route.h 9.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325
  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 <linux/in_route.h>
  31. #include <linux/rtnetlink.h>
  32. #include <linux/route.h>
  33. #include <linux/ip.h>
  34. #include <linux/cache.h>
  35. #include <linux/security.h>
  36. #define RTO_ONLINK 0x01
  37. #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
  38. struct fib_nh;
  39. struct inet_peer;
  40. struct fib_info;
  41. struct rtable {
  42. struct dst_entry dst;
  43. /* Lookup key. */
  44. __be32 rt_key_dst;
  45. __be32 rt_key_src;
  46. int rt_genid;
  47. unsigned rt_flags;
  48. __u16 rt_type;
  49. __u8 rt_key_tos;
  50. __be32 rt_dst; /* Path destination */
  51. __be32 rt_src; /* Path source */
  52. int rt_route_iif;
  53. int rt_iif;
  54. int rt_oif;
  55. __u32 rt_mark;
  56. /* Info on neighbour */
  57. __be32 rt_gateway;
  58. /* Miscellaneous cached information */
  59. __be32 rt_spec_dst; /* RFC1122 specific destination */
  60. u32 rt_peer_genid;
  61. struct inet_peer *peer; /* long-living peer info */
  62. struct fib_info *fi; /* for client ref to shared metrics */
  63. };
  64. static inline bool rt_is_input_route(struct rtable *rt)
  65. {
  66. return rt->rt_route_iif != 0;
  67. }
  68. static inline bool rt_is_output_route(struct rtable *rt)
  69. {
  70. return rt->rt_route_iif == 0;
  71. }
  72. struct ip_rt_acct {
  73. __u32 o_bytes;
  74. __u32 o_packets;
  75. __u32 i_bytes;
  76. __u32 i_packets;
  77. };
  78. struct rt_cache_stat {
  79. unsigned int in_hit;
  80. unsigned int in_slow_tot;
  81. unsigned int in_slow_mc;
  82. unsigned int in_no_route;
  83. unsigned int in_brd;
  84. unsigned int in_martian_dst;
  85. unsigned int in_martian_src;
  86. unsigned int out_hit;
  87. unsigned int out_slow_tot;
  88. unsigned int out_slow_mc;
  89. unsigned int gc_total;
  90. unsigned int gc_ignored;
  91. unsigned int gc_goal_miss;
  92. unsigned int gc_dst_overflow;
  93. unsigned int in_hlist_search;
  94. unsigned int out_hlist_search;
  95. };
  96. extern struct ip_rt_acct __percpu *ip_rt_acct;
  97. struct in_device;
  98. extern int ip_rt_init(void);
  99. extern void ip_rt_redirect(__be32 old_gw, __be32 dst, __be32 new_gw,
  100. __be32 src, struct net_device *dev);
  101. extern void rt_cache_flush(struct net *net, int how);
  102. extern void rt_cache_flush_batch(struct net *net);
  103. extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp);
  104. extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
  105. struct sock *sk);
  106. extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig);
  107. static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
  108. {
  109. return ip_route_output_flow(net, flp, NULL);
  110. }
  111. static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
  112. __be32 saddr, u8 tos, int oif)
  113. {
  114. struct flowi4 fl4 = {
  115. .flowi4_oif = oif,
  116. .daddr = daddr,
  117. .saddr = saddr,
  118. .flowi4_tos = tos,
  119. };
  120. return ip_route_output_key(net, &fl4);
  121. }
  122. static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
  123. struct sock *sk,
  124. __be32 daddr, __be32 saddr,
  125. __be16 dport, __be16 sport,
  126. __u8 proto, __u8 tos, int oif)
  127. {
  128. flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
  129. RT_SCOPE_UNIVERSE, proto,
  130. sk ? inet_sk_flowi_flags(sk) : 0,
  131. daddr, saddr, dport, sport);
  132. if (sk)
  133. security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
  134. return ip_route_output_flow(net, fl4, sk);
  135. }
  136. static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
  137. __be32 daddr, __be32 saddr,
  138. __be32 gre_key, __u8 tos, int oif)
  139. {
  140. memset(fl4, 0, sizeof(*fl4));
  141. fl4->flowi4_oif = oif;
  142. fl4->daddr = daddr;
  143. fl4->saddr = saddr;
  144. fl4->flowi4_tos = tos;
  145. fl4->flowi4_proto = IPPROTO_GRE;
  146. fl4->fl4_gre_key = gre_key;
  147. return ip_route_output_key(net, fl4);
  148. }
  149. extern int ip_route_input_common(struct sk_buff *skb, __be32 dst, __be32 src,
  150. u8 tos, struct net_device *devin, bool noref);
  151. static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
  152. u8 tos, struct net_device *devin)
  153. {
  154. return ip_route_input_common(skb, dst, src, tos, devin, false);
  155. }
  156. static inline int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
  157. u8 tos, struct net_device *devin)
  158. {
  159. return ip_route_input_common(skb, dst, src, tos, devin, true);
  160. }
  161. extern unsigned short ip_rt_frag_needed(struct net *net, const struct iphdr *iph,
  162. unsigned short new_mtu, struct net_device *dev);
  163. extern void ip_rt_send_redirect(struct sk_buff *skb);
  164. extern unsigned inet_addr_type(struct net *net, __be32 addr);
  165. extern unsigned inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr);
  166. extern void ip_rt_multicast_event(struct in_device *);
  167. extern int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
  168. extern void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
  169. extern int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb);
  170. struct in_ifaddr;
  171. extern void fib_add_ifaddr(struct in_ifaddr *);
  172. extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
  173. static inline void ip_rt_put(struct rtable * rt)
  174. {
  175. if (rt)
  176. dst_release(&rt->dst);
  177. }
  178. #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
  179. extern const __u8 ip_tos2prio[16];
  180. static inline char rt_tos2priority(u8 tos)
  181. {
  182. return ip_tos2prio[IPTOS_TOS(tos)>>1];
  183. }
  184. /* ip_route_connect() and ip_route_newports() work in tandem whilst
  185. * binding a socket for a new outgoing connection.
