xref: /freebsd-10-stable/lib/libc/net/name6.c

/*	$KAME: name6.c,v 1.25 2000/06/26 16:44:40 itojun Exp $	*/

/*
 * Copyright (C) 1995, 1996, 1997, 1998, and 1999 WIDE Project.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the project nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */
/*
 * ++Copyright++ 1985, 1988, 1993
 * -
 * Copyright (c) 1985, 1988, 1993
 *    The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 * -
 * Portions Copyright (c) 1993 by Digital Equipment Corporation.
 *
 * Permission to use, copy, modify, and distribute this software for any
 * purpose with or without fee is hereby granted, provided that the above
 * copyright notice and this permission notice appear in all copies, and that
 * the name of Digital Equipment Corporation not be used in advertising or
 * publicity pertaining to distribution of the document or software without
 * specific, written prior permission.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND DIGITAL EQUIPMENT CORP. DISCLAIMS ALL
 * WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES
 * OF MERCHANTABILITY AND FITNESS.   IN NO EVENT SHALL DIGITAL EQUIPMENT
 * CORPORATION BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
 * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
 * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
 * ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
 * SOFTWARE.
 * -
 * --Copyright--
 */

/*
 *	Atsushi Onoe <onoe@sm.sony.co.jp>
 */

#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");

#include "namespace.h"
#include <sys/param.h>
#include <sys/socket.h>
#include <sys/time.h>
#include <sys/queue.h>
#include <netinet/in.h>
#ifdef INET6
#include <net/if.h>
#include <net/if_var.h>
#include <sys/sysctl.h>
#include <sys/ioctl.h>
#include <netinet6/in6_var.h>	/* XXX */
#endif

#include <arpa/inet.h>
#include <arpa/nameser.h>

#include <errno.h>
#include <netdb.h>
#include <resolv.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdarg.h>
#include <nsswitch.h>
#include <unistd.h>
#include "un-namespace.h"
#include "netdb_private.h"
#include "res_private.h"

#ifndef MAXALIASES
#define	MAXALIASES	10
#endif
#ifndef	MAXADDRS
#define	MAXADDRS	20
#endif
#ifndef MAXDNAME
#define	MAXDNAME	1025
#endif

#ifdef INET6
#define	ADDRLEN(af)	((af) == AF_INET6 ? sizeof(struct in6_addr) : \
					    sizeof(struct in_addr))
#else
#define	ADDRLEN(af)	sizeof(struct in_addr)
#endif

#define	MAPADDR(ab, ina) \
do {									\
	memcpy(&(ab)->map_inaddr, ina, sizeof(struct in_addr));		\
	memset((ab)->map_zero, 0, sizeof((ab)->map_zero));		\
	memset((ab)->map_one, 0xff, sizeof((ab)->map_one));		\
} while (0)
#define	MAPADDRENABLED(flags) \
	(((flags) & AI_V4MAPPED) || \
	 (((flags) & AI_V4MAPPED_CFG)))

union inx_addr {
	struct in_addr	in_addr;
#ifdef INET6
	struct in6_addr	in6_addr;
#endif
	struct {
		u_char	mau_zero[10];
		u_char	mau_one[2];
		struct in_addr mau_inaddr;
	}		map_addr_un;
#define	map_zero	map_addr_un.mau_zero
#define	map_one		map_addr_un.mau_one
#define	map_inaddr	map_addr_un.mau_inaddr
};

struct policyqueue {
	TAILQ_ENTRY(policyqueue) pc_entry;
#ifdef INET6
	struct in6_addrpolicy pc_policy;
#endif
};
TAILQ_HEAD(policyhead, policyqueue);

#define AIO_SRCFLAG_DEPRECATED	0x1

struct hp_order {
	union {
		struct sockaddr_storage aiou_ss;
		struct sockaddr aiou_sa;
	} aio_src_un;
#define aio_srcsa aio_src_un.aiou_sa
	u_int32_t aio_srcflag;
	int aio_srcscope;
	int aio_dstscope;
	struct policyqueue *aio_srcpolicy;
	struct policyqueue *aio_dstpolicy;
	union {
		struct sockaddr_storage aiou_ss;
		struct sockaddr aiou_sa;
	} aio_un;
#define aio_sa aio_un.aiou_sa
	int aio_matchlen;
	char *aio_h_addr;
	int aio_initial_sequence;
};

static struct	 hostent *_hpcopy(struct hostent *, int *);
static struct	 hostent *_hpaddr(int, const char *, void *, int *);
#ifdef INET6
static struct	 hostent *_hpmerge(struct hostent *, struct hostent *, int *);
static struct	 hostent *_hpmapv6(struct hostent *, int *);
#endif
static struct	 hostent *_hpsort(struct hostent *, res_state);

