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/sysctl.h>
88#include <sys/syslog.h>
89
90#include <net/if.h>
91#include <net/if_var.h>
92#include <net/if_types.h>
93#include <net/route.h>
94#include <net/route/route_ctl.h>
95#include <net/route/nhop.h>
96#include <net/if_dl.h>
97#include <net/vnet.h>
98
99#include <netinet/in.h>
100#include <netinet/in_var.h>
101#include <net/if_llatbl.h>
102#include <netinet/if_ether.h>
103#include <netinet/in_systm.h>
104#include <netinet/ip.h>
105#include <netinet/in_pcb.h>
106#include <netinet/ip_carp.h>
107
108#include <netinet/ip6.h>
109#include <netinet6/ip6_var.h>
110#include <netinet6/nd6.h>
111#include <netinet6/mld6_var.h>
112#include <netinet6/ip6_mroute.h>
113#include <netinet6/in6_ifattach.h>
114#include <netinet6/scope6_var.h>
115#include <netinet6/in6_fib.h>
116#include <netinet6/in6_pcb.h>
117
118/*
119 * struct in6_ifreq and struct ifreq must be type punnable for common members
120 * of ifr_ifru to allow accessors to be shared.
121 */
122_Static_assert(offsetof(struct in6_ifreq, ifr_ifru) ==
123    offsetof(struct ifreq, ifr_ifru),
124    "struct in6_ifreq and struct ifreq are not type punnable");
125
126VNET_DECLARE(int, icmp6_nodeinfo_oldmcprefix);
127#define V_icmp6_nodeinfo_oldmcprefix	VNET(icmp6_nodeinfo_oldmcprefix)
128
129/*
130 * Definitions of some costant IP6 addresses.
131 */
132const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT;
133const struct in6_addr in6addr_loopback = IN6ADDR_LOOPBACK_INIT;
134const struct in6_addr in6addr_nodelocal_allnodes =
135	IN6ADDR_NODELOCAL_ALLNODES_INIT;
136const struct in6_addr in6addr_linklocal_allnodes =
137	IN6ADDR_LINKLOCAL_ALLNODES_INIT;
138const struct in6_addr in6addr_linklocal_allrouters =
139	IN6ADDR_LINKLOCAL_ALLROUTERS_INIT;
140const struct in6_addr in6addr_linklocal_allv2routers =
141	IN6ADDR_LINKLOCAL_ALLV2ROUTERS_INIT;
142
143const struct in6_addr in6mask0 = IN6MASK0;
144const struct in6_addr in6mask32 = IN6MASK32;
145const struct in6_addr in6mask64 = IN6MASK64;
146const struct in6_addr in6mask96 = IN6MASK96;
147const struct in6_addr in6mask128 = IN6MASK128;
148
149const struct sockaddr_in6 sa6_any =
150	{ sizeof(sa6_any), AF_INET6, 0, 0, IN6ADDR_ANY_INIT, 0 };
151
152static int in6_notify_ifa(struct ifnet *, struct in6_ifaddr *,
153	struct in6_aliasreq *, int);
154static void in6_unlink_ifa(struct in6_ifaddr *, struct ifnet *);
155
156static int in6_validate_ifra(struct ifnet *, struct in6_aliasreq *,
157    struct in6_ifaddr *, int);
158static struct in6_ifaddr *in6_alloc_ifa(struct ifnet *,
159    struct in6_aliasreq *, int flags);
160static int in6_update_ifa_internal(struct ifnet *, struct in6_aliasreq *,
161    struct in6_ifaddr *, int, int);
162static int in6_broadcast_ifa(struct ifnet *, struct in6_aliasreq *,
163    struct in6_ifaddr *, int);
164
165#define ifa2ia6(ifa)	((struct in6_ifaddr *)(ifa))
166#define ia62ifa(ia6)	(&((ia6)->ia_ifa))
167
168void
169in6_newaddrmsg(struct in6_ifaddr *ia, int cmd)
170{
171	struct rt_addrinfo info;
172	struct ifaddr *ifa;
173	struct sockaddr_dl gateway;
174	int fibnum;
175
176	ifa = &ia->ia_ifa;
177
178	/*
179	 * Prepare info data for the host route.
180	 * This code mimics one from ifa_maintain_loopback_route().
181	 */
182	bzero(&info, sizeof(struct rt_addrinfo));
183	info.rti_flags = ifa->ifa_flags | RTF_HOST | RTF_STATIC | RTF_PINNED;
184	info.rti_info[RTAX_DST] = ifa->ifa_addr;
185	info.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&gateway;
186	link_init_sdl(ifa->ifa_ifp, (struct sockaddr *)&gateway, ifa->ifa_ifp->if_type);
187	if (cmd != RTM_DELETE)
188		info.rti_ifp = V_loif;
189
190	fibnum = ia62ifa(ia)->ifa_ifp->if_fib;
191
192	if (cmd == RTM_ADD) {
193		rt_addrmsg(cmd, &ia->ia_ifa, fibnum);
194		rt_routemsg_info(cmd, &info, fibnum);
195	} else if (cmd == RTM_DELETE) {
196		rt_routemsg_info(cmd, &info, fibnum);
197		rt_addrmsg(cmd, &ia->ia_ifa, fibnum);
198	}
199}
200
201int
202in6_mask2len(struct in6_addr *mask, u_char *lim0)
203{
204	int x = 0, y;
205	u_char *lim = lim0, *p;
206
207	/* ignore the scope_id part */
208	if (lim0 == NULL || lim0 - (u_char *)mask > sizeof(*mask))
209		lim = (u_char *)mask + sizeof(*mask);
210	for (p = (u_char *)mask; p < lim; x++, p++) {
211		if (*p != 0xff)
212			break;
213	}
214	y = 0;
215	if (p < lim) {
216		for (y = 0; y < 8; y++) {
217			if ((*p & (0x80 >> y)) == 0)
218				break;
219		}
220	}
221
222	/*
223	 * when the limit pointer is given, do a stricter check on the
224	 * remaining bits.
225	 */
226	if (p < lim) {
227		if (y != 0 && (*p & (0x00ff >> y)) != 0)
228			return (-1);
229		for (p = p + 1; p < lim; p++)
230			if (*p != 0)
231				return (-1);
232	}
233
234	return x * 8 + y;
235}
236
237#ifdef COMPAT_FREEBSD32
238struct in6_ndifreq32 {
239	char ifname[IFNAMSIZ];
240	uint32_t ifindex;
241};
242#define	SIOCGDEFIFACE32_IN6	_IOWR('i', 86, struct in6_ndifreq32)
243#endif
244
245int
246in6_control(struct socket *so, u_long cmd, caddr_t data,
247    struct ifnet *ifp, struct thread *td)
248{
249	struct	in6_ifreq *ifr = (struct in6_ifreq *)data;
250	struct	in6_ifaddr *ia = NULL;
251	struct	in6_aliasreq *ifra = (struct in6_aliasreq *)data;
252	struct sockaddr_in6 *sa6;
253	int carp_attached = 0;
254	int error;
255	u_long ocmd = cmd;
256
257	/*
258	 * Compat to make pre-10.x ifconfig(8) operable.
259	 */
260	if (cmd == OSIOCAIFADDR_IN6)
261		cmd = SIOCAIFADDR_IN6;
262
263	switch (cmd) {
264	case SIOCGETSGCNT_IN6:
265	case SIOCGETMIFCNT_IN6:
266		/*
267		 * XXX mrt_ioctl has a 3rd, unused, FIB argument in route.c.
268		 * We cannot see how that would be needed, so do not adjust the
269		 * KPI blindly; more likely should clean up the IPv4 variant.
270		 */
271		return (mrt6_ioctl ? mrt6_ioctl(cmd, data) : EOPNOTSUPP);
272	}
273
274	switch (cmd) {
275	case SIOCAADDRCTL_POLICY:
276	case SIOCDADDRCTL_POLICY:
277		if (td != NULL) {
278			error = priv_check(td, PRIV_NETINET_ADDRCTRL6);
279			if (error)
280				return (error);
281		}
282		return (in6_src_ioctl(cmd, data));
283	}
284
285	if (ifp == NULL)
286		return (EOPNOTSUPP);
287
288	switch (cmd) {
289	case SIOCSNDFLUSH_IN6:
290	case SIOCSPFXFLUSH_IN6:
291	case SIOCSRTRFLUSH_IN6:
292	case SIOCSDEFIFACE_IN6:
293	case SIOCSIFINFO_FLAGS:
294	case SIOCSIFINFO_IN6:
295		if (td != NULL) {
296			error = priv_check(td, PRIV_NETINET_ND6);
297			if (error)
298				return (error);
299		}
300		/* FALLTHROUGH */
301	case OSIOCGIFINFO_IN6:
302	case SIOCGIFINFO_IN6:
303	case SIOCGNBRINFO_IN6:
304	case SIOCGDEFIFACE_IN6:
305		return (nd6_ioctl(cmd, data, ifp));
306
307#ifdef COMPAT_FREEBSD32
308	case SIOCGDEFIFACE32_IN6:
309		{
310			struct in6_ndifreq ndif;
311			struct in6_ndifreq32 *ndif32;
312
313			error = nd6_ioctl(SIOCGDEFIFACE_IN6, (caddr_t)&ndif,
314			    ifp);
315			if (error)
316				return (error);
317			ndif32 = (struct in6_ndifreq32 *)data;
318			ndif32->ifindex = ndif.ifindex;
319			return (0);
320		}
321#endif
322	}
323
324	switch (cmd) {
325	case SIOCSIFPREFIX_IN6:
326	case SIOCDIFPREFIX_IN6:
327	case SIOCAIFPREFIX_IN6:
328	case SIOCCIFPREFIX_IN6:
329	case SIOCSGIFPREFIX_IN6:
330	case SIOCGIFPREFIX_IN6:
331		log(LOG_NOTICE,
332		    "prefix ioctls are now invalidated. "
333		    "please use ifconfig.\n");
334		return (EOPNOTSUPP);
335	}
336
337	switch (cmd) {
338	case SIOCSSCOPE6:
339		if (td != NULL) {
340			error = priv_check(td, PRIV_NETINET_SCOPE6);
341			if (error)
342				return (error);
343		}
344		/* FALLTHROUGH */
345	case SIOCGSCOPE6:
346	case SIOCGSCOPE6DEF:
347		return (scope6_ioctl(cmd, data, ifp));
348	}
349
350	/*
351	 * Find address for this interface, if it exists.
352	 *
353	 * In netinet code, we have checked ifra_addr in SIOCSIF*ADDR operation
354	 * only, and used the first interface address as the target of other
355	 * operations (without checking ifra_addr).  This was because netinet
356	 * code/API assumed at most 1 interface address per interface.
357	 * Since IPv6 allows a node to assign multiple addresses
358	 * on a single interface, we almost always look and check the
359	 * presence of ifra_addr, and reject invalid ones here.
360	 * It also decreases duplicated code among SIOC*_IN6 operations.
361	 */
362	switch (cmd) {
363	case SIOCAIFADDR_IN6:
364	case SIOCSIFPHYADDR_IN6:
365		sa6 = &ifra->ifra_addr;
366		break;
367	case SIOCSIFADDR_IN6:
368	case SIOCGIFADDR_IN6:
369	case SIOCSIFDSTADDR_IN6:
370	case SIOCSIFNETMASK_IN6:
371	case SIOCGIFDSTADDR_IN6:
372	case SIOCGIFNETMASK_IN6:
373	case SIOCDIFADDR_IN6:
374	case SIOCGIFPSRCADDR_IN6:
375	case SIOCGIFPDSTADDR_IN6:
376	case SIOCGIFAFLAG_IN6:
377	case SIOCSNDFLUSH_IN6:
378	case SIOCSPFXFLUSH_IN6:
379	case SIOCSRTRFLUSH_IN6:
380	case SIOCGIFALIFETIME_IN6:
381	case SIOCGIFSTAT_IN6:
382	case SIOCGIFSTAT_ICMP6:
383		sa6 = &ifr->ifr_addr;
384		break;
385	case SIOCSIFADDR:
386	case SIOCSIFBRDADDR:
387	case SIOCSIFDSTADDR:
388	case SIOCSIFNETMASK:
389		/*
390		 * Although we should pass any non-INET6 ioctl requests
391		 * down to driver, we filter some legacy INET requests.
