name6.c revision 245225
1/*	$KAME: name6.c,v 1.25 2000/06/26 16:44:40 itojun Exp $	*/
2
3/*
4 * Copyright (C) 1995, 1996, 1997, 1998, and 1999 WIDE Project.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 *    notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 *    notice, this list of conditions and the following disclaimer in the
14 *    documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the project nor the names of its contributors
16 *    may be used to endorse or promote products derived from this software
17 *    without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 */
31/*
32 * ++Copyright++ 1985, 1988, 1993
33 * -
34 * Copyright (c) 1985, 1988, 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. All advertising materials mentioning features or use of this software
46 *    must display the following acknowledgement:
47 * 	This product includes software developed by the University of
48 * 	California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 *    may be used to endorse or promote products derived from this software
51 *    without specific prior written permission.
52 *
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63 * SUCH DAMAGE.
64 * -
65 * Portions Copyright (c) 1993 by Digital Equipment Corporation.
66 *
67 * Permission to use, copy, modify, and distribute this software for any
68 * purpose with or without fee is hereby granted, provided that the above
69 * copyright notice and this permission notice appear in all copies, and that
70 * the name of Digital Equipment Corporation not be used in advertising or
71 * publicity pertaining to distribution of the document or software without
72 * specific, written prior permission.
73 *
74 * THE SOFTWARE IS PROVIDED "AS IS" AND DIGITAL EQUIPMENT CORP. DISCLAIMS ALL
75 * WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES
76 * OF MERCHANTABILITY AND FITNESS.   IN NO EVENT SHALL DIGITAL EQUIPMENT
77 * CORPORATION BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
78 * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
79 * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
80 * ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
81 * SOFTWARE.
82 * -
83 * --Copyright--
84 */
85
86/*
87 *	Atsushi Onoe <onoe@sm.sony.co.jp>
88 */
89
90#include <sys/cdefs.h>
91__FBSDID("$FreeBSD: head/lib/libc/net/name6.c 245225 2013-01-09 15:22:37Z ume $");
92
93#include "namespace.h"
94#include <sys/param.h>
95#include <sys/socket.h>
96#include <sys/time.h>
97#include <sys/queue.h>
98#include <netinet/in.h>
99#ifdef INET6
100#include <net/if.h>
101#include <net/if_var.h>
102#include <sys/sysctl.h>
103#include <sys/ioctl.h>
104#include <netinet6/in6_var.h>	/* XXX */
105#endif
106
107#include <arpa/inet.h>
108#include <arpa/nameser.h>
109
110#include <errno.h>
111#include <netdb.h>
112#include <resolv.h>
113#include <stdio.h>
114#include <stdlib.h>
115#include <string.h>
116#include <stdarg.h>
117#include <nsswitch.h>
118#include <unistd.h>
119#include "un-namespace.h"
120#include "netdb_private.h"
121#include "res_private.h"
122
123#ifndef MAXALIASES
124#define	MAXALIASES	10
125#endif
126#ifndef	MAXADDRS
127#define	MAXADDRS	20
128#endif
129#ifndef MAXDNAME
130#define	MAXDNAME	1025
131#endif
132
133#ifdef INET6
134#define	ADDRLEN(af)	((af) == AF_INET6 ? sizeof(struct in6_addr) : \
135					    sizeof(struct in_addr))
136#else
137#define	ADDRLEN(af)	sizeof(struct in_addr)
138#endif
139
140#define	MAPADDR(ab, ina) \
141do {									\
142	memcpy(&(ab)->map_inaddr, ina, sizeof(struct in_addr));		\
143	memset((ab)->map_zero, 0, sizeof((ab)->map_zero));		\
144	memset((ab)->map_one, 0xff, sizeof((ab)->map_one));		\
