in6.c revision 337378e04f710de54c27deb773c544bf0eeda73a
1/*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (C) 1995, 1996, 1997, and 1998 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 *	$KAME: in6.c,v 1.259 2002/01/21 11:37:50 keiichi Exp $
32 */
33
34/*-
35 * Copyright (c) 1982, 1986, 1991, 1993
36 *	The Regents of the University of California.  All rights reserved.
37 *
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
40 * are met:
41 * 1. Redistributions of source code must retain the above copyright
42 *    notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 *    notice, this list of conditions and the following disclaimer in the
45 *    documentation and/or other materials provided with the distribution.
46 * 3. Neither the name of the University nor the names of its contributors
47 *    may be used to endorse or promote products derived from this software
48 *    without specific prior written permission.
49 *
50 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
51 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
52 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
53 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
54 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
55 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
56 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
57 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
58 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
59 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60 * SUCH DAMAGE.
61 *
62 *	@(#)in.c	8.2 (Berkeley) 11/15/93
63 */
64
65#include <sys/cdefs.h>
66__FBSDID("$FreeBSD$");
67
68#include "opt_inet.h"
69#include "opt_inet6.h"
70
71#include <sys/param.h>
72#include <sys/eventhandler.h>
73#include <sys/errno.h>
74#include <sys/jail.h>
75#include <sys/malloc.h>
76#include <sys/socket.h>
77#include <sys/socketvar.h>
78#include <sys/sockio.h>
79#include <sys/systm.h>
80#include <sys/priv.h>
81#include <sys/proc.h>
82#include <sys/protosw.h>
83#include <sys/time.h>
84#include <sys/kernel.h>
85#include <sys/lock.h>
86#include <sys/rmlock.h>
87#include <sys/syslog.h>
88
89#include <net/if.h>
90#include <net/if_var.h>
91#include <net/if_types.h>
92#include <net/route.h>
93#include <net/if_dl.h>
94#include <net/vnet.h>
95
96#include <netinet/in.h>
97#include <netinet/in_var.h>
98#include <net/if_llatbl.h>
99#include <netinet/if_ether.h>
100#include <netinet/in_systm.h>
101#include <netinet/ip.h>
102#include <netinet/in_pcb.h>
103#include <netinet/ip_carp.h>
104
105#include <netinet/ip6.h>
106#include <netinet6/ip6_var.h>
107#include <netinet6/nd6.h>
108#include <netinet6/mld6_var.h>
109#include <netinet6/ip6_mroute.h>
110#include <netinet6/in6_ifattach.h>
111#include <netinet6/scope6_var.h>
112#include <netinet6/in6_fib.h>
113#include <netinet6/in6_pcb.h>
114
115
116/*
117 * struct in6_ifreq and struct ifreq must be type punnable for common members
118 * of ifr_ifru to allow accessors to be shared.
119 */
120_Static_assert(offsetof(struct in6_ifreq, ifr_ifru) ==
121    offsetof(struct ifreq, ifr_ifru),
122    "struct in6_ifreq and struct ifreq are not type punnable");
123
124VNET_DECLARE(int, icmp6_nodeinfo_oldmcprefix);
125#define V_icmp6_nodeinfo_oldmcprefix	VNET(icmp6_nodeinfo_oldmcprefix)
126
127/*
128 * Definitions of some costant IP6 addresses.
129 */
130const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT;
131const struct in6_addr in6addr_loopback = IN6ADDR_LOOPBACK_INIT;
132const struct in6_addr in6addr_nodelocal_allnodes =
133	IN6ADDR_NODELOCAL_ALLNODES_INIT;
134const struct in6_addr in6addr_linklocal_allnodes =
135	IN6ADDR_LINKLOCAL_ALLNODES_INIT;
136const struct in6_addr in6addr_linklocal_allrouters =
137	IN6ADDR_LINKLOCAL_ALLROUTERS_INIT;
138const struct in6_addr in6addr_linklocal_allv2routers =
139	IN6ADDR_LINKLOCAL_ALLV2ROUTERS_INIT;
140
141const struct in6_addr in6mask0 = IN6MASK0;
142const struct in6_addr in6mask32 = IN6MASK32;
143const struct in6_addr in6mask64 = IN6MASK64;
144const struct in6_addr in6mask96 = IN6MASK96;
145const struct in6_addr in6mask128 = IN6MASK128;
146
147const struct sockaddr_in6 sa6_any =
148	{ sizeof(sa6_any), AF_INET6, 0, 0, IN6ADDR_ANY_INIT, 0 };
149
150static int in6_notify_ifa(struct ifnet *, struct in6_ifaddr *,
151	struct in6_aliasreq *, int);
152static void in6_unlink_ifa(struct in6_ifaddr *, struct ifnet *);
153
154static int in6_validate_ifra(struct ifnet *, struct in6_aliasreq *,
155    struct in6_ifaddr *, int);
156static struct in6_ifaddr *in6_alloc_ifa(struct ifnet *,
157    struct in6_aliasreq *, int flags);
158static int in6_update_ifa_internal(struct ifnet *, struct in6_aliasreq *,
159    struct in6_ifaddr *, int, int);
160static int in6_broadcast_ifa(struct ifnet *, struct in6_aliasreq *,
161    struct in6_ifaddr *, int);
162
163#define ifa2ia6(ifa)	((struct in6_ifaddr *)(ifa))
164#define ia62ifa(ia6)	(&((ia6)->ia_ifa))
165
166
167void
168in6_newaddrmsg(struct in6_ifaddr *ia, int cmd)
169{
170	struct sockaddr_dl gateway;
171	struct sockaddr_in6 mask, addr;
172	struct rtentry rt;
173	int fibnum;
174
175	/*
176	 * initialize for rtmsg generation
177	 */
178	bzero(&gateway, sizeof(gateway));
179	gateway.sdl_len = sizeof(gateway);
180	gateway.sdl_family = AF_LINK;
181
182	bzero(&rt, sizeof(rt));
183	rt.rt_gateway = (struct sockaddr *)&gateway;
184	memcpy(&mask, &ia->ia_prefixmask, sizeof(ia->ia_prefixmask));
185	memcpy(&addr, &ia->ia_addr, sizeof(ia->ia_addr));
186	rt_mask(&rt) = (struct sockaddr *)&mask;
187	rt_key(&rt) = (struct sockaddr *)&addr;
188	rt.rt_flags = RTF_HOST | RTF_STATIC;
189	if (cmd == RTM_ADD)
190		rt.rt_flags |= RTF_UP;
191	fibnum = V_rt_add_addr_allfibs ? RT_ALL_FIBS : ia62ifa(ia)->ifa_ifp->if_fib;
192	/* Announce arrival of local address to this FIB. */
193	rt_newaddrmsg_fib(cmd, &ia->ia_ifa, 0, &rt, fibnum);
194}
195
196int
197in6_mask2len(struct in6_addr *mask, u_char *lim0)
198{
199	int x = 0, y;
200	u_char *lim = lim0, *p;
201
202	/* ignore the scope_id part */
203	if (lim0 == NULL || lim0 - (u_char *)mask > sizeof(*mask))
204		lim = (u_char *)mask + sizeof(*mask);
205	for (p = (u_char *)mask; p < lim; x++, p++) {
206		if (*p != 0xff)
207			break;
208	}
209	y = 0;
210	if (p < lim) {
211		for (y = 0; y < 8; y++) {
212			if ((*p & (0x80 >> y)) == 0)
213				break;
214		}
215	}
216
217	/*
218	 * when the limit pointer is given, do a stricter check on the
219	 * remaining bits.
220	 */
221	if (p < lim) {
222		if (y != 0 && (*p & (0x00ff >> y)) != 0)
223			return (-1);
224		for (p = p + 1; p < lim; p++)
225			if (*p != 0)
226				return (-1);
227	}
228
229	return x * 8 + y;
230}
231
232#ifdef COMPAT_FREEBSD32
233struct in6_ndifreq32 {
234	char ifname[IFNAMSIZ];
235	uint32_t ifindex;
236};
237#define	SIOCGDEFIFACE32_IN6	_IOWR('i', 86, struct in6_ndifreq32)
238#endif
239
240int
241in6_control(struct socket *so, u_long cmd, caddr_t data,
242    struct ifnet *ifp, struct thread *td)
243{
244	struct	in6_ifreq *ifr = (struct in6_ifreq *)data;
245	struct	in6_ifaddr *ia = NULL;
246	struct	in6_aliasreq *ifra = (struct in6_aliasreq *)data;
247	struct sockaddr_in6 *sa6;
248	int carp_attached = 0;
249	int error;
250	u_long ocmd = cmd;
251
252	/*
253	 * Compat to make pre-10.x ifconfig(8) operable.
254	 */
255	if (cmd == OSIOCAIFADDR_IN6)
256		cmd = SIOCAIFADDR_IN6;
257
258	switch (cmd) {
259	case SIOCGETSGCNT_IN6:
260	case SIOCGETMIFCNT_IN6:
261		/*
262		 * XXX mrt_ioctl has a 3rd, unused, FIB argument in route.c.
263		 * We cannot see how that would be needed, so do not adjust the
264		 * KPI blindly; more likely should clean up the IPv4 variant.
265		 */
266		return (mrt6_ioctl ? mrt6_ioctl(cmd, data) : EOPNOTSUPP);
267	}
268
269	switch (cmd) {
270	case SIOCAADDRCTL_POLICY:
271	case SIOCDADDRCTL_POLICY:
272		if (td != NULL) {
273			error = priv_check(td, PRIV_NETINET_ADDRCTRL6);
274			if (error)
275				return (error);
276		}
277		return (in6_src_ioctl(cmd, data));
278	}
279
280	if (ifp == NULL)
281		return (EOPNOTSUPP);
282
283	switch (cmd) {
284	case SIOCSNDFLUSH_IN6:
285	case SIOCSPFXFLUSH_IN6:
286	case SIOCSRTRFLUSH_IN6:
287	case SIOCSDEFIFACE_IN6:
288	case SIOCSIFINFO_FLAGS:
289	case SIOCSIFINFO_IN6:
290		if (td != NULL) {
291			error = priv_check(td, PRIV_NETINET_ND6);
292			if (error)
293				return (error);
294		}
295		/* FALLTHROUGH */
296	case OSIOCGIFINFO_IN6:
297	case SIOCGIFINFO_IN6:
298	case SIOCGNBRINFO_IN6:
299	case SIOCGDEFIFACE_IN6:
300		return (nd6_ioctl(cmd, data, ifp));
301
302#ifdef COMPAT_FREEBSD32
303	case SIOCGDEFIFACE32_IN6:
304		{
305			struct in6_ndifreq ndif;
306			struct in6_ndifreq32 *ndif32;
307
308			error = nd6_ioctl(SIOCGDEFIFACE_IN6, (caddr_t)&ndif,
309			    ifp);
310			if (error)
311				return (error);
312			ndif32 = (struct in6_ndifreq32 *)data;
313			ndif32->ifindex = ndif.ifindex;
314			return (0);
315		}
316#endif
317	}
318
319	switch (cmd) {
320	case SIOCSIFPREFIX_IN6:
321	case SIOCDIFPREFIX_IN6:
322	case SIOCAIFPREFIX_IN6:
323	case SIOCCIFPREFIX_IN6:
324	case SIOCSGIFPREFIX_IN6:
325	case SIOCGIFPREFIX_IN6:
326		log(LOG_NOTICE,
327		    "prefix ioctls are now invalidated. "
328		    "please use ifconfig.\n");
329		return (EOPNOTSUPP);
330	}
331
332	switch (cmd) {
333	case SIOCSSCOPE6:
334		if (td != NULL) {
335			error = priv_check(td, PRIV_NETINET_SCOPE6);
336			if (error)
337				return (error);
338		}
339		/* FALLTHROUGH */
340	case SIOCGSCOPE6:
341	case SIOCGSCOPE6DEF:
342		return (scope6_ioctl(cmd, data, ifp));
343	}
344
345	/*
346	 * Find address for this interface, if it exists.
