xref: /freebsd-11-stable/sys/netinet6/in6.c (revision 75b55772be3d4d44a4e6fb07210ef45bd73b798b)
1 /*-
2  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. Neither the name of the project nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	$KAME: in6.c,v 1.259 2002/01/21 11:37:50 keiichi Exp $
30  */
31 
32 /*-
33  * Copyright (c) 1982, 1986, 1991, 1993
34  *	The Regents of the University of California.  All rights reserved.
35  *
36  * Redistribution and use in source and binary forms, with or without
37  * modification, are permitted provided that the following conditions
38  * are met:
39  * 1. Redistributions of source code must retain the above copyright
40  *    notice, this list of conditions and the following disclaimer.
41  * 2. Redistributions in binary form must reproduce the above copyright
42  *    notice, this list of conditions and the following disclaimer in the
43  *    documentation and/or other materials provided with the distribution.
44  * 4. Neither the name of the University nor the names of its contributors
45  *    may be used to endorse or promote products derived from this software
46  *    without specific prior written permission.
47  *
48  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58  * SUCH DAMAGE.
59  *
60  *	@(#)in.c	8.2 (Berkeley) 11/15/93
61  */
62 
63 #include <sys/cdefs.h>
64 __FBSDID("$FreeBSD$");
65 
66 #include "opt_compat.h"
67 #include "opt_inet.h"
68 #include "opt_inet6.h"
69 
70 #include <sys/param.h>
71 #include <sys/eventhandler.h>
72 #include <sys/errno.h>
73 #include <sys/jail.h>
74 #include <sys/malloc.h>
75 #include <sys/socket.h>
76 #include <sys/socketvar.h>
77 #include <sys/sockio.h>
78 #include <sys/systm.h>
79 #include <sys/priv.h>
80 #include <sys/proc.h>
81 #include <sys/time.h>
82 #include <sys/kernel.h>
83 #include <sys/lock.h>
84 #include <sys/rmlock.h>
85 #include <sys/syslog.h>
86 
87 #include <net/if.h>
88 #include <net/if_var.h>
89 #include <net/if_types.h>
90 #include <net/route.h>
91 #include <net/if_dl.h>
92 #include <net/vnet.h>
93 
94 #include <netinet/in.h>
95 #include <netinet/in_var.h>
96 #include <net/if_llatbl.h>
97 #include <netinet/if_ether.h>
98 #include <netinet/in_systm.h>
99 #include <netinet/ip.h>
100 #include <netinet/in_pcb.h>
101 #include <netinet/ip_carp.h>
102 
103 #include <netinet/ip6.h>
104 #include <netinet6/ip6_var.h>
105 #include <netinet6/nd6.h>
106 #include <netinet6/mld6_var.h>
107 #include <netinet6/ip6_mroute.h>
108 #include <netinet6/in6_ifattach.h>
109 #include <netinet6/scope6_var.h>
110 #include <netinet6/in6_fib.h>
111 #include <netinet6/in6_pcb.h>
112 
113 VNET_DECLARE(int, icmp6_nodeinfo_oldmcprefix);
114 #define V_icmp6_nodeinfo_oldmcprefix	VNET(icmp6_nodeinfo_oldmcprefix)
115 
116 /*
117  * Definitions of some costant IP6 addresses.
118  */
119 const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT;
120 const struct in6_addr in6addr_loopback = IN6ADDR_LOOPBACK_INIT;
121 const struct in6_addr in6addr_nodelocal_allnodes =
122 	IN6ADDR_NODELOCAL_ALLNODES_INIT;
123 const struct in6_addr in6addr_linklocal_allnodes =
124 	IN6ADDR_LINKLOCAL_ALLNODES_INIT;
125 const struct in6_addr in6addr_linklocal_allrouters =
126 	IN6ADDR_LINKLOCAL_ALLROUTERS_INIT;
127 const struct in6_addr in6addr_linklocal_allv2routers =
128 	IN6ADDR_LINKLOCAL_ALLV2ROUTERS_INIT;
129 
130 const struct in6_addr in6mask0 = IN6MASK0;
131 const struct in6_addr in6mask32 = IN6MASK32;
132 const struct in6_addr in6mask64 = IN6MASK64;
133 const struct in6_addr in6mask96 = IN6MASK96;
134 const struct in6_addr in6mask128 = IN6MASK128;
135 
136 const struct sockaddr_in6 sa6_any =
137 	{ sizeof(sa6_any), AF_INET6, 0, 0, IN6ADDR_ANY_INIT, 0 };
138 
139 static int in6_notify_ifa(struct ifnet *, struct in6_ifaddr *,
140 	struct in6_aliasreq *, int);
141 static void in6_unlink_ifa(struct in6_ifaddr *, struct ifnet *);
142 
143 static int in6_validate_ifra(struct ifnet *, struct in6_aliasreq *,
144     struct in6_ifaddr *, int);
145 static struct in6_ifaddr *in6_alloc_ifa(struct ifnet *,
146     struct in6_aliasreq *, int flags);
147 static int in6_update_ifa_internal(struct ifnet *, struct in6_aliasreq *,
148     struct in6_ifaddr *, int, int);
149 static int in6_broadcast_ifa(struct ifnet *, struct in6_aliasreq *,
150     struct in6_ifaddr *, int);
151 
152 #define ifa2ia6(ifa)	((struct in6_ifaddr *)(ifa))
153 #define ia62ifa(ia6)	(&((ia6)->ia_ifa))
154 
155 
156 void
in6_newaddrmsg(struct in6_ifaddr * ia,int cmd)157 in6_newaddrmsg(struct in6_ifaddr *ia, int cmd)
158 {
159 	struct sockaddr_dl gateway;
160 	struct sockaddr_in6 mask, addr;
161 	struct rtentry rt;
162 	int fibnum;
163 
164 	/*
165 	 * initialize for rtmsg generation
166 	 */
167 	bzero(&gateway, sizeof(gateway));
168 	gateway.sdl_len = sizeof(gateway);
169 	gateway.sdl_family = AF_LINK;
170 
171 	bzero(&rt, sizeof(rt));
172 	rt.rt_gateway = (struct sockaddr *)&gateway;
173 	memcpy(&mask, &ia->ia_prefixmask, sizeof(ia->ia_prefixmask));
174 	memcpy(&addr, &ia->ia_addr, sizeof(ia->ia_addr));
175 	rt_mask(&rt) = (struct sockaddr *)&mask;
176 	rt_key(&rt) = (struct sockaddr *)&addr;
177 	rt.rt_flags = RTF_HOST | RTF_STATIC;
178 	if (cmd == RTM_ADD)
179 		rt.rt_flags |= RTF_UP;
180 	fibnum = V_rt_add_addr_allfibs ? RT_ALL_FIBS : ia62ifa(ia)->ifa_ifp->if_fib;
181 	/* Announce arrival of local address to this FIB. */
182 	rt_newaddrmsg_fib(cmd, &ia->ia_ifa, 0, &rt, fibnum);
183 }
184 
185 int
in6_mask2len(struct in6_addr * mask,u_char * lim0)186 in6_mask2len(struct in6_addr *mask, u_char *lim0)
187 {
188 	int x = 0, y;
189 	u_char *lim = lim0, *p;
190 
191 	/* ignore the scope_id part */
192 	if (lim0 == NULL || lim0 - (u_char *)mask > sizeof(*mask))
193 		lim = (u_char *)mask + sizeof(*mask);
194 	for (p = (u_char *)mask; p < lim; x++, p++) {
195 		if (*p != 0xff)
196 			break;
197 	}
198 	y = 0;
199 	if (p < lim) {
200 		for (y = 0; y < 8; y++) {
201 			if ((*p & (0x80 >> y)) == 0)
202 				break;
203 		}
204 	}
205 
206 	/*
207 	 * when the limit pointer is given, do a stricter check on the
208 	 * remaining bits.
