11da177e4SLinus Torvalds /* 21da177e4SLinus Torvalds * INET An implementation of the TCP/IP protocol suite for the LINUX 31da177e4SLinus Torvalds * operating system. INET is implemented using the BSD Socket 41da177e4SLinus Torvalds * interface as the means of communication with the user level. 51da177e4SLinus Torvalds * 61da177e4SLinus Torvalds * Definitions for the IP router. 71da177e4SLinus Torvalds * 81da177e4SLinus Torvalds * Version: @(#)route.h 1.0.4 05/27/93 91da177e4SLinus Torvalds * 1002c30a84SJesper Juhl * Authors: Ross Biro 111da177e4SLinus Torvalds * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> 121da177e4SLinus Torvalds * Fixes: 131da177e4SLinus Torvalds * Alan Cox : Reformatted. Added ip_rt_local() 141da177e4SLinus Torvalds * Alan Cox : Support for TCP parameters. 151da177e4SLinus Torvalds * Alexey Kuznetsov: Major changes for new routing code. 161da177e4SLinus Torvalds * Mike McLagan : Routing by source 171da177e4SLinus Torvalds * Robert Olsson : Added rt_cache statistics 181da177e4SLinus Torvalds * 191da177e4SLinus Torvalds * This program is free software; you can redistribute it and/or 201da177e4SLinus Torvalds * modify it under the terms of the GNU General Public License 211da177e4SLinus Torvalds * as published by the Free Software Foundation; either version 221da177e4SLinus Torvalds * 2 of the License, or (at your option) any later version. 231da177e4SLinus Torvalds */ 241da177e4SLinus Torvalds #ifndef _ROUTE_H 251da177e4SLinus Torvalds #define _ROUTE_H 261da177e4SLinus Torvalds 271da177e4SLinus Torvalds #include <net/dst.h> 281da177e4SLinus Torvalds #include <net/inetpeer.h> 291da177e4SLinus Torvalds #include <net/flow.h> 3079876874SKOVACS Krisztian #include <net/inet_sock.h> 311da177e4SLinus Torvalds #include <linux/in_route.h> 321da177e4SLinus Torvalds #include <linux/rtnetlink.h> 331da177e4SLinus Torvalds #include <linux/route.h> 341da177e4SLinus Torvalds #include <linux/ip.h> 351da177e4SLinus Torvalds #include <linux/cache.h> 36beb8d13bSVenkat Yekkirala #include <linux/security.h> 371da177e4SLinus Torvalds 381da177e4SLinus Torvalds #define RTO_ONLINK 0x01 391da177e4SLinus Torvalds 401da177e4SLinus Torvalds #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE)) 411da177e4SLinus Torvalds 421da177e4SLinus Torvalds struct fib_nh; 431da177e4SLinus Torvalds struct inet_peer; 4462fa8a84SDavid S. Miller struct fib_info; 45fd2c3ef7SEric Dumazet struct rtable { 461da177e4SLinus Torvalds struct dst_entry dst; 471da177e4SLinus Torvalds 485e2b61f7SDavid S. Miller /* Lookup key. */ 495e2b61f7SDavid S. Miller __be32 rt_key_dst; 505e2b61f7SDavid S. Miller __be32 rt_key_src; 51093c2ca4SEric Dumazet 5229e75252SEric Dumazet int rt_genid; 5395c96174SEric Dumazet unsigned int rt_flags; 541da177e4SLinus Torvalds __u16 rt_type; 55475949d8SDavid S. Miller __u8 rt_key_tos; 561da177e4SLinus Torvalds 57f2c3fe24SAl Viro __be32 rt_dst; /* Path destination */ 58f2c3fe24SAl Viro __be32 rt_src; /* Path source */ 591b86a58fSOGAWA Hirofumi int rt_route_iif; 601da177e4SLinus Torvalds int rt_iif; 615e2b61f7SDavid S. Miller int rt_oif; 625e2b61f7SDavid S. Miller __u32 rt_mark; 631da177e4SLinus Torvalds 641da177e4SLinus Torvalds /* Info on neighbour */ 65f2c3fe24SAl Viro __be32 rt_gateway; 661da177e4SLinus Torvalds 671da177e4SLinus Torvalds /* Miscellaneous cached information */ 68f2c3fe24SAl Viro __be32 rt_spec_dst; /* RFC1122 specific destination */ 696431cbc2SDavid S. Miller u32 rt_peer_genid; 701da177e4SLinus Torvalds struct inet_peer *peer; /* long-living peer info */ 7162fa8a84SDavid S. Miller struct fib_info *fi; /* for