392		 * Drivers trust SIOCSIFADDR et al to come from an already
393		 * privileged layer, and do not perform any credentials
394		 * checks or input validation.
395		 */
396		return (EINVAL);
397	default:
398		sa6 = NULL;
399		break;
400	}
401	if (sa6 && sa6->sin6_family == AF_INET6) {
402		if (sa6->sin6_scope_id != 0)
403			error = sa6_embedscope(sa6, 0);
404		else
405			error = in6_setscope(&sa6->sin6_addr, ifp, NULL);
406		if (error != 0)
407			return (error);
408		if (td != NULL && (error = prison_check_ip6(td->td_ucred,
409		    &sa6->sin6_addr)) != 0)
410			return (error);
411		ia = in6ifa_ifpwithaddr(ifp, &sa6->sin6_addr);
412	} else
413		ia = NULL;
414
415	switch (cmd) {
416	case SIOCSIFADDR_IN6:
417	case SIOCSIFDSTADDR_IN6:
418	case SIOCSIFNETMASK_IN6:
419		/*
420		 * Since IPv6 allows a node to assign multiple addresses
421		 * on a single interface, SIOCSIFxxx ioctls are deprecated.
422		 */
423		/* we decided to obsolete this command (20000704) */
424		error = EINVAL;
425		goto out;
426
427	case SIOCDIFADDR_IN6:
428		/*
429		 * for IPv4, we look for existing in_ifaddr here to allow
430		 * "ifconfig if0 delete" to remove the first IPv4 address on
431		 * the interface.  For IPv6, as the spec allows multiple
432		 * interface address from the day one, we consider "remove the
433		 * first one" semantics to be not preferable.
434		 */
435		if (ia == NULL) {
436			error = EADDRNOTAVAIL;
437			goto out;
438		}
439		/* FALLTHROUGH */
440	case SIOCAIFADDR_IN6:
441		/*
442		 * We always require users to specify a valid IPv6 address for
443		 * the corresponding operation.
444		 */
445		if (ifra->ifra_addr.sin6_family != AF_INET6 ||
446		    ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6)) {
447			error = EAFNOSUPPORT;
448			goto out;
449		}
450
451		if (td != NULL) {
452			error = priv_check(td, (cmd == SIOCDIFADDR_IN6) ?
453			    PRIV_NET_DELIFADDR : PRIV_NET_ADDIFADDR);
454			if (error)
455				goto out;
456		}
457		/* FALLTHROUGH */
458	case SIOCGIFSTAT_IN6:
459	case SIOCGIFSTAT_ICMP6:
460		if (ifp->if_afdata[AF_INET6] == NULL) {
461			error = EPFNOSUPPORT;
462			goto out;
463		}
464		break;
465
466	case SIOCGIFADDR_IN6:
467		/* This interface is basically deprecated. use SIOCGIFCONF. */
468		/* FALLTHROUGH */
469	case SIOCGIFAFLAG_IN6:
470	case SIOCGIFNETMASK_IN6:
471	case SIOCGIFDSTADDR_IN6:
472	case SIOCGIFALIFETIME_IN6:
473		/* must think again about its semantics */
474		if (ia == NULL) {
475			error = EADDRNOTAVAIL;
476			goto out;
477		}
478		break;
479	}
480
481	switch (cmd) {
482	case SIOCGIFADDR_IN6:
483		ifr->ifr_addr = ia->ia_addr;
484		if ((error = sa6_recoverscope(&ifr->ifr_addr)) != 0)
485			goto out;
486		break;
487
488	case SIOCGIFDSTADDR_IN6:
489		if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
490			error = EINVAL;
491			goto out;
492		}
493		ifr->ifr_dstaddr = ia->ia_dstaddr;
494		if ((error = sa6_recoverscope(&ifr->ifr_dstaddr)) != 0)
495			goto out;
496		break;
497
498	case SIOCGIFNETMASK_IN6:
499		ifr->ifr_addr = ia->ia_prefixmask;
500		break;
501
502	case SIOCGIFAFLAG_IN6:
503		ifr->ifr_ifru.ifru_flags6 = ia->ia6_flags;
504		break;
505
506	case SIOCGIFSTAT_IN6:
507		COUNTER_ARRAY_COPY(((struct in6_ifextra *)
508		    ifp->if_afdata[AF_INET6])->in6_ifstat,
509		    &ifr->ifr_ifru.ifru_stat,
510		    sizeof(struct in6_ifstat) / sizeof(uint64_t));
511		break;
512
513	case SIOCGIFSTAT_ICMP6:
514		COUNTER_ARRAY_COPY(((struct in6_ifextra *)
515		    ifp->if_afdata[AF_INET6])->icmp6_ifstat,
516		    &ifr->ifr_ifru.ifru_icmp6stat,
517		    sizeof(struct icmp6_ifstat) / sizeof(uint64_t));
518		break;
519
520	case SIOCGIFALIFETIME_IN6:
521		ifr->ifr_ifru.ifru_lifetime = ia->ia6_lifetime;
522		if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
523			time_t maxexpire;
524			struct in6_addrlifetime *retlt =
525			    &ifr->ifr_ifru.ifru_lifetime;
526
527			/*
528			 * XXX: adjust expiration time assuming time_t is
529			 * signed.
530			 */
531			maxexpire = (-1) &
532			    ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
533			if (ia->ia6_lifetime.ia6t_vltime <
534			    maxexpire - ia->ia6_updatetime) {
535				retlt->ia6t_expire = ia->ia6_updatetime +
536				    ia->ia6_lifetime.ia6t_vltime;
537			} else
538				retlt->ia6t_expire = maxexpire;
539		}
540		if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
541			time_t maxexpire;
542			struct in6_addrlifetime *retlt =
543			    &ifr->ifr_ifru.ifru_lifetime;
544
545			/*
546			 * XXX: adjust expiration time assuming time_t is
547			 * signed.
548			 */
549			maxexpire = (-1) &
550			    ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
551			if (ia->ia6_lifetime.ia6t_pltime <
552			    maxexpire - ia->ia6_updatetime) {
553				retlt->ia6t_preferred = ia->ia6_updatetime +
554				    ia->ia6_lifetime.ia6t_pltime;
555			} else
556				retlt->ia6t_preferred = maxexpire;
557		}
558		break;
559
560	case SIOCAIFADDR_IN6:
561	{
562		struct nd_prefixctl pr0;
563		struct nd_prefix *pr;
564
565		/*
566		 * first, make or update the interface address structure,
567		 * and link it to the list.
568		 */
569		if ((error = in6_update_ifa(ifp, ifra, ia, 0)) != 0)
570			goto out;
571		if (ia != NULL) {
572			if (ia->ia_ifa.ifa_carp)
573				(*carp_detach_p)(&ia->ia_ifa, true);
574			ifa_free(&ia->ia_ifa);
575		}
576		if ((ia = in6ifa_ifpwithaddr(ifp, &ifra->ifra_addr.sin6_addr))
577		    == NULL) {
578			/*
579			 * this can happen when the user specify the 0 valid
580			 * lifetime.
581			 */
582			break;
583		}
584
585		if (cmd == ocmd && ifra->ifra_vhid > 0) {
586			if (carp_attach_p != NULL)
587				error = (*carp_attach_p)(&ia->ia_ifa,
588				    ifra->ifra_vhid);
589			else
590				error = EPROTONOSUPPORT;
591			if (error)
592				goto out;
593			else
594				carp_attached = 1;
595		}
596
597		/*
598		 * then, make the prefix on-link on the interface.
599		 * XXX: we'd rather create the prefix before the address, but
600		 * we need at least one address to install the corresponding
601		 * interface route, so we configure the address first.
602		 */
603
604		/*
605		 * convert mask to prefix length (prefixmask has already
606		 * been validated in in6_update_ifa().
607		 */
608		bzero(&pr0, sizeof(pr0));
609		pr0.ndpr_ifp = ifp;
610		pr0.ndpr_plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr,
611		    NULL);
612		if (pr0.ndpr_plen == 128) {
613			/* we don't need to install a host route. */
614			goto aifaddr_out;
615		}
616		pr0.ndpr_prefix = ifra->ifra_addr;
617		/* apply the mask for safety. */
618		IN6_MASK_ADDR(&pr0.ndpr_prefix.sin6_addr,
619		    &ifra->ifra_prefixmask.sin6_addr);
620
621		/*
622		 * XXX: since we don't have an API to set prefix (not address)
623		 * lifetimes, we just use the same lifetimes as addresses.
624		 * The (temporarily) installed lifetimes can be overridden by
625		 * later advertised RAs (when accept_rtadv is non 0), which is
626		 * an intended behavior.
627		 */
628		pr0.ndpr_raf_onlink = 1; /* should be configurable? */
629		pr0.ndpr_raf_auto =
630		    ((ifra->ifra_flags & IN6_IFF_AUTOCONF) != 0);
631		pr0.ndpr_vltime = ifra->ifra_lifetime.ia6t_vltime;
632		pr0.ndpr_pltime = ifra->ifra_lifetime.ia6t_pltime;
633
634		/* add the prefix if not yet. */
635		if ((pr = nd6_prefix_lookup(&pr0)) == NULL) {
636			/*
637			 * nd6_prelist_add will install the corresponding
638			 * interface route.
639			 */
640			if ((error = nd6_prelist_add(&pr0, NULL, &pr)) != 0) {
641				if (carp_attached)
642					(*carp_detach_p)(&ia->ia_ifa, false);
643				goto out;
644			}
645		}
646
647		/* relate the address to the prefix */
648		if (ia->ia6_ndpr == NULL) {
649			ia->ia6_ndpr = pr;
650			pr->ndpr_addrcnt++;
651
652			/*
653			 * If this is the first autoconf address from the
654			 * prefix, create a temporary address as well
655			 * (when required).
656			 */
657			if ((ia->ia6_flags & IN6_IFF_AUTOCONF) &&
658			    V_ip6_use_tempaddr && pr->ndpr_addrcnt == 1) {
659				int e;
660				if ((e = in6_tmpifadd(ia, 1, 0)) != 0) {
661					log(LOG_NOTICE, "in6_control: failed "
662					    "to create a temporary address, "
663					    "errno=%d\n", e);
664				}
665			}
666		}
667		nd6_prefix_rele(pr);
668
669		/*
670		 * this might affect the status of autoconfigured addresses,
671		 * that is, this address might make other addresses detached.
672		 */
673		pfxlist_onlink_check();
674
675aifaddr_out:
676		/*
677		 * Try to clear the flag when a new IPv6 address is added
678		 * onto an IFDISABLED interface and it succeeds.
679		 */
680		if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) {
681			struct in6_ndireq nd;
682
683			memset(&nd, 0, sizeof(nd));
684			nd.ndi.flags = ND_IFINFO(ifp)->flags;
685			nd.ndi.flags &= ~ND6_IFF_IFDISABLED;
686			if (nd6_ioctl(SIOCSIFINFO_FLAGS, (caddr_t)&nd, ifp) < 0)
687				log(LOG_NOTICE, "SIOCAIFADDR_IN6: "
688				    "SIOCSIFINFO_FLAGS for -ifdisabled "
689				    "failed.");
690			/*
691			 * Ignore failure of clearing the flag intentionally.
