name6.c revision 251071
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. 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 * Portions Copyright (c) 1993 by Digital Equipment Corporation.
62 *
63 * Permission to use, copy, modify, and distribute this software for any
64 * purpose with or without fee is hereby granted, provided that the above
65 * copyright notice and this permission notice appear in all copies, and that
66 * the name of Digital Equipment Corporation not be used in advertising or
67 * publicity pertaining to distribution of the document or software without
68 * specific, written prior permission.
69 *
70 * THE SOFTWARE IS PROVIDED "AS IS" AND DIGITAL EQUIPMENT CORP. DISCLAIMS ALL
71 * WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES
72 * OF MERCHANTABILITY AND FITNESS.   IN NO EVENT SHALL DIGITAL EQUIPMENT
73 * CORPORATION BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
74 * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
75 * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
76 * ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
77 * SOFTWARE.
78 * -
79 * --Copyright--
80 */
81
82/*
83 *	Atsushi Onoe <onoe@sm.sony.co.jp>
84 */
85
86#include <sys/cdefs.h>
87__FBSDID("$FreeBSD: head/lib/libc/net/name6.c 251071 2013-05-28 21:05:06Z emaste $");
88
89#include "namespace.h"
90#include <sys/param.h>
91#include <sys/socket.h>
92#include <sys/time.h>
93#include <sys/queue.h>
94#include <netinet/in.h>
95#ifdef INET6
96#include <net/if.h>
97#include <net/if_var.h>
98#include <sys/sysctl.h>
99#include <sys/ioctl.h>
100#include <netinet6/in6_var.h>	/* XXX */
101#endif
102
103#include <arpa/inet.h>
104#include <arpa/nameser.h>
105
106#include <errno.h>
107#include <netdb.h>
108#include <resolv.h>
109#include <stdio.h>
110#include <stdlib.h>
111#include <string.h>
112#include <stdarg.h>
113#include <nsswitch.h>
114#include <unistd.h>
115#include "un-namespace.h"
116#include "netdb_private.h"
117#include "res_private.h"
118
119#ifndef MAXALIASES
120#define	MAXALIASES	10
121#endif
122#ifndef	MAXADDRS
123#define	MAXADDRS	20
124#endif
125#ifndef MAXDNAME
126#define	MAXDNAME	1025
127#endif
128
129#ifdef INET6
130#define	ADDRLEN(af)	((af) == AF_INET6 ? sizeof(struct in6_addr) : \
131					    sizeof(struct in_addr))
132#else
133#define	ADDRLEN(af)	sizeof(struct in_addr)
134#endif
135
136#define	MAPADDR(ab, ina) \
137do {									\
138	memcpy(&(ab)->map_inaddr, ina, sizeof(struct in_addr));		\
139	memset((ab)->map_zero, 0, sizeof((ab)->map_zero));		\
140	memset((ab)->map_one, 0xff, sizeof((ab)->map_one));		\
141} while (0)
142#define	MAPADDRENABLED(flags) \
143	(((flags) & AI_V4MAPPED) || \
144	 (((flags) & AI_V4MAPPED_CFG)))
145
146union inx_addr {
147	struct in_addr	in_addr;
148#ifdef INET6
149	struct in6_addr	in6_addr;
150#endif
151	struct {
152		u_char	mau_zero[10];
153		u_char	mau_one[2];
154		struct in_addr mau_inaddr;
155	}		map_addr_un;
156#define	map_zero	map_addr_un.mau_zero
157#define	map_one		map_addr_un.mau_one
158#define	map_inaddr	map_addr_un.mau_inaddr
159};
160
161struct policyqueue {
162	TAILQ_ENTRY(policyqueue) pc_entry;
163#ifdef INET6
164	struct in6_addrpolicy pc_policy;
165#endif
166};
167TAILQ_HEAD(policyhead, policyqueue);
168
169#define AIO_SRCFLAG_DEPRECATED	0x1
170
171struct hp_order {
172	union {
173		struct sockaddr_storage aiou_ss;
174		struct sockaddr aiou_sa;
175	} aio_src_un;
176#define aio_srcsa aio_src_un.aiou_sa
177	u_int32_t aio_srcflag;
178	int aio_srcscope;
179	int aio_dstscope;
