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 /* IPv4 datagram length is stored into 16bit field (tot_len) */
39 #define IP_MAX_MTU 0xFFFFU
41 #define RTO_ONLINK 0x01
43 #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
44 #define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
46 struct ip_tunnel_info;
54 unsigned int rt_flags;
62 /* Info on neighbour */
65 struct in6_addr rt_gw6;
68 /* Miscellaneous cached information */
72 struct list_head rt_uncached;
73 struct uncached_list *rt_uncached_list;
76 static inline bool rt_is_input_route(const struct rtable *rt)
78 return rt->rt_is_input != 0;
81 static inline bool rt_is_output_route(const struct rtable *rt)
83 return rt->rt_is_input == 0;
86 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
88 if (rt->rt_gw_family == AF_INET)
100 struct rt_cache_stat {
101 unsigned int in_slow_tot;
102 unsigned int in_slow_mc;
103 unsigned int in_no_route;
105 unsigned int in_martian_dst;
106 unsigned int in_martian_src;
107 unsigned int out_slow_tot;
108 unsigned int out_slow_mc;
111 extern struct ip_rt_acct __percpu *ip_rt_acct;
115 int ip_rt_init(void);
116 void rt_cache_flush(struct net *net);
117 void rt_flush_dev(struct net_device *dev);
118 struct rtable *ip_route_output_key_hash(struct net *net, struct flowi4 *flp,
119 const struct sk_buff *skb);
120 struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *flp,
121 struct fib_result *res,
122 const struct sk_buff *skb);
124 static inline struct rtable *__ip_route_output_key(struct net *net,
127 return ip_route_output_key_hash(net, flp, NULL);
130 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
131 const struct sock *sk);
132 struct rtable *ip_route_output_tunnel(struct sk_buff *skb,
133 struct net_device *dev,
134 struct net *net, __be32 *saddr,
135 const struct ip_tunnel_info *info,
136 u8 protocol, bool use_cache);
138 struct dst_entry *ipv4_blackhole_route(struct net *net,
139 struct dst_entry *dst_orig);
141 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
143 return ip_route_output_flow(net, flp, NULL);
146 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
147 __be32 saddr, u8 tos, int oif)
149 struct flowi4 fl4 = {
155 return ip_route_output_key(net, &fl4);
158 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
160 __be32 daddr, __be32 saddr,
161 __be16 dport, __be16 sport,
162 __u8 proto, __u8 tos, int oif)
164 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
165 RT_SCOPE_UNIVERSE, proto,
166 sk ? inet_sk_flowi_flags(sk) : 0,
167 daddr, saddr, dport, sport, sock_net_uid(net, sk));
169 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
170 return ip_route_output_flow(net, fl4, sk);
173 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
174 __be32 daddr, __be32 saddr,
175 __be32 gre_key, __u8 tos, int oif)
177 memset(fl4, 0, sizeof(*fl4));
178 fl4->flowi4_oif = oif;
181 fl4->flowi4_tos = tos;
182 fl4->flowi4_proto = IPPROTO_GRE;
183 fl4->fl4_gre_key = gre_key;
184 return ip_route_output_key(net, fl4);
186 int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr,
187 u8 tos, struct net_device *dev,
188 struct in_device *in_dev, u32 *itag);
189 int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
190 u8 tos, struct net_device *devin);
191 int ip_route_input_rcu(struct sk_buff *skb, __be32 dst, __be32 src,
192 u8 tos, struct net_device *devin,
193 struct fib_result *res);
195 int ip_route_use_hint(struct sk_buff *skb, __be32 dst, __be32 src,
196 u8 tos, struct net_device *devin,
197 const struct sk_buff *hint);
199 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
200 u8 tos, struct net_device *devin)
205 err = ip_route_input_noref(skb, dst, src, tos, devin);
216 void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
218 void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
219 void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u8 protocol);
220 void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
221 void ip_rt_send_redirect(struct sk_buff *skb);
223 unsigned int inet_addr_type(struct net *net, __be32 addr);
224 unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
225 unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
227 unsigned int inet_addr_type_dev_table(struct net *net,
228 const struct net_device *dev,
230 void ip_rt_multicast_event(struct in_device *);
231 int ip_rt_ioctl(struct net *, unsigned int cmd, struct rtentry *rt);
232 void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
233 struct rtable *rt_dst_alloc(struct net_device *dev,
234 unsigned int flags, u16 type,
235 bool nopolicy, bool noxfrm);
236 struct rtable *rt_dst_clone(struct net_device *dev, struct rtable *rt);
239 void fib_add_ifaddr(struct in_ifaddr *);
240 void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
241 void fib_modify_prefix_metric(struct in_ifaddr *ifa, u32 new_metric);
243 void rt_add_uncached_list(struct rtable *rt);
244 void rt_del_uncached_list(struct rtable *rt);
246 int fib_dump_info_fnhe(struct sk_buff *skb, struct netlink_callback *cb,
247 u32 table_id, struct fib_info *fi,
248 int *fa_index, int fa_start, unsigned int flags);
250 static inline void ip_rt_put(struct rtable *rt)
252 /* dst_release() accepts a NULL parameter.