692			 * The failure means address duplication was detected.
693			 */
694		}
695		break;
696	}
697
698	case SIOCDIFADDR_IN6:
699		in6_purgeifaddr(ia);
700		EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa,
701		    IFADDR_EVENT_DEL);
702		break;
703
704	default:
705		if (ifp->if_ioctl == NULL) {
706			error = EOPNOTSUPP;
707			goto out;
708		}
709		error = (*ifp->if_ioctl)(ifp, cmd, data);
710		goto out;
711	}
712
713	error = 0;
714out:
715	if (ia != NULL)
716		ifa_free(&ia->ia_ifa);
717	return (error);
718}
719
720static struct in6_multi_mship *
721in6_joingroup_legacy(struct ifnet *ifp, const struct in6_addr *mcaddr,
722    int *errorp, int delay)
723{
724	struct in6_multi_mship *imm;
725	int error;
726
727	imm = malloc(sizeof(*imm), M_IP6MADDR, M_NOWAIT);
728	if (imm == NULL) {
729		*errorp = ENOBUFS;
730		return (NULL);
731	}
732
733	delay = (delay * PR_FASTHZ) / hz;
734
735	error = in6_joingroup(ifp, mcaddr, NULL, &imm->i6mm_maddr, delay);
736	if (error) {
737		*errorp = error;
738		free(imm, M_IP6MADDR);
739		return (NULL);
740	}
741
742	return (imm);
743}
744/*
745 * Join necessary multicast groups.  Factored out from in6_update_ifa().
746 * This entire work should only be done once, for the default FIB.
747 */
748static int
749in6_update_ifa_join_mc(struct ifnet *ifp, struct in6_aliasreq *ifra,
750    struct in6_ifaddr *ia, int flags, struct in6_multi **in6m_sol)
751{
752	char ip6buf[INET6_ADDRSTRLEN];
753	struct in6_addr mltaddr;
754	struct in6_multi_mship *imm;
755	int delay, error;
756
757	KASSERT(in6m_sol != NULL, ("%s: in6m_sol is NULL", __func__));
758
759	/* Join solicited multicast addr for new host id. */
760	bzero(&mltaddr, sizeof(struct in6_addr));
761	mltaddr.s6_addr32[0] = IPV6_ADDR_INT32_MLL;
762	mltaddr.s6_addr32[2] = htonl(1);
763	mltaddr.s6_addr32[3] = ifra->ifra_addr.sin6_addr.s6_addr32[3];
764	mltaddr.s6_addr8[12] = 0xff;
765	if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0) {
766		/* XXX: should not happen */
767		log(LOG_ERR, "%s: in6_setscope failed\n", __func__);
768		goto cleanup;
769	}
770	delay = error = 0;
771	if ((flags & IN6_IFAUPDATE_DADDELAY)) {
772		/*
773		 * We need a random delay for DAD on the address being
774		 * configured.  It also means delaying transmission of the
775		 * corresponding MLD report to avoid report collision.
776		 * [RFC 4861, Section 6.3.7]
777		 */
778		delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
779	}
780	imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
781	if (imm == NULL) {
782		nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
783		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr),
784		    if_name(ifp), error));
785		goto cleanup;
786	}
787	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
788	*in6m_sol = imm->i6mm_maddr;
789
790	/*
791	 * Join link-local all-nodes address.
792	 */
793	mltaddr = in6addr_linklocal_allnodes;
794	if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
795		goto cleanup; /* XXX: should not fail */
796
797	imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0);
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
806	/*
807	 * Join node information group address.
808	 */
809	delay = 0;
810	if ((flags & IN6_IFAUPDATE_DADDELAY)) {
811		/*
812		 * The spec does not say anything about delay for this group,
813		 * but the same logic should apply.
814		 */
815		delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
816	}
817	if (in6_nigroup(ifp, NULL, -1, &mltaddr) == 0) {
818		/* XXX jinmei */
819		imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
820		if (imm == NULL)
821			nd6log((LOG_WARNING,
822			    "%s: in6_joingroup failed for %s on %s "
823			    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
824			    &mltaddr), if_name(ifp), error));
825			/* XXX not very fatal, go on... */
826		else
827			LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
828	}
829	if (V_icmp6_nodeinfo_oldmcprefix &&
830	    in6_nigroup_oldmcprefix(ifp, NULL, -1, &mltaddr) == 0) {
831		imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
832		if (imm == NULL)
833			nd6log((LOG_WARNING,
834			    "%s: in6_joingroup failed for %s on %s "
835			    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
836			    &mltaddr), if_name(ifp), error));
837			/* XXX not very fatal, go on... */
838		else
839			LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
840	}
841
842	/*
843	 * Join interface-local all-nodes address.
844	 * (ff01::1%ifN, and ff01::%ifN/32)
845	 */
846	mltaddr = in6addr_nodelocal_allnodes;
847	if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
848		goto cleanup; /* XXX: should not fail */
849
850	imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0);
851	if (imm == NULL) {
852		nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
853		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
854		    &mltaddr), if_name(ifp), error));
855		goto cleanup;
856	}
857	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
858
859cleanup:
860	return (error);
861}
862
863/*
864 * Update parameters of an IPv6 interface address.
865 * If necessary, a new entry is created and linked into address chains.
866 * This function is separated from in6_control().
867 */
868int
869in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
870    struct in6_ifaddr *ia, int flags)
871{
872	int error, hostIsNew = 0;
873
874	if ((error = in6_validate_ifra(ifp, ifra, ia, flags)) != 0)
875		return (error);
876
877	if (ia == NULL) {
878		hostIsNew = 1;
879		if ((ia = in6_alloc_ifa(ifp, ifra, flags)) == NULL)
880			return (ENOBUFS);
881	}
882
883	error = in6_update_ifa_internal(ifp, ifra, ia, hostIsNew, flags);
884	if (error != 0) {
885		if (hostIsNew != 0) {
886			in6_unlink_ifa(ia, ifp);
887			ifa_free(&ia->ia_ifa);
888		}
889		return (error);
890	}
891
892	if (hostIsNew)
893		error = in6_broadcast_ifa(ifp, ifra, ia, flags);
894
895	return (error);
896}
897
898/*
899 * Fill in basic IPv6 address request info.
900 */
901void
902in6_prepare_ifra(struct in6_aliasreq *ifra, const struct in6_addr *addr,
903    const struct in6_addr *mask)
904{
905
906	memset(ifra, 0, sizeof(struct in6_aliasreq));
907
908	ifra->ifra_addr.sin6_family = AF_INET6;
909	ifra->ifra_addr.sin6_len = sizeof(struct sockaddr_in6);
910	if (addr != NULL)
911		ifra->ifra_addr.sin6_addr = *addr;
912
913	ifra->ifra_prefixmask.sin6_family = AF_INET6;
914	ifra->ifra_prefixmask.sin6_len = sizeof(struct sockaddr_in6);
915	if (mask != NULL)
916		ifra->ifra_prefixmask.sin6_addr = *mask;
917}
918
919static int
920in6_validate_ifra(struct ifnet *ifp, struct in6_aliasreq *ifra,
921    struct in6_ifaddr *ia, int flags)
922{
923	int plen = -1;
924	struct sockaddr_in6 dst6;
925	struct in6_addrlifetime *lt;
926	char ip6buf[INET6_ADDRSTRLEN];
927
928	/* Validate parameters */
929	if (ifp == NULL || ifra == NULL) /* this maybe redundant */
930		return (EINVAL);
931
932	/*
933	 * The destination address for a p2p link must have a family
934	 * of AF_UNSPEC or AF_INET6.
935	 */
936	if ((ifp->if_flags & IFF_POINTOPOINT) != 0 &&
937	    ifra->ifra_dstaddr.sin6_family != AF_INET6 &&
938	    ifra->ifra_dstaddr.sin6_family != AF_UNSPEC)
939		return (EAFNOSUPPORT);
940
941	/*
942	 * Validate address
943	 */
944	if (ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6) ||
945	    ifra->ifra_addr.sin6_family != AF_INET6)
946		return (EINVAL);
947
948	/*
949	 * validate ifra_prefixmask.  don't check sin6_family, netmask
950	 * does not carry fields other than sin6_len.
951	 */
952	if (ifra->ifra_prefixmask.sin6_len > sizeof(struct sockaddr_in6))
953		return (EINVAL);
954	/*
955	 * Because the IPv6 address architecture is classless, we require
956	 * users to specify a (non 0) prefix length (mask) for a new address.
957	 * We also require the prefix (when specified) mask is valid, and thus
958	 * reject a non-consecutive mask.
959	 */
960	if (ia == NULL && ifra->ifra_prefixmask.sin6_len == 0)
961		return (EINVAL);
962	if (ifra->ifra_prefixmask.sin6_len != 0) {
963		plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr,
964		    (u_char *)&ifra->ifra_prefixmask +
965		    ifra->ifra_prefixmask.sin6_len);
966		if (plen <= 0)
967			return (EINVAL);
968	} else {
969		/*
970		 * In this case, ia must not be NULL.  We just use its prefix
971		 * length.
972		 */
973		plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL);
974	}
975	/*
976	 * If the destination address on a p2p interface is specified,
977	 * and the address is a scoped one, validate/set the scope
978	 * zone identifier.
979	 */
980	dst6 = ifra->ifra_dstaddr;
981	if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) != 0 &&
982	    (dst6.sin6_family == AF_INET6)) {
983		struct in6_addr in6_tmp;
984		u_int32_t zoneid;
985
986		in6_tmp = dst6.sin6_addr;
987		if (in6_setscope(&in6_tmp, ifp, &zoneid))
988			return (EINVAL); /* XXX: should be impossible */
989
990		if (dst6.sin6_scope_id != 0) {
991			if (dst6.sin6_scope_id != zoneid)
992				return (EINVAL);
993		} else		/* user omit to specify the ID. */
994			dst6.sin6_scope_id = zoneid;
995
996		/* convert into the internal form */
997		if (sa6_embedscope(&dst6, 0))
998			return (EINVAL); /* XXX: should be impossible */
999	}
1000	/* Modify original ifra_dstaddr to reflect changes */
1001	ifra->ifra_dstaddr = dst6;
1002
1003	/*
1004	 * The destination address can be specified only for a p2p or a
1005	 * loopback interface.  If specified, the corresponding prefix length
1006	 * must be 128.
1007	 */
1008	if (ifra->ifra_dstaddr.sin6_family == AF_INET6) {
1009		if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) == 0) {
1010			/* XXX: noisy message */
1011			nd6log((LOG_INFO, "in6_update_ifa: a destination can "
1012			    "be specified for a p2p or a loopback IF only\n"));
1013			return (EINVAL);
1014		}
1015		if (plen != 128) {
1016			nd6log((LOG_INFO, "in6_update_ifa: prefixlen should "
1017			    "be 128 when dstaddr is specified\n"));
1018			return (EINVAL);
1019		}
1020	}
1021	/* lifetime consistency check */
1022	lt = &ifra->ifra_lifetime;
1023	if (lt->ia6t_pltime > lt->ia6t_vltime)
1024		return (EINVAL);
1025	if (lt->ia6t_vltime == 0) {
1026		/*
1027		 * the following log might be noisy, but this is a typical
1028		 * configuration mistake or a tool's bug.
