2 * Copyright (c) 2007-2012 Nicira, Inc.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of version 2 of the GNU General Public
6 * License as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful, but
9 * WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
13 * You should have received a copy of the GNU General Public License
14 * along with this program; if not, write to the Free Software
15 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
22 #include <linux/version.h>
23 #include <net/net_namespace.h>
24 #include <net/netns/generic.h>
27 #include "openvswitch/tunnel.h"
31 * The absolute minimum fragment size. Note that there are many other
32 * definitions of the minimum MTU.
37 * One of these goes in struct tnl_ops and in tnl_find_port().
38 * These values are in the same namespace as other TNL_T_* values, so
39 * only the least significant 10 bits are available to define protocol
42 #define TNL_T_PROTO_GRE 0
43 #define TNL_T_PROTO_GRE64 1
44 #define TNL_T_PROTO_CAPWAP 2
46 /* These flags are only needed when calling tnl_find_port(). */
47 #define TNL_T_KEY_EXACT (1 << 10)
48 #define TNL_T_KEY_MATCH (1 << 11)
50 /* Private flags not exposed to userspace in this form. */
51 #define TNL_F_IN_KEY_MATCH (1 << 16) /* Store the key in tun_id to
52 * match in flow table. */
53 #define TNL_F_OUT_KEY_ACTION (1 << 17) /* Get the key from a SET_TUNNEL
56 /* All public tunnel flags. */
57 #define TNL_F_PUBLIC (TNL_F_CSUM | TNL_F_TOS_INHERIT | TNL_F_TTL_INHERIT | \
58 TNL_F_DF_INHERIT | TNL_F_DF_DEFAULT | TNL_F_PMTUD | \
59 TNL_F_HDR_CACHE | TNL_F_IPSEC)
62 * struct port_lookup_key - Tunnel port key, used as hash table key.
63 * @in_key: Key to match on input, 0 for wildcard.
64 * @net: Network namespace of the port.
65 * @saddr: IPv4 source address to match, 0 to accept any source address.
66 * @daddr: IPv4 destination of tunnel.
67 * @tunnel_type: Set of TNL_T_* flags that define lookup.
69 struct port_lookup_key {
79 #define PORT_KEY_LEN (offsetof(struct port_lookup_key, tunnel_type) + \
80 FIELD_SIZEOF(struct port_lookup_key, tunnel_type))
82 static inline struct net *port_key_get_net(const struct port_lookup_key *key)
84 return read_pnet(&key->net);
87 static inline void port_key_set_net(struct port_lookup_key *key, struct net *net)
89 write_pnet(&key->net, net);
93 * struct tnl_mutable_config - modifiable configuration for a tunnel.
94 * @key: Used as key for tunnel port. Configured via OVS_TUNNEL_ATTR_*
96 * @rcu: RCU callback head for deferred destruction.
97 * @seq: Sequence number for distinguishing configuration versions.
98 * @tunnel_hlen: Tunnel header length.
99 * @eth_addr: Source address for packets generated by tunnel itself
100 * (e.g. ICMP fragmentation needed messages).
101 * @out_key: Key to use on output, 0 if this tunnel has no fixed output key.
102 * @flags: TNL_F_* flags.
103 * @tos: IPv4 TOS value to use for tunnel, 0 if no fixed TOS.
104 * @ttl: IPv4 TTL value to use for tunnel, 0 if no fixed TTL.
106 struct tnl_mutable_config {
107 struct port_lookup_key key;
112 unsigned char eth_addr[ETH_ALEN];
114 /* Configured via OVS_TUNNEL_ATTR_* attributes. */
120 /* Multicast configuration. */
125 u32 tunnel_type; /* Put the TNL_T_PROTO_* type in here. */
126 u8 ipproto; /* The IP protocol for the tunnel. */
129 * Returns the length of the tunnel header that will be added in
130 * build_header() (i.e. excludes the IP header). Returns a negative
131 * error code if the configuration is invalid.
133 int (*hdr_len)(const struct tnl_mutable_config *,
134 const struct ovs_key_ipv4_tunnel *);
137 * Builds the static portion of the tunnel header, which is stored in
138 * the header cache. In general the performance of this function is
139 * not too important as we try to only call it when building the cache
140 * so it is preferable to shift as much work as possible here. However,
141 * in some circumstances caching is disabled and this function will be
142 * called for every packet, so try not to make it too slow.
144 void (*build_header)(const struct vport *,
145 const struct tnl_mutable_config *,
146 const struct ovs_key_ipv4_tunnel *, void *header);
149 * Updates the cached header of a packet to match the actual packet
150 * data. Typical things that might need to be updated are length,
151 * checksum, etc. The IP header will have already been updated and this
152 * is the final step before transmission. Returns a linked list of
153 * completed SKBs (multiple packets may be generated in the event
156 struct sk_buff *(*update_header)(const struct vport *,
157 const struct tnl_mutable_config *,
158 struct dst_entry *, struct sk_buff *,
162 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
164 * On these kernels we have a fast mechanism to tell if the ARP cache for a
165 * particular destination has changed.
