route.h 9.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344
  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 <net/l3mdev.h>
  32. #include <linux/in_route.h>
  33. #include <linux/rtnetlink.h>
  34. #include <linux/rcupdate.h>
  35. #include <linux/route.h>
  36. #include <linux/ip.h>
  37. #include <linux/cache.h>
  38. #include <linux/security.h>
  39. /* IPv4 datagram length is stored into 16bit field (tot_len) */
  40. #define IP_MAX_MTU 0xFFFFU
  41. #define RTO_ONLINK 0x01
  42. #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
  43. #define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
  44. struct fib_nh;
  45. struct fib_info;
  46. struct uncached_list;
  47. struct rtable {
  48. struct dst_entry dst;
  49. int rt_genid;
  50. unsigned int rt_flags;
  51. __u16 rt_type;
  52. __u8 rt_is_input;
  53. __u8 rt_uses_gateway;
  54. int rt_iif;
  55. /* Info on neighbour */
  56. __be32 rt_gateway;
  57. /* Miscellaneous cached information */
  58. u32 rt_mtu_locked:1,
  59. rt_pmtu:31;
  60. u32 rt_table_id;
  61. struct list_head rt_uncached;
  62. struct uncached_list *rt_uncached_list;
  63. };
  64. static inline bool rt_is_input_route(const struct rtable *rt)
  65. {
  66. return rt->rt_is_input != 0;
  67. }
  68. static inline bool rt_is_output_route(const struct rtable *rt)
  69. {
  70. return rt->rt_is_input == 0;
  71. }
  72. static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
  73. {
  74. if (rt->rt_gateway)
  75. return rt->rt_gateway;
  76. return daddr;
  77. }
  78. struct ip_rt_acct {
  79. __u32 o_bytes;
  80. __u32 o_packets;
  81. __u32 i_bytes;
  82. __u32 i_packets;
  83. };
  84. struct rt_cache_stat {
  85. unsigned int in_slow_tot;
  86. unsigned int in_slow_mc;
  87. unsigned int in_no_route;
  88. unsigned int in_brd;
  89. unsigned int in_martian_dst;
  90. unsigned int in_martian_src;
  91. unsigned int out_slow_tot;
  92. unsigned int out_slow_mc;
  93. };
  94. extern struct ip_rt_acct __percpu *ip_rt_acct;
  95. struct in_device;
  96. int ip_rt_init(void);
  97. void rt_cache_flush(struct net *net);
  98. void rt_flush_dev(struct net_device *dev);
  99. struct rtable *__ip_route_output_key_hash(struct net *, struct flowi4 *flp,
  100. int mp_hash);
  101. static inline struct rtable *__ip_route_output_key(struct net *net,
  102. struct flowi4 *flp)
  103. {
  104. return __ip_route_output_key_hash(net, flp, -1);
  105. }
  106. struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
  107. const struct sock *sk);
  108. struct dst_entry *ipv4_blackhole_route(struct net *net,
  109. struct dst_entry *dst_orig);
  110. static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
  111. {
  112. return ip_route_output_flow(net, flp, NULL);
  113. }
  114. static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
  115. __be32 saddr, u8 tos, int oif)
  116. {
  117. struct flowi4 fl4 = {
  118. .flowi4_oif = oif,
  119. .flowi4_tos = tos,
  120. .daddr = daddr,
  121. .saddr = saddr,
  122. };
  123. return ip_route_output_key(net, &fl4);
  124. }
  125. static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
  126. struct sock *sk,
  127. __be32 daddr, __be32 saddr,
  128. __be16 dport, __be16 sport,
  129. __u8 proto, __u8 tos, int oif)
  130. {
  131. flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
  132. RT_SCOPE_UNIVERSE, proto,
  133. sk ? inet_sk_flowi_flags(sk) : 0,
  134. daddr, saddr, dport, sport);
  135. if (sk)
  136. security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
  137. return ip_route_output_flow(net, fl4, sk);
  138. }
  139. static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
  140. __be32 daddr, __be32 saddr,
  141. __be32 gre_key, __u8 tos, int oif)
  142. {
  143. memset(fl4, 0, sizeof(*fl4));
  144. fl4->flowi4_oif = oif;
  145. fl4->daddr = daddr;
  146. fl4->saddr = saddr;
  147. fl4->flowi4_tos = tos;
  148. fl4->flowi4_proto = IPPROTO_GRE;
  149. fl4->fl4_gre_key = gre_key;
  150. return ip_route_output_key(net, fl4);
  151. }
  152. int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
  153. u8 tos, struct net_device *devin);
  154. static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
  155. u8 tos, struct net_device *devin)
  156. {
  157. int err;
  158. rcu_read_lock();
  159. err = ip_route_input_noref(skb, dst, src, tos, devin);
  160. if (!err)
  161. skb_dst_force(skb);
  162. rcu_read_unlock();
  163. return err;
  164. }
  165. void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
  166. u32 mark, u8 protocol, int flow_flags);
  167. void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
  168. void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
  169. u8 protocol, int flow_flags);
  170. void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
  171. void ip_rt_send_redirect(struct sk_buff *skb);
  172. unsigned int inet_addr_type(struct net *net, __be32 addr);
  173. unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
  174. unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
  175. __be32 addr);
  176. unsigned int inet_addr_type_dev_table(struct net *net,
  177. const struct net_device *dev,
  178. __be32 addr);
  179. void ip_rt_multicast_event(struct in_device *);
  180. int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
  181. void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
  182. struct in_ifaddr;
  183. void fib_add_ifaddr(struct in_ifaddr *);
  184. void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
  185. static inline void ip_rt_put(struct rtable *rt)
  186. {
  187. /* dst_release() accepts a NULL parameter.
