]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blame - include/net/route.h
Merge tag 'dmaengine-5.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul...
[mirror_ubuntu-jammy-kernel.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);
571912c6
MV
131struct rtable *ip_route_output_tunnel(struct sk_buff *skb,
132 struct net_device *dev,
133 struct net *net, __be32 *saddr,
134 const struct ip_tunnel_info *info,
135 u8 protocol, bool use_cache);
136
2702c4bb
JP
137struct dst_entry *ipv4_blackhole_route(struct net *net,
138 struct dst_entry *dst_orig);
407eadd9 139
9d6ec938 140static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
5bfa787f
DM
141{
142 return ip_route_output_flow(net, flp, NULL);
143}
144
78fbfd8a
DM
145static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
146 __be32 saddr, u8 tos, int oif)
147{
9d6ec938
DM
148 struct flowi4 fl4 = {
149 .flowi4_oif = oif,
c5d21c4b 150 .flowi4_tos = tos,
9d6ec938
DM
151 .daddr = daddr,
152 .saddr = saddr,
78fbfd8a 153 };
9d6ec938 154 return ip_route_output_key(net, &fl4);
78fbfd8a
DM
155}
156
31e4543d
DM
157static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
158 struct sock *sk,
78fbfd8a
DM
159 __be32 daddr, __be32 saddr,
160 __be16 dport, __be16 sport,
161 __u8 proto, __u8 tos, int oif)
162{
31e4543d 163 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
94b92b88
DM
164 RT_SCOPE_UNIVERSE, proto,
165 sk ? inet_sk_flowi_flags(sk) : 0,
e2d118a1 166 daddr, saddr, dport, sport, sock_net_uid(net, sk));
78fbfd8a 167 if (sk)
3df98d79 168 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
31e4543d 169 return ip_route_output_flow(net, fl4, sk);
78fbfd8a
DM
170}
171
cbb1e85f 172static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
78fbfd8a
DM
173 __be32 daddr, __be32 saddr,
174 __be32 gre_key, __u8 tos, int oif)
175{
cbb1e85f
DM
176 memset(fl4, 0, sizeof(*fl4));
177 fl4->flowi4_oif = oif;
178 fl4->daddr = daddr;
179 fl4->saddr = saddr;
180 fl4->flowi4_tos = tos;
181 fl4->flowi4_proto = IPPROTO_GRE;
182 fl4->fl4_gre_key = gre_key;
183 return ip_route_output_key(net, fl4);
78fbfd8a 184}
bc044e8d
PA
185int ip_mc_validate_source(struct sk_buff *skb, __be32 daddr, __be32 saddr,
186 u8 tos, struct net_device *dev,
187 struct in_device *in_dev, u32 *itag);
2702c4bb
JP
188int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
189 u8 tos, struct net_device *devin);
5510cdf7
DA
190int ip_route_input_rcu(struct sk_buff *skb, __be32 dst, __be32 src,
191 u8 tos, struct net_device *devin,
192 struct fib_result *res);
c6cffba4 193
02b24941
PA
194int ip_route_use_hint(struct sk_buff *skb, __be32 dst, __be32 src,
195 u8 tos, struct net_device *devin,
196 const struct sk_buff *hint);
197
c6cffba4
DM
198static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
199 u8 tos, struct net_device *devin)
200{
201 int err;
202
203 rcu_read_lock();
204 err = ip_route_input_noref(skb, dst, src, tos, devin);
64327fc8 205 if (!err) {
222d7dbd 206 skb_dst_force(skb);
64327fc8
SB
207 if (!skb_dst(skb))
208 err = -EINVAL;
209 }
c6cffba4
DM
210 rcu_read_unlock();
211
212 return err;
213}
407eadd9 214
2702c4bb 215void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
d888f396 216 u8 protocol);
2702c4bb 217void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
1042caa7 218void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u8 protocol);
2702c4bb
JP
219void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
220void ip_rt_send_redirect(struct sk_buff *skb);
221
222unsigned int inet_addr_type(struct net *net, __be32 addr);
9b8ff518 223unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
2702c4bb
JP
224unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
225 __be32 addr);
30bbaa19
DA
226unsigned int inet_addr_type_dev_table(struct net *net,
227 const struct net_device *dev,
228 __be32 addr);
2702c4bb 229void ip_rt_multicast_event(struct in_device *);
ca25c300 230int ip_rt_ioctl(struct net *, unsigned int cmd, struct rtentry *rt);
2702c4bb 231void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
9ab179d8
DA
232struct rtable *rt_dst_alloc(struct net_device *dev,
233 unsigned int flags, u16 type,
af13b3c3 234 bool nopolicy, bool noxfrm);
5b18f128 235struct rtable *rt_dst_clone(struct net_device *dev, struct rtable *rt);
1da177e4 236
0ff60a45 237struct in_ifaddr;
2702c4bb
JP
238void fib_add_ifaddr(struct in_ifaddr *);
239void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
af4d768a 240void fib_modify_prefix_metric(struct in_ifaddr *ifa, u32 new_metric);
0ff60a45 241
510c321b
XL
242void rt_add_uncached_list(struct rtable *rt);
243void rt_del_uncached_list(struct rtable *rt);
244
ee28906f
SB
245int fib_dump_info_fnhe(struct sk_buff *skb, struct netlink_callback *cb,
246 u32 table_id, struct fib_info *fi,
e93fb3e9 247 int *fa_index, int fa_start, unsigned int flags);
ee28906f 248
6da025fa 249static inline void ip_rt_put(struct rtable *rt)
1da177e4 250{
6da025fa
ED
251 /* dst_release() accepts a NULL parameter.
