route.h 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353
  1. /*
  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.
  5. *
  6. * Definitions for the IP router.
  7. *
  8. * Version: @(#)route.h 1.0.4 05/27/93
  9. *
  10. * Authors: Ross Biro
  11. * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
  12. * Fixes:
  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
  18. *
  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.
  23. */
  24. #ifndef _ROUTE_H
  25. #define _ROUTE_H
  26. #include <net/dst.h>
  27. #include <net/inetpeer.h>
  28. #include <net/flow.h>
  29. #include <net/inet_sock.h>
  30. #include <net/ip_fib.h>
  31. #include <linux/in_route.h>
  32. #include <linux/rtnetlink.h>
  33. #include <linux/rcupdate.h>
  34. #include <linux/route.h>
  35. #include <linux/ip.h>
  36. #include <linux/cache.h>
  37. #include <linux/security.h>
  38. /* IPv4 datagram length is stored into 16bit field (tot_len) */
  39. #define IP_MAX_MTU 0xFFFFU
  40. #define RTO_ONLINK 0x01
  41. #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
  42. #define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
  43. struct fib_nh;
  44. struct fib_info;
  45. struct uncached_list;
  46. struct rtable {
  47. struct dst_entry dst;
  48. int rt_genid;
  49. unsigned int rt_flags;
  50. __u16 rt_type;
  51. __u8 rt_is_input;
  52. __u8 rt_uses_gateway;
  53. int rt_iif;
  54. /* Info on neighbour */
  55. __be32 rt_gateway;
  56. /* Miscellaneous cached information */
  57. u32 rt_mtu_locked:1,
  58. rt_pmtu:31;
  59. u32 rt_table_id;
  60. struct list_head rt_uncached;
  61. struct uncached_list *rt_uncached_list;
  62. };
  63. static inline bool rt_is_input_route(const struct rtable *rt)
  64. {
  65. return rt->rt_is_input != 0;
  66. }
  67. static inline bool rt_is_output_route(const struct rtable *rt)
  68. {
  69. return rt->rt_is_input == 0;
  70. }
  71. static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
  72. {
  73. if (rt->rt_gateway)
  74. return rt->rt_gateway;
  75. return daddr;
  76. }
  77. struct ip_rt_acct {
  78. __u32 o_bytes;
  79. __u32 o_packets;
  80. __u32 i_bytes;
  81. __u32 i_packets;
  82. };
  83. struct rt_cache_stat {
  84. unsigned int in_slow_tot;
  85. unsigned int in_slow_mc;
  86. unsigned int in_no_route;
  87. unsigned int in_brd;
  88. unsigned int in_martian_dst;
  89. unsigned int in_martian_src;
  90. unsigned int out_slow_tot;
  91. unsigned int out_slow_mc;
  92. };
  93. extern struct ip_rt_acct __percpu *ip_rt_acct;
  94. struct in_device;
  95. int ip_rt_init(void);
  96. void rt_cache_flush(struct net *net);
  97. void rt_flush_dev(struct net_device *dev);
  98. struct rtable *ip_route_output_key_hash(struct net *net, struct flowi4 *flp,
  99. const struct sk_buff *skb);
  100. struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *flp,
  101. struct fib_result *res,
  102. const struct sk_buff *skb);
  103. static inline struct rtable *__ip_route_output_key(struct net *net,
  104. struct flowi4 *flp)
  105. {
  106. return ip_route_output_key_hash(net, flp, NULL);
  107. }
  108. struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
  109. const struct sock *sk);
  110. struct dst_entry *ipv4_blackhole_route(struct net *net,
  111. struct dst_entry *dst_orig);
  112. static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
  113. {
  114. return ip_route_output_flow(net, flp, NULL);
  115. }
  116. static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
  117. __be32 saddr, u8 tos, int oif)
  118. {
  119. struct flowi4 fl4 = {
  120. .flowi4_oif = oif,
  121. .flowi4_tos = tos,
  122. .daddr = daddr,
  123. .saddr = saddr,
  124. };
  125. return ip_route_output_key(net, &fl4);
  126. }
  127. static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
  128. struct sock *sk,
  129. __be32 daddr, __be32 saddr,
  130. __be16 dport, __be16 sport,
  131. __u8 proto, __u8 tos, int oif)
  132. {
  133. flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
  134. RT_SCOPE_UNIVERSE, proto,
  135. sk ? inet_sk_flowi_flags(sk) : 0,
  136. daddr, saddr, dport, sport, sock_net_uid(net, sk));
  137. if (sk)
  138. security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
  139. return ip_route_output_flow(net, fl4, sk);
  140. }
  141. static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
  142. __be32 daddr, __be32 saddr,
  143. __be32 gre_key, __u8 tos, int oif)
  144. {
  145. memset(fl4, 0, sizeof(*fl4));
  146. fl4->flowi4_oif = oif;
  147. fl4->daddr = daddr;
  148. fl4->saddr = saddr;
  149. fl4->flowi4_tos = tos;
  150. fl4->flowi4_proto = IPPROTO_GRE;
  151. fl4->fl4_gre_key = gre_key;
  152. return ip_route_output_key(net, fl4);
  153. }
  154. int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr,
  155. u8 tos, struct net_device *dev,
  156. struct in_device *in_dev, u32 *itag);
  157. int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
  158. u8 tos, struct net_device *devin);
  159. int ip_route_input_rcu(struct sk_buff *skb, __be32 dst, __be32 src,
  160. u8 tos, struct net_device *devin,
  161. struct fib_result *res);
  162. static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
  163. u8 tos, struct net_device *devin)
  164. {
  165. int err;
  166. rcu_read_lock();
  167. err = ip_route_input_noref(skb, dst, src, tos, devin);
  168. if (!err) {