145} while (0)
146#define	MAPADDRENABLED(flags) \
147	(((flags) & AI_V4MAPPED) || \
148	 (((flags) & AI_V4MAPPED_CFG)))
149
150union inx_addr {
151	struct in_addr	in_addr;
152#ifdef INET6
153	struct in6_addr	in6_addr;
154#endif
155	struct {
156		u_char	mau_zero[10];
157		u_char	mau_one[2];
158		struct in_addr mau_inaddr;
159	}		map_addr_un;
160#define	map_zero	map_addr_un.mau_zero
161#define	map_one		map_addr_un.mau_one
162#define	map_inaddr	map_addr_un.mau_inaddr
163};
164
165struct policyqueue {
166	TAILQ_ENTRY(policyqueue) pc_entry;
167#ifdef INET6
168	struct in6_addrpolicy pc_policy;
169#endif
170};
171TAILQ_HEAD(policyhead, policyqueue);
172
173#define AIO_SRCFLAG_DEPRECATED	0x1
174
175struct hp_order {
176	union {
177		struct sockaddr_storage aiou_ss;
178		struct sockaddr aiou_sa;
179	} aio_src_un;
180#define aio_srcsa aio_src_un.aiou_sa
181	u_int32_t aio_srcflag;
182	int aio_srcscope;
183	int aio_dstscope;
184	struct policyqueue *aio_srcpolicy;
185	struct policyqueue *aio_dstpolicy;
186	union {
187		struct sockaddr_storage aiou_ss;
188		struct sockaddr aiou_sa;
189	} aio_un;
190#define aio_sa aio_un.aiou_sa
191	int aio_matchlen;
192	char *aio_h_addr;
193};
194
195static struct	 hostent *_hpcopy(struct hostent *, int *);
196static struct	 hostent *_hpaddr(int, const char *, void *, int *);
197#ifdef INET6
198static struct	 hostent *_hpmerge(struct hostent *, struct hostent *, int *);
199static struct	 hostent *_hpmapv6(struct hostent *, int *);
200#endif
201static struct	 hostent *_hpsort(struct hostent *, res_state);
202
203#ifdef ENABLE_IP6ADDRCTL
204static struct	 hostent *_hpreorder(struct hostent *);
205static int	 get_addrselectpolicy(struct policyhead *);
206static void	 free_addrselectpolicy(struct policyhead *);
207static struct	 policyqueue *match_addrselectpolicy(struct sockaddr *,
208	struct policyhead *);
209static void	 set_source(struct hp_order *, struct policyhead *);
210static int	 matchlen(struct sockaddr *, struct sockaddr *);
211static int	 comp_dst(const void *, const void *);
212static int	 gai_addr2scopetype(struct sockaddr *);
213#endif
214
215/*
216 * Functions defined in RFC2553
217 *	getipnodebyname, getipnodebyaddr, freehostent
218 */
219
220struct hostent *
221getipnodebyname(const char *name, int af, int flags, int *errp)
222{
223	struct hostent *hp;
224	union inx_addr addrbuf;
225	res_state statp;
226	u_long options;
227
228	switch (af) {
229	case AF_INET:
230#ifdef INET6
231	case AF_INET6:
232#endif
233		break;
234	default:
235		*errp = NO_RECOVERY;
236		return NULL;
237	}
238
239	if (flags & AI_ADDRCONFIG) {
240		int s;
241
242		if ((s = _socket(af, SOCK_DGRAM, 0)) < 0)
243			return NULL;
244		/*
245		 * TODO:
246		 * Note that implementation dependent test for address
247		 * configuration should be done everytime called
248		 * (or apropriate interval),
249		 * because addresses will be dynamically assigned or deleted.
250		 */
251		_close(s);
252	}
253
254#ifdef INET6
255	/* special case for literal address */
256	if (inet_pton(AF_INET6, name, &addrbuf) == 1) {
257		if (af != AF_INET6) {
258			*errp = HOST_NOT_FOUND;
259			return NULL;
260		}
261		return _hpaddr(af, name, &addrbuf, errp);
262	}
263#endif
264	if (inet_aton(name, (struct in_addr *)&addrbuf) == 1) {
265		if (af != AF_INET) {
266			if (MAPADDRENABLED(flags)) {
267				MAPADDR(&addrbuf, &addrbuf.in_addr);
268			} else {
269				*errp = HOST_NOT_FOUND;
270				return NULL;
271			}
272		}
273		return _hpaddr(af, name, &addrbuf, errp);
274	}
275
276
277	statp = __res_state();
278	if ((statp->options & RES_INIT) == 0) {
279		if (res_ninit(statp) < 0) {
280			*errp = NETDB_INTERNAL;
281			return NULL;
282		}