180	struct policyqueue *aio_srcpolicy;
181	struct policyqueue *aio_dstpolicy;
182	union {
183		struct sockaddr_storage aiou_ss;
184		struct sockaddr aiou_sa;
185	} aio_un;
186#define aio_sa aio_un.aiou_sa
187	int aio_matchlen;
188	char *aio_h_addr;
189};
190
191static struct	 hostent *_hpcopy(struct hostent *, int *);
192static struct	 hostent *_hpaddr(int, const char *, void *, int *);
193#ifdef INET6
194static struct	 hostent *_hpmerge(struct hostent *, struct hostent *, int *);
195static struct	 hostent *_hpmapv6(struct hostent *, int *);
196#endif
197static struct	 hostent *_hpsort(struct hostent *, res_state);
198
199#ifdef INET6
200static struct	 hostent *_hpreorder(struct hostent *);
201static int	 get_addrselectpolicy(struct policyhead *);
202static void	 free_addrselectpolicy(struct policyhead *);
203static struct	 policyqueue *match_addrselectpolicy(struct sockaddr *,
204	struct policyhead *);
205static void	 set_source(struct hp_order *, struct policyhead *);
206static int	 matchlen(struct sockaddr *, struct sockaddr *);
207static int	 comp_dst(const void *, const void *);
208static int	 gai_addr2scopetype(struct sockaddr *);
209#endif
210
211/*
212 * Functions defined in RFC2553
213 *	getipnodebyname, getipnodebyaddr, freehostent
214 */
215
216struct hostent *
217getipnodebyname(const char *name, int af, int flags, int *errp)
218{
219	struct hostent *hp;
220	union inx_addr addrbuf;
221	res_state statp;
222	u_long options;
223
224	switch (af) {
225	case AF_INET:
226#ifdef INET6
227	case AF_INET6:
228#endif
229		break;
230	default:
231		*errp = NO_RECOVERY;
232		return NULL;
233	}
234
235	if (flags & AI_ADDRCONFIG) {
236		int s;
237
238		if ((s = _socket(af, SOCK_DGRAM, 0)) < 0)
239			return NULL;
240		/*
241		 * TODO:
242		 * Note that implementation dependent test for address
243		 * configuration should be done everytime called
244		 * (or apropriate interval),
245		 * because addresses will be dynamically assigned or deleted.
246		 */
247		_close(s);
248	}
249
250#ifdef INET6
251	/* special case for literal address */
252	if (inet_pton(AF_INET6, name, &addrbuf) == 1) {
253		if (af != AF_INET6) {
254			*errp = HOST_NOT_FOUND;
255			return NULL;
256		}
257		return _hpaddr(af, name, &addrbuf, errp);
258	}
259#endif
260	if (inet_aton(name, (struct in_addr *)&addrbuf) == 1) {
261		if (af != AF_INET) {
262			if (MAPADDRENABLED(flags)) {
263				MAPADDR(&addrbuf, &addrbuf.in_addr);
264			} else {
265				*errp = HOST_NOT_FOUND;
266				return NULL;
267			}
268		}
269		return _hpaddr(af, name, &addrbuf, errp);
270	}
271
272
273	statp = __res_state();
274	if ((statp->options & RES_INIT) == 0) {
275		if (res_ninit(statp) < 0) {
276			*errp = NETDB_INTERNAL;
277			return NULL;
278		}
279	}
280
281	options = statp->options;
282	statp->options &= ~RES_USE_INET6;
283
284	hp = gethostbyname2(name, af);
285	hp = _hpcopy(hp, errp);
286#ifdef INET6
287	if (af == AF_INET6)
288		hp = _hpreorder(hp);
289
290	if (af == AF_INET6 && ((flags & AI_ALL) || hp == NULL) &&
291	    MAPADDRENABLED(flags)) {
292		struct hostent *hp2 = gethostbyname2(name, AF_INET);
293		if (hp == NULL)
294			if (hp2 == NULL)
295				*errp = statp->res_h_errno;
296			else
297				hp = _hpmapv6(hp2, errp);
298		else {
299			if (hp2 && strcmp(hp->h_name, hp2->h_name) == 0) {
300				struct hostent *hpb = hp;
301				hp = _hpmerge(hpb, hp2, errp);
302				freehostent(hpb);
303			}
304		}
305	}
306#endif
307
308	if (hp == NULL)
309		*errp = statp->res_h_errno;
310
311	statp->options = options;
312	return _hpsort(hp, statp);