client ref to shared metrics */ 721da177e4SLinus Torvalds }; 731da177e4SLinus Torvalds 74b8400f37SSteffen Klassert static inline bool rt_is_input_route(const struct rtable *rt) 75c7537967SDavid S. Miller { 761b86a58fSOGAWA Hirofumi return rt->rt_route_iif != 0; 77c7537967SDavid S. Miller } 78c7537967SDavid S. Miller 79b8400f37SSteffen Klassert static inline bool rt_is_output_route(const struct rtable *rt) 80c7537967SDavid S. Miller { 811b86a58fSOGAWA Hirofumi return rt->rt_route_iif == 0; 82c7537967SDavid S. Miller } 83c7537967SDavid S. Miller 84fd2c3ef7SEric Dumazet struct ip_rt_acct { 851da177e4SLinus Torvalds __u32 o_bytes; 861da177e4SLinus Torvalds __u32 o_packets; 871da177e4SLinus Torvalds __u32 i_bytes; 881da177e4SLinus Torvalds __u32 i_packets; 891da177e4SLinus Torvalds }; 901da177e4SLinus Torvalds 91fd2c3ef7SEric Dumazet struct rt_cache_stat { 921da177e4SLinus Torvalds unsigned int in_hit; 931da177e4SLinus Torvalds unsigned int in_slow_tot; 941da177e4SLinus Torvalds unsigned int in_slow_mc; 951da177e4SLinus Torvalds unsigned int in_no_route; 961da177e4SLinus Torvalds unsigned int in_brd; 971da177e4SLinus Torvalds unsigned int in_martian_dst; 981da177e4SLinus Torvalds unsigned int in_martian_src; 991da177e4SLinus Torvalds unsigned int out_hit; 1001da177e4SLinus Torvalds unsigned int out_slow_tot; 1011da177e4SLinus Torvalds unsigned int out_slow_mc; 1021da177e4SLinus Torvalds unsigned int gc_total; 1031da177e4SLinus Torvalds unsigned int gc_ignored; 1041da177e4SLinus Torvalds unsigned int gc_goal_miss; 1051da177e4SLinus Torvalds unsigned int gc_dst_overflow; 1061da177e4SLinus Torvalds unsigned int in_hlist_search; 1071da177e4SLinus Torvalds unsigned int out_hlist_search; 1081da177e4SLinus Torvalds }; 1091da177e4SLinus Torvalds 1107d720c3eSTejun Heo extern struct ip_rt_acct __percpu *ip_rt_acct; 1111da177e4SLinus Torvalds 1121da177e4SLinus Torvalds struct in_device; 1131da177e4SLinus Torvalds extern int ip_rt_init(void); 114f7655229SAl Viro extern void ip_rt_redirect(__be32 old_gw, __be32 dst, __be32 new_gw, 115f7655229SAl Viro __be32 src, struct net_device *dev); 11676e6ebfbSDenis V. Lunev extern void rt_cache_flush(struct net *net, int how); 1176561a3b1SDavid S. Miller extern void rt_cache_flush_batch(struct net *net); 118813b3b5dSDavid S. Miller extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp); 1199d6ec938SDavid S. Miller extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp, 120b23dd4feSDavid S. Miller struct sock *sk); 1212774c131SDavid S. Miller extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig); 122407eadd9SEric Dumazet 1239d6ec938SDavid S. Miller static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp) 1245bfa787fSDavid S. Miller { 1255bfa787fSDavid S. Miller return ip_route_output_flow(net, flp, NULL); 1265bfa787fSDavid S. Miller } 1275bfa787fSDavid S. Miller 12878fbfd8aSDavid S. Miller static inline struct rtable *ip_route_output(struct net *net, __be32 daddr, 12978fbfd8aSDavid S. Miller __be32 saddr, u8 tos, int oif) 13078fbfd8aSDavid S. Miller { 1319d6ec938SDavid S. Miller struct flowi4 fl4 = { 1329d6ec938SDavid S. Miller .flowi4_oif = oif, 1339d6ec938SDavid S. Miller .daddr = daddr, 1349d6ec938SDavid S. Miller .saddr = saddr, 1359d6ec938SDavid S. Miller .flowi4_tos = tos, 13678fbfd8aSDavid S. Miller }; 1379d6ec938SDavid S. Miller return ip_route_output_key(net, &fl4); 13878fbfd8aSDavid S. Miller } 13978fbfd8aSDavid S. Miller 14031e4543dSDavid S. Miller static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4, 14131e4543dSDavid S. Miller struct sock *sk, 14278fbfd8aSDavid S. Miller __be32 daddr, __be32 saddr, 14378fbfd8aSDavid S. Miller __be16 dport, __be16 sport, 14478fbfd8aSDavid S. Miller __u8 proto, __u8 tos, int oif) 14578fbfd8aSDavid S. Miller { 14631e4543dSDavid S. Miller flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos, 14794b92b88SDavid S. Miller RT_SCOPE_UNIVERSE, proto, 14894b92b88SDavid S. Miller sk ? inet_sk_flowi_flags(sk) : 0, 14994b92b88SDavid S. Miller daddr, saddr, dport, sport); 15078fbfd8aSDavid S. Miller if (sk) 15131e4543dSDavid S. Miller security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 15231e4543dSDavid S. Miller return ip_route_output_flow(net, fl4, sk); 15378fbfd8aSDavid S. Miller } 15478fbfd8aSDavid S. Miller 155cbb1e85fSDavid S. Miller static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4, 15678fbfd8aSDavid S. Miller __be32 daddr, __be32 saddr, 15778fbfd8aSDavid S. Miller __be32 gre_key, __u8 tos, int oif) 15878fbfd8aSDavid S. Miller { 159cbb1e85fSDavid S. Miller memset(fl4, 0, sizeof(*fl4)); 160cbb1e85fSDavid S. Miller fl4->flowi4_oif = oif; 161cbb1e85fSDavid S. Miller fl4->daddr = daddr; 162cbb1e85fSDavid S. Miller fl4->saddr = saddr; 163cbb1e85fSDavid S. Miller fl4->flowi4_tos = tos; 164cbb1e85fSDavid S. Miller fl4->flowi4_proto = IPPROTO_GRE; 165cbb1e85fSDavid S. Miller fl4->fl4_gre_key = gre_key; 166cbb1e85fSDavid S. Miller return ip_route_output_key(net, fl4); 16778fbfd8aSDavid S. Miller } 16878fbfd8aSDavid S. Miller 169407eadd9SEric Dumazet extern int ip_route_input_common(struct sk_buff *skb, __be32 dst, __be32 src, 170407eadd9SEric Dumazet u8 tos, struct net_device *devin, bool noref); 171407eadd9SEric Dumazet 172407eadd9SEric Dumazet static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src, 173407eadd9SEric Dumazet u8 tos, struct net_device *devin) 174407eadd9SEric Dumazet { 175407eadd9SEric Dumazet return ip_route_input_common(skb, dst, src, tos, devin, false); 176407eadd9SEric Dumazet } 177407eadd9SEric Dumazet 178407eadd9SEric Dumazet static inline int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src, 179407eadd9SEric Dumazet u8 tos, struct net_device *devin) 180407eadd9SEric Dumazet { 181407eadd9SEric Dumazet return ip_route_input_common(skb, dst, src, tos, devin, true); 182407eadd9SEric Dumazet } 183407eadd9SEric Dumazet 184b71d1d42SEric Dumazet extern unsigned short ip_rt_frag_needed(struct net *net, const struct iphdr *iph, 185b71d1d42SEric Dumazet unsigned short new_mtu, struct net_device *dev); 1861da177e4SLinus Torvalds extern void ip_rt_send_redirect(struct sk_buff *skb); 1871da177e4SLinus Torvalds 18895c96174SEric Dumazet extern unsigned int inet_addr_type(struct net *net, __be32 addr); 18995c96174SEric Dumazet extern unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr); 1901da177e4SLinus Torvalds extern void ip_rt_multicast_event(struct in_device *); 1911bad118aSDenis V. Lunev extern int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg); 1928e36360aSDavid S. Miller extern void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt); 1931da177e4SLinus Torvalds extern int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb); 1941da177e4SLinus Torvalds 1950ff60a45SJamal Hadi Salim struct in_ifaddr; 1960ff60a45SJamal Hadi Salim extern void fib_add_ifaddr(struct in_ifaddr *); 197e6abbaa2SJulian Anastasov extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *); 1980ff60a45SJamal Hadi Salim 1991da177e4SLinus Torvalds static inline void ip_rt_put(struct rtable * rt) 2001da177e4SLinus Torvalds { 2011da177e4SLinus Torvalds if (rt) 202d8d1f30bSChangli Gao dst_release(&rt->dst); 2031da177e4SLinus Torvalds } 2041da177e4SLinus Torvalds 2051da177e4SLinus Torvalds #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3) 2061da177e4SLinus Torvalds 2074839c52bSPhilippe De Muyter extern const __u8 ip_tos2prio[16]; 2081da177e4SLinus Torvalds 2091da177e4SLinus Torvalds static inline char rt_tos2priority(u8 tos) 2101da177e4SLinus Torvalds { 2111da177e4SLinus Torvalds return ip_tos2prio[IPTOS_TOS(tos)>>1]; 2121da177e4SLinus Torvalds } 2131da177e4SLinus Torvalds 2142d7192d6SDavid S. Miller /* ip_route_connect() and ip_route_newports() work in tandem whilst 2152d7192d6SDavid S. Miller * binding a socket for a new outgoing connection. 2162d7192d6SDavid S. Miller * 2172d7192d6SDavid S. Miller * In order to use IPSEC properly, we must, in the end, have a 2182d7192d6SDavid S. Miller * route that was looked up using all available keys including source 2192d7192d6SDavid S. Miller * and destination ports. 2202d7192d6SDavid S. Miller * 2212d7192d6SDavid S. Miller * However, if a source port needs to be allocated (the user specified 2222d7192d6SDavid S. Miller * a wildcard source port) we need to obtain addressing information 2232d7192d6SDavid S. Miller * in order to perform that allocation. 2242d7192d6SDavid S. Miller * 2252d7192d6SDavid S. Miller * So ip_route_connect() looks up a route using wildcarded source and 2262d7192d6SDavid S. Miller * destination ports in the key, simply so that we can get a pair of 2272d7192d6SDavid S. Miller * addresses to use for port allocation. 2282d7192d6SDavid S. Miller * 2292d7192d6SDavid S. Miller * Later, once the ports are allocated, ip_route_newports() will make 2302d7192d6SDavid S. Miller * another route lookup if needed to make sure we catch any IPSEC 2312d7192d6SDavid S. Miller * rules keyed on the port information. 2322d7192d6SDavid S. Miller * 2332d7192d6SDavid S. Miller * The callers allocate the flow key on their stack, and must pass in 2342d7192d6SDavid S. Miller * the same flowi4 object to both the ip_route_connect() and the 2352d7192d6SDavid S. Miller * ip_route_newports() calls. 2362d7192d6SDavid S. Miller */ 2372d7192d6SDavid S. Miller 2382d7192d6SDavid S. Miller static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src, 2392d7192d6SDavid S. Miller u32 tos, int oif, u8 protocol, 240b23dd4feSDavid S. Miller __be16 sport, __be16 dport, 241b23dd4feSDavid S. Miller struct sock *sk, bool can_sleep) 2421da177e4SLinus Torvalds { 2432d7192d6SDavid S. Miller __u8 flow_flags = 0; 24479876874SKOVACS Krisztian 24579876874SKOVACS Krisztian if (inet_sk(sk)->transparent) 24694b92b88SDavid S. Miller flow_flags |= FLOWI_FLAG_ANYSRC; 247a4daad6bSDavid S. Miller if (protocol == IPPROTO_TCP) 24894b92b88SDavid S. Miller flow_flags |= FLOWI_FLAG_PRECOW_METRICS; 2495df65e55SDavid S. Miller if (can_sleep) 25094b92b88SDavid S. Miller flow_flags |= FLOWI_FLAG_CAN_SLEEP; 25194b92b88SDavid S. Miller 2522d7192d6SDavid S. Miller flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE, 25394b92b88SDavid S. Miller protocol, flow_flags, dst, src, dport, sport); 2542d7192d6SDavid S. Miller } 2552d7192d6SDavid S. Miller 2562d7192d6SDavid S. Miller static inline struct rtable *ip_route_connect(struct flowi4 *fl4, 2572d7192d6SDavid S. Miller __be32 dst, __be32 src, u32 tos, 2582d7192d6SDavid S. Miller int oif, u8 protocol, 2592d7192d6SDavid S. Miller __be16 sport, __be16 dport, 2602d7192d6SDavid S. Miller struct sock *sk, bool can_sleep) 2612d7192d6SDavid S. Miller { 2622d7192d6SDavid S. Miller struct net *net = sock_net(sk); 2632d7192d6SDavid S. Miller struct rtable *rt; 2642d7192d6SDavid S. Miller 2652d7192d6SDavid S. Miller