347	 *
348	 * In netinet code, we have checked ifra_addr in SIOCSIF*ADDR operation
349	 * only, and used the first interface address as the target of other
350	 * operations (without checking ifra_addr).  This was because netinet
351	 * code/API assumed at most 1 interface address per interface.
352	 * Since IPv6 allows a node to assign multiple addresses
353	 * on a single interface, we almost always look and check the
354	 * presence of ifra_addr, and reject invalid ones here.
355	 * It also decreases duplicated code among SIOC*_IN6 operations.
356	 */
357	switch (cmd) {
358	case SIOCAIFADDR_IN6:
359	case SIOCSIFPHYADDR_IN6:
360		sa6 = &ifra->ifra_addr;
361		break;
362	case SIOCSIFADDR_IN6:
363	case SIOCGIFADDR_IN6:
364	case SIOCSIFDSTADDR_IN6:
365	case SIOCSIFNETMASK_IN6:
366	case SIOCGIFDSTADDR_IN6:
367	case SIOCGIFNETMASK_IN6:
368	case SIOCDIFADDR_IN6:
369	case SIOCGIFPSRCADDR_IN6:
370	case SIOCGIFPDSTADDR_IN6:
371	case SIOCGIFAFLAG_IN6:
372	case SIOCSNDFLUSH_IN6:
373	case SIOCSPFXFLUSH_IN6:
374	case SIOCSRTRFLUSH_IN6:
375	case SIOCGIFALIFETIME_IN6:
376	case SIOCGIFSTAT_IN6:
377	case SIOCGIFSTAT_ICMP6:
378		sa6 = &ifr->ifr_addr;
379		break;
380	case SIOCSIFADDR:
381	case SIOCSIFBRDADDR:
382	case SIOCSIFDSTADDR:
383	case SIOCSIFNETMASK:
384		/*
385		 * Although we should pass any non-INET6 ioctl requests
386		 * down to driver, we filter some legacy INET requests.
387		 * Drivers trust SIOCSIFADDR et al to come from an already
388		 * privileged layer, and do not perform any credentials
389		 * checks or input validation.
390		 */
391		return (EINVAL);
392	default:
393		sa6 = NULL;
394		break;
395	}
396	if (sa6 && sa6->sin6_family == AF_INET6) {
397		if (sa6->sin6_scope_id != 0)
398			error = sa6_embedscope(sa6, 0);
399		else
400			error = in6_setscope(&sa6->sin6_addr, ifp, NULL);
401		if (error != 0)
402			return (error);
403		if (td != NULL && (error = prison_check_ip6(td->td_ucred,
404		    &sa6->sin6_addr)) != 0)
405			return (error);
406		ia = in6ifa_ifpwithaddr(ifp, &sa6->sin6_addr);
407	} else
408		ia = NULL;
409
410	switch (cmd) {
411	case SIOCSIFADDR_IN6:
412	case SIOCSIFDSTADDR_IN6:
413	case SIOCSIFNETMASK_IN6:
414		/*
415		 * Since IPv6 allows a node to assign multiple addresses
416		 * on a single interface, SIOCSIFxxx ioctls are deprecated.
417		 */
418		/* we decided to obsolete this command (20000704) */
419		error = EINVAL;
420		goto out;
421
422	case SIOCDIFADDR_IN6:
423		/*
424		 * for IPv4, we look for existing in_ifaddr here to allow
425		 * "ifconfig if0 delete" to remove the first IPv4 address on
426		 * the interface.  For IPv6, as the spec allows multiple
427		 * interface address from the day one, we consider "remove the
428		 * first one" semantics to be not preferable.
429		 */
430		if (ia == NULL) {
431			error = EADDRNOTAVAIL;
432			goto out;
433		}
434		/* FALLTHROUGH */
435	case SIOCAIFADDR_IN6:
436		/*
437		 * We always require users to specify a valid IPv6 address for
438		 * the corresponding operation.
439		 */
440		if (ifra->ifra_addr.sin6_family != AF_INET6 ||
441		    ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6)) {
442			error = EAFNOSUPPORT;
443			goto out;
444		}
445
446		if (td != NULL) {
447			error = priv_check(td, (cmd == SIOCDIFADDR_IN6) ?
448			    PRIV_NET_DELIFADDR : PRIV_NET_ADDIFADDR);
449			if (error)
450				goto out;
451		}
452		/* FALLTHROUGH */
453	case SIOCGIFSTAT_IN6:
454	case SIOCGIFSTAT_ICMP6:
455		if (ifp->if_afdata[AF_INET6] == NULL) {
456			error = EPFNOSUPPORT;
457			goto out;
458		}
459		break;
460
461	case SIOCGIFADDR_IN6:
462		/* This interface is basically deprecated. use SIOCGIFCONF. */
463		/* FALLTHROUGH */
464	case SIOCGIFAFLAG_IN6:
465	case SIOCGIFNETMASK_IN6:
466	case SIOCGIFDSTADDR_IN6:
467	case SIOCGIFALIFETIME_IN6:
468		/* must think again about its semantics */
469		if (ia == NULL) {
470			error = EADDRNOTAVAIL;
471			goto out;
472		}
473		break;
474	}
475
476	switch (cmd) {
477	case SIOCGIFADDR_IN6:
478		ifr->ifr_addr = ia->ia_addr;
479		if ((error = sa6_recoverscope(&ifr->ifr_addr)) != 0)
480			goto out;
481		break;
482
483	case SIOCGIFDSTADDR_IN6:
484		if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
485			error = EINVAL;
486			goto out;
487		}
488		ifr->ifr_dstaddr = ia->ia_dstaddr;
489		if ((error = sa6_recoverscope(&ifr->ifr_dstaddr)) != 0)
490			goto out;
491		break;
492
493	case SIOCGIFNETMASK_IN6:
494		ifr->ifr_addr = ia->ia_prefixmask;
495		break;
496
497	case SIOCGIFAFLAG_IN6:
498		ifr->ifr_ifru.ifru_flags6 = ia->ia6_flags;
499		break;
500
501	case SIOCGIFSTAT_IN6:
502		COUNTER_ARRAY_COPY(((struct in6_ifextra *)
503		    ifp->if_afdata[AF_INET6])->in6_ifstat,
504		    &ifr->ifr_ifru.ifru_stat,
505		    sizeof(struct in6_ifstat) / sizeof(uint64_t));
506		break;
507
508	case SIOCGIFSTAT_ICMP6:
509		COUNTER_ARRAY_COPY(((struct in6_ifextra *)
510		    ifp->if_afdata[AF_INET6])->icmp6_ifstat,
511		    &ifr->ifr_ifru.ifru_icmp6stat,
512		    sizeof(struct icmp6_ifstat) / sizeof(uint64_t));
513		break;
514
515	case SIOCGIFALIFETIME_IN6:
516		ifr->ifr_ifru.ifru_lifetime = ia->ia6_lifetime;
517		if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
518			time_t maxexpire;
519			struct in6_addrlifetime *retlt =
520			    &ifr->ifr_ifru.ifru_lifetime;
521
522			/*
523			 * XXX: adjust expiration time assuming time_t is
524			 * signed.
525			 */
526			maxexpire = (-1) &
527			    ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
528			if (ia->ia6_lifetime.ia6t_vltime <
529			    maxexpire - ia->ia6_updatetime) {
530				retlt->ia6t_expire = ia->ia6_updatetime +
531				    ia->ia6_lifetime.ia6t_vltime;
532			} else
533				retlt->ia6t_expire = maxexpire;
534		}
535		if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
536			time_t maxexpire;
537			struct in6_addrlifetime *retlt =
538			    &ifr->ifr_ifru.ifru_lifetime;
539
540			/*
541			 * XXX: adjust expiration time assuming time_t is
542			 * signed.
543			 */
544			maxexpire = (-1) &
545			    ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
546			if (ia->ia6_lifetime.ia6t_pltime <
547			    maxexpire - ia->ia6_updatetime) {
548				retlt->ia6t_preferred = ia->ia6_updatetime +
549				    ia->ia6_lifetime.ia6t_pltime;
550			} else
551				retlt->ia6t_preferred = maxexpire;
552		}
553		break;
554
555	case SIOCAIFADDR_IN6:
556	{
557		struct nd_prefixctl pr0;
558		struct nd_prefix *pr;
559
560		/*
561		 * first, make or update the interface address structure,
562		 * and link it to the list.
563		 */
564		if ((error = in6_update_ifa(ifp, ifra, ia, 0)) != 0)
565			goto out;
566		if (ia != NULL) {
567			if (ia->ia_ifa.ifa_carp)
568				(*carp_detach_p)(&ia->ia_ifa, true);
569			ifa_free(&ia->ia_ifa);
570		}
571		if ((ia = in6ifa_ifpwithaddr(ifp, &ifra->ifra_addr.sin6_addr))
572		    == NULL) {
573			/*
574			 * this can happen when the user specify the 0 valid
575			 * lifetime.
576			 */
577			break;
578		}
579
580		if (cmd == ocmd && ifra->ifra_vhid > 0) {
581			if (carp_attach_p != NULL)
582				error = (*carp_attach_p)(&ia->ia_ifa,
583				    ifra->ifra_vhid);
584			else
585				error = EPROTONOSUPPORT;
586			if (error)
587				goto out;
588			else
589				carp_attached = 1;
590		}
591
592		/*
593		 * then, make the prefix on-link on the interface.
594		 * XXX: we'd rather create the prefix before the address, but
595		 * we need at least one address to install the corresponding
596		 * interface route, so we configure the address first.
597		 */
598
599		/*
600		 * convert mask to prefix length (prefixmask has already
601		 * been validated in in6_update_ifa().
602		 */
603		bzero(&pr0, sizeof(pr0));
604		pr0.ndpr_ifp = ifp;
605		pr0.ndpr_plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr,
606		    NULL);
607		if (pr0.ndpr_plen == 128) {
608			/* we don't need to install a host route. */
609			goto aifaddr_out;
610		}
611		pr0.ndpr_prefix = ifra->ifra_addr;
612		/* apply the mask for safety. */
613		IN6_MASK_ADDR(&pr0.ndpr_prefix.sin6_addr,
614		    &ifra->ifra_prefixmask.sin6_addr);
615
616		/*
617		 * XXX: since we don't have an API to set prefix (not address)
618		 * lifetimes, we just use the same lifetimes as addresses.
619		 * The (temporarily) installed lifetimes can be overridden by
620		 * later advertised RAs (when accept_rtadv is non 0), which is
621		 * an intended behavior.
622		 */
623		pr0.ndpr_raf_onlink = 1; /* should be configurable? */
624		pr0.ndpr_raf_auto =
625		    ((ifra->ifra_flags & IN6_IFF_AUTOCONF) != 0);
626		pr0.ndpr_vltime = ifra->ifra_lifetime.ia6t_vltime;
627		pr0.ndpr_pltime = ifra->ifra_lifetime.ia6t_pltime;
628
629		/* add the prefix if not yet. */
630		if ((pr = nd6_prefix_lookup(&pr0)) == NULL) {
631			/*
632			 * nd6_prelist_add will install the corresponding
633			 * interface route.
634			 */
635			if ((error = nd6_prelist_add(&pr0, NULL, &pr)) != 0) {
636				if (carp_attached)
637					(*carp_detach_p)(&ia->ia_ifa, false);
638				goto out;
639			}
640		}
641
642		/* relate the address to the prefix */
643		if (ia->ia6_ndpr == NULL) {
644			ia->ia6_ndpr = pr;
645			pr->ndpr_addrcnt++;
646
647			/*
648			 * If this is the first autoconf address from the
649			 * prefix, create a temporary address as well
650			 * (when required).