#ifdef INET6
static struct	 hostent *_hpreorder(struct hostent *);
static int	 get_addrselectpolicy(struct policyhead *);
static void	 free_addrselectpolicy(struct policyhead *);
static struct	 policyqueue *match_addrselectpolicy(struct sockaddr *,
	struct policyhead *);
static void	 set_source(struct hp_order *, struct policyhead *);
static int	 matchlen(struct sockaddr *, struct sockaddr *);
static int	 comp_dst(const void *, const void *);
static int	 gai_addr2scopetype(struct sockaddr *);
#endif

/*
 * Functions defined in RFC2553
 *	getipnodebyname, getipnodebyaddr, freehostent
 */

struct hostent *
getipnodebyname(const char *name, int af, int flags, int *errp)
{
	struct hostent *hp;
	union inx_addr addrbuf;
	res_state statp;
	u_long options;

	switch (af) {
	case AF_INET:
#ifdef INET6
	case AF_INET6:
#endif
		break;
	default:
		*errp = NO_RECOVERY;
		return NULL;
	}

	if (flags & AI_ADDRCONFIG) {
		int s;

		if ((s = _socket(af, SOCK_DGRAM | SOCK_CLOEXEC, 0)) < 0)
			return NULL;
		/*
		 * TODO:
		 * Note that implementation dependent test for address
		 * configuration should be done everytime called
		 * (or apropriate interval),
		 * because addresses will be dynamically assigned or deleted.
		 */
		_close(s);
	}

#ifdef INET6
	/* special case for literal address */
	if (inet_pton(AF_INET6, name, &addrbuf) == 1) {
		if (af != AF_INET6) {
			*errp = HOST_NOT_FOUND;
			return NULL;
		}
		return _hpaddr(af, name, &addrbuf, errp);
	}
#endif
	if (inet_aton(name, (struct in_addr *)&addrbuf) == 1) {
		if (af != AF_INET) {
			if (MAPADDRENABLED(flags)) {
				MAPADDR(&addrbuf, &addrbuf.in_addr);
			} else {
				*errp = HOST_NOT_FOUND;
				return NULL;
			}
		}
		return _hpaddr(af, name, &addrbuf, errp);
	}


	statp = __res_state();
	if ((statp->options & RES_INIT) == 0) {
		if (res_ninit(statp) < 0) {
			*errp = NETDB_INTERNAL;
			return NULL;
		}
	}

	options = statp->options;
	statp->options &= ~RES_USE_INET6;

	hp = gethostbyname2(name, af);
	hp = _hpcopy(hp, errp);
#ifdef INET6
	if (af == AF_INET6)
		hp = _hpreorder(hp);

	if (af == AF_INET6 && ((flags & AI_ALL) || hp == NULL) &&
	    MAPADDRENABLED(flags)) {
		struct hostent *hp2 = gethostbyname2(name, AF_INET);
		if (hp == NULL)
			if (hp2 == NULL)
				*errp = statp->res_h_errno;
			else
				hp = _hpmapv6(hp2, errp);
		else {
			if (hp2 && strcmp(hp->h_name, hp2->h_name) == 0) {
				struct hostent *hpb = hp;
				hp = _hpmerge(hpb, hp2, errp);
				freehostent(hpb);
			}
		}
	}
#endif

	if (hp == NULL)
		*errp = statp->res_h_errno;

	statp->options = options;
	return _hpsort(hp, statp);
}

struct hostent *
getipnodebyaddr(const void *src, size_t len, int af, int *errp)
{
	struct hostent *hp;
	res_state statp;
	u_long options;

