1 /* SPDX-License-Identifier: GPL-2.0 */ 2 #ifndef _NET_IP6_ROUTE_H 3 #define _NET_IP6_ROUTE_H 4 5 #include <net/addrconf.h> 6 #include <net/flow.h> 7 #include <net/ip6_fib.h> 8 #include <net/sock.h> 9 #include <net/lwtunnel.h> 10 #include <linux/ip.h> 11 #include <linux/ipv6.h> 12 #include <linux/route.h> 13 #include <net/nexthop.h> 14 15 struct route_info { 16 __u8 type; 17 __u8 length; 18 __u8 prefix_len; 19 #if defined(__BIG_ENDIAN_BITFIELD) 20 __u8 reserved_h:3, 21 route_pref:2, 22 reserved_l:3; 23 #elif defined(__LITTLE_ENDIAN_BITFIELD) 24 __u8 reserved_l:3, 25 route_pref:2, 26 reserved_h:3; 27 #endif 28 __be32 lifetime; 29 __u8 prefix[]; /* 0,8 or 16 */ 30 }; 31 32 #define RT6_LOOKUP_F_IFACE 0x00000001 33 #define RT6_LOOKUP_F_REACHABLE 0x00000002 34 #define RT6_LOOKUP_F_HAS_SADDR 0x00000004 35 #define RT6_LOOKUP_F_SRCPREF_TMP 0x00000008 36 #define RT6_LOOKUP_F_SRCPREF_PUBLIC 0x00000010 37 #define RT6_LOOKUP_F_SRCPREF_COA 0x00000020 38 #define RT6_LOOKUP_F_IGNORE_LINKSTATE 0x00000040 39 #define RT6_LOOKUP_F_DST_NOREF 0x00000080 40 41 /* We do not (yet ?) support IPv6 jumbograms (RFC 2675) 42 * Unlike IPv4, hdr->seg_len doesn't include the IPv6 header 43 */ 44 #define IP6_MAX_MTU (0xFFFF + sizeof(struct ipv6hdr)) 45 46 /* 47 * rt6_srcprefs2flags() and rt6_flags2srcprefs() translate 48 * between IPV6_ADDR_PREFERENCES socket option values 49 * IPV6_PREFER_SRC_TMP = 0x1 50 * IPV6_PREFER_SRC_PUBLIC = 0x2 51 * IPV6_PREFER_SRC_COA = 0x4 52 * and above RT6_LOOKUP_F_SRCPREF_xxx flags. 53 */ 54 static inline int rt6_srcprefs2flags(unsigned int srcprefs) 55 { 56 return (srcprefs & IPV6_PREFER_SRC_MASK) << 3; 57 } 58 59 static inline unsigned int rt6_flags2srcprefs(int flags) 60 { 61 return (flags >> 3) & IPV6_PREFER_SRC_MASK; 62 } 63 64 static inline bool rt6_need_strict(const struct in6_addr *daddr) 65 { 66 return ipv6_addr_type(daddr) & 67 (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK); 68 } 69 70 /* fib entries using a nexthop object can not be coalesced into 71 * a multipath route 72 */ 73 static inline bool rt6_qualify_for_ecmp(const struct fib6_info *f6i) 74 { 75 /* the RTF_ADDRCONF flag filters out RA's */ 76 return !(f6i->fib6_flags & RTF_ADDRCONF) && !f6i->nh && 77 f6i->fib6_nh->fib_nh_gw_family; 78 } 79 80 #if IS_ENABLED(CONFIG_IPV6) 81 void ip6_route_input(struct sk_buff *skb); 82 #else 83 static inline void ip6_route_input(struct sk_buff *skb) 84 { 85 } 86 #endif 87 88 struct dst_entry *ip6_route_input_lookup(struct net *net, 89 struct net_device *dev, 90 struct flowi6 *fl6, 91 const struct sk_buff *skb, int flags); 92 93 struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk, 94 struct flowi6 *fl6, int flags); 95 96 static inline struct dst_entry *ip6_route_output(struct net *net, 97 const struct sock *sk, 98 struct flowi6 *fl6) 99 { 100 return ip6_route_output_flags(net, sk, fl6, 0); 101 } 102 103 /* Only conditionally release dst if flags indicates 104 * !RT6_LOOKUP_F_DST_NOREF or dst is uncached. 105 */ 106 static inline void ip6_rt_put_flags(struct rt6_info *rt, int flags) 107 { 108 if (!