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;
56 /* Info on neighbour */
59 /* Miscellaneous cached information */
62 struct list_head rt_uncached;
65 static inline bool rt_is_input_route(const struct rtable *rt)
67 return rt->rt_is_input != 0;
70 static inline bool rt_is_output_route(const struct rtable *rt)
72 return rt->rt_is_input == 0;
75 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
78 return rt->rt_gateway;
89 struct rt_cache_stat {
91 unsigned int in_slow_tot;
92 unsigned int in_slow_mc;
93 unsigned int in_no_route;
95 unsigned int in_martian_dst;
96 unsigned int in_martian_src;
98 unsigned int out_slow_tot;
99 unsigned int out_slow_mc;
100 unsigned int gc_total;
101 unsigned int gc_ignored;
102 unsigned int gc_goal_miss;
103 unsigned int gc_dst_overflow;
104 unsigned int in_hlist_search;
105 unsigned int out_hlist_search;
108 extern struct ip_rt_acct __percpu *ip_rt_acct;
111 extern int ip_rt_init(void);
112 extern void rt_cache_flush(struct net *net);
113 extern void rt_flush_dev(struct net_device *dev);
114 extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp);
115 extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
117 extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig);
119 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
121 return ip_route_output_flow(net, flp, NULL);
124 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
125 __be32 saddr, u8 tos, int oif)
127 struct flowi4 fl4 = {
133 return ip_route_output_key(net, &fl4);
136 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
138 __be32 daddr, __be32 saddr,
139 __be16 dport, __be16 sport,
140 __u8 proto, __u8 tos, int oif)
142 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
143 RT_SCOPE_UNIVERSE, proto,
144 sk ? inet_sk_flowi_flags(sk) : 0,
145 daddr, saddr, dport, sport);
147 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
148 return ip_route_output_flow(net, fl4, sk);
151 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
152 __be32 daddr, __be32 saddr,
153 __be32 gre_key, __u8 tos, int oif)
155 memset(fl4, 0, sizeof(*fl4));
156 fl4->flowi4_oif = oif;
159 fl4->flowi4_tos = tos;
160 fl4->flowi4_proto = IPPROTO_GRE;
161 fl4->fl4_gre_key = gre_key;
162 return ip_route_output_key(net, fl4);
165 extern int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
166 u8 tos, struct net_device *devin);
168 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
169 u8 tos, struct net_device *devin)
174 err = ip_route_input_noref(skb, dst, src, tos, devin);
182 extern void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu,
183 int oif, u32 mark, u8 protocol, int flow_flags);
184 extern void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
185 extern void ipv4_redirect(struct sk_buff *skb, struct net *net,
186 int oif, u32 mark, u8 protocol, int flow_flags);
187 extern void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
188 extern void ip_rt_send_redirect(struct sk_buff *skb);
190 extern unsigned int inet_addr_type(struct net *net, __be32 addr);
191 extern unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr);
192 extern void ip_rt_multicast_event(struct in_device *);
193 extern int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
194 extern void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
195 extern int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb);
198 extern void fib_add_ifaddr(struct in_ifaddr *);
199 extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
201 static inline void ip_rt_put(struct rtable *rt)
203 /* dst_release() accepts a NULL parameter.
204 * We rely on dst being first structure in struct rtable
206 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
207 dst_release(&rt->dst);
210 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
212 extern const __u8 ip_tos2prio[16];
214 static inline char rt_tos2priority(u8 tos)
216 return ip_tos2prio[IPTOS_TOS(tos)>>1];
219 /* ip_route_connect() and ip_route_newports() work in tandem whilst
220 * binding a socket for a new outgoing connection.
222 * In order to use IPSEC properly, we must, in the end, have a
223 * route that was looked up using all available keys including source
224 * and destination ports.
226 * However, if a source port needs to be allocated (the user specified
227 * a wildcard source port) we need to obtain addressing information
228 * in order to perform that allocation.
230 * So ip_route_connect() looks up a route using wildcarded source and
231 * destination ports in the key, simply so that we can get a pair of
232 * addresses to use for port allocation.
234 * Later, once the ports are allocated, ip_route_newports() will make
235 * another route lookup if needed to make sure we catch any IPSEC
236 * rules keyed on the port information.
238 * The callers allocate the flow key on their stack, and must pass in
239 * the same flowi4 object to both the ip_route_connect() and the
240 * ip_route_newports() calls.
243 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
244 u32 tos, int oif, u8 protocol,
245 __be16 sport, __be16 dport,
246 struct sock *sk, bool can_sleep)
250 if (inet_sk(sk)->transparent)
251 flow_flags |= FLOWI_FLAG_ANYSRC;
253 flow_flags |= FLOWI_FLAG_CAN_SLEEP;
255 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
256 protocol, flow_flags, dst, src, dport, sport);
259 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
260 __be32 dst, __be32 src, u32 tos,
261 int oif, u8 protocol,
262 __be16 sport, __be16 dport,
263 struct sock *sk, bool can_sleep)
265 struct net *net = sock_net(sk);
268 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
269 sport, dport, sk, can_sleep);
272 rt = __ip_route_output_key(net, fl4);
276 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
278 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
279 return ip_route_output_flow(net, fl4, sk);
282 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
283 __be16 orig_sport, __be16 orig_dport,
284 __be16 sport, __be16 dport,
287 if (sport != orig_sport || dport != orig_dport) {
288 fl4->fl4_dport = dport;
289 fl4->fl4_sport = sport;
291 flowi4_update_output(fl4, sk->sk_bound_dev_if,
292 RT_CONN_FLAGS(sk), fl4->daddr,
294 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
295 return ip_route_output_flow(sock_net(sk), fl4, sk);
300 static inline int inet_iif(const struct sk_buff *skb)
302 int iif = skb_rtable(skb)->rt_iif;
309 extern int sysctl_ip_default_ttl;
311 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
313 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
316 hoplimit = sysctl_ip_default_ttl;
320 #endif /* _ROUTE_H */