283	}
284
285	options = statp->options;
286	statp->options &= ~RES_USE_INET6;
287
288	hp = gethostbyname2(name, af);
289	hp = _hpcopy(hp, errp);
290
291#ifdef INET6
292	if (af == AF_INET6 && ((flags & AI_ALL) || hp == NULL) &&
293	    MAPADDRENABLED(flags)) {
294		struct hostent *hp2 = gethostbyname2(name, AF_INET);
295		if (hp == NULL)
296			if (hp2 == NULL)
297				*errp = statp->res_h_errno;
298			else
299				hp = _hpmapv6(hp2, errp);
300		else {
301			if (hp2 && strcmp(hp->h_name, hp2->h_name) == 0) {
302				struct hostent *hpb = hp;
303				hp = _hpmerge(hpb, hp2, errp);
304				freehostent(hpb);
305			}
306		}
307	}
308#endif
309
310	if (hp == NULL)
311		*errp = statp->res_h_errno;
312
313	statp->options = options;
314#ifdef ENABLE_IP6ADDRCTL
315	return _hpreorder(_hpsort(hp, statp));
316#else
317	return _hpsort(hp, statp);
318#endif
319}
320
321struct hostent *
322getipnodebyaddr(const void *src, size_t len, int af, int *errp)
323{
324	struct hostent *hp;
325	res_state statp;
326	u_long options;
327
328#ifdef INET6
329	struct in6_addr addrbuf;
330#else
331	struct in_addr addrbuf;
332#endif
333
334	switch (af) {
335	case AF_INET:
336		if (len != sizeof(struct in_addr)) {
337			*errp = NO_RECOVERY;
338			return NULL;
339		}
340		if ((long)src & ~(sizeof(struct in_addr) - 1)) {
341			memcpy(&addrbuf, src, len);
342			src = &addrbuf;
343		}
344		if (((struct in_addr *)src)->s_addr == 0)
345			return NULL;
346		break;
347#ifdef INET6
348	case AF_INET6:
349		if (len != sizeof(struct in6_addr)) {
350			*errp = NO_RECOVERY;
351			return NULL;
352		}
353		if ((long)src & ~(sizeof(struct in6_addr) / 2 - 1)) {	/*XXX*/
354			memcpy(&addrbuf, src, len);
355			src = &addrbuf;
356		}
357		if (IN6_IS_ADDR_UNSPECIFIED((struct in6_addr *)src))
358			return NULL;
359		if (IN6_IS_ADDR_V4MAPPED((struct in6_addr *)src)
360		||  IN6_IS_ADDR_V4COMPAT((struct in6_addr *)src)) {
361			src = (char *)src +
362			    (sizeof(struct in6_addr) - sizeof(struct in_addr));
363			af = AF_INET;
364			len = sizeof(struct in_addr);
365		}
366		break;
367#endif
368	default:
369		*errp = NO_RECOVERY;
370		return NULL;
371	}
372
373	statp = __res_state();
374	if ((statp->options & RES_INIT) == 0) {
375		if (res_ninit(statp) < 0) {
376			RES_SET_H_ERRNO(statp, NETDB_INTERNAL);
377			return NULL;
378		}
379	}
380
381	options = statp->options;
382	statp->options &= ~RES_USE_INET6;
383
384	hp = gethostbyaddr(src, len, af);
385	if (hp == NULL)
386		*errp = statp->res_h_errno;
387
388	statp->options = options;
389	return (_hpcopy(hp, errp));
390}
391
392void
393freehostent(struct hostent *ptr)
394{
395	free(ptr);
396}
397
398/*
399 * Private utility functions
400 */
401
402/*
403 * _hpcopy: allocate and copy hostent structure
404 */
405static struct hostent *
406_hpcopy(struct hostent *hp, int *errp)
407{
408	struct hostent *nhp;
409	char *cp, **pp;
410	int size, addrsize;
411	int nalias = 0, naddr = 0;
412	int al_off;
413	int i;
414
415	if (hp == NULL)
416		return hp;
417
418	/* count size to be allocated */
419	size = sizeof(struct hostent);
420	if (hp->h_name != NULL)
421		size += strlen(hp->h_name) + 1;
422	if ((pp = hp->h_aliases) != NULL) {
423		for (i = 0; *pp != NULL; i++, pp++) {
424			if (**pp != '\0') {
425				size += strlen(*pp) + 1;
426				nalias++;
427			}
428		}
429	}
430	/* adjust alignment */
431	size = ALIGN(size);
432	al_off = size;
433	size += sizeof(char *) * (nalias + 1);
434	addrsize = ALIGN(hp->h_length);
435	if ((pp = hp->h_addr_list) != NULL) {
436		while (*pp++ != NULL)
437			naddr++;
438	}
439	size += addrsize * naddr;
440	size += sizeof(char *) * (naddr + 1);
441
442	/* copy */
443	if ((nhp = (struct hostent *)malloc(size)) == NULL) {
