Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
[linux-2.6-block.git] / include / net / route.h
CommitLineData
2874c5fd 1/* SPDX-License-Identifier: GPL-2.0-or-later */
1da177e4
LT
2/*
3 * INET An implementation of the TCP/IP protocol suite for the LINUX
4 * operating system. INET is implemented using the BSD Socket
5 * interface as the means of communication with the user level.
6 *
7 * Definitions for the IP router.
8 *
9 * Version: @(#)route.h 1.0.4 05/27/93
10 *
02c30a84 11 * Authors: Ross Biro
1da177e4
LT
12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
13 * Fixes:
14 * Alan Cox : Reformatted. Added ip_rt_local()
15 * Alan Cox : Support for TCP parameters.
16 * Alexey Kuznetsov: Major changes for new routing code.
17 * Mike McLagan : Routing by source
18 * Robert Olsson : Added rt_cache statistics
1da177e4
LT
19 */
20#ifndef _ROUTE_H
21#define _ROUTE_H
22
1da177e4
LT
23#include <net/dst.h>
24#include <net/inetpeer.h>
25#include <net/flow.h>
79876874 26#include <net/inet_sock.h>
79a13159 27#include <net/ip_fib.h>
5c9f7c1d
DA
28#include <net/arp.h>
29#include <net/ndisc.h>
1da177e4
LT
30#include <linux/in_route.h>
31#include <linux/rtnetlink.h>
c6cffba4 32#include <linux/rcupdate.h>
1da177e4
LT
33#include <linux/route.h>
34#include <linux/ip.h>
35#include <linux/cache.h>
beb8d13b 36#include <linux/security.h>
1da177e4 37
f87c10a8
HFS
38/* IPv4 datagram length is stored into 16bit field (tot_len) */
39#define IP_MAX_MTU 0xFFFFU
40
1da177e4
LT
41#define RTO_ONLINK 0x01
42
1da177e4 43#define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
aa661581 44#define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
1da177e4
LT
45
46struct fib_nh;
62fa8a84 47struct fib_info;
5055c371 48struct uncached_list;
fd2c3ef7 49struct rtable {
d8d1f30b 50 struct dst_entry dst;
1da177e4 51
29e75252 52 int rt_genid;
95c96174 53 unsigned int rt_flags;
1da177e4 54 __u16 rt_type;
155e8336 55 __u8 rt_is_input;
77d5bc7e 56 __u8 rt_uses_gateway;
1da177e4 57
1da177e4
LT
58 int rt_iif;
59
77d5bc7e 60 u8 rt_gw_family;
1da177e4 61 /* Info on neighbour */
0f5f7d7b
DA
62 union {
63 __be32 rt_gw4;
64 struct in6_addr rt_gw6;
65 };
1da177e4 66
1da177e4 67 /* Miscellaneous cached information */
d52e5a7e
SD
68 u32 rt_mtu_locked:1,
69 rt_pmtu:31;
caacf05e
DM
70
71 struct list_head rt_uncached;
5055c371 72 struct uncached_list *rt_uncached_list;
1da177e4
LT
73};
74
b8400f37 75static inline bool rt_is_input_route(const struct rtable *rt)
c7537967 76{
9917e1e8 77 return rt->rt_is_input != 0;
c7537967
DM
78}
79
b8400f37 80static inline bool rt_is_output_route(const struct rtable *rt)
c7537967 81{
9917e1e8 82 return rt->rt_is_input == 0;
c7537967
DM
83}
84
f8126f1d
DM
85static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
86{
1550c171
DA
87 if (rt->rt_gw_family == AF_INET)
88 return rt->rt_gw4;
f8126f1d
DM
89 return daddr;
90}
91
fd2c3ef7 92struct ip_rt_acct {
1da177e4
LT
93 __u32 o_bytes;
94 __u32 o_packets;
95 __u32 i_bytes;
96 __u32 i_packets;
97};
98
fd2c3ef7 99struct rt_cache_stat {
1da177e4
LT
100 unsigned int in_slow_tot;
101 unsigned int in_slow_mc;
102 unsigned int in_no_route;
103 unsigned int in_brd;
104 unsigned int in_martian_dst;
105 unsigned int in_martian_src;
1da177e4
LT
106 unsigned int out_slow_tot;
107 unsigned int out_slow_mc;
1da177e4
LT
108};
109
7d720c3e 110extern struct ip_rt_acct __percpu *ip_rt_acct;
