_vport *tnl_vport = tnl_vport_priv(vxlan_port->vport);
80 if (tnl_vport->dst_port == port &&
81 net_eq(sock_net(vxlan_port->vxlan_rcv_socket->sk), net))
88 static inline struct vxlanhdr *vxlan_hdr(const struct sk_buff *skb)
90 return (struct vxlanhdr *)(udp_hdr(skb) + 1);
93 static void vxlan_build_header(const struct vport *vport,
97 struct tnl_vport *tnl_vport = tnl_vport_priv(vport);
98 struct udphdr *udph = udp_hdr(skb);
99 struct vxlanhdr *vxh = (struct vxlanhdr *)(udph + 1);
100 const struct ovs_key_ipv4_tunnel *tun_key = OVS_CB(skb)->tun_key;
102 udph->dest = tnl_vport->dst_port;
103 udph->source = htons(ovs_tnl_get_src_port(skb));
105 udph->len = htons(skb->len - skb_transport_offset(skb));
107 vxh->vx_flags = htonl(VXLAN_FLAGS);
108 vxh->vx_vni = htonl(be64_to_cpu(tun_key->tun_id) << 8);
111 * Allow our local IP stack to fragment the outer packet even if the
112 * DF bit is set as a last resort. We also need to force selection of
113 * an IP ID here because Linux will otherwise leave it at 0 if the
114 * packet originally had DF set.
117 __ip_select_ident(ip_hdr(skb), skb_dst(skb), 0);
120 /* Called with rcu_read_lock and BH disabled. */
121 static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
123 struct vxlan_port *vxlan_vport;
124 struct vxlanhdr *vxh;
126 struct ovs_key_ipv4_tunnel tun_key;
129 vxlan_vport = vxlan_find_port(dev_net(skb->dev), udp_hdr(skb)->dest);
130 if (unlikely(!vxlan_vport))
133 if (unlikely(!pskb_may_pull(skb, VXLAN_HLEN + ETH_HLEN)))
136 vxh = vxlan_hdr(skb);
137 if (unlikely(vxh->vx_flags != htonl(VXLAN_FLAGS) ||
138 vxh->vx_vni & htonl(0xff)))
141 __skb_pull(skb, VXLAN_HLEN);
142 skb_postpull_rcsum(skb, skb_transport_header(skb), VXLAN_HLEN + ETH_HLEN);
144 key = cpu_to_be64(ntohl(vxh->vx_vni) >> 8);
146 /* Save outer tunnel values */
148 tnl_tun_key_init(&tun_key, iph, key, OVS_TNL_F_KEY);
149 OVS_CB(skb)->tun_key = &tun_key;
151 ovs_tnl_rcv(vxlan_vport->vport, skb);
160 /* Random value. Irrelevant as long as it's not 0 since we set the handler. */
161 #define UDP_ENCAP_VXLAN 1
162 static int vxlan_socket_init(struct vxlan_port *vxlan_port, struct net *net)
165 struct sockaddr_in sin;
166 struct tnl_vport *tnl_vport = tnl_vport_priv(vxlan_port->vport);
168 err = sock_create_kern(AF_INET, SOCK_DGRAM, 0,
169 &vxlan_port->vxlan_rcv_socket);
173 /* release net ref. */
174 sk_change_net(vxlan_port->vxlan_rcv_socket->sk, net);
176 sin.sin_family = AF_INET;
177 sin.sin_addr.s_addr = htonl(INADDR_ANY);
178 sin.sin_port = tnl_vport->dst_port;
180 err = kernel_bind(vxlan_port->vxlan_rcv_socket, (struct sockaddr *)&sin,
181 sizeof(struct sockaddr_in));
185 udp_sk(vxlan_port->vxlan_rcv_socket->sk)->encap_type = UDP_ENCAP_VXLAN;
186 udp_sk(vxlan_port->vxlan_rcv_socket->sk)->encap_rcv = vxlan_rcv;
193 sk_release_kernel(vxlan_port->vxlan_rcv_socket->sk);
195 pr_warn("cannot register vxlan protocol handler\n");
199 static void free_port_rcu(struct rcu_head *rcu)
201 struct vxlan_port *vxlan_port = container_of(rcu,
202 struct vxlan_port, rcu);
207 static void vxlan_tunnel_release(struct vxlan_port *vxlan_port)
212 list_del_rcu(&vxlan_port->list);
214 sk_release_kernel(vxlan_port->vxlan_rcv_socket->sk);
215 call_rcu(&vxlan_port->rcu, free_port_rcu);
218 static int vxlan_tunnel_setup(struct net *net, struct vport *vport,
219 struct nlattr *options)
221 struct vxlan_port *vxlan_port;
222 struct tnl_vport *tnl_vport = tnl_vport_priv(vport);
232 a = nla_find_nested(options, OVS_TUNNEL_ATTR_DST_PORT);
233 if (a && nla_len(a) == sizeof(u16)) {
234 dst_port = nla_get_u16(a);
236 /* Require destination port from userspace. */
241 /* Verify if we already have a socket created for this port */
242 vxlan_port = vxlan_find_port(net, htons(dst_port));
248 /* Add a new socket for this port */
249 vxlan_port = kzalloc(sizeof(struct vxlan_port), GFP_KERNEL);
255 tnl_vport->dst_port = htons(dst_port);
256 vxlan_port->vport = vport;
257 list_add_tail_rcu(&vxlan_port->list, &vxlan_ports);
259 err = vxlan_socket_init(vxlan_port, net);
266 list_del_rcu(&vxlan_port->list);
272 static int vxlan_get_options(const struct vport *vport, struct sk_buff *skb)
274 const struct tnl_vport *tnl_vport = tnl_vport_priv(vport);
276 if (nla_put_u16(skb, OVS_TUNNEL_ATTR_DST_PORT, ntohs(tnl_vport->dst_port)))
281 static const struct tnl_ops ovs_vxlan_tnl_ops = {
282 .ipproto = IPPROTO_UDP,
283 .hdr_len = vxlan_hdr_len,
284 .build_header = vxlan_build_header,
287 static void vxlan_tnl_destroy(struct vport *vport)
289 struct vxlan_port *vxlan_port;
290 struct tnl_vport *tnl_vport = tnl_vport_priv(vport);
292 vxlan_port = vxlan_find_port(ovs_dp_get_net(vport->dp),
293 tnl_vport->dst_port);
295 vxlan_tunnel_release(vxlan_port);
296 ovs_tnl_destroy(vport);
299 static struct vport *vxlan_tnl_create(const struct vport_parms *parms)
304 vport = ovs_tnl_create(parms, &ovs_vxlan_vport_ops, &ovs_vxlan_tnl_ops);
308 err = vxlan_tunnel_setup(ovs_dp_get_net(parms->dp), vport,
311 ovs_tnl_destroy(vport);
318 const struct vport_ops ovs_vxlan_vport_ops = {
319 .type = OVS_VPORT_TYPE_VXLAN,
320 .flags = VPORT_F_TUN_ID,
321 .create = vxlan_tnl_create,
322 .destroy = vxlan_tnl_destroy,
323 .get_name = ovs_tnl_get_name,
324 .get_options = vxlan_get_options,
325 .send = ovs_tnl_send,
328 #warning VXLAN tunneling will not be available on kernels before 2.6.26
329 #endif /* Linux kernel < 2.6.26 */