#ifdef INET6
	struct in6_addr addrbuf;
#else
	struct in_addr addrbuf;
#endif

	switch (af) {
	case AF_INET:
		if (len != sizeof(struct in_addr)) {
			*errp = NO_RECOVERY;
			return NULL;
		}
		if ((long)src & ~(sizeof(struct in_addr) - 1)) {
			memcpy(&addrbuf, src, len);
			src = &addrbuf;
		}
		if (((struct in_addr *)src)->s_addr == 0)
			return NULL;
		break;
#ifdef INET6
	case AF_INET6:
		if (len != sizeof(struct in6_addr)) {
			*errp = NO_RECOVERY;
			return NULL;
		}
		if ((long)src & ~(sizeof(struct in6_addr) / 2 - 1)) {	/*XXX*/
			memcpy(&addrbuf, src, len);
			src = &addrbuf;
		}
		if (IN6_IS_ADDR_UNSPECIFIED((struct in6_addr *)src))
			return NULL;
		if (IN6_IS_ADDR_V4MAPPED((struct in6_addr *)src)
		||  IN6_IS_ADDR_V4COMPAT((struct in6_addr *)src)) {
			src = (char *)src +
			    (sizeof(struct in6_addr) - sizeof(struct in_addr));
			af = AF_INET;
			len = sizeof(struct in_addr);
		}
		break;
#endif
	default:
		*errp = NO_RECOVERY;
		return NULL;
	}

	statp = __res_state();
	if ((statp->options & RES_INIT) == 0) {
		if (res_ninit(statp) < 0) {
			RES_SET_H_ERRNO(statp, NETDB_INTERNAL);
			return NULL;
		}
	}

	options = statp->options;
	statp->options &= ~RES_USE_INET6;

	hp = gethostbyaddr(src, len, af);
	if (hp == NULL)
		*errp = statp->res_h_errno;

	statp->options = options;
	return (_hpcopy(hp, errp));
}

void
freehostent(struct hostent *ptr)
{
	free(ptr);
}

/*
 * Private utility functions
 */

/*
 * _hpcopy: allocate and copy hostent structure
 */
static struct hostent *
_hpcopy(struct hostent *hp, int *errp)
{
	struct hostent *nhp;
	char *cp, **pp;
	int size, addrsize;
	int nalias = 0, naddr = 0;
	int al_off;
	int i;

	if (hp == NULL)
		return hp;

	/* count size to be allocated */
	size = sizeof(struct hostent);
	if (hp->h_name != NULL)
		size += strlen(hp->h_name) + 1;
	if ((pp = hp->h_aliases) != NULL) {
		for (i = 0; *pp != NULL; i++, pp++) {
			if (**pp != '\0') {
				size += strlen(*pp) + 1;
				nalias++;
			}
		}
	}
	/* adjust alignment */
	size = ALIGN(size);
	al_off = size;
	size += sizeof(char *) * (nalias + 1);
	addrsize = ALIGN(hp->h_length);
	if ((pp = hp->h_addr_list) != NULL) {
		while (*pp++ != NULL)
			naddr++;
	}
	size += addrsize * naddr;
	size += sizeof(char *) * (naddr + 1);

	/* copy */
	if ((nhp = (struct hostent *)malloc(size)) == NULL) {
		*errp = TRY_AGAIN;
		return NULL;
	}
	cp = (char *)&nhp[1];
	if (hp->h_name != NULL) {
		nhp->h_name = cp;
		strcpy(cp, hp->h_name);
		cp += strlen(cp) + 1;
	} else
		nhp->h_name = NULL;
	nhp->h_aliases = (char **)((char *)nhp + al_off);
	if ((pp = hp->h_aliases) != NULL) {
		for (i = 0; *pp != NULL; pp++) {
			if (**pp != '\0') {
				nhp->h_aliases[i++] = cp;
				strcpy(cp, *pp);
				cp += strlen(cp) + 1;
			}
		}
	}
	nhp->h_aliases[nalias] = NULL;
	cp = (char *)&nhp->h_aliases[nalias + 1];
	nhp->h_addrtype = hp->h_addrtype;
	nhp->h_length = hp->h_length;
	nhp->h_addr_list = (char **)cp;
	if ((pp = hp->h_addr_list) != NULL) {
		cp = (char *)&nhp->h_addr_list[naddr + 1];
		for (i = 0; *pp != NULL; pp++) {
			nhp->h_addr_list[i++] = cp;
			memcpy(cp, *pp, hp->h_length);
			cp += addrsize;
		}
	}
	nhp->h_addr_list[naddr] = NULL;
	return nhp;
}

/*
 * _hpaddr: construct hostent structure with one address
 */
static struct hostent *
_hpaddr(int af, const char *name, void *addr, int *errp)
{
	struct hostent *hp, hpbuf;
	char *addrs[2];

	hp = &hpbuf;
	hp->h_name = (char *)name;
	hp->h_aliases = NULL;
	hp->h_addrtype = af;
	hp->h_length = ADDRLEN(af);
	hp->h_addr_list = addrs;
	addrs[0] = (char *)addr;
	addrs[1] = NULL;
	return (_hpcopy(hp, errp));
}

