xref: /openbmc/linux/net/openvswitch/vport-vxlan.c (revision b34e08d5)
1 /*
2  * Copyright (c) 2013 Nicira, Inc.
3  * Copyright (c) 2013 Cisco Systems, Inc.
4  *
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.
8  *
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.
13  *
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
17  * 02110-1301, USA
18  */
19 
20 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
21 
22 #include <linux/in.h>
23 #include <linux/ip.h>
24 #include <linux/net.h>
25 #include <linux/rculist.h>
26 #include <linux/udp.h>
27 
28 #include <net/icmp.h>
29 #include <net/ip.h>
30 #include <net/udp.h>
31 #include <net/ip_tunnels.h>
32 #include <net/rtnetlink.h>
33 #include <net/route.h>
34 #include <net/dsfield.h>
35 #include <net/inet_ecn.h>
36 #include <net/net_namespace.h>
37 #include <net/netns/generic.h>
38 #include <net/vxlan.h>
39 
40 #include "datapath.h"
41 #include "vport.h"
42 
43 /**
44  * struct vxlan_port - Keeps track of open UDP ports
45  * @vs: vxlan_sock created for the port.
46  * @name: vport name.
47  */
48 struct vxlan_port {
49 	struct vxlan_sock *vs;
50 	char name[IFNAMSIZ];
51 };
52 
53 static inline struct vxlan_port *vxlan_vport(const struct vport *vport)
54 {
55 	return vport_priv(vport);
56 }
57 
58 /* Called with rcu_read_lock and BH disabled. */
59 static void vxlan_rcv(struct vxlan_sock *vs, struct sk_buff *skb, __be32 vx_vni)
60 {
61 	struct ovs_key_ipv4_tunnel tun_key;
62 	struct vport *vport = vs->data;
63 	struct iphdr *iph;
64 	__be64 key;
65 
66 	/* Save outer tunnel values */
67 	iph = ip_hdr(skb);
68 	key = cpu_to_be64(ntohl(vx_vni) >> 8);
69 	ovs_flow_tun_key_init(&tun_key, iph, key, TUNNEL_KEY);
70 
71 	ovs_vport_receive(vport, skb, &tun_key);
72 }
73 
74 static int vxlan_get_options(const struct vport *vport, struct sk_buff *skb)
75 {
76 	struct vxlan_port *vxlan_port = vxlan_vport(vport);
77 	__be16 dst_port = inet_sk(vxlan_port->vs->sock->sk)->inet_sport;
78 
79 	if (nla_put_u16(skb, OVS_TUNNEL_ATTR_DST_PORT, ntohs(dst_port)))
80 		return -EMSGSIZE;
81 	return 0;
82 }
83 
84 static void vxlan_tnl_destroy(struct vport *vport)
85 {
86 	struct vxlan_port *vxlan_port = vxlan_vport(vport);
87 
88 	vxlan_sock_release(vxlan_port->vs);
89 
90 	ovs_vport_deferred_free(vport);
91 }
92 
93 static struct vport *vxlan_tnl_create(const struct vport_parms *parms)
94 {
95 	struct net *net = ovs_dp_get_net(parms->dp);
96 	struct nlattr *options = parms->options;
97 	struct vxlan_port *vxlan_port;
98 	struct vxlan_sock *vs;
99 	struct vport *vport;
100 	struct nlattr *a;
101 	u16 dst_port;
102 	int err;
103 
104 	if (!options) {
105 		err = -EINVAL;
106 		goto error;
107 	}
108 	a = nla_find_nested(options, OVS_TUNNEL_ATTR_DST_PORT);
109 	if (a && nla_len(a) == sizeof(u16)) {
110 		dst_port = nla_get_u16(a);
111 	} else {
112 		/* Require destination port from userspace. */
113 		err = -EINVAL;
114 		goto error;
115 	}
116 
117 	vport = ovs_vport_alloc(sizeof(struct vxlan_port),
118 				&ovs_vxlan_vport_ops, parms);
119 	if (IS_ERR(vport))
120 		return vport;
121 
122 	vxlan_port = vxlan_vport(vport);
123 	strncpy(vxlan_port->name, parms->name, IFNAMSIZ);
124 
125 	vs = vxlan_sock_add(net, htons(dst_port), vxlan_rcv, vport, true, false);
126 	if (IS_ERR(vs)) {
127 		ovs_vport_free(vport);
128 		return (void *)vs;
129 	}
130 	vxlan_port->vs = vs;
131 
132 	return vport;
133 
134 error:
135 	return ERR_PTR(err);
136 }
137 
138 static int vxlan_tnl_send(struct vport *vport, struct sk_buff *skb)
139 {
140 	struct net *net = ovs_dp_get_net(vport->dp);
141 	struct vxlan_port *vxlan_port = vxlan_vport(vport);
142 	__be16 dst_port = inet_sk(vxlan_port->vs->sock->sk)->inet_sport;
143 	struct rtable *rt;
144 	struct flowi4 fl;
145 	__be16 src_port;
146 	int port_min;
147 	int port_max;
148 	__be16 df;
149 	int err;
150 
151 	if (unlikely(!OVS_CB(skb)->tun_key)) {
152 		err = -EINVAL;
153 		goto error;
154 	}
155 
156 	/* Route lookup */
157 	memset(&fl, 0, sizeof(fl));
158 	fl.daddr = OVS_CB(skb)->tun_key->ipv4_dst;
159 	fl.saddr = OVS_CB(skb)->tun_key->ipv4_src;
160 	fl.flowi4_tos = RT_TOS(OVS_CB(skb)->tun_key->ipv4_tos);
161 	fl.flowi4_mark = skb->mark;
162 	fl.flowi4_proto = IPPROTO_UDP;
163 
164 	rt = ip_route_output_key(net, &fl);
165 	if (IS_ERR(rt)) {
166 		err = PTR_ERR(rt);
167 		goto error;
168 	}
169 
170 	df = OVS_CB(skb)->tun_key->tun_flags & TUNNEL_DONT_FRAGMENT ?
171 		htons(IP_DF) : 0;
172 
173 	skb->local_df = 1;
174 
175 	inet_get_local_port_range(net, &port_min, &port_max);
176 	src_port = vxlan_src_port(port_min, port_max, skb);
177 
178 	err = vxlan_xmit_skb(vxlan_port->vs, rt, skb,
179 			     fl.saddr, OVS_CB(skb)->tun_key->ipv4_dst,
180 			     OVS_CB(skb)->tun_key->ipv4_tos,
181 			     OVS_CB(skb)->tun_key->ipv4_ttl, df,
182 			     src_port, dst_port,
183 			     htonl(be64_to_cpu(OVS_CB(skb)->tun_key->tun_id) << 8));
184 	if (err < 0)
185 		ip_rt_put(rt);
186 error:
187 	return err;
188 }
189 
190 static const char *vxlan_get_name(const struct vport *vport)
191 {
192 	struct vxlan_port *vxlan_port = vxlan_vport(vport);
193 	return vxlan_port->name;
194 }
195 
196 const struct vport_ops ovs_vxlan_vport_ops = {
197 	.type		= OVS_VPORT_TYPE_VXLAN,
198 	.create		= vxlan_tnl_create,
199 	.destroy	= vxlan_tnl_destroy,
200 	.get_name	= vxlan_get_name,
201 	.get_options	= vxlan_get_options,
202 	.send		= vxlan_tnl_send,
203 };
204