444		*errp = TRY_AGAIN;
445		return NULL;
446	}
447	cp = (char *)&nhp[1];
448	if (hp->h_name != NULL) {
449		nhp->h_name = cp;
450		strcpy(cp, hp->h_name);
451		cp += strlen(cp) + 1;
452	} else
453		nhp->h_name = NULL;
454	nhp->h_aliases = (char **)((char *)nhp + al_off);
455	if ((pp = hp->h_aliases) != NULL) {
456		for (i = 0; *pp != NULL; pp++) {
457			if (**pp != '\0') {
458				nhp->h_aliases[i++] = cp;
459				strcpy(cp, *pp);
460				cp += strlen(cp) + 1;
461			}
462		}
463	}
464	nhp->h_aliases[nalias] = NULL;
465	cp = (char *)&nhp->h_aliases[nalias + 1];
466	nhp->h_addrtype = hp->h_addrtype;
467	nhp->h_length = hp->h_length;
468	nhp->h_addr_list = (char **)cp;
469	if ((pp = hp->h_addr_list) != NULL) {
470		cp = (char *)&nhp->h_addr_list[naddr + 1];
471		for (i = 0; *pp != NULL; pp++) {
472			nhp->h_addr_list[i++] = cp;
473			memcpy(cp, *pp, hp->h_length);
474			cp += addrsize;
475		}
476	}
477	nhp->h_addr_list[naddr] = NULL;
478	return nhp;
479}
480
481/*
482 * _hpaddr: construct hostent structure with one address
483 */
484static struct hostent *
485_hpaddr(int af, const char *name, void *addr, int *errp)
486{
487	struct hostent *hp, hpbuf;
488	char *addrs[2];
489
490	hp = &hpbuf;
491	hp->h_name = (char *)name;
492	hp->h_aliases = NULL;
493	hp->h_addrtype = af;
494	hp->h_length = ADDRLEN(af);
495	hp->h_addr_list = addrs;
496	addrs[0] = (char *)addr;
497	addrs[1] = NULL;
498	return (_hpcopy(hp, errp));
499}
500
501#ifdef INET6
502/*
503 * _hpmerge: merge 2 hostent structure, arguments will be freed
504 */
505static struct hostent *
506_hpmerge(struct hostent *hp1, struct hostent *hp2, int *errp)
507{
508	int i, j;
509	int naddr, nalias;
510	char **pp;
511	struct hostent *hp, hpbuf;
512	char *aliases[MAXALIASES + 1], *addrs[MAXADDRS + 1];
513	union inx_addr addrbuf[MAXADDRS];
514
515	if (hp1 == NULL)
516		return _hpcopy(hp2, errp);
517	if (hp2 == NULL)
518		return _hpcopy(hp1, errp);
519
520#define	HP(i)	(i == 1 ? hp1 : hp2)
521	hp = &hpbuf;
522	hp->h_name = (hp1->h_name != NULL ? hp1->h_name : hp2->h_name);
523	hp->h_aliases = aliases;
524	nalias = 0;
525	for (i = 1; i <= 2; i++) {
526		if ((pp = HP(i)->h_aliases) == NULL)
527			continue;
528		for (; nalias < MAXALIASES && *pp != NULL; pp++) {
529			/* check duplicates */
530			for (j = 0; j < nalias; j++)
531				if (strcasecmp(*pp, aliases[j]) == 0)
532					break;
533			if (j == nalias)
534				aliases[nalias++] = *pp;
535		}
536	}
537	aliases[nalias] = NULL;
538	if (hp1->h_length != hp2->h_length) {
539		hp->h_addrtype = AF_INET6;
540		hp->h_length = sizeof(struct in6_addr);
541	} else {
542		hp->h_addrtype = hp1->h_addrtype;
543		hp->h_length = hp1->h_length;
544	}
545
546	hp->h_addr_list = addrs;
547	naddr = 0;
548	for (i = 1; i <= 2; i++) {
549		if ((pp = HP(i)->h_addr_list) == NULL)
550			continue;
551		if (HP(i)->h_length == hp->h_length) {
552			while (naddr < MAXADDRS && *pp != NULL)
553				addrs[naddr++] = *pp++;
554		} else {
555			/* copy IPv4 addr as mapped IPv6 addr */
556			while (naddr < MAXADDRS && *pp != NULL) {
557				MAPADDR(&addrbuf[naddr], *pp++);
558				addrs[naddr] = (char *)&addrbuf[naddr];
559				naddr++;
560			}
561		}
562	}
563	addrs[naddr] = NULL;
564	return (_hpcopy(hp, errp));
565}
566#endif
567
568/*
569 * _hpmapv6: convert IPv4 hostent into IPv4-mapped IPv6 addresses
570 */
571#ifdef INET6
572static struct hostent *
573_hpmapv6(struct hostent *hp, int *errp)
574{
575	struct hostent hp6;
576
577	if (hp == NULL)
578		return NULL;
579	if (hp->h_addrtype == AF_INET6)
580		return _hpcopy(hp, errp);
581
582	memset(&hp6, 0, sizeof(struct hostent));
583	hp6.h_addrtype = AF_INET6;
584	hp6.h_length = sizeof(struct in6_addr);
