2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
6 * Definitions for the IP router.
8 * Version: @(#)route.h 1.0.4 05/27/93
11 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
13 * Alan Cox : Reformatted. Added ip_rt_local()
14 * Alan Cox : Support for TCP parameters.
15 * Alexey Kuznetsov: Major changes for new routing code.
16 * Mike McLagan : Routing by source
17 * Robert Olsson : Added rt_cache statistics
19 * This program is free software; you can redistribute it and/or
20 * modify it under the terms of the GNU General Public License
21 * as published by the Free Software Foundation; either version
22 * 2 of the License, or (at your option) any later version.
28 #include <net/inetpeer.h>
30 #include <net/inet_sock.h>
31 #include <net/ip_fib.h>
33 #include <net/ndisc.h>
34 #include <linux/in_route.h>
35 #include <linux/rtnetlink.h>
36 #include <linux/rcupdate.h>
37 #include <linux/route.h>
39 #include <linux/cache.h>
40 #include <linux/security.h>
42 /* IPv4 datagram length is stored into 16bit field (tot_len) */
43 #define IP_MAX_MTU 0xFFFFU
45 #define RTO_ONLINK 0x01
47 #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
48 #define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
57 unsigned int rt_flags;
64 /* Info on neighbour */
67 struct in6_addr rt_gw6;
70 /* Miscellaneous cached information */
74 struct list_head rt_uncached;
75 struct uncached_list *rt_uncached_list;
78 static inline bool rt_is_input_route(const struct rtable *rt)
80 return rt->rt_is_input != 0;
83 static inline bool rt_is_output_route(const struct rtable *rt)
85 return rt->rt_is_input == 0;
88 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
90 if (rt->rt_gw_family == AF_INET)
102 struct rt_cache_stat {
103 unsigned int in_slow_tot;
104 unsigned int in_slow_mc;
105 unsigned int in_no_route;
107 unsigned int in_martian_dst;
108 unsigned int in_martian_src;
109 unsigned int out_slow_tot;
110 unsigned int out_slow_mc;
113 extern struct ip_rt_acct __percpu *ip_rt_acct;
117 int ip_rt_init(void);
118 void rt_cache_flush(struct net *net);
119 void rt_flush_dev(struct net_device *dev);
120 struct rtable *ip_route_output_key_hash(struct net *net, struct flowi4 *flp,
121 const struct sk_buff *skb);
122 struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *flp,
123 struct fib_result *res,
124 const struct sk_buff *skb);
126 static inline struct rtable *__ip_route_output_key(struct net *net,
129 return ip_route_output_key_hash(net, flp, NULL);
132 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
133 const struct sock *sk);
134 struct dst_entry *ipv4_blackhole_route(struct net *net,
135 struct dst_entry *dst_orig);
137 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
139 return ip_route_output_flow(net, flp, NULL);
142 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
143 __be32 saddr, u8 tos, int oif)
145 struct flowi4 fl4 = {
151 return ip_route_output_key(net, &fl4);
154 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
156 __be32 daddr, __be32 saddr,
157 __be16 dport, __be16 sport,
158 __u8 proto, __u8 tos, int oif)
160 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
161 RT_SCOPE_UNIVERSE, proto,
162 sk ? inet_sk_flowi_flags(sk) : 0,
163 daddr, saddr, dport, sport, sock_net_uid(net, sk));
165 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
166 return ip_route_output_flow(net, fl4, sk);
169 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
170 __be32 daddr, __be32 saddr,
171 __be32 gre_key, __u8 tos, int oif)
173 memset(fl4, 0, sizeof(*fl4));
174 fl4->flowi4_oif = oif;
177 fl4->flowi4_tos = tos;
178 fl4->flowi4_proto = IPPROTO_GRE;
179 fl4->fl4_gre_key = gre_key;
180 return ip_route_output_key(net, fl4);
182 int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr,
183 u8 tos, struct net_device *dev,
184 struct in_device *in_dev, u32 *itag);
185 int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
186 u8 tos, struct net_device *devin);
187 int ip_route_input_rcu(struct sk_buff *skb, __be32 dst, __be32 src,
188 u8 tos, struct net_device *devin,
189 struct fib_result *res);
191 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
192 u8 tos, struct net_device *devin)
197 err = ip_route_input_noref(skb, dst, src, tos, devin);
208 void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
210 void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
211 void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u8 protocol);
212 void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
213 void ip_rt_send_redirect(struct sk_buff *skb);
215 unsigned int inet_addr_type(struct net *net, __be32 addr);
216 unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
217 unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
219 unsigned int inet_addr_type_dev_table(struct net *net,
220 const struct net_device *dev,
222 void ip_rt_multicast_event(struct in_device *);
223 int ip_rt_ioctl(struct net *, unsigned int cmd, struct rtentry *rt);
224 void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
225 struct rtable *rt_dst_alloc(struct net_device *dev,
226 unsigned int flags, u16 type,
227 bool nopolicy, bool noxfrm, bool will_cache);
230 void fib_add_ifaddr(struct in_ifaddr *);
231 void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
232 void fib_modify_prefix_metric(struct in_ifaddr *ifa, u32 new_metric);
234 void rt_add_uncached_list(struct rtable *rt);
235 void rt_del_uncached_list(struct rtable *rt);
237 static inline void ip_rt_put(struct rtable *rt)
239 /* dst_release() accepts a NULL parameter.
