1/* SPDX-License-Identifier: GPL-2.0-or-later */ 2/* 3 * INET An implementation of the TCP/IP protocol suite for the LINUX 4 * operating system. INET is implemented using the BSD Socket 5 * interface as the means of communication with the user level. 6 * 7 * Definitions for the IP router. 8 * 9 * Version: @(#)route.h 1.0.4 05/27/93 10 * 11 * Authors: Ross Biro 12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> 13 * Fixes: 14 * Alan Cox : Reformatted. Added ip_rt_local() 15 * Alan Cox : Support for TCP parameters. 16 * Alexey Kuznetsov: Major changes for new routing code. 17 * Mike McLagan : Routing by source 18 * Robert Olsson : Added rt_cache statistics 19 */ 20#ifndef _ROUTE_H 21#define _ROUTE_H 22 23#include <net/dst.h> 24#include <net/inetpeer.h> 25#include <net/flow.h> 26#include <net/inet_sock.h> 27#include <net/ip_fib.h> 28#include <net/arp.h> 29#include <net/ndisc.h> 30#include <linux/in_route.h> 31#include <linux/rtnetlink.h> 32#include <linux/rcupdate.h> 33#include <linux/route.h> 34#include <linux/ip.h> 35#include <linux/cache.h> 36#include <linux/security.h> 37 38#define RTO_ONLINK 0x01 39 40#define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE)) 41#define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE)) 42 43static inline __u8 ip_sock_rt_scope(const struct sock *sk) 44{ 45 if (sock_flag(sk, SOCK_LOCALROUTE)) 46 return RT_SCOPE_LINK; 47 48 return RT_SCOPE_UNIVERSE; 49} 50 51static inline __u8 ip_sock_rt_tos(const struct sock *sk) 52{ 53 return RT_TOS(inet_sk(sk)->tos); 54} 55 56struct ip_tunnel_info; 57struct fib_nh; 58struct fib_info; 59struct uncached_list; 60struct rtable { 61 struct dst_entry dst; 62 63 int rt_genid; 64 unsigned int rt_flags; 65 __u16 rt_type; 66 __u8 rt_is_input; 67 __u8 rt_uses_gateway; 68 69 int rt_iif; 70 71 u8 rt_gw_family; 72 /* Info on neighbour */ 73 union { 74 __be32 rt_gw4; 75 struct in6_addr rt_gw6; 76 }; 77 78 /* Miscellaneous cached information */ 79 u32 rt_mtu_locked:1, 80 rt_pmtu:31; 81}; 82 83static inline bool rt_is_input_route(const struct rtable *rt) 84{ 85 return rt->rt_is_input != 0; 86} 87 88static inline bool rt_is_output_route(const struct rtable *rt) 89{ 90 return rt->rt_is_input == 0; 91} 92 93static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr) 94{ 95 if (rt->rt_gw_family == AF_INET) 96 return rt->rt_gw4; 97 return daddr; 98} 99 100struct ip_rt_acct { 101 __u32 o_bytes; 102 __u32 o_packets; 103 __u32 i_bytes; 104 __u32 i_packets; 105}; 106 107struct rt_cache_stat { 108 unsigned int in_slow_tot; 109 unsigned int in_slow_mc; 110 unsigned int in_no_route; 111 unsigned int in_brd; 112 unsigned int in_martian_dst; 113 unsigned int in_martian_src; 114 unsigned int out_slow_tot; 115 unsigned int out_slow_mc; 116}; 117 118extern struct ip_rt_acct __percpu *ip_rt_acct; 119 120struct in_device; 121 122int ip_rt_init(void); 123void rt_cache_flush(struct net *net); 124void rt_flush_dev(struct net_device *dev); 125struct rtable *ip_route_output_key_hash(struct net *net, struct flowi4 *flp, 126 const struct sk_buff *skb); 127struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *flp, 128 struct fib_result *res, 129 const struct sk_buff *skb); 130 131static inline struct rtable *__ip_route_output_key(struct net *net, 132 struct flowi4 *flp) 133{ 134 return ip_route_output_key_hash(net, flp, NULL); 135} 136 137struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp, 138 const struct sock *sk); 139struct rtable *ip_route_output_tunnel(struct sk_buff *skb, 140 struct net_device *dev, 141 struct net *net, __be32 *saddr, 142 const struct ip_tunnel_info *info, 143 u8 protocol, bool use_cache); 144 145struct dst_entry *ipv4_blackhole_route(struct net *net, 146 struct dst_entry *dst_orig); 147 148static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp) 149{ 150 return ip_route_output_flow(net, flp, NULL); 151} 152 153static inline struct rtable *ip_route_output(struct net *net, __be32 daddr, 154 __be32 saddr, u8 tos, int oif) 155{ 156 struct flowi4 fl4 = { 157 .flowi4_oif = oif, 158 .flowi4_tos = tos, 159 .daddr = daddr, 160 .saddr = saddr, 161 }; 162 return ip_route_output_key(net, &fl4); 163} 164 165static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4, 166 const struct sock *sk, 167 __be32 daddr, __be32 saddr, 168 __be16 dport, __be16 sport, 169 __u8 proto, __u8 tos, int oif) 170{ 171 flowi4_init_output(fl4, oif, sk ? READ_ONCE(sk->sk_mark) : 0, tos, 172 RT_SCOPE_UNIVERSE, proto, 173 sk ? inet_sk_flowi_flags(sk) : 0, 174 daddr, saddr, dport, sport, sock_net_uid(net, sk)); 175 if (sk) 176 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4)); 177 return ip_route_output_flow(net, fl4, sk); 178} 179 180static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4, 181 __be32 daddr, __be32 saddr, 182 __be32 gre_key, __u8 tos, int oif) 183{ 184 memset(fl4, 0, sizeof(*fl4)); 185 fl4->flowi4_oif = oif; 186 fl4->daddr = daddr; 187 fl4->saddr = saddr; 188 fl4->flowi4_tos = tos; 189 fl4->flowi4_proto = IPPROTO_GRE; 190 fl4->fl4_gre_key = gre_key; 191 return ip_route_output_key(net, fl4); 192} 193int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr, 194 u8 tos, struct net_device *dev, 195 struct in_device *in_dev, u32 *itag); 196int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src, 197 u8 tos, struct net_device *devin); 198int ip_route_use_hint(struct sk_buff *skb, __be32 dst, __be32 src, 199 u8 tos, struct net_device *devin, 200 const struct sk_buff *hint); 201 202static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src, 203 u8 tos, struct net_device *devin) 204{ 205 int err; 206 207 rcu_read_lock(); 208 err = ip_route_input_noref(skb, dst, src, tos, devin); 209 if (!err) { 210 skb_dst_force(skb); 211 if (!skb_dst(skb)) 212 err = -EINVAL; 213 } 214 rcu_read_unlock(); 215 216 return err; 217} 218 219void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif, 220 u8 protocol); 221void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu); 222void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u8 protocol); 223void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk); 224void ip_rt_send_redirect(struct sk_buff *skb); 225 226unsigned int inet_addr_type(struct net *net, __be32 addr); 227unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id); 228unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev, 229 __be32 addr); 230unsigned int inet_addr_type_dev_table(struct net *net, 231 const struct net_device *dev, 232 __be32 addr); 233void ip_rt_multicast_event(struct in_device *); 234int ip_rt_ioctl(struct net *, unsigned int cmd, struct rtentry *rt); 235void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt); 236struct rtable *rt_dst_alloc(struct net_device *dev, 237 unsigned int flags, u16 type, bool noxfrm); 238struct rtable *rt_dst_clone(struct net_device *dev, struct rtable *rt); 239 240struct in_ifaddr; 241void fib_add_ifaddr(struct in_ifaddr *); 242void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *); 243void fib_modify_prefix_metric(struct in_ifaddr *ifa, u32 new_metric); 244 245void rt_add_uncached_list(struct rtable *rt); 246void rt_del_uncached_list(struct rtable *rt); 247 248int fib_dump_info_fnhe(struct sk_buff *skb, struct netlink_callback *cb, 249 u32 table_id, struct fib_info *fi, 250 int *fa_index, int fa_start, unsigned int flags); 251 252static inline void ip_rt_put(struct rtable *rt) 253{ 254 /* dst_release() accepts a NULL parameter. 255 * We rely on dst being first structure in struct rtable 256 */ 257 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0); 258 dst_release(&rt->dst); 259} 260 261#define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3) 262 263extern const __u8 ip_tos2prio[16]; 264 265static inline char rt_tos2priority(u8 tos) 266{ 267 return ip_tos2prio[IPTOS_TOS(tos)>>1]; 268} 269 270/* ip_route_connect() and ip_route_newports() work in tandem whilst 271 * binding a socket for a new outgoing connection. 