xref: /openbmc/linux/include/net/ip6_route.h (revision 0e6774ec)
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(struct net *net, const struct sock *sk,
88 					 struct flowi6 *fl6, int flags);
89 
90 static inline struct dst_entry *ip6_route_output(struct net *net,
91 						 const struct sock *sk,
92 						 struct flowi6 *fl6)
93 {
94 	return ip6_route_output_flags(net, sk, fl6, 0);
95 }
96 
97 /* Only conditionally release dst if flags indicates
98  * !RT6_LOOKUP_F_DST_NOREF or dst is in uncached_list.
99  */
100 static inline void ip6_rt_put_flags(struct rt6_info *rt, int flags)
101 {
102 	if (!(flags & RT6_LOOKUP_F_DST_NOREF) ||
103 	    !list_empty(&rt->dst.rt_uncached))
104 		ip6_rt_put(rt);
105 }
106 
107 struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
108 				   const struct sk_buff *skb, int flags);
109 struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
110 			       int ifindex, struct flowi6 *fl6,
111 			       const struct sk_buff *skb, int flags);
112 
113 void ip6_route_init_special_entries(void);
114 int ip6_route_init(void);
115 void ip6_route_cleanup(void);
116 
117 int ipv6_route_ioctl(struct net *net, unsigned int cmd,
118 		struct in6_rtmsg *rtmsg);
119 
120 int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags,
121 		  struct netlink_ext_ack *extack);
122 int ip6_ins_rt(struct net *net, struct fib6_info *f6i);
123 int ip6_del_rt(struct net *net, struct fib6_info *f6i, bool skip_notify);
124 
125 void rt6_flush_exceptions(struct fib6_info *f6i);
126 void rt6_age_exceptions(struct fib6_info *f6i, struct fib6_gc_args *gc_args,
127 			unsigned long now);
128 
129 static inline int ip6_route_get_saddr(struct net *net, struct fib6_info *f6i,
130 				      const struct in6_addr *daddr,
131 				      unsigned int prefs, int l3mdev_index,
132 				      struct in6_addr *saddr)
133 {
134 	struct net_device *l3mdev;
135 	struct net_device *dev;
136 	bool same_vrf;
137 	int err = 0;
138 
139 	rcu_read_lock();
140 
141 	l3mdev = dev_get_by_index_rcu(net, l3mdev_index);
142 	if (!f6i || !f6i->fib6_prefsrc.plen || l3mdev)
143 		dev = f6i ? fib6_info_nh_dev(f6i) : NULL;
144 	same_vrf = !l3mdev || l3mdev_master_dev_rcu(dev) == l3mdev;
145 	if (f6i && f6i->fib6_prefsrc.plen && same_vrf)
146 		*saddr = f6i->fib6_prefsrc.addr;
147 	else
148 		err = ipv6_dev_get_saddr(net, same_vrf ? dev : l3mdev, daddr, prefs, saddr);
149 
150 	rcu_read_unlock();
151 
152 	return err;
153 }
154 
155 struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
156 			    const struct in6_addr *saddr, int oif,
157 			    const struct sk_buff *skb, int flags);
158 u32 rt6_multipath_hash(const struct net *net, const struct flowi6 *fl6,
159 		       const struct sk_buff *skb, struct flow_keys *hkeys);
160 
161 struct dst_entry *icmp6_dst_alloc(struct net_device *dev, struct flowi6 *fl6);
162 
163 void fib6_force_start_gc(struct net *net);
164 
165 struct fib6_info *addrconf_f6i_alloc(struct net *net, struct inet6_dev *idev,
166 				     const struct in6_addr *addr, bool anycast,
167 				     gfp_t gfp_flags, struct netlink_ext_ack *extack);
168 
169 struct rt6_info *ip6_dst_alloc(struct net *net, struct net_device *dev,
170 			       int flags);
171 
172 /*
173  *	support functions for ND
174  *
175  */
176 struct fib6_info *rt6_get_dflt_router(struct net *net,
177 				     const struct in6_addr *addr,
178 				     struct net_device *dev);
179 struct fib6_info *rt6_add_dflt_router(struct net *net,
180 				     const struct in6_addr *gwaddr,
181 				     struct net_device *dev, unsigned int pref,
182 				     u32 defrtr_usr_metric);
183 
184 void rt6_purge_dflt_routers(struct net *net);
185 
186 int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
187 		  const struct in6_addr *gwaddr);
188 
189 void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, int oif,
190 		     u32 mark, kuid_t uid);
191 void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu);
192 void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
193 		  kuid_t uid);
194 void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif);
195 void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk);
196 
197 struct netlink_callback;
198 
199 struct rt6_rtnl_dump_arg {
200 	struct sk_buff *skb;
201 	struct netlink_callback *cb;
202 	struct net *net;
203 	struct fib_dump_filter filter;
204 };
205 
206 int rt6_dump_route(struct fib6_info *f6i, void *p_arg, unsigned int skip);
207 void rt6_mtu_change(struct net_device *dev, unsigned int mtu);
208 void rt6_remove_prefsrc(struct inet6_ifaddr *ifp);