209 	 */
210 	if (p < lim) {
211 		if (y != 0 && (*p & (0x00ff >> y)) != 0)
212 			return (-1);
213 		for (p = p + 1; p < lim; p++)
214 			if (*p != 0)
215 				return (-1);
216 	}
217 
218 	return x * 8 + y;
219 }
220 
221 #ifdef COMPAT_FREEBSD32
222 struct in6_ndifreq32 {
223 	char ifname[IFNAMSIZ];
224 	uint32_t ifindex;
225 };
226 #define	SIOCGDEFIFACE32_IN6	_IOWR('i', 86, struct in6_ndifreq32)
227 #endif
228 
229 int
in6_control(struct socket * so,u_long cmd,caddr_t data,struct ifnet * ifp,struct thread * td)230 in6_control(struct socket *so, u_long cmd, caddr_t data,
231     struct ifnet *ifp, struct thread *td)
232 {
233 	struct	in6_ifreq *ifr = (struct in6_ifreq *)data;
234 	struct	in6_ifaddr *ia = NULL;
235 	struct	in6_aliasreq *ifra = (struct in6_aliasreq *)data;
236 	struct sockaddr_in6 *sa6;
237 	int carp_attached = 0;
238 	int error;
239 	u_long ocmd = cmd;
240 
241 	/*
242 	 * Compat to make pre-10.x ifconfig(8) operable.
243 	 */
244 	if (cmd == OSIOCAIFADDR_IN6)
245 		cmd = SIOCAIFADDR_IN6;
246 
247 	switch (cmd) {
248 	case SIOCGETSGCNT_IN6:
249 	case SIOCGETMIFCNT_IN6:
250 		/*
251 		 * XXX mrt_ioctl has a 3rd, unused, FIB argument in route.c.
252 		 * We cannot see how that would be needed, so do not adjust the
253 		 * KPI blindly; more likely should clean up the IPv4 variant.
254 		 */
255 		return (mrt6_ioctl ? mrt6_ioctl(cmd, data) : EOPNOTSUPP);
256 	}
257 
258 	switch (cmd) {
259 	case SIOCAADDRCTL_POLICY:
260 	case SIOCDADDRCTL_POLICY:
261 		if (td != NULL) {
262 			error = priv_check(td, PRIV_NETINET_ADDRCTRL6);
263 			if (error)
264 				return (error);
265 		}
266 		return (in6_src_ioctl(cmd, data));
267 	}
268 
269 	if (ifp == NULL)
270 		return (EOPNOTSUPP);
271 
272 	switch (cmd) {
273 	case SIOCSNDFLUSH_IN6:
274 	case SIOCSPFXFLUSH_IN6:
275 	case SIOCSRTRFLUSH_IN6:
276 	case SIOCSDEFIFACE_IN6:
277 	case SIOCSIFINFO_FLAGS:
278 	case SIOCSIFINFO_IN6:
279 		if (td != NULL) {
280 			error = priv_check(td, PRIV_NETINET_ND6);
281 			if (error)
282 				return (error);
283 		}
284 		/* FALLTHROUGH */
285 	case OSIOCGIFINFO_IN6:
286 	case SIOCGIFINFO_IN6:
287 	case SIOCGNBRINFO_IN6:
288 	case SIOCGDEFIFACE_IN6:
289 		return (nd6_ioctl(cmd, data, ifp));
290 
291 #ifdef COMPAT_FREEBSD32
292 	case SIOCGDEFIFACE32_IN6:
293 		{
294 			struct in6_ndifreq ndif;
295 			struct in6_ndifreq32 *ndif32;
296 
297 			error = nd6_ioctl(SIOCGDEFIFACE_IN6, (caddr_t)&ndif,
298 			    ifp);
299 			if (error)
300 				return (error);
301 			ndif32 = (struct in6_ndifreq32 *)data;
302 			ndif32->ifindex = ndif.ifindex;
303 			return (0);
304 		}
305 #endif
306 	}
307 
308 	switch (cmd) {
309 	case SIOCSIFPREFIX_IN6:
310 	case SIOCDIFPREFIX_IN6:
311 	case SIOCAIFPREFIX_IN6:
312 	case SIOCCIFPREFIX_IN6:
313 	case SIOCSGIFPREFIX_IN6:
314 	case SIOCGIFPREFIX_IN6:
315 		log(LOG_NOTICE,
316 		    "prefix ioctls are now invalidated. "
317 		    "please use ifconfig.\n");
318 		return (EOPNOTSUPP);
319 	}
320 
321 	switch (cmd) {
322 	case SIOCSSCOPE6:
323 		if (td != NULL) {
324 			error = priv_check(td, PRIV_NETINET_SCOPE6);
325 			if (error)
326 				return (error);
327 		}
328 		/* FALLTHROUGH */
329 	case SIOCGSCOPE6:
330 	case SIOCGSCOPE6DEF:
331 		return (scope6_ioctl(cmd, data, ifp));
332 	}
333 
334 	/*
335 	 * Find address for this interface, if it exists.
336 	 *
337 	 * In netinet code, we have checked ifra_addr in SIOCSIF*ADDR operation
338 	 * only, and used the first interface address as the target of other
339 	 * operations (without checking ifra_addr).  This was because netinet
340 	 * code/API assumed at most 1 interface address per interface.
341 	 * Since IPv6 allows a node to assign multiple addresses
342 	 * on a single interface, we almost always look and check the
343 	 * presence of ifra_addr, and reject invalid ones here.
344 	 * It also decreases duplicated code among SIOC*_IN6 operations.
345 	 */
346 	switch (cmd) {
347 	case SIOCAIFADDR_IN6:
348 	case SIOCSIFPHYADDR_IN6:
349 		sa6 = &ifra->ifra_addr;
350 		break;
351 	case SIOCSIFADDR_IN6:
352 	case SIOCGIFADDR_IN6:
353 	case SIOCSIFDSTADDR_IN6:
354 	case SIOCSIFNETMASK_IN6:
355 	case SIOCGIFDSTADDR_IN6:
356 	case SIOCGIFNETMASK_IN6:
357 	case SIOCDIFADDR_IN6:
358 	case SIOCGIFPSRCADDR_IN6:
359 	case SIOCGIFPDSTADDR_IN6:
360 	case SIOCGIFAFLAG_IN6:
361 	case SIOCSNDFLUSH_IN6:
362 	case SIOCSPFXFLUSH_IN6:
363 	case SIOCSRTRFLUSH_IN6:
364 	case SIOCGIFALIFETIME_IN6:
365 	case SIOCGIFSTAT_IN6:
366 	case SIOCGIFSTAT_ICMP6:
367 		sa6 = &ifr->ifr_addr;
368 		break;
369 	case SIOCSIFADDR:
370 	case SIOCSIFBRDADDR:
371 	case SIOCSIFDSTADDR:
372 	case SIOCSIFNETMASK:
373 		/*
374 		 * Although we should pass any non-INET6 ioctl requests
375 		 * down to driver, we filter some legacy INET requests.
376 		 * Drivers trust SIOCSIFADDR et al to come from an already
377 		 * privileged layer, and do not perform any credentials
378 		 * checks or input validation.