651			 */
652			if ((ia->ia6_flags & IN6_IFF_AUTOCONF) &&
653			    V_ip6_use_tempaddr && pr->ndpr_addrcnt == 1) {
654				int e;
655				if ((e = in6_tmpifadd(ia, 1, 0)) != 0) {
656					log(LOG_NOTICE, "in6_control: failed "
657					    "to create a temporary address, "
658					    "errno=%d\n", e);
659				}
660			}
661		}
662		nd6_prefix_rele(pr);
663
664		/*
665		 * this might affect the status of autoconfigured addresses,
666		 * that is, this address might make other addresses detached.
667		 */
668		pfxlist_onlink_check();
669
670aifaddr_out:
671		/*
672		 * Try to clear the flag when a new IPv6 address is added
673		 * onto an IFDISABLED interface and it succeeds.
674		 */
675		if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) {
676			struct in6_ndireq nd;
677
678			memset(&nd, 0, sizeof(nd));
679			nd.ndi.flags = ND_IFINFO(ifp)->flags;
680			nd.ndi.flags &= ~ND6_IFF_IFDISABLED;
681			if (nd6_ioctl(SIOCSIFINFO_FLAGS, (caddr_t)&nd, ifp) < 0)
682				log(LOG_NOTICE, "SIOCAIFADDR_IN6: "
683				    "SIOCSIFINFO_FLAGS for -ifdisabled "
684				    "failed.");
685			/*
686			 * Ignore failure of clearing the flag intentionally.
687			 * The failure means address duplication was detected.
688			 */
689		}
690		break;
691	}
692
693	case SIOCDIFADDR_IN6:
694	{
695		struct nd_prefix *pr;
696
697		/*
698		 * If the address being deleted is the only one that owns
699		 * the corresponding prefix, expire the prefix as well.
700		 * XXX: theoretically, we don't have to worry about such
701		 * relationship, since we separate the address management
702		 * and the prefix management.  We do this, however, to provide
703		 * as much backward compatibility as possible in terms of
704		 * the ioctl operation.
705		 * Note that in6_purgeaddr() will decrement ndpr_addrcnt.
706		 */
707		pr = ia->ia6_ndpr;
708		in6_purgeaddr(&ia->ia_ifa);
709		if (pr != NULL && pr->ndpr_addrcnt == 0) {
710			ND6_WLOCK();
711			nd6_prefix_unlink(pr, NULL);
712			ND6_WUNLOCK();
713			nd6_prefix_del(pr);
714		}
715		EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa,
716		    IFADDR_EVENT_DEL);
717		break;
718	}
719
720	default:
721		if (ifp->if_ioctl == NULL) {
722			error = EOPNOTSUPP;
723			goto out;
724		}
725		error = (*ifp->if_ioctl)(ifp, cmd, data);
726		goto out;
727	}
728
729	error = 0;
730out:
731	if (ia != NULL)
732		ifa_free(&ia->ia_ifa);
733	return (error);
734}
735
736
737static struct in6_multi_mship *
738in6_joingroup_legacy(struct ifnet *ifp, const struct in6_addr *mcaddr,
739    int *errorp, int delay)
740{
741	struct in6_multi_mship *imm;
742	int error;
743
744	imm = malloc(sizeof(*imm), M_IP6MADDR, M_NOWAIT);
745	if (imm == NULL) {
746		*errorp = ENOBUFS;
747		return (NULL);
748	}
749
750	delay = (delay * PR_FASTHZ) / hz;
751
752	error = in6_joingroup(ifp, mcaddr, NULL, &imm->i6mm_maddr, delay);
753	if (error) {
754		*errorp = error;
755		free(imm, M_IP6MADDR);
756		return (NULL);
757	}
758
759	return (imm);
760}
761/*
762 * Join necessary multicast groups.  Factored out from in6_update_ifa().
763 * This entire work should only be done once, for the default FIB.
764 */
765static int
766in6_update_ifa_join_mc(struct ifnet *ifp, struct in6_aliasreq *ifra,
767    struct in6_ifaddr *ia, int flags, struct in6_multi **in6m_sol)
768{
769	char ip6buf[INET6_ADDRSTRLEN];
770	struct in6_addr mltaddr;
771	struct in6_multi_mship *imm;
772	int delay, error;
773
774	KASSERT(in6m_sol != NULL, ("%s: in6m_sol is NULL", __func__));
775
776	/* Join solicited multicast addr for new host id. */
777	bzero(&mltaddr, sizeof(struct in6_addr));
778	mltaddr.s6_addr32[0] = IPV6_ADDR_INT32_MLL;
779	mltaddr.s6_addr32[2] = htonl(1);
780	mltaddr.s6_addr32[3] = ifra->ifra_addr.sin6_addr.s6_addr32[3];
781	mltaddr.s6_addr8[12] = 0xff;
782	if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0) {
783		/* XXX: should not happen */
784		log(LOG_ERR, "%s: in6_setscope failed\n", __func__);
785		goto cleanup;
786	}
787	delay = error = 0;
788	if ((flags & IN6_IFAUPDATE_DADDELAY)) {
789		/*
790		 * We need a random delay for DAD on the address being
791		 * configured.  It also means delaying transmission of the
792		 * corresponding MLD report to avoid report collision.
793		 * [RFC 4861, Section 6.3.7]
794		 */
795		delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
796	}
797	imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
798	if (imm == NULL) {
799		nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
800		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr),
801		    if_name(ifp), error));
802		goto cleanup;
803	}
804	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
805	*in6m_sol = imm->i6mm_maddr;
806
807	/*
808	 * Join link-local all-nodes address.
809	 */
810	mltaddr = in6addr_linklocal_allnodes;
811	if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
812		goto cleanup; /* XXX: should not fail */
813
814	imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0);
815	if (imm == NULL) {
816		nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
817		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr),
818		    if_name(ifp), error));
819		goto cleanup;
820	}
821	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
822
823	/*
824	 * Join node information group address.
825	 */
826	delay = 0;
827	if ((flags & IN6_IFAUPDATE_DADDELAY)) {
828		/*
829		 * The spec does not say anything about delay for this group,
830		 * but the same logic should apply.
831		 */
832		delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
833	}
834	if (in6_nigroup(ifp, NULL, -1, &mltaddr) == 0) {
835		/* XXX jinmei */
836		imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
837		if (imm == NULL)
838			nd6log((LOG_WARNING,
839			    "%s: in6_joingroup failed for %s on %s "
840			    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
841			    &mltaddr), if_name(ifp), error));
842			/* XXX not very fatal, go on... */
843		else
844			LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
845	}
846	if (V_icmp6_nodeinfo_oldmcprefix &&
847	    in6_nigroup_oldmcprefix(ifp, NULL, -1, &mltaddr) == 0) {
848		imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
849		if (imm == NULL)
850			nd6log((LOG_WARNING,
851			    "%s: in6_joingroup failed for %s on %s "
852			    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
853			    &mltaddr), if_name(ifp), error));
854			/* XXX not very fatal, go on... */
855		else
856			LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
857	}
858
859	/*
860	 * Join interface-local all-nodes address.
861	 * (ff01::1%ifN, and ff01::%ifN/32)
862	 */
863	mltaddr = in6addr_nodelocal_allnodes;
864	if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
865		goto cleanup; /* XXX: should not fail */
866
867	imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0);
868	if (imm == NULL) {
869		nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
870		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
871		    &mltaddr), if_name(ifp), error));
872		goto cleanup;
873	}
874	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
875
876cleanup:
877	return (error);
878}
879
880/*
881 * Update parameters of an IPv6 interface address.
882 * If necessary, a new entry is created and linked into address chains.
883 * This function is separated from in6_control().
884 */
885int
886in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
887    struct in6_ifaddr *ia, int flags)
888{
889	int error, hostIsNew = 0;
890
891	if ((error = in6_validate_ifra(ifp, ifra, ia, flags)) != 0)
892		return (error);
893
894	if (ia == NULL) {
895		hostIsNew = 1;
896		if ((ia = in6_alloc_ifa(ifp, ifra, flags)) == NULL)
897			return (ENOBUFS);
898	}
899
900	error = in6_update_ifa_internal(ifp, ifra, ia, hostIsNew, flags);
901	if (error != 0) {
902		if (hostIsNew != 0) {
903			in6_unlink_ifa(ia, ifp);
904			ifa_free(&ia->ia_ifa);
905		}
906		return (error);
907	}
908
909	if (hostIsNew)
910		error = in6_broadcast_ifa(ifp, ifra, ia, flags);
911
912	return (error);
913}
914
915/*
916 * Fill in basic IPv6 address request info.
917 */
918void
919in6_prepare_ifra(struct in6_aliasreq *ifra, const struct in6_addr *addr,
920    const struct in6_addr *mask)
921{
922
923	memset(ifra, 0, sizeof(struct in6_aliasreq));
924
925	ifra->ifra_addr.sin6_family = AF_INET6;
926	ifra->ifra_addr.sin6_len = sizeof(struct sockaddr_in6);
927	if (addr != NULL)
928		ifra->ifra_addr.sin6_addr = *addr;
929
930	ifra->ifra_prefixmask.sin6_family = AF_INET6;
931	ifra->ifra_prefixmask.sin6_len = sizeof(struct sockaddr_in6);
932	if (mask != NULL)
933		ifra->ifra_prefixmask.sin6_addr = *mask;
934}
935
936static int
937in6_validate_ifra(struct ifnet *ifp, struct in6_aliasreq *ifra,
938    struct in6_ifaddr *ia, int flags)
939{
940	int plen = -1;
941	struct sockaddr_in6 dst6;
942	struct in6_addrlifetime *lt;
943	char ip6buf[INET6_ADDRSTRLEN];
944
945	/* Validate parameters */
946	if (ifp == NULL || ifra == NULL) /* this maybe redundant */
947		return (EINVAL);
948
949	/*
950	 * The destination address for a p2p link must have a family
951	 * of AF_UNSPEC or AF_INET6.
952	 */
953	if ((ifp->if_flags & IFF_POINTOPOINT) != 0 &&
954	    ifra->ifra_dstaddr.sin6_family != AF_INET6 &&
955	    ifra->ifra_dstaddr.sin6_family != AF_UNSPEC)
956		return (EAFNOSUPPORT);
957
958	/*
959	 * Validate address
960	 */
961	if (ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6) ||
962	    ifra->ifra_addr.sin6_family != AF_INET6)
963		return (EINVAL);
964
965	/*
966	 * validate ifra_prefixmask.  don't check sin6_family, netmask
967	 * does not carry fields other than sin6_len.
968	 */
969	if (ifra->ifra_prefixmask.sin6_len > sizeof(struct sockaddr_in6))
970		return (EINVAL);
971	/*
972	 * Because the IPv6 address architecture is classless, we require
973	 * users to specify a (non 0) prefix length (mask) for a new address.
974	 * We also require the prefix (when specified) mask is valid, and thus
975	 * reject a non-consecutive mask.
976	 */
977	if (ia == NULL && ifra->ifra_prefixmask.sin6_len == 0)
978		return (EINVAL);
979	if (ifra->ifra_prefixmask.sin6_len != 0) {
980		plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr,
981		    (u_char *)&ifra->ifra_prefixmask +
982		    ifra->ifra_prefixmask.sin6_len);
983		if (plen <= 0)
984			return (EINVAL);
985	} else {
986		/*
987		 * In this case, ia must not be NULL.  We just use its prefix
988		 * length.
