]>
Commit | Line | Data |
---|---|---|
2874c5fd | 1 | /* SPDX-License-Identifier: GPL-2.0-or-later */ |
1da177e4 LT |
2 | /* |
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. | |
6 | * | |
7 | * Definitions for the IP router. | |
8 | * | |
9 | * Version: @(#)route.h 1.0.4 05/27/93 | |
10 | * | |
02c30a84 | 11 | * Authors: Ross Biro |
1da177e4 LT |
12 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
13 | * Fixes: | |
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 | |
1da177e4 LT |
19 | */ |
20 | #ifndef _ROUTE_H | |
21 | #define _ROUTE_H | |
22 | ||
1da177e4 LT |
23 | #include <net/dst.h> |
24 | #include <net/inetpeer.h> | |
25 | #include <net/flow.h> | |
79876874 | 26 | #include <net/inet_sock.h> |
79a13159 | 27 | #include <net/ip_fib.h> |
5c9f7c1d DA |
28 | #include <net/arp.h> |
29 | #include <net/ndisc.h> | |
1da177e4 LT |
30 | #include <linux/in_route.h> |
31 | #include <linux/rtnetlink.h> | |
c6cffba4 | 32 | #include <linux/rcupdate.h> |
1da177e4 LT |
33 | #include <linux/route.h> |
34 | #include <linux/ip.h> | |
35 | #include <linux/cache.h> | |
beb8d13b | 36 | #include <linux/security.h> |
1da177e4 | 37 | |
f87c10a8 HFS |
38 | /* IPv4 datagram length is stored into 16bit field (tot_len) */ |
39 | #define IP_MAX_MTU 0xFFFFU | |
40 | ||
1da177e4 LT |
41 | #define RTO_ONLINK 0x01 |
42 | ||
1da177e4 | 43 | #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE)) |
aa661581 | 44 | #define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE)) |
1da177e4 LT |
45 | |
46 | struct fib_nh; | |
62fa8a84 | 47 | struct fib_info; |
5055c371 | 48 | struct uncached_list; |
fd2c3ef7 | 49 | struct rtable { |
d8d1f30b | 50 | struct dst_entry dst; |
1da177e4 | 51 | |
29e75252 | 52 | int rt_genid; |
95c96174 | 53 | unsigned int rt_flags; |
1da177e4 | 54 | __u16 rt_type; |
155e8336 | 55 | __u8 rt_is_input; |
1550c171 | 56 | u8 rt_gw_family; |
1da177e4 | 57 | |
1da177e4 LT |
58 | int rt_iif; |
59 | ||
60 | /* Info on neighbour */ | |
0f5f7d7b DA |
61 | union { |
62 | __be32 rt_gw4; | |
63 | struct in6_addr rt_gw6; | |
64 | }; | |
1da177e4 | 65 | |
1da177e4 | 66 | /* Miscellaneous cached information */ |
d52e5a7e SD |
67 | u32 rt_mtu_locked:1, |
68 | rt_pmtu:31; | |
caacf05e DM |
69 | |
70 | struct list_head rt_uncached; | |
5055c371 | 71 | struct uncached_list *rt_uncached_list; |
1da177e4 LT |
72 | }; |
73 | ||
b8400f37 | 74 | static inline bool rt_is_input_route(const struct rtable *rt) |
c7537967 | 75 | { |
9917e1e8 | 76 | return rt->rt_is_input != 0; |
c7537967 DM |
77 | } |
78 | ||
b8400f37 | 79 | static inline bool rt_is_output_route(const struct rtable *rt) |
c7537967 | 80 | { |
9917e1e8 | 81 | return rt->rt_is_input == 0; |
c7537967 DM |
82 | } |
83 | ||
f8126f1d DM |
84 | static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr) |
85 | { | |
1550c171 DA |
86 | if (rt->rt_gw_family == AF_INET) |
87 | return rt->rt_gw4; | |
f8126f1d DM |
88 | return daddr; |
89 | } | |
90 | ||
fd2c3ef7 | 91 | struct ip_rt_acct { |
1da177e4 LT |
92 | __u32 o_bytes; |
93 | __u32 o_packets; | |
94 | __u32 i_bytes; | |
95 | __u32 i_packets; | |
96 | }; | |
97 | ||
fd2c3ef7 | 98 | struct rt_cache_stat { |
1da177e4 LT |
99 | unsigned int in_slow_tot; |
100 | unsigned int in_slow_mc; | |
101 | unsigned int in_no_route; | |
102 | unsigned int in_brd; | |
103 | unsigned int in_martian_dst; | |
104 | unsigned int in_martian_src; | |
1da177e4 LT |
105 | unsigned int out_slow_tot; |
106 | unsigned int out_slow_mc; | |
1da177e4 LT |
107 | }; |
108 | ||
7d720c3e | 109 | extern struct ip_rt_acct __percpu *ip_rt_acct; |
1da177e4 LT |
110 | |
111 | struct in_device; | |
2702c4bb JP |
112 | |
113 | int ip_rt_init(void); | |
114 | void rt_cache_flush(struct net *net); | |