1029		 */
1030		nd6log((LOG_INFO,
1031		    "in6_update_ifa: valid lifetime is 0 for %s\n",
1032		    ip6_sprintf(ip6buf, &ifra->ifra_addr.sin6_addr)));
1033
1034		if (ia == NULL)
1035			return (0); /* there's nothing to do */
1036	}
1037
1038	/* Check prefix mask */
1039	if (ia != NULL && ifra->ifra_prefixmask.sin6_len != 0) {
1040		/*
1041		 * We prohibit changing the prefix length of an existing
1042		 * address, because
1043		 * + such an operation should be rare in IPv6, and
1044		 * + the operation would confuse prefix management.
1045		 */
1046		if (ia->ia_prefixmask.sin6_len != 0 &&
1047		    in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL) != plen) {
1048			nd6log((LOG_INFO, "in6_validate_ifa: the prefix length "
1049			    "of an existing %s address should not be changed\n",
1050			    ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
1051
1052			return (EINVAL);
1053		}
1054	}
1055
1056	return (0);
1057}
1058
1059/*
1060 * Allocate a new ifaddr and link it into chains.
1061 */
1062static struct in6_ifaddr *
1063in6_alloc_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, int flags)
1064{
1065	struct in6_ifaddr *ia;
1066
1067	/*
1068	 * When in6_alloc_ifa() is called in a process of a received
1069	 * RA, it is called under an interrupt context.  So, we should
1070	 * call malloc with M_NOWAIT.
1071	 */
1072	ia = (struct in6_ifaddr *)ifa_alloc(sizeof(*ia), M_NOWAIT);
1073	if (ia == NULL)
1074		return (NULL);
1075	LIST_INIT(&ia->ia6_memberships);
1076	/* Initialize the address and masks, and put time stamp */
1077	ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr;
1078	ia->ia_addr.sin6_family = AF_INET6;
1079	ia->ia_addr.sin6_len = sizeof(ia->ia_addr);
1080	/* XXX: Can we assign ,sin6_addr and skip the rest? */
1081	ia->ia_addr = ifra->ifra_addr;
1082	ia->ia6_createtime = time_uptime;
1083	if ((ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) != 0) {
1084		/*
1085		 * Some functions expect that ifa_dstaddr is not
1086		 * NULL for p2p interfaces.
1087		 */
1088		ia->ia_ifa.ifa_dstaddr =
1089		    (struct sockaddr *)&ia->ia_dstaddr;
1090	} else {
1091		ia->ia_ifa.ifa_dstaddr = NULL;
1092	}
1093
1094	/* set prefix mask if any */
1095	ia->ia_ifa.ifa_netmask = (struct sockaddr *)&ia->ia_prefixmask;
1096	if (ifra->ifra_prefixmask.sin6_len != 0) {
1097		ia->ia_prefixmask.sin6_family = AF_INET6;
1098		ia->ia_prefixmask.sin6_len = ifra->ifra_prefixmask.sin6_len;
1099		ia->ia_prefixmask.sin6_addr = ifra->ifra_prefixmask.sin6_addr;
1100	}
1101
1102	ia->ia_ifp = ifp;
1103	ifa_ref(&ia->ia_ifa);			/* if_addrhead */
1104	IF_ADDR_WLOCK(ifp);
1105	CK_STAILQ_INSERT_TAIL(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
1106	IF_ADDR_WUNLOCK(ifp);
1107
1108	ifa_ref(&ia->ia_ifa);			/* in6_ifaddrhead */
1109	IN6_IFADDR_WLOCK();
1110	CK_STAILQ_INSERT_TAIL(&V_in6_ifaddrhead, ia, ia_link);
1111	CK_LIST_INSERT_HEAD(IN6ADDR_HASH(&ia->ia_addr.sin6_addr), ia, ia6_hash);
1112	IN6_IFADDR_WUNLOCK();
1113
1114	return (ia);
1115}
1116
1117/*
1118 * Update/configure interface address parameters:
1119 *
1120 * 1) Update lifetime
1121 * 2) Update interface metric ad flags
1122 * 3) Notify other subsystems
1123 */
1124static int
1125in6_update_ifa_internal(struct ifnet *ifp, struct in6_aliasreq *ifra,
1126    struct in6_ifaddr *ia, int hostIsNew, int flags)
1127{
1128	int error;
1129
1130	/* update timestamp */
1131	ia->ia6_updatetime = time_uptime;
1132
1133	/*
1134	 * Set lifetimes.  We do not refer to ia6t_expire and ia6t_preferred
1135	 * to see if the address is deprecated or invalidated, but initialize
1136	 * these members for applications.
1137	 */
1138	ia->ia6_lifetime = ifra->ifra_lifetime;
1139	if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
1140		ia->ia6_lifetime.ia6t_expire =
1141		    time_uptime + ia->ia6_lifetime.ia6t_vltime;
1142	} else
1143		ia->ia6_lifetime.ia6t_expire = 0;
1144	if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
1145		ia->ia6_lifetime.ia6t_preferred =
1146		    time_uptime + ia->ia6_lifetime.ia6t_pltime;
1147	} else
1148		ia->ia6_lifetime.ia6t_preferred = 0;
1149
1150	/*
1151	 * backward compatibility - if IN6_IFF_DEPRECATED is set from the
1152	 * userland, make it deprecated.
1153	 */
1154	if ((ifra->ifra_flags & IN6_IFF_DEPRECATED) != 0) {
1155		ia->ia6_lifetime.ia6t_pltime = 0;
1156		ia->ia6_lifetime.ia6t_preferred = time_uptime;
1157	}
1158
1159	/*
1160	 * configure address flags.
1161	 */
1162	ia->ia6_flags = ifra->ifra_flags;
1163
1164	/*
1165	 * Make the address tentative before joining multicast addresses,
1166	 * so that corresponding MLD responses would not have a tentative
1167	 * source address.
1168	 */
1169	ia->ia6_flags &= ~IN6_IFF_DUPLICATED;	/* safety */
1170
1171	/*
1172	 * DAD should be performed for an new address or addresses on
1173	 * an interface with ND6_IFF_IFDISABLED.
1174	 */
1175	if (in6if_do_dad(ifp) &&
1176	    (hostIsNew || (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)))
1177		ia->ia6_flags |= IN6_IFF_TENTATIVE;
1178
1179	/* notify other subsystems */
1180	error = in6_notify_ifa(ifp, ia, ifra, hostIsNew);
1181
1182	return (error);
1183}
1184
1185/*
1186 * Do link-level ifa job:
1187 * 1) Add lle entry for added address
1188 * 2) Notifies routing socket users about new address
1189 * 3) join appropriate multicast group
1190 * 4) start DAD if enabled
1191 */
1192static int
1193in6_broadcast_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
1194    struct in6_ifaddr *ia, int flags)
1195{
1196	struct in6_multi *in6m_sol;
1197	int error = 0;
1198
1199	/* Add local address to lltable, if necessary (ex. on p2p link). */
1200	if ((error = nd6_add_ifa_lle(ia)) != 0) {
1201		in6_purgeaddr(&ia->ia_ifa);
1202		ifa_free(&ia->ia_ifa);
1203		return (error);
1204	}
1205
1206	/* Join necessary multicast groups. */
1207	in6m_sol = NULL;
1208	if ((ifp->if_flags & IFF_MULTICAST) != 0) {
1209		error = in6_update_ifa_join_mc(ifp, ifra, ia, flags, &in6m_sol);
1210		if (error != 0) {
1211			in6_purgeaddr(&ia->ia_ifa);
1212			ifa_free(&ia->ia_ifa);
1213			return (error);
1214		}
1215	}
1216
1217	/* Perform DAD, if the address is TENTATIVE. */
1218	if ((ia->ia6_flags & IN6_IFF_TENTATIVE)) {
1219		int delay, mindelay, maxdelay;
1220
1221		delay = 0;
1222		if ((flags & IN6_IFAUPDATE_DADDELAY)) {
1223			/*
1224			 * We need to impose a delay before sending an NS
1225			 * for DAD.  Check if we also needed a delay for the
1226			 * corresponding MLD message.  If we did, the delay
1227			 * should be larger than the MLD delay (this could be
1228			 * relaxed a bit, but this simple logic is at least
1229			 * safe).
1230			 * XXX: Break data hiding guidelines and look at
1231			 * state for the solicited multicast group.
1232			 */
1233			mindelay = 0;
1234			if (in6m_sol != NULL &&
1235			    in6m_sol->in6m_state == MLD_REPORTING_MEMBER) {
1236				mindelay = in6m_sol->in6m_timer;
1237			}
1238			maxdelay = MAX_RTR_SOLICITATION_DELAY * hz;
1239			if (maxdelay - mindelay == 0)
1240				delay = 0;
1241			else {
1242				delay =
1243				    (arc4random() % (maxdelay - mindelay)) +
1244				    mindelay;
1245			}
1246		}
1247		nd6_dad_start((struct ifaddr *)ia, delay);
1248	}
1249
1250	in6_newaddrmsg(ia, RTM_ADD);
1251	ifa_free(&ia->ia_ifa);
1252	return (error);
1253}
1254
1255/*
1256 * Adds or deletes interface route for p2p ifa.
1257 * Returns 0 on success or errno.
1258 */
1259static int
1260in6_handle_dstaddr_rtrequest(int cmd, struct in6_ifaddr *ia)
1261{
1262	struct epoch_tracker et;
1263	struct ifaddr *ifa = &ia->ia_ifa;
1264	int error;
1265
1266	/* Prepare gateway */
1267	struct sockaddr_dl_short sdl = {
1268		.sdl_family = AF_LINK,
1269		.sdl_len = sizeof(struct sockaddr_dl_short),
1270		.sdl_type = ifa->ifa_ifp->if_type,
1271		.sdl_index = ifa->ifa_ifp->if_index,
1272	};
1273
1274	struct sockaddr_in6 dst = {
1275		.sin6_family = AF_INET6,
1276		.sin6_len = sizeof(struct sockaddr_in6),
1277		.sin6_addr = ia->ia_dstaddr.sin6_addr,
1278	};
1279
1280	struct rt_addrinfo info = {
1281		.rti_ifa = ifa,
1282		.rti_flags = RTF_PINNED | RTF_HOST,
1283		.rti_info = {
1284			[RTAX_DST] = (struct sockaddr *)&dst,
1285			[RTAX_GATEWAY] = (struct sockaddr *)&sdl,
1286		},
1287	};
1288	/* Don't set additional per-gw filters on removal */
1289
1290	NET_EPOCH_ENTER(et);
1291	error = rib_handle_ifaddr_info(ifa->ifa_ifp->if_fib, cmd, &info);
1292	NET_EPOCH_EXIT(et);
1293
1294	return (error);
1295}
1296
1297static bool
1298ifa_is_p2p(struct in6_ifaddr *ia)
1299{
1300	int plen;
1301
1302	plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */
1303
1304	if ((plen == 128) && (ia->ia_dstaddr.sin6_family == AF_INET6) &&
1305	    !IN6_ARE_ADDR_EQUAL(&ia->ia_addr.sin6_addr, &ia->ia_dstaddr.sin6_addr))
1306		return (true);
1307
1308	return (false);
1309}
1310
1311void
1312in6_purgeaddr(struct ifaddr *ifa)
1313{
1314	struct ifnet *ifp = ifa->ifa_ifp;
1315	struct in6_ifaddr *ia = (struct in6_ifaddr *) ifa;
1316	struct in6_multi_mship *imm;
1317	int error;
1318
1319	if (ifa->ifa_carp)
1320		(*carp_detach_p)(ifa, false);
1321
1322	/*
1323	 * Remove the loopback route to the interface address.
1324	 * The check for the current setting of "nd6_useloopback"
1325	 * is not needed.