  169. skb_dst_force(skb);
  170. if (!skb_dst(skb))
  171. err = -EINVAL;
  172. }
  173. rcu_read_unlock();
  174. return err;
  175. }
  176. void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
  177. u32 mark, u8 protocol, int flow_flags);
  178. void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
  179. void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
  180. u8 protocol, int flow_flags);
  181. void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
  182. void ip_rt_send_redirect(struct sk_buff *skb);
  183. unsigned int inet_addr_type(struct net *net, __be32 addr);
  184. unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
  185. unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
  186. __be32 addr);
  187. unsigned int inet_addr_type_dev_table(struct net *net,
  188. const struct net_device *dev,
  189. __be32 addr);
  190. void ip_rt_multicast_event(struct in_device *);
  191. int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
  192. void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
  193. struct rtable *rt_dst_alloc(struct net_device *dev,
  194. unsigned int flags, u16 type,
  195. bool nopolicy, bool noxfrm, bool will_cache);
  196. struct in_ifaddr;
  197. void fib_add_ifaddr(struct in_ifaddr *);
  198. void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
  199. void rt_add_uncached_list(struct rtable *rt);
  200. void rt_del_uncached_list(struct rtable *rt);
  201. static inline void __attribute__((no_sanitize("null"))) ip_rt_put(struct rtable *rt)
  202. {
  203. /* dst_release() accepts a NULL parameter.
  204. * We rely on dst being first structure in struct rtable
  205. */
  206. BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
  207. dst_release(&rt->dst);
  208. }
  209. #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
  210. extern const __u8 ip_tos2prio[16];
  211. static inline char rt_tos2priority(u8 tos)
  212. {
  213. return ip_tos2prio[IPTOS_TOS(tos)>>1];
  214. }
  215. /* ip_route_connect() and ip_route_newports() work in tandem whilst
  216. * binding a socket for a new outgoing connection.
  217. *
  218. * In order to use IPSEC properly, we must, in the end, have a
  219. * route that was looked up using all available keys including source
  220. * and destination ports.
  221. *
  222. * However, if a source port needs to be allocated (the user specified
  223. * a wildcard source port) we need to obtain addressing information
  224. * in order to perform that allocation.
  225. *
  226. * So ip_route_connect() looks up a route using wildcarded source and
  227. * destination ports in the key, simply so that we can get a pair of
  228. * addresses to use for port allocation.
  229. *
  230. * Later, once the ports are allocated, ip_route_newports() will make
  231. * another route lookup if needed to make sure we catch any IPSEC
  232. * rules keyed on the port information.
  233. *
  234. * The callers allocate the flow key on their stack, and must pass in
  235. * the same flowi4 object to both the ip_route_connect() and the
  236. * ip_route_newports() calls.
  237. */
  238. static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
  239. u32 tos, int oif, u8 protocol,
  240. __be16 sport, __be16 dport,
  241. struct sock *sk)
  242. {
  243. __u8 flow_flags = 0;
  244. if (inet_sk(sk)->transparent)
  245. flow_flags |= FLOWI_FLAG_ANYSRC;
  246. flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
  247. protocol, flow_flags, dst, src, dport, sport,
  248. sk->sk_uid);
  249. }
  250. static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
  251. __be32 dst, __be32 src, u32 tos,
  252. int oif, u8 protocol,
  253. __be16 sport, __be16 dport,
  254. struct sock *sk)
  255. {
  256. struct net *net = sock_net(sk);
  257. struct rtable *rt;
  258. ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
  259. sport, dport, sk);
  260. if (!dst || !src) {
  261. rt = __ip_route_output_key(net, fl4);
  262. if (IS_ERR(rt))
  263. return rt;
  264. ip_rt_put(rt);
  265. flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
  266. }
  267. security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
  268. return ip_route_output_flow(net, fl4, sk);
  269. }
  270. static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
  271. __be16 orig_sport, __be16 orig_dport,
  272. __be16 sport, __be16 dport,
  273. struct sock *sk)
  274. {
  275. if (sport != orig_sport || dport != orig_dport) {
  276. fl4->fl4_dport = dport;
  277. fl4->fl4_sport = sport;
  278. ip_rt_put(rt);
  279. flowi4_update_output(fl4, sk->sk_bound_dev_if,
  280. RT_CONN_FLAGS(sk), fl4->daddr,
  281. fl4->saddr);
  282. security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
  283. return ip_route_output_flow(sock_net(sk), fl4, sk);
  284. }
  285. return rt;
  286. }
  287. static inline int inet_iif(const struct sk_buff *skb)
  288. {
  289. struct rtable *rt = skb_rtable(skb);
  290. if (rt && rt->rt_iif)
  291. return rt->rt_iif;
  292. return skb->skb_iif;
  293. }
  294. static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
  295. {
  296. int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
  297. struct net *net = dev_net(dst->dev);
  298. if (hoplimit == 0)
  299. hoplimit = net->ipv4.sysctl_ip_default_ttl;
  300. return hoplimit;
  301. }
  302. #endif /* _ROUTE_H */