379 		 */
380 		return (EINVAL);
381 	default:
382 		sa6 = NULL;
383 		break;
384 	}
385 	if (sa6 && sa6->sin6_family == AF_INET6) {
386 		if (sa6->sin6_scope_id != 0)
387 			error = sa6_embedscope(sa6, 0);
388 		else
389 			error = in6_setscope(&sa6->sin6_addr, ifp, NULL);
390 		if (error != 0)
391 			return (error);
392 		if (td != NULL && (error = prison_check_ip6(td->td_ucred,
393 		    &sa6->sin6_addr)) != 0)
394 			return (error);
395 		ia = in6ifa_ifpwithaddr(ifp, &sa6->sin6_addr);
396 	} else
397 		ia = NULL;
398 
399 	switch (cmd) {
400 	case SIOCSIFADDR_IN6:
401 	case SIOCSIFDSTADDR_IN6:
402 	case SIOCSIFNETMASK_IN6:
403 		/*
404 		 * Since IPv6 allows a node to assign multiple addresses
405 		 * on a single interface, SIOCSIFxxx ioctls are deprecated.
406 		 */
407 		/* we decided to obsolete this command (20000704) */
408 		error = EINVAL;
409 		goto out;
410 
411 	case SIOCDIFADDR_IN6:
412 		/*
413 		 * for IPv4, we look for existing in_ifaddr here to allow
414 		 * "ifconfig if0 delete" to remove the first IPv4 address on
415 		 * the interface.  For IPv6, as the spec allows multiple
416 		 * interface address from the day one, we consider "remove the
417 		 * first one" semantics to be not preferable.
418 		 */
419 		if (ia == NULL) {
420 			error = EADDRNOTAVAIL;
421 			goto out;
422 		}
423 		/* FALLTHROUGH */
424 	case SIOCAIFADDR_IN6:
425 		/*
426 		 * We always require users to specify a valid IPv6 address for
427 		 * the corresponding operation.
428 		 */
429 		if (ifra->ifra_addr.sin6_family != AF_INET6 ||
430 		    ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6)) {
431 			error = EAFNOSUPPORT;
432 			goto out;
433 		}
434 
435 		if (td != NULL) {
436 			error = priv_check(td, (cmd == SIOCDIFADDR_IN6) ?
437 			    PRIV_NET_DELIFADDR : PRIV_NET_ADDIFADDR);
438 			if (error)
439 				goto out;
440 		}
441 		/* FALLTHROUGH */
442 	case SIOCGIFSTAT_IN6:
443 	case SIOCGIFSTAT_ICMP6:
444 		if (ifp->if_afdata[AF_INET6] == NULL) {
445 			error = EPFNOSUPPORT;
446 			goto out;
447 		}
448 		break;
449 
450 	case SIOCGIFADDR_IN6:
451 		/* This interface is basically deprecated. use SIOCGIFCONF. */
452 		/* FALLTHROUGH */
453 	case SIOCGIFAFLAG_IN6:
454 	case SIOCGIFNETMASK_IN6:
455 	case SIOCGIFDSTADDR_IN6:
456 	case SIOCGIFALIFETIME_IN6:
457 		/* must think again about its semantics */
458 		if (ia == NULL) {
459 			error = EADDRNOTAVAIL;
460 			goto out;
461 		}
462 		break;
463 	}
464 
465 	switch (cmd) {
466 	case SIOCGIFADDR_IN6:
467 		ifr->ifr_addr = ia->ia_addr;
468 		if ((error = sa6_recoverscope(&ifr->ifr_addr)) != 0)
469 			goto out;
470 		break;
471 
472 	case SIOCGIFDSTADDR_IN6:
473 		if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
474 			error = EINVAL;
475 			goto out;
476 		}
477 		/*
478 		 * XXX: should we check if ifa_dstaddr is NULL and return
479 		 * an error?
480 		 */
481 		ifr->ifr_dstaddr = ia->ia_dstaddr;
482 		if ((error = sa6_recoverscope(&ifr->ifr_dstaddr)) != 0)
483 			goto out;
484 		break;
485 
486 	case SIOCGIFNETMASK_IN6:
487 		ifr->ifr_addr = ia->ia_prefixmask;
488 		break;
489 
490 	case SIOCGIFAFLAG_IN6:
491 		ifr->ifr_ifru.ifru_flags6 = ia->ia6_flags;
492 		break;
493 
494 	case SIOCGIFSTAT_IN6:
495 		COUNTER_ARRAY_COPY(((struct in6_ifextra *)
496 		    ifp->if_afdata[AF_INET6])->in6_ifstat,
497 		    &ifr->ifr_ifru.ifru_stat,
498 		    sizeof(struct in6_ifstat) / sizeof(uint64_t));
499 		break;
500 
501 	case SIOCGIFSTAT_ICMP6:
502 		COUNTER_ARRAY_COPY(((struct in6_ifextra *)
503 		    ifp->if_afdata[AF_INET6])->icmp6_ifstat,
504 		    &ifr->ifr_ifru.ifru_icmp6stat,
505 		    sizeof(struct icmp6_ifstat) / sizeof(uint64_t));
506 		break;
507 
508 	case SIOCGIFALIFETIME_IN6:
509 		ifr->ifr_ifru.ifru_lifetime = ia->ia6_lifetime;
510 		if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
511 			time_t maxexpire;
512 			struct in6_addrlifetime *retlt =
513 			    &ifr->ifr_ifru.ifru_lifetime;
514 
515 			/*
516 			 * XXX: adjust expiration time assuming time_t is
517 			 * signed.
518 			 */
519 			maxexpire = (-1) &
520 			    ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
521 			if (ia->ia6_lifetime.ia6t_vltime <
522 			    maxexpire - ia->ia6_updatetime) {
523 				retlt->ia6t_expire = ia->ia6_updatetime +
524 				    ia->ia6_lifetime.ia6t_vltime;
525 			} else
526 				retlt->ia6t_expire = maxexpire;
527 		}
528 		if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
529 			time_t maxexpire;
530 			struct in6_addrlifetime *retlt =
531 			    &ifr->ifr_ifru.ifru_lifetime;
532 
533 			/*
534 			 * XXX: adjust expiration time assuming time_t is
535 			 * signed.
536 			 */
537 			maxexpire = (-1) &
538 			    ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
539 			if (ia->ia6_lifetime.ia6t_pltime <
540 			    maxexpire - ia->ia6_updatetime) {
541 				retlt->ia6t_preferred = ia->ia6_updatetime +
542 				    ia->ia6_lifetime.ia6t_pltime;
543 			} else
544 				retlt->ia6t_preferred = maxexpire;
545 		}
546 		break;
547 
548 	case SIOCAIFADDR_IN6:
549 	{
550 		struct nd_prefixctl pr0;
551 		struct nd_prefix *pr;
552 
553 		/*
554 		 * first, make or update the interface address structure,
555 		 * and link it to the list.
556 		 */
557 		if ((error = in6_update_ifa(ifp, ifra, ia, 0)) != 0)
558 			goto out;
559 		if (ia != NULL) {
560 			if (ia->ia_ifa.ifa_carp)
561 				(*carp_detach_p)(&ia->ia_ifa, true);
562 			ifa_free(&ia->ia_ifa);
563 		}
564 		if ((ia = in6ifa_ifpwithaddr(ifp, &ifra->ifra_addr.sin6_addr))
565 		    == NULL) {
566 			/*
567 			 * this can happen when the user specify the 0 valid
568 			 * lifetime.