989		 */
990		plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL);
991	}
992	/*
993	 * If the destination address on a p2p interface is specified,
994	 * and the address is a scoped one, validate/set the scope
995	 * zone identifier.
996	 */
997	dst6 = ifra->ifra_dstaddr;
998	if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) != 0 &&
999	    (dst6.sin6_family == AF_INET6)) {
1000		struct in6_addr in6_tmp;
1001		u_int32_t zoneid;
1002
1003		in6_tmp = dst6.sin6_addr;
1004		if (in6_setscope(&in6_tmp, ifp, &zoneid))
1005			return (EINVAL); /* XXX: should be impossible */
1006
1007		if (dst6.sin6_scope_id != 0) {
1008			if (dst6.sin6_scope_id != zoneid)
1009				return (EINVAL);
1010		} else		/* user omit to specify the ID. */
1011			dst6.sin6_scope_id = zoneid;
1012
1013		/* convert into the internal form */
1014		if (sa6_embedscope(&dst6, 0))
1015			return (EINVAL); /* XXX: should be impossible */
1016	}
1017	/* Modify original ifra_dstaddr to reflect changes */
1018	ifra->ifra_dstaddr = dst6;
1019
1020	/*
1021	 * The destination address can be specified only for a p2p or a
1022	 * loopback interface.  If specified, the corresponding prefix length
1023	 * must be 128.
1024	 */
1025	if (ifra->ifra_dstaddr.sin6_family == AF_INET6) {
1026		if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) == 0) {
1027			/* XXX: noisy message */
1028			nd6log((LOG_INFO, "in6_update_ifa: a destination can "
1029			    "be specified for a p2p or a loopback IF only\n"));
1030			return (EINVAL);
1031		}
1032		if (plen != 128) {
1033			nd6log((LOG_INFO, "in6_update_ifa: prefixlen should "
1034			    "be 128 when dstaddr is specified\n"));
1035			return (EINVAL);
1036		}
1037	}
1038	/* lifetime consistency check */
1039	lt = &ifra->ifra_lifetime;
1040	if (lt->ia6t_pltime > lt->ia6t_vltime)
1041		return (EINVAL);
1042	if (lt->ia6t_vltime == 0) {
1043		/*
1044		 * the following log might be noisy, but this is a typical
1045		 * configuration mistake or a tool's bug.
1046		 */
1047		nd6log((LOG_INFO,
1048		    "in6_update_ifa: valid lifetime is 0 for %s\n",
1049		    ip6_sprintf(ip6buf, &ifra->ifra_addr.sin6_addr)));
1050
1051		if (ia == NULL)
1052			return (0); /* there's nothing to do */
1053	}
1054
1055	/* Check prefix mask */
1056	if (ia != NULL && ifra->ifra_prefixmask.sin6_len != 0) {
1057		/*
1058		 * We prohibit changing the prefix length of an existing
1059		 * address, because
1060		 * + such an operation should be rare in IPv6, and
1061		 * + the operation would confuse prefix management.
1062		 */
1063		if (ia->ia_prefixmask.sin6_len != 0 &&
1064		    in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL) != plen) {
1065			nd6log((LOG_INFO, "in6_validate_ifa: the prefix length "
1066			    "of an existing %s address should not be changed\n",
1067			    ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
1068
1069			return (EINVAL);
1070		}
1071	}
1072
1073	return (0);
1074}
1075
1076
1077/*
1078 * Allocate a new ifaddr and link it into chains.
1079 */
1080static struct in6_ifaddr *
1081in6_alloc_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, int flags)
1082{
1083	struct in6_ifaddr *ia;
1084
1085	/*
1086	 * When in6_alloc_ifa() is called in a process of a received
1087	 * RA, it is called under an interrupt context.  So, we should
1088	 * call malloc with M_NOWAIT.
1089	 */
1090	ia = (struct in6_ifaddr *)ifa_alloc(sizeof(*ia), M_NOWAIT);
1091	if (ia == NULL)
1092		return (NULL);
1093	LIST_INIT(&ia->ia6_memberships);
1094	/* Initialize the address and masks, and put time stamp */
1095	ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr;
1096	ia->ia_addr.sin6_family = AF_INET6;
1097	ia->ia_addr.sin6_len = sizeof(ia->ia_addr);
1098	/* XXX: Can we assign ,sin6_addr and skip the rest? */
1099	ia->ia_addr = ifra->ifra_addr;
1100	ia->ia6_createtime = time_uptime;
1101	if ((ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) != 0) {
1102		/*
1103		 * Some functions expect that ifa_dstaddr is not
1104		 * NULL for p2p interfaces.
1105		 */
1106		ia->ia_ifa.ifa_dstaddr =
1107		    (struct sockaddr *)&ia->ia_dstaddr;
1108	} else {
1109		ia->ia_ifa.ifa_dstaddr = NULL;
1110	}
1111
1112	/* set prefix mask if any */
1113	ia->ia_ifa.ifa_netmask = (struct sockaddr *)&ia->ia_prefixmask;
1114	if (ifra->ifra_prefixmask.sin6_len != 0) {
1115		ia->ia_prefixmask.sin6_family = AF_INET6;
1116		ia->ia_prefixmask.sin6_len = ifra->ifra_prefixmask.sin6_len;
1117		ia->ia_prefixmask.sin6_addr = ifra->ifra_prefixmask.sin6_addr;
1118	}
1119
1120	ia->ia_ifp = ifp;
1121	ifa_ref(&ia->ia_ifa);			/* if_addrhead */
1122	IF_ADDR_WLOCK(ifp);
1123	CK_STAILQ_INSERT_TAIL(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
1124	IF_ADDR_WUNLOCK(ifp);
1125
1126	ifa_ref(&ia->ia_ifa);			/* in6_ifaddrhead */
1127	IN6_IFADDR_WLOCK();
1128	CK_STAILQ_INSERT_TAIL(&V_in6_ifaddrhead, ia, ia_link);
1129	CK_LIST_INSERT_HEAD(IN6ADDR_HASH(&ia->ia_addr.sin6_addr), ia, ia6_hash);
1130	IN6_IFADDR_WUNLOCK();
1131
1132	return (ia);
1133}
1134
1135/*
1136 * Update/configure interface address parameters:
1137 *
1138 * 1) Update lifetime
1139 * 2) Update interface metric ad flags
1140 * 3) Notify other subsystems
1141 */
1142static int
1143in6_update_ifa_internal(struct ifnet *ifp, struct in6_aliasreq *ifra,
1144    struct in6_ifaddr *ia, int hostIsNew, int flags)
1145{
1146	int error;
1147
1148	/* update timestamp */
1149	ia->ia6_updatetime = time_uptime;
1150
1151	/*
1152	 * Set lifetimes.  We do not refer to ia6t_expire and ia6t_preferred
1153	 * to see if the address is deprecated or invalidated, but initialize
1154	 * these members for applications.
1155	 */
1156	ia->ia6_lifetime = ifra->ifra_lifetime;
1157	if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
1158		ia->ia6_lifetime.ia6t_expire =
1159		    time_uptime + ia->ia6_lifetime.ia6t_vltime;
1160	} else
1161		ia->ia6_lifetime.ia6t_expire = 0;
1162	if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
1163		ia->ia6_lifetime.ia6t_preferred =
1164		    time_uptime + ia->ia6_lifetime.ia6t_pltime;
1165	} else
1166		ia->ia6_lifetime.ia6t_preferred = 0;
1167
1168	/*
1169	 * backward compatibility - if IN6_IFF_DEPRECATED is set from the
1170	 * userland, make it deprecated.
1171	 */
1172	if ((ifra->ifra_flags & IN6_IFF_DEPRECATED) != 0) {
1173		ia->ia6_lifetime.ia6t_pltime = 0;
1174		ia->ia6_lifetime.ia6t_preferred = time_uptime;
1175	}
1176
1177	/*
1178	 * configure address flags.
1179	 */
1180	ia->ia6_flags = ifra->ifra_flags;
1181
1182	/*
1183	 * Make the address tentative before joining multicast addresses,
1184	 * so that corresponding MLD responses would not have a tentative
1185	 * source address.
1186	 */
1187	ia->ia6_flags &= ~IN6_IFF_DUPLICATED;	/* safety */
1188
1189	/*
1190	 * DAD should be performed for an new address or addresses on
1191	 * an interface with ND6_IFF_IFDISABLED.
1192	 */
1193	if (in6if_do_dad(ifp) &&
1194	    (hostIsNew || (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)))
1195		ia->ia6_flags |= IN6_IFF_TENTATIVE;
1196
1197	/* notify other subsystems */
1198	error = in6_notify_ifa(ifp, ia, ifra, hostIsNew);
1199
1200	return (error);
1201}
1202
1203/*
1204 * Do link-level ifa job:
1205 * 1) Add lle entry for added address
1206 * 2) Notifies routing socket users about new address
1207 * 3) join appropriate multicast group
1208 * 4) start DAD if enabled
1209 */
1210static int
1211in6_broadcast_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
1212    struct in6_ifaddr *ia, int flags)
1213{
1214	struct in6_multi *in6m_sol;
1215	int error = 0;
1216
1217	/* Add local address to lltable, if necessary (ex. on p2p link). */
1218	if ((error = nd6_add_ifa_lle(ia)) != 0) {
1219		in6_purgeaddr(&ia->ia_ifa);
1220		ifa_free(&ia->ia_ifa);
1221		return (error);
1222	}
1223
1224	/* Join necessary multicast groups. */
1225	in6m_sol = NULL;
1226	if ((ifp->if_flags & IFF_MULTICAST) != 0) {
1227		error = in6_update_ifa_join_mc(ifp, ifra, ia, flags, &in6m_sol);
1228		if (error != 0) {
1229			in6_purgeaddr(&ia->ia_ifa);
1230			ifa_free(&ia->ia_ifa);
1231			return (error);
1232		}
1233	}
1234
1235	/* Perform DAD, if the address is TENTATIVE. */
1236	if ((ia->ia6_flags & IN6_IFF_TENTATIVE)) {
1237		int delay, mindelay, maxdelay;
1238
1239		delay = 0;
1240		if ((flags & IN6_IFAUPDATE_DADDELAY)) {
1241			/*
1242			 * We need to impose a delay before sending an NS
1243			 * for DAD.  Check if we also needed a delay for the
1244			 * corresponding MLD message.  If we did, the delay
1245			 * should be larger than the MLD delay (this could be
1246			 * relaxed a bit, but this simple logic is at least
1247			 * safe).
1248			 * XXX: Break data hiding guidelines and look at
1249			 * state for the solicited multicast group.
1250			 */
1251			mindelay = 0;
1252			if (in6m_sol != NULL &&
1253			    in6m_sol->in6m_state == MLD_REPORTING_MEMBER) {
1254				mindelay = in6m_sol->in6m_timer;
1255			}
1256			maxdelay = MAX_RTR_SOLICITATION_DELAY * hz;
1257			if (maxdelay - mindelay == 0)
1258				delay = 0;
1259			else {
1260				delay =
1261				    (arc4random() % (maxdelay - mindelay)) +
1262				    mindelay;
1263			}
1264		}
1265		nd6_dad_start((struct ifaddr *)ia, delay);
1266	}
1267
1268	in6_newaddrmsg(ia, RTM_ADD);
1269	ifa_free(&ia->ia_ifa);
1270	return (error);
1271}
1272
1273void
1274in6_purgeaddr(struct ifaddr *ifa)
1275{
1276	struct ifnet *ifp = ifa->ifa_ifp;
1277	struct in6_ifaddr *ia = (struct in6_ifaddr *) ifa;
1278	struct in6_multi_mship *imm;
1279	int plen, error;
1280
1281	if (ifa->ifa_carp)
1282		(*carp_detach_p)(ifa, false);
1283
1284	/*
1285	 * Remove the loopback route to the interface address.