169 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,27)
171 * On these kernels we have a fast mechanism to tell if the routing table
174 #define HAVE_RT_GENID
176 #if !defined(HAVE_HH_SEQ) || !defined(HAVE_RT_GENID)
177 /* If we can't detect all system changes directly we need to use a timeout. */
178 #define NEED_CACHE_TIMEOUT
183 int len; /* Length of data to be memcpy'd from cache. */
184 int hh_len; /* Hardware hdr length, cached from hh_cache. */
186 /* Sequence number of mutable->seq from which this cache was
188 unsigned mutable_seq;
192 * The sequence number from the seqlock protecting the hardware header
193 * cache (in the ARP cache). Since every write increments the counter
194 * this gives us an easy way to tell if it has changed.
199 #ifdef NEED_CACHE_TIMEOUT
201 * If we don't have direct mechanisms to detect all important changes in
202 * the system fall back to an expiration time. This expiration time
203 * can be relatively short since at high rates there will be millions of
204 * packets per second, so we'll still get plenty of benefit from the
205 * cache. Note that if something changes we may blackhole packets
206 * until the expiration time (depending on what changed and the kernel
207 * version we may be able to detect the change sooner). Expiration is
208 * expressed as a time in jiffies.
210 unsigned long expiration;
214 * The routing table entry that is the result of looking up the tunnel
215 * endpoints. It also contains a sequence number (called a generation
216 * ID) that can be compared to a global sequence to tell if the routing
217 * table has changed (and therefore there is a potential that this
218 * cached route has been invalidated).
223 * If the output device for tunnel traffic is an OVS internal device,
224 * the flow of that datapath. Since all tunnel traffic will have the
225 * same headers this allows us to cache the flow lookup. NULL if the
226 * output device is not OVS or if there is no flow installed.
228 struct sw_flow *flow;
230 /* The cached header follows after padding for alignment. */
235 struct hlist_node hash_node;
238 const struct tnl_ops *tnl_ops;
240 struct tnl_mutable_config __rcu *mutable;
243 * ID of last fragment sent (for tunnel protocols with direct support
244 * fragmentation). If the protocol relies on IP fragmentation then
245 * this is not needed.
249 spinlock_t cache_lock;
250 struct tnl_cache __rcu *cache; /* Protected by RCU/cache_lock. */
252 #ifdef NEED_CACHE_TIMEOUT
254 * If we must rely on expiration time to invalidate the cache, this is
255 * the interval. It is randomized within a range (defined by
256 * MAX_CACHE_EXP in tunnel.c) to avoid synchronized expirations caused
257 * by creation of a large number of tunnels at a one time.
259 unsigned long cache_exp_interval;
263 struct vport *ovs_tnl_create(const struct vport_parms *, const struct vport_ops *,
264 const struct tnl_ops *);
265 void ovs_tnl_destroy(struct vport *);
267 int ovs_tnl_set_options(struct vport *, struct nlattr *);
268 int ovs_tnl_get_options(const struct vport *, struct sk_buff *);
270 int ovs_tnl_set_addr(struct vport *vport, const unsigned char *addr);
271 const char *ovs_tnl_get_name(const struct vport *vport);
272 const unsigned char *ovs_tnl_get_addr(const struct vport *vport);
273 int ovs_tnl_send(struct vport *vport, struct sk_buff *skb);
274 void ovs_tnl_rcv(struct vport *vport, struct sk_buff *skb);
276 struct vport *ovs_tnl_find_port(struct net *net, __be32 saddr, __be32 daddr,
277 __be64 key, int tunnel_type,
278 const struct tnl_mutable_config **mutable);
279 bool ovs_tnl_frag_needed(struct vport *vport,
280 const struct tnl_mutable_config *mutable,
281 struct sk_buff *skb, unsigned int mtu);
282 void ovs_tnl_free_linked_skbs(struct sk_buff *skb);
284 int ovs_tnl_init(void);
285 void ovs_tnl_exit(void);
286 static inline struct tnl_vport *tnl_vport_priv(const struct vport *vport)
288 return vport_priv(vport);
291 static inline void tnl_tun_key_init(struct ovs_key_ipv4_tunnel *tun_key,
292 const struct iphdr *iph, __be64 tun_id, u32 tun_flags)
294 tun_key->tun_id = tun_id;
295 tun_key->ipv4_src = iph->saddr;
296 tun_key->ipv4_dst = iph->daddr;
297 tun_key->ipv4_tos = iph->tos;
298 tun_key->ipv4_ttl = iph->ttl;
299 tun_key->tun_flags = tun_flags;
302 #endif /* tunnel.h */