(flags & RT6_LOOKUP_F_DST_NOREF) || 109 rt->dst.rt_uncached_list) 110 ip6_rt_put(rt); 111 } 112 113 struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6, 114 const struct sk_buff *skb, int flags); 115 struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, 116 int ifindex, struct flowi6 *fl6, 117 const struct sk_buff *skb, int flags); 118 119 void ip6_route_init_special_entries(void); 120 int ip6_route_init(void); 121 void ip6_route_cleanup(void); 122 123 int ipv6_route_ioctl(struct net *net, unsigned int cmd, 124 struct in6_rtmsg *rtmsg); 125 126 int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags, 127 struct netlink_ext_ack *extack); 128 int ip6_ins_rt(struct net *net, struct fib6_info *f6i); 129 int ip6_del_rt_reason(struct net *net, struct fib6_info *f6i, 130 enum rt_del_reason del_reason); 131 #if IS_ENABLED(CONFIG_IPV6) 132 int ip6_del_rt(struct net *net, struct fib6_info *f6i, bool skip_notify); 133 #else 134 static inline int ip6_del_rt(struct net *net, struct fib6_info *f6i, 135 bool skip_notify) 136 { 137 return -EAFNOSUPPORT; 138 } 139 #endif 140 141 void rt6_flush_exceptions(struct fib6_info *f6i); 142 void rt6_age_exceptions(struct fib6_info *f6i, struct fib6_gc_args *gc_args, 143 unsigned long now); 144 145 static inline int ip6_route_get_saddr(struct net *net, struct fib6_info *f6i, 146 const struct in6_addr *daddr, 147 unsigned int prefs, int l3mdev_index, 148 struct in6_addr *saddr) 149 { 150 struct net_device *l3mdev; 151 struct net_device *dev; 152 bool same_vrf; 153 int err = 0; 154 155 rcu_read_lock(); 156 157 l3mdev = dev_get_by_index_rcu(net, l3mdev_index); 158 if (!f6i || !f6i->fib6_prefsrc.plen || l3mdev) 159 dev = f6i ? fib6_info_nh_dev(f6i) : NULL; 160 same_vrf = !l3mdev || l3mdev_master_dev_rcu(dev) == l3mdev; 161 if (f6i && f6i->fib6_prefsrc.plen && same_vrf) 162 *saddr = f6i->fib6_prefsrc.addr; 163 else 164 err = ipv6_dev_get_saddr(net, same_vrf ? dev : l3mdev, daddr, prefs, saddr); 165 166 rcu_read_unlock(); 167 168 return err; 169 } 170 171 struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr, 172 const struct in6_addr *saddr, int oif, 173 const struct sk_buff *skb, int flags); 174 u32 rt6_multipath_hash(const struct net *net, const struct flowi6 *fl6, 175 const struct sk_buff *skb, struct flow_keys *hkeys); 176 177 struct dst_entry *icmp6_dst_alloc(struct net_device *dev, struct flowi6 *fl6); 178 179 void fib6_force_start_gc(struct net *net); 180 181 struct fib6_info *addrconf_f6i_alloc(struct net *net, struct inet6_dev *idev, 182 const struct in6_addr *addr, bool anycast, 183 gfp_t gfp_flags, struct netlink_ext_ack *extack); 184 185 struct rt6_info *ip6_dst_alloc(struct net *net, struct net_device *dev, 186 int flags); 187 188 /* 189 * support functions for ND 190 * 191 */ 192 struct fib6_info *rt6_get_dflt_router(struct net *net, 193 const struct in6_addr *addr, 194 struct net_device *dev); 195 struct fib6_info *rt6_add_dflt_router(struct net *net, 196 const struct in6_addr *gwaddr, 197 struct net_device *dev, unsigned int pref, 198 u32 defrtr_usr_metric, 199 int lifetime); 200 201 void rt6_purge_dflt_routers(struct net *net); 202 203 int rt6_route_rcv(struct net_device *dev, u8 *opt, int len, 204 const struct in6_addr *gwaddr); 205 206 void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, int oif, 207 u32 mark, kuid_t uid); 208 void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu); 209 void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark, 210 kuid_t uid); 211 void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif); 212 void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk); 213 214 struct netlink_callback; 215 216 struct rt6_rtnl_dump_arg { 217 struct sk_buff *skb; 218 struct netlink_callback *cb; 219 struct net *net; 220 struct fib_dump_filter filter; 221 }; 222 223 int rt6_dump_route(struct fib6_info *f6i, void *p_arg, unsigned int skip); 224 void rt6_mtu_change(struct net_device *dev, unsigned int mtu); 225 void rt6_remove_prefsrc(struct inet6_ifaddr *ifp); 226 void rt6_clean_tohost(struct net *net, struct in6_addr *gateway); 227 void rt6_sync_up(struct net_device *dev, unsigned char nh_flags); 228 void rt6_disable_ip(struct net_device *dev, unsigned long event); 229 void rt6_sync_down_dev(struct net_device *dev, unsigned long event); 230 void rt6_multipath_rebalance(struct fib6_info *f6i); 231 232 void rt6_uncached_list_add(struct rt6_info *rt); 233 void rt6_uncached_list_del(struct rt6_info *rt); 234 #ifdef CONFIG_IPV6 235 void rt6_uncached_list_flush_dev(struct net_device *dev); 236 #else 237 static inline void rt6_uncached_list_flush_dev(struct net_device *dev) 238 { 239 } 240 #endif 241 242 static inline const struct rt6_info *skb_rt6_info(const struct sk_buff *skb) 243 { 244 const struct dst_entry *dst = skb_dst(skb); 245 246 if (dst) 247 return dst_rt6_info(dst); 248 249 return NULL; 250 } 251 252 /* 253 * Store a destination cache entry in a socket 254 */ 255 static inline void ip6_dst_store(struct sock *sk, struct dst_entry *dst, 256 bool daddr_set, 257 bool saddr_set) 258 { 259 struct ipv6_pinfo *np = inet6_sk(sk); 260 261 np->dst_cookie = rt6_get_cookie(dst_rt6_info(dst)); 262 sk_setup_caps(sk, dst); 263 np->daddr_cache = daddr_set; 264 #ifdef CONFIG_IPV6_SUBTREES 265 np->saddr_cache = saddr_set; 266 #endif 267 } 268 269 void ip6_sk_dst_store_flow(struct sock *sk, struct dst_entry *dst, 270 const struct flowi6 *fl6); 271 272 static inline bool ipv6_unicast_destination(const struct sk_buff *skb) 273 { 274 const struct rt6_info *rt = dst_rt6_info(skb_dst(skb)); 275 276 return rt->rt6i_flags & RTF_LOCAL; 277 } 278 279 static inline bool __ipv6_anycast_destination(const struct rt6key *rt6i_dst, 280 u32 rt6i_flags, 281 const struct in6_addr *daddr) 282 { 283 return rt6i_flags & RTF_ANYCAST || 284 (rt6i_dst->plen < 127 && 285 !(rt6i_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) && 286 ipv6_addr_equal(&rt6i_dst->addr, daddr)); 287 } 288 289 static inline bool ipv6_anycast_destination(const struct dst_entry *dst, 290 const struct in6_addr *daddr) 291 { 292 const struct rt6_info *rt = dst_rt6_info(dst); 293 294 return __ipv6_anycast_destination(&rt->rt6i_dst, rt->rt6i_flags, daddr); 295 } 296 297 #if IS_ENABLED(CONFIG_IPV6) 298 int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, 299 int (*output)(struct net *, struct sock *, struct sk_buff *)); 300 #else 301 static inline int ip6_fragment(struct net *net, struct sock *sk, 302 struct sk_buff *skb, 303 int (*output)(struct net *, struct sock *, 304 struct sk_buff *)) 305 { 306 kfree_skb(skb); 307 return -EAFNOSUPPORT; 308 } 309 #endif 310 311 /* Variant of dst_mtu() for IPv6 users */ 312 static inline u32 dst6_mtu(const struct dst_entry *dst) 313 { 314 return INDIRECT_CALL_1(dst->ops->mtu, ip6_mtu, dst); 315 } 316 317 static inline unsigned int ip6_skb_dst_mtu(const struct sk_buff *skb) 318 { 319 const struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ? 