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 inet_peer;
44 struct fib_info;
45 struct rtable {
46 struct dst_entry dst;
47
48 /* Lookup key. */
49 __be32 rt_key_dst;
50 __be32 rt_key_src;
51
52 int rt_genid;
53 unsigned rt_flags;
54 __u16 rt_type;
55 __u8 rt_key_tos;
56
57 __be32 rt_dst; /* Path destination */
58 __be32 rt_src; /* Path source */
59 int rt_route_iif;
60 int rt_iif;
61 int rt_oif;
62 __u32 rt_mark;
63 uid_t rt_uid;
64
65 /* Info on neighbour */
66 __be32 rt_gateway;
67
68 /* Miscellaneous cached information */
69 __be32 rt_spec_dst; /* RFC1122 specific destination */
70 u32 rt_peer_genid;
71 struct inet_peer *peer; /* long-living peer info */
72 struct fib_info *fi; /* for client ref to shared metrics */
73 };
74
rt_is_input_route(const struct rtable * rt)75 static inline bool rt_is_input_route(const struct rtable *rt)
76 {
77 return rt->rt_route_iif != 0;
78 }
79
rt_is_output_route(const struct rtable * rt)80 static inline bool rt_is_output_route(const struct rtable *rt)
81 {
82 return rt->rt_route_iif == 0;
83 }
84
85 struct ip_rt_acct {
86 __u32 o_bytes;
87 __u32 o_packets;
88 __u32 i_bytes;
89 __u32 i_packets;
90 };
91
92 struct rt_cache_stat {
93 unsigned int in_hit;
94 unsigned int in_slow_tot;
95 unsigned int in_slow_mc;
96 unsigned int in_no_route;
97 unsigned int in_brd;
98 unsigned int in_martian_dst;
99 unsigned int in_martian_src;
100 unsigned int out_hit;
101 unsigned int out_slow_tot;
102 unsigned int out_slow_mc;
103 unsigned int gc_total;
104 unsigned int gc_ignored;
105 unsigned int gc_goal_miss;
106 unsigned int gc_dst_overflow;
107 unsigned int in_hlist_search;
108 unsigned int out_hlist_search;
109 };
110
111 extern struct ip_rt_acct __percpu *ip_rt_acct;
112
113 struct in_device;
114 extern int ip_rt_init(void);
115 extern void ip_rt_redirect(__be32 old_gw, __be32 dst, __be32 new_gw,
116 __be32 src, struct net_device *dev);
117 extern void rt_cache_flush(struct net *net, int how);
118 extern void rt_cache_flush_batch(struct net *net);
119 extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp);
120 extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
121 struct sock *sk);
122 extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig);
123
ip_route_output_key(struct net * net,struct flowi4 * flp)124 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
125 {
126 return ip_route_output_flow(net, flp, NULL);
127 }
128
ip_route_output(struct net * net,__be32 daddr,__be32 saddr,u8 tos,int oif)129 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
130 __be32 saddr, u8 tos, int oif)
131 {
132 struct flowi4 fl4 = {
133 .flowi4_oif = oif,
134 .daddr = daddr,
135 .saddr = saddr,
136 .flowi4_tos = tos,
137 };
138 return ip_route_output_key(net, &fl4);
139 }
140
ip_route_output_ports(struct net * net,struct flowi4 * fl4,struct sock * sk,__be32 daddr,__be32 saddr,__be16 dport,__be16 sport,__u8 proto,__u8 tos,int oif)141 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
142 struct sock *sk,
143 __be32 daddr, __be32 saddr,
144 __be16 dport, __be16 sport,
145 __u8 proto, __u8 tos, int oif)
146 {
147 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
148 RT_SCOPE_UNIVERSE, proto,
149 sk ? inet_sk_flowi_flags(sk) : 0,
150 daddr, saddr, dport, sport, sk ? sock_i_uid(sk) : 0);
151 if (sk)
152 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
153 return ip_route_output_flow(net, fl4, sk);
154 }
155
ip_route_output_gre(struct net * net,struct flowi4 * fl4,__be32 daddr,__be32 saddr,__be32 gre_key,__u8 tos,int oif)156 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
157 __be32 daddr, __be32 saddr,
158 __be32 gre_key, __u8 tos, int oif)
159 {
160 memset(fl4, 0, sizeof(*fl4));
161 fl4->flowi4_oif = oif;
162 fl4->daddr = daddr;
163 fl4->saddr = saddr;
164 fl4->flowi4_tos = tos;
165 fl4->flowi4_proto = IPPROTO_GRE;
166 fl4->fl4_gre_key = gre_key;
167 return ip_route_output_key(net, fl4);
168 }
169
170 extern int ip_route_input_common(struct sk_buff *skb, __be32 dst, __be32 src,
171 u8 tos, struct net_device *devin, bool noref);
172
ip_route_input(struct sk_buff * skb,__be32 dst,__be32 src,u8 tos,struct net_device * devin)173 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