313}
314
315struct hostent *
316getipnodebyaddr(const void *src, size_t len, int af, int *errp)
317{
318	struct hostent *hp;
319	res_state statp;
320	u_long options;
321
322#ifdef INET6
323	struct in6_addr addrbuf;
324#else
325	struct in_addr addrbuf;
326#endif
327
328	switch (af) {
329	case AF_INET:
330		if (len != sizeof(struct in_addr)) {
331			*errp = NO_RECOVERY;
332			return NULL;
333		}
334		if ((long)src & ~(sizeof(struct in_addr) - 1)) {
335			memcpy(&addrbuf, src, len);
336			src = &addrbuf;
337		}
338		if (((struct in_addr *)src)->s_addr == 0)
339			return NULL;
340		break;
341#ifdef INET6
342	case AF_INET6:
343		if (len != sizeof(struct in6_addr)) {
344			*errp = NO_RECOVERY;
345			return NULL;
346		}
347		if ((long)src & ~(sizeof(struct in6_addr) / 2 - 1)) {	/*XXX*/
348			memcpy(&addrbuf, src, len);
349			src = &addrbuf;
350		}
351		if (IN6_IS_ADDR_UNSPECIFIED((struct in6_addr *)src))
352			return NULL;
353		if (IN6_IS_ADDR_V4MAPPED((struct in6_addr *)src)
354		||  IN6_IS_ADDR_V4COMPAT((struct in6_addr *)src)) {
355			src = (char *)src +
356			    (sizeof(struct in6_addr) - sizeof(struct in_addr));
357			af = AF_INET;
358			len = sizeof(struct in_addr);
359		}
360		break;
361#endif
362	default:
363		*errp = NO_RECOVERY;
364		return NULL;
365	}
366
367	statp = __res_state();
368	if ((statp->options & RES_INIT) == 0) {
369		if (res_ninit(statp) < 0) {
370			RES_SET_H_ERRNO(statp, NETDB_INTERNAL);
371			return NULL;
372		}
373	}
374
375	options = statp->options;
376	statp->options &= ~RES_USE_INET6;
377
378	hp = gethostbyaddr(src, len, af);
379	if (hp == NULL)
380		*errp = statp->res_h_errno;
381
382	statp->options = options;
383	return (_hpcopy(hp, errp));
384}
385
386void
387freehostent(struct hostent *ptr)
388{
389	free(ptr);
390}
391
392/*
393 * Private utility functions
394 */
395
396/*
397 * _hpcopy: allocate and copy hostent structure
398 */
399static struct hostent *
400_hpcopy(struct hostent *hp, int *errp)
401{
402	struct hostent *nhp;
403	char *cp, **pp;
404	int size, addrsize;
405	int nalias = 0, naddr = 0;
406	int al_off;
407	int i;
408
409	if (hp == NULL)
410		return hp;
411
412	/* count size to be allocated */
413	size = sizeof(struct hostent);
414	if (hp->h_name != NULL)
415		size += strlen(hp->h_name) + 1;
416	if ((pp = hp->h_aliases) != NULL) {
417		for (i = 0; *pp != NULL; i++, pp++) {
418			if (**pp != '\0') {
419				size += strlen(*pp) + 1;
420				nalias++;
421			}
422		}
423	}
424	/* adjust alignment */
425	size = ALIGN(size);
426	al_off = size;
427	size += sizeof(char *) * (nalias + 1);
428	addrsize = ALIGN(hp->h_length);
429	if ((pp = hp->h_addr_list) != NULL) {
430		while (*pp++ != NULL)
431			naddr++;
432	}
433	size += addrsize * naddr;
434	size += sizeof(char *) * (naddr + 1);
435
436	/* copy */
437	if ((nhp = (struct hostent *)malloc(size)) == NULL) {
438		*errp = TRY_AGAIN;
439		return NULL;
440	}
441	cp = (char *)&nhp[1];
442	if (hp->h_name != NULL) {
443		nhp->h_name = cp;
444		strcpy(cp, hp->h_name);
445		cp += strlen(cp) + 1;
446	} else
447		nhp->h_name = NULL;
448	nhp->h_aliases = (char **)((char *)nhp + al_off);
449	if ((pp = hp->h_aliases) != NULL) {
450		for (i = 0; *pp != NULL; pp++) {
451			if (**pp != '\0') {
452				nhp->h_aliases[i++] = cp;
453				strcpy(cp, *pp);
454				cp += strlen(cp) + 1;
455			}
456		}
457	}
458	nhp->h_aliases[nalias] = NULL;
459	cp = (char *)&nhp->h_aliases[nalias + 1];
460	nhp->h_addrtype = hp->h_addrtype;
461	nhp->h_length = hp->h_length;
462	nhp->h_addr_list = (char **)cp;