1da177e4
LT
111
112struct in_device;
2702c4bb
JP
113
114int ip_rt_init(void);
115void rt_cache_flush(struct net *net);
116void rt_flush_dev(struct net_device *dev);
3abd1ade
DA
117struct rtable *ip_route_output_key_hash(struct net *net, struct flowi4 *flp,
118 const struct sk_buff *skb);
119struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *flp,
120 struct fib_result *res,
121 const struct sk_buff *skb);
79a13159
PN
122
123static inline struct rtable *__ip_route_output_key(struct net *net,
124 struct flowi4 *flp)
125{
3abd1ade 126 return ip_route_output_key_hash(net, flp, NULL);
79a13159
PN
127}
128
2702c4bb 129struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
6f9c9615 130 const struct sock *sk);
2702c4bb
JP
131struct dst_entry *ipv4_blackhole_route(struct net *net,
132 struct dst_entry *dst_orig);
407eadd9 133
9d6ec938 134static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
5bfa787f
DM
135{
136 return ip_route_output_flow(net, flp, NULL);
137}
138
78fbfd8a
DM
139static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
140 __be32 saddr, u8 tos, int oif)
141{
9d6ec938
DM
142 struct flowi4 fl4 = {
143 .flowi4_oif = oif,
c5d21c4b 144 .flowi4_tos = tos,
9d6ec938
DM
145 .daddr = daddr,
146 .saddr = saddr,
78fbfd8a 147 };
9d6ec938 148 return ip_route_output_key(net, &fl4);
78fbfd8a
DM
149}
150
31e4543d
DM
151static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
152 struct sock *sk,
78fbfd8a
DM
153 __be32 daddr, __be32 saddr,
154 __be16 dport, __be16 sport,
155 __u8 proto, __u8 tos, int oif)
156{
31e4543d 157 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
94b92b88
DM
158 RT_SCOPE_UNIVERSE, proto,
159 sk ? inet_sk_flowi_flags(sk) : 0,
e2d118a1 160 daddr, saddr, dport, sport, sock_net_uid(net, sk));
78fbfd8a 161 if (sk)
31e4543d
DM
162 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
163 return ip_route_output_flow(net, fl4, sk);
78fbfd8a
DM
164}
165
cbb1e85f 166static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
78fbfd8a
DM
167 __be32 daddr, __be32 saddr,
168 __be32 gre_key, __u8 tos, int oif)
169{
cbb1e85f
DM
170 memset(fl4, 0, sizeof(*fl4));
171 fl4->flowi4_oif = oif;
172 fl4->daddr = daddr;
173 fl4->saddr = saddr;
174 fl4->flowi4_tos = tos;
175 fl4->flowi4_proto = IPPROTO_GRE;
176 fl4->fl4_gre_key = gre_key;
177 return ip_route_output_key(net, fl4);
78fbfd8a 178}
bc044e8d
PA
179int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr,
180 u8 tos, struct net_device *dev,
181 struct in_device *in_dev, u32 *itag);
2702c4bb
JP
182int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
183 u8 tos, struct net_device *devin);
5510cdf7
DA
184int ip_route_input_rcu(struct sk_buff *skb, __be32 dst, __be32 src,
185 u8 tos, struct net_device *devin,
186 struct fib_result *res);
c6cffba4 187
02b24941
PA
188int ip_route_use_hint(struct sk_buff *skb, __be32 dst, __be32 src,
189 u8 tos, struct net_device *devin,
190 const struct sk_buff *hint);
191
c6cffba4
DM
192static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
193 u8 tos, struct net_device *devin)
194{
195 int err;
196
197 rcu_read_lock();
198 err = ip_route_input_noref(skb, dst, src, tos, devin);
64327fc8 199 if (!err) {
222d7dbd 200 skb_dst_force(skb);
64327fc8
SB
201 if (!skb_dst(skb))
202 err = -EINVAL;
203 }
c6cffba4
DM
204 rcu_read_unlock();
205
206 return err;
207}
407eadd9 208
