ip6_route.h 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362
  1. /* SPDX-License-Identifier: GPL-2.0 */
  2. #ifndef _NET_IP6_ROUTE_H
  3. #define _NET_IP6_ROUTE_H
  4. #include <net/addrconf.h>
  5. #include <net/flow.h>
  6. #include <net/ip6_fib.h>
  7. #include <net/sock.h>
  8. #include <net/lwtunnel.h>
  9. #include <linux/ip.h>
  10. #include <linux/ipv6.h>
  11. #include <linux/route.h>
  12. #include <net/nexthop.h>
  13. struct route_info {
  14. __u8 type;
  15. __u8 length;
  16. __u8 prefix_len;
  17. #if defined(__BIG_ENDIAN_BITFIELD)
  18. __u8 reserved_h:3,
  19. route_pref:2,
  20. reserved_l:3;
  21. #elif defined(__LITTLE_ENDIAN_BITFIELD)
  22. __u8 reserved_l:3,
  23. route_pref:2,
  24. reserved_h:3;
  25. #endif
  26. __be32 lifetime;
  27. __u8 prefix[]; /* 0,8 or 16 */
  28. };
  29. #define RT6_LOOKUP_F_IFACE 0x00000001
  30. #define RT6_LOOKUP_F_REACHABLE 0x00000002
  31. #define RT6_LOOKUP_F_HAS_SADDR 0x00000004
  32. #define RT6_LOOKUP_F_SRCPREF_TMP 0x00000008
  33. #define RT6_LOOKUP_F_SRCPREF_PUBLIC 0x00000010
  34. #define RT6_LOOKUP_F_SRCPREF_COA 0x00000020
  35. #define RT6_LOOKUP_F_IGNORE_LINKSTATE 0x00000040
  36. #define RT6_LOOKUP_F_DST_NOREF 0x00000080
  37. /* We do not (yet ?) support IPv6 jumbograms (RFC 2675)
  38. * Unlike IPv4, hdr->seg_len doesn't include the IPv6 header
  39. */
  40. #define IP6_MAX_MTU (0xFFFF + sizeof(struct ipv6hdr))
  41. /*
  42. * rt6_srcprefs2flags() and rt6_flags2srcprefs() translate
  43. * between IPV6_ADDR_PREFERENCES socket option values
  44. * IPV6_PREFER_SRC_TMP = 0x1
  45. * IPV6_PREFER_SRC_PUBLIC = 0x2
  46. * IPV6_PREFER_SRC_COA = 0x4
  47. * and above RT6_LOOKUP_F_SRCPREF_xxx flags.
  48. */
  49. static inline int rt6_srcprefs2flags(unsigned int srcprefs)
  50. {
  51. return (srcprefs & IPV6_PREFER_SRC_MASK) << 3;
  52. }
  53. static inline unsigned int rt6_flags2srcprefs(int flags)
  54. {
  55. return (flags >> 3) & IPV6_PREFER_SRC_MASK;
  56. }
  57. static inline bool rt6_need_strict(const struct in6_addr *daddr)
  58. {
  59. return ipv6_addr_type(daddr) &
  60. (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK);
  61. }
  62. /* fib entries using a nexthop object can not be coalesced into
  63. * a multipath route
  64. */
  65. static inline bool rt6_qualify_for_ecmp(const struct fib6_info *f6i)
  66. {
  67. /* the RTF_ADDRCONF flag filters out RA's */
  68. return !(f6i->fib6_flags & RTF_ADDRCONF) && !f6i->nh &&
  69. f6i->fib6_nh->fib_nh_gw_family;
  70. }
  71. void ip6_route_input(struct sk_buff *skb);
  72. struct dst_entry *ip6_route_input_lookup(struct net *net,
  73. struct net_device *dev,
  74. struct flowi6 *fl6,
  75. const struct sk_buff *skb, int flags);
  76. struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
  77. struct flowi6 *fl6, int flags);
  78. static inline struct dst_entry *ip6_route_output(struct net *net,
  79. const struct sock *sk,
  80. struct flowi6 *fl6)
  81. {
  82. return ip6_route_output_flags(net, sk, fl6, 0);
  83. }
  84. /* Only conditionally release dst if flags indicates
  85. * !RT6_LOOKUP_F_DST_NOREF or dst is in uncached_list.
