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(READ_ONCE(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(READ_ONCE(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 */
83 static inline bool rt_is_input_route(const struct rtable *rt)
85 return rt->rt_is_input != 0;
88 static inline bool rt_is_output_route(const struct rtable *rt)
90 return rt->rt_is_input == 0;
93 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
95 if (rt->rt_gw_family == AF_INET)
107 struct rt_cache_stat {
108 unsigned int in_slow_tot;
109 unsigned int in_slow_mc;
110 unsigned int in_no_route;
112 unsigned int in_martian_dst;
113 unsigned int in_martian_src;
114 unsigned int out_slow_tot;
115 unsigned int out_slow_mc;
118 extern struct ip_rt_acct __percpu *ip_rt_acct;
122 int ip_rt_init(void);
123 void rt_cache_flush(struct net *net);
124 void rt_flush_dev(struct net_device *dev);
125 struct rtable *ip_route_output_key_hash(struct net *net, struct flowi4 *flp,
126 const struct sk_buff *skb);
127 struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *flp,
128 struct fib_result *res,
129 const struct sk_buff *skb);
131 static inline struct rtable *__ip_route_output_key(struct net *net,
134 return ip_route_output_key_hash(net, flp, NULL);
137 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
138 const struct sock *sk);
139 struct rtable *ip_route_output_tunnel(struct sk_buff *skb,
140 struct net_device *dev,
141 struct net *net, __be32 *saddr,
142 const struct ip_tunnel_info *info,
143 u8 protocol, bool use_cache);
145 struct dst_entry *ipv4_blackhole_route(struct net *net,
146 struct dst_entry *dst_orig);
148 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
150 return ip_route_output_flow(net, flp, NULL);
153 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
154 __be32 saddr, u8 tos, int oif)
156 struct flowi4 fl4 = {
162 return ip_route_output_key(net, &fl4);
165 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
166 const struct sock *sk,
167 __be32 daddr, __be32 saddr,
168 __be16 dport, __be16 sport,
169 __u8 proto, __u8 tos, int oif)
171 flowi4_init_output(fl4, oif, sk ? READ_ONCE(sk->sk_mark) : 0, tos,
172 RT_SCOPE_UNIVERSE, proto,
173 sk ? inet_sk_flowi_flags(sk) : 0,
174 daddr, saddr, dport, sport, sock_net_uid(net, sk));
176 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
177 return ip_route_output_flow(net, fl4, sk);
180 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
181 __be32 daddr, __be32 saddr,
182 __be32 gre_key, __u8 tos, int oif)
184 memset(fl4, 0, sizeof(*fl4));
185 fl4->flowi4_oif = oif;
188 fl4->flowi4_tos = tos;
189 fl4->flowi4_proto = IPPROTO_GRE;
190 fl4->fl4_gre_key = gre_key;
191 return ip_route_output_key(net, fl4);
193 int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr,
194 u8 tos, struct net_device *dev,
195 struct in_device *in_dev, u32 *itag);
196 int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
197 u8 tos, struct net_device *devin);
198 int ip_route_use_hint(struct sk_buff *skb, __be32 dst, __be32 src,
199 u8 tos, struct net_device *devin,
200 const struct sk_buff *hint);
202 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
203 u8 tos, struct net_device *devin)
208 err = ip_route_input_noref(skb, dst, src, tos, devin);
219 void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
221 void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
222 void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u8 protocol);
223 void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
224 void ip_rt_send_redirect(struct sk_buff *skb);
226 unsigned int inet_addr_type(struct net *net, __be32 addr);
227 unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
228 unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
230 unsigned int inet_addr_type_dev_table(struct net *net,
231 const struct net_device *dev,
233 void ip_rt_multicast_event(struct in_device *);
234 int ip_rt_ioctl(struct net *, unsigned int cmd, struct rtentry *rt);
235 void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
236 struct rtable *rt_dst_alloc(struct net_device *dev,
237 unsigned int flags, u16 type, bool noxfrm);
238 struct rtable *rt_dst_clone(struct net_device *dev, struct rtable *rt);
241 void fib_add_ifaddr(struct in_ifaddr *);
242 void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
243 void fib_modify_prefix_metric(struct in_ifaddr *ifa, u32 new_metric);
245 void rt_add_uncached_list(struct rtable *rt);
246 void rt_del_uncached_list(struct rtable *rt);
248 int fib_dump_info_fnhe(struct sk_buff *skb, struct netlink_callback *cb,
249 u32 table_id, struct fib_info *fi,
250 int *fa_index, int fa_start, unsigned int flags);
252 static inline void ip_rt_put(struct rtable *rt)
254 /* dst_release() accepts a NULL parameter.
