xref: /linux/include/net/ip6_route.h (revision e6a901a00822659181c93c86d8bbc2a17779fddc)
1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _NET_IP6_ROUTE_H
3 #define _NET_IP6_ROUTE_H
4 
5 #include <net/addrconf.h>
6 #include <net/flow.h>
7 #include <net/ip6_fib.h>
8 #include <net/sock.h>
9 #include <net/lwtunnel.h>
10 #include <linux/ip.h>
11 #include <linux/ipv6.h>
12 #include <linux/route.h>
13 #include <net/nexthop.h>
14 
15 struct route_info {
16 	__u8			type;
17 	__u8			length;
18 	__u8			prefix_len;
19 #if defined(__BIG_ENDIAN_BITFIELD)
20 	__u8			reserved_h:3,
21 				route_pref:2,
22 				reserved_l:3;
23 #elif defined(__LITTLE_ENDIAN_BITFIELD)
24 	__u8			reserved_l:3,
25 				route_pref:2,
26 				reserved_h:3;
27 #endif
28 	__be32			lifetime;
29 	__u8			prefix[];	/* 0,8 or 16 */
30 };
31 
32 #define RT6_LOOKUP_F_IFACE		0x00000001
33 #define RT6_LOOKUP_F_REACHABLE		0x00000002
34 #define RT6_LOOKUP_F_HAS_SADDR		0x00000004
35 #define RT6_LOOKUP_F_SRCPREF_TMP	0x00000008
36 #define RT6_LOOKUP_F_SRCPREF_PUBLIC	0x00000010
37 #define RT6_LOOKUP_F_SRCPREF_COA	0x00000020
38 #define RT6_LOOKUP_F_IGNORE_LINKSTATE	0x00000040
39 #define RT6_LOOKUP_F_DST_NOREF		0x00000080
40 
41 /* We do not (yet ?) support IPv6 jumbograms (RFC 2675)
42  * Unlike IPv4, hdr->seg_len doesn't include the IPv6 header
43  */
44 #define IP6_MAX_MTU (0xFFFF + sizeof(struct ipv6hdr))
45 
46 /*
47  * rt6_srcprefs2flags() and rt6_flags2srcprefs() translate
48  * between IPV6_ADDR_PREFERENCES socket option values
49  *	IPV6_PREFER_SRC_TMP    = 0x1
50  *	IPV6_PREFER_SRC_PUBLIC = 0x2
51  *	IPV6_PREFER_SRC_COA    = 0x4
52  * and above RT6_LOOKUP_F_SRCPREF_xxx flags.
53  */
54 static inline int rt6_srcprefs2flags(unsigned int srcprefs)
55 {
56 	return (srcprefs & IPV6_PREFER_SRC_MASK) << 3;
57 }
58 
59 static inline unsigned int rt6_flags2srcprefs(int flags)
60 {
61 	return (flags >> 3) & IPV6_PREFER_SRC_MASK;
62 }
63 
64 static inline bool rt6_need_strict(const struct in6_addr *daddr)
65 {
66 	return ipv6_addr_type(daddr) &
67 		(IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK);
68 }
69 
70 /* fib entries using a nexthop object can not be coalesced into
71  * a multipath route
72  */
73 static inline bool rt6_qualify_for_ecmp(const struct fib6_info *f6i)
74 {
75 	/* the RTF_ADDRCONF flag filters out RA's */
76 	return !(f6i->fib6_flags & RTF_ADDRCONF) && !f6i->nh &&
77 		f6i->fib6_nh->fib_nh_gw_family;
78 }
79 
80 void ip6_route_input(struct sk_buff *skb);
81 struct dst_entry *ip6_route_input_lookup(struct net *net,
82 					 struct net_device *dev,
83 					 struct flowi6 *fl6,
84 					 const struct sk_buff *skb, int flags);
85 
86 struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
87 					 struct flowi6 *fl6, int flags);
88 
89 static inline struct dst_entry *ip6_route_output(struct net *net,
90 						 const struct sock *sk,
91 						 struct flowi6 *fl6)
92 {
93 	return ip6_route_output_flags(net, sk, fl6, 0);
94 }
95 
96 /* Only conditionally release dst if flags indicates
97  * !RT6_LOOKUP_F_DST_NOREF or dst is in uncached_list.