320 inet6_sk(skb->sk) : NULL; 321 const struct dst_entry *dst = skb_dst(skb); 322 unsigned int mtu; 323 324 if (np && READ_ONCE(np->pmtudisc) >= IPV6_PMTUDISC_PROBE) { 325 mtu = READ_ONCE(dst_dev(dst)->mtu); 326 mtu -= lwtunnel_headroom(dst->lwtstate, mtu); 327 } else { 328 mtu = dst_mtu(dst); 329 } 330 return mtu; 331 } 332 333 static inline bool ip6_sk_accept_pmtu(const struct sock *sk) 334 { 335 u8 pmtudisc = READ_ONCE(inet6_sk(sk)->pmtudisc); 336 337 return pmtudisc != IPV6_PMTUDISC_INTERFACE && 338 pmtudisc != IPV6_PMTUDISC_OMIT; 339 } 340 341 static inline bool ip6_sk_ignore_df(const struct sock *sk) 342 { 343 u8 pmtudisc = READ_ONCE(inet6_sk(sk)->pmtudisc); 344 345 return pmtudisc < IPV6_PMTUDISC_DO || 346 pmtudisc == IPV6_PMTUDISC_OMIT; 347 } 348 349 static inline const struct in6_addr *rt6_nexthop(const struct rt6_info *rt, 350 const struct in6_addr *daddr) 351 { 352 if (rt->rt6i_flags & RTF_GATEWAY) 353 return &rt->rt6i_gateway; 354 else if (unlikely(rt->rt6i_flags & RTF_CACHE)) 355 return &rt->rt6i_dst.addr; 356 else 357 return daddr; 358 } 359 360 static inline bool rt6_duplicate_nexthop(struct fib6_info *a, struct fib6_info *b) 361 { 362 struct fib6_nh *nha, *nhb; 363 364 if (a->nh || b->nh) 365 return nexthop_cmp(a->nh, b->nh); 366 367 nha = a->fib6_nh; 368 nhb = b->fib6_nh; 369 return nha->fib_nh_dev == nhb->fib_nh_dev && 370 ipv6_addr_equal(&nha->fib_nh_gw6, &nhb->fib_nh_gw6) && 371 !lwtunnel_cmp_encap(nha->fib_nh_lws, nhb->fib_nh_lws); 372 } 373 374 static inline unsigned int ip6_dst_mtu_maybe_forward(const struct dst_entry *dst, 375 bool forwarding) 376 { 377 struct inet6_dev *idev; 378 unsigned int mtu; 379 380 if (!forwarding || dst_metric_locked(dst, RTAX_MTU)) { 381 mtu = dst_metric_raw(dst, RTAX_MTU); 382 if (mtu) 383 goto out; 384 } 385 386 mtu = IPV6_MIN_MTU; 387 rcu_read_lock(); 388 idev = __in6_dev_get(dst_dev_rcu(dst)); 389 if (idev) 390 mtu = READ_ONCE(idev->cnf.mtu6); 391 rcu_read_unlock(); 392 393 out: 394 mtu = min_t(unsigned int, mtu, IP6_MAX_MTU); 395 396 return mtu - lwtunnel_headroom(dst->lwtstate, mtu); 397 } 398 399 /* Configured/administrative MTU of a route, for advertising the TCP MSS. 400 * 401 * Unlike ip6_dst_mtu_maybe_forward(), this ignores any ICMPv6-learned path 402 * MTU (which is kept on the RTF_CACHE exception route) and returns the MTU of 403 * the underlying route (fib6_pmtu) or the egress device. The advertised MSS 404 * bounds what the peer may send to us and must reflect our receive 405 * capability, not a path MTU learned on the reverse (send) direction. See 406 * RFC 2923 section 2.3 and the comment above tcp_advertise_mss(). 407 */ 408 static inline unsigned int ip6_dst_mtu_configured(const struct dst_entry *dst) 409 { 410 const struct rt6_info *rt = dst_rt6_info(dst); 411 const struct fib6_info *from; 412 struct inet6_dev *idev; 413 unsigned int mtu = 0; 414 415 rcu_read_lock(); 416 /* IPv6 keeps the learned PMTU and the configured MTU in the same 417 * RTAX_MTU slot: the learned value sits on this (possibly RTF_CACHE) 418 * dst, the configured one on the underlying route. Reach the latter 419 * via ->from (fib6_pmtu), populated by ip6_route_info_create(). 420 */ 421 from = rcu_dereference(rt->from); 422 if (from) 423 mtu = from->fib6_pmtu; 424 if (!mtu) { 425 mtu = IPV6_MIN_MTU; 426 idev = __in6_dev_get(dst_dev_rcu(dst)); 427 if (idev) 428 mtu = max_t(unsigned int, mtu, READ_ONCE(idev->cnf.mtu6)); 429 } 430 rcu_read_unlock(); 431 432 mtu = min_t(unsigned int, mtu, IP6_MAX_MTU); 433 return mtu - lwtunnel_headroom(dst->lwtstate, mtu); 434 } 435 436 u32 ip6_mtu_from_fib6(const struct fib6_result *res, 437 const struct in6_addr *daddr, 438 const struct in6_addr *saddr); 439 440 struct neighbour *__ip6_dst_neigh_lookup(const struct in6_addr *gw, 441 struct net_device *dev, struct sk_buff *skb, 442 const void *daddr); 443 #endif 444