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/rcupdate.h> 34 #include <linux/route.h> 35 #include <linux/ip.h> 36 #include <linux/cache.h> 37 #include <linux/security.h> 38 39 #define RTO_ONLINK 0x01 40 41 #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE)) 42 #define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE)) 43 44 struct fib_nh; 45 struct fib_info; 46 struct rtable { 47 struct dst_entry dst; 48 49 int rt_genid; 50 unsigned int rt_flags; 51 __u16 rt_type; 52 __u8 rt_is_input; 53 __u8 rt_uses_gateway; 54 55 int rt_iif; 56 57 /* Info on neighbour */ 58 __be32 rt_gateway; 59 60 /* Miscellaneous cached information */ 61 u32 rt_pmtu; 62 63 struct list_head rt_uncached; 64 }; 65 66 static inline bool rt_is_input_route(const struct rtable *rt) 67 { 68 return rt->rt_is_input != 0; 69 } 70 71 static inline bool rt_is_output_route(const struct rtable *rt) 72 { 73 return rt->rt_is_input == 0; 74 } 75 76 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr) 77 { 78 if (rt->rt_gateway) 79 return rt->rt_gateway; 80 return daddr; 81 } 82 83 struct ip_rt_acct { 84 __u32 o_bytes; 85 __u32 o_packets; 86 __u32 i_bytes; 87 __u32 i_packets; 88 }; 89 90 struct rt_cache_stat { 91 unsigned int in_slow_tot; 92 unsigned int in_slow_mc; 93 unsigned int in_no_route; 94 unsigned int in_brd; 95 unsigned int in_martian_dst; 96 unsigned int in_martian_src; 97 unsigned int out_slow_tot; 98 unsigned int out_slow_mc; 99 }; 100 101 extern struct ip_rt_acct __percpu *ip_rt_acct; 102 103 struct in_device; 104 105 int ip_rt_init(void); 106 void rt_cache_flush(struct net *net); 107 void rt_flush_dev(struct net_device *dev); 108 struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp); 109 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp, 110 struct sock *sk); 111 struct dst_entry *ipv4_blackhole_route(struct net *net, 112 struct dst_entry *dst_orig); 113 114 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp) 115 { 116 return ip_route_output_flow(net, flp, NULL); 117 } 118 119 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr, 120 __be32 saddr, u8 tos, int oif) 121 { 122 struct flowi4 fl4 = { 123 .flowi4_oif = oif, 124 .flowi4_tos = tos, 125 .daddr = daddr, 126 .saddr = saddr, 127 }; 128 return ip_route_output_key(net, &fl4); 129 } 130 131 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4, 132 struct sock *sk, 133 __be32 daddr, __be32 saddr, 134 __be16 dport, __be16 sport, 135 __u8 proto, __u8 tos, int oif) 136 { 137 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos, 138 RT_SCOPE_UNIVERSE, proto, 139 sk ? inet_sk_flowi_flags(sk) : 0, 140 daddr, saddr, dport, sport); 141 if (sk) 142 security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 143 return ip_route_output_flow(net, fl4, sk); 144 } 145 146 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4, 147 __be32 daddr, __be32 saddr, 148 __be32 gre_key, __u8 tos, int oif) 149 { 150 memset(fl4, 0, sizeof(*fl4)); 151 fl4->flowi4_oif = oif; 152 fl4->daddr = daddr; 153 fl4->saddr = saddr; 154 fl4->flowi4_tos = tos; 155 fl4->flowi4_proto = IPPROTO_GRE; 156 fl4->fl4_gre_key = gre_key; 157 return ip_route_output_key(net, fl4); 158 } 159 160 int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src, 161 u8 tos, struct net_device *devin); 162 163 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src, 164 u8 tos, struct net_device *devin) 165 { 166 int err; 167 168 rcu_read_lock(); 169 err = ip_route_input_noref(skb, dst, src, tos, devin); 170 if (!err) 171 skb_dst_force(skb); 172 rcu_read_unlock(); 173 174 return err; 175 } 176 177 void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif, 178 u32 mark, u8 protocol, int flow_flags); 179 void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu); 180 void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark, 181 u8 protocol, int flow_flags); 182 void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk); 183 void ip_rt_send_redirect(struct sk_buff *skb); 184 185 unsigned int inet_addr_type(struct net *net, __be32 addr); 186 unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev, 187 __be32 addr); 188 void ip_rt_multicast_event(struct in_device *); 189 int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg); 190 void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt); 191 int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb); 192 193 struct in_ifaddr; 194 void fib_add_ifaddr(struct in_ifaddr *); 195 void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *); 196 197 static inline void ip_rt_put(struct rtable *rt) 198 { 199 /* dst_release() accepts a NULL parameter. 