xref: /openbmc/linux/include/net/ip6_route.h (revision b868a02e)
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 	/* No need to bitmask because srcprefs have only 3 bits. */
57 	return srcprefs << 3;
58 }
59 
60 static inline unsigned int rt6_flags2srcprefs(int flags)
61 {
62 	return (flags >> 3) & 7;
63 }
64 
65 static inline bool rt6_need_strict(const struct in6_addr *daddr)
66 {
67 	return ipv6_addr_type(daddr) &
68 		(IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK);
69 }
70 
71 /* fib entries using a nexthop object can not be coalesced into
72  * a multipath route
73  */
74 static inline bool rt6_qualify_for_ecmp(const struct fib6_info *f6i)
75 {
76 	/* the RTF_ADDRCONF flag filters out RA's */
77 	return !(f6i->fib6_flags & RTF_ADDRCONF) && !f6i->nh &&
78 		f6i->fib6_nh->fib_nh_gw_family;
79 }
80 
81 void ip6_route_input(struct sk_buff *skb);
82 struct dst_entry *ip6_route_input_lookup(struct net *net,
83 					 struct net_device *dev,
84 					 struct flowi6 *fl6,
85 					 const struct sk_buff *skb, int flags);
86 
87 struct dst_entry *ip6_route_output_flags_noref(struct net *net,
88 					       const struct sock *sk,
89 					       struct flowi6 *fl6, int flags);
90 
91 struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
92 					 struct flowi6 *fl6, int flags);
93 
94 static inline struct dst_entry *ip6_route_output(struct net *net,
95 						 const struct sock *sk,
96 						 struct flowi6 *fl6)
97 {
98 	return ip6_route_output_flags(net, sk, fl6, 0);
99 }
100 
101 /* Only conditionally release dst if flags indicates
102  * !RT6_LOOKUP_F_DST_NOREF or dst is in uncached_list.
103  */
104 static inline void ip6_rt_put_flags(struct rt6_info *rt, int flags)
105 {
106 	if (!(flags & RT6_LOOKUP_F_DST_NOREF) ||
107 	    !list_empty(&rt->rt6i_uncached))
108 		ip6_rt_put(rt);
109 }
110 
111 struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
112 				   const struct sk_buff *skb, int flags);
113 struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
114 			       int ifindex, struct flowi6 *fl6,
115 			       const struct sk_buff *skb, int flags);
116 
117 void ip6_route_init_special_entries(void);
118 int ip6_route_init(void);
119 void ip6_route_cleanup(void);
120 
121 int ipv6_route_ioctl(struct net *net, unsigned int cmd,
122 		struct in6_rtmsg *rtmsg);
123 
124 int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags,
125 		  struct netlink_ext_ack *extack);
126 int ip6_ins_rt(struct net *net, struct fib6_info *f6i);
127 int ip6_del_rt(struct net *net, struct fib6_info *f6i, bool skip_notify);
128 
129 void rt6_flush_exceptions(struct fib6_info *f6i);
130 void rt6_age_exceptions(struct fib6_info *f6i, struct fib6_gc_args *gc_args,
131 			unsigned long now);
132 
133 static inline int ip6_route_get_saddr(struct net *net, struct fib6_info *f6i,
134 				      const struct in6_addr *daddr,
135 				      unsigned int prefs,
136 				      struct in6_addr *saddr)
137 {
138 	int err = 0;
139 
140 	if (f6i && f6i->fib6_prefsrc.plen) {
141 		*saddr = f6i->fib6_prefsrc.addr;
142 	} else {
143 		struct net_device *dev = f6i ? fib6_info_nh_dev(f6i) : NULL;
144 
145 		err = ipv6_dev_get_saddr(net, dev, daddr, prefs, saddr);
146 	}
147 
148 	return err;
149 }
150 
151 struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
152 			    const struct in6_addr *saddr, int oif,
153 			    const struct sk_buff *skb, int flags);
154 u32 rt6_multipath_hash(const struct net *net, const struct flowi6 *fl6,
155 		       const struct sk_buff *skb, struct flow_keys *hkeys);
156 
157 struct dst_entry *icmp6_dst_alloc(struct net_device *dev, struct flowi6 *fl6);
158 
159 void fib6_force_start_gc(struct net *net);
160 
161 struct fib6_info *addrconf_f6i_alloc(struct net *net, struct inet6_dev *idev,
162 				     const struct in6_addr *addr, bool anycast,
163 				     gfp_t gfp_flags);
164 
165 struct rt6_info *ip6_dst_alloc(struct net *net, struct net_device *dev,
166 			       int flags);
167 
168 /*
169  *	support functions for ND
170  *
171  */
172 struct fib6_info *rt6_get_dflt_router(struct net *net,
173 				     const struct in6_addr *addr,
174 				     struct net_device *dev);
