1 /* SPDX-License-Identifier: GPL-2.0-or-later */
3 * INET An implementation of the TCP/IP protocol suite for the LINUX
4 * operating system. INET is implemented using the BSD Socket
5 * interface as the means of communication with the user level.
7 * Definitions for the IP router.
9 * Version: @(#)route.h 1.0.4 05/27/93
12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
14 * Alan Cox : Reformatted. Added ip_rt_local()
15 * Alan Cox : Support for TCP parameters.
16 * Alexey Kuznetsov: Major changes for new routing code.
17 * Mike McLagan : Routing by source
18 * Robert Olsson : Added rt_cache statistics
24 #include <net/inetpeer.h>
26 #include <net/inet_sock.h>
27 #include <net/ip_fib.h>
29 #include <net/ndisc.h>
30 #include <linux/in_route.h>
31 #include <linux/rtnetlink.h>
32 #include <linux/rcupdate.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))
41 #define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
43 static inline __u8 ip_sock_rt_scope(const struct sock *sk)
45 if (sock_flag(sk, SOCK_LOCALROUTE))
48 return RT_SCOPE_UNIVERSE;
51 static inline __u8 ip_sock_rt_tos(const struct sock *sk)
53 return RT_TOS(inet_sk(sk)->tos);
56 struct ip_tunnel_info;
64 unsigned int rt_flags;
72 /* Info on neighbour */
75 struct in6_addr rt_gw6;
78 /* Miscellaneous cached information */
82 struct list_head rt_uncached;
83 struct uncached_list *rt_uncached_list;
86 static inline bool rt_is_input_route(const struct rtable *rt)
88 return rt->rt_is_input != 0;
91 static inline bool rt_is_output_route(const struct rtable *rt)
93 return rt->rt_is_input == 0;
96 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
98 if (rt->rt_gw_family == AF_INET)
110 struct rt_cache_stat {
111 unsigned int in_slow_tot;
112 unsigned int in_slow_mc;
113 unsigned int in_no_route;
115 unsigned int in_martian_dst;
116 unsigned int in_martian_src;
117 unsigned int out_slow_tot;
118 unsigned int out_slow_mc;
121 extern struct ip_rt_acct __percpu *ip_rt_acct;
125 int ip_rt_init(void);
126 void rt_cache_flush(struct net *net);
127 void rt_flush_dev(struct net_device *dev);
128 struct rtable *ip_route_output_key_hash(struct net *net, struct flowi4 *flp,
129 const struct sk_buff *skb);
130 struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *flp,
131 struct fib_result *res,
132 const struct sk_buff *skb);
134 static inline struct rtable *__ip_route_output_key(struct net *net,
137 return ip_route_output_key_hash(net, flp, NULL);
140 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
141 const struct sock *sk);
142 struct rtable *ip_route_output_tunnel(struct sk_buff *skb,
143 struct net_device *dev,
144 struct net *net, __be32 *saddr,
145 const struct ip_tunnel_info *info,
146 u8 protocol, bool use_cache);
148 struct dst_entry *ipv4_blackhole_route(struct net *net,
149 struct dst_entry *dst_orig);
151 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
153 return ip_route_output_flow(net, flp, NULL);
156 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
157 __be32 saddr, u8 tos, int oif)
159 struct flowi4 fl4 = {
165 return ip_route_output_key(net, &fl4);
168 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
170 __be32 daddr, __be32 saddr,
171 __be16 dport, __be16 sport,
172 __u8 proto, __u8 tos, int oif)
174 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
175 RT_SCOPE_UNIVERSE, proto,
176 sk ? inet_sk_flowi_flags(sk) : 0,
177 daddr, saddr, dport, sport, sock_net_uid(net, sk));
179 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
180 return ip_route_output_flow(net, fl4, sk);
183 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
184 __be32 daddr, __be32 saddr,
185 __be32 gre_key, __u8 tos, int oif)
187 memset(fl4, 0, sizeof(*fl4));
188 fl4->flowi4_oif = oif;
191 fl4->flowi4_tos = tos;
192 fl4->flowi4_proto = IPPROTO_GRE;
193 fl4->fl4_gre_key = gre_key;
194 return ip_route_output_key(net, fl4);
196 int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr,
197 u8 tos, struct net_device *dev,
198 struct in_device *in_dev, u32 *itag);
199 int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
200 u8 tos, struct net_device *devin);
201 int ip_route_use_hint(struct sk_buff *skb, __be32 dst, __be32 src,
202 u8 tos, struct net_device *devin,
203 const struct sk_buff *hint);
205 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
206 u8 tos, struct net_device *devin)
211 err = ip_route_input_noref(skb, dst, src, tos, devin);
222 void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
224 void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
225 void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u8 protocol);
226 void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
227 void ip_rt_send_redirect(struct sk_buff *skb);
229 unsigned int inet_addr_type(struct net *net, __be32 addr);
230 unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
231 unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
233 unsigned int inet_addr_type_dev_table(struct net *net,
234 const struct net_device *dev,
236 void ip_rt_multicast_event(struct in_device *);
237 int ip_rt_ioctl(struct net *, unsigned int cmd, struct rtentry *rt);
238 void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
239 struct rtable *rt_dst_alloc(struct net_device *dev,
240 unsigned int flags, u16 type, bool noxfrm);
241 struct rtable *rt_dst_clone(struct net_device *dev, struct rtable *rt);
244 void fib_add_ifaddr(struct in_ifaddr *);
245 void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
246 void fib_modify_prefix_metric(struct in_ifaddr *ifa, u32 new_metric);
248 void rt_add_uncached_list(struct rtable *rt);
249 void rt_del_uncached_list(struct rtable *rt);
251 int fib_dump_info_fnhe(struct sk_buff *skb, struct netlink_callback *cb,
252 u32 table_id, struct fib_info *fi,
253 int *fa_index, int fa_start, unsigned int flags);
255 static inline void ip_rt_put(struct rtable *rt)
257 /* dst_release() accepts a NULL parameter.