98  */
99 static inline void ip6_rt_put_flags(struct rt6_info *rt, int flags)
100 {
101 	if (!(flags & RT6_LOOKUP_F_DST_NOREF) ||
102 	    !list_empty(&rt->dst.rt_uncached))
103 		ip6_rt_put(rt);
104 }
105 
106 struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
107 				   const struct sk_buff *skb, int flags);
108 struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
109 			       int ifindex, struct flowi6 *fl6,
110 			       const struct sk_buff *skb, int flags);
111 
112 void ip6_route_init_special_entries(void);
113 int ip6_route_init(void);
114 void ip6_route_cleanup(void);
115 
116 int ipv6_route_ioctl(struct net *net, unsigned int cmd,
117 		struct in6_rtmsg *rtmsg);
118 
119 int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags,
120 		  struct netlink_ext_ack *extack);
121 int ip6_ins_rt(struct net *net, struct fib6_info *f6i);
122 int ip6_del_rt(struct net *net, struct fib6_info *f6i, bool skip_notify);
123 
124 void rt6_flush_exceptions(struct fib6_info *f6i);
125 void rt6_age_exceptions(struct fib6_info *f6i, struct fib6_gc_args *gc_args,
126 			unsigned long now);
127 
128 static inline int ip6_route_get_saddr(struct net *net, struct fib6_info *f6i,
129 				      const struct in6_addr *daddr,
130 				      unsigned int prefs,
131 				      struct in6_addr *saddr)
132 {
133 	int err = 0;
134 
135 	if (f6i && f6i->fib6_prefsrc.plen) {
136 		*saddr = f6i->fib6_prefsrc.addr;
137 	} else {
138 		struct net_device *dev = f6i ? fib6_info_nh_dev(f6i) : NULL;
139 
140 		err = ipv6_dev_get_saddr(net, dev, daddr, prefs, saddr);
141 	}
142 
143 	return err;
144 }
145 
146 struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
147 			    const struct in6_addr *saddr, int oif,
148 			    const struct sk_buff *skb, int flags);
149 u32 rt6_multipath_hash(const struct net *net, const struct flowi6 *fl6,
150 		       const struct sk_buff *skb, struct flow_keys *hkeys);
151 
152 struct dst_entry *icmp6_dst_alloc(struct net_device *dev, struct flowi6 *fl6);
153 
154 void fib6_force_start_gc(struct net *net);
155 
156 struct fib6_info *addrconf_f6i_alloc(struct net *net, struct inet6_dev *idev,
157 				     const struct in6_addr *addr, bool anycast,
158 				     gfp_t gfp_flags, struct netlink_ext_ack *extack);
159 
160 struct rt6_info *ip6_dst_alloc(struct net *net, struct net_device *dev,
161 			       int flags);
162 
163 /*
164  *	support functions for ND
165  *
166  */
167 struct fib6_info *rt6_get_dflt_router(struct net *net,
168 				     const struct in6_addr *addr,
169 				     struct net_device *dev);
170 struct fib6_info *rt6_add_dflt_router(struct net *net,
171 				     const struct in6_addr *gwaddr,
172 				     struct net_device *dev, unsigned int pref,
173 				     u32 defrtr_usr_metric,
174 				     int lifetime);
175 
176 void rt6_purge_dflt_routers(struct net *net);
177 
178 int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
179 		  const struct in6_addr *gwaddr);
180 
181 void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, int oif,
182 		     u32 mark, kuid_t uid);
183 void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu);
184 void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
185 		  kuid_t uid);
186 void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif);
187 void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk);
188 
189 struct netlink_callback;
190 
191 struct rt6_rtnl_dump_arg {
192 	struct sk_buff *skb;
193 	struct netlink_callback *cb;
194 	struct net *net;
195 	struct fib_dump_filter filter;
196 };
197 
198 int rt6_dump_route(struct fib6_info *f6i, void *p_arg, unsigned int skip);
199 void rt6_mtu_change(struct net_device *dev, unsigned int mtu);
200 void rt6_remove_prefsrc(struct inet6_ifaddr *ifp);
201 void rt6_clean_tohost(struct net *net, struct in6_addr *gateway);
202 void rt6_sync_up(struct net_device *dev, unsigned char nh_flags);
203 void rt6_disable_ip(struct net_device *dev, unsigned long event);
204 void rt6_sync_down_dev(struct net_device *dev, unsigned long event);
205 void rt6_multipath_rebalance(struct fib6_info *f6i);
206 
207 void rt6_uncached_list_add(struct rt6_info *rt);
208 void rt6_uncached_list_del(struct rt6_info *rt);
209 
210 static inline const struct rt6_info *skb_rt6_info(const struct sk_buff *skb)
211 {
212 	const struct dst_entry *dst = skb_dst(skb);
213 	const struct rt6_info *rt6 = NULL;
214 
215 	if (dst)
216 		rt6 = container_of(dst, struct rt6_info, dst);
217 
218 	return rt6;
219 }
220 
221 /*
222  *	Store a destination cache entry in a socket
223  */
224 static inline void ip6_dst_store(struct sock *sk, struct dst_entry *dst,
225 				 const struct in6_addr *daddr,
226 				 const struct in6_addr *saddr)
227 {
228 	struct ipv6_pinfo *np = inet6_sk(sk);
229 
230 	np->dst_cookie = rt6_get_cookie((struct rt6_info *)dst);
231 	sk_setup_caps(sk, dst);
232 	np->daddr_cache = daddr;
233 #ifdef CONFIG_IPV6_SUBTREES
234 	np->saddr_cache = saddr;
235 #endif
236 }
237 
238 void ip6_sk_dst_store_flow(struct sock *sk, struct dst_entry *dst,
239 			   const struct flowi6 *fl6);
240 
241 static inline bool ipv6_unicast_destination(const struct sk_buff *skb)
242 {
243 	struct rt6_info *rt = (struct rt6_info *) skb_dst(skb);
244 
245 	return rt->rt6i_flags & RTF_LOCAL;
246 }
247 
248 static inline bool ipv6_anycast_destination(const struct dst_entry *dst,
249 					    const struct in6_addr *daddr)
250 {
251 	struct rt6_info *rt = (struct rt6_info *)dst;
252 
253 	return rt->rt6i_flags & RTF_ANYCAST ||
254 		(rt->rt6i_dst.plen < 127 &&
255 		 !(rt->rt6i_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) &&
256 		 ipv6_addr_equal(&rt->rt6i_dst.addr, daddr));
257 }
258 
259 int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
260 		 int (*output)(struct net *, struct sock *, struct sk_buff *));
261 
262 static inline unsigned int ip6_skb_dst_mtu(const struct sk_buff *skb)
263 {
264 	const struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ?
