blob: b1c0d5b564c2c1936455086cbf37dbbc761de0dd [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
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 *
Jesper Juhl02c30a82005-05-05 16:16:16 -070010 * Authors: Ross Biro
Linus Torvalds1da177e2005-04-16 15:20:36 -070011 * 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
Linus Torvalds1da177e2005-04-16 15:20:36 -070027#include <net/dst.h>
28#include <net/inetpeer.h>
29#include <net/flow.h>
KOVACS Krisztian79876872008-10-01 07:35:39 -070030#include <net/inet_sock.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070031#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>
Venkat Yekkiralabeb8d132006-08-04 23:12:42 -070036#include <linux/security.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070037
Linus Torvalds1da177e2005-04-16 15:20:36 -070038#define RTO_ONLINK 0x01
39
Linus Torvalds1da177e2005-04-16 15:20:36 -070040#define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
41
42struct fib_nh;
43struct inet_peer;
David S. Miller62fa8a82011-01-26 20:51:05 -080044struct fib_info;
Eric Dumazetfd2c3ef2009-11-03 03:26:03 +000045struct rtable {
Changli Gaod8d1f302010-06-10 23:31:35 -070046 struct dst_entry dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -070047
David S. Miller5e2b61f2011-03-04 21:47:09 -080048 /* Lookup key. */
49 __be32 rt_key_dst;
50 __be32 rt_key_src;
Eric Dumazet093c2ca2007-02-09 16:19:26 -080051
Eric Dumazet29e75252008-01-31 17:05:09 -080052 int rt_genid;
Linus Torvalds1da177e2005-04-16 15:20:36 -070053 unsigned rt_flags;
54 __u16 rt_type;
David S. Miller475949d2011-05-03 19:45:15 -070055 __u8 rt_key_tos;
Linus Torvalds1da177e2005-04-16 15:20:36 -070056
Al Virof2c3fe22006-09-26 21:26:42 -070057 __be32 rt_dst; /* Path destination */
58 __be32 rt_src; /* Path source */
OGAWA Hirofumi1b86a582011-04-07 14:04:08 -070059 int rt_route_iif;
Linus Torvalds1da177e2005-04-16 15:20:36 -070060 int rt_iif;
David S. Miller5e2b61f2011-03-04 21:47:09 -080061 int rt_oif;
62 __u32 rt_mark;
Linus Torvalds1da177e2005-04-16 15:20:36 -070063
64 /* Info on neighbour */
Al Virof2c3fe22006-09-26 21:26:42 -070065 __be32 rt_gateway;
Linus Torvalds1da177e2005-04-16 15:20:36 -070066
Linus Torvalds1da177e2005-04-16 15:20:36 -070067 /* Miscellaneous cached information */
Al Virof2c3fe22006-09-26 21:26:42 -070068 __be32 rt_spec_dst; /* RFC1122 specific destination */
David S. Miller6431cbc2011-02-07 20:38:06 -080069 u32 rt_peer_genid;
Linus Torvalds1da177e2005-04-16 15:20:36 -070070 struct inet_peer *peer; /* long-living peer info */
David S. Miller62fa8a82011-01-26 20:51:05 -080071 struct fib_info *fi; /* for client ref to shared metrics */
Linus Torvalds1da177e2005-04-16 15:20:36 -070072};
73
Steffen Klassertb8400f32011-11-23 02:14:15 +000074static inline bool rt_is_input_route(const struct rtable *rt)
David S. Millerc7537962010-11-11 17:07:48 -080075{
OGAWA Hirofumi1b86a582011-04-07 14:04:08 -070076 return rt->rt_route_iif != 0;
David S. Millerc7537962010-11-11 17:07:48 -080077}
78
Steffen Klassertb8400f32011-11-23 02:14:15 +000079static inline bool rt_is_output_route(const struct rtable *rt)
David S. Millerc7537962010-11-11 17:07:48 -080080{
OGAWA Hirofumi1b86a582011-04-07 14:04:08 -070081 return rt->rt_route_iif == 0;
David S. Millerc7537962010-11-11 17:07:48 -080082}
83
Eric Dumazetfd2c3ef2009-11-03 03:26:03 +000084struct ip_rt_acct {
Linus Torvalds1da177e2005-04-16 15:20:36 -070085 __u32 o_bytes;
86 __u32 o_packets;
87 __u32 i_bytes;
88 __u32 i_packets;
89};
90