569 			 */
570 			break;
571 		}
572 
573 		if (cmd == ocmd && ifra->ifra_vhid > 0) {
574 			if (carp_attach_p != NULL)
575 				error = (*carp_attach_p)(&ia->ia_ifa,
576 				    ifra->ifra_vhid);
577 			else
578 				error = EPROTONOSUPPORT;
579 			if (error)
580 				goto out;
581 			else
582 				carp_attached = 1;
583 		}
584 
585 		/*
586 		 * then, make the prefix on-link on the interface.
587 		 * XXX: we'd rather create the prefix before the address, but
588 		 * we need at least one address to install the corresponding
589 		 * interface route, so we configure the address first.
590 		 */
591 
592 		/*
593 		 * convert mask to prefix length (prefixmask has already
594 		 * been validated in in6_update_ifa().
595 		 */
596 		bzero(&pr0, sizeof(pr0));
597 		pr0.ndpr_ifp = ifp;
598 		pr0.ndpr_plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr,
599 		    NULL);
600 		if (pr0.ndpr_plen == 128) {
601 			/* we don't need to install a host route. */
602 			goto aifaddr_out;
603 		}
604 		pr0.ndpr_prefix = ifra->ifra_addr;
605 		/* apply the mask for safety. */
606 		IN6_MASK_ADDR(&pr0.ndpr_prefix.sin6_addr,
607 		    &ifra->ifra_prefixmask.sin6_addr);
608 
609 		/*
610 		 * XXX: since we don't have an API to set prefix (not address)
611 		 * lifetimes, we just use the same lifetimes as addresses.
612 		 * The (temporarily) installed lifetimes can be overridden by
613 		 * later advertised RAs (when accept_rtadv is non 0), which is
614 		 * an intended behavior.
615 		 */
616 		pr0.ndpr_raf_onlink = 1; /* should be configurable? */
617 		pr0.ndpr_raf_auto =
618 		    ((ifra->ifra_flags & IN6_IFF_AUTOCONF) != 0);
619 		pr0.ndpr_vltime = ifra->ifra_lifetime.ia6t_vltime;
620 		pr0.ndpr_pltime = ifra->ifra_lifetime.ia6t_pltime;
621 
622 		/* add the prefix if not yet. */
623 		if ((pr = nd6_prefix_lookup(&pr0)) == NULL) {
624 			/*
625 			 * nd6_prelist_add will install the corresponding
626 			 * interface route.
627 			 */
628 			if ((error = nd6_prelist_add(&pr0, NULL, &pr)) != 0) {
629 				if (carp_attached)
630 					(*carp_detach_p)(&ia->ia_ifa, false);
631 				goto out;
632 			}
633 		}
634 
635 		/* relate the address to the prefix */
636 		if (ia->ia6_ndpr == NULL) {
637 			ia->ia6_ndpr = pr;
638 			pr->ndpr_addrcnt++;
639 
640 			/*
641 			 * If this is the first autoconf address from the
642 			 * prefix, create a temporary address as well
643 			 * (when required).
644 			 */
645 			if ((ia->ia6_flags & IN6_IFF_AUTOCONF) &&
646 			    V_ip6_use_tempaddr && pr->ndpr_addrcnt == 1) {
647 				int e;
648 				if ((e = in6_tmpifadd(ia, 1, 0)) != 0) {
649 					log(LOG_NOTICE, "in6_control: failed "
650 					    "to create a temporary address, "
651 					    "errno=%d\n", e);
652 				}
653 			}
654 		}
655 		nd6_prefix_rele(pr);
656 
657 		/*
658 		 * this might affect the status of autoconfigured addresses,
659 		 * that is, this address might make other addresses detached.
660 		 */
661 		pfxlist_onlink_check();
662 
663 aifaddr_out:
664 		/*
665 		 * Try to clear the flag when a new IPv6 address is added
666 		 * onto an IFDISABLED interface and it succeeds.
667 		 */
668 		if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) {
669 			struct in6_ndireq nd;
670 
671 			memset(&nd, 0, sizeof(nd));
672 			nd.ndi.flags = ND_IFINFO(ifp)->flags;
673 			nd.ndi.flags &= ~ND6_IFF_IFDISABLED;
674 			if (nd6_ioctl(SIOCSIFINFO_FLAGS, (caddr_t)&nd, ifp) < 0)
675 				log(LOG_NOTICE, "SIOCAIFADDR_IN6: "
676 				    "SIOCSIFINFO_FLAGS for -ifdisabled "
677 				    "failed.");
678 			/*
679 			 * Ignore failure of clearing the flag intentionally.
680 			 * The failure means address duplication was detected.
681 			 */
682 		}
683 		break;
684 	}
685 
686 	case SIOCDIFADDR_IN6:
687 	{
688 		struct nd_prefix *pr;
689 
690 		/*
691 		 * If the address being deleted is the only one that owns
692 		 * the corresponding prefix, expire the prefix as well.
693 		 * XXX: theoretically, we don't have to worry about such
694 		 * relationship, since we separate the address management
695 		 * and the prefix management.  We do this, however, to provide
696 		 * as much backward compatibility as possible in terms of
697 		 * the ioctl operation.
698 		 * Note that in6_purgeaddr() will decrement ndpr_addrcnt.
699 		 */
700 		pr = ia->ia6_ndpr;
701 		in6_purgeaddr(&ia->ia_ifa);
702 		if (pr != NULL && pr->ndpr_addrcnt == 0) {
703 			ND6_WLOCK();
704 			nd6_prefix_unlink(pr, NULL);
705 			ND6_WUNLOCK();
706 			nd6_prefix_del(pr);
707 		}
708 		EVENTHANDLER_INVOKE(ifaddr_event, ifp);
709 		break;
710 	}
711 
712 	default:
713 		if (ifp->if_ioctl == NULL) {
714 			error = EOPNOTSUPP;
715 			goto out;
716 		}
717 		error = (*ifp->if_ioctl)(ifp, cmd, data);
718 		goto out;
719 	}
720 
721 	error = 0;
722 out:
723 	if (ia != NULL)
724 		ifa_free(&ia->ia_ifa);
725 	return (error);
726 }
727 
728 
729 /*
730  * Join necessary multicast groups.  Factored out from in6_update_ifa().
731  * This entire work should only be done once, for the default FIB.
732  */
733 static int
in6_update_ifa_join_mc(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia,int flags,struct in6_multi ** in6m_sol)734 in6_update_ifa_join_mc(struct ifnet *ifp, struct in6_aliasreq *ifra,
735     struct in6_ifaddr *ia, int flags, struct in6_multi **in6m_sol)
736 {
737 	char ip6buf[INET6_ADDRSTRLEN];
738 	struct in6_addr mltaddr;
739 	struct in6_multi_mship *imm;
740 	int delay, error;
741 
742 	KASSERT(in6m_sol != NULL, ("%s: in6m_sol is NULL", __func__));
743 
744 	/* Join solicited multicast addr for new host id. */
745 	bzero(&mltaddr, sizeof(struct in6_addr));
746 	mltaddr.s6_addr32[0] = IPV6_ADDR_INT32_MLL;
747 	mltaddr.s6_addr32[2] = htonl(1);
748 	mltaddr.s6_addr32[3] = ifra->ifra_addr.sin6_addr.s6_addr32[3];
749 	mltaddr.s6_addr8[12] = 0xff;
750 	if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0) {
751 		/* XXX: should not happen */
752 		log(LOG_ERR, "%s: in6_setscope failed\n", __func__);
753 		goto cleanup;
754 	}
755 	delay = error = 0;
756 	if ((flags & IN6_IFAUPDATE_DADDELAY)) {
757 		/*
758 		 * We need a random delay for DAD on the address being
759 		 * configured.  It also means delaying transmission of the
760 		 * corresponding MLD report to avoid report collision.