585	return _hpmerge(&hp6, hp, errp);
586}
587#endif
588
589/*
590 * _hpsort: sort address by sortlist
591 */
592static struct hostent *
593_hpsort(struct hostent *hp, res_state statp)
594{
595	int i, j, n;
596	u_char *ap, *sp, *mp, **pp;
597	char t;
598	char order[MAXADDRS];
599	int nsort = statp->nsort;
600
601	if (hp == NULL || hp->h_addr_list[1] == NULL || nsort == 0)
602		return hp;
603	for (i = 0; (ap = (u_char *)hp->h_addr_list[i]); i++) {
604		for (j = 0; j < nsort; j++) {
605#ifdef INET6
606			if (statp->_u._ext.ext->sort_list[j].af !=
607			    hp->h_addrtype)
608				continue;
609			sp = (u_char *)&statp->_u._ext.ext->sort_list[j].addr;
610			mp = (u_char *)&statp->_u._ext.ext->sort_list[j].mask;
611#else
612			sp = (u_char *)&statp->sort_list[j].addr;
613			mp = (u_char *)&statp->sort_list[j].mask;
614#endif
615			for (n = 0; n < hp->h_length; n++) {
616				if ((ap[n] & mp[n]) != sp[n])
617					break;
618			}
619			if (n == hp->h_length)
620				break;
621		}
622		order[i] = j;
623	}
624	n = i;
625	pp = (u_char **)hp->h_addr_list;
626	for (i = 0; i < n - 1; i++) {
627		for (j = i + 1; j < n; j++) {
628			if (order[i] > order[j]) {
629				ap = pp[i];
630				pp[i] = pp[j];
631				pp[j] = ap;
632				t = order[i];
633				order[i] = order[j];
634				order[j] = t;
635			}
636		}
637	}
638	return hp;
639}
640
641#ifdef ENABLE_IP6ADDRCTL
642/*
643 * _hpreorder: sort address by default address selection
644 */
645static struct hostent *
646_hpreorder(struct hostent *hp)
647{
648	struct hp_order *aio;
649	int i, n;
650	char *ap;
651	struct sockaddr *sa;
652	struct policyhead policyhead;
653
654	if (hp == NULL)
655		return hp;
656
657	switch (hp->h_addrtype) {
658	case AF_INET:
659#ifdef INET6
660	case AF_INET6:
661#endif
662		break;
663	default:
664		free_addrselectpolicy(&policyhead);
665		return hp;
666	}
667
668	/* count the number of addrinfo elements for sorting. */
669	for (n = 0; hp->h_addr_list[n] != NULL; n++)
670		;
671
672	/*
673	 * If the number is small enough, we can skip the reordering process.
674	 */
675	if (n <= 1)
676		return hp;
677
678	/* allocate a temporary array for sort and initialization of it. */
679	if ((aio = malloc(sizeof(*aio) * n)) == NULL)
680		return hp;	/* give up reordering */
681	memset(aio, 0, sizeof(*aio) * n);
682
683	/* retrieve address selection policy from the kernel */
684	TAILQ_INIT(&policyhead);
685	if (!get_addrselectpolicy(&policyhead)) {
686		/* no policy is installed into kernel, we don't sort. */
687		free(aio);
688		return hp;
689	}
690
691	for (i = 0; i < n; i++) {
692		ap = hp->h_addr_list[i];
693		aio[i].aio_h_addr = ap;
694		sa = &aio[i].aio_sa;
695		switch (hp->h_addrtype) {
696		case AF_INET:
697			sa->sa_family = AF_INET;
698			sa->sa_len = sizeof(struct sockaddr_in);
699			memcpy(&((struct sockaddr_in *)sa)->sin_addr, ap,
700			    sizeof(struct in_addr));
701			break;
702#ifdef INET6
703		case AF_INET6:
704			if (IN6_IS_ADDR_V4MAPPED((struct in6_addr *)ap)) {
705				sa->sa_family = AF_INET;
706				sa->sa_len = sizeof(struct sockaddr_in);
707				memcpy(&((struct sockaddr_in *)sa)->sin_addr,
708				    &ap[12], sizeof(struct in_addr));
709			} else {
710				sa->sa_family = AF_INET6;
711				sa->sa_len = sizeof(struct sockaddr_in6);
712				memcpy(&((struct sockaddr_in6 *)sa)->sin6_addr,
713				    ap, sizeof(struct in6_addr));
714			}
715			break;
716#endif
717		}
718		aio[i].aio_dstscope = gai_addr2scopetype(sa);
719		aio[i].aio_dstpolicy = match_addrselectpolicy(sa, &policyhead);
720		set_source(&aio[i], &policyhead);
721	}
722
723	/* perform sorting. */
724	qsort(aio, n, sizeof(*aio), comp_dst);
725
726	/* reorder the h_addr_list. */
