in6_src.c revision 1.87
1/*	$OpenBSD: in6_src.c,v 1.87 2023/11/28 13:23:20 bluhm Exp $	*/
2/*	$KAME: in6_src.c,v 1.36 2001/02/06 04:08:17 itojun Exp $	*/
3
4/*
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 *    notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 *    notice, this list of conditions and the following disclaimer in the
15 *    documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 *    may be used to endorse or promote products derived from this software
18 *    without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32
33/*
34 * Copyright (c) 1982, 1986, 1991, 1993
35 *	The Regents of the University of California.  All rights reserved.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
40 * 1. Redistributions of source code must retain the above copyright
41 *    notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 *    notice, this list of conditions and the following disclaimer in the
44 *    documentation and/or other materials provided with the distribution.
45 * 3. Neither the name of the University nor the names of its contributors
46 *    may be used to endorse or promote products derived from this software
47 *    without specific prior written permission.
48 *
49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
59 * SUCH DAMAGE.
60 *
61 *	@(#)in_pcb.c	8.2 (Berkeley) 1/4/94
62 */
63
64#include <sys/param.h>
65#include <sys/systm.h>
66#include <sys/mbuf.h>
67#include <sys/socket.h>
68#include <sys/socketvar.h>
69#include <sys/ioctl.h>
70#include <sys/errno.h>
71#include <sys/time.h>
72
73#include <net/if.h>
74#include <net/if_var.h>
75#include <net/route.h>
76
77#include <netinet/in.h>
78#include <netinet/ip.h>
79#include <netinet/in_pcb.h>
80#include <netinet6/in6_var.h>
81#include <netinet/ip6.h>
82#include <netinet6/ip6_var.h>
83#include <netinet6/nd6.h>
84
85int in6_selectif(struct sockaddr_in6 *, struct ip6_pktopts *,
86    struct ip6_moptions *, struct route_in6 *, struct ifnet **, u_int);
87
88/*
89 * Return an IPv6 address, which is the most appropriate for a given
90 * destination and pcb. We need the additional opt parameter because
91 * the values set at pcb level can be overridden via cmsg.
92 */
93int
94in6_pcbselsrc(struct in6_addr **in6src, struct sockaddr_in6 *dstsock,
95    struct inpcb *inp, struct ip6_pktopts *opts)
96{
97	struct ip6_moptions *mopts = inp->inp_moptions6;
98	struct route_in6 *ro = &inp->inp_route6;
99	struct in6_addr *laddr = &inp->inp_laddr6;
100	u_int rtableid = inp->inp_rtableid;
101	struct ifnet *ifp = NULL;
102	struct sockaddr	*ip6_source = NULL;
103	struct in6_addr *dst;
104	struct in6_ifaddr *ia6 = NULL;
105	struct in6_pktinfo *pi = NULL;
106	int	error;
107
108	dst = &dstsock->sin6_addr;
109
110	/*
111	 * If the source address is explicitly specified by the caller,
112	 * check if the requested source address is indeed a unicast address
113	 * assigned to the node, and can be used as the packet's source
114	 * address.  If everything is okay, use the address as source.
115	 */
116	if (opts && (pi = opts->ip6po_pktinfo) &&
117	    !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) {
118		struct sockaddr_in6 sa6;
119
120		/* get the outgoing interface */
121		error = in6_selectif(dstsock, opts, mopts, ro, &ifp, rtableid);
122		if (error)
123			return (error);
124
125		bzero(&sa6, sizeof(sa6));
126		sa6.sin6_family = AF_INET6;
127		sa6.sin6_len = sizeof(sa6);
128		sa6.sin6_addr = pi->ipi6_addr;
129
130		if (ifp && IN6_IS_SCOPE_EMBED(&sa6.sin6_addr))
131			sa6.sin6_addr.s6_addr16[1] = htons(ifp->if_index);
132		if_put(ifp); /* put reference from in6_selectif */
133
134		ia6 = ifatoia6(ifa_ifwithaddr(sin6tosa(&sa6), rtableid));
135		if (ia6 == NULL || (ia6->ia6_flags &
136		     (IN6_IFF_ANYCAST|IN6_IFF_TENTATIVE|IN6_IFF_DUPLICATED)))
137			return (EADDRNOTAVAIL);
138
139		pi->ipi6_addr = sa6.sin6_addr; /* XXX: this overrides pi */
140
141		*in6src = &pi->ipi6_addr;
142		return (0);
143	}
144
145	/*
146	 * If the source address is not specified but the socket(if any)
147	 * is already bound, use the bound address.
