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 27 #include <net/dst.h> 28 #include <net/inetpeer.h> 29 #include <net/flow.h> 30 #include <net/inet_sock.h> 31 #include <linux/in_route.h> 32 #include <linux/rtnetlink.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 42 struct fib_nh; 43 struct inet_peer; 44 struct fib_info; 45 struct rtable { 46 struct dst_entry dst; 47 48 /* Lookup key. */ 49 __be32 rt_key_dst; 50 __be32 rt_key_src; 51 52 int rt_genid; 53 unsigned int rt_flags; 54 __u16 rt_type; 55 __u8 rt_key_tos; 56 57 __be32 rt_dst; /* Path destination */ 58 __be32 rt_src; /* Path source */ 59 int rt_route_iif; 60 int rt_iif; 61 int rt_oif; 62 __u32 rt_mark; 63 64 /* Info on neighbour */ 65 __be32 rt_gateway; 66 67 /* Miscellaneous cached information */ 68 u32 rt_peer_genid; 69 unsigned long _peer; /* long-living peer info */ 70 struct fib_info *fi; /* for client ref to shared metrics */ 71 }; 72 73 static inline struct inet_peer *rt_peer_ptr(struct rtable *rt) 74 { 75 return inetpeer_ptr(rt->_peer); 76 } 77 78 static inline bool rt_has_peer(struct rtable *rt) 79 { 80 return inetpeer_ptr_is_peer(rt->_peer); 81 } 82 83 static inline void __rt_set_peer(struct rtable *rt, struct inet_peer *peer) 84 { 85 __inetpeer_ptr_set_peer(&rt->_peer, peer); 86 } 87 88 static inline bool rt_set_peer(struct rtable *rt, struct inet_peer *peer) 89 { 90 return inetpeer_ptr_set_peer(&rt->_peer, peer); 91 } 92 93 static inline void rt_init_peer(struct rtable *rt, struct inet_peer_base *base) 94 { 95 inetpeer_init_ptr(&rt->_peer, base); 96 } 97 98 static inline void rt_transfer_peer(struct rtable *rt, struct rtable *ort) 99 { 100 rt->_peer = ort->_peer; 101 if (rt_has_peer(ort)) { 102 struct inet_peer *peer = rt_peer_ptr(ort); 103 atomic_inc(&peer->refcnt); 104 } 105 } 106 107 static inline bool rt_is_input_route(const struct rtable *rt) 108 { 109 return rt->rt_route_iif != 0; 110 } 111 112 static inline bool rt_is_output_route(const struct rtable *rt) 113 { 114 return rt->rt_route_iif == 0; 115 } 116 117 struct ip_rt_acct { 118 __u32 o_bytes; 119 __u32 o_packets; 120 __u32 i_bytes; 121 __u32 i_packets; 122 }; 123 124 struct rt_cache_stat { 125 unsigned int in_hit; 126 unsigned int in_slow_tot; 127 unsigned int in_slow_mc; 128 unsigned int in_no_route; 129 unsigned int in_brd; 130 unsigned int in_martian_dst; 131 unsigned int in_martian_src; 132 unsigned int out_hit; 133 unsigned int out_slow_tot; 134 unsigned int out_slow_mc; 135 unsigned int gc_total; 136 unsigned int gc_ignored; 137 unsigned int gc_goal_miss; 138 unsigned int gc_dst_overflow; 139 unsigned int in_hlist_search; 140 unsigned int out_hlist_search; 141 }; 142 143 extern struct ip_rt_acct __percpu *ip_rt_acct; 144 145 struct in_device; 146 extern int ip_rt_init(void); 147 extern void ip_rt_redirect(__be32 old_gw, __be32 dst, __be32 new_gw, 148 __be32 src, struct net_device *dev); 149 extern void rt_cache_flush(struct net *net, int how); 150 extern void rt_cache_flush_batch(struct net *net); 151 extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp); 152 extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp, 153 struct sock *sk); 154 extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig); 155 156 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp) 157 { 158 return ip_route_output_flow(net, flp, NULL); 159 } 160 161 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr, 162 __be32 saddr, u8 tos, int oif) 163 { 164 struct flowi4 fl4 = { 165 .flowi4_oif = oif, 166 .flowi4_tos = tos, 167 .daddr = daddr, 168 .saddr = saddr, 169 }; 170 return ip_route_output_key(net, &fl4); 171 } 172 173 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4, 174 struct sock *sk, 175 __be32 daddr, __be32 saddr, 176 __be16 dport, __be16 sport, 177 __u8 proto, __u8 tos, int oif) 178 { 179 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos, 180 RT_SCOPE_UNIVERSE, proto, 181 sk ? inet_sk_flowi_flags(sk) : 0, 182 daddr, saddr, dport, sport); 183 if (sk) 184 security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 185 return ip_route_output_flow(net, fl4, sk); 186 } 187 188 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4, 189 __be32 daddr, __be32 saddr, 190 __be32 gre_key, __u8 tos, int oif) 191 { 192 memset(fl4, 0, sizeof(*fl4)); 193 fl4->flowi4_oif = oif; 194 fl4->daddr = daddr; 195 fl4->saddr = saddr; 196 fl4->flowi4_tos = tos; 197 fl4->flowi4_proto = IPPROTO_GRE; 198 fl4->fl4_gre_key = gre_key; 199 return ip_route_output_key(net, fl4); 200 } 201 202 extern int ip_route_input_common(struct sk_buff *skb, __be32 dst, __be32 src, 203 u8 tos, struct net_device *devin, bool noref); 204 205 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src, 206 u8 tos, struct net_device *devin) 207 { 208 return ip_route_input_common(skb, dst, src, tos, devin, false); 209 } 210 211 static inline int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src, 212 u8 tos, struct net_device *devin) 213 { 214 return ip_route_input_common(skb, dst, src, tos, devin, true); 215 } 216 217 extern void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, 218 int oif, u32 mark, u8 protocol, int flow_flags); 219 extern void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu); 220 extern void ip_rt_send_redirect(struct sk_buff *skb); 221 222 extern unsigned int inet_addr_type(struct net *net, __be32 addr); 223 extern unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr); 224 extern void ip_rt_multicast_event(struct in_device *); 225 extern int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg); 226 extern void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt); 227 extern int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb); 228 229 struct in_ifaddr; 230 extern void fib_add_ifaddr(struct in_ifaddr *); 231 extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *); 232 233 static inline void ip_rt_put(struct rtable * rt) 234 { 235 if (rt) 236 dst_release(&rt->dst); 237 } 238 239 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3) 240 241 extern const __u8 ip_tos2prio[16]; 242 243 static inline char rt_tos2priority(u8 tos) 244 { 245 return ip_tos2prio[IPTOS_TOS(tos)>>1]; 246 } 247 248 /* ip_route_connect() and ip_route_newports() work in tandem whilst 249 * binding a socket for a new outgoing connection. 250 * 251 * In order to use IPSEC properly, we must, in the end, have a 252 * route that was looked up using all available keys including source 253 * and destination ports. 254 * 255 * However, if a source port needs to be allocated (the user specified 256 * a wildcard source port) we need to obtain addressing information 257 * in order to perform that allocation. 258 * 259 * So ip_route_connect() looks up a route using wildcarded source and 260 * destination ports in the key, simply so that we can get a pair of 261 * addresses to use for port allocation. 262 * 263 * Later, once the ports are allocated, ip_route_newports() will make 264 * another route lookup if needed to make sure we catch any IPSEC 265 * rules keyed on the port information. 266 * 267 * The callers allocate the flow key on their stack, and must pass in 268 * the same flowi4 object to both the ip_route_connect() and the 269 * ip_route_newports() calls. 270 */ 271 272 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src, 273 u32 tos, int oif, u8 protocol, 274 __be16 sport, __be16 dport, 275 struct sock *sk, bool can_sleep) 276 { 277 __u8 flow_flags = 0; 278 279 if (inet_sk(sk)->transparent) 280 flow_flags |= FLOWI_FLAG_ANYSRC; 281 if (protocol == IPPROTO_TCP) 282 flow_flags |= FLOWI_FLAG_PRECOW_METRICS; 283 if (can_sleep) 284 flow_flags |= FLOWI_FLAG_CAN_SLEEP; 285 286 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE, 287 protocol, flow_flags, dst, src, dport, sport); 288 } 289 290 static inline struct rtable *ip_route_connect(struct flowi4 *fl4, 291 __be32 dst, __be32 src, u32 tos, 292 int oif, u8 protocol, 293 __be16 sport, __be16 dport, 294 struct sock *sk, bool can_sleep) 295 { 296 struct net *net = sock_net(sk); 297 struct rtable *rt; 298 299 ip_route_connect_init(fl4, dst, src, tos, oif, protocol, 300 sport, dport, sk, can_sleep); 301 302 if (!dst || !src) { 303 rt = __ip_route_output_key(net, fl4); 304 if (IS_ERR(rt)) 305 return rt; 306 ip_rt_put(rt); 307 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr); 308 } 309 security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 310 return ip_route_output_flow(net, fl4, sk); 311 } 312 313 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt, 314 __be16 orig_sport, __be16 orig_dport, 315 __be16 sport, __be16 dport, 316 struct sock *sk) 317 { 318 if (sport != orig_sport || dport != orig_dport) { 319 fl4->fl4_dport = dport; 320 fl4->fl4_sport = sport; 321 ip_rt_put(rt); 322 flowi4_update_output(fl4, sk->sk_bound_dev_if, 323 RT_CONN_FLAGS(sk), fl4->daddr, 324 fl4->saddr); 325 security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 326 return ip_route_output_flow(sock_net(sk), fl4, sk); 327 } 328 return rt; 329 } 330 331 extern void rt_bind_peer(struct rtable *rt, __be32 daddr, int create); 332 333 static inline struct inet_peer *__rt_get_peer(struct rtable *rt, __be32 daddr, int create) 334 { 335 if (rt_has_peer(rt)) 336 return rt_peer_ptr(rt); 337 338 rt_bind_peer(rt, daddr, create); 339 return (rt_has_peer(rt) ? rt_peer_ptr(rt) : NULL); 340 } 341 342 static inline struct inet_peer *rt_get_peer(struct rtable *rt, __be32 daddr) 343 { 344 return __rt_get_peer(rt, daddr, 0); 345 } 346 347 static inline struct inet_peer *rt_get_peer_create(struct rtable *rt, __be32 daddr) 348 { 349 return __rt_get_peer(rt, daddr, 1); 350 } 351 352 static inline int inet_iif(const struct sk_buff *skb) 353 { 354 return skb_rtable(skb)->rt_iif; 355 } 356 357 extern int sysctl_ip_default_ttl; 358 359 static inline int ip4_dst_hoplimit(const struct dst_entry *dst) 360 { 361 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); 362 363 if (hoplimit == 0) 364 hoplimit = sysctl_ip_default_ttl; 365 return hoplimit; 366 } 367 368 #endif /* _ROUTE_H */ 369