258 * We rely on dst being first structure in struct rtable
260 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
261 dst_release(&rt->dst);
264 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
266 extern const __u8 ip_tos2prio[16];
268 static inline char rt_tos2priority(u8 tos)
270 return ip_tos2prio[IPTOS_TOS(tos)>>1];
273 /* ip_route_connect() and ip_route_newports() work in tandem whilst
274 * binding a socket for a new outgoing connection.
276 * In order to use IPSEC properly, we must, in the end, have a
277 * route that was looked up using all available keys including source
278 * and destination ports.
280 * However, if a source port needs to be allocated (the user specified
281 * a wildcard source port) we need to obtain addressing information
282 * in order to perform that allocation.
284 * So ip_route_connect() looks up a route using wildcarded source and
285 * destination ports in the key, simply so that we can get a pair of
286 * addresses to use for port allocation.
288 * Later, once the ports are allocated, ip_route_newports() will make
289 * another route lookup if needed to make sure we catch any IPSEC
290 * rules keyed on the port information.
292 * The callers allocate the flow key on their stack, and must pass in
293 * the same flowi4 object to both the ip_route_connect() and the
294 * ip_route_newports() calls.
297 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst,
298 __be32 src, int oif, u8 protocol,
299 __be16 sport, __be16 dport,
300 const struct sock *sk)
304 if (inet_sk(sk)->transparent)
305 flow_flags |= FLOWI_FLAG_ANYSRC;
307 flowi4_init_output(fl4, oif, sk->sk_mark, ip_sock_rt_tos(sk),
308 ip_sock_rt_scope(sk), protocol, flow_flags, dst,
309 src, dport, sport, sk->sk_uid);
312 static inline struct rtable *ip_route_connect(struct flowi4 *fl4, __be32 dst,
313 __be32 src, int oif, u8 protocol,
314 __be16 sport, __be16 dport,
317 struct net *net = sock_net(sk);
320 ip_route_connect_init(fl4, dst, src, oif, protocol, sport, dport, sk);
323 rt = __ip_route_output_key(net, fl4);
327 flowi4_update_output(fl4, oif, fl4->flowi4_tos, fl4->daddr,
330 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
331 return ip_route_output_flow(net, fl4, sk);
334 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
335 __be16 orig_sport, __be16 orig_dport,
336 __be16 sport, __be16 dport,
339 if (sport != orig_sport || dport != orig_dport) {
340 fl4->fl4_dport = dport;
341 fl4->fl4_sport = sport;
343 flowi4_update_output(fl4, sk->sk_bound_dev_if,
344 RT_CONN_FLAGS(sk), fl4->daddr,
346 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
347 return ip_route_output_flow(sock_net(sk), fl4, sk);
352 static inline int inet_iif(const struct sk_buff *skb)
354 struct rtable *rt = skb_rtable(skb);
356 if (rt && rt->rt_iif)
362 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
364 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
365 struct net *net = dev_net(dst->dev);
368 hoplimit = READ_ONCE(net->ipv4.sysctl_ip_default_ttl);
372 static inline struct neighbour *ip_neigh_gw4(struct net_device *dev,
375 struct neighbour *neigh;
377 neigh = __ipv4_neigh_lookup_noref(dev, (__force u32)daddr);
378 if (unlikely(!neigh))
379 neigh = __neigh_create(&arp_tbl, &daddr, dev, false);
384 static inline struct neighbour *ip_neigh_for_gw(struct rtable *rt,
388 struct net_device *dev = rt->dst.dev;
389 struct neighbour *neigh;
391 if (likely(rt->rt_gw_family == AF_INET)) {
392 neigh = ip_neigh_gw4(dev, rt->rt_gw4);
393 } else if (rt->rt_gw_family == AF_INET6) {
394 neigh = ip_neigh_gw6(dev, &rt->rt_gw6);
397 neigh = ip_neigh_gw4(dev, ip_hdr(skb)->daddr);
402 #endif /* _ROUTE_H */