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 fib_info; 44 struct rtable { 45 struct dst_entry dst; 46 47 int rt_genid; 48 unsigned int rt_flags; 49 __u16 rt_type; 50 51 int rt_route_iif; 52 int rt_iif; 53 int rt_oif; 54 55 /* Info on neighbour */ 56 __be32 rt_gateway; 57 58 /* Miscellaneous cached information */ 59 u32 rt_pmtu; 60 struct fib_info *fi; /* for client ref to shared metrics */ 61 }; 62 63 static inline bool rt_is_input_route(const struct rtable *rt) 64 { 65 return rt->rt_route_iif != 0; 66 } 67 68 static inline bool rt_is_output_route(const struct rtable *rt) 69 { 70 return rt->rt_route_iif == 0; 71 } 72 73 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr) 74 { 75 if (rt->rt_gateway) 76 return rt->rt_gateway; 77 return daddr; 78 } 79 80 struct ip_rt_acct { 81 __u32 o_bytes; 82 __u32 o_packets; 83 __u32 i_bytes; 84 __u32 i_packets; 85 }; 86 87 struct rt_cache_stat { 88 unsigned int in_hit; 89 unsigned int in_slow_tot; 90 unsigned int in_slow_mc; 91 unsigned int in_no_route; 92 unsigned int in_brd; 93 unsigned int in_martian_dst; 94 unsigned int in_martian_src; 95 unsigned int out_hit; 96 unsigned int out_slow_tot; 97 unsigned int out_slow_mc; 98 unsigned int gc_total; 99 unsigned int gc_ignored; 100 unsigned int gc_goal_miss; 101 unsigned int gc_dst_overflow; 102 unsigned int in_hlist_search; 103 unsigned int out_hlist_search; 104 }; 105 106 extern struct ip_rt_acct __percpu *ip_rt_acct; 107 108 struct in_device; 109 extern int ip_rt_init(void); 110 extern void rt_cache_flush(struct net *net, int how); 111 extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp); 112 extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp, 113 struct sock *sk); 114 extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig); 115 116 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp) 117 { 118 return ip_route_output_flow(net, flp, NULL); 119 } 120 121 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr, 122 __be32 saddr, u8 tos, int oif) 123 { 124 struct flowi4 fl4 = { 125 .flowi4_oif = oif, 126 .flowi4_tos = tos, 127 .daddr = daddr, 128 .saddr = saddr, 129 }; 130 return ip_route_output_key(net, &fl4); 131 } 132 133 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4, 134 struct sock *sk, 135 __be32 daddr, __be32 saddr, 136 __be16 dport, __be16 sport, 137 __u8 proto, __u8 tos, int oif) 138 { 139 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos, 140 RT_SCOPE_UNIVERSE, proto, 141 sk ? inet_sk_flowi_flags(sk) : 0, 142 daddr, saddr, dport, sport); 143 if (sk) 144 security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 145 return ip_route_output_flow(net, fl4, sk); 146 } 147 148 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4, 149 __be32 daddr, __be32 saddr, 150 __be32 gre_key, __u8 tos, int oif) 151 { 152 memset(fl4, 0, sizeof(*fl4)); 153 fl4->flowi4_oif = oif; 154 fl4->daddr = daddr; 155 fl4->saddr = saddr; 156 fl4->flowi4_tos = tos; 157 fl4->flowi4_proto = IPPROTO_GRE; 158 fl4->fl4_gre_key = gre_key; 159 return ip_route_output_key(net, fl4); 160 } 161 162 extern int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src, 163 u8 tos, struct net_device *devin); 164 165 extern void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, 166 int oif, u32 mark, u8 protocol, int flow_flags); 167 extern void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu); 168 extern void ipv4_redirect(struct sk_buff *skb, struct net *net, 169 int oif, u32 mark, u8 protocol, int flow_flags); 170 extern void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk); 171 extern void ip_rt_send_redirect(struct sk_buff *skb); 172 173 extern unsigned int inet_addr_type(struct net *net, __be32 addr); 174 extern unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr); 175 extern void ip_rt_multicast_event(struct in_device *); 176 extern