761 		 * [RFC 4861, Section 6.3.7]
762 		 */
763 		delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
764 	}
765 	imm = in6_joingroup(ifp, &mltaddr, &error, delay);
766 	if (imm == NULL) {
767 		nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
768 		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr),
769 		    if_name(ifp), error));
770 		goto cleanup;
771 	}
772 	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
773 	*in6m_sol = imm->i6mm_maddr;
774 
775 	/*
776 	 * Join link-local all-nodes address.
777 	 */
778 	mltaddr = in6addr_linklocal_allnodes;
779 	if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
780 		goto cleanup; /* XXX: should not fail */
781 
782 	imm = in6_joingroup(ifp, &mltaddr, &error, 0);
783 	if (imm == NULL) {
784 		nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
785 		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr),
786 		    if_name(ifp), error));
787 		goto cleanup;
788 	}
789 	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
790 
791 	/*
792 	 * Join node information group address.
793 	 */
794 	delay = 0;
795 	if ((flags & IN6_IFAUPDATE_DADDELAY)) {
796 		/*
797 		 * The spec does not say anything about delay for this group,
798 		 * but the same logic should apply.
799 		 */
800 		delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
801 	}
802 	if (in6_nigroup(ifp, NULL, -1, &mltaddr) == 0) {
803 		/* XXX jinmei */
804 		imm = in6_joingroup(ifp, &mltaddr, &error, delay);
805 		if (imm == NULL)
806 			nd6log((LOG_WARNING,
807 			    "%s: in6_joingroup failed for %s on %s "
808 			    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
809 			    &mltaddr), if_name(ifp), error));
810 			/* XXX not very fatal, go on... */
811 		else
812 			LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
813 	}
814 	if (V_icmp6_nodeinfo_oldmcprefix &&
815 	    in6_nigroup_oldmcprefix(ifp, NULL, -1, &mltaddr) == 0) {
816 		imm = in6_joingroup(ifp, &mltaddr, &error, delay);
817 		if (imm == NULL)
818 			nd6log((LOG_WARNING,
819 			    "%s: in6_joingroup failed for %s on %s "
820 			    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
821 			    &mltaddr), if_name(ifp), error));
822 			/* XXX not very fatal, go on... */
823 		else
824 			LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
825 	}
826 
827 	/*
828 	 * Join interface-local all-nodes address.
829 	 * (ff01::1%ifN, and ff01::%ifN/32)
830 	 */
831 	mltaddr = in6addr_nodelocal_allnodes;
832 	if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
833 		goto cleanup; /* XXX: should not fail */
834 
835 	imm = in6_joingroup(ifp, &mltaddr, &error, 0);
836 	if (imm == NULL) {
837 		nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
838 		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
839 		    &mltaddr), if_name(ifp), error));
840 		goto cleanup;
841 	}
842 	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
843 
844 cleanup:
845 	return (error);
846 }
847 
848 /*
849  * Update parameters of an IPv6 interface address.
850  * If necessary, a new entry is created and linked into address chains.
851  * This function is separated from in6_control().
852  */
853 int
in6_update_ifa(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia,int flags)854 in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
855     struct in6_ifaddr *ia, int flags)
856 {
857 	int error, hostIsNew = 0;
858 
859 	if ((error = in6_validate_ifra(ifp, ifra, ia, flags)) != 0)
860 		return (error);
861 
862 	if (ia == NULL) {
863 		hostIsNew = 1;
864 		if ((ia = in6_alloc_ifa(ifp, ifra, flags)) == NULL)
865 			return (ENOBUFS);
866 	}
867 
868 	error = in6_update_ifa_internal(ifp, ifra, ia, hostIsNew, flags);
869 	if (error != 0) {
870 		if (hostIsNew != 0) {
871 			in6_unlink_ifa(ia, ifp);
872 			ifa_free(&ia->ia_ifa);
873 		}
874 		return (error);
875 	}
876 
877 	if (hostIsNew)
878 		error = in6_broadcast_ifa(ifp, ifra, ia, flags);
879 
880 	return (error);
881 }
882 
883 /*
884  * Fill in basic IPv6 address request info.
885  */
886 void
in6_prepare_ifra(struct in6_aliasreq * ifra,const struct in6_addr * addr,const struct in6_addr * mask)887 in6_prepare_ifra(struct in6_aliasreq *ifra, const struct in6_addr *addr,
888     const struct in6_addr *mask)
889 {
890 
891 	memset(ifra, 0, sizeof(struct in6_aliasreq));
892 
893 	ifra->ifra_addr.sin6_family = AF_INET6;
894 	ifra->ifra_addr.sin6_len = sizeof(struct sockaddr_in6);
895 	if (addr != NULL)
896 		ifra->ifra_addr.sin6_addr = *addr;
897 
898 	ifra->ifra_prefixmask.sin6_family = AF_INET6;
899 	ifra->ifra_prefixmask.sin6_len = sizeof(struct sockaddr_in6);
900 	if (mask != NULL)
901 		ifra->ifra_prefixmask.sin6_addr = *mask;
902 }
903 
904 static int
in6_validate_ifra(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia,int flags)905 in6_validate_ifra(struct ifnet *ifp, struct in6_aliasreq *ifra,
906     struct in6_ifaddr *ia, int flags)
907 {
908 	int plen = -1;
909 	struct sockaddr_in6 dst6;
910 	struct in6_addrlifetime *lt;
911 	char ip6buf[INET6_ADDRSTRLEN];
912 
913 	/* Validate parameters */
914 	if (ifp == NULL || ifra == NULL) /* this maybe redundant */
915 		return (EINVAL);
916 
917 	/*
918 	 * The destination address for a p2p link must have a family
919 	 * of AF_UNSPEC or AF_INET6.
920 	 */
921 	if ((ifp->if_flags & IFF_POINTOPOINT) != 0 &&
922 	    ifra->ifra_dstaddr.sin6_family != AF_INET6 &&
923 	    ifra->ifra_dstaddr.sin6_family != AF_UNSPEC)
924 		return (EAFNOSUPPORT);
925 
926 	/*
927 	 * Validate address
928 	 */
929 	if (ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6) ||
930 	    ifra->ifra_addr.sin6_family != AF_INET6)
931 		return (EINVAL);
932 
933 	/*
934 	 * validate ifra_prefixmask.  don't check sin6_family, netmask
935 	 * does not carry fields other than sin6_len.
936 	 */
937 	if (ifra->ifra_prefixmask.sin6_len > sizeof(struct sockaddr_in6))
938 		return (EINVAL);
939 	/*
940 	 * Because the IPv6 address architecture is classless, we require
941 	 * users to specify a (non 0) prefix length (mask) for a new address.
942 	 * We also require the prefix (when specified) mask is valid, and thus
943 	 * reject a non-consecutive mask.
944 	 */
945 	if (ia == NULL && ifra->ifra_prefixmask.sin6_len == 0)
946 		return (EINVAL);
947 	if (ifra->ifra_prefixmask.sin6_len != 0) {
948 		plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr,
949 		    (u_char *)&ifra->ifra_prefixmask +
950 		    ifra->ifra_prefixmask.sin6_len);
951 		if (plen <= 0)
952 			return (EINVAL);
953 	} else {
954 		/*
955 		 * In this case, ia must not be NULL.  We just use its prefix
956 		 * length.