272 * 273 * In order to use IPSEC properly, we must, in the end, have a 274 * route that was looked up using all available keys including source 275 * and destination ports. 276 * 277 * However, if a source port needs to be allocated (the user specified 278 * a wildcard source port) we need to obtain addressing information 279 * in order to perform that allocation. 280 * 281 * So ip_route_connect() looks up a route using wildcarded source and 282 * destination ports in the key, simply so that we can get a pair of 283 * addresses to use for port allocation. 284 * 285 * Later, once the ports are allocated, ip_route_newports() will make 286 * another route lookup if needed to make sure we catch any IPSEC 287 * rules keyed on the port information. 288 * 289 * The callers allocate the flow key on their stack, and must pass in 290 * the same flowi4 object to both the ip_route_connect() and the 291 * ip_route_newports() calls. 292 */ 293 294static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, 295 __be32 src, int oif, u8 protocol, 296 __be16 sport, __be16 dport, 297 const struct sock *sk) 298{ 299 __u8 flow_flags = 0; 300 301 if (inet_test_bit(TRANSPARENT, sk)) 302 flow_flags |= FLOWI_FLAG_ANYSRC; 303 304 flowi4_init_output(fl4, oif, READ_ONCE(sk->sk_mark), ip_sock_rt_tos(sk), 305 ip_sock_rt_scope(sk), protocol, flow_flags, dst, 306 src, dport, sport, sk->sk_uid); 307} 308 309static inline struct rtable *ip_route_connect(struct flowi4 *fl4, __be32 dst, 310 __be32 src, int oif, u8 protocol, 311 __be16 sport, __be16 dport, 312 const struct sock *sk) 313{ 314 struct net *net = sock_net(sk); 315 struct rtable *rt; 316 317 ip_route_connect_init(fl4, dst, src, oif, protocol, sport, dport, sk); 318 319 if (!dst || !src) { 320 rt = __ip_route_output_key(net, fl4); 321 if (IS_ERR(rt)) 322 return rt; 323 ip_rt_put(rt); 324 flowi4_update_output(fl4, oif, fl4->daddr, fl4->saddr); 325 } 326 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4)); 327 return ip_route_output_flow(net, fl4, sk); 328} 329 330static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt, 331 __be16 orig_sport, __be16 orig_dport, 332 __be16 sport, __be16 dport, 333 const struct sock *sk) 334{ 335 if (sport != orig_sport || dport != orig_dport) { 336 fl4->fl4_dport = dport; 337 fl4->fl4_sport = sport; 338 ip_rt_put(rt); 339 flowi4_update_output(fl4, sk->sk_bound_dev_if, fl4->daddr, 340 fl4->saddr); 341 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4)); 342 return ip_route_output_flow(sock_net(sk), fl4, sk); 343 } 344 return rt; 345} 346 347static inline int inet_iif(const struct sk_buff *skb) 348{ 349 struct rtable *rt = skb_rtable(skb); 350 351 if (rt && rt->rt_iif) 352 return rt->rt_iif; 353 354 return skb->skb_iif; 355} 356 357static inline int ip4_dst_hoplimit(const struct dst_entry *dst) 358{ 359 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); 360 struct net *net = dev_net(dst->dev); 361 362 if (hoplimit == 0) 363 hoplimit = READ_ONCE(net->ipv4.sysctl_ip_default_ttl); 364 return hoplimit; 365} 366 367static inline struct neighbour *ip_neigh_gw4(struct net_device *dev, 368 __be32 daddr) 369{ 370 struct neighbour *neigh; 371 372 neigh = __ipv4_neigh_lookup_noref(dev, (__force u32)daddr); 373 if (unlikely(!neigh)) 374 neigh = __neigh_create(&arp_tbl, &daddr, dev, false); 375 376 return neigh; 377} 378 379static inline struct neighbour *ip_neigh_for_gw(struct rtable *rt, 380 struct sk_buff *skb, 381 bool *is_v6gw) 382{ 383 struct net_device *dev = rt->dst.dev; 384 struct neighbour *neigh; 385 386 if (likely(rt->rt_gw_family == AF_INET)) { 387 neigh = ip_neigh_gw4(dev, rt->rt_gw4); 388 } else if (rt->rt_gw_family == AF_INET6) { 389 neigh = ip_neigh_gw6(dev, &rt->rt_gw6); 390 *is_v6gw = true; 391 } else { 392 neigh = ip_neigh_gw4(dev, ip_hdr(skb)->daddr); 393 } 394 return neigh; 395} 396 397#endif /* _ROUTE_H */ 398