#ifdef INET6
/*
 * _hpmerge: merge 2 hostent structure, arguments will be freed
 */
static struct hostent *
_hpmerge(struct hostent *hp1, struct hostent *hp2, int *errp)
{
	int i, j;
	int naddr, nalias;
	char **pp;
	struct hostent *hp, hpbuf;
	char *aliases[MAXALIASES + 1], *addrs[MAXADDRS + 1];
	union inx_addr addrbuf[MAXADDRS];

	if (hp1 == NULL)
		return _hpcopy(hp2, errp);
	if (hp2 == NULL)
		return _hpcopy(hp1, errp);

#define	HP(i)	(i == 1 ? hp1 : hp2)
	hp = &hpbuf;
	hp->h_name = (hp1->h_name != NULL ? hp1->h_name : hp2->h_name);
	hp->h_aliases = aliases;
	nalias = 0;
	for (i = 1; i <= 2; i++) {
		if ((pp = HP(i)->h_aliases) == NULL)
			continue;
		for (; nalias < MAXALIASES && *pp != NULL; pp++) {
			/* check duplicates */
			for (j = 0; j < nalias; j++)
				if (strcasecmp(*pp, aliases[j]) == 0)
					break;
			if (j == nalias)
				aliases[nalias++] = *pp;
		}
	}
	aliases[nalias] = NULL;
	if (hp1->h_length != hp2->h_length) {
		hp->h_addrtype = AF_INET6;
		hp->h_length = sizeof(struct in6_addr);
	} else {
		hp->h_addrtype = hp1->h_addrtype;
		hp->h_length = hp1->h_length;
	}

	hp->h_addr_list = addrs;
	naddr = 0;
	for (i = 1; i <= 2; i++) {
		if ((pp = HP(i)->h_addr_list) == NULL)
			continue;
		if (HP(i)->h_length == hp->h_length) {
			while (naddr < MAXADDRS && *pp != NULL)
				addrs[naddr++] = *pp++;
		} else {
			/* copy IPv4 addr as mapped IPv6 addr */
			while (naddr < MAXADDRS && *pp != NULL) {
				MAPADDR(&addrbuf[naddr], *pp++);
				addrs[naddr] = (char *)&addrbuf[naddr];
				naddr++;
			}
		}
	}
	addrs[naddr] = NULL;
	return (_hpcopy(hp, errp));
}
#endif

/*
 * _hpmapv6: convert IPv4 hostent into IPv4-mapped IPv6 addresses
 */
#ifdef INET6
static struct hostent *
_hpmapv6(struct hostent *hp, int *errp)
{
	struct hostent hp6;

	if (hp == NULL)
		return NULL;
	if (hp->h_addrtype == AF_INET6)
		return _hpcopy(hp, errp);

	memset(&hp6, 0, sizeof(struct hostent));
	hp6.h_addrtype = AF_INET6;
	hp6.h_length = sizeof(struct in6_addr);
	return _hpmerge(&hp6, hp, errp);
}
#endif

/*
 * _hpsort: sort address by sortlist
 */
static struct hostent *
_hpsort(struct hostent *hp, res_state statp)
{
	int i, j, n;
	u_char *ap, *sp, *mp, **pp;
	char t;
	char order[MAXADDRS];
	int nsort = statp->nsort;

	if (hp == NULL || hp->h_addr_list[1] == NULL || nsort == 0)
		return hp;
	for (i = 0; (ap = (u_char *)hp->h_addr_list[i]); i++) {
		for (j = 0; j < nsort; j++) {
#ifdef INET6
			if (statp->_u._ext.ext->sort_list[j].af !=
			    hp->h_addrtype)
				continue;
			sp = (u_char *)&statp->_u._ext.ext->sort_list[j].addr;
			mp = (u_char *)&statp->_u._ext.ext->sort_list[j].mask;
#else
			sp = (u_char *)&statp->sort_list[j].addr;
			mp = (u_char *)&statp->sort_list[j].mask;
#endif
			for (n = 0; n < hp->h_length; n++) {
				if ((ap[n] & mp[n]) != sp[n])
					break;
			}
			if (n == hp->h_length)
				break;
		}
		order[i] = j;
	}
	n = i;
	pp = (u_char **)hp->h_addr_list;
	for (i = 0; i < n - 1; i++) {
		for (j = i + 1; j < n; j++) {
			if (order[i] > order[j]) {
				ap = pp[i];
				pp[i] = pp[j];
				pp[j] = ap;
				t = order[i];
				order[i] = order[j];
				order[j] = t;
			}
		}
	}
	return hp;
}