int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg); 177 extern void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt); 178 extern int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb); 179 180 struct in_ifaddr; 181 extern void fib_add_ifaddr(struct in_ifaddr *); 182 extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *); 183 184 static inline void ip_rt_put(struct rtable * rt) 185 { 186 if (rt) 187 dst_release(&rt->dst); 188 } 189 190 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3) 191 192 extern const __u8 ip_tos2prio[16]; 193 194 static inline char rt_tos2priority(u8 tos) 195 { 196 return ip_tos2prio[IPTOS_TOS(tos)>>1]; 197 } 198 199 /* ip_route_connect() and ip_route_newports() work in tandem whilst 200 * binding a socket for a new outgoing connection. 201 * 202 * In order to use IPSEC properly, we must, in the end, have a 203 * route that was looked up using all available keys including source 204 * and destination ports. 205 * 206 * However, if a source port needs to be allocated (the user specified 207 * a wildcard source port) we need to obtain addressing information 208 * in order to perform that allocation. 209 * 210 * So ip_route_connect() looks up a route using wildcarded source and 211 * destination ports in the key, simply so that we can get a pair of 212 * addresses to use for port allocation. 213 * 214 * Later, once the ports are allocated, ip_route_newports() will make 215 * another route lookup if needed to make sure we catch any IPSEC 216 * rules keyed on the port information. 217 * 218 * The callers allocate the flow key on their stack, and must pass in 219 * the same flowi4 object to both the ip_route_connect() and the 220 * ip_route_newports() calls. 221 */ 222 223 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src, 224 u32 tos, int oif, u8 protocol, 225 __be16 sport, __be16 dport, 226 struct sock *sk, bool can_sleep) 227 { 228 __u8 flow_flags = 0; 229 230 if (inet_sk(sk)->transparent) 231 flow_flags |= FLOWI_FLAG_ANYSRC; 232 if (can_sleep) 233 flow_flags |= FLOWI_FLAG_CAN_SLEEP; 234 235 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE, 236 protocol, flow_flags, dst, src, dport, sport); 237 } 238 239 static inline struct rtable *ip_route_connect(struct flowi4 *fl4, 240 __be32 dst, __be32 src, u32 tos, 241 int oif, u8 protocol, 242 __be16 sport, __be16 dport, 243 struct sock *sk, bool can_sleep) 244 { 245 struct net *net = sock_net(sk); 246 struct rtable *rt; 247 248 ip_route_connect_init(fl4, dst, src, tos, oif, protocol, 249 sport, dport, sk, can_sleep); 250 251 if (!dst || !src) { 252 rt = __ip_route_output_key(net, fl4); 253 if (IS_ERR(rt)) 254 return rt; 255 ip_rt_put(rt); 256 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr); 257 } 258 security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 259 return ip_route_output_flow(net, fl4, sk); 260 } 261 262 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt, 263 __be16 orig_sport, __be16 orig_dport, 264 __be16 sport, __be16 dport, 265 struct sock *sk) 266 { 267 if (sport != orig_sport || dport != orig_dport) { 268 fl4->fl4_dport = dport; 269 fl4->fl4_sport = sport; 270 ip_rt_put(rt); 271 flowi4_update_output(fl4, sk->sk_bound_dev_if, 272 RT_CONN_FLAGS(sk), fl4->daddr, 273 fl4->saddr); 274 security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 275 return ip_route_output_flow(sock_net(sk), fl4, sk); 276 } 277 return rt; 278 } 279 280 static inline int inet_iif(const struct sk_buff *skb) 281 { 282 return skb_rtable(skb)->rt_iif; 283 } 284 285 extern int sysctl_ip_default_ttl; 286 287 static inline int ip4_dst_hoplimit(const struct dst_entry *dst) 288 { 289 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); 290 291 if (hoplimit == 0) 292 hoplimit = sysctl_ip_default_ttl; 293 return hoplimit; 294 } 295 296 #endif /* _ROUTE_H */ 297