2702c4bb 209void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
d888f396 210 u8 protocol);
2702c4bb 211void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
1042caa7 212void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u8 protocol);
2702c4bb
JP
213void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
214void ip_rt_send_redirect(struct sk_buff *skb);
215
216unsigned int inet_addr_type(struct net *net, __be32 addr);
9b8ff518 217unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
2702c4bb
JP
218unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
219 __be32 addr);
30bbaa19
DA
220unsigned int inet_addr_type_dev_table(struct net *net,
221 const struct net_device *dev,
222 __be32 addr);
2702c4bb 223void ip_rt_multicast_event(struct in_device *);
ca25c300 224int ip_rt_ioctl(struct net *, unsigned int cmd, struct rtentry *rt);
2702c4bb 225void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
9ab179d8
DA
226struct rtable *rt_dst_alloc(struct net_device *dev,
227 unsigned int flags, u16 type,
228 bool nopolicy, bool noxfrm, bool will_cache);
5b18f128 229struct rtable *rt_dst_clone(struct net_device *dev, struct rtable *rt);
1da177e4 230
0ff60a45 231struct in_ifaddr;
2702c4bb
JP
232void fib_add_ifaddr(struct in_ifaddr *);
233void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
af4d768a 234void fib_modify_prefix_metric(struct in_ifaddr *ifa, u32 new_metric);
0ff60a45 235
510c321b
XL
236void rt_add_uncached_list(struct rtable *rt);
237void rt_del_uncached_list(struct rtable *rt);
238
ee28906f
SB
239int fib_dump_info_fnhe(struct sk_buff *skb, struct netlink_callback *cb,
240 u32 table_id, struct fib_info *fi,
e93fb3e9 241 int *fa_index, int fa_start, unsigned int flags);
ee28906f 242
6da025fa 243static inline void ip_rt_put(struct rtable *rt)
1da177e4 244{
6da025fa
ED
245 /* dst_release() accepts a NULL parameter.
246 * We rely on dst being first structure in struct rtable
247 */
248 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
249 dst_release(&rt->dst);
1da177e4
LT
250}
251
252#define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
253
4839c52b 254extern const __u8 ip_tos2prio[16];
1da177e4
LT
255
256static inline char rt_tos2priority(u8 tos)
257{
258 return ip_tos2prio[IPTOS_TOS(tos)>>1];
259}
260
2d7192d6
DM
261/* ip_route_connect() and ip_route_newports() work in tandem whilst
262 * binding a socket for a new outgoing connection.
263 *
264 * In order to use IPSEC properly, we must, in the end, have a
265 * route that was looked up using all available keys including source
266 * and destination ports.
267 *
268 * However, if a source port needs to be allocated (the user specified
269 * a wildcard source port) we need to obtain addressing information
270 * in order to perform that allocation.
271 *
272 * So ip_route_connect() looks up a route using wildcarded source and
273 * destination ports in the key, simply so that we can get a pair of
274 * addresses to use for port allocation.
275 *
276 * Later, once the ports are allocated, ip_route_newports() will make
277 * another route lookup if needed to make sure we catch any IPSEC
278 * rules keyed on the port information.
279 *
280 * The callers allocate the flow key on their stack, and must pass in
281 * the same flowi4 object to both the ip_route_connect() and the
282 * ip_route_newports() calls.