1326	 */
1327	if (ia->ia_flags & IFA_RTSELF) {
1328		error = ifa_del_loopback_route((struct ifaddr *)ia,
1329		    (struct sockaddr *)&ia->ia_addr);
1330		if (error == 0)
1331			ia->ia_flags &= ~IFA_RTSELF;
1332	}
1333
1334	/* stop DAD processing */
1335	nd6_dad_stop(ifa);
1336
1337	/* Leave multicast groups. */
1338	while ((imm = LIST_FIRST(&ia->ia6_memberships)) != NULL) {
1339		LIST_REMOVE(imm, i6mm_chain);
1340		if (imm->i6mm_maddr != NULL)
1341			in6_leavegroup(imm->i6mm_maddr, NULL);
1342		free(imm, M_IP6MADDR);
1343	}
1344	/* Check if we need to remove p2p route */
1345	if ((ia->ia_flags & IFA_ROUTE) && ifa_is_p2p(ia)) {
1346		error = in6_handle_dstaddr_rtrequest(RTM_DELETE, ia);
1347		if (error != 0)
1348			log(LOG_INFO, "%s: err=%d, destination address delete "
1349			    "failed\n", __func__, error);
1350		ia->ia_flags &= ~IFA_ROUTE;
1351	}
1352
1353	in6_newaddrmsg(ia, RTM_DELETE);
1354	in6_unlink_ifa(ia, ifp);
1355}
1356
1357/*
1358 * Removes @ia from the corresponding interfaces and unlinks corresponding
1359 *  prefix if no addresses are using it anymore.
1360 */
1361void
1362in6_purgeifaddr(struct in6_ifaddr *ia)
1363{
1364	struct nd_prefix *pr;
1365
1366	/*
1367	 * If the address being deleted is the only one that owns
1368	 * the corresponding prefix, expire the prefix as well.
1369	 * XXX: theoretically, we don't have to worry about such
1370	 * relationship, since we separate the address management
1371	 * and the prefix management.  We do this, however, to provide
1372	 * as much backward compatibility as possible in terms of
1373	 * the ioctl operation.
1374	 * Note that in6_purgeaddr() will decrement ndpr_addrcnt.
1375	 */
1376	pr = ia->ia6_ndpr;
1377	in6_purgeaddr(&ia->ia_ifa);
1378	if (pr != NULL && pr->ndpr_addrcnt == 0) {
1379		ND6_WLOCK();
1380		nd6_prefix_unlink(pr, NULL);
1381		ND6_WUNLOCK();
1382		nd6_prefix_del(pr);
1383	}
1384}
1385
1386
1387static void
1388in6_unlink_ifa(struct in6_ifaddr *ia, struct ifnet *ifp)
1389{
1390	char ip6buf[INET6_ADDRSTRLEN];
1391	int remove_lle;
1392
1393	IF_ADDR_WLOCK(ifp);
1394	CK_STAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifaddr, ifa_link);
1395	IF_ADDR_WUNLOCK(ifp);
1396	ifa_free(&ia->ia_ifa);			/* if_addrhead */
1397
1398	/*
1399	 * Defer the release of what might be the last reference to the
1400	 * in6_ifaddr so that it can't be freed before the remainder of the
1401	 * cleanup.
1402	 */
1403	IN6_IFADDR_WLOCK();
1404	CK_STAILQ_REMOVE(&V_in6_ifaddrhead, ia, in6_ifaddr, ia_link);
1405	CK_LIST_REMOVE(ia, ia6_hash);
1406	IN6_IFADDR_WUNLOCK();
1407
1408	/*
1409	 * Release the reference to the base prefix.  There should be a
1410	 * positive reference.
1411	 */
1412	remove_lle = 0;
1413	if (ia->ia6_ndpr == NULL) {
1414		nd6log((LOG_NOTICE,
1415		    "in6_unlink_ifa: autoconf'ed address "
1416		    "%s has no prefix\n", ip6_sprintf(ip6buf, IA6_IN6(ia))));
1417	} else {
1418		ia->ia6_ndpr->ndpr_addrcnt--;
1419		/* Do not delete lles within prefix if refcont != 0 */
1420		if (ia->ia6_ndpr->ndpr_addrcnt == 0)
1421			remove_lle = 1;
1422		ia->ia6_ndpr = NULL;
1423	}
1424
1425	nd6_rem_ifa_lle(ia, remove_lle);
1426
1427	/*
1428	 * Also, if the address being removed is autoconf'ed, call
1429	 * pfxlist_onlink_check() since the release might affect the status of
1430	 * other (detached) addresses.
1431	 */
1432	if ((ia->ia6_flags & IN6_IFF_AUTOCONF)) {
1433		pfxlist_onlink_check();
1434	}
1435	ifa_free(&ia->ia_ifa);			/* in6_ifaddrhead */
1436}
1437
1438/*
1439 * Notifies other subsystems about address change/arrival:
1440 * 1) Notifies device handler on the first IPv6 address assignment
1441 * 2) Handle routing table changes for P2P links and route
1442 * 3) Handle routing table changes for address host route
1443 */
1444static int
1445in6_notify_ifa(struct ifnet *ifp, struct in6_ifaddr *ia,
1446    struct in6_aliasreq *ifra, int hostIsNew)
1447{
1448	int	error = 0, ifacount = 0;
1449	struct ifaddr *ifa;
1450	struct sockaddr_in6 *pdst;
1451	char ip6buf[INET6_ADDRSTRLEN];
1452
1453	/*
1454	 * Give the interface a chance to initialize
1455	 * if this is its first address,
1456	 */
1457	if (hostIsNew != 0) {
1458		struct epoch_tracker et;
1459
1460		NET_EPOCH_ENTER(et);
1461		CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1462			if (ifa->ifa_addr->sa_family != AF_INET6)
1463				continue;
1464			ifacount++;
1465		}
1466		NET_EPOCH_EXIT(et);
1467	}
1468
1469	if (ifacount <= 1 && ifp->if_ioctl) {
1470		error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
1471		if (error)
1472			goto done;
1473	}
1474
1475	/*
1476	 * If a new destination address is specified, scrub the old one and
1477	 * install the new destination.  Note that the interface must be
1478	 * p2p or loopback.
1479	 */
1480	pdst = &ifra->ifra_dstaddr;
1481	if (pdst->sin6_family == AF_INET6 &&
1482	    !IN6_ARE_ADDR_EQUAL(&pdst->sin6_addr, &ia->ia_dstaddr.sin6_addr)) {
1483		if ((ia->ia_flags & IFA_ROUTE) != 0 &&
1484		    (in6_handle_dstaddr_rtrequest(RTM_DELETE, ia) != 0)) {
1485			nd6log((LOG_ERR, "in6_update_ifa_internal: failed to "
1486			    "remove a route to the old destination: %s\n",
1487			    ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
1488			/* proceed anyway... */
1489		} else
1490			ia->ia_flags &= ~IFA_ROUTE;
1491		ia->ia_dstaddr = *pdst;
1492	}
1493
1494	/*
1495	 * If a new destination address is specified for a point-to-point
1496	 * interface, install a route to the destination as an interface
1497	 * direct route.
1498	 * XXX: the logic below rejects assigning multiple addresses on a p2p
1499	 * interface that share the same destination.
1500	 */
1501	if (!(ia->ia_flags & IFA_ROUTE) && ifa_is_p2p(ia)) {
1502		error = in6_handle_dstaddr_rtrequest(RTM_ADD, ia);
1503		if (error)
1504			goto done;
1505		ia->ia_flags |= IFA_ROUTE;
1506	}
1507
1508	/*
1509	 * add a loopback route to self if not exists
1510	 */
1511	if (!(ia->ia_flags & IFA_RTSELF) && V_nd6_useloopback) {
1512		error = ifa_add_loopback_route((struct ifaddr *)ia,
1513		    (struct sockaddr *)&ia->ia_addr);
1514		if (error == 0)
1515			ia->ia_flags |= IFA_RTSELF;
1516	}
1517done:
1518	WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
1519	    "Invoking IPv6 network device address event may sleep");
1520
1521	ifa_ref(&ia->ia_ifa);
1522	EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa,
1523	    IFADDR_EVENT_ADD);
1524	ifa_free(&ia->ia_ifa);
1525
1526	return (error);
1527}
1528
1529/*
1530 * Find an IPv6 interface link-local address specific to an interface.
1531 * ifaddr is returned referenced.
1532 */
1533struct in6_ifaddr *
1534in6ifa_ifpforlinklocal(struct ifnet *ifp, int ignoreflags)
1535{
1536	struct ifaddr *ifa;
1537
1538	NET_EPOCH_ASSERT();
1539
1540	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1541		if (ifa->ifa_addr->sa_family != AF_INET6)
1542			continue;
1543		if (IN6_IS_ADDR_LINKLOCAL(IFA_IN6(ifa))) {
1544			if ((((struct in6_ifaddr *)ifa)->ia6_flags &
1545			    ignoreflags) != 0)
1546				continue;
1547			ifa_ref(ifa);
1548			break;
1549		}
1550	}
1551
1552	return ((struct in6_ifaddr *)ifa);
1553}
1554
1555/*
1556 * find the interface address corresponding to a given IPv6 address.
1557 * ifaddr is returned referenced if @referenced flag is set.
1558 */
1559struct in6_ifaddr *
1560in6ifa_ifwithaddr(const struct in6_addr *addr, uint32_t zoneid, bool referenced)
1561{
1562	struct rm_priotracker in6_ifa_tracker;
1563	struct in6_ifaddr *ia;
1564
1565	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1566	CK_LIST_FOREACH(ia, IN6ADDR_HASH(addr), ia6_hash) {
1567		if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), addr)) {
1568			if (zoneid != 0 &&
1569			    zoneid != ia->ia_addr.sin6_scope_id)
1570				continue;
1571			if (referenced)
1572				ifa_ref(&ia->ia_ifa);
1573			break;
1574		}
1575	}
1576	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1577	return (ia);
1578}
1579
1580/*
1581 * find the internet address corresponding to a given interface and address.
1582 * ifaddr is returned referenced.
1583 */
1584struct in6_ifaddr *
1585in6ifa_ifpwithaddr(struct ifnet *ifp, const struct in6_addr *addr)
1586{
1587	struct epoch_tracker et;
1588	struct ifaddr *ifa;
1589
1590	NET_EPOCH_ENTER(et);
1591	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1592		if (ifa->ifa_addr->sa_family != AF_INET6)
1593			continue;
1594		if (IN6_ARE_ADDR_EQUAL(addr, IFA_IN6(ifa))) {
1595			ifa_ref(ifa);
1596			break;
1597		}
1598	}
1599	NET_EPOCH_EXIT(et);
1600
1601	return ((struct in6_ifaddr *)ifa);
1602}
1603
1604/*
1605 * Find a link-local scoped address on ifp and return it if any.
1606 */
1607struct in6_ifaddr *
1608in6ifa_llaonifp(struct ifnet *ifp)
1609{
1610	struct epoch_tracker et;
1611	struct sockaddr_in6 *sin6;
1612	struct ifaddr *ifa;
1613
1614	if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)
1615		return (NULL);
1616	NET_EPOCH_ENTER(et);
1617	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1618		if (ifa->ifa_addr->sa_family != AF_INET6)
1619			continue;
1620		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
1621		if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr) ||
1622		    IN6_IS_ADDR_MC_INTFACELOCAL(&sin6->sin6_addr) ||
1623		    IN6_IS_ADDR_MC_NODELOCAL(&sin6->sin6_addr))
1624			break;
1625	}
1626	NET_EPOCH_EXIT(et);
1627
1628	return ((struct in6_ifaddr *)ifa);
1629}
1630
1631/*
1632 * Convert IP6 address to printable (loggable) representation. Caller
1633 * has to make sure that ip6buf is at least INET6_ADDRSTRLEN long.