#ifdef INET6
/*
 * _hpreorder: sort address by default address selection
 */
static struct hostent *
_hpreorder(struct hostent *hp)
{
	struct hp_order *aio;
	int i, n;
	char *ap;
	struct sockaddr *sa;
	struct policyhead policyhead;

	if (hp == NULL)
		return hp;

	switch (hp->h_addrtype) {
	case AF_INET:
#ifdef INET6
	case AF_INET6:
#endif
		break;
	default:
		return hp;
	}

	/* count the number of addrinfo elements for sorting. */
	for (n = 0; hp->h_addr_list[n] != NULL; n++)
		;

	/*
	 * If the number is small enough, we can skip the reordering process.
	 */
	if (n <= 1)
		return hp;

	/* allocate a temporary array for sort and initialization of it. */
	if ((aio = malloc(sizeof(*aio) * n)) == NULL)
		return hp;	/* give up reordering */
	memset(aio, 0, sizeof(*aio) * n);

	/* retrieve address selection policy from the kernel */
	TAILQ_INIT(&policyhead);
	if (!get_addrselectpolicy(&policyhead)) {
		/* no policy is installed into kernel, we don't sort. */
		free(aio);
		return hp;
	}

	for (i = 0; i < n; i++) {
		ap = hp->h_addr_list[i];
		aio[i].aio_h_addr = ap;
		sa = &aio[i].aio_sa;
		switch (hp->h_addrtype) {
		case AF_INET:
			sa->sa_family = AF_INET;
			sa->sa_len = sizeof(struct sockaddr_in);
			memcpy(&((struct sockaddr_in *)sa)->sin_addr, ap,
			    sizeof(struct in_addr));
			break;
#ifdef INET6
		case AF_INET6:
			if (IN6_IS_ADDR_V4MAPPED((struct in6_addr *)ap)) {
				sa->sa_family = AF_INET;
				sa->sa_len = sizeof(struct sockaddr_in);
				memcpy(&((struct sockaddr_in *)sa)->sin_addr,
				    &ap[12], sizeof(struct in_addr));
			} else {
				sa->sa_family = AF_INET6;
				sa->sa_len = sizeof(struct sockaddr_in6);
				memcpy(&((struct sockaddr_in6 *)sa)->sin6_addr,
				    ap, sizeof(struct in6_addr));
			}
			break;
#endif
		}
		aio[i].aio_dstscope = gai_addr2scopetype(sa);
		aio[i].aio_dstpolicy = match_addrselectpolicy(sa, &policyhead);
		set_source(&aio[i], &policyhead);
		aio[i].aio_initial_sequence = i;
	}

	/* perform sorting. */
	qsort(aio, n, sizeof(*aio), comp_dst);

	/* reorder the h_addr_list. */
	for (i = 0; i < n; i++)
		hp->h_addr_list[i] = aio[i].aio_h_addr;

	/* cleanup and return */
	free(aio);
	free_addrselectpolicy(&policyhead);
	return hp;
}

static int
get_addrselectpolicy(struct policyhead *head)
{
#ifdef INET6
	int mib[] = { CTL_NET, PF_INET6, IPPROTO_IPV6, IPV6CTL_ADDRCTLPOLICY };
	size_t l;
	char *buf;
	struct in6_addrpolicy *pol, *ep;

	if (sysctl(mib, sizeof(mib) / sizeof(mib[0]), NULL, &l, NULL, 0) < 0)
		return (0);
	if ((buf = malloc(l)) == NULL)
		return (0);
	if (sysctl(mib, sizeof(mib) / sizeof(mib[0]), buf, &l, NULL, 0) < 0) {
		free(buf);
		return (0);
	}

	ep = (struct in6_addrpolicy *)(buf + l);
	for (pol = (struct in6_addrpolicy *)buf; pol + 1 <= ep; pol++) {
		struct policyqueue *new;