115 | void rt_flush_dev(struct net_device *dev); | |
3abd1ade DA |
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); | |
79a13159 PN |
121 | |
122 | static inline struct rtable *__ip_route_output_key(struct net *net, | |
123 | struct flowi4 *flp) | |
124 | { | |
3abd1ade | 125 | return ip_route_output_key_hash(net, flp, NULL); |
79a13159 PN |
126 | } |
127 | ||
2702c4bb | 128 | struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp, |
6f9c9615 | 129 | const struct sock *sk); |
2702c4bb JP |
130 | struct dst_entry *ipv4_blackhole_route(struct net *net, |
131 | struct dst_entry *dst_orig); | |
407eadd9 | 132 | |
9d6ec938 | 133 | static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp) |
5bfa787f DM |
134 | { |
135 | return ip_route_output_flow(net, flp, NULL); | |
136 | } | |
137 | ||
78fbfd8a DM |
138 | static inline struct rtable *ip_route_output(struct net *net, __be32 daddr, |
139 | __be32 saddr, u8 tos, int oif) | |
140 | { | |
9d6ec938 DM |
141 | struct flowi4 fl4 = { |
142 | .flowi4_oif = oif, | |
c5d21c4b | 143 | .flowi4_tos = tos, |
9d6ec938 DM |
144 | .daddr = daddr, |
145 | .saddr = saddr, | |
78fbfd8a | 146 | }; |
9d6ec938 | 147 | return ip_route_output_key(net, &fl4); |
78fbfd8a DM |
148 | } |
149 | ||
31e4543d DM |
150 | static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4, |
151 | struct sock *sk, | |
78fbfd8a DM |
152 | __be32 daddr, __be32 saddr, |
153 | __be16 dport, __be16 sport, | |
154 | __u8 proto, __u8 tos, int oif) | |
155 | { | |
31e4543d | 156 | flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos, |
94b92b88 DM |
157 | RT_SCOPE_UNIVERSE, proto, |
158 | sk ? inet_sk_flowi_flags(sk) : 0, | |
e2d118a1 | 159 | daddr, saddr, dport, sport, sock_net_uid(net, sk)); |
78fbfd8a | 160 | if (sk) |
31e4543d DM |
161 | security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); |
162 | return ip_route_output_flow(net, fl4, sk); | |
78fbfd8a DM |
163 | } |
164 | ||
cbb1e85f | 165 | static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4, |
78fbfd8a DM |
166 | __be32 daddr, __be32 saddr, |
167 | __be32 gre_key, __u8 tos, int oif) | |
168 | { | |
cbb1e85f DM |
169 | memset(fl4, 0, sizeof(*fl4)); |
170 | fl4->flowi4_oif = oif; | |
171 | fl4->daddr = daddr; | |
172 | fl4->saddr = saddr; | |
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); | |
78fbfd8a | 177 | } |
bc044e8d PA |
178 | int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr, |
179 | u8 tos, struct net_device *dev, | |
180 | struct in_device *in_dev, u32 *itag); | |
2702c4bb JP |
181 | int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src, |
182 | u8 tos, struct net_device *devin); | |
5510cdf7 DA |
183 | int ip_route_input_rcu(struct sk_buff *skb, __be32 dst, __be32 src, |
184 | u8 tos, struct net_device *devin, | |
185 | struct fib_result *res); | |
c6cffba4 DM |
186 | |
187 | static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src, | |
188 | u8 tos, struct net_device *devin) | |
189 | { | |
190 | int err; | |
191 | ||
192 | rcu_read_lock(); | |
193 | err = ip_route_input_noref(skb, dst, src, tos, devin); | |
64327fc8 | 194 | if (!err) { |
222d7dbd | 195 | skb_dst_force(skb); |
64327fc8 SB |
196 | if (!skb_dst(skb)) |
197 | err = -EINVAL; | |
198 | } | |
c6cffba4 DM |
199 | rcu_read_unlock(); |
200 | ||
201 | return err; | |
202 | } | |
407eadd9 | 203 | |
2702c4bb | 204 | void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif, |
d888f396 | 205 | u8 protocol); |
2702c4bb | 206 | void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu); |
1042caa7 | 207 | void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u8 protocol); |
2702c4bb JP |
208 | void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk); |