255 * We rely on dst being first structure in struct rtable
257 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
258 dst_release(&rt->dst);
261 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
263 extern const __u8 ip_tos2prio[16];
265 static inline char rt_tos2priority(u8 tos)
267 return ip_tos2prio[IPTOS_TOS(tos)>>1];
270 /* ip_route_connect() and ip_route_newports() work in tandem whilst
271 * binding a socket for a new outgoing connection.
273 * In order to use IPSEC properly, we must, in the end, have a
274 * route that was looked up using all available keys including source
275 * and destination ports.
277 * However, if a source port needs to be allocated (the user specified
278 * a wildcard source port) we need to obtain addressing information
279 * in order to perform that allocation.
281 * So ip_route_connect() looks up a route using wildcarded source and
282 * destination ports in the key, simply so that we can get a pair of
283 * addresses to use for port allocation.
285 * Later, once the ports are allocated, ip_route_newports() will make
286 * another route lookup if needed to make sure we catch any IPSEC
287 * rules keyed on the port information.
289 * The callers allocate the flow key on their stack, and must pass in
290 * the same flowi4 object to both the ip_route_connect() and the
291 * ip_route_newports() calls.
294 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst,
295 __be32 src, int oif, u8 protocol,
296 __be16 sport, __be16 dport,
297 const struct sock *sk)
301 if (inet_test_bit(TRANSPARENT, sk))
302 flow_flags |= FLOWI_FLAG_ANYSRC;
304 flowi4_init_output(fl4, oif, READ_ONCE(sk->sk_mark), ip_sock_rt_tos(sk),
305 ip_sock_rt_scope(sk), protocol, flow_flags, dst,
306 src, dport, sport, sk->sk_uid);
309 static inline struct rtable *ip_route_connect(struct flowi4 *fl4, __be32 dst,
310 __be32 src, int oif, u8 protocol,
311 __be16 sport, __be16 dport,
312 const struct sock *sk)
314 struct net *net = sock_net(sk);
317 ip_route_connect_init(fl4, dst, src, oif, protocol, sport, dport, sk);
320 rt = __ip_route_output_key(net, fl4);
324 flowi4_update_output(fl4, oif, fl4->daddr, fl4->saddr);
326 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
327 return ip_route_output_flow(net, fl4, sk);
330 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
331 __be16 orig_sport, __be16 orig_dport,
332 __be16 sport, __be16 dport,
333 const struct sock *sk)
335 if (sport != orig_sport || dport != orig_dport) {
336 fl4->fl4_dport = dport;
337 fl4->fl4_sport = sport;
339 flowi4_update_output(fl4, sk->sk_bound_dev_if, fl4->daddr,
341 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
342 return ip_route_output_flow(sock_net(sk), fl4, sk);
347 static inline int inet_iif(const struct sk_buff *skb)
349 struct rtable *rt = skb_rtable(skb);
351 if (rt && rt->rt_iif)
357 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
359 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
360 struct net *net = dev_net(dst->dev);
363 hoplimit = READ_ONCE(net->ipv4.sysctl_ip_default_ttl);
367 static inline struct neighbour *ip_neigh_gw4(struct net_device *dev,
370 struct neighbour *neigh;
372 neigh = __ipv4_neigh_lookup_noref(dev, (__force u32)daddr);
373 if (unlikely(!neigh))
374 neigh = __neigh_create(&arp_tbl, &daddr, dev, false);
379 static inline struct neighbour *ip_neigh_for_gw(struct rtable *rt,
383 struct net_device *dev = rt->dst.dev;
384 struct neighbour *neigh;
386 if (likely(rt->rt_gw_family == AF_INET)) {
387 neigh = ip_neigh_gw4(dev, rt->rt_gw4);
388 } else if (rt->rt_gw_family == AF_INET6) {
389 neigh = ip_neigh_gw6(dev, &rt->rt_gw6);
392 neigh = ip_neigh_gw4(dev, ip_hdr(skb)->daddr);
397 #endif /* _ROUTE_H */