148	 */
149	if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) {
150		*in6src = laddr;
151		return (0);
152	}
153
154	/*
155	 * If the caller doesn't specify the source address but
156	 * the outgoing interface, use an address associated with
157	 * the interface.
158	 */
159	if (pi && pi->ipi6_ifindex) {
160		ifp = if_get(pi->ipi6_ifindex);
161		if (ifp == NULL)
162			return (ENXIO); /* XXX: better error? */
163
164		ia6 = in6_ifawithscope(ifp, dst, rtableid);
165		if_put(ifp);
166
167		if (ia6 == NULL)
168			return (EADDRNOTAVAIL);
169
170		*in6src = &ia6->ia_addr.sin6_addr;
171		return (0);
172	}
173
174	error = in6_selectsrc(in6src, dstsock, mopts, rtableid);
175	if (error != EADDRNOTAVAIL)
176		return (error);
177
178	/*
179	 * If route is known or can be allocated now,
180	 * our src addr is taken from the i/f, else punt.
181	 */
182	if (!rtisvalid(ro->ro_rt) || (ro->ro_tableid != rtableid) ||
183	    !IN6_ARE_ADDR_EQUAL(&ro->ro_dst.sin6_addr, dst)) {
184		rtfree(ro->ro_rt);
185		ro->ro_rt = NULL;
186	}
187	if (ro->ro_rt == NULL) {
188		struct sockaddr_in6 *sa6;
189
190		/* No route yet, so try to acquire one */
191		bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
192		ro->ro_tableid = rtableid;
193		sa6 = &ro->ro_dst;
194		sa6->sin6_family = AF_INET6;
195		sa6->sin6_len = sizeof(struct sockaddr_in6);
196		sa6->sin6_addr = *dst;
197		sa6->sin6_scope_id = dstsock->sin6_scope_id;
198		ro->ro_rt = rtalloc(sin6tosa(&ro->ro_dst),
199		    RT_RESOLVE, ro->ro_tableid);
200	}
201
202	/*
203	 * in_pcbconnect() checks out IFF_LOOPBACK to skip using
204	 * the address. But we don't know why it does so.
205	 * It is necessary to ensure the scope even for lo0
206	 * so doesn't check out IFF_LOOPBACK.
207	 */
208
209	if (ro->ro_rt) {
210		ifp = if_get(ro->ro_rt->rt_ifidx);
211		if (ifp != NULL) {
212			ia6 = in6_ifawithscope(ifp, dst, rtableid);
213			if_put(ifp);
214		}
215		if (ia6 == NULL) /* xxx scope error ?*/
216			ia6 = ifatoia6(ro->ro_rt->rt_ifa);
217	}
218
219	/*
220	 * Use preferred source address if :
221	 * - destination is not onlink
222	 * - preferred source address is set
223	 * - output interface is UP
224	 */
225	if (ro->ro_rt && !(ro->ro_rt->rt_flags & RTF_LLINFO) &&
226	    !(ro->ro_rt->rt_flags & RTF_HOST)) {
227		ip6_source = rtable_getsource(rtableid, AF_INET6);
228		if (ip6_source != NULL) {
229			struct ifaddr *ifa;
230			if ((ifa = ifa_ifwithaddr(ip6_source, rtableid)) !=
231			    NULL && ISSET(ifa->ifa_ifp->if_flags, IFF_UP)) {
232				*in6src = &satosin6(ip6_source)->sin6_addr;
233				return (0);
234			}
235		}
236	}
237
238	if (ia6 == NULL)
239		return (EHOSTUNREACH);	/* no route */
240
241	*in6src = &ia6->ia_addr.sin6_addr;
242	return (0);
243}
244
245/*
246 * Return an IPv6 address, which is the most appropriate for a given
247 * destination and multicast options.
248 * If necessary, this function lookups the routing table and returns
249 * an entry to the caller for later use.
250 */
251int
252in6_selectsrc(struct in6_addr **in6src, struct sockaddr_in6 *dstsock,
253    struct ip6_moptions *mopts, unsigned int rtableid)
254{
255	struct ifnet *ifp = NULL;
256	struct in6_addr *dst;
257	struct in6_ifaddr *ia6 = NULL;
258
259	dst = &dstsock->sin6_addr;
260
261	/*
262	 * If the destination address is a link-local unicast address or
263	 * a link/interface-local multicast address, and if the outgoing
264	 * interface is specified by the sin6_scope_id filed, use an address
265	 * associated with the interface.
266	 * XXX: We're now trying to define more specific semantics of
267	 *      sin6_scope_id field, so this part will be rewritten in
268	 *      the near future.