1634 */
1635static char digits[] = "0123456789abcdef";
1636char *
1637ip6_sprintf(char *ip6buf, const struct in6_addr *addr)
1638{
1639	int i, cnt = 0, maxcnt = 0, idx = 0, index = 0;
1640	char *cp;
1641	const u_int16_t *a = (const u_int16_t *)addr;
1642	const u_int8_t *d;
1643	int dcolon = 0, zero = 0;
1644
1645	cp = ip6buf;
1646
1647	for (i = 0; i < 8; i++) {
1648		if (*(a + i) == 0) {
1649			cnt++;
1650			if (cnt == 1)
1651				idx = i;
1652		}
1653		else if (maxcnt < cnt) {
1654			maxcnt = cnt;
1655			index = idx;
1656			cnt = 0;
1657		}
1658	}
1659	if (maxcnt < cnt) {
1660		maxcnt = cnt;
1661		index = idx;
1662	}
1663
1664	for (i = 0; i < 8; i++) {
1665		if (dcolon == 1) {
1666			if (*a == 0) {
1667				if (i == 7)
1668					*cp++ = ':';
1669				a++;
1670				continue;
1671			} else
1672				dcolon = 2;
1673		}
1674		if (*a == 0) {
1675			if (dcolon == 0 && *(a + 1) == 0 && i == index) {
1676				if (i == 0)
1677					*cp++ = ':';
1678				*cp++ = ':';
1679				dcolon = 1;
1680			} else {
1681				*cp++ = '0';
1682				*cp++ = ':';
1683			}
1684			a++;
1685			continue;
1686		}
1687		d = (const u_char *)a;
1688		/* Try to eliminate leading zeros in printout like in :0001. */
1689		zero = 1;
1690		*cp = digits[*d >> 4];
1691		if (*cp != '0') {
1692			zero = 0;
1693			cp++;
1694		}
1695		*cp = digits[*d++ & 0xf];
1696		if (zero == 0 || (*cp != '0')) {
1697			zero = 0;
1698			cp++;
1699		}
1700		*cp = digits[*d >> 4];
1701		if (zero == 0 || (*cp != '0')) {
1702			zero = 0;
1703			cp++;
1704		}
1705		*cp++ = digits[*d & 0xf];
1706		*cp++ = ':';
1707		a++;
1708	}
1709	*--cp = '\0';
1710	return (ip6buf);
1711}
1712
1713int
1714in6_localaddr(struct in6_addr *in6)
1715{
1716	struct rm_priotracker in6_ifa_tracker;
1717	struct in6_ifaddr *ia;
1718
1719	if (IN6_IS_ADDR_LOOPBACK(in6) || IN6_IS_ADDR_LINKLOCAL(in6))
1720		return 1;
1721
1722	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1723	CK_STAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) {
1724		if (IN6_ARE_MASKED_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr,
1725		    &ia->ia_prefixmask.sin6_addr)) {
1726			IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1727			return 1;
1728		}
1729	}
1730	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1731
1732	return (0);
1733}
1734
1735/*
1736 * Return 1 if an internet address is for the local host and configured
1737 * on one of its interfaces.
1738 */
1739int
1740in6_localip(struct in6_addr *in6)
1741{
1742	struct rm_priotracker in6_ifa_tracker;
1743	struct in6_ifaddr *ia;
1744
1745	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1746	CK_LIST_FOREACH(ia, IN6ADDR_HASH(in6), ia6_hash) {
1747		if (IN6_ARE_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr)) {
1748			IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1749			return (1);
1750		}
1751	}
1752	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1753	return (0);
1754}
1755
1756/*
1757 * Return 1 if an internet address is configured on an interface.
1758 */
1759int
1760in6_ifhasaddr(struct ifnet *ifp, struct in6_addr *addr)
1761{
1762	struct in6_addr in6;
1763	struct ifaddr *ifa;
1764	struct in6_ifaddr *ia6;
1765
1766	NET_EPOCH_ASSERT();
1767
1768	in6 = *addr;
1769	if (in6_clearscope(&in6))
1770		return (0);
1771	in6_setscope(&in6, ifp, NULL);
1772
1773	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1774		if (ifa->ifa_addr->sa_family != AF_INET6)
1775			continue;
1776		ia6 = (struct in6_ifaddr *)ifa;
1777		if (IN6_ARE_ADDR_EQUAL(&ia6->ia_addr.sin6_addr, &in6))
1778			return (1);
1779	}
1780
1781	return (0);
1782}
1783
1784int
1785in6_is_addr_deprecated(struct sockaddr_in6 *sa6)
1786{
1787	struct rm_priotracker in6_ifa_tracker;
1788	struct in6_ifaddr *ia;
1789
1790	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1791	CK_LIST_FOREACH(ia, IN6ADDR_HASH(&sa6->sin6_addr), ia6_hash) {
1792		if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), &sa6->sin6_addr)) {
1793			if (ia->ia6_flags & IN6_IFF_DEPRECATED) {
1794				IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1795				return (1); /* true */
1796			}
1797			break;
1798		}
1799	}
1800	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1801
1802	return (0);		/* false */
1803}
1804
1805/*
1806 * return length of part which dst and src are equal
1807 * hard coding...
1808 */
1809int
1810in6_matchlen(struct in6_addr *src, struct in6_addr *dst)
1811{
1812	int match = 0;
1813	u_char *s = (u_char *)src, *d = (u_char *)dst;
1814	u_char *lim = s + 16, r;
1815
1816	while (s < lim)
1817		if ((r = (*d++ ^ *s++)) != 0) {
1818			while (r < 128) {
1819				match++;
1820				r <<= 1;
1821			}
1822			break;
1823		} else
1824			match += 8;
1825	return match;
1826}
1827
1828/* XXX: to be scope conscious */
1829int
1830in6_are_prefix_equal(struct in6_addr *p1, struct in6_addr *p2, int len)
1831{
1832	int bytelen, bitlen;
1833
1834	/* sanity check */
1835	if (0 > len || len > 128) {
1836		log(LOG_ERR, "in6_are_prefix_equal: invalid prefix length(%d)\n",
1837		    len);
1838		return (0);
1839	}
1840
1841	bytelen = len / 8;
1842	bitlen = len % 8;
1843
1844	if (bcmp(&p1->s6_addr, &p2->s6_addr, bytelen))
1845		return (0);
1846	if (bitlen != 0 &&
1847	    p1->s6_addr[bytelen] >> (8 - bitlen) !=
1848	    p2->s6_addr[bytelen] >> (8 - bitlen))
1849		return (0);
1850
1851	return (1);
1852}
1853
1854void
1855in6_prefixlen2mask(struct in6_addr *maskp, int len)
1856{
1857	u_char maskarray[8] = {0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, 0xff};
1858	int bytelen, bitlen, i;
1859
1860	/* sanity check */
1861	if (0 > len || len > 128) {
1862		log(LOG_ERR, "in6_prefixlen2mask: invalid prefix length(%d)\n",
1863		    len);
1864		return;
1865	}
1866
1867	bzero(maskp, sizeof(*maskp));
1868	bytelen = len / 8;
1869	bitlen = len % 8;
1870	for (i = 0; i < bytelen; i++)
1871		maskp->s6_addr[i] = 0xff;
1872	if (bitlen)
1873		maskp->s6_addr[bytelen] = maskarray[bitlen - 1];
1874}
1875
1876/*
1877 * return the best address out of the same scope. if no address was
1878 * found, return the first valid address from designated IF.
1879 */
1880struct in6_ifaddr *
1881in6_ifawithifp(struct ifnet *ifp, struct in6_addr *dst)
1882{
1883	int dst_scope =	in6_addrscope(dst), blen = -1, tlen;
1884	struct ifaddr *ifa;
1885	struct in6_ifaddr *besta = NULL;
1886	struct in6_ifaddr *dep[2];	/* last-resort: deprecated */
1887
1888	NET_EPOCH_ASSERT();
1889
1890	dep[0] = dep[1] = NULL;
1891
1892	/*
1893	 * We first look for addresses in the same scope.
1894	 * If there is one, return it.
1895	 * If two or more, return one which matches the dst longest.
1896	 * If none, return one of global addresses assigned other ifs.
1897	 */
1898	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1899		if (ifa->ifa_addr->sa_family != AF_INET6)
1900			continue;
1901		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
1902			continue; /* XXX: is there any case to allow anycast? */
1903		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
1904			continue; /* don't use this interface */
1905		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
1906			continue;
1907		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
1908			if (V_ip6_use_deprecated)
1909				dep[0] = (struct in6_ifaddr *)ifa;
1910			continue;
1911		}
1912
1913		if (dst_scope == in6_addrscope(IFA_IN6(ifa))) {
1914			/*
1915			 * call in6_matchlen() as few as possible
1916			 */
1917			if (besta) {
1918				if (blen == -1)
1919					blen = in6_matchlen(&besta->ia_addr.sin6_addr, dst);
1920				tlen = in6_matchlen(IFA_IN6(ifa), dst);
1921				if (tlen > blen) {
1922					blen = tlen;
1923					besta = (struct in6_ifaddr *)ifa;
1924				}
1925			} else
1926				besta = (struct in6_ifaddr *)ifa;
1927		}
1928	}
1929	if (besta)
1930		return (besta);
1931
1932	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1933		if (ifa->ifa_addr->sa_family != AF_INET6)
1934			continue;
1935		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
1936			continue; /* XXX: is there any case to allow anycast? */
1937		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
1938			continue; /* don't use this interface */
1939		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
1940			continue;
1941		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
1942			if (V_ip6_use_deprecated)
1943				dep[1] = (struct in6_ifaddr *)ifa;
1944			continue;
1945		}
1946
1947		return (struct in6_ifaddr *)ifa;
1948	}
1949
1950	/* use the last-resort values, that are, deprecated addresses */
1951	if (dep[0])
1952		return dep[0];
1953	if (dep[1])
1954		return dep[1];
1955
1956	return NULL;
1957}
1958
1959/*
1960 * perform DAD when interface becomes IFF_UP.
1961 */
1962void
1963in6_if_up(struct ifnet *ifp)
1964{
1965	struct epoch_tracker et;
1966	struct ifaddr *ifa;
1967	struct in6_ifaddr *ia;
1968
1969	NET_EPOCH_ENTER(et);
1970	CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1971		if (ifa->ifa_addr->sa_family != AF_INET6)
1972			continue;
1973		ia = (struct in6_ifaddr *)ifa;
1974		if (ia->ia6_flags & IN6_IFF_TENTATIVE) {
1975			/*
1976			 * The TENTATIVE flag was likely set by hand
1977			 * beforehand, implicitly indicating the need for DAD.
1978			 * We may be able to skip the random delay in this
1979			 * case, but we impose delays just in case.
1980			 */
1981			nd6_dad_start(ifa,
1982			    arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz));
1983		}
1984	}
1985	NET_EPOCH_EXIT(et);
1986
1987	/*
1988	 * special cases, like 6to4, are handled in in6_ifattach
1989	 */
1990	in6_ifattach(ifp, NULL);
1991}
1992
1993int
1994in6if_do_dad(struct ifnet *ifp)
1995{
1996
1997	if ((ifp->if_flags & IFF_LOOPBACK) != 0)
1998		return (0);
1999	if ((ifp->if_flags & IFF_MULTICAST) == 0)
2000		return (0);
2001	if ((ND_IFINFO(ifp)->flags &
2002	    (ND6_IFF_IFDISABLED | ND6_IFF_NO_DAD)) != 0)
2003		return (0);
2004	return (1);
2005}
2006
2007/*
2008 * Calculate max IPv6 MTU through all the interfaces and store it
2009 * to in6_maxmtu.