463	if ((pp = hp->h_addr_list) != NULL) {
464		cp = (char *)&nhp->h_addr_list[naddr + 1];
465		for (i = 0; *pp != NULL; pp++) {
466			nhp->h_addr_list[i++] = cp;
467			memcpy(cp, *pp, hp->h_length);
468			cp += addrsize;
469		}
470	}
471	nhp->h_addr_list[naddr] = NULL;
472	return nhp;
473}
474
475/*
476 * _hpaddr: construct hostent structure with one address
477 */
478static struct hostent *
479_hpaddr(int af, const char *name, void *addr, int *errp)
480{
481	struct hostent *hp, hpbuf;
482	char *addrs[2];
483
484	hp = &hpbuf;
485	hp->h_name = (char *)name;
486	hp->h_aliases = NULL;
487	hp->h_addrtype = af;
488	hp->h_length = ADDRLEN(af);
489	hp->h_addr_list = addrs;
490	addrs[0] = (char *)addr;
491	addrs[1] = NULL;
492	return (_hpcopy(hp, errp));
493}
494
495#ifdef INET6
496/*
497 * _hpmerge: merge 2 hostent structure, arguments will be freed
498 */
499static struct hostent *
500_hpmerge(struct hostent *hp1, struct hostent *hp2, int *errp)
501{
502	int i, j;
503	int naddr, nalias;
504	char **pp;
505	struct hostent *hp, hpbuf;
506	char *aliases[MAXALIASES + 1], *addrs[MAXADDRS + 1];
507	union inx_addr addrbuf[MAXADDRS];
508
509	if (hp1 == NULL)
510		return _hpcopy(hp2, errp);
511	if (hp2 == NULL)
512		return _hpcopy(hp1, errp);
513
514#define	HP(i)	(i == 1 ? hp1 : hp2)
515	hp = &hpbuf;
516	hp->h_name = (hp1->h_name != NULL ? hp1->h_name : hp2->h_name);
517	hp->h_aliases = aliases;
518	nalias = 0;
519	for (i = 1; i <= 2; i++) {
520		if ((pp = HP(i)->h_aliases) == NULL)
521			continue;
522		for (; nalias < MAXALIASES && *pp != NULL; pp++) {
523			/* check duplicates */
524			for (j = 0; j < nalias; j++)
525				if (strcasecmp(*pp, aliases[j]) == 0)
526					break;
527			if (j == nalias)
528				aliases[nalias++] = *pp;
529		}
530	}
531	aliases[nalias] = NULL;
532	if (hp1->h_length != hp2->h_length) {
533		hp->h_addrtype = AF_INET6;
534		hp->h_length = sizeof(struct in6_addr);
535	} else {
536		hp->h_addrtype = hp1->h_addrtype;
537		hp->h_length = hp1->h_length;
538	}
539
540	hp->h_addr_list = addrs;
541	naddr = 0;
542	for (i = 1; i <= 2; i++) {
543		if ((pp = HP(i)->h_addr_list) == NULL)
544			continue;
545		if (HP(i)->h_length == hp->h_length) {
546			while (naddr < MAXADDRS && *pp != NULL)
547				addrs[naddr++] = *pp++;
548		} else {
549			/* copy IPv4 addr as mapped IPv6 addr */
550			while (naddr < MAXADDRS && *pp != NULL) {
551				MAPADDR(&addrbuf[naddr], *pp++);
552				addrs[naddr] = (char *)&addrbuf[naddr];
553				naddr++;
554			}
555		}
556	}
557	addrs[naddr] = NULL;
558	return (_hpcopy(hp, errp));
559}
560#endif
561
562/*
563 * _hpmapv6: convert IPv4 hostent into IPv4-mapped IPv6 addresses
564 */
565#ifdef INET6
566static struct hostent *
567_hpmapv6(struct hostent *hp, int *errp)
568{
569	struct hostent hp6;
570
571	if (hp == NULL)
572		return NULL;
573	if (hp->h_addrtype == AF_INET6)
574		return _hpcopy(hp, errp);
575
576	memset(&hp6, 0, sizeof(struct hostent));
577	hp6.h_addrtype = AF_INET6;
578	hp6.h_length = sizeof(struct in6_addr);
579	return _hpmerge(&hp6, hp, errp);
580}
581#endif
582
583/*
584 * _hpsort: sort address by sortlist
585 */
586static struct hostent *
587_hpsort(struct hostent *hp, res_state statp)
588{
589	int i, j, n;
590	u_char *ap, *sp, *mp, **pp;
591	char t;
592	char order[MAXADDRS];
593	int nsort = statp->nsort;
594
595	if (hp == NULL || hp->h_addr_list[1] == NULL || nsort == 0)
596		return hp;
597	for (i = 0; (ap = (u_char *)hp->h_addr_list[i]); i++) {
598		for (j = 0; j < nsort; j++) {
599#ifdef INET6
600			if (statp->_u._ext.ext->sort_list[j].af !=