209 | void ip_rt_send_redirect(struct sk_buff *skb); | |
210 | ||
211 | unsigned int inet_addr_type(struct net *net, __be32 addr); | |
9b8ff518 | 212 | unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id); |
2702c4bb JP |
213 | unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev, |
214 | __be32 addr); | |
30bbaa19 DA |
215 | unsigned int inet_addr_type_dev_table(struct net *net, |
216 | const struct net_device *dev, | |
217 | __be32 addr); | |
2702c4bb | 218 | void ip_rt_multicast_event(struct in_device *); |
ca25c300 | 219 | int ip_rt_ioctl(struct net *, unsigned int cmd, struct rtentry *rt); |
2702c4bb | 220 | void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt); |
9ab179d8 DA |
221 | struct rtable *rt_dst_alloc(struct net_device *dev, |
222 | unsigned int flags, u16 type, | |
223 | bool nopolicy, bool noxfrm, bool will_cache); | |
5b18f128 | 224 | struct rtable *rt_dst_clone(struct net_device *dev, struct rtable *rt); |
1da177e4 | 225 | |
0ff60a45 | 226 | struct in_ifaddr; |
2702c4bb JP |
227 | void fib_add_ifaddr(struct in_ifaddr *); |
228 | void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *); | |
af4d768a | 229 | void fib_modify_prefix_metric(struct in_ifaddr *ifa, u32 new_metric); |
0ff60a45 | 230 | |
510c321b XL |
231 | void rt_add_uncached_list(struct rtable *rt); |
232 | void rt_del_uncached_list(struct rtable *rt); | |
233 | ||
ee28906f SB |
234 | int fib_dump_info_fnhe(struct sk_buff *skb, struct netlink_callback *cb, |
235 | u32 table_id, struct fib_info *fi, | |
e93fb3e9 | 236 | int *fa_index, int fa_start, unsigned int flags); |
ee28906f | 237 | |
6da025fa | 238 | static inline void ip_rt_put(struct rtable *rt) |
1da177e4 | 239 | { |
6da025fa ED |
240 | /* dst_release() accepts a NULL parameter. |
241 | * We rely on dst being first structure in struct rtable | |
242 | */ | |
243 | BUILD_BUG_ON(offsetof(struct rtable, dst) != 0); | |
244 | dst_release(&rt->dst); | |
1da177e4 LT |
245 | } |
246 | ||
247 | #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3) | |
248 | ||
4839c52b | 249 | extern const __u8 ip_tos2prio[16]; |
1da177e4 LT |
250 | |
251 | static inline char rt_tos2priority(u8 tos) | |
252 | { | |
253 | return ip_tos2prio[IPTOS_TOS(tos)>>1]; | |
254 | } | |
255 | ||
2d7192d6 DM |
256 | /* ip_route_connect() and ip_route_newports() work in tandem whilst |
257 | * binding a socket for a new outgoing connection. | |
258 | * | |
259 | * In order to use IPSEC properly, we must, in the end, have a | |
260 | * route that was looked up using all available keys including source | |
261 | * and destination ports. | |
262 | * | |
263 | * However, if a source port needs to be allocated (the user specified | |
264 | * a wildcard source port) we need to obtain addressing information | |
265 | * in order to perform that allocation. | |
266 | * | |
267 | * So ip_route_connect() looks up a route using wildcarded source and | |
268 | * destination ports in the key, simply so that we can get a pair of | |
269 | * addresses to use for port allocation. | |
270 | * | |
271 | * Later, once the ports are allocated, ip_route_newports() will make | |
272 | * another route lookup if needed to make sure we catch any IPSEC | |
273 | * rules keyed on the port information. | |
274 | * | |
275 | * The callers allocate the flow key on their stack, and must pass in | |
276 | * the same flowi4 object to both the ip_route_connect() and the | |
277 | * ip_route_newports() calls. | |
278 | */ | |
279 | ||
280 | static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src, | |
281 | u32 tos, int oif, u8 protocol, | |
282 | __be16 sport, __be16 dport, | |
0e0d44ab | 283 | struct sock *sk) |
1da177e4 | 284 | { |
2d7192d6 | 285 | __u8 flow_flags = 0; |
79876874 KK |