269	 */
270	if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MC_LINKLOCAL(dst) ||
271	     IN6_IS_ADDR_MC_INTFACELOCAL(dst)) && dstsock->sin6_scope_id) {
272		ifp = if_get(dstsock->sin6_scope_id);
273		if (ifp == NULL)
274			return (ENXIO); /* XXX: better error? */
275
276		ia6 = in6_ifawithscope(ifp, dst, rtableid);
277		if_put(ifp);
278
279		if (ia6 == NULL)
280			return (EADDRNOTAVAIL);
281
282		*in6src = &ia6->ia_addr.sin6_addr;
283		return (0);
284	}
285
286	/*
287	 * If the destination address is a multicast address and
288	 * the outgoing interface for the address is specified
289	 * by the caller, use an address associated with the interface.
290	 * Even if the outgoing interface is not specified, we also
291	 * choose a loopback interface as the outgoing interface.
292	 */
293	if (IN6_IS_ADDR_MULTICAST(dst)) {
294		ifp = mopts ? if_get(mopts->im6o_ifidx) : NULL;
295
296		if (!ifp && dstsock->sin6_scope_id)
297			ifp = if_get(htons(dstsock->sin6_scope_id));
298
299		if (ifp) {
300			ia6 = in6_ifawithscope(ifp, dst, rtableid);
301			if_put(ifp);
302
303			if (ia6 == NULL)
304				return (EADDRNOTAVAIL);
305
306			*in6src = &ia6->ia_addr.sin6_addr;
307			return (0);
308		}
309	}
310
311	return (EADDRNOTAVAIL);
312}
313
314struct rtentry *
315in6_selectroute(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
316    struct route_in6 *ro, unsigned int rtableid)
317{
318	struct in6_addr *dst;
319
320	dst = &dstsock->sin6_addr;
321
322	/*
323	 * Use a cached route if it exists and is valid, else try to allocate
324	 * a new one.
325	 */
326	if (ro) {
327		if (rtisvalid(ro->ro_rt))
328			KASSERT(sin6tosa(&ro->ro_dst)->sa_family == AF_INET6);
329		if (!rtisvalid(ro->ro_rt) ||
330		    !IN6_ARE_ADDR_EQUAL(&ro->ro_dst.sin6_addr, dst)) {
331			rtfree(ro->ro_rt);
332			ro->ro_rt = NULL;
333		}
334		if (ro->ro_rt == NULL) {
335			struct sockaddr_in6 *sa6;
336
337			/* No route yet, so try to acquire one */
338			bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
339			ro->ro_tableid = rtableid;
340			sa6 = &ro->ro_dst;
341			*sa6 = *dstsock;
342			sa6->sin6_scope_id = 0;
343			ro->ro_tableid = rtableid;
344			ro->ro_rt = rtalloc_mpath(sin6tosa(&ro->ro_dst),
345			    NULL, ro->ro_tableid);
346		}
347
348		/*
349		 * Check if the outgoing interface conflicts with
350		 * the interface specified by ipi6_ifindex (if specified).
351		 * Note that loopback interface is always okay.
352		 * (this may happen when we are sending a packet to one of
353		 *  our own addresses.)
354		 */
355		if (opts && opts->ip6po_pktinfo &&
356		    opts->ip6po_pktinfo->ipi6_ifindex) {
357			if (ro->ro_rt != NULL &&
358			    !ISSET(ro->ro_rt->rt_flags, RTF_LOCAL) &&
359			    ro->ro_rt->rt_ifidx !=
360			    opts->ip6po_pktinfo->ipi6_ifindex) {
361			    	return (NULL);
362			}
363		}
364
365		return (ro->ro_rt);
366	}
367
368	return (NULL);
369}
370
371int
372in6_selectif(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
373    struct ip6_moptions *mopts, struct route_in6 *ro, struct ifnet **retifp,
374    u_int rtableid)
375{
376	struct rtentry *rt = NULL;
377	struct in6_pktinfo *pi = NULL;
378
379	/* If the caller specify the outgoing interface explicitly, use it. */
380	if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) {
381		*retifp = if_get(pi->ipi6_ifindex);
382		if (*retifp != NULL)
383			return (0);
384	}
385
386	/*
387	 * If the destination address is a multicast address and the outgoing
388	 * interface for the address is specified by the caller, use it.
389	 */
390	if (IN6_IS_ADDR_MULTICAST(&dstsock->sin6_addr) &&
391	    mopts != NULL && (*retifp = if_get(mopts->im6o_ifidx)) != NULL)
392	    	return (0);
393
394	rt = in6_selectroute(dstsock, opts, ro, rtableid);
395	if (rt == NULL)
396		return (EHOSTUNREACH);
397
398	/*
399	 * do not use a rejected or black hole route.