601			    hp->h_addrtype)
602				continue;
603			sp = (u_char *)&statp->_u._ext.ext->sort_list[j].addr;
604			mp = (u_char *)&statp->_u._ext.ext->sort_list[j].mask;
605#else
606			sp = (u_char *)&statp->sort_list[j].addr;
607			mp = (u_char *)&statp->sort_list[j].mask;
608#endif
609			for (n = 0; n < hp->h_length; n++) {
610				if ((ap[n] & mp[n]) != sp[n])
611					break;
612			}
613			if (n == hp->h_length)
614				break;
615		}
616		order[i] = j;
617	}
618	n = i;
619	pp = (u_char **)hp->h_addr_list;
620	for (i = 0; i < n - 1; i++) {
621		for (j = i + 1; j < n; j++) {
622			if (order[i] > order[j]) {
623				ap = pp[i];
624				pp[i] = pp[j];
625				pp[j] = ap;
626				t = order[i];
627				order[i] = order[j];
628				order[j] = t;
629			}
630		}
631	}
632	return hp;
633}
634
635#ifdef INET6
636/*
637 * _hpreorder: sort address by default address selection
638 */
639static struct hostent *
640_hpreorder(struct hostent *hp)
641{
642	struct hp_order *aio;
643	int i, n;
644	char *ap;
645	struct sockaddr *sa;
646	struct policyhead policyhead;
647
648	if (hp == NULL)
649		return hp;
650
651	switch (hp->h_addrtype) {
652	case AF_INET:
653#ifdef INET6
654	case AF_INET6:
655#endif
656		break;
657	default:
658		free_addrselectpolicy(&policyhead);
659		return hp;
660	}
661
662	/* count the number of addrinfo elements for sorting. */
663	for (n = 0; hp->h_addr_list[n] != NULL; n++)
664		;
665
666	/*
667	 * If the number is small enough, we can skip the reordering process.
668	 */
669	if (n <= 1)
670		return hp;
671
672	/* allocate a temporary array for sort and initialization of it. */
673	if ((aio = malloc(sizeof(*aio) * n)) == NULL)
674		return hp;	/* give up reordering */
675	memset(aio, 0, sizeof(*aio) * n);
676
677	/* retrieve address selection policy from the kernel */
678	TAILQ_INIT(&policyhead);
679	if (!get_addrselectpolicy(&policyhead)) {
680		/* no policy is installed into kernel, we don't sort. */
681		free(aio);
682		return hp;
683	}
684
685	for (i = 0; i < n; i++) {
686		ap = hp->h_addr_list[i];
687		aio[i].aio_h_addr = ap;
688		sa = &aio[i].aio_sa;
689		switch (hp->h_addrtype) {
690		case AF_INET:
691			sa->sa_family = AF_INET;
692			sa->sa_len = sizeof(struct sockaddr_in);
693			memcpy(&((struct sockaddr_in *)sa)->sin_addr, ap,
694			    sizeof(struct in_addr));
695			break;
696#ifdef INET6
697		case AF_INET6:
698			if (IN6_IS_ADDR_V4MAPPED((struct in6_addr *)ap)) {
699				sa->sa_family = AF_INET;
700				sa->sa_len = sizeof(struct sockaddr_in);
701				memcpy(&((struct sockaddr_in *)sa)->sin_addr,
702				    &ap[12], sizeof(struct in_addr));
703			} else {
704				sa->sa_family = AF_INET6;
705				sa->sa_len = sizeof(struct sockaddr_in6);
706				memcpy(&((struct sockaddr_in6 *)sa)->sin6_addr,
707				    ap, sizeof(struct in6_addr));
708			}
709			break;
710#endif
711		}
712		aio[i].aio_dstscope = gai_addr2scopetype(sa);
713		aio[i].aio_dstpolicy = match_addrselectpolicy(sa, &policyhead);
714		set_source(&aio[i], &policyhead);
715	}
716
717	/* perform sorting. */
718	qsort(aio, n, sizeof(*aio), comp_dst);
719
720	/* reorder the h_addr_list. */
721	for (i = 0; i < n; i++)
722		hp->h_addr_list[i] = aio[i].aio_h_addr;
723
724	/* cleanup and return */
725	free(aio);
726	free_addrselectpolicy(&policyhead);
727	return hp;
728}
729
730static int
731get_addrselectpolicy(struct policyhead *head)
732{
733#ifdef INET6
734	int mib[] = { CTL_NET, PF_INET6, IPPROTO_IPV6, IPV6CTL_ADDRCTLPOLICY };
735	size_t l;
736	char *buf;
737	struct in6_addrpolicy *pol, *ep;
738
