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/route.h>
35 #include <linux/cache.h>
36 #include <linux/security.h>
38 #define RTO_ONLINK 0x01
40 #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
53 unsigned int rt_flags;
57 __be32 rt_dst; /* Path destination */
58 __be32 rt_src; /* Path source */
64 /* Info on neighbour */
67 /* Miscellaneous cached information */
68 __be32 rt_spec_dst; /* RFC1122 specific destination */
70 unsigned long _peer; /* long-living peer info */
71 struct fib_info *fi; /* for client ref to shared metrics */
74 static inline struct inet_peer *rt_peer_ptr(struct rtable *rt)
76 return inetpeer_ptr(rt->_peer);
79 static inline bool rt_has_peer(struct rtable *rt)
81 return inetpeer_ptr_is_peer(rt->_peer);
84 static inline void __rt_set_peer(struct rtable *rt, struct inet_peer *peer)
86 __inetpeer_ptr_set_peer(&rt->_peer, peer);
89 static inline bool rt_set_peer(struct rtable *rt, struct inet_peer *peer)
91 return inetpeer_ptr_set_peer(&rt->_peer, peer);
94 static inline void rt_init_peer(struct rtable *rt, struct inet_peer_base *base)
96 inetpeer_init_ptr(&rt->_peer, base);
99 static inline void rt_transfer_peer(struct rtable *rt, struct rtable *ort)
101 rt->_peer = ort->_peer;
102 if (rt_has_peer(ort)) {
103 struct inet_peer *peer = rt_peer_ptr(ort);
104 atomic_inc(&peer->refcnt);
108 static inline bool rt_is_input_route(const struct rtable *rt)
110 return rt->rt_route_iif != 0;
113 static inline bool rt_is_output_route(const struct rtable *rt)
115 return rt->rt_route_iif == 0;
125 struct rt_cache_stat {
127 unsigned int in_slow_tot;
128 unsigned int in_slow_mc;
129 unsigned int in_no_route;
131 unsigned int in_martian_dst;
132 unsigned int in_martian_src;
133 unsigned int out_hit;
134 unsigned int out_slow_tot;
135 unsigned int out_slow_mc;
136 unsigned int gc_total;
137 unsigned int gc_ignored;
138 unsigned int gc_goal_miss;
139 unsigned int gc_dst_overflow;
140 unsigned int in_hlist_search;
141 unsigned int out_hlist_search;
144 extern struct ip_rt_acct __percpu *ip_rt_acct;
147 extern int ip_rt_init(void);
148 extern void ip_rt_redirect(__be32 old_gw, __be32 dst, __be32 new_gw,
149 __be32 src, struct net_device *dev);
150 extern void rt_cache_flush(struct net *net, int how);
151 extern void rt_cache_flush_batch(struct net *net);
152 extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp);
153 extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
155 extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig);
157 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
159 return ip_route_output_flow(net, flp, NULL);
162 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
163 __be32 saddr, u8 tos, int oif)
165 struct flowi4 fl4 = {
171 return ip_route_output_key(net, &fl4);
174 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
176 __be32 daddr, __be32 saddr,
177 __be16 dport, __be16 sport,
178 __u8 proto, __u8 tos, int oif)
180 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
181 RT_SCOPE_UNIVERSE, proto,
182 sk ? inet_sk_flowi_flags(sk) : 0,
183 daddr, saddr, dport, sport);
185 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
186 return ip_route_output_flow(net, fl4, sk);
189 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
190 __be32 daddr, __be32 saddr,
191 __be32 gre_key, __u8 tos, int oif)
193 memset(fl4, 0, sizeof(*fl4));
194 fl4->flowi4_oif = oif;
197 fl4->flowi4_tos = tos;
198 fl4->flowi4_proto = IPPROTO_GRE;
199 fl4->fl4_gre_key = gre_key;
200 return ip_route_output_key(net, fl4);
203 extern int ip_route_input_common(struct sk_buff *skb, __be32 dst, __be32 src,
204 u8 tos, struct net_device *devin, bool noref);
206 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
207 u8 tos, struct net_device *devin)
209 return ip_route_input_common(skb, dst, src, tos, devin, false);
212 static inline int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
213 u8 tos, struct net_device *devin)
215 return ip_route_input_common(skb, dst, src, tos, devin, true);
218 extern void ip_rt_send_redirect(struct sk_buff *skb);
220 extern unsigned int inet_addr_type(struct net *net, __be32 addr);
221 extern unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr);
222 extern void ip_rt_multicast_event(struct in_device *);
223 extern int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
224 extern void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
225 extern int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb);
228 extern void fib_add_ifaddr(struct in_ifaddr *);
229 extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
231 static inline void ip_rt_put(struct rtable * rt)
234 dst_release(&rt->dst);
237 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
239 extern const __u8 ip_tos2prio[16];
241 static inline char rt_tos2priority(u8 tos)
243 return ip_tos2prio[IPTOS_TOS(tos)>>1];
246 /* ip_route_connect() and ip_route_newports() work in tandem whilst
247 * binding a socket for a new outgoing connection.