283 */
284
285static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
286 u32 tos, int oif, u8 protocol,
287 __be16 sport, __be16 dport,
0e0d44ab 288 struct sock *sk)
1da177e4 289{
2d7192d6 290 __u8 flow_flags = 0;
79876874
KK
291
292 if (inet_sk(sk)->transparent)
94b92b88 293 flow_flags |= FLOWI_FLAG_ANYSRC;
94b92b88 294
2d7192d6 295 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
e2d118a1
LC
296 protocol, flow_flags, dst, src, dport, sport,
297 sk->sk_uid);
2d7192d6
DM
298}
299
300static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
301 __be32 dst, __be32 src, u32 tos,
302 int oif, u8 protocol,
303 __be16 sport, __be16 dport,
0e0d44ab 304 struct sock *sk)
2d7192d6
DM
305{
306 struct net *net = sock_net(sk);
307 struct rtable *rt;
308
309 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
0e0d44ab 310 sport, dport, sk);
79876874 311
1da177e4 312 if (!dst || !src) {
2d7192d6 313 rt = __ip_route_output_key(net, fl4);
b23dd4fe
DM
314 if (IS_ERR(rt))
315 return rt;
b23dd4fe 316 ip_rt_put(rt);
e6b45241 317 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
1da177e4 318 }
2d7192d6
DM
319 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
320 return ip_route_output_flow(net, fl4, sk);
1da177e4
LT
321}
322
2d7192d6
DM
323static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
324 __be16 orig_sport, __be16 orig_dport,
325 __be16 sport, __be16 dport,
326 struct sock *sk)
1da177e4 327{
dca8b089 328 if (sport != orig_sport || dport != orig_dport) {
2d7192d6
DM
329 fl4->fl4_dport = dport;
330 fl4->fl4_sport = sport;
b23dd4fe 331 ip_rt_put(rt);
e6b45241
JA
332 flowi4_update_output(fl4, sk->sk_bound_dev_if,
333 RT_CONN_FLAGS(sk), fl4->daddr,
334 fl4->saddr);
2d7192d6
DM
335 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
336 return ip_route_output_flow(sock_net(sk), fl4, sk);
1da177e4 337 }
b23dd4fe 338 return rt;
1da177e4
LT
339}
340
1668e010
KK
341static inline int inet_iif(const struct sk_buff *skb)
342{
0340d0b9
TH
343 struct rtable *rt = skb_rtable(skb);
344
345 if (rt && rt->rt_iif)
346 return rt->rt_iif;
13378cad 347
13378cad 348 return skb->skb_iif;
1668e010
KK
349}
350
323e126f
DM
351static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
352{
353 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
fa50d974 354 struct net *net = dev_net(dst->dev);
323e126f
DM
355
356 if (hoplimit == 0)
fa50d974 357 hoplimit = net->ipv4.sysctl_ip_default_ttl;
323e126f
DM
358 return hoplimit;
359}
360
5c9f7c1d
DA
361static inline struct neighbour *ip_neigh_gw4(struct net_device *dev,
362 __be32 daddr)
363{
364 struct neighbour *neigh;
365
366 neigh = __ipv4_neigh_lookup_noref(dev, daddr);
367 if (unlikely(!neigh))
368 neigh = __neigh_create(&arp_tbl, &daddr, dev, false);
369
370 return neigh;
371}
372
373static inline struct neighbour *ip_neigh_for_gw(struct rtable *rt,
374 struct sk_buff *skb,
375 bool *is_v6gw)
376{
377 struct net_device *dev = rt->dst.dev;
378 struct neighbour *neigh;
379
380 if (likely(rt->rt_gw_family == AF_INET)) {
381 neigh = ip_neigh_gw4(dev, rt->rt_gw4);
382 } else if (rt->rt_gw_family == AF_INET6) {
383 neigh = ip_neigh_gw6(dev, &rt->rt_gw6);
384 *is_v6gw = true;
385 } else {
386 neigh = ip_neigh_gw4(dev, ip_hdr(skb)->daddr);
387 }
388 return neigh;
389}
390
1da177e4 391#endif /* _ROUTE_H */