200 * We rely on dst being first structure in struct rtable 201 */ 202 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0); 203 dst_release(&rt->dst); 204 } 205 206 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3) 207 208 extern const __u8 ip_tos2prio[16]; 209 210 static inline char rt_tos2priority(u8 tos) 211 { 212 return ip_tos2prio[IPTOS_TOS(tos)>>1]; 213 } 214 215 /* ip_route_connect() and ip_route_newports() work in tandem whilst 216 * binding a socket for a new outgoing connection. 217 * 218 * In order to use IPSEC properly, we must, in the end, have a 219 * route that was looked up using all available keys including source 220 * and destination ports. 221 * 222 * However, if a source port needs to be allocated (the user specified 223 * a wildcard source port) we need to obtain addressing information 224 * in order to perform that allocation. 225 * 226 * So ip_route_connect() looks up a route using wildcarded source and 227 * destination ports in the key, simply so that we can get a pair of 228 * addresses to use for port allocation. 229 * 230 * Later, once the ports are allocated, ip_route_newports() will make 231 * another route lookup if needed to make sure we catch any IPSEC 232 * rules keyed on the port information. 233 * 234 * The callers allocate the flow key on their stack, and must pass in 235 * the same flowi4 object to both the ip_route_connect() and the 236 * ip_route_newports() calls. 237 */ 238 239 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src, 240 u32 tos, int oif, u8 protocol, 241 __be16 sport, __be16 dport, 242 struct sock *sk, bool can_sleep) 243 { 244 __u8 flow_flags = 0; 245 246 if (inet_sk(sk)->transparent) 247 flow_flags |= FLOWI_FLAG_ANYSRC; 248 if (can_sleep) 249 flow_flags |= FLOWI_FLAG_CAN_SLEEP; 250 251 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE, 252 protocol, flow_flags, dst, src, dport, sport); 253 } 254 255 static inline struct rtable *ip_route_connect(struct flowi4 *fl4, 256 __be32 dst, __be32 src, u32 tos, 257 int oif, u8 protocol, 258 __be16 sport, __be16 dport, 259 struct sock *sk, bool can_sleep) 260 { 261 struct net *net = sock_net(sk); 262 struct rtable *rt; 263 264 ip_route_connect_init(fl4, dst, src, tos, oif, protocol, 265 sport, dport, sk, can_sleep); 266 267 if (!dst || !src) { 268 rt = __ip_route_output_key(net, fl4); 269 if (IS_ERR(rt)) 270 return rt; 271 ip_rt_put(rt); 272 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr); 273 } 274 security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 275 return ip_route_output_flow(net, fl4, sk); 276 } 277 278 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt, 279 __be16 orig_sport, __be16 orig_dport, 280 __be16 sport, __be16 dport, 281 struct sock *sk) 282 { 283 if (sport != orig_sport || dport != orig_dport) { 284 fl4->fl4_dport = dport; 285 fl4->fl4_sport = sport; 286 ip_rt_put(rt); 287 flowi4_update_output(fl4, sk->sk_bound_dev_if, 288 RT_CONN_FLAGS(sk), fl4->daddr, 289 fl4->saddr); 290 security_sk_classify_flow(sk, flowi4_to_flowi(fl4)); 291 return ip_route_output_flow(sock_net(sk), fl4, sk); 292 } 293 return rt; 294 } 295 296 static inline int inet_iif(const struct sk_buff *skb) 297 { 298 int iif = skb_rtable(skb)->rt_iif; 299 300 if (iif) 301 return iif; 302 return skb->skb_iif; 303 } 304 305 extern int sysctl_ip_default_ttl; 306 307 static inline int ip4_dst_hoplimit(const struct dst_entry *dst) 308 { 309 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); 310 311 if (hoplimit == 0) 312 hoplimit = sysctl_ip_default_ttl; 313 return hoplimit; 314 } 315 316 static inline bool ip_sk_accept_pmtu(const struct sock *sk) 317 { 318 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE; 319 } 320 321 static inline bool ip_sk_use_pmtu(const struct sock *sk) 322 { 323 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE; 324 } 325 326 static inline int ip_skb_dst_mtu(const struct sk_buff *skb) 327 { 328 return (!skb->sk || ip_sk_use_pmtu(skb->sk)) ? 329 dst_mtu(skb_dst(skb)) : skb_dst(skb)->dev->mtu; 330 } 331 332 #endif /* _ROUTE_H */ 333