Eric Dumazetfd2c3ef2009-11-03 03:26:03 +000091struct rt_cache_stat {
Linus Torvalds1da177e2005-04-16 15:20:36 -070092 unsigned int in_hit;
93 unsigned int in_slow_tot;
94 unsigned int in_slow_mc;
95 unsigned int in_no_route;
96 unsigned int in_brd;
97 unsigned int in_martian_dst;
98 unsigned int in_martian_src;
99 unsigned int out_hit;
100 unsigned int out_slow_tot;
101 unsigned int out_slow_mc;
102 unsigned int gc_total;
103 unsigned int gc_ignored;
104 unsigned int gc_goal_miss;
105 unsigned int gc_dst_overflow;
106 unsigned int in_hlist_search;
107 unsigned int out_hlist_search;
108};
109
Tejun Heo7d720c32010-02-16 15:20:26 +0000110extern struct ip_rt_acct __percpu *ip_rt_acct;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700111
112struct in_device;
113extern int ip_rt_init(void);
Al Virof7655222006-09-26 21:25:43 -0700114extern void ip_rt_redirect(__be32 old_gw, __be32 dst, __be32 new_gw,
115 __be32 src, struct net_device *dev);
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700116extern void rt_cache_flush(struct net *net, int how);
David S. Miller6561a3b2010-12-19 21:11:20 -0800117extern void rt_cache_flush_batch(struct net *net);
David S. Miller813b3b52011-04-28 14:48:42 -0700118extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp);
David S. Miller9d6ec932011-03-12 01:12:47 -0500119extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
David S. Millerb23dd4f2011-03-02 14:31:35 -0800120 struct sock *sk);
David S. Miller2774c132011-03-01 14:59:04 -0800121extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig);
Eric Dumazet407eadd2010-05-10 11:32:55 +0000122
David S. Miller9d6ec932011-03-12 01:12:47 -0500123static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
David S. Miller5bfa7872011-03-02 14:56:30 -0800124{
125 return ip_route_output_flow(net, flp, NULL);
126}
127
David S. Miller78fbfd82011-03-12 00:00:52 -0500128static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
129 __be32 saddr, u8 tos, int oif)
130{
David S. Miller9d6ec932011-03-12 01:12:47 -0500131 struct flowi4 fl4 = {
132 .flowi4_oif = oif,
133 .daddr = daddr,
134 .saddr = saddr,
135 .flowi4_tos = tos,
David S. Miller78fbfd82011-03-12 00:00:52 -0500136 };
David S. Miller9d6ec932011-03-12 01:12:47 -0500137 return ip_route_output_key(net, &fl4);
David S. Miller78fbfd82011-03-12 00:00:52 -0500138}
139
David S. Miller31e45432011-05-03 20:25:42 -0700140static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
141 struct sock *sk,
David S. Miller78fbfd82011-03-12 00:00:52 -0500142 __be32 daddr, __be32 saddr,
143 __be16 dport, __be16 sport,
144 __u8 proto, __u8 tos, int oif)
145{
David S. Miller31e45432011-05-03 20:25:42 -0700146 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
David S. Miller94b92b82011-03-31 04:52:59 -0700147 RT_SCOPE_UNIVERSE, proto,
148 sk ? inet_sk_flowi_flags(sk) : 0,
149 daddr, saddr, dport, sport);
David S. Miller78fbfd82011-03-12 00:00:52 -0500150 if (sk)
David S. Miller31e45432011-05-03 20:25:42 -0700151 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
152 return ip_route_output_flow(net, fl4, sk);
David S. Miller78fbfd82011-03-12 00:00:52 -0500153}
154
David S. Millercbb1e852011-05-04 12:33:34 -0700155static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
David S. Miller78fbfd82011-03-12 00:00:52 -0500156 __be32 daddr, __be32 saddr,
157 __be32 gre_key, __u8 tos, int oif)
158{
David S. Millercbb1e852011-05-04 12:33:34 -0700159 memset(fl4, 0, sizeof(*fl4));
160 fl4->flowi4_oif = oif;
161 fl4->daddr = daddr;
162 fl4->saddr = saddr;
163 fl4->flowi4_tos = tos;