957 		 */
958 		plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL);
959 	}
960 	/*
961 	 * If the destination address on a p2p interface is specified,
962 	 * and the address is a scoped one, validate/set the scope
963 	 * zone identifier.
964 	 */
965 	dst6 = ifra->ifra_dstaddr;
966 	if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) != 0 &&
967 	    (dst6.sin6_family == AF_INET6)) {
968 		struct in6_addr in6_tmp;
969 		u_int32_t zoneid;
970 
971 		in6_tmp = dst6.sin6_addr;
972 		if (in6_setscope(&in6_tmp, ifp, &zoneid))
973 			return (EINVAL); /* XXX: should be impossible */
974 
975 		if (dst6.sin6_scope_id != 0) {
976 			if (dst6.sin6_scope_id != zoneid)
977 				return (EINVAL);
978 		} else		/* user omit to specify the ID. */
979 			dst6.sin6_scope_id = zoneid;
980 
981 		/* convert into the internal form */
982 		if (sa6_embedscope(&dst6, 0))
983 			return (EINVAL); /* XXX: should be impossible */
984 	}
985 	/* Modify original ifra_dstaddr to reflect changes */
986 	ifra->ifra_dstaddr = dst6;
987 
988 	/*
989 	 * The destination address can be specified only for a p2p or a
990 	 * loopback interface.  If specified, the corresponding prefix length
991 	 * must be 128.
992 	 */
993 	if (ifra->ifra_dstaddr.sin6_family == AF_INET6) {
994 		if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) == 0) {
995 			/* XXX: noisy message */
996 			nd6log((LOG_INFO, "in6_update_ifa: a destination can "
997 			    "be specified for a p2p or a loopback IF only\n"));
998 			return (EINVAL);
999 		}
1000 		if (plen != 128) {
1001 			nd6log((LOG_INFO, "in6_update_ifa: prefixlen should "
1002 			    "be 128 when dstaddr is specified\n"));
1003 			return (EINVAL);
1004 		}
1005 	}
1006 	/* lifetime consistency check */
1007 	lt = &ifra->ifra_lifetime;
1008 	if (lt->ia6t_pltime > lt->ia6t_vltime)
1009 		return (EINVAL);
1010 	if (lt->ia6t_vltime == 0) {
1011 		/*
1012 		 * the following log might be noisy, but this is a typical
1013 		 * configuration mistake or a tool's bug.
1014 		 */
1015 		nd6log((LOG_INFO,
1016 		    "in6_update_ifa: valid lifetime is 0 for %s\n",
1017 		    ip6_sprintf(ip6buf, &ifra->ifra_addr.sin6_addr)));
1018 
1019 		if (ia == NULL)
1020 			return (0); /* there's nothing to do */
1021 	}
1022 
1023 	/* Check prefix mask */
1024 	if (ia != NULL && ifra->ifra_prefixmask.sin6_len != 0) {
1025 		/*
1026 		 * We prohibit changing the prefix length of an existing
1027 		 * address, because
1028 		 * + such an operation should be rare in IPv6, and
1029 		 * + the operation would confuse prefix management.
1030 		 */
1031 		if (ia->ia_prefixmask.sin6_len != 0 &&
1032 		    in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL) != plen) {
1033 			nd6log((LOG_INFO, "in6_validate_ifa: the prefix length "
1034 			    "of an existing %s address should not be changed\n",
1035 			    ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
1036 
1037 			return (EINVAL);
1038 		}
1039 	}
1040 
1041 	return (0);
1042 }
1043 
1044 
1045 /*
1046  * Allocate a new ifaddr and link it into chains.
1047  */
1048 static struct in6_ifaddr *
in6_alloc_ifa(struct ifnet * ifp,struct in6_aliasreq * ifra,int flags)1049 in6_alloc_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, int flags)
1050 {
1051 	struct in6_ifaddr *ia;
1052 
1053 	/*
1054 	 * When in6_alloc_ifa() is called in a process of a received
1055 	 * RA, it is called under an interrupt context.  So, we should
1056 	 * call malloc with M_NOWAIT.
1057 	 */
1058 	ia = (struct in6_ifaddr *)ifa_alloc(sizeof(*ia), M_NOWAIT);
1059 	if (ia == NULL)
1060 		return (NULL);
1061 	LIST_INIT(&ia->ia6_memberships);
1062 	/* Initialize the address and masks, and put time stamp */
1063 	ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr;
1064 	ia->ia_addr.sin6_family = AF_INET6;
1065 	ia->ia_addr.sin6_len = sizeof(ia->ia_addr);
1066 	/* XXX: Can we assign ,sin6_addr and skip the rest? */
1067 	ia->ia_addr = ifra->ifra_addr;
1068 	ia->ia6_createtime = time_uptime;
1069 	if ((ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) != 0) {
1070 		/*
1071 		 * Some functions expect that ifa_dstaddr is not
1072 		 * NULL for p2p interfaces.
1073 		 */
1074 		ia->ia_ifa.ifa_dstaddr =
1075 		    (struct sockaddr *)&ia->ia_dstaddr;
1076 	} else {
1077 		ia->ia_ifa.ifa_dstaddr = NULL;
1078 	}
1079 
1080 	/* set prefix mask if any */
1081 	ia->ia_ifa.ifa_netmask = (struct sockaddr *)&ia->ia_prefixmask;
1082 	if (ifra->ifra_prefixmask.sin6_len != 0) {
1083 		ia->ia_prefixmask.sin6_family = AF_INET6;
1084 		ia->ia_prefixmask.sin6_len = ifra->ifra_prefixmask.sin6_len;
1085 		ia->ia_prefixmask.sin6_addr = ifra->ifra_prefixmask.sin6_addr;
1086 	}
1087 
1088 	ia->ia_ifp = ifp;
1089 	ifa_ref(&ia->ia_ifa);			/* if_addrhead */
1090 	IF_ADDR_WLOCK(ifp);
1091 	TAILQ_INSERT_TAIL(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
1092 	IF_ADDR_WUNLOCK(ifp);
1093 
1094 	ifa_ref(&ia->ia_ifa);			/* in6_ifaddrhead */
1095 	IN6_IFADDR_WLOCK();
1096 	TAILQ_INSERT_TAIL(&V_in6_ifaddrhead, ia, ia_link);
1097 	LIST_INSERT_HEAD(IN6ADDR_HASH(&ia->ia_addr.sin6_addr), ia, ia6_hash);
1098 	IN6_IFADDR_WUNLOCK();
1099 
1100 	return (ia);
1101 }
1102 
1103 /*
1104  * Update/configure interface address parameters:
1105  *
1106  * 1) Update lifetime
1107  * 2) Update interface metric ad flags
1108  * 3) Notify other subsystems
1109  */
1110 static int
in6_update_ifa_internal(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia,int hostIsNew,int flags)1111 in6_update_ifa_internal(struct ifnet *ifp, struct in6_aliasreq *ifra,
1112     struct in6_ifaddr *ia, int hostIsNew, int flags)
1113 {
1114 	int error;
1115 
1116 	/* update timestamp */
1117 	ia->ia6_updatetime = time_uptime;
1118 
1119 	/*
1120 	 * Set lifetimes.  We do not refer to ia6t_expire and ia6t_preferred
1121 	 * to see if the address is deprecated or invalidated, but initialize
1122 	 * these members for applications.