1286	 * The check for the current setting of "nd6_useloopback"
1287	 * is not needed.
1288	 */
1289	if (ia->ia_flags & IFA_RTSELF) {
1290		error = ifa_del_loopback_route((struct ifaddr *)ia,
1291		    (struct sockaddr *)&ia->ia_addr);
1292		if (error == 0)
1293			ia->ia_flags &= ~IFA_RTSELF;
1294	}
1295
1296	/* stop DAD processing */
1297	nd6_dad_stop(ifa);
1298
1299	/* Leave multicast groups. */
1300	while ((imm = LIST_FIRST(&ia->ia6_memberships)) != NULL) {
1301		LIST_REMOVE(imm, i6mm_chain);
1302		if (imm->i6mm_maddr != NULL)
1303			in6_leavegroup(imm->i6mm_maddr, NULL);
1304		free(imm, M_IP6MADDR);
1305	}
1306	plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */
1307	if ((ia->ia_flags & IFA_ROUTE) && plen == 128) {
1308		error = rtinit(&(ia->ia_ifa), RTM_DELETE, ia->ia_flags |
1309		    (ia->ia_dstaddr.sin6_family == AF_INET6 ? RTF_HOST : 0));
1310		if (error != 0)
1311			log(LOG_INFO, "%s: err=%d, destination address delete "
1312			    "failed\n", __func__, error);
1313		ia->ia_flags &= ~IFA_ROUTE;
1314	}
1315
1316	in6_newaddrmsg(ia, RTM_DELETE);
1317	in6_unlink_ifa(ia, ifp);
1318}
1319
1320static void
1321in6_unlink_ifa(struct in6_ifaddr *ia, struct ifnet *ifp)
1322{
1323	char ip6buf[INET6_ADDRSTRLEN];
1324	int remove_lle;
1325
1326	IF_ADDR_WLOCK(ifp);
1327	CK_STAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifaddr, ifa_link);
1328	IF_ADDR_WUNLOCK(ifp);
1329	ifa_free(&ia->ia_ifa);			/* if_addrhead */
1330
1331	/*
1332	 * Defer the release of what might be the last reference to the
1333	 * in6_ifaddr so that it can't be freed before the remainder of the
1334	 * cleanup.
1335	 */
1336	IN6_IFADDR_WLOCK();
1337	CK_STAILQ_REMOVE(&V_in6_ifaddrhead, ia, in6_ifaddr, ia_link);
1338	CK_LIST_REMOVE(ia, ia6_hash);
1339	IN6_IFADDR_WUNLOCK();
1340
1341	/*
1342	 * Release the reference to the base prefix.  There should be a
1343	 * positive reference.
1344	 */
1345	remove_lle = 0;
1346	if (ia->ia6_ndpr == NULL) {
1347		nd6log((LOG_NOTICE,
1348		    "in6_unlink_ifa: autoconf'ed address "
1349		    "%s has no prefix\n", ip6_sprintf(ip6buf, IA6_IN6(ia))));
1350	} else {
1351		ia->ia6_ndpr->ndpr_addrcnt--;
1352		/* Do not delete lles within prefix if refcont != 0 */
1353		if (ia->ia6_ndpr->ndpr_addrcnt == 0)
1354			remove_lle = 1;
1355		ia->ia6_ndpr = NULL;
1356	}
1357
1358	nd6_rem_ifa_lle(ia, remove_lle);
1359
1360	/*
1361	 * Also, if the address being removed is autoconf'ed, call
1362	 * pfxlist_onlink_check() since the release might affect the status of
1363	 * other (detached) addresses.
1364	 */
1365	if ((ia->ia6_flags & IN6_IFF_AUTOCONF)) {
1366		pfxlist_onlink_check();
1367	}
1368	ifa_free(&ia->ia_ifa);			/* in6_ifaddrhead */
1369}
1370
1371/*
1372 * Notifies other subsystems about address change/arrival:
1373 * 1) Notifies device handler on the first IPv6 address assignment
1374 * 2) Handle routing table changes for P2P links and route
1375 * 3) Handle routing table changes for address host route
1376 */
1377static int
1378in6_notify_ifa(struct ifnet *ifp, struct in6_ifaddr *ia,
1379    struct in6_aliasreq *ifra, int hostIsNew)
1380{
1381	int	error = 0, plen, ifacount = 0;
1382	struct ifaddr *ifa;
1383	struct sockaddr_in6 *pdst;
1384	char ip6buf[INET6_ADDRSTRLEN];
1385
1386	/*
1387	 * Give the interface a chance to initialize
1388	 * if this is its first address,
1389	 */
1390	if (hostIsNew != 0) {
1391		struct epoch_tracker et;
1392
1393		NET_EPOCH_ENTER(et);
1394		CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1395			if (ifa->ifa_addr->sa_family != AF_INET6)
1396				continue;
1397			ifacount++;
1398		}
1399		NET_EPOCH_EXIT(et);
1400	}
1401
1402	if (ifacount <= 1 && ifp->if_ioctl) {
1403		error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
1404		if (error)
1405			goto done;
1406	}
1407
1408	/*
1409	 * If a new destination address is specified, scrub the old one and
1410	 * install the new destination.  Note that the interface must be
1411	 * p2p or loopback.
1412	 */
1413	pdst = &ifra->ifra_dstaddr;
1414	if (pdst->sin6_family == AF_INET6 &&
1415	    !IN6_ARE_ADDR_EQUAL(&pdst->sin6_addr, &ia->ia_dstaddr.sin6_addr)) {
1416		if ((ia->ia_flags & IFA_ROUTE) != 0 &&
1417		    (rtinit(&(ia->ia_ifa), (int)RTM_DELETE, RTF_HOST) != 0)) {
1418			nd6log((LOG_ERR, "in6_update_ifa_internal: failed to "
1419			    "remove a route to the old destination: %s\n",
1420			    ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
1421			/* proceed anyway... */
1422		} else
1423			ia->ia_flags &= ~IFA_ROUTE;
1424		ia->ia_dstaddr = *pdst;
1425	}
1426
1427	/*
1428	 * If a new destination address is specified for a point-to-point
1429	 * interface, install a route to the destination as an interface
1430	 * direct route.
1431	 * XXX: the logic below rejects assigning multiple addresses on a p2p
1432	 * interface that share the same destination.
1433	 */
1434	plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */
1435	if (!(ia->ia_flags & IFA_ROUTE) && plen == 128 &&
1436	    ia->ia_dstaddr.sin6_family == AF_INET6) {
1437		int rtflags = RTF_UP | RTF_HOST;
1438		/*
1439		 * Handle the case for ::1 .
1440		 */
1441		if (ifp->if_flags & IFF_LOOPBACK)
1442			ia->ia_flags |= IFA_RTSELF;
1443		error = rtinit(&ia->ia_ifa, RTM_ADD, ia->ia_flags | rtflags);
1444		if (error)
1445			goto done;
1446		ia->ia_flags |= IFA_ROUTE;
1447	}
1448
1449	/*
1450	 * add a loopback route to self if not exists
1451	 */
1452	if (!(ia->ia_flags & IFA_RTSELF) && V_nd6_useloopback) {
1453		error = ifa_add_loopback_route((struct ifaddr *)ia,
1454		    (struct sockaddr *)&ia->ia_addr);
1455		if (error == 0)
1456			ia->ia_flags |= IFA_RTSELF;
1457	}
1458done:
1459	WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
1460	    "Invoking IPv6 network device address event may sleep");
1461
1462	ifa_ref(&ia->ia_ifa);
1463	EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa,
1464	    IFADDR_EVENT_ADD);
1465	ifa_free(&ia->ia_ifa);
1466
1467	return (error);
1468}
1469
1470/*
1471 * Find an IPv6 interface link-local address specific to an interface.
1472 * ifaddr is returned referenced.
1473 */
1474struct in6_ifaddr *
1475in6ifa_ifpforlinklocal(struct ifnet *ifp, int ignoreflags)
1476{
1477	struct ifaddr *ifa;
1478
1479	NET_EPOCH_ASSERT();
1480
1481	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1482		if (ifa->ifa_addr->sa_family != AF_INET6)
1483			continue;
1484		if (IN6_IS_ADDR_LINKLOCAL(IFA_IN6(ifa))) {
1485			if ((((struct in6_ifaddr *)ifa)->ia6_flags &
1486			    ignoreflags) != 0)
1487				continue;
1488			ifa_ref(ifa);
1489			break;
1490		}
1491	}
1492
1493	return ((struct in6_ifaddr *)ifa);
1494}
1495
1496
1497/*
1498 * find the interface address corresponding to a given IPv6 address.
1499 * ifaddr is returned referenced.
1500 */
1501struct in6_ifaddr *
1502in6ifa_ifwithaddr(const struct in6_addr *addr, uint32_t zoneid)
1503{
1504	struct rm_priotracker in6_ifa_tracker;
1505	struct in6_ifaddr *ia;
1506
1507	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1508	CK_LIST_FOREACH(ia, IN6ADDR_HASH(addr), ia6_hash) {
1509		if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), addr)) {
1510			if (zoneid != 0 &&
1511			    zoneid != ia->ia_addr.sin6_scope_id)
1512				continue;
1513			ifa_ref(&ia->ia_ifa);
1514			break;
1515		}
1516	}
1517	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1518	return (ia);
1519}
1520
1521/*
1522 * find the internet address corresponding to a given interface and address.
1523 * ifaddr is returned referenced.
1524 */
1525struct in6_ifaddr *
1526in6ifa_ifpwithaddr(struct ifnet *ifp, const struct in6_addr *addr)
1527{
1528	struct epoch_tracker et;
1529	struct ifaddr *ifa;
1530
1531	NET_EPOCH_ENTER(et);
1532	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1533		if (ifa->ifa_addr->sa_family != AF_INET6)
1534			continue;
1535		if (IN6_ARE_ADDR_EQUAL(addr, IFA_IN6(ifa))) {
1536			ifa_ref(ifa);
1537			break;
1538		}
1539	}
1540	NET_EPOCH_EXIT(et);
1541
1542	return ((struct in6_ifaddr *)ifa);
1543}
1544
1545/*
1546 * Find a link-local scoped address on ifp and return it if any.
1547 */
1548struct in6_ifaddr *
1549in6ifa_llaonifp(struct ifnet *ifp)
1550{
1551	struct sockaddr_in6 *sin6;
1552	struct ifaddr *ifa;
1553
1554	NET_EPOCH_ASSERT();
1555
1556	if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)
1557		return (NULL);
1558	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1559		if (ifa->ifa_addr->sa_family != AF_INET6)
1560			continue;
1561		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
1562		if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr) ||
1563		    IN6_IS_ADDR_MC_INTFACELOCAL(&sin6->sin6_addr) ||
1564		    IN6_IS_ADDR_MC_NODELOCAL(&sin6->sin6_addr))
1565			break;
1566	}
1567
1568	return ((struct in6_ifaddr *)ifa);
1569}
1570
1571/*
1572 * Convert IP6 address to printable (loggable) representation. Caller
1573 * has to make sure that ip6buf is at least INET6_ADDRSTRLEN long.