727	for (i = 0; i < n; i++)
728		hp->h_addr_list[i] = aio[i].aio_h_addr;
729
730	/* cleanup and return */
731	free(aio);
732	free_addrselectpolicy(&policyhead);
733	return hp;
734}
735
736static int
737get_addrselectpolicy(struct policyhead *head)
738{
739#ifdef INET6
740	int mib[] = { CTL_NET, PF_INET6, IPPROTO_IPV6, IPV6CTL_ADDRCTLPOLICY };
741	size_t l;
742	char *buf;
743	struct in6_addrpolicy *pol, *ep;
744
745	if (sysctl(mib, sizeof(mib) / sizeof(mib[0]), NULL, &l, NULL, 0) < 0)
746		return (0);
747	if ((buf = malloc(l)) == NULL)
748		return (0);
749	if (sysctl(mib, sizeof(mib) / sizeof(mib[0]), buf, &l, NULL, 0) < 0) {
750		free(buf);
751		return (0);
752	}
753
754	ep = (struct in6_addrpolicy *)(buf + l);
755	for (pol = (struct in6_addrpolicy *)buf; pol + 1 <= ep; pol++) {
756		struct policyqueue *new;
757
758		if ((new = malloc(sizeof(*new))) == NULL) {
759			free_addrselectpolicy(head); /* make the list empty */
760			break;
761		}
762		new->pc_policy = *pol;
763		TAILQ_INSERT_TAIL(head, new, pc_entry);
764	}
765
766	free(buf);
767	return (1);
768#else
769	return (0);
770#endif
771}
772
773static void
774free_addrselectpolicy(struct policyhead *head)
775{
776	struct policyqueue *ent, *nent;
777
778	for (ent = TAILQ_FIRST(head); ent; ent = nent) {
779		nent = TAILQ_NEXT(ent, pc_entry);
780		TAILQ_REMOVE(head, ent, pc_entry);
781		free(ent);
782	}
783}
784
785static struct policyqueue *
786match_addrselectpolicy(struct sockaddr *addr, struct policyhead *head)
787{
788#ifdef INET6
789	struct policyqueue *ent, *bestent = NULL;
790	struct in6_addrpolicy *pol;
791	int matchlen, bestmatchlen = -1;
792	u_char *mp, *ep, *k, *p, m;
793	struct sockaddr_in6 key;
794
795	switch(addr->sa_family) {
796	case AF_INET6:
797		key = *(struct sockaddr_in6 *)addr;
798		break;
799	case AF_INET:
800		/* convert the address into IPv4-mapped IPv6 address. */
801		memset(&key, 0, sizeof(key));
802		key.sin6_family = AF_INET6;
803		key.sin6_len = sizeof(key);
804		key.sin6_addr.s6_addr[10] = 0xff;
805		key.sin6_addr.s6_addr[11] = 0xff;
806		memcpy(&key.sin6_addr.s6_addr[12],
807		       &((struct sockaddr_in *)addr)->sin_addr, 4);
808		break;
809	default:
810		return(NULL);
811	}
812
813	for (ent = TAILQ_FIRST(head); ent; ent = TAILQ_NEXT(ent, pc_entry)) {
814		pol = &ent->pc_policy;
815		matchlen = 0;
816
817		mp = (u_char *)&pol->addrmask.sin6_addr;
818		ep = mp + 16;	/* XXX: scope field? */
819		k = (u_char *)&key.sin6_addr;
820		p = (u_char *)&pol->addr.sin6_addr;
821		for (; mp < ep && *mp; mp++, k++, p++) {
822			m = *mp;
823			if ((*k & m) != *p)
824				goto next; /* not match */
825			if (m == 0xff) /* short cut for a typical case */
826				matchlen += 8;
827			else {
828				while (m >= 0x80) {
829					matchlen++;
830					m <<= 1;
831				}
832			}
833		}
834
835		/* matched.  check if this is better than the current best. */
836		if (matchlen > bestmatchlen) {
837			bestent = ent;
838			bestmatchlen = matchlen;
839		}
840
841	  next:
842		continue;
843	}
844
845	return(bestent);
846#else
847	return(NULL);
848#endif
849
850}
851
852static void
853set_source(struct hp_order *aio, struct policyhead *ph)
854{
855	struct sockaddr_storage ss = aio->aio_un.aiou_ss;
856	socklen_t srclen;
857	int s;
858
859	/* set unspec ("no source is available"), just in case */
860	aio->aio_srcsa.sa_family = AF_UNSPEC;
861	aio->aio_srcscope = -1;
862
863	switch(ss.ss_family) {
864	case AF_INET:
865		((struct sockaddr_in *)&ss)->sin_port = htons(1);
866		break;
867#ifdef INET6
868	case AF_INET6:
869		((struct sockaddr_in6 *)&ss)->sin6_port = htons(1);
870		break;
871#endif
872	default:		/* ignore unsupported AFs explicitly */