739	if (sysctl(mib, sizeof(mib) / sizeof(mib[0]), NULL, &l, NULL, 0) < 0)
740		return (0);
741	if ((buf = malloc(l)) == NULL)
742		return (0);
743	if (sysctl(mib, sizeof(mib) / sizeof(mib[0]), buf, &l, NULL, 0) < 0) {
744		free(buf);
745		return (0);
746	}
747
748	ep = (struct in6_addrpolicy *)(buf + l);
749	for (pol = (struct in6_addrpolicy *)buf; pol + 1 <= ep; pol++) {
750		struct policyqueue *new;
751
752		if ((new = malloc(sizeof(*new))) == NULL) {
753			free_addrselectpolicy(head); /* make the list empty */
754			break;
755		}
756		new->pc_policy = *pol;
757		TAILQ_INSERT_TAIL(head, new, pc_entry);
758	}
759
760	free(buf);
761	return (1);
762#else
763	return (0);
764#endif
765}
766
767static void
768free_addrselectpolicy(struct policyhead *head)
769{
770	struct policyqueue *ent, *nent;
771
772	for (ent = TAILQ_FIRST(head); ent; ent = nent) {
773		nent = TAILQ_NEXT(ent, pc_entry);
774		TAILQ_REMOVE(head, ent, pc_entry);
775		free(ent);
776	}
777}
778
779static struct policyqueue *
780match_addrselectpolicy(struct sockaddr *addr, struct policyhead *head)
781{
782#ifdef INET6
783	struct policyqueue *ent, *bestent = NULL;
784	struct in6_addrpolicy *pol;
785	int matchlen, bestmatchlen = -1;
786	u_char *mp, *ep, *k, *p, m;
787	struct sockaddr_in6 key;
788
789	switch(addr->sa_family) {
790	case AF_INET6:
791		key = *(struct sockaddr_in6 *)addr;
792		break;
793	case AF_INET:
794		/* convert the address into IPv4-mapped IPv6 address. */
795		memset(&key, 0, sizeof(key));
796		key.sin6_family = AF_INET6;
797		key.sin6_len = sizeof(key);
798		key.sin6_addr.s6_addr[10] = 0xff;
799		key.sin6_addr.s6_addr[11] = 0xff;
800		memcpy(&key.sin6_addr.s6_addr[12],
801		       &((struct sockaddr_in *)addr)->sin_addr, 4);
802		break;
803	default:
804		return(NULL);
805	}
806
807	for (ent = TAILQ_FIRST(head); ent; ent = TAILQ_NEXT(ent, pc_entry)) {
808		pol = &ent->pc_policy;
809		matchlen = 0;
810
811		mp = (u_char *)&pol->addrmask.sin6_addr;
812		ep = mp + 16;	/* XXX: scope field? */
813		k = (u_char *)&key.sin6_addr;
814		p = (u_char *)&pol->addr.sin6_addr;
815		for (; mp < ep && *mp; mp++, k++, p++) {
816			m = *mp;
817			if ((*k & m) != *p)
818				goto next; /* not match */
819			if (m == 0xff) /* short cut for a typical case */
820				matchlen += 8;
821			else {
822				while (m >= 0x80) {
823					matchlen++;
824					m <<= 1;
825				}
826			}
827		}
828
829		/* matched.  check if this is better than the current best. */
830		if (matchlen > bestmatchlen) {
831			bestent = ent;
832			bestmatchlen = matchlen;
833		}
834
835	  next:
836		continue;
837	}
838
839	return(bestent);
840#else
841	return(NULL);
842#endif
843
844}
845
846static void
847set_source(struct hp_order *aio, struct policyhead *ph)
848{
849	struct sockaddr_storage ss = aio->aio_un.aiou_ss;
850	socklen_t srclen;
851	int s;
852
853	/* set unspec ("no source is available"), just in case */
854	aio->aio_srcsa.sa_family = AF_UNSPEC;
855	aio->aio_srcscope = -1;
856
857	switch(ss.ss_family) {
858	case AF_INET:
859		((struct sockaddr_in *)&ss)->sin_port = htons(1);
860		break;
861#ifdef INET6
862	case AF_INET6:
863		((struct sockaddr_in6 *)&ss)->sin6_port = htons(1);
864		break;
865#endif
866	default:		/* ignore unsupported AFs explicitly */
867		return;
868	}
869
870	/* open a socket to get the source address for the given dst */
871	if ((s = _socket(ss.ss_family, SOCK_DGRAM, IPPROTO_UDP)) < 0)
872		return;		/* give up */
873	if (_connect(s, (struct sockaddr *)&ss, ss.ss_len) < 0)
874		goto cleanup;
875	srclen = ss.ss_len;