		if ((new = malloc(sizeof(*new))) == NULL) {
			free_addrselectpolicy(head); /* make the list empty */
			break;
		}
		new->pc_policy = *pol;
		TAILQ_INSERT_TAIL(head, new, pc_entry);
	}

	free(buf);
	return (1);
#else
	return (0);
#endif
}

static void
free_addrselectpolicy(struct policyhead *head)
{
	struct policyqueue *ent, *nent;

	for (ent = TAILQ_FIRST(head); ent; ent = nent) {
		nent = TAILQ_NEXT(ent, pc_entry);
		TAILQ_REMOVE(head, ent, pc_entry);
		free(ent);
	}
}

static struct policyqueue *
match_addrselectpolicy(struct sockaddr *addr, struct policyhead *head)
{
#ifdef INET6
	struct policyqueue *ent, *bestent = NULL;
	struct in6_addrpolicy *pol;
	int matchlen, bestmatchlen = -1;
	u_char *mp, *ep, *k, *p, m;
	struct sockaddr_in6 key;

	switch(addr->sa_family) {
	case AF_INET6:
		key = *(struct sockaddr_in6 *)addr;
		break;
	case AF_INET:
		/* convert the address into IPv4-mapped IPv6 address. */
		memset(&key, 0, sizeof(key));
		key.sin6_family = AF_INET6;
		key.sin6_len = sizeof(key);
		_map_v4v6_address(
		    (char *)&((struct sockaddr_in *)addr)->sin_addr,
		    (char *)&key.sin6_addr);
		break;
	default:
		return(NULL);
	}

	for (ent = TAILQ_FIRST(head); ent; ent = TAILQ_NEXT(ent, pc_entry)) {
		pol = &ent->pc_policy;
		matchlen = 0;

		mp = (u_char *)&pol->addrmask.sin6_addr;
		ep = mp + 16;	/* XXX: scope field? */
		k = (u_char *)&key.sin6_addr;
		p = (u_char *)&pol->addr.sin6_addr;
		for (; mp < ep && *mp; mp++, k++, p++) {
			m = *mp;
			if ((*k & m) != *p)
				goto next; /* not match */
			if (m == 0xff) /* short cut for a typical case */
				matchlen += 8;
			else {
				while (m >= 0x80) {
					matchlen++;
					m <<= 1;
				}
			}
		}

		/* matched.  check if this is better than the current best. */
		if (matchlen > bestmatchlen) {
			bestent = ent;
			bestmatchlen = matchlen;
		}

	  next:
		continue;
	}

	return(bestent);
#else
	return(NULL);
#endif

}

static void
set_source(struct hp_order *aio, struct policyhead *ph)
{
	struct sockaddr_storage ss = aio->aio_un.aiou_ss;
	socklen_t srclen;
	int s;

	/* set unspec ("no source is available"), just in case */
	aio->aio_srcsa.sa_family = AF_UNSPEC;
	aio->aio_srcscope = -1;

	switch(ss.ss_family) {
	case AF_INET:
		((struct sockaddr_in *)&ss)->sin_port = htons(1);
		break;
#ifdef INET6
	case AF_INET6:
		((struct sockaddr_in6 *)&ss)->sin6_port = htons(1);
		break;
#endif
	default:		/* ignore unsupported AFs explicitly */
		return;
	}

	/* open a socket to get the source address for the given dst */
	if ((s = _socket(ss.ss_family, SOCK_DGRAM | SOCK_CLOEXEC,
	    IPPROTO_UDP)) < 0)
		return;		/* give up */
	if (_connect(s, (struct sockaddr *)&ss, ss.ss_len) < 0)
		goto cleanup;
	srclen = ss.ss_len;
	if (_getsockname(s, &aio->aio_srcsa, &srclen) < 0) {
		aio->aio_srcsa.sa_family = AF_UNSPEC;
		goto cleanup;
	}
	aio->aio_srcscope = gai_addr2scopetype(&aio->aio_srcsa);
	aio->aio_srcpolicy = match_addrselectpolicy(&aio->aio_srcsa, ph);
	aio->aio_matchlen = matchlen(&aio->aio_srcsa, (struct sockaddr *)&ss);
#ifdef INET6
	if (ss.ss_family == AF_INET6) {
		struct in6_ifreq ifr6;
		u_int32_t flags6;