1574 */
1575static char digits[] = "0123456789abcdef";
1576char *
1577ip6_sprintf(char *ip6buf, const struct in6_addr *addr)
1578{
1579	int i, cnt = 0, maxcnt = 0, idx = 0, index = 0;
1580	char *cp;
1581	const u_int16_t *a = (const u_int16_t *)addr;
1582	const u_int8_t *d;
1583	int dcolon = 0, zero = 0;
1584
1585	cp = ip6buf;
1586
1587	for (i = 0; i < 8; i++) {
1588		if (*(a + i) == 0) {
1589			cnt++;
1590			if (cnt == 1)
1591				idx = i;
1592		}
1593		else if (maxcnt < cnt) {
1594			maxcnt = cnt;
1595			index = idx;
1596			cnt = 0;
1597		}
1598	}
1599	if (maxcnt < cnt) {
1600		maxcnt = cnt;
1601		index = idx;
1602	}
1603
1604	for (i = 0; i < 8; i++) {
1605		if (dcolon == 1) {
1606			if (*a == 0) {
1607				if (i == 7)
1608					*cp++ = ':';
1609				a++;
1610				continue;
1611			} else
1612				dcolon = 2;
1613		}
1614		if (*a == 0) {
1615			if (dcolon == 0 && *(a + 1) == 0 && i == index) {
1616				if (i == 0)
1617					*cp++ = ':';
1618				*cp++ = ':';
1619				dcolon = 1;
1620			} else {
1621				*cp++ = '0';
1622				*cp++ = ':';
1623			}
1624			a++;
1625			continue;
1626		}
1627		d = (const u_char *)a;
1628		/* Try to eliminate leading zeros in printout like in :0001. */
1629		zero = 1;
1630		*cp = digits[*d >> 4];
1631		if (*cp != '0') {
1632			zero = 0;
1633			cp++;
1634		}
1635		*cp = digits[*d++ & 0xf];
1636		if (zero == 0 || (*cp != '0')) {
1637			zero = 0;
1638			cp++;
1639		}
1640		*cp = digits[*d >> 4];
1641		if (zero == 0 || (*cp != '0')) {
1642			zero = 0;
1643			cp++;
1644		}
1645		*cp++ = digits[*d & 0xf];
1646		*cp++ = ':';
1647		a++;
1648	}
1649	*--cp = '\0';
1650	return (ip6buf);
1651}
1652
1653int
1654in6_localaddr(struct in6_addr *in6)
1655{
1656	struct rm_priotracker in6_ifa_tracker;
1657	struct in6_ifaddr *ia;
1658
1659	if (IN6_IS_ADDR_LOOPBACK(in6) || IN6_IS_ADDR_LINKLOCAL(in6))
1660		return 1;
1661
1662	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1663	CK_STAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) {
1664		if (IN6_ARE_MASKED_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr,
1665		    &ia->ia_prefixmask.sin6_addr)) {
1666			IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1667			return 1;
1668		}
1669	}
1670	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1671
1672	return (0);
1673}
1674
1675/*
1676 * Return 1 if an internet address is for the local host and configured
1677 * on one of its interfaces.
1678 */
1679int
1680in6_localip(struct in6_addr *in6)
1681{
1682	struct rm_priotracker in6_ifa_tracker;
1683	struct in6_ifaddr *ia;
1684
1685	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1686	CK_LIST_FOREACH(ia, IN6ADDR_HASH(in6), ia6_hash) {
1687		if (IN6_ARE_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr)) {
1688			IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1689			return (1);
1690		}
1691	}
1692	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1693	return (0);
1694}
1695
1696/*
1697 * Return 1 if an internet address is configured on an interface.
1698 */
1699int
1700in6_ifhasaddr(struct ifnet *ifp, struct in6_addr *addr)
1701{
1702	struct in6_addr in6;
1703	struct ifaddr *ifa;
1704	struct in6_ifaddr *ia6;
1705
1706	NET_EPOCH_ASSERT();
1707
1708	in6 = *addr;
1709	if (in6_clearscope(&in6))
1710		return (0);
1711	in6_setscope(&in6, ifp, NULL);
1712
1713	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1714		if (ifa->ifa_addr->sa_family != AF_INET6)
1715			continue;
1716		ia6 = (struct in6_ifaddr *)ifa;
1717		if (IN6_ARE_ADDR_EQUAL(&ia6->ia_addr.sin6_addr, &in6))
1718			return (1);
1719	}
1720
1721	return (0);
1722}
1723
1724int
1725in6_is_addr_deprecated(struct sockaddr_in6 *sa6)
1726{
1727	struct rm_priotracker in6_ifa_tracker;
1728	struct in6_ifaddr *ia;
1729
1730	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1731	CK_LIST_FOREACH(ia, IN6ADDR_HASH(&sa6->sin6_addr), ia6_hash) {
1732		if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), &sa6->sin6_addr)) {
1733			if (ia->ia6_flags & IN6_IFF_DEPRECATED) {
1734				IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1735				return (1); /* true */
1736			}
1737			break;
1738		}
1739	}
1740	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1741
1742	return (0);		/* false */
1743}
1744
1745/*
1746 * return length of part which dst and src are equal
1747 * hard coding...
1748 */
1749int
1750in6_matchlen(struct in6_addr *src, struct in6_addr *dst)
1751{
1752	int match = 0;
1753	u_char *s = (u_char *)src, *d = (u_char *)dst;
1754	u_char *lim = s + 16, r;
1755
1756	while (s < lim)
1757		if ((r = (*d++ ^ *s++)) != 0) {
1758			while (r < 128) {
1759				match++;
1760				r <<= 1;
1761			}
1762			break;
1763		} else
1764			match += 8;
1765	return match;
1766}
1767
1768/* XXX: to be scope conscious */
1769int
1770in6_are_prefix_equal(struct in6_addr *p1, struct in6_addr *p2, int len)
1771{
1772	int bytelen, bitlen;
1773
1774	/* sanity check */
1775	if (0 > len || len > 128) {
1776		log(LOG_ERR, "in6_are_prefix_equal: invalid prefix length(%d)\n",
1777		    len);
1778		return (0);
1779	}
1780
1781	bytelen = len / 8;
1782	bitlen = len % 8;
1783
1784	if (bcmp(&p1->s6_addr, &p2->s6_addr, bytelen))
1785		return (0);
1786	if (bitlen != 0 &&
1787	    p1->s6_addr[bytelen] >> (8 - bitlen) !=
1788	    p2->s6_addr[bytelen] >> (8 - bitlen))
1789		return (0);
1790
1791	return (1);
1792}
1793
1794void
1795in6_prefixlen2mask(struct in6_addr *maskp, int len)
1796{
1797	u_char maskarray[8] = {0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, 0xff};
1798	int bytelen, bitlen, i;
1799
1800	/* sanity check */
1801	if (0 > len || len > 128) {
1802		log(LOG_ERR, "in6_prefixlen2mask: invalid prefix length(%d)\n",
1803		    len);
1804		return;
1805	}
1806
1807	bzero(maskp, sizeof(*maskp));
1808	bytelen = len / 8;
1809	bitlen = len % 8;
1810	for (i = 0; i < bytelen; i++)
1811		maskp->s6_addr[i] = 0xff;
1812	if (bitlen)
1813		maskp->s6_addr[bytelen] = maskarray[bitlen - 1];
1814}
1815
1816/*
1817 * return the best address out of the same scope. if no address was
1818 * found, return the first valid address from designated IF.
1819 */
1820struct in6_ifaddr *
1821in6_ifawithifp(struct ifnet *ifp, struct in6_addr *dst)
1822{
1823	int dst_scope =	in6_addrscope(dst), blen = -1, tlen;
1824	struct ifaddr *ifa;
1825	struct in6_ifaddr *besta = NULL;
1826	struct in6_ifaddr *dep[2];	/* last-resort: deprecated */
1827
1828	NET_EPOCH_ASSERT();
1829
1830	dep[0] = dep[1] = NULL;
1831
1832	/*
1833	 * We first look for addresses in the same scope.
1834	 * If there is one, return it.
1835	 * If two or more, return one which matches the dst longest.
1836	 * If none, return one of global addresses assigned other ifs.
1837	 */
1838	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1839		if (ifa->ifa_addr->sa_family != AF_INET6)
1840			continue;
1841		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
1842			continue; /* XXX: is there any case to allow anycast? */
1843		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
1844			continue; /* don't use this interface */
1845		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
1846			continue;
1847		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
1848			if (V_ip6_use_deprecated)
1849				dep[0] = (struct in6_ifaddr *)ifa;
1850			continue;
1851		}
1852
1853		if (dst_scope == in6_addrscope(IFA_IN6(ifa))) {
1854			/*
1855			 * call in6_matchlen() as few as possible
1856			 */
1857			if (besta) {
1858				if (blen == -1)
1859					blen = in6_matchlen(&besta->ia_addr.sin6_addr, dst);
1860				tlen = in6_matchlen(IFA_IN6(ifa), dst);
1861				if (tlen > blen) {
1862					blen = tlen;
1863					besta = (struct in6_ifaddr *)ifa;
1864				}
1865			} else
1866				besta = (struct in6_ifaddr *)ifa;
1867		}
1868	}
1869	if (besta) {
1870		ifa_ref(&besta->ia_ifa);
1871		return (besta);
1872	}
1873
1874	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1875		if (ifa->ifa_addr->sa_family != AF_INET6)
1876			continue;
1877		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
1878			continue; /* XXX: is there any case to allow anycast? */
1879		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
1880			continue; /* don't use this interface */
1881		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
1882			continue;
1883		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
1884			if (V_ip6_use_deprecated)
1885				dep[1] = (struct in6_ifaddr *)ifa;
1886			continue;
1887		}
1888
1889		if (ifa != NULL)
1890			ifa_ref(ifa);
1891		return (struct in6_ifaddr *)ifa;
1892	}
1893
1894	/* use the last-resort values, that are, deprecated addresses */
1895	if (dep[0]) {
1896		ifa_ref((struct ifaddr *)dep[0]);
1897		return dep[0];
1898	}
1899	if (dep[1]) {
1900		ifa_ref((struct ifaddr *)dep[1]);
1901		return dep[1];
1902	}
1903
1904	return NULL;
1905}
1906
1907/*
1908 * perform DAD when interface becomes IFF_UP.
1909 */
1910void
1911in6_if_up(struct ifnet *ifp)
1912{
1913	struct epoch_tracker et;
1914	struct ifaddr *ifa;
1915	struct in6_ifaddr *ia;
1916
1917	NET_EPOCH_ENTER(et);
1918	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1919		if (ifa->ifa_addr->sa_family != AF_INET6)
1920			continue;
1921		ia = (struct in6_ifaddr *)ifa;
1922		if (ia->ia6_flags & IN6_IFF_TENTATIVE) {
1923			/*
1924			 * The TENTATIVE flag was likely set by hand
1925			 * beforehand, implicitly indicating the need for DAD.
1926			 * We may be able to skip the random delay in this
1927			 * case, but we impose delays just in case.
1928			 */
1929			nd6_dad_start(ifa,
1930			    arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz));
1931		}
1932	}
1933
1934	/*
1935	 * special cases, like 6to4, are handled in in6_ifattach
1936	 */
1937	in6_ifattach(ifp, NULL);
1938	NET_EPOCH_EXIT(et);
1939}
1940
1941int
1942in6if_do_dad(struct ifnet *ifp)
1943{
1944
1945	if ((ifp->if_flags & IFF_LOOPBACK) != 0)
1946		return (0);
1947	if ((ifp->if_flags & IFF_MULTICAST) == 0)
1948		return (0);
1949	if ((ND_IFINFO(ifp)->flags &
1950	    (ND6_IFF_IFDISABLED | ND6_IFF_NO_DAD)) != 0)
1951		return (0);
1952	return (1);
1953}
1954
1955/*
1956 * Calculate max IPv6 MTU through all the interfaces and store it
1957 * to in6_maxmtu.