252 * We rely on dst being first structure in struct rtable
253 */
254 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
255 dst_release(&rt->dst);
1da177e4
LT
256}
257
258#define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
259
4839c52b 260extern const __u8 ip_tos2prio[16];
1da177e4
LT
261
262static inline char rt_tos2priority(u8 tos)
263{
264 return ip_tos2prio[IPTOS_TOS(tos)>>1];
265}
266
2d7192d6
DM
267/* ip_route_connect() and ip_route_newports() work in tandem whilst
268 * binding a socket for a new outgoing connection.
269 *
270 * In order to use IPSEC properly, we must, in the end, have a
271 * route that was looked up using all available keys including source
272 * and destination ports.
273 *
274 * However, if a source port needs to be allocated (the user specified
275 * a wildcard source port) we need to obtain addressing information
276 * in order to perform that allocation.
277 *
278 * So ip_route_connect() looks up a route using wildcarded source and
279 * destination ports in the key, simply so that we can get a pair of
280 * addresses to use for port allocation.
281 *
282 * Later, once the ports are allocated, ip_route_newports() will make
283 * another route lookup if needed to make sure we catch any IPSEC
284 * rules keyed on the port information.
285 *
286 * The callers allocate the flow key on their stack, and must pass in
287 * the same flowi4 object to both the ip_route_connect() and the
288 * ip_route_newports() calls.
289 */
290
291static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
292 u32 tos, int oif, u8 protocol,
293 __be16 sport, __be16 dport,
0e0d44ab 294 struct sock *sk)
1da177e4 295{
2d7192d6 296 __u8 flow_flags = 0;
79876874
KK
297
298 if (inet_sk(sk)->transparent)
94b92b88 299 flow_flags |= FLOWI_FLAG_ANYSRC;
94b92b88 300
2d7192d6 301 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
e2d118a1
LC
302 protocol, flow_flags, dst, src, dport, sport,
303 sk->sk_uid);
2d7192d6
DM
304}
305
306static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
307 __be32 dst, __be32 src, u32 tos,
308 int oif, u8 protocol,
309 __be16 sport, __be16 dport,
0e0d44ab 310 struct sock *sk)
2d7192d6
DM
311{
312 struct net *net = sock_net(sk);
313 struct rtable *rt;
314
315 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
0e0d44ab 316 sport, dport, sk);
79876874 317
1da177e4 318 if (!dst || !src) {
2d7192d6 319 rt = __ip_route_output_key(net, fl4);
b23dd4fe
DM
320 if (IS_ERR(rt))
321 return rt;
b23dd4fe 322 ip_rt_put(rt);
e6b45241 323 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
1da177e4 324 }
3df98d79 325 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
2d7192d6 326 return ip_route_output_flow(net, fl4, sk);
1da177e4
LT
327}
328
2d7192d6
DM
329static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
330 __be16 orig_sport, __be16 orig_dport,
331 __be16 sport, __be16 dport,
332 struct sock *sk)
1da177e4 333{
dca8b089 334 if (sport != orig_sport || dport != orig_dport) {
2d7192d6
DM
335 fl4->fl4_dport = dport;
336 fl4->fl4_sport = sport;
b23dd4fe 337 ip_rt_put(rt);
e6b45241
JA
338 flowi4_update_output(fl4, sk->sk_bound_dev_if,
339 RT_CONN_FLAGS(sk), fl4->daddr,
340 fl4->saddr);
3df98d79 341 security_sk_classify_flow(sk, flowi4_to_flowi_common(fl4));
2d7192d6 342 return ip_route_output_flow(sock_net(sk), fl4, sk);
1da177e4 343 }
b23dd4fe 344 return rt;
1da177e4
LT
345}
346
1668e010
KK
347static inline int inet_iif(const struct sk_buff *skb)
348{
0340d0b9
TH
349 struct rtable *rt = skb_rtable(skb);
350
351 if (rt && rt->rt_iif)
352 return rt->rt_iif;
13378cad 353
13378cad 354 return skb->skb_iif;
1668e010
KK
355}
356
323e126f
DM
357static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
358{
359 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
fa50d974 360 struct net *net = dev_net(dst->dev);
323e126f
DM
361
362 if (hoplimit == 0)
fa50d974 363 hoplimit = net->ipv4.sysctl_ip_default_ttl;
323e126f
DM
364 return hoplimit;
365}
366
5c9f7c1d
DA
367static inline struct neighbour *ip_neigh_gw4(struct net_device *dev,
368 __be32 daddr)
369{
370 struct neighbour *neigh;
371
372 neigh = __ipv4_neigh_lookup_noref(dev, daddr);
373 if (unlikely(!neigh))
374 neigh = __neigh_create(&arp_tbl, &daddr, dev, false);
375
376 return neigh;
377}
378
379static inline struct neighbour *ip_neigh_for_gw(struct rtable *rt,
380 struct sk_buff *skb,
381 bool *is_v6gw)
382{
383 struct net_device *dev = rt->dst.dev;
384 struct neighbour *neigh;
385
386 if (likely(rt->rt_gw_family == AF_INET)) {
387 neigh = ip_neigh_gw4(dev, rt->rt_gw4);
388 } else if (rt->rt_gw_family == AF_INET6) {
389 neigh = ip_neigh_gw6(dev, &rt->rt_gw6);
390 *is_v6gw = true;
391 } else {
392 neigh = ip_neigh_gw4(dev, ip_hdr(skb)->daddr);
393 }
394 return neigh;
395}
396
1da177e4 397#endif /* _ROUTE_H */