  188. * We rely on dst being first structure in struct rtable
  189. */
  190. BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
  191. dst_release(&rt->dst);
  192. }
  193. #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
  194. extern const __u8 ip_tos2prio[16];
  195. static inline char rt_tos2priority(u8 tos)
  196. {
  197. return ip_tos2prio[IPTOS_TOS(tos)>>1];
  198. }
  199. /* ip_route_connect() and ip_route_newports() work in tandem whilst
  200. * binding a socket for a new outgoing connection.
  201. *
  202. * In order to use IPSEC properly, we must, in the end, have a
  203. * route that was looked up using all available keys including source
  204. * and destination ports.
  205. *
  206. * However, if a source port needs to be allocated (the user specified
  207. * a wildcard source port) we need to obtain addressing information
  208. * in order to perform that allocation.
  209. *
  210. * So ip_route_connect() looks up a route using wildcarded source and
  211. * destination ports in the key, simply so that we can get a pair of
  212. * addresses to use for port allocation.
  213. *
  214. * Later, once the ports are allocated, ip_route_newports() will make
  215. * another route lookup if needed to make sure we catch any IPSEC
  216. * rules keyed on the port information.
  217. *
  218. * The callers allocate the flow key on their stack, and must pass in
  219. * the same flowi4 object to both the ip_route_connect() and the
  220. * ip_route_newports() calls.
  221. */
  222. static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
  223. u32 tos, int oif, u8 protocol,
  224. __be16 sport, __be16 dport,
  225. struct sock *sk)
  226. {
  227. __u8 flow_flags = 0;
  228. if (inet_sk(sk)->transparent)
  229. flow_flags |= FLOWI_FLAG_ANYSRC;
  230. flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
  231. protocol, flow_flags, dst, src, dport, sport);
  232. }
  233. static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
  234. __be32 dst, __be32 src, u32 tos,
  235. int oif, u8 protocol,
  236. __be16 sport, __be16 dport,
  237. struct sock *sk)
  238. {
  239. struct net *net = sock_net(sk);
  240. struct rtable *rt;
  241. ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
  242. sport, dport, sk);
  243. if (!src && oif) {
  244. int rc;
  245. rc = l3mdev_get_saddr(net, oif, fl4);
  246. if (rc < 0)
  247. return ERR_PTR(rc);
  248. src = fl4->saddr;
  249. }
  250. if (!dst || !src) {
  251. rt = __ip_route_output_key(net, fl4);
  252. if (IS_ERR(rt))
  253. return rt;
  254. ip_rt_put(rt);
  255. flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
  256. }
  257. security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
  258. return ip_route_output_flow(net, fl4, sk);
  259. }
  260. static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
  261. __be16 orig_sport, __be16 orig_dport,
  262. __be16 sport, __be16 dport,
  263. struct sock *sk)
  264. {
  265. if (sport != orig_sport || dport != orig_dport) {
  266. fl4->fl4_dport = dport;
  267. fl4->fl4_sport = sport;
  268. ip_rt_put(rt);
  269. flowi4_update_output(fl4, sk->sk_bound_dev_if,
  270. RT_CONN_FLAGS(sk), fl4->daddr,
  271. fl4->saddr);
  272. security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
  273. return ip_route_output_flow(sock_net(sk), fl4, sk);
  274. }
  275. return rt;
  276. }
  277. static inline int inet_iif(const struct sk_buff *skb)
  278. {
  279. int iif = skb_rtable(skb)->rt_iif;
  280. if (iif)
  281. return iif;
  282. return skb->skb_iif;
  283. }
  284. extern int sysctl_ip_default_ttl;
  285. static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
  286. {
  287. int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
  288. if (hoplimit == 0)
  289. hoplimit = sysctl_ip_default_ttl;
  290. return hoplimit;
  291. }
  292. #endif /* _ROUTE_H */