ipv4: reset rt_iif for recirculated mcast/bcast out pkts
[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;
1550c171 56 u8 rt_gw_family;
1da177e4 57
1da177e4
LT
58 int rt_iif;
59
60 /* Info on neighbour */
0f5f7d7b
DA
61 union {
62 __be32 rt_gw4;
63 struct in6_addr rt_gw6;
64 };
1da177e4 65
1da177e4 66 /* Miscellaneous cached information */
d52e5a7e
SD
67 u32 rt_mtu_locked:1,
68 rt_pmtu:31;
caacf05e
DM
69
70 struct list_head rt_uncached;
5055c371 71 struct uncached_list *rt_uncached_list;
1da177e4
LT
72};
73
b8400f37 74static inline bool rt_is_input_route(const struct rtable *rt)
c7537967 75{
9917e1e8 76 return rt->rt_is_input != 0;
c7537967
DM
77}
78
b8400f37 79static inline bool rt_is_output_route(const struct rtable *rt)
c7537967 80{
9917e1e8 81 return rt->rt_is_input == 0;
c7537967
DM
82}
83
f8126f1d
DM
84static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
85{
1550c171
DA
86 if (rt->rt_gw_family == AF_INET)
87 return rt->rt_gw4;
f8126f1d
DM
88 return daddr;
89}
90
fd2c3ef7 91struct ip_rt_acct {
1da177e4
LT
92 __u32 o_bytes;
93 __u32 o_packets;
94 __u32 i_bytes;
95 __u32 i_packets;
96};
97
fd2c3ef7 98struct rt_cache_stat {
1da177e4
LT
99 unsigned int in_slow_tot;
100 unsigned int in_slow_mc;
101 unsigned int in_no_route;
102 unsigned int in_brd;
103 unsigned int in_martian_dst;
104 unsigned int in_martian_src;
1da177e4
LT
105 unsigned int out_slow_tot;
106 unsigned int out_slow_mc;
1da177e4
LT
107};
108
7d720c3e 109extern struct ip_rt_acct __percpu *ip_rt_acct;
1da177e4
LT
110
111struct in_device;
2702c4bb
JP
112
113int ip_rt_init(void);
114void rt_cache_flush(struct net *net);
115void rt_flush_dev(struct net_device *dev);
3abd1ade
DA
116struct rtable *ip_route_output_key_hash(struct net *net, struct flowi4 *flp,
117 const struct sk_buff *skb);
118struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *flp,
119 struct fib_result *res,
120 const struct sk_buff *skb);
79a13159
PN
121
122static inline struct rtable *__ip_route_output_key(struct net *net,
123 struct flowi4 *flp)
124{
3abd1ade 125 return ip_route_output_key_hash(net, flp, NULL);
79a13159
PN
126}
127
2702c4bb 128struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
6f9c9615 129 const struct sock *sk);
2702c4bb
JP
130struct dst_entry *ipv4_blackhole_route(struct net *net,
131 struct dst_entry *dst_orig);
407eadd9 132
9d6ec938 133static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
5bfa787f
DM
134{
135 return ip_route_output_flow(net, flp, NULL);
136}
137
78fbfd8a
DM
138static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
139 __be32 saddr, u8 tos, int oif)
140{
9d6ec938
DM
141 struct flowi4 fl4 = {
142 .flowi4_oif = oif,
c5d21c4b 143 .flowi4_tos = tos,
9d6ec938
DM
144 .daddr = daddr,
145 .saddr = saddr,
78fbfd8a 146 };
9d6ec938 147 return ip_route_output_key(net, &fl4);
78fbfd8a
DM
148}
149
31e4543d
DM
150static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
151 struct sock *sk,
78fbfd8a
DM
152 __be32 daddr, __be32 saddr,
153 __be16 dport, __be16 sport,
154 __u8 proto, __u8 tos, int oif)
155{
31e4543d 156 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
94b92b88
DM
157 RT_SCOPE_UNIVERSE, proto,
158 sk ? inet_sk_flowi_flags(sk) : 0,
e2d118a1 159 daddr, saddr, dport, sport, sock_net_uid(net, sk));
78fbfd8a 160 if (sk)
31e4543d
DM
161 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
162 return ip_route_output_flow(net, fl4, sk);
78fbfd8a
DM
163}
164
cbb1e85f 165static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
78fbfd8a
DM
166 __be32 daddr, __be32 saddr,
167 __be32 gre_key, __u8 tos, int oif)
168{
cbb1e85f
DM
169 memset(fl4, 0, sizeof(*fl4));
170 fl4->flowi4_oif = oif;
171 fl4->daddr = daddr;
172 fl4->saddr = saddr;
173 fl4->flowi4_tos = tos;
174 fl4->flowi4_proto = IPPROTO_GRE;
175 fl4->fl4_gre_key = gre_key;
176 return ip_route_output_key(net, fl4);