286 | |
287 | if (inet_sk(sk)->transparent) | |
94b92b88 | 288 | flow_flags |= FLOWI_FLAG_ANYSRC; |
94b92b88 | 289 | |
2d7192d6 | 290 | flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE, |
e2d118a1 LC |
291 | protocol, flow_flags, dst, src, dport, sport, |
292 | sk->sk_uid); | |
2d7192d6 DM |
293 | } |
294 | ||
295 | static inline struct rtable *ip_route_connect(struct flowi4 *fl4, | |
296 | __be32 dst, __be32 src, u32 tos, | |
297 | int oif, u8 protocol, | |
298 | __be16 sport, __be16 dport, | |
0e0d44ab | 299 | struct sock *sk) |
2d7192d6 DM |
300 | { |
301 | struct net *net = sock_net(sk); | |
302 | struct rtable *rt; | |
303 | ||
304 | ip_route_connect_init(fl4, dst, src, tos, oif, protocol, | |
0e0d44ab | 305 | sport, dport, sk); |
79876874 | 306 | |
1da177e4 | 307 | if (!dst || !src) { |
2d7192d6 | 308 | rt = __ip_route_output_key(net, fl4); |
b23dd4fe DM |
309 | if (IS_ERR(rt)) |
310 | return rt; | |
b23dd4fe | 311 | ip_rt_put(rt); |
e6b45241 | 312 | flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr); |
1da177e4 | 313 | } |
2d7192d6 DM |
314 | security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); |
315 | return ip_route_output_flow(net, fl4, sk); | |
1da177e4 LT |
316 | } |
317 | ||
2d7192d6 DM |
318 | static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt, |
319 | __be16 orig_sport, __be16 orig_dport, | |
320 | __be16 sport, __be16 dport, | |
321 | struct sock *sk) | |
1da177e4 | 322 | { |
dca8b089 | 323 | if (sport != orig_sport || dport != orig_dport) { |
2d7192d6 DM |
324 | fl4->fl4_dport = dport; |
325 | fl4->fl4_sport = sport; | |
b23dd4fe | 326 | ip_rt_put(rt); |
e6b45241 JA |
327 | flowi4_update_output(fl4, sk->sk_bound_dev_if, |
328 | RT_CONN_FLAGS(sk), fl4->daddr, | |
329 | fl4->saddr); | |
2d7192d6 DM |
330 | security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); |
331 | return ip_route_output_flow(sock_net(sk), fl4, sk); | |
1da177e4 | 332 | } |
b23dd4fe | 333 | return rt; |
1da177e4 LT |
334 | } |
335 | ||
1668e010 KK |
336 | static inline int inet_iif(const struct sk_buff *skb) |
337 | { | |
0340d0b9 TH |
338 | struct rtable *rt = skb_rtable(skb); |
339 | ||
340 | if (rt && rt->rt_iif) | |
341 | return rt->rt_iif; | |
13378cad | 342 | |
13378cad | 343 | return skb->skb_iif; |
1668e010 KK |
344 | } |
345 | ||
323e126f DM |
346 | static inline int ip4_dst_hoplimit(const struct dst_entry *dst) |
347 | { | |
348 | int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); | |
fa50d974 | 349 | struct net *net = dev_net(dst->dev); |
323e126f DM |
350 | |
351 | if (hoplimit == 0) | |
fa50d974 | 352 | hoplimit = net->ipv4.sysctl_ip_default_ttl; |
323e126f DM |
353 | return hoplimit; |
354 | } | |
355 | ||
5c9f7c1d DA |
356 | static inline struct neighbour *ip_neigh_gw4(struct net_device *dev, |
357 | __be32 daddr) | |
358 | { | |
359 | struct neighbour *neigh; | |
360 | ||
361 | neigh = __ipv4_neigh_lookup_noref(dev, daddr); | |
362 | if (unlikely(!neigh)) | |
363 | neigh = __neigh_create(&arp_tbl, &daddr, dev, false); | |
364 | ||
365 | return neigh; | |
366 | } | |
367 | ||
368 | static inline struct neighbour *ip_neigh_for_gw(struct rtable *rt, | |
369 | struct sk_buff *skb, | |
370 | bool *is_v6gw) | |
371 | { | |
372 | struct net_device *dev = rt->dst.dev; | |
373 | struct neighbour *neigh; | |
374 | ||
375 | if (likely(rt->rt_gw_family == AF_INET)) { | |
376 | neigh = ip_neigh_gw4(dev, rt->rt_gw4); | |
377 | } else if (rt->rt_gw_family == AF_INET6) { | |
378 | neigh = ip_neigh_gw6(dev, &rt->rt_gw6); | |
379 | *is_v6gw = true; | |
380 | } else { | |
381 | neigh = ip_neigh_gw4(dev, ip_hdr(skb)->daddr); | |
382 | } | |
383 | return neigh; | |
384 | } | |
385 | ||
1da177e4 | 386 | #endif /* _ROUTE_H */ |