209 void rt6_clean_tohost(struct net *net, struct in6_addr *gateway);
210 void rt6_sync_up(struct net_device *dev, unsigned char nh_flags);
211 void rt6_disable_ip(struct net_device *dev, unsigned long event);
212 void rt6_sync_down_dev(struct net_device *dev, unsigned long event);
213 void rt6_multipath_rebalance(struct fib6_info *f6i);
214 
215 void rt6_uncached_list_add(struct rt6_info *rt);
216 void rt6_uncached_list_del(struct rt6_info *rt);
217 
218 static inline const struct rt6_info *skb_rt6_info(const struct sk_buff *skb)
219 {
220 	const struct dst_entry *dst = skb_dst(skb);
221 	const struct rt6_info *rt6 = NULL;
222 
223 	if (dst)
224 		rt6 = container_of(dst, struct rt6_info, dst);
225 
226 	return rt6;
227 }
228 
229 /*
230  *	Store a destination cache entry in a socket
231  */
232 static inline void ip6_dst_store(struct sock *sk, struct dst_entry *dst,
233 				 const struct in6_addr *daddr,
234 				 const struct in6_addr *saddr)
235 {
236 	struct ipv6_pinfo *np = inet6_sk(sk);
237 
238 	np->dst_cookie = rt6_get_cookie((struct rt6_info *)dst);
239 	sk_setup_caps(sk, dst);
240 	np->daddr_cache = daddr;
241 #ifdef CONFIG_IPV6_SUBTREES
242 	np->saddr_cache = saddr;
243 #endif
244 }
245 
246 void ip6_sk_dst_store_flow(struct sock *sk, struct dst_entry *dst,
247 			   const struct flowi6 *fl6);
248 
249 static inline bool ipv6_unicast_destination(const struct sk_buff *skb)
250 {
251 	struct rt6_info *rt = (struct rt6_info *) skb_dst(skb);
252 
253 	return rt->rt6i_flags & RTF_LOCAL;
254 }
255 
256 static inline bool ipv6_anycast_destination(const struct dst_entry *dst,
257 					    const struct in6_addr *daddr)
258 {
259 	struct rt6_info *rt = (struct rt6_info *)dst;
260 
261 	return rt->rt6i_flags & RTF_ANYCAST ||
262 		(rt->rt6i_dst.plen < 127 &&
263 		 !(rt->rt6i_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) &&
264 		 ipv6_addr_equal(&rt->rt6i_dst.addr, daddr));
265 }
266 
267 int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
268 		 int (*output)(struct net *, struct sock *, struct sk_buff *));
269 
270 static inline unsigned int ip6_skb_dst_mtu(const struct sk_buff *skb)
271 {
272 	const struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ?
273 				inet6_sk(skb->sk) : NULL;
274 	const struct dst_entry *dst = skb_dst(skb);
275 	unsigned int mtu;
276 
277 	if (np && np->pmtudisc >= IPV6_PMTUDISC_PROBE) {
278 		mtu = READ_ONCE(dst->dev->mtu);
279 		mtu -= lwtunnel_headroom(dst->lwtstate, mtu);
280 	} else {
281 		mtu = dst_mtu(dst);
282 	}
283 	return mtu;
284 }
285 
286 static inline bool ip6_sk_accept_pmtu(const struct sock *sk)
287 {
288 	return inet6_sk(sk)->pmtudisc != IPV6_PMTUDISC_INTERFACE &&
289 	       inet6_sk(sk)->pmtudisc != IPV6_PMTUDISC_OMIT;
290 }
291 
292 static inline bool ip6_sk_ignore_df(const struct sock *sk)
293 {
294 	return inet6_sk(sk)->pmtudisc < IPV6_PMTUDISC_DO ||
295 	       inet6_sk(sk)->pmtudisc == IPV6_PMTUDISC_OMIT;
296 }
297 
298 static inline const struct in6_addr *rt6_nexthop(const struct rt6_info *rt,
299 						 const struct in6_addr *daddr)
300 {
301 	if (rt->rt6i_flags & RTF_GATEWAY)
302 		return &rt->rt6i_gateway;
303 	else if (unlikely(rt->rt6i_flags & RTF_CACHE))
304 		return &rt->rt6i_dst.addr;
305 	else
306 		return daddr;
307 }
308 
309 static inline bool rt6_duplicate_nexthop(struct fib6_info *a, struct fib6_info *b)
310 {
311 	struct fib6_nh *nha, *nhb;
312 
313 	if (a->nh || b->nh)
314 		return nexthop_cmp(a->nh, b->nh);
315 
316 	nha = a->fib6_nh;
317 	nhb = b->fib6_nh;
318 	return nha->fib_nh_dev == nhb->fib_nh_dev &&
319 	       ipv6_addr_equal(&nha->fib_nh_gw6, &nhb->fib_nh_gw6) &&
320 	       !lwtunnel_cmp_encap(nha->fib_nh_lws, nhb->fib_nh_lws);
321 }
322 
323 static inline unsigned int ip6_dst_mtu_maybe_forward(const struct dst_entry *dst,
324 						     bool forwarding)
325 {
326 	struct inet6_dev *idev;
327 	unsigned int mtu;
328 
329 	if (!forwarding || dst_metric_locked(dst, RTAX_MTU)) {
330 		mtu = dst_metric_raw(dst, RTAX_MTU);
331 		if (mtu)
332 			goto out;
333 	}
334 
335 	mtu = IPV6_MIN_MTU;
336 	rcu_read_lock();
337 	idev = __in6_dev_get(dst->dev);
338 	if (idev)
339 		mtu = idev->cnf.mtu6;
340 	rcu_read_unlock();
341 
342 out:
343 	return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
344 }
345 
346 u32 ip6_mtu_from_fib6(const struct fib6_result *res,
347 		      const struct in6_addr *daddr,
348 		      const struct in6_addr *saddr);
349 
350 struct neighbour *ip6_neigh_lookup(const struct in6_addr *gw,
351 				   struct net_device *dev, struct sk_buff *skb,
352 				   const void *daddr);
353 #endif
354