1958 */
1959void
1960in6_setmaxmtu(void)
1961{
1962	struct epoch_tracker et;
1963	unsigned long maxmtu = 0;
1964	struct ifnet *ifp;
1965
1966	NET_EPOCH_ENTER(et);
1967	CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
1968		/* this function can be called during ifnet initialization */
1969		if (!ifp->if_afdata[AF_INET6])
1970			continue;
1971		if ((ifp->if_flags & IFF_LOOPBACK) == 0 &&
1972		    IN6_LINKMTU(ifp) > maxmtu)
1973			maxmtu = IN6_LINKMTU(ifp);
1974	}
1975	NET_EPOCH_EXIT(et);
1976	if (maxmtu)	/* update only when maxmtu is positive */
1977		V_in6_maxmtu = maxmtu;
1978}
1979
1980/*
1981 * Provide the length of interface identifiers to be used for the link attached
1982 * to the given interface.  The length should be defined in "IPv6 over
1983 * xxx-link" document.  Note that address architecture might also define
1984 * the length for a particular set of address prefixes, regardless of the
1985 * link type.  As clarified in rfc2462bis, those two definitions should be
1986 * consistent, and those really are as of August 2004.
1987 */
1988int
1989in6_if2idlen(struct ifnet *ifp)
1990{
1991	switch (ifp->if_type) {
1992	case IFT_ETHER:		/* RFC2464 */
1993	case IFT_PROPVIRTUAL:	/* XXX: no RFC. treat it as ether */
1994	case IFT_L2VLAN:	/* ditto */
1995	case IFT_BRIDGE:	/* bridge(4) only does Ethernet-like links */
1996	case IFT_INFINIBAND:
1997		return (64);
1998	case IFT_PPP:		/* RFC2472 */
1999		return (64);
2000	case IFT_FRELAY:	/* RFC2590 */
2001		return (64);
2002	case IFT_IEEE1394:	/* RFC3146 */
2003		return (64);
2004	case IFT_GIF:
2005		return (64);	/* draft-ietf-v6ops-mech-v2-07 */
2006	case IFT_LOOP:
2007		return (64);	/* XXX: is this really correct? */
2008	default:
2009		/*
2010		 * Unknown link type:
2011		 * It might be controversial to use the today's common constant
2012		 * of 64 for these cases unconditionally.  For full compliance,
2013		 * we should return an error in this case.  On the other hand,
2014		 * if we simply miss the standard for the link type or a new
2015		 * standard is defined for a new link type, the IFID length
2016		 * is very likely to be the common constant.  As a compromise,
2017		 * we always use the constant, but make an explicit notice
2018		 * indicating the "unknown" case.
2019		 */
2020		printf("in6_if2idlen: unknown link type (%d)\n", ifp->if_type);
2021		return (64);
2022	}
2023}
2024
2025#include <sys/sysctl.h>
2026
2027struct in6_llentry {
2028	struct llentry		base;
2029};
2030
2031#define	IN6_LLTBL_DEFAULT_HSIZE	32
2032#define	IN6_LLTBL_HASH(k, h) \
2033	(((((((k >> 8) ^ k) >> 8) ^ k) >> 8) ^ k) & ((h) - 1))
2034
2035/*
2036 * Do actual deallocation of @lle.
2037 */
2038static void
2039in6_lltable_destroy_lle_unlocked(epoch_context_t ctx)
2040{
2041	struct llentry *lle;
2042
2043	lle = __containerof(ctx, struct llentry, lle_epoch_ctx);
2044	LLE_LOCK_DESTROY(lle);
2045	LLE_REQ_DESTROY(lle);
2046	free(lle, M_LLTABLE);
2047}
2048
2049/*
2050 * Called by LLE_FREE_LOCKED when number of references
2051 * drops to zero.
2052 */
2053static void
2054in6_lltable_destroy_lle(struct llentry *lle)
2055{
2056
2057	LLE_WUNLOCK(lle);
2058	epoch_call(net_epoch_preempt,  &lle->lle_epoch_ctx, in6_lltable_destroy_lle_unlocked);
2059}
2060
2061static struct llentry *
2062in6_lltable_new(const struct in6_addr *addr6, u_int flags)
2063{
2064	struct in6_llentry *lle;
2065
2066	lle = malloc(sizeof(struct in6_llentry), M_LLTABLE, M_NOWAIT | M_ZERO);
2067	if (lle == NULL)		/* NB: caller generates msg */
2068		return NULL;
2069
2070	lle->base.r_l3addr.addr6 = *addr6;
2071	lle->base.lle_refcnt = 1;
2072	lle->base.lle_free = in6_lltable_destroy_lle;
2073	LLE_LOCK_INIT(&lle->base);
2074	LLE_REQ_INIT(&lle->base);
2075	callout_init(&lle->base.lle_timer, 1);
2076
2077	return (&lle->base);
2078}
2079
2080static int
2081in6_lltable_match_prefix(const struct sockaddr *saddr,
2082    const struct sockaddr *smask, u_int flags, struct llentry *lle)
2083{
2084	const struct in6_addr *addr, *mask, *lle_addr;
2085
2086	addr = &((const struct sockaddr_in6 *)saddr)->sin6_addr;
2087	mask = &((const struct sockaddr_in6 *)smask)->sin6_addr;
2088	lle_addr = &lle->r_l3addr.addr6;
2089
2090	if (IN6_ARE_MASKED_ADDR_EQUAL(lle_addr, addr, mask) == 0)
2091		return (0);
2092
2093	if (lle->la_flags & LLE_IFADDR) {
2094
2095		/*
2096		 * Delete LLE_IFADDR records IFF address & flag matches.
2097		 * Note that addr is the interface address within prefix
2098		 * being matched.
2099		 */
2100		if (IN6_ARE_ADDR_EQUAL(addr, lle_addr) &&
2101		    (flags & LLE_STATIC) != 0)
2102			return (1);
2103		return (0);
2104	}
2105
2106	/* flags & LLE_STATIC means deleting both dynamic and static entries */
2107	if ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))
2108		return (1);
2109
2110	return (0);
2111}
2112
2113static void
2114in6_lltable_free_entry(struct lltable *llt, struct llentry *lle)
2115{
2116	struct ifnet *ifp;
2117
2118	LLE_WLOCK_ASSERT(lle);
2119	KASSERT(llt != NULL, ("lltable is NULL"));
2120
2121	/* Unlink entry from table */
2122	if ((lle->la_flags & LLE_LINKED) != 0) {
2123
2124		ifp = llt->llt_ifp;
2125		IF_AFDATA_WLOCK_ASSERT(ifp);
2126		lltable_unlink_entry(llt, lle);
2127	}
2128
2129	llentry_free(lle);
2130}
2131
2132static int
2133in6_lltable_rtcheck(struct ifnet *ifp,
2134		    u_int flags,
2135		    const struct sockaddr *l3addr)
2136{
2137	const struct sockaddr_in6 *sin6;
2138	struct nhop6_basic nh6;
2139	struct in6_addr dst;
2140	uint32_t scopeid;
2141	int error;
2142	char ip6buf[INET6_ADDRSTRLEN];
2143	int fibnum;
2144
2145	NET_EPOCH_ASSERT();
2146	KASSERT(l3addr->sa_family == AF_INET6,
2147	    ("sin_family %d", l3addr->sa_family));
2148
2149	sin6 = (const struct sockaddr_in6 *)l3addr;
2150	in6_splitscope(&sin6->sin6_addr, &dst, &scopeid);
2151	fibnum = V_rt_add_addr_allfibs ? RT_DEFAULT_FIB : ifp->if_fib;
2152	error = fib6_lookup_nh_basic(fibnum, &dst, scopeid, 0, 0, &nh6);
2153	if (error != 0 || (nh6.nh_flags & NHF_GATEWAY) || nh6.nh_ifp != ifp) {
2154		struct ifaddr *ifa;
2155		/*
2156		 * Create an ND6 cache for an IPv6 neighbor
2157		 * that is not covered by our own prefix.
2158		 */
2159		ifa = ifaof_ifpforaddr(l3addr, ifp);
2160		if (ifa != NULL) {
2161			return 0;
2162		}
2163		log(LOG_INFO, "IPv6 address: \"%s\" is not on the network\n",
2164		    ip6_sprintf(ip6buf, &sin6->sin6_addr));
2165		return EINVAL;
2166	}
2167	return 0;
2168}
2169
2170/*
2171 * Called by the datapath to indicate that the entry was used.
2172 */
2173static void
2174in6_lltable_mark_used(struct llentry *lle)
2175{
2176
2177	LLE_REQ_LOCK(lle);
2178	lle->r_skip_req = 0;
2179
2180	/*
2181	 * Set the hit time so the callback function
2182	 * can determine the remaining time before
2183	 * transiting to the DELAY state.
2184	 */
2185	lle->lle_hittime = time_uptime;
2186	LLE_REQ_UNLOCK(lle);
2187}
2188
2189static inline uint32_t
2190in6_lltable_hash_dst(const struct in6_addr *dst, uint32_t hsize)
2191{
2192
2193	return (IN6_LLTBL_HASH(dst->s6_addr32[3], hsize));
2194}
2195
2196static uint32_t
2197in6_lltable_hash(const struct llentry *lle, uint32_t hsize)
2198{
2199
2200	return (in6_lltable_hash_dst(&lle->r_l3addr.addr6, hsize));
2201}
2202
2203static void
2204in6_lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
2205{
2206	struct sockaddr_in6 *sin6;
2207
2208	sin6 = (struct sockaddr_in6 *)sa;
2209	bzero(sin6, sizeof(*sin6));
2210	sin6->sin6_family = AF_INET6;
2211	sin6->sin6_len = sizeof(*sin6);
2212	sin6->sin6_addr = lle->r_l3addr.addr6;
2213}
2214
2215static inline struct llentry *
2216in6_lltable_find_dst(struct lltable *llt, const struct in6_addr *dst)
2217{
2218	struct llentry *lle;
2219	struct llentries *lleh;
2220	u_int hashidx;
2221
2222	hashidx = in6_lltable_hash_dst(dst, llt->llt_hsize);
2223	lleh = &llt->lle_head[hashidx];
2224	CK_LIST_FOREACH(lle, lleh, lle_next) {
2225		if (lle->la_flags & LLE_DELETED)
2226			continue;
2227		if (IN6_ARE_ADDR_EQUAL(&lle->r_l3addr.addr6, dst))
2228			break;
2229	}
2230
2231	return (lle);
2232}
2233
2234static void
2235in6_lltable_delete_entry(struct lltable *llt, struct llentry *lle)
2236{
2237
2238	lle->la_flags |= LLE_DELETED;
2239	EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
2240#ifdef DIAGNOSTIC
2241	log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
2242#endif
2243	llentry_free(lle);
2244}
2245
2246static struct llentry *
2247in6_lltable_alloc(struct lltable *llt, u_int flags,
2248	const struct sockaddr *l3addr)
2249{
2250	const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr;
2251	struct ifnet *ifp = llt->llt_ifp;
2252	struct llentry *lle;
2253	char linkhdr[LLE_MAX_LINKHDR];
2254	size_t linkhdrsize;
2255	int lladdr_off;
2256
2257	KASSERT(l3addr->sa_family == AF_INET6,
2258	    ("sin_family %d", l3addr->sa_family));
2259
2260	/*
2261	 * A route that covers the given address must have
2262	 * been installed 1st because we are doing a resolution,
2263	 * verify this.