164 fl4->flowi4_proto = IPPROTO_GRE;
165 fl4->fl4_gre_key = gre_key;
166 return ip_route_output_key(net, fl4);
David S. Miller78fbfd82011-03-12 00:00:52 -0500167}
168
Eric Dumazet407eadd2010-05-10 11:32:55 +0000169extern int ip_route_input_common(struct sk_buff *skb, __be32 dst, __be32 src,
170 u8 tos, struct net_device *devin, bool noref);
171
172static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
173 u8 tos, struct net_device *devin)
174{
175 return ip_route_input_common(skb, dst, src, tos, devin, false);
176}
177
178static inline int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
179 u8 tos, struct net_device *devin)
180{
181 return ip_route_input_common(skb, dst, src, tos, devin, true);
182}
183
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000184extern unsigned short ip_rt_frag_needed(struct net *net, const struct iphdr *iph,
185 unsigned short new_mtu, struct net_device *dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700186extern void ip_rt_send_redirect(struct sk_buff *skb);
187
Eric W. Biederman6b175b22008-01-10 03:25:28 -0800188extern unsigned inet_addr_type(struct net *net, __be32 addr);
189extern unsigned inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700190extern void ip_rt_multicast_event(struct in_device *);
Denis V. Lunev1bad1182008-01-10 03:29:53 -0800191extern int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
David S. Miller8e363602011-05-13 17:29:41 -0400192extern void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700193extern int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb);
194
Jamal Hadi Salim0ff60a42005-11-22 14:47:37 -0800195struct in_ifaddr;
196extern void fib_add_ifaddr(struct in_ifaddr *);
Julian Anastasove6abbaa2011-03-19 12:13:49 +0000197extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
Jamal Hadi Salim0ff60a42005-11-22 14:47:37 -0800198
Linus Torvalds1da177e2005-04-16 15:20:36 -0700199static inline void ip_rt_put(struct rtable * rt)
200{
201 if (rt)
Changli Gaod8d1f302010-06-10 23:31:35 -0700202 dst_release(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700203}
204
205#define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
206
Philippe De Muyter4839c522007-07-09 15:32:57 -0700207extern const __u8 ip_tos2prio[16];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700208
209static inline char rt_tos2priority(u8 tos)
210{
211 return ip_tos2prio[IPTOS_TOS(tos)>>1];
212}
213
David S. Miller2d7192d2011-04-26 13:28:44 -0700214/* ip_route_connect() and ip_route_newports() work in tandem whilst
215 * binding a socket for a new outgoing connection.
216 *
217 * In order to use IPSEC properly, we must, in the end, have a
218 * route that was looked up using all available keys including source
219 * and destination ports.
220 *
221 * However, if a source port needs to be allocated (the user specified
222 * a wildcard source port) we need to obtain addressing information
223 * in order to perform that allocation.
224 *
225 * So ip_route_connect() looks up a route using wildcarded source and
226 * destination ports in the key, simply so that we can get a pair of
227 * addresses to use for port allocation.
228 *
229 * Later, once the ports are allocated, ip_route_newports() will make
230 * another route lookup if needed to make sure we catch any IPSEC
231 * rules keyed on the port information.
232 *
233 * The callers allocate the flow key on their stack, and must pass in
234 * the same flowi4 object to both the ip_route_connect() and the
235 * ip_route_newports() calls.