  186. *
  187. * In order to use IPSEC properly, we must, in the end, have a
  188. * route that was looked up using all available keys including source
  189. * and destination ports.
  190. *
  191. * However, if a source port needs to be allocated (the user specified
  192. * a wildcard source port) we need to obtain addressing information
  193. * in order to perform that allocation.
  194. *
  195. * So ip_route_connect() looks up a route using wildcarded source and
  196. * destination ports in the key, simply so that we can get a pair of
  197. * addresses to use for port allocation.
  198. *
  199. * Later, once the ports are allocated, ip_route_newports() will make
  200. * another route lookup if needed to make sure we catch any IPSEC
  201. * rules keyed on the port information.
  202. *
  203. * The callers allocate the flow key on their stack, and must pass in
  204. * the same flowi4 object to both the ip_route_connect() and the
  205. * ip_route_newports() calls.
  206. */
  207. static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
  208. u32 tos, int oif, u8 protocol,
  209. __be16 sport, __be16 dport,
  210. struct sock *sk, bool can_sleep)
  211. {
  212. __u8 flow_flags = 0;
  213. if (inet_sk(sk)->transparent)
  214. flow_flags |= FLOWI_FLAG_ANYSRC;
  215. if (protocol == IPPROTO_TCP)
  216. flow_flags |= FLOWI_FLAG_PRECOW_METRICS;
  217. if (can_sleep)
  218. flow_flags |= FLOWI_FLAG_CAN_SLEEP;
  219. flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
  220. protocol, flow_flags, dst, src, dport, sport);
  221. }
  222. static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
  223. __be32 dst, __be32 src, u32 tos,
  224. int oif, u8 protocol,
  225. __be16 sport, __be16 dport,
  226. struct sock *sk, bool can_sleep)
  227. {
  228. struct net *net = sock_net(sk);
  229. struct rtable *rt;
  230. ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
  231. sport, dport, sk, can_sleep);
  232. if (!dst || !src) {
  233. rt = __ip_route_output_key(net, fl4);
  234. if (IS_ERR(rt))
  235. return rt;
  236. ip_rt_put(rt);
  237. flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
  238. }
  239. security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
  240. return ip_route_output_flow(net, fl4, sk);
  241. }
  242. static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
  243. __be16 orig_sport, __be16 orig_dport,
  244. __be16 sport, __be16 dport,
  245. struct sock *sk)
  246. {
  247. if (sport != orig_sport || dport != orig_dport) {
  248. fl4->fl4_dport = dport;
  249. fl4->fl4_sport = sport;
  250. ip_rt_put(rt);
  251. flowi4_update_output(fl4, sk->sk_bound_dev_if,
  252. RT_CONN_FLAGS(sk), fl4->daddr,
  253. fl4->saddr);
  254. security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
  255. return ip_route_output_flow(sock_net(sk), fl4, sk);
  256. }
  257. return rt;
  258. }
  259. extern void rt_bind_peer(struct rtable *rt, __be32 daddr, int create);
  260. static inline struct inet_peer *rt_get_peer(struct rtable *rt, __be32 daddr)
  261. {
  262. if (rt->peer)
  263. return rt->peer;
  264. rt_bind_peer(rt, daddr, 0);
  265. return rt->peer;
  266. }
  267. static inline int inet_iif(const struct sk_buff *skb)
  268. {
  269. return skb_rtable(skb)->rt_iif;
  270. }
  271. extern int sysctl_ip_default_ttl;
  272. static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
  273. {
  274. int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
  275. if (hoplimit == 0)
  276. hoplimit = sysctl_ip_default_ttl;
  277. return hoplimit;
  278. }
  279. #endif /* _ROUTE_H */