876	if (_getsockname(s, &aio->aio_srcsa, &srclen) < 0) {
877		aio->aio_srcsa.sa_family = AF_UNSPEC;
878		goto cleanup;
879	}
880	aio->aio_srcscope = gai_addr2scopetype(&aio->aio_srcsa);
881	aio->aio_srcpolicy = match_addrselectpolicy(&aio->aio_srcsa, ph);
882	aio->aio_matchlen = matchlen(&aio->aio_srcsa, (struct sockaddr *)&ss);
883#ifdef INET6
884	if (ss.ss_family == AF_INET6) {
885		struct in6_ifreq ifr6;
886		u_int32_t flags6;
887
888		memset(&ifr6, 0, sizeof(ifr6));
889		memcpy(&ifr6.ifr_addr, &ss, ss.ss_len);
890		if (_ioctl(s, SIOCGIFAFLAG_IN6, &ifr6) == 0) {
891			flags6 = ifr6.ifr_ifru.ifru_flags6;
892			if ((flags6 & IN6_IFF_DEPRECATED))
893				aio->aio_srcflag |= AIO_SRCFLAG_DEPRECATED;
894		}
895	}
896#endif
897
898  cleanup:
899	_close(s);
900	return;
901}
902
903static int
904matchlen(struct sockaddr *src, struct sockaddr *dst)
905{
906	int match = 0;
907	u_char *s, *d;
908	u_char *lim, r;
909	int addrlen;
910
911	switch (src->sa_family) {
912#ifdef INET6
913	case AF_INET6:
914		s = (u_char *)&((struct sockaddr_in6 *)src)->sin6_addr;
915		d = (u_char *)&((struct sockaddr_in6 *)dst)->sin6_addr;
916		addrlen = sizeof(struct in6_addr);
917		lim = s + addrlen;
918		break;
919#endif
920	case AF_INET:
921		s = (u_char *)&((struct sockaddr_in *)src)->sin_addr;
922		d = (u_char *)&((struct sockaddr_in *)dst)->sin_addr;
923		addrlen = sizeof(struct in_addr);
924		lim = s + addrlen;
925		break;
926	default:
927		return(0);
928	}
929
930	while (s < lim)
931		if ((r = (*d++ ^ *s++)) != 0) {
932			while (r < addrlen * 8) {
933				match++;
934				r <<= 1;
935			}
936			break;
937		} else
938			match += 8;
939	return(match);
940}
941
942static int
943comp_dst(const void *arg1, const void *arg2)
944{
945	const struct hp_order *dst1 = arg1, *dst2 = arg2;
946
947	/*
948	 * Rule 1: Avoid unusable destinations.
949	 * XXX: we currently do not consider if an appropriate route exists.
950	 */
951	if (dst1->aio_srcsa.sa_family != AF_UNSPEC &&
952	    dst2->aio_srcsa.sa_family == AF_UNSPEC) {
953		return(-1);
954	}
955	if (dst1->aio_srcsa.sa_family == AF_UNSPEC &&
956	    dst2->aio_srcsa.sa_family != AF_UNSPEC) {
957		return(1);
958	}
959
960	/* Rule 2: Prefer matching scope. */
961	if (dst1->aio_dstscope == dst1->aio_srcscope &&
962	    dst2->aio_dstscope != dst2->aio_srcscope) {
963		return(-1);
964	}
965	if (dst1->aio_dstscope != dst1->aio_srcscope &&
966	    dst2->aio_dstscope == dst2->aio_srcscope) {
967		return(1);
968	}
969
970	/* Rule 3: Avoid deprecated addresses. */
971	if (dst1->aio_srcsa.sa_family != AF_UNSPEC &&
972	    dst2->aio_srcsa.sa_family != AF_UNSPEC) {
973		if (!(dst1->aio_srcflag & AIO_SRCFLAG_DEPRECATED) &&
974		    (dst2->aio_srcflag & AIO_SRCFLAG_DEPRECATED)) {
975			return(-1);
976		}
977		if ((dst1->aio_srcflag & AIO_SRCFLAG_DEPRECATED) &&
978		    !(dst2->aio_srcflag & AIO_SRCFLAG_DEPRECATED)) {
979			return(1);
980		}
981	}
982
983	/* Rule 4: Prefer home addresses. */
984	/* XXX: not implemented yet */
985
986	/* Rule 5: Prefer matching label. */
987#ifdef INET6
988	if (dst1->aio_srcpolicy && dst1->aio_dstpolicy &&
989	    dst1->aio_srcpolicy->pc_policy.label ==
990	    dst1->aio_dstpolicy->pc_policy.label &&
991	    (dst2->aio_srcpolicy == NULL || dst2->aio_dstpolicy == NULL ||
992	     dst2->aio_srcpolicy->pc_policy.label !=
993	     dst2->aio_dstpolicy->pc_policy.label)) {
994		return(-1);
995	}
996	if (dst2->aio_srcpolicy && dst2->aio_dstpolicy &&
997	    dst2->aio_srcpolicy->pc_policy.label ==
998	    dst2->aio_dstpolicy->pc_policy.label &&