2264	 */
2265	if (!(flags & LLE_IFADDR) &&
2266	    in6_lltable_rtcheck(ifp, flags, l3addr) != 0)
2267		return (NULL);
2268
2269	lle = in6_lltable_new(&sin6->sin6_addr, flags);
2270	if (lle == NULL) {
2271		log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
2272		return (NULL);
2273	}
2274	lle->la_flags = flags;
2275	if ((flags & LLE_IFADDR) == LLE_IFADDR) {
2276		linkhdrsize = LLE_MAX_LINKHDR;
2277		if (lltable_calc_llheader(ifp, AF_INET6, IF_LLADDR(ifp),
2278		    linkhdr, &linkhdrsize, &lladdr_off) != 0) {
2279			epoch_call(net_epoch_preempt,  &lle->lle_epoch_ctx, in6_lltable_destroy_lle_unlocked);
2280			return (NULL);
2281		}
2282		lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
2283		    lladdr_off);
2284		lle->la_flags |= LLE_STATIC;
2285	}
2286
2287	if ((lle->la_flags & LLE_STATIC) != 0)
2288		lle->ln_state = ND6_LLINFO_REACHABLE;
2289
2290	return (lle);
2291}
2292
2293static struct llentry *
2294in6_lltable_lookup(struct lltable *llt, u_int flags,
2295	const struct sockaddr *l3addr)
2296{
2297	const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr;
2298	struct llentry *lle;
2299
2300	IF_AFDATA_LOCK_ASSERT(llt->llt_ifp);
2301	KASSERT(l3addr->sa_family == AF_INET6,
2302	    ("sin_family %d", l3addr->sa_family));
2303	KASSERT((flags & (LLE_UNLOCKED | LLE_EXCLUSIVE)) !=
2304	    (LLE_UNLOCKED | LLE_EXCLUSIVE),
2305	    ("wrong lle request flags: %#x", flags));
2306
2307	lle = in6_lltable_find_dst(llt, &sin6->sin6_addr);
2308	if (lle == NULL)
2309		return (NULL);
2310	if (flags & LLE_UNLOCKED)
2311		return (lle);
2312
2313	if (flags & LLE_EXCLUSIVE)
2314		LLE_WLOCK(lle);
2315	else
2316		LLE_RLOCK(lle);
2317
2318	/*
2319	 * If the afdata lock is not held, the LLE may have been unlinked while
2320	 * we were blocked on the LLE lock.  Check for this case.
2321	 */
2322	if (__predict_false((lle->la_flags & LLE_LINKED) == 0)) {
2323		if (flags & LLE_EXCLUSIVE)
2324			LLE_WUNLOCK(lle);
2325		else
2326			LLE_RUNLOCK(lle);
2327		return (NULL);
2328	}
2329	return (lle);
2330}
2331
2332static int
2333in6_lltable_dump_entry(struct lltable *llt, struct llentry *lle,
2334    struct sysctl_req *wr)
2335{
2336	struct ifnet *ifp = llt->llt_ifp;
2337	/* XXX stack use */
2338	struct {
2339		struct rt_msghdr	rtm;
2340		struct sockaddr_in6	sin6;
2341		/*
2342		 * ndp.c assumes that sdl is word aligned
2343		 */
2344#ifdef __LP64__
2345		uint32_t		pad;
2346#endif
2347		struct sockaddr_dl	sdl;
2348	} ndpc;
2349	struct sockaddr_dl *sdl;
2350	int error;
2351
2352	bzero(&ndpc, sizeof(ndpc));
2353	/* skip deleted entries */
2354	if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
2355		return (0);
2356	/* Skip if jailed and not a valid IP of the prison. */
2357	lltable_fill_sa_entry(lle, (struct sockaddr *)&ndpc.sin6);
2358	if (prison_if(wr->td->td_ucred, (struct sockaddr *)&ndpc.sin6) != 0)
2359		return (0);
2360	/*
2361	 * produce a msg made of:
2362	 *  struct rt_msghdr;
2363	 *  struct sockaddr_in6 (IPv6)
2364	 *  struct sockaddr_dl;
2365	 */
2366	ndpc.rtm.rtm_msglen = sizeof(ndpc);
2367	ndpc.rtm.rtm_version = RTM_VERSION;
2368	ndpc.rtm.rtm_type = RTM_GET;
2369	ndpc.rtm.rtm_flags = RTF_UP;
2370	ndpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
2371	if (V_deembed_scopeid)
2372		sa6_recoverscope(&ndpc.sin6);
2373
2374	/* publish */
2375	if (lle->la_flags & LLE_PUB)
2376		ndpc.rtm.rtm_flags |= RTF_ANNOUNCE;
2377
2378	sdl = &ndpc.sdl;
2379	sdl->sdl_family = AF_LINK;
2380	sdl->sdl_len = sizeof(*sdl);
2381	sdl->sdl_index = ifp->if_index;
2382	sdl->sdl_type = ifp->if_type;
2383	if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
2384		sdl->sdl_alen = ifp->if_addrlen;
2385		bcopy(lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
2386	} else {
2387		sdl->sdl_alen = 0;
2388		bzero(LLADDR(sdl), ifp->if_addrlen);
2389	}
2390	if (lle->la_expire != 0)
2391		ndpc.rtm.rtm_rmx.rmx_expire = lle->la_expire +
2392		    lle->lle_remtime / hz + time_second - time_uptime;
2393	ndpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
2394	if (lle->la_flags & LLE_STATIC)
2395		ndpc.rtm.rtm_flags |= RTF_STATIC;
2396	if (lle->la_flags & LLE_IFADDR)
2397		ndpc.rtm.rtm_flags |= RTF_PINNED;
2398	if (lle->ln_router != 0)
2399		ndpc.rtm.rtm_flags |= RTF_GATEWAY;
2400	ndpc.rtm.rtm_rmx.rmx_pksent = lle->la_asked;
2401	/* Store state in rmx_weight value */
2402	ndpc.rtm.rtm_rmx.rmx_state = lle->ln_state;
2403	ndpc.rtm.rtm_index = ifp->if_index;
2404	error = SYSCTL_OUT(wr, &ndpc, sizeof(ndpc));
2405
2406	return (error);
2407}
2408
2409static struct lltable *
2410in6_lltattach(struct ifnet *ifp)
2411{
2412	struct lltable *llt;
2413
2414	llt = lltable_allocate_htbl(IN6_LLTBL_DEFAULT_HSIZE);
2415	llt->llt_af = AF_INET6;
2416	llt->llt_ifp = ifp;
2417
2418	llt->llt_lookup = in6_lltable_lookup;
2419	llt->llt_alloc_entry = in6_lltable_alloc;
2420	llt->llt_delete_entry = in6_lltable_delete_entry;
2421	llt->llt_dump_entry = in6_lltable_dump_entry;
2422	llt->llt_hash = in6_lltable_hash;
2423	llt->llt_fill_sa_entry = in6_lltable_fill_sa_entry;
2424	llt->llt_free_entry = in6_lltable_free_entry;
2425	llt->llt_match_prefix = in6_lltable_match_prefix;
2426	llt->llt_mark_used = in6_lltable_mark_used;
2427 	lltable_link(llt);
2428
2429	return (llt);
2430}
2431
2432void *
2433in6_domifattach(struct ifnet *ifp)
2434{
2435	struct in6_ifextra *ext;
2436
2437	/* There are not IPv6-capable interfaces. */
2438	switch (ifp->if_type) {
2439	case IFT_PFLOG:
2440	case IFT_PFSYNC:
2441	case IFT_USB:
2442		return (NULL);
2443	}
2444	ext = (struct in6_ifextra *)malloc(sizeof(*ext), M_IFADDR, M_WAITOK);
2445	bzero(ext, sizeof(*ext));
2446
2447	ext->in6_ifstat = malloc(sizeof(counter_u64_t) *
2448	    sizeof(struct in6_ifstat) / sizeof(uint64_t), M_IFADDR, M_WAITOK);
2449	COUNTER_ARRAY_ALLOC(ext->in6_ifstat,
2450	    sizeof(struct in6_ifstat) / sizeof(uint64_t), M_WAITOK);
2451
2452	ext->icmp6_ifstat = malloc(sizeof(counter_u64_t) *
2453	    sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_IFADDR,
2454	    M_WAITOK);
2455	COUNTER_ARRAY_ALLOC(ext->icmp6_ifstat,
2456	    sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_WAITOK);
2457
2458	ext->nd_ifinfo = nd6_ifattach(ifp);
2459	ext->scope6_id = scope6_ifattach(ifp);
2460	ext->lltable = in6_lltattach(ifp);
2461
2462	ext->mld_ifinfo = mld_domifattach(ifp);
2463
2464	return ext;
2465}
2466
2467int
2468in6_domifmtu(struct ifnet *ifp)
2469{
2470	if (ifp->if_afdata[AF_INET6] == NULL)
2471		return ifp->if_mtu;
2472
2473	return (IN6_LINKMTU(ifp));
2474}
2475
2476void
2477in6_domifdetach(struct ifnet *ifp, void *aux)
2478{
2479	struct in6_ifextra *ext = (struct in6_ifextra *)aux;
2480
2481	mld_domifdetach(ifp);
2482	scope6_ifdetach(ext->scope6_id);
2483	nd6_ifdetach(ifp, ext->nd_ifinfo);
2484	lltable_free(ext->lltable);
2485	COUNTER_ARRAY_FREE(ext->in6_ifstat,
2486	    sizeof(struct in6_ifstat) / sizeof(uint64_t));
2487	free(ext->in6_ifstat, M_IFADDR);
2488	COUNTER_ARRAY_FREE(ext->icmp6_ifstat,
2489	    sizeof(struct icmp6_ifstat) / sizeof(uint64_t));
2490	free(ext->icmp6_ifstat, M_IFADDR);
2491	free(ext, M_IFADDR);
2492}
2493
2494/*
2495 * Convert sockaddr_in6 to sockaddr_in.  Original sockaddr_in6 must be
2496 * v4 mapped addr or v4 compat addr
2497 */
2498void
2499in6_sin6_2_sin(struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
2500{
2501
2502	bzero(sin, sizeof(*sin));
2503	sin->sin_len = sizeof(struct sockaddr_in);
2504	sin->sin_family = AF_INET;
2505	sin->sin_port = sin6->sin6_port;
2506	sin->sin_addr.s_addr = sin6->sin6_addr.s6_addr32[3];
2507}
2508
2509/* Convert sockaddr_in to sockaddr_in6 in v4 mapped addr format. */
2510void
2511in6_sin_2_v4mapsin6(struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
2512{
2513	bzero(sin6, sizeof(*sin6));
2514	sin6->sin6_len = sizeof(struct sockaddr_in6);
2515	sin6->sin6_family = AF_INET6;
2516	sin6->sin6_port = sin->sin_port;
2517	sin6->sin6_addr.s6_addr32[0] = 0;
2518	sin6->sin6_addr.s6_addr32[1] = 0;
2519	sin6->sin6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
2520	sin6->sin6_addr.s6_addr32[3] = sin->sin_addr.s_addr;
2521}
2522
2523/* Convert sockaddr_in6 into sockaddr_in. */
2524void
2525in6_sin6_2_sin_in_sock(struct sockaddr *nam)
2526{
2527	struct sockaddr_in *sin_p;
2528	struct sockaddr_in6 sin6;
2529
2530	/*
2531	 * Save original sockaddr_in6 addr and convert it
2532	 * to sockaddr_in.
2533	 */
2534	sin6 = *(struct sockaddr_in6 *)nam;
2535	sin_p = (struct sockaddr_in *)nam;
2536	in6_sin6_2_sin(sin_p, &sin6);
2537}
2538
2539/* Convert sockaddr_in into sockaddr_in6 in v4 mapped addr format. */
2540void
2541in6_sin_2_v4mapsin6_in_sock(struct sockaddr **nam)
2542{
2543	struct sockaddr_in *sin_p;
2544	struct sockaddr_in6 *sin6_p;
2545
2546	sin6_p = malloc(sizeof *sin6_p, M_SONAME, M_WAITOK);
2547	sin_p = (struct sockaddr_in *)*nam;
2548	in6_sin_2_v4mapsin6(sin_p, sin6_p);
2549	free(*nam, M_SONAME);
2550	*nam = (struct sockaddr *)sin6_p;
2551}
2552