873		return;
874	}
875
876	/* open a socket to get the source address for the given dst */
877	if ((s = _socket(ss.ss_family, SOCK_DGRAM, IPPROTO_UDP)) < 0)
878		return;		/* give up */
879	if (_connect(s, (struct sockaddr *)&ss, ss.ss_len) < 0)
880		goto cleanup;
881	srclen = ss.ss_len;
882	if (_getsockname(s, &aio->aio_srcsa, &srclen) < 0) {
883		aio->aio_srcsa.sa_family = AF_UNSPEC;
884		goto cleanup;
885	}
886	aio->aio_srcscope = gai_addr2scopetype(&aio->aio_srcsa);
887	aio->aio_srcpolicy = match_addrselectpolicy(&aio->aio_srcsa, ph);
888	aio->aio_matchlen = matchlen(&aio->aio_srcsa, (struct sockaddr *)&ss);
889#ifdef INET6
890	if (ss.ss_family == AF_INET6) {
891		struct in6_ifreq ifr6;
892		u_int32_t flags6;
893
894		memset(&ifr6, 0, sizeof(ifr6));
895		memcpy(&ifr6.ifr_addr, &ss, ss.ss_len);
896		if (_ioctl(s, SIOCGIFAFLAG_IN6, &ifr6) == 0) {
897			flags6 = ifr6.ifr_ifru.ifru_flags6;
898			if ((flags6 & IN6_IFF_DEPRECATED))
899				aio->aio_srcflag |= AIO_SRCFLAG_DEPRECATED;
900		}
901	}
902#endif
903
904  cleanup:
905	_close(s);
906	return;
907}
908
909static int
910matchlen(struct sockaddr *src, struct sockaddr *dst)
911{
912	int match = 0;
913	u_char *s, *d;
914	u_char *lim, r;
915	int addrlen;
916
917	switch (src->sa_family) {
918#ifdef INET6
919	case AF_INET6:
920		s = (u_char *)&((struct sockaddr_in6 *)src)->sin6_addr;
921		d = (u_char *)&((struct sockaddr_in6 *)dst)->sin6_addr;
922		addrlen = sizeof(struct in6_addr);
923		lim = s + addrlen;
924		break;
925#endif
926	case AF_INET:
927		s = (u_char *)&((struct sockaddr_in *)src)->sin_addr;
928		d = (u_char *)&((struct sockaddr_in *)dst)->sin_addr;
929		addrlen = sizeof(struct in_addr);
930		lim = s + addrlen;
931		break;
932	default:
933		return(0);
934	}
935
936	while (s < lim)
937		if ((r = (*d++ ^ *s++)) != 0) {
938			while (r < addrlen * 8) {
939				match++;
940				r <<= 1;
941			}
942			break;
943		} else
944			match += 8;
945	return(match);
946}
947
948static int
949comp_dst(const void *arg1, const void *arg2)
950{
951	const struct hp_order *dst1 = arg1, *dst2 = arg2;
952
953	/*
954	 * Rule 1: Avoid unusable destinations.
955	 * XXX: we currently do not consider if an appropriate route exists.
956	 */
957	if (dst1->aio_srcsa.sa_family != AF_UNSPEC &&
958	    dst2->aio_srcsa.sa_family == AF_UNSPEC) {
959		return(-1);
960	}
961	if (dst1->aio_srcsa.sa_family == AF_UNSPEC &&
962	    dst2->aio_srcsa.sa_family != AF_UNSPEC) {
963		return(1);
964	}
965
966	/* Rule 2: Prefer matching scope. */
967	if (dst1->aio_dstscope == dst1->aio_srcscope &&
968	    dst2->aio_dstscope != dst2->aio_srcscope) {
969		return(-1);
970	}
971	if (dst1->aio_dstscope != dst1->aio_srcscope &&
972	    dst2->aio_dstscope == dst2->aio_srcscope) {
973		return(1);
974	}
975
976	/* Rule 3: Avoid deprecated addresses. */
977	if (dst1->aio_srcsa.sa_family != AF_UNSPEC &&
978	    dst2->aio_srcsa.sa_family != AF_UNSPEC) {
979		if (!(dst1->aio_srcflag & AIO_SRCFLAG_DEPRECATED) &&
980		    (dst2->aio_srcflag & AIO_SRCFLAG_DEPRECATED)) {
981			return(-1);
982		}
983		if ((dst1->aio_srcflag & AIO_SRCFLAG_DEPRECATED) &&
984		    !(dst2->aio_srcflag & AIO_SRCFLAG_DEPRECATED)) {
985			return(1);
986		}
987	}
988
989	/* Rule 4: Prefer home addresses. */
990	/* XXX: not implemented yet */
991
992	/* Rule 5: Prefer matching label. */
993#ifdef INET6
994	if (dst1->aio_srcpolicy && dst1->aio_dstpolicy &&
995	    dst1->aio_srcpolicy->pc_policy.label ==
996	    dst1->aio_dstpolicy->pc_policy.label &&
997	    (dst2->aio_srcpolicy == NULL || dst2->aio_dstpolicy == NULL ||
998	     dst2->aio_srcpolicy->pc_policy.label !=
999	     dst2->aio_dstpolicy->pc_policy.label)) {
1000		return(-1);
1001	}