174 u8 tos, struct net_device *devin)
175 {
176 return ip_route_input_common(skb, dst, src, tos, devin, false);
177 }
178
ip_route_input_noref(struct sk_buff * skb,__be32 dst,__be32 src,u8 tos,struct net_device * devin)179 static inline int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
180 u8 tos, struct net_device *devin)
181 {
182 return ip_route_input_common(skb, dst, src, tos, devin, true);
183 }
184
185 extern unsigned short ip_rt_frag_needed(struct net *net, const struct iphdr *iph,
186 unsigned short new_mtu, struct net_device *dev);
187 extern void ip_rt_send_redirect(struct sk_buff *skb);
188
189 extern unsigned inet_addr_type(struct net *net, __be32 addr);
190 extern unsigned inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr);
191 extern void ip_rt_multicast_event(struct in_device *);
192 extern int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
193 extern void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
194 extern int ip_rt_dump(struct sk_buff *skb, struct netlink_callback *cb);
195
196 struct in_ifaddr;
197 extern void fib_add_ifaddr(struct in_ifaddr *);
198 extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
199
ip_rt_put(struct rtable * rt)200 static inline void ip_rt_put(struct rtable * rt)
201 {
202 if (rt)
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
rt_tos2priority(u8 tos)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
ip_route_connect_init(struct flowi4 * fl4,__be32 dst,__be32 src,u32 tos,int oif,u8 protocol,__be16 sport,__be16 dport,struct sock * sk,bool can_sleep)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 (protocol == IPPROTO_TCP)
249 flow_flags |= FLOWI_FLAG_PRECOW_METRICS;
250 if (can_sleep)
251 flow_flags |= FLOWI_FLAG_CAN_SLEEP;
252
253 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
254 protocol, flow_flags, dst, src, dport, sport,
255 sock_i_uid(sk));
256 }
257
ip_route_connect(struct flowi4 * fl4,__be32 dst,__be32 src,u32 tos,int oif,u8 protocol,__be16 sport,__be16 dport,struct sock * sk,bool can_sleep)258 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
259 __be32 dst, __be32 src, u32 tos,
260 int oif, u8 protocol,
261 __be16 sport, __be16 dport,
262 struct sock *sk, bool can_sleep)
263 {
264 struct net *net = sock_net(sk);
265 struct rtable *rt;
266
267 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
268 sport, dport, sk, can_sleep);
269
270 if (!dst || !src) {
271 rt = __ip_route_output_key(net, fl4);
272 if (IS_ERR(rt))
273 return rt;
274 ip_rt_put(rt);
275 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
276 }
277 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
278 return ip_route_output_flow(net, fl4, sk);
279 }
280
ip_route_newports(struct flowi4 * fl4,struct rtable * rt,__be16 orig_sport,__be16 orig_dport,__be16 sport,__be16 dport,struct sock * sk)281 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
282 __be16 orig_sport, __be16 orig_dport,
283 __be16 sport, __be16 dport,
284 struct sock *sk)
285 {
286 if (sport != orig_sport || dport != orig_dport) {
287 fl4->fl4_dport = dport;
288 fl4->fl4_sport = sport;
289 ip_rt_put(rt);
290 flowi4_update_output(fl4, sk->sk_bound_dev_if,
291 RT_CONN_FLAGS(sk), fl4->daddr,
292 fl4->saddr);
293 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
294 return ip_route_output_flow(sock_net(sk), fl4, sk);
295 }
296 return rt;
297 }
298
299 extern void rt_bind_peer(struct rtable *rt, __be32 daddr, int create);
300
rt_get_peer(struct rtable * rt,__be32 daddr)301 static inline struct inet_peer *rt_get_peer(struct rtable *rt, __be32 daddr)
302 {
303 if (rt->peer)
304 return rt->peer;
305
306 rt_bind_peer(rt, daddr, 0);
307 return rt->peer;
308 }
309
inet_iif(const struct sk_buff * skb)310 static inline int inet_iif(const struct sk_buff *skb)
311 {
312 return skb_rtable(skb)->rt_iif;
313 }
314
315 extern int sysctl_ip_default_ttl;
316
ip4_dst_hoplimit(const struct dst_entry * dst)317 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
318 {
319 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
320
321 if (hoplimit == 0)
322 hoplimit = sysctl_ip_default_ttl;
323 return hoplimit;
324 }
325
326 #endif /* _ROUTE_H */
327