		memset(&ifr6, 0, sizeof(ifr6));
		memcpy(&ifr6.ifr_addr, &ss, ss.ss_len);
		if (_ioctl(s, SIOCGIFAFLAG_IN6, &ifr6) == 0) {
			flags6 = ifr6.ifr_ifru.ifru_flags6;
			if ((flags6 & IN6_IFF_DEPRECATED))
				aio->aio_srcflag |= AIO_SRCFLAG_DEPRECATED;
		}
	}
#endif

  cleanup:
	_close(s);
	return;
}

static int
matchlen(struct sockaddr *src, struct sockaddr *dst)
{
	int match = 0;
	u_char *s, *d;
	u_char *lim, r;
	int addrlen;

	switch (src->sa_family) {
#ifdef INET6
	case AF_INET6:
		s = (u_char *)&((struct sockaddr_in6 *)src)->sin6_addr;
		d = (u_char *)&((struct sockaddr_in6 *)dst)->sin6_addr;
		addrlen = sizeof(struct in6_addr);
		lim = s + addrlen;
		break;
#endif
	case AF_INET:
		s = (u_char *)&((struct sockaddr_in *)src)->sin_addr;
		d = (u_char *)&((struct sockaddr_in *)dst)->sin_addr;
		addrlen = sizeof(struct in_addr);
		lim = s + addrlen;
		break;
	default:
		return(0);
	}

	while (s < lim)
		if ((r = (*d++ ^ *s++)) != 0) {
			while ((r & 0x80) == 0) {
				match++;
				r <<= 1;
			}
			break;
		} else
			match += 8;
	return(match);
}

static int
comp_dst(const void *arg1, const void *arg2)
{
	const struct hp_order *dst1 = arg1, *dst2 = arg2;

	/*
	 * Rule 1: Avoid unusable destinations.
	 * XXX: we currently do not consider if an appropriate route exists.
	 */
	if (dst1->aio_srcsa.sa_family != AF_UNSPEC &&
	    dst2->aio_srcsa.sa_family == AF_UNSPEC) {
		return(-1);
	}
	if (dst1->aio_srcsa.sa_family == AF_UNSPEC &&
	    dst2->aio_srcsa.sa_family != AF_UNSPEC) {
		return(1);
	}

	/* Rule 2: Prefer matching scope. */
	if (dst1->aio_dstscope == dst1->aio_srcscope &&
	    dst2->aio_dstscope != dst2->aio_srcscope) {
		return(-1);
	}
	if (dst1->aio_dstscope != dst1->aio_srcscope &&
	    dst2->aio_dstscope == dst2->aio_srcscope) {
		return(1);
	}

	/* Rule 3: Avoid deprecated addresses. */
	if (dst1->aio_srcsa.sa_family != AF_UNSPEC &&
	    dst2->aio_srcsa.sa_family != AF_UNSPEC) {
		if (!(dst1->aio_srcflag & AIO_SRCFLAG_DEPRECATED) &&
		    (dst2->aio_srcflag & AIO_SRCFLAG_DEPRECATED)) {
			return(-1);
		}
		if ((dst1->aio_srcflag & AIO_SRCFLAG_DEPRECATED) &&
		    !(dst2->aio_srcflag & AIO_SRCFLAG_DEPRECATED)) {
			return(1);
		}
	}

	/* Rule 4: Prefer home addresses. */
	/* XXX: not implemented yet */

	/* Rule 5: Prefer matching label. */
#ifdef INET6
	if (dst1->aio_srcpolicy && dst1->aio_dstpolicy &&
	    dst1->aio_srcpolicy->pc_policy.label ==
	    dst1->aio_dstpolicy->pc_policy.label &&
	    (dst2->aio_srcpolicy == NULL || dst2->aio_dstpolicy == NULL ||
	     dst2->aio_srcpolicy->pc_policy.label !=
	     dst2->aio_dstpolicy->pc_policy.label)) {
		return(-1);
	}
	if (dst2->aio_srcpolicy && dst2->aio_dstpolicy &&
	    dst2->aio_srcpolicy->pc_policy.label ==
	    dst2->aio_dstpolicy->pc_policy.label &&
	    (dst1->aio_srcpolicy == NULL || dst1->aio_dstpolicy == NULL ||
	     dst1->aio_srcpolicy->pc_policy.label !=
	     dst1->aio_dstpolicy->pc_policy.label)) {
		return(1);
	}
#endif