236 */
KOVACS Krisztian79876872008-10-01 07:35:39 -0700237
David S. Miller2d7192d2011-04-26 13:28:44 -0700238static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
239 u32 tos, int oif, u8 protocol,
240 __be16 sport, __be16 dport,
241 struct sock *sk, bool can_sleep)
242{
243 __u8 flow_flags = 0;
244
KOVACS Krisztian79876872008-10-01 07:35:39 -0700245 if (inet_sk(sk)->transparent)
David S. Miller94b92b82011-03-31 04:52:59 -0700246 flow_flags |= FLOWI_FLAG_ANYSRC;
David S. Millera4daad62011-01-27 22:01:53 -0800247 if (protocol == IPPROTO_TCP)
David S. Miller94b92b82011-03-31 04:52:59 -0700248 flow_flags |= FLOWI_FLAG_PRECOW_METRICS;
David S. Miller5df65e52011-03-01 14:22:19 -0800249 if (can_sleep)
David S. Miller94b92b82011-03-31 04:52:59 -0700250 flow_flags |= FLOWI_FLAG_CAN_SLEEP;
251
David S. Miller2d7192d2011-04-26 13:28:44 -0700252 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
David S. Miller94b92b82011-03-31 04:52:59 -0700253 protocol, flow_flags, dst, src, dport, sport);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700254}
255
David S. Miller2d7192d2011-04-26 13:28:44 -0700256static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
257 __be32 dst, __be32 src, u32 tos,
258 int oif, u8 protocol,
259 __be16 sport, __be16 dport,
260 struct sock *sk, bool can_sleep)
261{
262 struct net *net = sock_net(sk);
263 struct rtable *rt;
264
265 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
266 sport, dport, sk, can_sleep);
267
268 if (!dst || !src) {
269 rt = __ip_route_output_key(net, fl4);
270 if (IS_ERR(rt))
271 return rt;
David S. Miller2d7192d2011-04-26 13:28:44 -0700272 ip_rt_put(rt);
Julian Anastasove6b45242012-02-04 13:04:46 +0000273 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
David S. Miller2d7192d2011-04-26 13:28:44 -0700274 }
275 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
276 return ip_route_output_flow(net, fl4, sk);
277}
278
279static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
280 __be16 orig_sport, __be16 orig_dport,
281 __be16 sport, __be16 dport,
282 struct sock *sk)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700283{
David S. Millerdca8b082011-02-24 13:38:12 -0800284 if (sport != orig_sport || dport != orig_dport) {
David S. Miller2d7192d2011-04-26 13:28:44 -0700285 fl4->fl4_dport = dport;
286 fl4->fl4_sport = sport;
David S. Millerb23dd4f2011-03-02 14:31:35 -0800287 ip_rt_put(rt);
Julian Anastasove6b45242012-02-04 13:04:46 +0000288 flowi4_update_output(fl4, sk->sk_bound_dev_if,
289 RT_CONN_FLAGS(sk), fl4->daddr,
290 fl4->saddr);
David S. Miller2d7192d2011-04-26 13:28:44 -0700291 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
292 return ip_route_output_flow(sock_net(sk), fl4, sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700293 }
David S. Millerb23dd4f2011-03-02 14:31:35 -0800294 return rt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700295}
296
David S. Millera48eff12011-05-18 18:42:43 -0400297extern void rt_bind_peer(struct rtable *rt, __be32 daddr, int create);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700298
David S. Millered2361e2011-05-18 18:38:54 -0400299static inline struct inet_peer *rt_get_peer(struct rtable *rt, __be32 daddr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700300{
301 if (rt->peer)
302 return rt->peer;
303
David S. Millera48eff12011-05-18 18:42:43 -0400304 rt_bind_peer(rt, daddr, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700305 return rt->peer;
306}
307
KOVACS Krisztian1668e012008-10-01 07:33:10 -0700308static inline int inet_iif(const struct sk_buff *skb)
309{
Eric Dumazet511c3f92009-06-02 05:14:27 +0000310 return skb_rtable(skb)->rt_iif;
KOVACS Krisztian1668e012008-10-01 07:33:10 -0700311}
312
David S. Miller323e1262010-12-12 21:55:08 -0800313extern int sysctl_ip_default_ttl;
314
315static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
316{
317 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
318
319 if (hoplimit == 0)
320 hoplimit = sysctl_ip_default_ttl;
321 return hoplimit;
322}
323
Linus Torvalds1da177e2005-04-16 15:20:36 -0700324#endif /* _ROUTE_H */