240 * We rely on dst being first structure in struct rtable
242 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
243 dst_release(&rt->dst);
246 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
248 extern const __u8 ip_tos2prio[16];
250 static inline char rt_tos2priority(u8 tos)
252 return ip_tos2prio[IPTOS_TOS(tos)>>1];
255 /* ip_route_connect() and ip_route_newports() work in tandem whilst
256 * binding a socket for a new outgoing connection.
258 * In order to use IPSEC properly, we must, in the end, have a
259 * route that was looked up using all available keys including source
260 * and destination ports.
262 * However, if a source port needs to be allocated (the user specified
263 * a wildcard source port) we need to obtain addressing information
264 * in order to perform that allocation.
266 * So ip_route_connect() looks up a route using wildcarded source and
267 * destination ports in the key, simply so that we can get a pair of
268 * addresses to use for port allocation.
270 * Later, once the ports are allocated, ip_route_newports() will make
271 * another route lookup if needed to make sure we catch any IPSEC
272 * rules keyed on the port information.
274 * The callers allocate the flow key on their stack, and must pass in
275 * the same flowi4 object to both the ip_route_connect() and the
276 * ip_route_newports() calls.
279 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
280 u32 tos, int oif, u8 protocol,
281 __be16 sport, __be16 dport,
286 if (inet_sk(sk)->transparent)
287 flow_flags |= FLOWI_FLAG_ANYSRC;
289 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
290 protocol, flow_flags, dst, src, dport, sport,
294 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
295 __be32 dst, __be32 src, u32 tos,
296 int oif, u8 protocol,
297 __be16 sport, __be16 dport,
300 struct net *net = sock_net(sk);
303 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
307 rt = __ip_route_output_key(net, fl4);
311 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
313 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
314 return ip_route_output_flow(net, fl4, sk);
317 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
318 __be16 orig_sport, __be16 orig_dport,
319 __be16 sport, __be16 dport,
322 if (sport != orig_sport || dport != orig_dport) {
323 fl4->fl4_dport = dport;
324 fl4->fl4_sport = sport;
326 flowi4_update_output(fl4, sk->sk_bound_dev_if,
327 RT_CONN_FLAGS(sk), fl4->daddr,
329 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
330 return ip_route_output_flow(sock_net(sk), fl4, sk);
335 static inline int inet_iif(const struct sk_buff *skb)
337 struct rtable *rt = skb_rtable(skb);
339 if (rt && rt->rt_iif)
345 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
347 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
348 struct net *net = dev_net(dst->dev);
351 hoplimit = net->ipv4.sysctl_ip_default_ttl;
355 static inline struct neighbour *ip_neigh_gw4(struct net_device *dev,
358 struct neighbour *neigh;
360 neigh = __ipv4_neigh_lookup_noref(dev, daddr);
361 if (unlikely(!neigh))
362 neigh = __neigh_create(&arp_tbl, &daddr, dev, false);
367 static inline struct neighbour *ip_neigh_for_gw(struct rtable *rt,
371 struct net_device *dev = rt->dst.dev;
372 struct neighbour *neigh;
374 if (likely(rt->rt_gw_family == AF_INET)) {
375 neigh = ip_neigh_gw4(dev, rt->rt_gw4);
376 } else if (rt->rt_gw_family == AF_INET6) {
377 neigh = ip_neigh_gw6(dev, &rt->rt_gw6);
380 neigh = ip_neigh_gw4(dev, ip_hdr(skb)->daddr);
385 #endif /* _ROUTE_H */