78fbfd8a 177}
bc044e8d
PA
178int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr,
179 u8 tos, struct net_device *dev,
180 struct in_device *in_dev, u32 *itag);
2702c4bb
JP
181int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
182 u8 tos, struct net_device *devin);
5510cdf7
DA
183int ip_route_input_rcu(struct sk_buff *skb, __be32 dst, __be32 src,
184 u8 tos, struct net_device *devin,
185 struct fib_result *res);
c6cffba4
DM
186
187static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
188 u8 tos, struct net_device *devin)
189{
190 int err;
191
192 rcu_read_lock();
193 err = ip_route_input_noref(skb, dst, src, tos, devin);
64327fc8 194 if (!err) {
222d7dbd 195 skb_dst_force(skb);
64327fc8
SB
196 if (!skb_dst(skb))
197 err = -EINVAL;
198 }
c6cffba4
DM
199 rcu_read_unlock();
200
201 return err;
202}
407eadd9 203
2702c4bb 204void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
d888f396 205 u8 protocol);
2702c4bb 206void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
1042caa7 207void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u8 protocol);
2702c4bb
JP
208void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
209void ip_rt_send_redirect(struct sk_buff *skb);
210
211unsigned int inet_addr_type(struct net *net, __be32 addr);
9b8ff518 212unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
2702c4bb
JP
213unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
214 __be32 addr);
30bbaa19
DA
215unsigned int inet_addr_type_dev_table(struct net *net,
216 const struct net_device *dev,
217 __be32 addr);
2702c4bb 218void ip_rt_multicast_event(struct in_device *);
ca25c300 219int ip_rt_ioctl(struct net *, unsigned int cmd, struct rtentry *rt);
2702c4bb 220void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
9ab179d8
DA
221struct rtable *rt_dst_alloc(struct net_device *dev,
222 unsigned int flags, u16 type,
223 bool nopolicy, bool noxfrm, bool will_cache);
5b18f128 224struct rtable *rt_dst_clone(struct net_device *dev, struct rtable *rt);
1da177e4 225
0ff60a45 226struct in_ifaddr;
2702c4bb
JP
227void fib_add_ifaddr(struct in_ifaddr *);
228void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
af4d768a 229void fib_modify_prefix_metric(struct in_ifaddr *ifa, u32 new_metric);
0ff60a45 230
510c321b
XL
231void rt_add_uncached_list(struct rtable *rt);
232void rt_del_uncached_list(struct rtable *rt);
233
6da025fa 234static inline void ip_rt_put(struct rtable *rt)
1da177e4 235{
6da025fa
ED
236 /* dst_release() accepts a NULL parameter.
237 * We rely on dst being first structure in struct rtable
238 */
239 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
240 dst_release(&rt->dst);
1da177e4
LT
241}
242
243#define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
244
4839c52b 245extern const __u8 ip_tos2prio[16];
1da177e4
LT
246
247static inline char rt_tos2priority(u8 tos)
248{
249 return ip_tos2prio[IPTOS_TOS(tos)>>1];
250}
251
2d7192d6
DM
252/* ip_route_connect() and ip_route_newports() work in tandem whilst
253 * binding a socket for a new outgoing connection.
254 *
255 * In order to use IPSEC properly, we must, in the end, have a
256 * route that was looked up using all available keys including source
257 * and destination ports.
258 *
259 * However, if a source port needs to be allocated (the user specified
260 * a wildcard source port) we need to obtain addressing information
261 * in order to perform that allocation.
262 *
263 * So ip_route_connect() looks up a route using wildcarded source and
264 * destination ports in the key, simply so that we can get a pair of
265 * addresses to use for port allocation.
266 *
267 * Later, once the ports are allocated, ip_route_newports() will make
268 * another route lookup if needed to make sure we catch any IPSEC
269 * rules keyed on the port information.