1002	if (dst2->aio_srcpolicy && dst2->aio_dstpolicy &&
1003	    dst2->aio_srcpolicy->pc_policy.label ==
1004	    dst2->aio_dstpolicy->pc_policy.label &&
1005	    (dst1->aio_srcpolicy == NULL || dst1->aio_dstpolicy == NULL ||
1006	     dst1->aio_srcpolicy->pc_policy.label !=
1007	     dst1->aio_dstpolicy->pc_policy.label)) {
1008		return(1);
1009	}
1010#endif
1011
1012	/* Rule 6: Prefer higher precedence. */
1013#ifdef INET6
1014	if (dst1->aio_dstpolicy &&
1015	    (dst2->aio_dstpolicy == NULL ||
1016	     dst1->aio_dstpolicy->pc_policy.preced >
1017	     dst2->aio_dstpolicy->pc_policy.preced)) {
1018		return(-1);
1019	}
1020	if (dst2->aio_dstpolicy &&
1021	    (dst1->aio_dstpolicy == NULL ||
1022	     dst2->aio_dstpolicy->pc_policy.preced >
1023	     dst1->aio_dstpolicy->pc_policy.preced)) {
1024		return(1);
1025	}
1026#endif
1027
1028	/* Rule 7: Prefer native transport. */
1029	/* XXX: not implemented yet */
1030
1031	/* Rule 8: Prefer smaller scope. */
1032	if (dst1->aio_dstscope >= 0 &&
1033	    dst1->aio_dstscope < dst2->aio_dstscope) {
1034		return(-1);
1035	}
1036	if (dst2->aio_dstscope >= 0 &&
1037	    dst2->aio_dstscope < dst1->aio_dstscope) {
1038		return(1);
1039	}
1040
1041	/*
1042	 * Rule 9: Use longest matching prefix.
1043	 * We compare the match length in a same AF only.
1044	 */
1045	if (dst1->aio_sa.sa_family == dst2->aio_sa.sa_family) {
1046		if (dst1->aio_matchlen > dst2->aio_matchlen) {
1047			return(-1);
1048		}
1049		if (dst1->aio_matchlen < dst2->aio_matchlen) {
1050			return(1);
1051		}
1052	}
1053
1054	/* Rule 10: Otherwise, leave the order unchanged. */
1055	return(-1);
1056}
1057
1058/*
1059 * Copy from scope.c.
1060 * XXX: we should standardize the functions and link them as standard
1061 * library.
1062 */
1063static int
1064gai_addr2scopetype(struct sockaddr *sa)
1065{
1066#ifdef INET6
1067	struct sockaddr_in6 *sa6;
1068#endif
1069	struct sockaddr_in *sa4;
1070
1071	switch(sa->sa_family) {
1072#ifdef INET6
1073	case AF_INET6:
1074		sa6 = (struct sockaddr_in6 *)sa;
1075		if (IN6_IS_ADDR_MULTICAST(&sa6->sin6_addr)) {
1076			/* just use the scope field of the multicast address */
1077			return(sa6->sin6_addr.s6_addr[2] & 0x0f);
1078		}
1079		/*
1080		 * Unicast addresses: map scope type to corresponding scope
1081		 * value defined for multcast addresses.
1082		 * XXX: hardcoded scope type values are bad...
1083		 */
1084		if (IN6_IS_ADDR_LOOPBACK(&sa6->sin6_addr))
1085			return(1); /* node local scope */
1086		if (IN6_IS_ADDR_LINKLOCAL(&sa6->sin6_addr))
1087			return(2); /* link-local scope */
1088		if (IN6_IS_ADDR_SITELOCAL(&sa6->sin6_addr))
1089			return(5); /* site-local scope */
1090		return(14);	/* global scope */
1091		break;
1092#endif
1093	case AF_INET:
1094		/*
1095		 * IPv4 pseudo scoping according to RFC 3484.
1096		 */
1097		sa4 = (struct sockaddr_in *)sa;
1098		/* IPv4 autoconfiguration addresses have link-local scope. */
1099		if (((u_char *)&sa4->sin_addr)[0] == 169 &&
1100		    ((u_char *)&sa4->sin_addr)[1] == 254)
1101			return(2);
1102		/* Private addresses have site-local scope. */
1103		if (((u_char *)&sa4->sin_addr)[0] == 10 ||
1104		    (((u_char *)&sa4->sin_addr)[0] == 172 &&
1105		     (((u_char *)&sa4->sin_addr)[1] & 0xf0) == 16) ||
1106		    (((u_char *)&sa4->sin_addr)[0] == 192 &&
1107		     ((u_char *)&sa4->sin_addr)[1] == 168))
1108			return(14);	/* XXX: It should be 5 unless NAT */
1109		/* Loopback addresses have link-local scope. */
1110		if (((u_char *)&sa4->sin_addr)[0] == 127)
1111			return(2);
1112		return(14);
1113		break;
1114	default:
1115		errno = EAFNOSUPPORT; /* is this a good error? */
1116		return(-1);
1117	}
1118}
1119#endif
1120