	/* Rule 6: Prefer higher precedence. */
#ifdef INET6
	if (dst1->aio_dstpolicy &&
	    (dst2->aio_dstpolicy == NULL ||
	     dst1->aio_dstpolicy->pc_policy.preced >
	     dst2->aio_dstpolicy->pc_policy.preced)) {
		return(-1);
	}
	if (dst2->aio_dstpolicy &&
	    (dst1->aio_dstpolicy == NULL ||
	     dst2->aio_dstpolicy->pc_policy.preced >
	     dst1->aio_dstpolicy->pc_policy.preced)) {
		return(1);
	}
#endif

	/* Rule 7: Prefer native transport. */
	/* XXX: not implemented yet */

	/* Rule 8: Prefer smaller scope. */
	if (dst1->aio_dstscope >= 0 &&
	    dst1->aio_dstscope < dst2->aio_dstscope) {
		return(-1);
	}
	if (dst2->aio_dstscope >= 0 &&
	    dst2->aio_dstscope < dst1->aio_dstscope) {
		return(1);
	}

	/*
	 * Rule 9: Use longest matching prefix.
	 * We compare the match length in a same AF only.
	 */
	if (dst1->aio_sa.sa_family == dst2->aio_sa.sa_family) {
		if (dst1->aio_matchlen > dst2->aio_matchlen) {
			return(-1);
		}
		if (dst1->aio_matchlen < dst2->aio_matchlen) {
			return(1);
		}
	}

	/* Rule 10: Otherwise, leave the order unchanged. */

	/*
	 * Note that qsort is unstable; so, we can't return zero and
	 * expect the order to be unchanged.
	 * That also means we can't depend on the current position of
	 * dst2 being after dst1.  We must enforce the initial order
	 * with an explicit compare on the original position.
	 * The qsort specification requires that "When the same objects
	 * (consisting of width bytes, irrespective of their current
	 * positions in the array) are passed more than once to the
	 * comparison function, the results shall be consistent with one
	 * another."
	 * In other words, If A < B, then we must also return B > A.
	 */
	if (dst2->aio_initial_sequence < dst1->aio_initial_sequence)
		return(1);

	return(-1);
}

/*
 * Copy from scope.c.
 * XXX: we should standardize the functions and link them as standard
 * library.
 */
static int
gai_addr2scopetype(struct sockaddr *sa)
{
#ifdef INET6
	struct sockaddr_in6 *sa6;
#endif
	struct sockaddr_in *sa4;

	switch(sa->sa_family) {
#ifdef INET6
	case AF_INET6:
		sa6 = (struct sockaddr_in6 *)sa;
		if (IN6_IS_ADDR_MULTICAST(&sa6->sin6_addr)) {
			/* just use the scope field of the multicast address */
			return(sa6->sin6_addr.s6_addr[2] & 0x0f);
		}
		/*
		 * Unicast addresses: map scope type to corresponding scope
		 * value defined for multcast addresses.
		 * XXX: hardcoded scope type values are bad...
		 */
		if (IN6_IS_ADDR_LOOPBACK(&sa6->sin6_addr))
			return(1); /* node local scope */
		if (IN6_IS_ADDR_LINKLOCAL(&sa6->sin6_addr))
			return(2); /* link-local scope */
		if (IN6_IS_ADDR_SITELOCAL(&sa6->sin6_addr))
			return(5); /* site-local scope */
		return(14);	/* global scope */
		break;
#endif
	case AF_INET:
		/*
		 * IPv4 pseudo scoping according to RFC 3484.
		 */
		sa4 = (struct sockaddr_in *)sa;
		/* IPv4 autoconfiguration addresses have link-local scope. */
		if (((u_char *)&sa4->sin_addr)[0] == 169 &&
		    ((u_char *)&sa4->sin_addr)[1] == 254)
			return(2);
		/* Private addresses have site-local scope. */
		if (((u_char *)&sa4->sin_addr)[0] == 10 ||
		    (((u_char *)&sa4->sin_addr)[0] == 172 &&
		     (((u_char *)&sa4->sin_addr)[1] & 0xf0) == 16) ||
		    (((u_char *)&sa4->sin_addr)[0] == 192 &&
		     ((u_char *)&sa4->sin_addr)[1] == 168))
			return(14);	/* XXX: It should be 5 unless NAT */
		/* Loopback addresses have link-local scope. */
		if (((u_char *)&sa4->sin_addr)[0] == 127)
			return(2);
		return(14);
		break;
	default:
		errno = EAFNOSUPPORT; /* is this a good error? */
		return(-1);
	}
}
#endif