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 <linux/in_route.h>
32 #include <linux/rtnetlink.h>
33 #include <linux/rcupdate.h>
34 #include <linux/route.h>
36 #include <linux/cache.h>
37 #include <linux/security.h>
39 #define RTO_ONLINK 0x01
41 #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
49 unsigned int rt_flags;
55 /* Info on neighbour */
58 /* Miscellaneous cached information */
61 struct list_head rt_uncached;
64 static inline bool rt_is_input_route(const struct rtable *rt)
66 return rt->rt_is_input != 0;
69 static inline bool rt_is_output_route(const struct rtable *rt)
71 return rt->rt_is_input == 0;
74 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
77 return rt->rt_gateway;
88 struct rt_cache_stat {
90 unsigned int in_slow_tot;
91 unsigned int in_slow_mc;
92 unsigned int in_no_route;
94 unsigned int in_martian_dst;
95 unsigned int in_martian_src;
97 unsigned int out_slow_tot;
98 unsigned int out_slow_mc;
99 unsigned int gc_total;
100 unsigned int gc_ignored;
101 unsigned int gc_goal_miss;
102 unsigned int gc_dst_overflow;
103 unsigned int in_hlist_search;
104 unsigned int out_hlist_search;
107 extern struct ip_rt_acct __percpu *ip_rt_acct;
110 extern int ip_rt_init(void);
111 extern void rt_cache_flush(struct net *net);
112 extern void rt_flush_dev(struct net_device *dev);
113 extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp);
114 extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
116 extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig);
118 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
120 return ip_route_output_flow(net, flp, NULL);
123 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
124 __be32 saddr, u8 tos, int oif)
126 struct flowi4 fl4 = {
132 return ip_route_output_key(net, &fl4);
135 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
137 __be32 daddr, __be32 saddr,
138 __be16 dport, __be16 sport,
139 __u8 proto, __u8 tos, int oif)
141 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
142 RT_SCOPE_UNIVERSE, proto,
143 sk ? inet_sk_flowi_flags(sk) : 0,
144 daddr, saddr, dport, sport);
146 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
147 return ip_route_output_flow(net, fl4, sk);
150 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
151 __be32 daddr, __be32 saddr,
152 __be32 gre_key, __u8 tos, int oif)
154 memset(fl4, 0, sizeof(*fl4));
155 fl4->flowi4_oif = oif;
158 fl4->flowi4_tos = tos;
159 fl4->flowi4_proto = IPPROTO_GRE;
160 fl4->fl4_gre_key = gre_key;
161 return ip_route_output_key(net, fl4);
164 extern int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
165 u8 tos, struct net_device *devin);
167 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
168 u8 tos, struct net_device *devin)
173 err = ip_route_input_noref(skb, dst, src, tos, devin);
181 extern void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu,
182 int oif, u32 mark, u8 protocol, int flow_flags);
183 extern void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
184 extern void ipv4_redirect(struct sk_buff *skb, struct net *net,
185 int oif, u32 mark, u8 protocol, int flow_flags);
186 extern void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
187 extern void ip_rt_send_redirect(struct sk_buff *skb);
189 extern unsigned int inet_addr_type(struct net *net, __be32 addr);
190 extern unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr);
191 extern void ip_rt_multicast_event(struct in_device *);
192 extern int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
193 extern void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
194 extern int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb);
197 extern void fib_add_ifaddr(struct in_ifaddr *);
198 extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
200 static inline void ip_rt_put(struct rtable * rt)
203 dst_release(&rt->dst);
206 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
208 extern const __u8 ip_tos2prio[16];
210 static inline char rt_tos2priority(u8 tos)
212 return ip_tos2prio[IPTOS_TOS(tos)>>1];
215 /* ip_route_connect() and ip_route_newports() work in tandem whilst
216 * binding a socket for a new outgoing connection.
218 * In order to use IPSEC properly, we must, in the end, have a
219 * route that was looked up using all available keys including source
220 * and destination ports.
222 * However, if a source port needs to be allocated (the user specified
223 * a wildcard source port) we need to obtain addressing information
224 * in order to perform that allocation.
226 * So ip_route_connect() looks up a route using wildcarded source and
227 * destination ports in the key, simply so that we can get a pair of
228 * addresses to use for port allocation.
230 * Later, once the ports are allocated, ip_route_newports() will make
231 * another route lookup if needed to make sure we catch any IPSEC
232 * rules keyed on the port information.
234 * The callers allocate the flow key on their stack, and must pass in
235 * the same flowi4 object to both the ip_route_connect() and the
236 * ip_route_newports() calls.
239 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
240 u32 tos, int oif, u8 protocol,
241 __be16 sport, __be16 dport,
242 struct sock *sk, bool can_sleep)
246 if (inet_sk(sk)->transparent)
247 flow_flags |= FLOWI_FLAG_ANYSRC;
249 flow_flags |= FLOWI_FLAG_CAN_SLEEP;
251 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
252 protocol, flow_flags, dst, src, dport, sport);
255 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
256 __be32 dst, __be32 src, u32 tos,
257 int oif, u8 protocol,
258 __be16 sport, __be16 dport,
259 struct sock *sk, bool can_sleep)
261 struct net *net = sock_net(sk);
264 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
265 sport, dport, sk, can_sleep);
268 rt = __ip_route_output_key(net, fl4);
272 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
274 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
275 return ip_route_output_flow(net, fl4, sk);
278 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
279 __be16 orig_sport, __be16 orig_dport,
280 __be16 sport, __be16 dport,
283 if (sport != orig_sport || dport != orig_dport) {
284 fl4->fl4_dport = dport;
285 fl4->fl4_sport = sport;
287 flowi4_update_output(fl4, sk->sk_bound_dev_if,
288 RT_CONN_FLAGS(sk), fl4->daddr,
290 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
291 return ip_route_output_flow(sock_net(sk), fl4, sk);
296 static inline int inet_iif(const struct sk_buff *skb)
298 int iif = skb_rtable(skb)->rt_iif;
305 extern int sysctl_ip_default_ttl;
307 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
309 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
312 hoplimit = sysctl_ip_default_ttl;
316 #endif /* _ROUTE_H */