175 struct fib6_info *rt6_add_dflt_router(struct net *net,
176 				     const struct in6_addr *gwaddr,
177 				     struct net_device *dev, unsigned int pref,
178 				     u32 defrtr_usr_metric);
179 
180 void rt6_purge_dflt_routers(struct net *net);
181 
182 int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
183 		  const struct in6_addr *gwaddr);
184 
185 void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, int oif,
186 		     u32 mark, kuid_t uid);
187 void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu);
188 void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
189 		  kuid_t uid);
190 void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif);
191 void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk);
192 
193 struct netlink_callback;
194 
195 struct rt6_rtnl_dump_arg {
196 	struct sk_buff *skb;
197 	struct netlink_callback *cb;
198 	struct net *net;
199 	struct fib_dump_filter filter;
200 };
201 
202 int rt6_dump_route(struct fib6_info *f6i, void *p_arg, unsigned int skip);
203 void rt6_mtu_change(struct net_device *dev, unsigned int mtu);
204 void rt6_remove_prefsrc(struct inet6_ifaddr *ifp);
205 void rt6_clean_tohost(struct net *net, struct in6_addr *gateway);
206 void rt6_sync_up(struct net_device *dev, unsigned char nh_flags);
207 void rt6_disable_ip(struct net_device *dev, unsigned long event);
208 void rt6_sync_down_dev(struct net_device *dev, unsigned long event);
209 void rt6_multipath_rebalance(struct fib6_info *f6i);
210 
211 void rt6_uncached_list_add(struct rt6_info *rt);
212 void rt6_uncached_list_del(struct rt6_info *rt);
213 
214 static inline const struct rt6_info *skb_rt6_info(const struct sk_buff *skb)
215 {
216 	const struct dst_entry *dst = skb_dst(skb);
217 	const struct rt6_info *rt6 = NULL;
218 
219 	if (dst)
220 		rt6 = container_of(dst, struct rt6_info, dst);
221 
222 	return rt6;
223 }
224 
225 /*
226  *	Store a destination cache entry in a socket
227  */
228 static inline void ip6_dst_store(struct sock *sk, struct dst_entry *dst,
229 				 const struct in6_addr *daddr,
230 				 const struct in6_addr *saddr)
231 {
232 	struct ipv6_pinfo *np = inet6_sk(sk);
233 
234 	np->dst_cookie = rt6_get_cookie((struct rt6_info *)dst);
235 	sk_setup_caps(sk, dst);
236 	np->daddr_cache = daddr;
237 #ifdef CONFIG_IPV6_SUBTREES
238 	np->saddr_cache = saddr;
239 #endif
240 }
241 
242 void ip6_sk_dst_store_flow(struct sock *sk, struct dst_entry *dst,
243 			   const struct flowi6 *fl6);
244 
245 static inline bool ipv6_unicast_destination(const struct sk_buff *skb)
246 {
247 	struct rt6_info *rt = (struct rt6_info *) skb_dst(skb);
248 
249 	return rt->rt6i_flags & RTF_LOCAL;
250 }
251 
252 static inline bool ipv6_anycast_destination(const struct dst_entry *dst,
253 					    const struct in6_addr *daddr)
254 {
255 	struct rt6_info *rt = (struct rt6_info *)dst;
256 
257 	return rt->rt6i_flags & RTF_ANYCAST ||
258 		(rt->rt6i_dst.plen < 127 &&
259 		 !(rt->rt6i_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) &&
260 		 ipv6_addr_equal(&rt->rt6i_dst.addr, daddr));
261 }
262 
263 int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
264 		 int (*output)(struct net *, struct sock *, struct sk_buff *));
265 
266 static inline unsigned int ip6_skb_dst_mtu(const struct sk_buff *skb)
267 {
268 	const struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ?
269 				inet6_sk(skb->sk) : NULL;
270 	const struct dst_entry *dst = skb_dst(skb);
271 	unsigned int mtu;
272 
273 	if (np && np->pmtudisc >= IPV6_PMTUDISC_PROBE) {
274 		mtu = READ_ONCE(dst->dev->mtu);
275 		mtu -= lwtunnel_headroom(dst->lwtstate, mtu);
276 	} else {
277 		mtu = dst_mtu(dst);
278 	}
279 	return mtu;
280 }
281 
282 static inline bool ip6_sk_accept_pmtu(const struct sock *sk)
283 {
284 	return inet6_sk(sk)->pmtudisc != IPV6_PMTUDISC_INTERFACE &&
285 	       inet6_sk(sk)->pmtudisc != IPV6_PMTUDISC_OMIT;
286 }
287 
288 static inline bool ip6_sk_ignore_df(const struct sock *sk)
289 {
290 	return inet6_sk(sk)->pmtudisc < IPV6_PMTUDISC_DO ||
291 	       inet6_sk(sk)->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 = 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