999	    (dst1->aio_srcpolicy == NULL || dst1->aio_dstpolicy == NULL ||
1000	     dst1->aio_srcpolicy->pc_policy.label !=
1001	     dst1->aio_dstpolicy->pc_policy.label)) {
1002		return(1);
1003	}
1004#endif
1005
1006	/* Rule 6: Prefer higher precedence. */
1007#ifdef INET6
1008	if (dst1->aio_dstpolicy &&
1009	    (dst2->aio_dstpolicy == NULL ||
1010	     dst1->aio_dstpolicy->pc_policy.preced >
1011	     dst2->aio_dstpolicy->pc_policy.preced)) {
1012		return(-1);
1013	}
1014	if (dst2->aio_dstpolicy &&
1015	    (dst1->aio_dstpolicy == NULL ||
1016	     dst2->aio_dstpolicy->pc_policy.preced >
1017	     dst1->aio_dstpolicy->pc_policy.preced)) {
1018		return(1);
1019	}
1020#endif
1021
1022	/* Rule 7: Prefer native transport. */
1023	/* XXX: not implemented yet */
1024
1025	/* Rule 8: Prefer smaller scope. */
1026	if (dst1->aio_dstscope >= 0 &&
1027	    dst1->aio_dstscope < dst2->aio_dstscope) {
1028		return(-1);
1029	}
1030	if (dst2->aio_dstscope >= 0 &&
1031	    dst2->aio_dstscope < dst1->aio_dstscope) {
1032		return(1);
1033	}
1034
1035	/*
1036	 * Rule 9: Use longest matching prefix.
1037	 * We compare the match length in a same AF only.
1038	 */
1039	if (dst1->aio_sa.sa_family == dst2->aio_sa.sa_family) {
1040		if (dst1->aio_matchlen > dst2->aio_matchlen) {
1041			return(-1);
1042		}
1043		if (dst1->aio_matchlen < dst2->aio_matchlen) {
1044			return(1);
1045		}
1046	}
1047
1048	/* Rule 10: Otherwise, leave the order unchanged. */
1049	return(-1);
1050}
1051
1052/*
1053 * Copy from scope.c.
1054 * XXX: we should standardize the functions and link them as standard
1055 * library.
1056 */
1057static int
1058gai_addr2scopetype(struct sockaddr *sa)
1059{
1060#ifdef INET6
1061	struct sockaddr_in6 *sa6;
1062#endif
1063	struct sockaddr_in *sa4;
1064
1065	switch(sa->sa_family) {
1066#ifdef INET6
1067	case AF_INET6:
1068		sa6 = (struct sockaddr_in6 *)sa;
1069		if (IN6_IS_ADDR_MULTICAST(&sa6->sin6_addr)) {
1070			/* just use the scope field of the multicast address */
1071			return(sa6->sin6_addr.s6_addr[2] & 0x0f);
1072		}
1073		/*
1074		 * Unicast addresses: map scope type to corresponding scope
1075		 * value defined for multcast addresses.
1076		 * XXX: hardcoded scope type values are bad...
1077		 */
1078		if (IN6_IS_ADDR_LOOPBACK(&sa6->sin6_addr))
1079			return(1); /* node local scope */
1080		if (IN6_IS_ADDR_LINKLOCAL(&sa6->sin6_addr))
1081			return(2); /* link-local scope */
1082		if (IN6_IS_ADDR_SITELOCAL(&sa6->sin6_addr))
1083			return(5); /* site-local scope */
1084		return(14);	/* global scope */
1085		break;
1086#endif
1087	case AF_INET:
1088		/*
1089		 * IPv4 pseudo scoping according to RFC 3484.
1090		 */
1091		sa4 = (struct sockaddr_in *)sa;
1092		/* IPv4 autoconfiguration addresses have link-local scope. */
1093		if (((u_char *)&sa4->sin_addr)[0] == 169 &&
1094		    ((u_char *)&sa4->sin_addr)[1] == 254)
1095			return(2);
1096		/* Private addresses have site-local scope. */
1097		if (((u_char *)&sa4->sin_addr)[0] == 10 ||
1098		    (((u_char *)&sa4->sin_addr)[0] == 172 &&
1099		     (((u_char *)&sa4->sin_addr)[1] & 0xf0) == 16) ||
1100		    (((u_char *)&sa4->sin_addr)[0] == 192 &&
1101		     ((u_char *)&sa4->sin_addr)[1] == 168))
1102			return(14);	/* XXX: It should be 5 unless NAT */
1103		/* Loopback addresses have link-local scope. */
1104		if (((u_char *)&sa4->sin_addr)[0] == 127)
1105			return(2);
1106		return(14);
1107		break;
1108	default:
1109		errno = EAFNOSUPPORT; /* is this a good error? */
1110		return(-1);
1111	}
1112}
1113#endif
1114