2010 */
2011void
2012in6_setmaxmtu(void)
2013{
2014	struct epoch_tracker et;
2015	unsigned long maxmtu = 0;
2016	struct ifnet *ifp;
2017
2018	NET_EPOCH_ENTER(et);
2019	CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
2020		/* this function can be called during ifnet initialization */
2021		if (!ifp->if_afdata[AF_INET6])
2022			continue;
2023		if ((ifp->if_flags & IFF_LOOPBACK) == 0 &&
2024		    IN6_LINKMTU(ifp) > maxmtu)
2025			maxmtu = IN6_LINKMTU(ifp);
2026	}
2027	NET_EPOCH_EXIT(et);
2028	if (maxmtu)	/* update only when maxmtu is positive */
2029		V_in6_maxmtu = maxmtu;
2030}
2031
2032/*
2033 * Provide the length of interface identifiers to be used for the link attached
2034 * to the given interface.  The length should be defined in "IPv6 over
2035 * xxx-link" document.  Note that address architecture might also define
2036 * the length for a particular set of address prefixes, regardless of the
2037 * link type.  As clarified in rfc2462bis, those two definitions should be
2038 * consistent, and those really are as of August 2004.
2039 */
2040int
2041in6_if2idlen(struct ifnet *ifp)
2042{
2043	switch (ifp->if_type) {
2044	case IFT_ETHER:		/* RFC2464 */
2045	case IFT_PROPVIRTUAL:	/* XXX: no RFC. treat it as ether */
2046	case IFT_L2VLAN:	/* ditto */
2047	case IFT_BRIDGE:	/* bridge(4) only does Ethernet-like links */
2048	case IFT_INFINIBAND:
2049		return (64);
2050	case IFT_PPP:		/* RFC2472 */
2051		return (64);
2052	case IFT_FRELAY:	/* RFC2590 */
2053		return (64);
2054	case IFT_IEEE1394:	/* RFC3146 */
2055		return (64);
2056	case IFT_GIF:
2057		return (64);	/* draft-ietf-v6ops-mech-v2-07 */
2058	case IFT_LOOP:
2059		return (64);	/* XXX: is this really correct? */
2060	default:
2061		/*
2062		 * Unknown link type:
2063		 * It might be controversial to use the today's common constant
2064		 * of 64 for these cases unconditionally.  For full compliance,
2065		 * we should return an error in this case.  On the other hand,
2066		 * if we simply miss the standard for the link type or a new
2067		 * standard is defined for a new link type, the IFID length
2068		 * is very likely to be the common constant.  As a compromise,
2069		 * we always use the constant, but make an explicit notice
2070		 * indicating the "unknown" case.
2071		 */
2072		printf("in6_if2idlen: unknown link type (%d)\n", ifp->if_type);
2073		return (64);
2074	}
2075}
2076
2077struct in6_llentry {
2078	struct llentry		base;
2079};
2080
2081#define	IN6_LLTBL_DEFAULT_HSIZE	32
2082#define	IN6_LLTBL_HASH(k, h) \
2083	(((((((k >> 8) ^ k) >> 8) ^ k) >> 8) ^ k) & ((h) - 1))
2084
2085/*
2086 * Do actual deallocation of @lle.
2087 */
2088static void
2089in6_lltable_destroy_lle_unlocked(epoch_context_t ctx)
2090{
2091	struct llentry *lle;
2092
2093	lle = __containerof(ctx, struct llentry, lle_epoch_ctx);
2094	LLE_LOCK_DESTROY(lle);
2095	LLE_REQ_DESTROY(lle);
2096	free(lle, M_LLTABLE);
2097}
2098
2099/*
2100 * Called by LLE_FREE_LOCKED when number of references
2101 * drops to zero.
2102 */
2103static void
2104in6_lltable_destroy_lle(struct llentry *lle)
2105{
2106
2107	LLE_WUNLOCK(lle);
2108	NET_EPOCH_CALL(in6_lltable_destroy_lle_unlocked, &lle->lle_epoch_ctx);
2109}
2110
2111static struct llentry *
2112in6_lltable_new(const struct in6_addr *addr6, u_int flags)
2113{
2114	struct in6_llentry *lle;
2115
2116	lle = malloc(sizeof(struct in6_llentry), M_LLTABLE, M_NOWAIT | M_ZERO);
2117	if (lle == NULL)		/* NB: caller generates msg */
2118		return NULL;
2119
2120	lle->base.r_l3addr.addr6 = *addr6;
2121	lle->base.lle_refcnt = 1;
2122	lle->base.lle_free = in6_lltable_destroy_lle;
2123	LLE_LOCK_INIT(&lle->base);
2124	LLE_REQ_INIT(&lle->base);
2125	callout_init(&lle->base.lle_timer, 1);
2126
2127	return (&lle->base);
2128}
2129
2130static int
2131in6_lltable_match_prefix(const struct sockaddr *saddr,
2132    const struct sockaddr *smask, u_int flags, struct llentry *lle)
2133{
2134	const struct in6_addr *addr, *mask, *lle_addr;
2135
2136	addr = &((const struct sockaddr_in6 *)saddr)->sin6_addr;
2137	mask = &((const struct sockaddr_in6 *)smask)->sin6_addr;
2138	lle_addr = &lle->r_l3addr.addr6;
2139
2140	if (IN6_ARE_MASKED_ADDR_EQUAL(lle_addr, addr, mask) == 0)
2141		return (0);
2142
2143	if (lle->la_flags & LLE_IFADDR) {
2144		/*
2145		 * Delete LLE_IFADDR records IFF address & flag matches.
2146		 * Note that addr is the interface address within prefix
2147		 * being matched.
2148		 */
2149		if (IN6_ARE_ADDR_EQUAL(addr, lle_addr) &&
2150		    (flags & LLE_STATIC) != 0)
2151			return (1);
2152		return (0);
2153	}
2154
2155	/* flags & LLE_STATIC means deleting both dynamic and static entries */
2156	if ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))
2157		return (1);
2158
2159	return (0);
2160}
2161
2162static void
2163in6_lltable_free_entry(struct lltable *llt, struct llentry *lle)
2164{
2165	struct ifnet *ifp;
2166
2167	LLE_WLOCK_ASSERT(lle);
2168	KASSERT(llt != NULL, ("lltable is NULL"));
2169
2170	/* Unlink entry from table */
2171	if ((lle->la_flags & LLE_LINKED) != 0) {
2172		ifp = llt->llt_ifp;
2173		IF_AFDATA_WLOCK_ASSERT(ifp);
2174		lltable_unlink_entry(llt, lle);
2175	}
2176
2177	llentry_free(lle);
2178}
2179
2180static int
2181in6_lltable_rtcheck(struct ifnet *ifp,
2182		    u_int flags,
2183		    const struct sockaddr *l3addr)
2184{
2185	const struct sockaddr_in6 *sin6;
2186	struct nhop_object *nh;
2187	struct in6_addr dst;
2188	uint32_t scopeid;
2189	char ip6buf[INET6_ADDRSTRLEN];
2190	int fibnum;
2191
2192	NET_EPOCH_ASSERT();
2193	KASSERT(l3addr->sa_family == AF_INET6,
2194	    ("sin_family %d", l3addr->sa_family));
2195
2196	sin6 = (const struct sockaddr_in6 *)l3addr;
2197	in6_splitscope(&sin6->sin6_addr, &dst, &scopeid);
2198	fibnum = V_rt_add_addr_allfibs ? RT_DEFAULT_FIB : ifp->if_fib;
2199	nh = fib6_lookup(fibnum, &dst, scopeid, NHR_NONE, 0);
2200	if (nh && ((nh->nh_flags & NHF_GATEWAY) || nh->nh_ifp != ifp)) {
2201		struct ifaddr *ifa;
2202		/*
2203		 * Create an ND6 cache for an IPv6 neighbor
2204		 * that is not covered by our own prefix.
2205		 */
2206		ifa = ifaof_ifpforaddr(l3addr, ifp);
2207		if (ifa != NULL) {
2208			return 0;
2209		}
2210		log(LOG_INFO, "IPv6 address: \"%s\" is not on the network\n",
2211		    ip6_sprintf(ip6buf, &sin6->sin6_addr));
2212		return EINVAL;
2213	}
2214	return 0;
2215}
2216
2217/*
2218 * Called by the datapath to indicate that the entry was used.
2219 */
2220static void
2221in6_lltable_mark_used(struct llentry *lle)
2222{
2223
2224	LLE_REQ_LOCK(lle);
2225	lle->r_skip_req = 0;
2226
2227	/*
2228	 * Set the hit time so the callback function
2229	 * can determine the remaining time before
2230	 * transiting to the DELAY state.
2231	 */
2232	lle->lle_hittime = time_uptime;
2233	LLE_REQ_UNLOCK(lle);
2234}
2235
2236static inline uint32_t
2237in6_lltable_hash_dst(const struct in6_addr *dst, uint32_t hsize)
2238{
2239
2240	return (IN6_LLTBL_HASH(dst->s6_addr32[3], hsize));
2241}
2242
2243static uint32_t
2244in6_lltable_hash(const struct llentry *lle, uint32_t hsize)
2245{
2246
2247	return (in6_lltable_hash_dst(&lle->r_l3addr.addr6, hsize));
2248}
2249
2250static void
2251in6_lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
2252{
2253	struct sockaddr_in6 *sin6;
2254
2255	sin6 = (struct sockaddr_in6 *)sa;
2256	bzero(sin6, sizeof(*sin6));
2257	sin6->sin6_family = AF_INET6;
2258	sin6->sin6_len = sizeof(*sin6);
2259	sin6->sin6_addr = lle->r_l3addr.addr6;
2260}
2261
2262static inline struct llentry *
2263in6_lltable_find_dst(struct lltable *llt, const struct in6_addr *dst)
2264{
2265	struct llentry *lle;
2266	struct llentries *lleh;
2267	u_int hashidx;
2268
2269	hashidx = in6_lltable_hash_dst(dst, llt->llt_hsize);
2270	lleh = &llt->lle_head[hashidx];
2271	CK_LIST_FOREACH(lle, lleh, lle_next) {
2272		if (lle->la_flags & LLE_DELETED)
2273			continue;
2274		if (IN6_ARE_ADDR_EQUAL(&lle->r_l3addr.addr6, dst))
2275			break;
2276	}
2277
2278	return (lle);
2279}
2280
2281static void
2282in6_lltable_delete_entry(struct lltable *llt, struct llentry *lle)
2283{
2284
2285	lle->la_flags |= LLE_DELETED;
2286	EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
2287#ifdef DIAGNOSTIC
2288	log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
2289#endif
2290	llentry_free(lle);
2291}
2292
2293static struct llentry *
2294in6_lltable_alloc(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 ifnet *ifp = llt->llt_ifp;
2299	struct llentry *lle;
2300	char linkhdr[LLE_MAX_LINKHDR];
2301	size_t linkhdrsize;
2302	int lladdr_off;
2303
2304	KASSERT(l3addr->sa_family == AF_INET6,
2305	    ("sin_family %d", l3addr->sa_family));
2306
2307	/*
2308	 * A route that covers the given address must have
2309	 * been installed 1st because we are doing a resolution,
2310	 * verify this.