ip_route_connect_init(fl4, dst, src, tos, oif, protocol, 2662d7192d6SDavid S. Miller sport, dport, sk, can_sleep); 26779876874SKOVACS Krisztian 2681da177e4SLinus Torvalds if (!dst || !src) { 2692d7192d6SDavid S. Miller rt = __ip_route_output_key(net, fl4); 270b23dd4feSDavid S. Miller if (IS_ERR(rt)) 271b23dd4feSDavid S. Miller return rt; 272b23dd4feSDavid S. Miller ip_rt_put(rt); 273e6b45241SJulian Anastasov flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr); 2741da177e4SLinus Torvalds } 2752d7192d6SDavid S. Miller security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 2762d7192d6SDavid S. Miller return ip_route_output_flow(net, fl4, sk); 2771da177e4SLinus Torvalds } 2781da177e4SLinus Torvalds 2792d7192d6SDavid S. Miller static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt, 2802d7192d6SDavid S. Miller __be16 orig_sport, __be16 orig_dport, 2812d7192d6SDavid S. Miller __be16 sport, __be16 dport, 2822d7192d6SDavid S. Miller struct sock *sk) 2831da177e4SLinus Torvalds { 284dca8b089SDavid S. Miller if (sport != orig_sport || dport != orig_dport) { 2852d7192d6SDavid S. Miller fl4->fl4_dport = dport; 2862d7192d6SDavid S. Miller fl4->fl4_sport = sport; 287b23dd4feSDavid S. Miller ip_rt_put(rt); 288e6b45241SJulian Anastasov flowi4_update_output(fl4, sk->sk_bound_dev_if, 289e6b45241SJulian Anastasov RT_CONN_FLAGS(sk), fl4->daddr, 290e6b45241SJulian Anastasov fl4->saddr); 2912d7192d6SDavid S. Miller security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 2922d7192d6SDavid S. Miller return ip_route_output_flow(sock_net(sk), fl4, sk); 2931da177e4SLinus Torvalds } 294b23dd4feSDavid S. Miller return rt; 2951da177e4SLinus Torvalds } 2961da177e4SLinus Torvalds 297a48eff12SDavid S. Miller extern void rt_bind_peer(struct rtable *rt, __be32 daddr, int create); 2981da177e4SLinus Torvalds 299fbfe95a4SDavid S. Miller static inline struct inet_peer *__rt_get_peer(struct rtable *rt, __be32 daddr, int create) 3001da177e4SLinus Torvalds { 3011da177e4SLinus Torvalds if (rt->peer) 3021da177e4SLinus Torvalds return rt->peer; 3031da177e4SLinus Torvalds 304fbfe95a4SDavid S. Miller rt_bind_peer(rt, daddr, create); 3051da177e4SLinus Torvalds return rt->peer; 3061da177e4SLinus Torvalds } 3071da177e4SLinus Torvalds 308fbfe95a4SDavid S. Miller static inline struct inet_peer *rt_get_peer(struct rtable *rt, __be32 daddr) 309fbfe95a4SDavid S. Miller { 310fbfe95a4SDavid S. Miller return __rt_get_peer(rt, daddr, 0); 311fbfe95a4SDavid S. Miller } 312fbfe95a4SDavid S. Miller 313fbfe95a4SDavid S. Miller static inline struct inet_peer *rt_get_peer_create(struct rtable *rt, __be32 daddr) 314fbfe95a4SDavid S. Miller { 315fbfe95a4SDavid S. Miller return __rt_get_peer(rt, daddr, 1); 316fbfe95a4SDavid S. Miller } 317fbfe95a4SDavid S. Miller 3181668e010SKOVACS Krisztian static inline int inet_iif(const struct sk_buff *skb) 3191668e010SKOVACS Krisztian { 320511c3f92SEric Dumazet return skb_rtable(skb)->rt_iif; 3211668e010SKOVACS Krisztian } 3221668e010SKOVACS Krisztian 323323e126fSDavid S. Miller extern int sysctl_ip_default_ttl; 324323e126fSDavid S. Miller 325323e126fSDavid S. Miller static inline int ip4_dst_hoplimit(const struct dst_entry *dst) 326323e126fSDavid S. Miller { 327323e126fSDavid S. Miller int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); 328323e126fSDavid S. Miller 329323e126fSDavid S. Miller if (hoplimit == 0) 330323e126fSDavid S. Miller hoplimit = sysctl_ip_default_ttl; 331323e126fSDavid S. Miller return hoplimit; 332323e126fSDavid S. Miller } 333323e126fSDavid S. Miller 3341da177e4SLinus Torvalds #endif /* _ROUTE_H */ 335