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