400	 * XXX: this check should be done in the L2 output routine.
401	 * However, if we skipped this check here, we'd see the following
402	 * scenario:
403	 * - install a rejected route for a scoped address prefix
404	 *   (like fe80::/10)
405	 * - send a packet to a destination that matches the scoped prefix,
406	 *   with ambiguity about the scope zone.
407	 * - pick the outgoing interface from the route, and disambiguate the
408	 *   scope zone with the interface.
409	 * - ip6_output() would try to get another route with the "new"
410	 *   destination, which may be valid.
411	 * - we'd see no error on output.
412	 * Although this may not be very harmful, it should still be confusing.
413	 * We thus reject the case here.
414	 */
415	if (rt && (rt->rt_flags & (RTF_REJECT | RTF_BLACKHOLE)))
416		return (rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH);
417
418	if (rt != NULL)
419		*retifp = if_get(rt->rt_ifidx);
420
421	return (0);
422}
423
424int
425in6_selecthlim(struct inpcb *in6p)
426{
427	if (in6p && in6p->inp_hops >= 0)
428		return (in6p->inp_hops);
429
430	return (ip6_defhlim);
431}
432
433/*
434 * generate kernel-internal form (scopeid embedded into s6_addr16[1]).
435 * If the address scope of is link-local, embed the interface index in the
436 * address.  The routine determines our precedence
437 * between advanced API scope/interface specification and basic API
438 * specification.
439 *
440 * this function should be nuked in the future, when we get rid of
441 * embedded scopeid thing.
442 *
443 * XXX actually, it is over-specification to return ifp against sin6_scope_id.
444 * there can be multiple interfaces that belong to a particular scope zone
445 * (in specification, we have 1:N mapping between a scope zone and interfaces).
446 * we may want to change the function to return something other than ifp.
447 */
448int
449in6_embedscope(struct in6_addr *in6, const struct sockaddr_in6 *sin6,
450    const struct ip6_pktopts *outputopts6, const struct ip6_moptions *moptions6)
451{
452	u_int32_t scopeid;
453
454	*in6 = sin6->sin6_addr;
455
456	/*
457	 * don't try to read sin6->sin6_addr beyond here, since the caller may
458	 * ask us to overwrite existing sockaddr_in6
459	 */
460
461	if (IN6_IS_SCOPE_EMBED(in6)) {
462		struct in6_pktinfo *pi;
463
464		/*
465		 * KAME assumption: link id == interface id
466		 */
467
468		if (outputopts6 && (pi = outputopts6->ip6po_pktinfo) &&
469		    pi->ipi6_ifindex)
470			scopeid = pi->ipi6_ifindex;
471		else if (moptions6 && IN6_IS_ADDR_MULTICAST(in6) &&
472		    moptions6->im6o_ifidx)
473			scopeid = moptions6->im6o_ifidx;
474		else
475			scopeid = sin6->sin6_scope_id;
476
477		if (scopeid) {
478			struct ifnet *ifp;
479
480			ifp = if_get(scopeid);
481			if (ifp == NULL)
482				return ENXIO;  /* XXX EINVAL? */
483			/*XXX assignment to 16bit from 32bit variable */
484			in6->s6_addr16[1] = htons(scopeid & 0xffff);
485			if_put(ifp);
486		}
487	}
488
489	return 0;
490}
491
492/*
493 * generate standard sockaddr_in6 from embedded form.
494 * touches sin6_addr and sin6_scope_id only.
495 *
496 * this function should be nuked in the future, when we get rid of
497 * embedded scopeid thing.
498 */
499void
500in6_recoverscope(struct sockaddr_in6 *sin6, const struct in6_addr *in6)
501{
502	u_int32_t scopeid;
503
504	sin6->sin6_addr = *in6;
505
506	/*
507	 * don't try to read *in6 beyond here, since the caller may
508	 * ask us to overwrite existing sockaddr_in6
509	 */
510
511	sin6->sin6_scope_id = 0;
512	if (IN6_IS_SCOPE_EMBED(in6)) {
513		/*
514		 * KAME assumption: link id == interface id
515		 */
516		scopeid = ntohs(sin6->sin6_addr.s6_addr16[1]);
517		if (scopeid) {
518			sin6->sin6_addr.s6_addr16[1] = 0;
519			sin6->sin6_scope_id = scopeid;
520		}
521	}
522}
523
524/*
525 * just clear the embedded scope identifier.
526 */
527void
528in6_clearscope(struct in6_addr *addr)
529{
530	if (IN6_IS_SCOPE_EMBED(addr))
531		addr->s6_addr16[1] = 0;
532}
533