270 *
271 * The callers allocate the flow key on their stack, and must pass in
272 * the same flowi4 object to both the ip_route_connect() and the
273 * ip_route_newports() calls.
274 */
275
276static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
277 u32 tos, int oif, u8 protocol,
278 __be16 sport, __be16 dport,
0e0d44ab 279 struct sock *sk)
1da177e4 280{
2d7192d6 281 __u8 flow_flags = 0;
79876874
KK
282
283 if (inet_sk(sk)->transparent)
94b92b88 284 flow_flags |= FLOWI_FLAG_ANYSRC;
94b92b88 285
2d7192d6 286 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
e2d118a1
LC
287 protocol, flow_flags, dst, src, dport, sport,
288 sk->sk_uid);
2d7192d6
DM
289}
290
291static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
292 __be32 dst, __be32 src, u32 tos,
293 int oif, u8 protocol,
294 __be16 sport, __be16 dport,
0e0d44ab 295 struct sock *sk)
2d7192d6
DM
296{
297 struct net *net = sock_net(sk);
298 struct rtable *rt;
299
300 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
0e0d44ab 301 sport, dport, sk);
79876874 302
1da177e4 303 if (!dst || !src) {
2d7192d6 304 rt = __ip_route_output_key(net, fl4);
b23dd4fe
DM
305 if (IS_ERR(rt))
306 return rt;
b23dd4fe 307 ip_rt_put(rt);
e6b45241 308 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
1da177e4 309 }
2d7192d6
DM
310 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
311 return ip_route_output_flow(net, fl4, sk);
1da177e4
LT
312}
313
2d7192d6
DM
314static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
315 __be16 orig_sport, __be16 orig_dport,
316 __be16 sport, __be16 dport,
317 struct sock *sk)
1da177e4 318{
dca8b089 319 if (sport != orig_sport || dport != orig_dport) {
2d7192d6
DM
320 fl4->fl4_dport = dport;
321 fl4->fl4_sport = sport;
b23dd4fe 322 ip_rt_put(rt);
e6b45241
JA
323 flowi4_update_output(fl4, sk->sk_bound_dev_if,
324 RT_CONN_FLAGS(sk), fl4->daddr,
325 fl4->saddr);
2d7192d6
DM
326 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
327 return ip_route_output_flow(sock_net(sk), fl4, sk);
1da177e4 328 }
b23dd4fe 329 return rt;
1da177e4
LT
330}
331
1668e010
KK
332static inline int inet_iif(const struct sk_buff *skb)
333{
0340d0b9
TH
334 struct rtable *rt = skb_rtable(skb);
335
336 if (rt && rt->rt_iif)
337 return rt->rt_iif;
13378cad 338
13378cad 339 return skb->skb_iif;
1668e010
KK
340}
341
323e126f
DM
342static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
343{
344 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
fa50d974 345 struct net *net = dev_net(dst->dev);
323e126f
DM
346
347 if (hoplimit == 0)
fa50d974 348 hoplimit = net->ipv4.sysctl_ip_default_ttl;
323e126f
DM
349 return hoplimit;
350}
351
5c9f7c1d
DA
352static inline struct neighbour *ip_neigh_gw4(struct net_device *dev,
353 __be32 daddr)
354{
355 struct neighbour *neigh;
356
357 neigh = __ipv4_neigh_lookup_noref(dev, daddr);
358 if (unlikely(!neigh))
359 neigh = __neigh_create(&arp_tbl, &daddr, dev, false);
360
361 return neigh;
362}
363
364static inline struct neighbour *ip_neigh_for_gw(struct rtable *rt,
365 struct sk_buff *skb,
366 bool *is_v6gw)
367{
368 struct net_device *dev = rt->dst.dev;
369 struct neighbour *neigh;
370
371 if (likely(rt->rt_gw_family == AF_INET)) {
372 neigh = ip_neigh_gw4(dev, rt->rt_gw4);
373 } else if (rt->rt_gw_family == AF_INET6) {
374 neigh = ip_neigh_gw6(dev, &rt->rt_gw6);
375 *is_v6gw = true;
376 } else {
377 neigh = ip_neigh_gw4(dev, ip_hdr(skb)->daddr);
378 }
379 return neigh;
380}
381
1da177e4 382#endif /* _ROUTE_H */