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>
32 #include <linux/in_route.h>
33 #include <linux/rtnetlink.h>
34 #include <linux/rcupdate.h>
35 #include <linux/route.h>
37 #include <linux/cache.h>
38 #include <linux/security.h>
40 /* IPv4 datagram length is stored into 16bit field (tot_len) */
41 #define IP_MAX_MTU 0xFFFFU
43 #define RTO_ONLINK 0x01
45 #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
46 #define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
55 unsigned int rt_flags;
62 /* Info on neighbour */
65 /* Miscellaneous cached information */
70 struct list_head rt_uncached;
71 struct uncached_list *rt_uncached_list;
74 static inline bool rt_is_input_route(const struct rtable *rt)
76 return rt->rt_is_input != 0;
79 static inline bool rt_is_output_route(const struct rtable *rt)
81 return rt->rt_is_input == 0;
84 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
87 return rt->rt_gateway;
98 struct rt_cache_stat {
99 unsigned int in_slow_tot;
100 unsigned int in_slow_mc;
101 unsigned int in_no_route;
103 unsigned int in_martian_dst;
104 unsigned int in_martian_src;
105 unsigned int out_slow_tot;
106 unsigned int out_slow_mc;
109 extern struct ip_rt_acct __percpu *ip_rt_acct;
113 int ip_rt_init(void);
114 void rt_cache_flush(struct net *net);
115 void rt_flush_dev(struct net_device *dev);
116 struct rtable *ip_route_output_key_hash(struct net *net, struct flowi4 *flp,
117 const struct sk_buff *skb);
118 struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *flp,
119 struct fib_result *res,
120 const struct sk_buff *skb);
122 static inline struct rtable *__ip_route_output_key(struct net *net,
125 return ip_route_output_key_hash(net, flp, NULL);
128 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
129 const struct sock *sk);
130 struct dst_entry *ipv4_blackhole_route(struct net *net,
131 struct dst_entry *dst_orig);
133 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
135 return ip_route_output_flow(net, flp, NULL);
138 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
139 __be32 saddr, u8 tos, int oif)
141 struct flowi4 fl4 = {
147 return ip_route_output_key(net, &fl4);
150 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
152 __be32 daddr, __be32 saddr,
153 __be16 dport, __be16 sport,
154 __u8 proto, __u8 tos, int oif)
156 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
157 RT_SCOPE_UNIVERSE, proto,
158 sk ? inet_sk_flowi_flags(sk) : 0,
159 daddr, saddr, dport, sport, sock_net_uid(net, sk));
161 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
162 return ip_route_output_flow(net, fl4, sk);
165 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
166 __be32 daddr, __be32 saddr,
167 __be32 gre_key, __u8 tos, int oif)
169 memset(fl4, 0, sizeof(*fl4));
170 fl4->flowi4_oif = oif;
173 fl4->flowi4_tos = tos;
174 fl4->flowi4_proto = IPPROTO_GRE;
175 fl4->fl4_gre_key = gre_key;
176 return ip_route_output_key(net, fl4);
179 int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
180 u8 tos, struct net_device *devin);
181 int ip_route_input_rcu(struct sk_buff *skb, __be32 dst, __be32 src,
182 u8 tos, struct net_device *devin,
183 struct fib_result *res);
185 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
186 u8 tos, struct net_device *devin)
191 err = ip_route_input_noref(skb, dst, src, tos, devin);
199 void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
200 u32 mark, u8 protocol, int flow_flags);
201 void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
202 void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
203 u8 protocol, int flow_flags);
204 void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
205 void ip_rt_send_redirect(struct sk_buff *skb);
207 unsigned int inet_addr_type(struct net *net, __be32 addr);
208 unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
209 unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
211 unsigned int inet_addr_type_dev_table(struct net *net,
212 const struct net_device *dev,
214 void ip_rt_multicast_event(struct in_device *);
215 int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
216 void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
217 struct rtable *rt_dst_alloc(struct net_device *dev,
218 unsigned int flags, u16 type,
219 bool nopolicy, bool noxfrm, bool will_cache);
222 void fib_add_ifaddr(struct in_ifaddr *);
223 void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
225 static inline void ip_rt_put(struct rtable *rt)
227 /* dst_release() accepts a NULL parameter.
228 * We rely on dst being first structure in struct rtable
230 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
231 dst_release(&rt->dst);
234 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
236 extern const __u8 ip_tos2prio[16];
238 static inline char rt_tos2priority(u8 tos)
240 return ip_tos2prio[IPTOS_TOS(tos)>>1];
243 /* ip_route_connect() and ip_route_newports() work in tandem whilst
244 * binding a socket for a new outgoing connection.
246 * In order to use IPSEC properly, we must, in the end, have a
247 * route that was looked up using all available keys including source
248 * and destination ports.
250 * However, if a source port needs to be allocated (the user specified
251 * a wildcard source port) we need to obtain addressing information
252 * in order to perform that allocation.
254 * So ip_route_connect() looks up a route using wildcarded source and
255 * destination ports in the key, simply so that we can get a pair of
256 * addresses to use for port allocation.
258 * Later, once the ports are allocated, ip_route_newports() will make
259 * another route lookup if needed to make sure we catch any IPSEC
260 * rules keyed on the port information.
262 * The callers allocate the flow key on their stack, and must pass in
263 * the same flowi4 object to both the ip_route_connect() and the
264 * ip_route_newports() calls.
267 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
268 u32 tos, int oif, u8 protocol,
269 __be16 sport, __be16 dport,
274 if (inet_sk(sk)->transparent)
275 flow_flags |= FLOWI_FLAG_ANYSRC;
277 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
278 protocol, flow_flags, dst, src, dport, sport,
282 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
283 __be32 dst, __be32 src, u32 tos,
284 int oif, u8 protocol,
285 __be16 sport, __be16 dport,
288 struct net *net = sock_net(sk);
291 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
295 rt = __ip_route_output_key(net, fl4);
299 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
301 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
302 return ip_route_output_flow(net, fl4, sk);
305 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
306 __be16 orig_sport, __be16 orig_dport,
307 __be16 sport, __be16 dport,
310 if (sport != orig_sport || dport != orig_dport) {
311 fl4->fl4_dport = dport;
312 fl4->fl4_sport = sport;
314 flowi4_update_output(fl4, sk->sk_bound_dev_if,
315 RT_CONN_FLAGS(sk), fl4->daddr,
317 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
318 return ip_route_output_flow(sock_net(sk), fl4, sk);
323 static inline int inet_iif(const struct sk_buff *skb)
325 struct rtable *rt = skb_rtable(skb);
327 if (rt && rt->rt_iif)
333 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
335 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
336 struct net *net = dev_net(dst->dev);
339 hoplimit = net->ipv4.sysctl_ip_default_ttl;
343 #endif /* _ROUTE_H */