265 				inet6_sk(skb->sk) : NULL;
266 	const struct dst_entry *dst = skb_dst(skb);
267 	unsigned int mtu;
268 
269 	if (np && READ_ONCE(np->pmtudisc) >= IPV6_PMTUDISC_PROBE) {
270 		mtu = READ_ONCE(dst->dev->mtu);
271 		mtu -= lwtunnel_headroom(dst->lwtstate, mtu);
272 	} else {
273 		mtu = dst_mtu(dst);
274 	}
275 	return mtu;
276 }
277 
278 static inline bool ip6_sk_accept_pmtu(const struct sock *sk)
279 {
280 	u8 pmtudisc = READ_ONCE(inet6_sk(sk)->pmtudisc);
281 
282 	return pmtudisc != IPV6_PMTUDISC_INTERFACE &&
283 	       pmtudisc != IPV6_PMTUDISC_OMIT;
284 }
285 
286 static inline bool ip6_sk_ignore_df(const struct sock *sk)
287 {
288 	u8 pmtudisc = READ_ONCE(inet6_sk(sk)->pmtudisc);
289 
290 	return pmtudisc < IPV6_PMTUDISC_DO ||
291 	       pmtudisc == IPV6_PMTUDISC_OMIT;
292 }
293 
294 static inline const struct in6_addr *rt6_nexthop(const struct rt6_info *rt,
295 						 const struct in6_addr *daddr)
296 {
297 	if (rt->rt6i_flags & RTF_GATEWAY)
298 		return &rt->rt6i_gateway;
299 	else if (unlikely(rt->rt6i_flags & RTF_CACHE))
300 		return &rt->rt6i_dst.addr;
301 	else
302 		return daddr;
303 }
304 
305 static inline bool rt6_duplicate_nexthop(struct fib6_info *a, struct fib6_info *b)
306 {
307 	struct fib6_nh *nha, *nhb;
308 
309 	if (a->nh || b->nh)
310 		return nexthop_cmp(a->nh, b->nh);
311 
312 	nha = a->fib6_nh;
313 	nhb = b->fib6_nh;
314 	return nha->fib_nh_dev == nhb->fib_nh_dev &&
315 	       ipv6_addr_equal(&nha->fib_nh_gw6, &nhb->fib_nh_gw6) &&
316 	       !lwtunnel_cmp_encap(nha->fib_nh_lws, nhb->fib_nh_lws);
317 }
318 
319 static inline unsigned int ip6_dst_mtu_maybe_forward(const struct dst_entry *dst,
320 						     bool forwarding)
321 {
322 	struct inet6_dev *idev;
323 	unsigned int mtu;
324 
325 	if (!forwarding || dst_metric_locked(dst, RTAX_MTU)) {
326 		mtu = dst_metric_raw(dst, RTAX_MTU);
327 		if (mtu)
328 			goto out;
329 	}
330 
331 	mtu = IPV6_MIN_MTU;
332 	rcu_read_lock();
333 	idev = __in6_dev_get(dst->dev);
334 	if (idev)
335 		mtu = READ_ONCE(idev->cnf.mtu6);
336 	rcu_read_unlock();
337 
338 out:
339 	return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
340 }
341 
342 u32 ip6_mtu_from_fib6(const struct fib6_result *res,
343 		      const struct in6_addr *daddr,
344 		      const struct in6_addr *saddr);
345 
346 struct neighbour *ip6_neigh_lookup(const struct in6_addr *gw,
347 				   struct net_device *dev, struct sk_buff *skb,
348 				   const void *daddr);
349 #endif
350