  86. */
  87. static inline void ip6_rt_put_flags(struct rt6_info *rt, int flags)
  88. {
  89. if (!(flags & RT6_LOOKUP_F_DST_NOREF) ||
  90. !list_empty(&rt->dst.rt_uncached))
  91. ip6_rt_put(rt);
  92. }
  93. struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
  94. const struct sk_buff *skb, int flags);
  95. struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
  96. int ifindex, struct flowi6 *fl6,
  97. const struct sk_buff *skb, int flags);
  98. void ip6_route_init_special_entries(void);
  99. int ip6_route_init(void);
  100. void ip6_route_cleanup(void);
  101. int ipv6_route_ioctl(struct net *net, unsigned int cmd,
  102. struct in6_rtmsg *rtmsg);
  103. int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags,
  104. struct netlink_ext_ack *extack);
  105. int ip6_ins_rt(struct net *net, struct fib6_info *f6i);
  106. int ip6_del_rt(struct net *net, struct fib6_info *f6i, bool skip_notify);
  107. void rt6_flush_exceptions(struct fib6_info *f6i);
  108. void rt6_age_exceptions(struct fib6_info *f6i, struct fib6_gc_args *gc_args,
  109. unsigned long now);
  110. static inline int ip6_route_get_saddr(struct net *net, struct fib6_info *f6i,
  111. const struct in6_addr *daddr,
  112. unsigned int prefs, int l3mdev_index,
  113. struct in6_addr *saddr)
  114. {
  115. struct net_device *l3mdev;
  116. struct net_device *dev;
  117. bool same_vrf;
  118. int err = 0;
  119. rcu_read_lock();
  120. l3mdev = dev_get_by_index_rcu(net, l3mdev_index);
  121. if (!f6i || !f6i->fib6_prefsrc.plen || l3mdev)
  122. dev = f6i ? fib6_info_nh_dev(f6i) : NULL;
  123. same_vrf = !l3mdev || l3mdev_master_dev_rcu(dev) == l3mdev;
  124. if (f6i && f6i->fib6_prefsrc.plen && same_vrf)
  125. *saddr = f6i->fib6_prefsrc.addr;
  126. else
  127. err = ipv6_dev_get_saddr(net, same_vrf ? dev : l3mdev, daddr, prefs, saddr);
  128. rcu_read_unlock();
  129. return err;
  130. }
  131. struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
  132. const struct in6_addr *saddr, int oif,
  133. const struct sk_buff *skb, int flags);
  134. u32 rt6_multipath_hash(const struct net *net, const struct flowi6 *fl6,
  135. const struct sk_buff *skb, struct flow_keys *hkeys);
  136. struct dst_entry *icmp6_dst_alloc(struct net_device *dev, struct flowi6 *fl6);
  137. void fib6_force_start_gc(struct net *net);
  138. struct fib6_info *addrconf_f6i_alloc(struct net *net, struct inet6_dev *idev,
  139. const struct in6_addr *addr, bool anycast,
  140. gfp_t gfp_flags, struct netlink_ext_ack *extack);
  141. struct rt6_info *ip6_dst_alloc(struct net *net, struct net_device *dev,
  142. int flags);
  143. /*
  144. * support functions for ND
  145. *
  146. */
  147. struct fib6_info *rt6_get_dflt_router(struct net *net,
  148. const struct in6_addr *addr,
  149. struct net_device *dev);
  150. struct fib6_info *rt6_add_dflt_router(struct net *net,
  151. const struct in6_addr *gwaddr,
  152. struct net_device *dev, unsigned int pref,
  153. u32 defrtr_usr_metric,
  154. int lifetime);
  155. void rt6_purge_dflt_routers(struct net *net);
  156. int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
  157. const struct in6_addr *gwaddr);
  158. void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, int oif,
  159. u32 mark, kuid_t uid);
  160. void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu);
  161. void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
  162. kuid_t uid);
  163. void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif);
  164. void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk);
  165. struct netlink_callback;
  166. struct rt6_rtnl_dump_arg {
  167. struct sk_buff *skb;
  168. struct netlink_callback *cb;
  169. struct net *net;
  170. struct fib_dump_filter filter;
  171. };
  172. int rt6_dump_route(struct fib6_info *f6i, void *p_arg, unsigned int skip);
  173. void rt6_mtu_change(struct net_device *dev, unsigned int mtu);
  174. void rt6_remove_prefsrc(struct inet6_ifaddr *ifp);
  175. void rt6_clean_tohost(struct net *net, struct in6_addr *gateway);
  176. void rt6_sync_up(struct net_device *dev, unsigned char nh_flags);
  177. void rt6_disable_ip(struct net_device *dev, unsigned long event);
  178. void rt6_sync_down_dev(struct net_device *dev, unsigned long event);
  179. void rt6_multipath_rebalance(struct fib6_info *f6i);
  180. void rt6_uncached_list_add(struct rt6_info *rt);