2311	 */
2312	if (!(flags & LLE_IFADDR) &&
2313	    in6_lltable_rtcheck(ifp, flags, l3addr) != 0)
2314		return (NULL);
2315
2316	lle = in6_lltable_new(&sin6->sin6_addr, flags);
2317	if (lle == NULL) {
2318		log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
2319		return (NULL);
2320	}
2321	lle->la_flags = flags;
2322	if ((flags & LLE_IFADDR) == LLE_IFADDR) {
2323		linkhdrsize = LLE_MAX_LINKHDR;
2324		if (lltable_calc_llheader(ifp, AF_INET6, IF_LLADDR(ifp),
2325		    linkhdr, &linkhdrsize, &lladdr_off) != 0) {
2326			NET_EPOCH_CALL(in6_lltable_destroy_lle_unlocked, &lle->lle_epoch_ctx);
2327			return (NULL);
2328		}
2329		lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
2330		    lladdr_off);
2331		lle->la_flags |= LLE_STATIC;
2332	}
2333
2334	if ((lle->la_flags & LLE_STATIC) != 0)
2335		lle->ln_state = ND6_LLINFO_REACHABLE;
2336
2337	return (lle);
2338}
2339
2340static struct llentry *
2341in6_lltable_lookup(struct lltable *llt, u_int flags,
2342	const struct sockaddr *l3addr)
2343{
2344	const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr;
2345	struct llentry *lle;
2346
2347	IF_AFDATA_LOCK_ASSERT(llt->llt_ifp);
2348	KASSERT(l3addr->sa_family == AF_INET6,
2349	    ("sin_family %d", l3addr->sa_family));
2350	KASSERT((flags & (LLE_UNLOCKED | LLE_EXCLUSIVE)) !=
2351	    (LLE_UNLOCKED | LLE_EXCLUSIVE),
2352	    ("wrong lle request flags: %#x", flags));
2353
2354	lle = in6_lltable_find_dst(llt, &sin6->sin6_addr);
2355	if (lle == NULL)
2356		return (NULL);
2357	if (flags & LLE_UNLOCKED)
2358		return (lle);
2359
2360	if (flags & LLE_EXCLUSIVE)
2361		LLE_WLOCK(lle);
2362	else
2363		LLE_RLOCK(lle);
2364
2365	/*
2366	 * If the afdata lock is not held, the LLE may have been unlinked while
2367	 * we were blocked on the LLE lock.  Check for this case.
2368	 */
2369	if (__predict_false((lle->la_flags & LLE_LINKED) == 0)) {
2370		if (flags & LLE_EXCLUSIVE)
2371			LLE_WUNLOCK(lle);
2372		else
2373			LLE_RUNLOCK(lle);
2374		return (NULL);
2375	}
2376	return (lle);
2377}
2378
2379static int
2380in6_lltable_dump_entry(struct lltable *llt, struct llentry *lle,
2381    struct sysctl_req *wr)
2382{
2383	struct ifnet *ifp = llt->llt_ifp;
2384	/* XXX stack use */
2385	struct {
2386		struct rt_msghdr	rtm;
2387		struct sockaddr_in6	sin6;
2388		/*
2389		 * ndp.c assumes that sdl is word aligned
2390		 */
2391#ifdef __LP64__
2392		uint32_t		pad;
2393#endif
2394		struct sockaddr_dl	sdl;
2395	} ndpc;
2396	struct sockaddr_dl *sdl;
2397	int error;
2398
2399	bzero(&ndpc, sizeof(ndpc));
2400	/* skip deleted entries */
2401	if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
2402		return (0);
2403	/* Skip if jailed and not a valid IP of the prison. */
2404	lltable_fill_sa_entry(lle, (struct sockaddr *)&ndpc.sin6);
2405	if (prison_if(wr->td->td_ucred, (struct sockaddr *)&ndpc.sin6) != 0)
2406		return (0);
2407	/*
2408	 * produce a msg made of:
2409	 *  struct rt_msghdr;
2410	 *  struct sockaddr_in6 (IPv6)
2411	 *  struct sockaddr_dl;
2412	 */
2413	ndpc.rtm.rtm_msglen = sizeof(ndpc);
2414	ndpc.rtm.rtm_version = RTM_VERSION;
2415	ndpc.rtm.rtm_type = RTM_GET;
2416	ndpc.rtm.rtm_flags = RTF_UP;
2417	ndpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
2418	sa6_recoverscope(&ndpc.sin6);
2419
2420	/* publish */
2421	if (lle->la_flags & LLE_PUB)
2422		ndpc.rtm.rtm_flags |= RTF_ANNOUNCE;
2423
2424	sdl = &ndpc.sdl;
2425	sdl->sdl_family = AF_LINK;
2426	sdl->sdl_len = sizeof(*sdl);
2427	sdl->sdl_index = ifp->if_index;
2428	sdl->sdl_type = ifp->if_type;
2429	if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
2430		sdl->sdl_alen = ifp->if_addrlen;
2431		bcopy(lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
2432	} else {
2433		sdl->sdl_alen = 0;
2434		bzero(LLADDR(sdl), ifp->if_addrlen);
2435	}
2436	if (lle->la_expire != 0)
2437		ndpc.rtm.rtm_rmx.rmx_expire = lle->la_expire +
2438		    lle->lle_remtime / hz + time_second - time_uptime;
2439	ndpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
2440	if (lle->la_flags & LLE_STATIC)
2441		ndpc.rtm.rtm_flags |= RTF_STATIC;
2442	if (lle->la_flags & LLE_IFADDR)
2443		ndpc.rtm.rtm_flags |= RTF_PINNED;
2444	if (lle->ln_router != 0)
2445		ndpc.rtm.rtm_flags |= RTF_GATEWAY;
2446	ndpc.rtm.rtm_rmx.rmx_pksent = lle->la_asked;
2447	/* Store state in rmx_weight value */
2448	ndpc.rtm.rtm_rmx.rmx_state = lle->ln_state;
2449	ndpc.rtm.rtm_index = ifp->if_index;
2450	error = SYSCTL_OUT(wr, &ndpc, sizeof(ndpc));
2451
2452	return (error);
2453}
2454
2455static struct lltable *
2456in6_lltattach(struct ifnet *ifp)
2457{
2458	struct lltable *llt;
2459
2460	llt = lltable_allocate_htbl(IN6_LLTBL_DEFAULT_HSIZE);
2461	llt->llt_af = AF_INET6;
2462	llt->llt_ifp = ifp;
2463
2464	llt->llt_lookup = in6_lltable_lookup;
2465	llt->llt_alloc_entry = in6_lltable_alloc;
2466	llt->llt_delete_entry = in6_lltable_delete_entry;
2467	llt->llt_dump_entry = in6_lltable_dump_entry;
2468	llt->llt_hash = in6_lltable_hash;
2469	llt->llt_fill_sa_entry = in6_lltable_fill_sa_entry;
2470	llt->llt_free_entry = in6_lltable_free_entry;
2471	llt->llt_match_prefix = in6_lltable_match_prefix;
2472	llt->llt_mark_used = in6_lltable_mark_used;
2473 	lltable_link(llt);
2474
2475	return (llt);
2476}
2477
2478void *
2479in6_domifattach(struct ifnet *ifp)
2480{
2481	struct in6_ifextra *ext;
2482
2483	/* There are not IPv6-capable interfaces. */
2484	switch (ifp->if_type) {
2485	case IFT_PFLOG:
2486	case IFT_PFSYNC:
2487	case IFT_USB:
2488		return (NULL);
2489	}
2490	ext = (struct in6_ifextra *)malloc(sizeof(*ext), M_IFADDR, M_WAITOK);
2491	bzero(ext, sizeof(*ext));
2492
2493	ext->in6_ifstat = malloc(sizeof(counter_u64_t) *
2494	    sizeof(struct in6_ifstat) / sizeof(uint64_t), M_IFADDR, M_WAITOK);
2495	COUNTER_ARRAY_ALLOC(ext->in6_ifstat,
2496	    sizeof(struct in6_ifstat) / sizeof(uint64_t), M_WAITOK);
2497
2498	ext->icmp6_ifstat = malloc(sizeof(counter_u64_t) *
2499	    sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_IFADDR,
2500	    M_WAITOK);
2501	COUNTER_ARRAY_ALLOC(ext->icmp6_ifstat,
2502	    sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_WAITOK);
2503
2504	ext->nd_ifinfo = nd6_ifattach(ifp);
2505	ext->scope6_id = scope6_ifattach(ifp);
2506	ext->lltable = in6_lltattach(ifp);
2507
2508	ext->mld_ifinfo = mld_domifattach(ifp);
2509
2510	return ext;
2511}
2512
2513int
2514in6_domifmtu(struct ifnet *ifp)
2515{
2516	if (ifp->if_afdata[AF_INET6] == NULL)
2517		return ifp->if_mtu;
2518
2519	return (IN6_LINKMTU(ifp));
2520}
2521
2522void
2523in6_domifdetach(struct ifnet *ifp, void *aux)
2524{
2525	struct in6_ifextra *ext = (struct in6_ifextra *)aux;
2526
2527	mld_domifdetach(ifp);
2528	scope6_ifdetach(ext->scope6_id);
2529	nd6_ifdetach(ifp, ext->nd_ifinfo);
2530	lltable_free(ext->lltable);
2531	COUNTER_ARRAY_FREE(ext->in6_ifstat,
2532	    sizeof(struct in6_ifstat) / sizeof(uint64_t));
2533	free(ext->in6_ifstat, M_IFADDR);
2534	COUNTER_ARRAY_FREE(ext->icmp6_ifstat,
2535	    sizeof(struct icmp6_ifstat) / sizeof(uint64_t));
2536	free(ext->icmp6_ifstat, M_IFADDR);
2537	free(ext, M_IFADDR);
2538}
2539
2540/*
2541 * Convert sockaddr_in6 to sockaddr_in.  Original sockaddr_in6 must be
2542 * v4 mapped addr or v4 compat addr
2543 */
2544void
2545in6_sin6_2_sin(struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
2546{
2547
2548	bzero(sin, sizeof(*sin));
2549	sin->sin_len = sizeof(struct sockaddr_in);
2550	sin->sin_family = AF_INET;
2551	sin->sin_port = sin6->sin6_port;
2552	sin->sin_addr.s_addr = sin6->sin6_addr.s6_addr32[3];
2553}
2554
2555/* Convert sockaddr_in to sockaddr_in6 in v4 mapped addr format. */
2556void
2557in6_sin_2_v4mapsin6(struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
2558{
2559	bzero(sin6, sizeof(*sin6));
2560	sin6->sin6_len = sizeof(struct sockaddr_in6);
2561	sin6->sin6_family = AF_INET6;
2562	sin6->sin6_port = sin->sin_port;
2563	sin6->sin6_addr.s6_addr32[0] = 0;
2564	sin6->sin6_addr.s6_addr32[1] = 0;
2565	sin6->sin6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
2566	sin6->sin6_addr.s6_addr32[3] = sin->sin_addr.s_addr;
2567}
2568
2569/* Convert sockaddr_in6 into sockaddr_in. */
2570void
2571in6_sin6_2_sin_in_sock(struct sockaddr *nam)
2572{
2573	struct sockaddr_in *sin_p;
2574	struct sockaddr_in6 sin6;
2575
2576	/*
2577	 * Save original sockaddr_in6 addr and convert it
2578	 * to sockaddr_in.
2579	 */
2580	sin6 = *(struct sockaddr_in6 *)nam;
2581	sin_p = (struct sockaddr_in *)nam;
2582	in6_sin6_2_sin(sin_p, &sin6);
2583}
2584
2585/* Convert sockaddr_in into sockaddr_in6 in v4 mapped addr format. */
2586void
2587in6_sin_2_v4mapsin6_in_sock(struct sockaddr **nam)
2588{
2589	struct sockaddr_in *sin_p;
2590	struct sockaddr_in6 *sin6_p;
2591
2592	sin6_p = malloc(sizeof *sin6_p, M_SONAME, M_WAITOK);
2593	sin_p = (struct sockaddr_in *)*nam;
2594	in6_sin_2_v4mapsin6(sin_p, sin6_p);
2595	free(*nam, M_SONAME);
2596	*nam = (struct sockaddr *)sin6_p;
2597}
2598