1123 	 */
1124 	ia->ia6_lifetime = ifra->ifra_lifetime;
1125 	if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
1126 		ia->ia6_lifetime.ia6t_expire =
1127 		    time_uptime + ia->ia6_lifetime.ia6t_vltime;
1128 	} else
1129 		ia->ia6_lifetime.ia6t_expire = 0;
1130 	if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
1131 		ia->ia6_lifetime.ia6t_preferred =
1132 		    time_uptime + ia->ia6_lifetime.ia6t_pltime;
1133 	} else
1134 		ia->ia6_lifetime.ia6t_preferred = 0;
1135 
1136 	/*
1137 	 * backward compatibility - if IN6_IFF_DEPRECATED is set from the
1138 	 * userland, make it deprecated.
1139 	 */
1140 	if ((ifra->ifra_flags & IN6_IFF_DEPRECATED) != 0) {
1141 		ia->ia6_lifetime.ia6t_pltime = 0;
1142 		ia->ia6_lifetime.ia6t_preferred = time_uptime;
1143 	}
1144 
1145 	/*
1146 	 * configure address flags.
1147 	 */
1148 	ia->ia6_flags = ifra->ifra_flags;
1149 
1150 	/*
1151 	 * Make the address tentative before joining multicast addresses,
1152 	 * so that corresponding MLD responses would not have a tentative
1153 	 * source address.
1154 	 */
1155 	ia->ia6_flags &= ~IN6_IFF_DUPLICATED;	/* safety */
1156 
1157 	/*
1158 	 * DAD should be performed for an new address or addresses on
1159 	 * an interface with ND6_IFF_IFDISABLED.
1160 	 */
1161 	if (in6if_do_dad(ifp) &&
1162 	    (hostIsNew || (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)))
1163 		ia->ia6_flags |= IN6_IFF_TENTATIVE;
1164 
1165 	/* notify other subsystems */
1166 	error = in6_notify_ifa(ifp, ia, ifra, hostIsNew);
1167 
1168 	return (error);
1169 }
1170 
1171 /*
1172  * Do link-level ifa job:
1173  * 1) Add lle entry for added address
1174  * 2) Notifies routing socket users about new address
1175  * 3) join appropriate multicast group
1176  * 4) start DAD if enabled
1177  */
1178 static int
in6_broadcast_ifa(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia,int flags)1179 in6_broadcast_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
1180     struct in6_ifaddr *ia, int flags)
1181 {
1182 	struct in6_multi *in6m_sol;
1183 	int error = 0;
1184 
1185 	/* Add local address to lltable, if necessary (ex. on p2p link). */
1186 	if ((error = nd6_add_ifa_lle(ia)) != 0) {
1187 		in6_purgeaddr(&ia->ia_ifa);
1188 		ifa_free(&ia->ia_ifa);
1189 		return (error);
1190 	}
1191 
1192 	/* Join necessary multicast groups. */
1193 	in6m_sol = NULL;
1194 	if ((ifp->if_flags & IFF_MULTICAST) != 0) {
1195 		error = in6_update_ifa_join_mc(ifp, ifra, ia, flags, &in6m_sol);
1196 		if (error != 0) {
1197 			in6_purgeaddr(&ia->ia_ifa);
1198 			ifa_free(&ia->ia_ifa);
1199 			return (error);
1200 		}
1201 	}
1202 
1203 	/* Perform DAD, if the address is TENTATIVE. */
1204 	if ((ia->ia6_flags & IN6_IFF_TENTATIVE)) {
1205 		int delay, mindelay, maxdelay;
1206 
1207 		delay = 0;
1208 		if ((flags & IN6_IFAUPDATE_DADDELAY)) {
1209 			/*
1210 			 * We need to impose a delay before sending an NS
1211 			 * for DAD.  Check if we also needed a delay for the
1212 			 * corresponding MLD message.  If we did, the delay
1213 			 * should be larger than the MLD delay (this could be
1214 			 * relaxed a bit, but this simple logic is at least
1215 			 * safe).
1216 			 * XXX: Break data hiding guidelines and look at
1217 			 * state for the solicited multicast group.
1218 			 */
1219 			mindelay = 0;
1220 			if (in6m_sol != NULL &&
1221 			    in6m_sol->in6m_state == MLD_REPORTING_MEMBER) {
1222 				mindelay = in6m_sol->in6m_timer;
1223 			}
1224 			maxdelay = MAX_RTR_SOLICITATION_DELAY * hz;
1225 			if (maxdelay - mindelay == 0)
1226 				delay = 0;
1227 			else {
1228 				delay =
1229 				    (arc4random() % (maxdelay - mindelay)) +
1230 				    mindelay;
1231 			}
1232 		}
1233 		nd6_dad_start((struct ifaddr *)ia, delay);
1234 	}
1235 
1236 	in6_newaddrmsg(ia, RTM_ADD);
1237 	ifa_free(&ia->ia_ifa);
1238 	return (error);
1239 }
1240 
1241 void
in6_purgeaddr(struct ifaddr * ifa)1242 in6_purgeaddr(struct ifaddr *ifa)
1243 {
1244 	struct ifnet *ifp = ifa->ifa_ifp;
1245 	struct in6_ifaddr *ia = (struct in6_ifaddr *) ifa;
1246 	struct in6_multi_mship *imm;
1247 	int plen, error;
1248 
1249 	if (ifa->ifa_carp)
1250 		(*carp_detach_p)(ifa, false);
1251 
1252 	/*
1253 	 * Remove the loopback route to the interface address.
1254 	 * The check for the current setting of "nd6_useloopback"
1255 	 * is not needed.
1256 	 */
1257 	if (ia->ia_flags & IFA_RTSELF) {
1258 		error = ifa_del_loopback_route((struct ifaddr *)ia,
1259 		    (struct sockaddr *)&ia->ia_addr);
1260 		if (error == 0)
1261 			ia->ia_flags &= ~IFA_RTSELF;
1262 	}
1263 
1264 	/* stop DAD processing */
1265 	nd6_dad_stop(ifa);
1266 
1267 	/* Leave multicast groups. */
1268 	while ((imm = LIST_FIRST(&ia->ia6_memberships)) != NULL) {
1269 		LIST_REMOVE(imm, i6mm_chain);
1270 		in6_leavegroup(imm);
1271 	}
1272 	plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */
1273 	if ((ia->ia_flags & IFA_ROUTE) && plen == 128) {
1274 		error = rtinit(&(ia->ia_ifa), RTM_DELETE, ia->ia_flags |
1275 		    (ia->ia_dstaddr.sin6_family == AF_INET6 ? RTF_HOST : 0));
1276 		if (error != 0)
1277 			log(LOG_INFO, "%s: err=%d, destination address delete "
1278 			    "failed\n", __func__, error);
1279 		ia->ia_flags &= ~IFA_ROUTE;
1280 	}
1281 
1282 	in6_newaddrmsg(ia, RTM_DELETE);
1283 	in6_unlink_ifa(ia, ifp);
1284 }
1285 
1286 static void
in6_unlink_ifa(struct in6_ifaddr * ia,struct ifnet * ifp)1287 in6_unlink_ifa(struct in6_ifaddr *ia, struct ifnet *ifp)
1288 {
1289 	char ip6buf[INET6_ADDRSTRLEN];
1290 	int remove_lle;
1291 
1292 	IF_ADDR_WLOCK(ifp);
1293 	TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
1294 	IF_ADDR_WUNLOCK(ifp);
1295 	ifa_free(&ia->ia_ifa);			/* if_addrhead */
1296 
1297 	/*
1298 	 * Defer the release of what might be the last reference to the
1299 	 * in6_ifaddr so that it can't be freed before the remainder of the
1300 	 * cleanup.