249 * In order to use IPSEC properly, we must, in the end, have a
250 * route that was looked up using all available keys including source
251 * and destination ports.
253 * However, if a source port needs to be allocated (the user specified
254 * a wildcard source port) we need to obtain addressing information
255 * in order to perform that allocation.
257 * So ip_route_connect() looks up a route using wildcarded source and
258 * destination ports in the key, simply so that we can get a pair of
259 * addresses to use for port allocation.
261 * Later, once the ports are allocated, ip_route_newports() will make
262 * another route lookup if needed to make sure we catch any IPSEC
263 * rules keyed on the port information.
265 * The callers allocate the flow key on their stack, and must pass in
266 * the same flowi4 object to both the ip_route_connect() and the
267 * ip_route_newports() calls.
270 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
271 u32 tos, int oif, u8 protocol,
272 __be16 sport, __be16 dport,
273 struct sock *sk, bool can_sleep)
277 if (inet_sk(sk)->transparent)
278 flow_flags |= FLOWI_FLAG_ANYSRC;
279 if (protocol == IPPROTO_TCP)
280 flow_flags |= FLOWI_FLAG_PRECOW_METRICS;
282 flow_flags |= FLOWI_FLAG_CAN_SLEEP;
284 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
285 protocol, flow_flags, dst, src, dport, sport);
288 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
289 __be32 dst, __be32 src, u32 tos,
290 int oif, u8 protocol,
291 __be16 sport, __be16 dport,
292 struct sock *sk, bool can_sleep)
294 struct net *net = sock_net(sk);
297 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
298 sport, dport, sk, can_sleep);
301 rt = __ip_route_output_key(net, fl4);
305 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
307 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
308 return ip_route_output_flow(net, fl4, sk);
311 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
312 __be16 orig_sport, __be16 orig_dport,
313 __be16 sport, __be16 dport,
316 if (sport != orig_sport || dport != orig_dport) {
317 fl4->fl4_dport = dport;
318 fl4->fl4_sport = sport;
320 flowi4_update_output(fl4, sk->sk_bound_dev_if,
321 RT_CONN_FLAGS(sk), fl4->daddr,
323 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
324 return ip_route_output_flow(sock_net(sk), fl4, sk);
329 extern void rt_bind_peer(struct rtable *rt, __be32 daddr, int create);
331 static inline struct inet_peer *__rt_get_peer(struct rtable *rt, __be32 daddr, int create)
334 return rt_peer_ptr(rt);
336 rt_bind_peer(rt, daddr, create);
337 return (rt_has_peer(rt) ? rt_peer_ptr(rt) : NULL);
340 static inline struct inet_peer *rt_get_peer(struct rtable *rt, __be32 daddr)
342 return __rt_get_peer(rt, daddr, 0);
345 static inline struct inet_peer *rt_get_peer_create(struct rtable *rt, __be32 daddr)
347 return __rt_get_peer(rt, daddr, 1);
350 static inline int inet_iif(const struct sk_buff *skb)
352 return skb_rtable(skb)->rt_iif;
355 extern int sysctl_ip_default_ttl;
357 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
359 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
362 hoplimit = sysctl_ip_default_ttl;
366 #endif /* _ROUTE_H */