253 * We rely on dst being first structure in struct rtable
255 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
256 dst_release(&rt->dst);
259 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
261 extern const __u8 ip_tos2prio[16];
263 static inline char rt_tos2priority(u8 tos)
265 return ip_tos2prio[IPTOS_TOS(tos)>>1];
268 /* ip_route_connect() and ip_route_newports() work in tandem whilst
269 * binding a socket for a new outgoing connection.
271 * In order to use IPSEC properly, we must, in the end, have a
272 * route that was looked up using all available keys including source
273 * and destination ports.
275 * However, if a source port needs to be allocated (the user specified
276 * a wildcard source port) we need to obtain addressing information
277 * in order to perform that allocation.
279 * So ip_route_connect() looks up a route using wildcarded source and
280 * destination ports in the key, simply so that we can get a pair of
281 * addresses to use for port allocation.
283 * Later, once the ports are allocated, ip_route_newports() will make
284 * another route lookup if needed to make sure we catch any IPSEC
285 * rules keyed on the port information.
287 * The callers allocate the flow key on their stack, and must pass in
288 * the same flowi4 object to both the ip_route_connect() and the
289 * ip_route_newports() calls.
292 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
293 u32 tos, int oif, u8 protocol,
294 __be16 sport, __be16 dport,
299 if (inet_sk(sk)->transparent)
300 flow_flags |= FLOWI_FLAG_ANYSRC;
302 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
303 protocol, flow_flags, dst, src, dport, sport,
307 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
308 __be32 dst, __be32 src, u32 tos,
309 int oif, u8 protocol,
310 __be16 sport, __be16 dport,
313 struct net *net = sock_net(sk);
316 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
320 rt = __ip_route_output_key(net, fl4);
324 flowi4_update_output(fl4, oif, tos, 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,
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,
340 RT_CONN_FLAGS(sk), fl4->daddr,
342 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
343 return ip_route_output_flow(sock_net(sk), fl4, sk);
348 static inline int inet_iif(const struct sk_buff *skb)
350 struct rtable *rt = skb_rtable(skb);
352 if (rt && rt->rt_iif)
358 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
360 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
361 struct net *net = dev_net(dst->dev);
364 hoplimit = net->ipv4.sysctl_ip_default_ttl;
368 static inline struct neighbour *ip_neigh_gw4(struct net_device *dev,
371 struct neighbour *neigh;
373 neigh = __ipv4_neigh_lookup_noref(dev, (__force u32)daddr);
374 if (unlikely(!neigh))
375 neigh = __neigh_create(&arp_tbl, &daddr, dev, false);
380 static inline struct neighbour *ip_neigh_for_gw(struct rtable *rt,
384 struct net_device *dev = rt->dst.dev;
385 struct neighbour *neigh;
387 if (likely(rt->rt_gw_family == AF_INET)) {
388 neigh = ip_neigh_gw4(dev, rt->rt_gw4);
389 } else if (rt->rt_gw_family == AF_INET6) {
390 neigh = ip_neigh_gw6(dev, &rt->rt_gw6);
393 neigh = ip_neigh_gw4(dev, ip_hdr(skb)->daddr);
398 #endif /* _ROUTE_H */