  181. void rt6_uncached_list_del(struct rt6_info *rt);
  182. static inline const struct rt6_info *skb_rt6_info(const struct sk_buff *skb)
  183. {
  184. const struct dst_entry *dst = skb_dst(skb);
  185. if (dst)
  186. return dst_rt6_info(dst);
  187. return NULL;
  188. }
  189. /*
  190. * Store a destination cache entry in a socket
  191. */
  192. static inline void ip6_dst_store(struct sock *sk, struct dst_entry *dst,
  193. bool daddr_set,
  194. bool saddr_set)
  195. {
  196. struct ipv6_pinfo *np = inet6_sk(sk);
  197. np->dst_cookie = rt6_get_cookie(dst_rt6_info(dst));
  198. sk_setup_caps(sk, dst);
  199. np->daddr_cache = daddr_set;
  200. #ifdef CONFIG_IPV6_SUBTREES
  201. np->saddr_cache = saddr_set;
  202. #endif
  203. }
  204. void ip6_sk_dst_store_flow(struct sock *sk, struct dst_entry *dst,
  205. const struct flowi6 *fl6);
  206. static inline bool ipv6_unicast_destination(const struct sk_buff *skb)
  207. {
  208. const struct rt6_info *rt = dst_rt6_info(skb_dst(skb));
  209. return rt->rt6i_flags & RTF_LOCAL;
  210. }
  211. static inline bool ipv6_anycast_destination(const struct dst_entry *dst,
  212. const struct in6_addr *daddr)
  213. {
  214. const struct rt6_info *rt = dst_rt6_info(dst);
  215. return rt->rt6i_flags & RTF_ANYCAST ||
  216. (rt->rt6i_dst.plen < 127 &&
  217. !(rt->rt6i_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) &&
  218. ipv6_addr_equal(&rt->rt6i_dst.addr, daddr));
  219. }
  220. int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
  221. int (*output)(struct net *, struct sock *, struct sk_buff *));
  222. /* Variant of dst_mtu() for IPv6 users */
  223. static inline u32 dst6_mtu(const struct dst_entry *dst)
  224. {
  225. return INDIRECT_CALL_1(dst->ops->mtu, ip6_mtu, dst);
  226. }
  227. static inline unsigned int ip6_skb_dst_mtu(const struct sk_buff *skb)
  228. {
  229. const struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ?
  230. inet6_sk(skb->sk) : NULL;
  231. const struct dst_entry *dst = skb_dst(skb);
  232. unsigned int mtu;
  233. if (np && READ_ONCE(np->pmtudisc) >= IPV6_PMTUDISC_PROBE) {
  234. mtu = READ_ONCE(dst_dev(dst)->mtu);
  235. mtu -= lwtunnel_headroom(dst->lwtstate, mtu);
  236. } else {
  237. mtu = dst_mtu(dst);
  238. }
  239. return mtu;
  240. }
  241. static inline bool ip6_sk_accept_pmtu(const struct sock *sk)
  242. {
  243. u8 pmtudisc = READ_ONCE(inet6_sk(sk)->pmtudisc);
  244. return pmtudisc != IPV6_PMTUDISC_INTERFACE &&
  245. pmtudisc != IPV6_PMTUDISC_OMIT;
  246. }
  247. static inline bool ip6_sk_ignore_df(const struct sock *sk)
  248. {
  249. u8 pmtudisc = READ_ONCE(inet6_sk(sk)->pmtudisc);
  250. return pmtudisc < IPV6_PMTUDISC_DO ||
  251. pmtudisc == IPV6_PMTUDISC_OMIT;
  252. }
  253. static inline const struct in6_addr *rt6_nexthop(const struct rt6_info *rt,
  254. const struct in6_addr *daddr)
  255. {
  256. if (rt->rt6i_flags & RTF_GATEWAY)
  257. return &rt->rt6i_gateway;
  258. else if (unlikely(rt->rt6i_flags & RTF_CACHE))
  259. return &rt->rt6i_dst.addr;
  260. else
  261. return daddr;
  262. }
  263. static inline bool rt6_duplicate_nexthop(struct fib6_info *a, struct fib6_info *b)
  264. {
  265. struct fib6_nh *nha, *nhb;
  266. if (a->nh || b->nh)
  267. return nexthop_cmp(a->nh, b->nh);
  268. nha = a->fib6_nh;
  269. nhb = b->fib6_nh;
  270. return nha->fib_nh_dev == nhb->fib_nh_dev &&
  271. ipv6_addr_equal(&nha->fib_nh_gw6, &nhb->fib_nh_gw6) &&
  272. !lwtunnel_cmp_encap(nha->fib_nh_lws, nhb->fib_nh_lws);
  273. }
  274. static inline unsigned int ip6_dst_mtu_maybe_forward(const struct dst_entry *dst,
  275. bool forwarding)
  276. {
  277. struct inet6_dev *idev;
  278. unsigned int mtu;
  279. if (!forwarding || dst_metric_locked(dst, RTAX_MTU)) {
  280. mtu = dst_metric_raw(dst, RTAX_MTU);
  281. if (mtu)
  282. goto out;
  283. }
  284. mtu = IPV6_MIN_MTU;
  285. rcu_read_lock();
  286. idev = __in6_dev_get(dst_dev_rcu(dst));
  287. if (idev)
  288. mtu = READ_ONCE(idev->cnf.mtu6);
  289. rcu_read_unlock();
  290. out:
  291. return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
  292. }
  293. u32 ip6_mtu_from_fib6(const struct fib6_result *res,
  294. const struct in6_addr *daddr,
  295. const struct in6_addr *saddr);
  296. struct neighbour *ip6_neigh_lookup(const struct in6_addr *gw,
  297. struct net_device *dev, struct sk_buff *skb,
  298. const void *daddr);
  299. #endif