1301 	 */
1302 	IN6_IFADDR_WLOCK();
1303 	TAILQ_REMOVE(&V_in6_ifaddrhead, ia, ia_link);
1304 	LIST_REMOVE(ia, ia6_hash);
1305 	IN6_IFADDR_WUNLOCK();
1306 
1307 	/*
1308 	 * Release the reference to the base prefix.  There should be a
1309 	 * positive reference.
1310 	 */
1311 	remove_lle = 0;
1312 	if (ia->ia6_ndpr == NULL) {
1313 		nd6log((LOG_NOTICE,
1314 		    "in6_unlink_ifa: autoconf'ed address "
1315 		    "%s has no prefix\n", ip6_sprintf(ip6buf, IA6_IN6(ia))));
1316 	} else {
1317 		ia->ia6_ndpr->ndpr_addrcnt--;
1318 		/* Do not delete lles within prefix if refcont != 0 */
1319 		if (ia->ia6_ndpr->ndpr_addrcnt == 0)
1320 			remove_lle = 1;
1321 		ia->ia6_ndpr = NULL;
1322 	}
1323 
1324 	nd6_rem_ifa_lle(ia, remove_lle);
1325 
1326 	/*
1327 	 * Also, if the address being removed is autoconf'ed, call
1328 	 * pfxlist_onlink_check() since the release might affect the status of
1329 	 * other (detached) addresses.
1330 	 */
1331 	if ((ia->ia6_flags & IN6_IFF_AUTOCONF)) {
1332 		pfxlist_onlink_check();
1333 	}
1334 	ifa_free(&ia->ia_ifa);			/* in6_ifaddrhead */
1335 }
1336 
1337 /*
1338  * Notifies other subsystems about address change/arrival:
1339  * 1) Notifies device handler on the first IPv6 address assignment
1340  * 2) Handle routing table changes for P2P links and route
1341  * 3) Handle routing table changes for address host route
1342  */
1343 static int
in6_notify_ifa(struct ifnet * ifp,struct in6_ifaddr * ia,struct in6_aliasreq * ifra,int hostIsNew)1344 in6_notify_ifa(struct ifnet *ifp, struct in6_ifaddr *ia,
1345     struct in6_aliasreq *ifra, int hostIsNew)
1346 {
1347 	int	error = 0, plen, ifacount = 0;
1348 	struct ifaddr *ifa;
1349 	struct sockaddr_in6 *pdst;
1350 	char ip6buf[INET6_ADDRSTRLEN];
1351 
1352 	/*
1353 	 * Give the interface a chance to initialize
1354 	 * if this is its first address,
1355 	 */
1356 	if (hostIsNew != 0) {
1357 		IF_ADDR_RLOCK(ifp);
1358 		TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1359 			if (ifa->ifa_addr->sa_family != AF_INET6)
1360 				continue;
1361 			ifacount++;
1362 		}
1363 		IF_ADDR_RUNLOCK(ifp);
1364 	}
1365 
1366 	if (ifacount <= 1 && ifp->if_ioctl) {
1367 		error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
1368 		if (error)
1369 			goto done;
1370 	}
1371 
1372 	/*
1373 	 * If a new destination address is specified, scrub the old one and
1374 	 * install the new destination.  Note that the interface must be
1375 	 * p2p or loopback.
1376 	 */
1377 	pdst = &ifra->ifra_dstaddr;
1378 	if (pdst->sin6_family == AF_INET6 &&
1379 	    !IN6_ARE_ADDR_EQUAL(&pdst->sin6_addr, &ia->ia_dstaddr.sin6_addr)) {
1380 		if ((ia->ia_flags & IFA_ROUTE) != 0 &&
1381 		    (rtinit(&(ia->ia_ifa), (int)RTM_DELETE, RTF_HOST) != 0)) {
1382 			nd6log((LOG_ERR, "in6_update_ifa_internal: failed to "
1383 			    "remove a route to the old destination: %s\n",
1384 			    ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
1385 			/* proceed anyway... */
1386 		} else
1387 			ia->ia_flags &= ~IFA_ROUTE;
1388 		ia->ia_dstaddr = *pdst;
1389 	}
1390 
1391 	/*
1392 	 * If a new destination address is specified for a point-to-point
1393 	 * interface, install a route to the destination as an interface
1394 	 * direct route.
1395 	 * XXX: the logic below rejects assigning multiple addresses on a p2p
1396 	 * interface that share the same destination.
1397 	 */
1398 	plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */
1399 	if (!(ia->ia_flags & IFA_ROUTE) && plen == 128 &&
1400 	    ia->ia_dstaddr.sin6_family == AF_INET6) {
1401 		int rtflags = RTF_UP | RTF_HOST;
1402 		/*
1403 		 * Handle the case for ::1 .
1404 		 */
1405 		if (ifp->if_flags & IFF_LOOPBACK)
1406 			ia->ia_flags |= IFA_RTSELF;
1407 		error = rtinit(&ia->ia_ifa, RTM_ADD, ia->ia_flags | rtflags);
1408 		if (error)
1409 			goto done;
1410 		ia->ia_flags |= IFA_ROUTE;
1411 	}
1412 
1413 	/*
1414 	 * add a loopback route to self if not exists
1415 	 */
1416 	if (!(ia->ia_flags & IFA_RTSELF) && V_nd6_useloopback) {
1417 		error = ifa_add_loopback_route((struct ifaddr *)ia,
1418 		    (struct sockaddr *)&ia->ia_addr);
1419 		if (error == 0)
1420 			ia->ia_flags |= IFA_RTSELF;
1421 	}
1422 done:
1423 	WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
1424 	    "Invoking IPv6 network device address event may sleep");
1425 
1426 	EVENTHANDLER_INVOKE(ifaddr_event, ifp);
1427 
1428 	return (error);
1429 }
1430 
1431 /*
1432  * Find an IPv6 interface link-local address specific to an interface.
1433  * ifaddr is returned referenced.
1434  */
1435 struct in6_ifaddr *
in6ifa_ifpforlinklocal(struct ifnet * ifp,int ignoreflags)1436 in6ifa_ifpforlinklocal(struct ifnet *ifp, int ignoreflags)
1437 {
1438 	struct ifaddr *ifa;
1439 
1440 	IF_ADDR_RLOCK(ifp);
1441 	TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1442 		if (ifa->ifa_addr->sa_family != AF_INET6)
1443 			continue;
1444 		if (IN6_IS_ADDR_LINKLOCAL(IFA_IN6(ifa))) {
1445 			if ((((struct in6_ifaddr *)ifa)->ia6_flags &
1446 			    ignoreflags) != 0)
1447 				continue;
1448 			ifa_ref(ifa);
1449 			break;
1450 		}
1451 	}
1452 	IF_ADDR_RUNLOCK(ifp);
1453 
1454 	return ((struct in6_ifaddr *)ifa);
1455 }
1456 
1457 
1458 /*
1459  * find the interface address corresponding to a given IPv6 address.
1460  * ifaddr is returned referenced.
1461  */
1462 struct in6_ifaddr *
in6ifa_ifwithaddr(const struct in6_addr * addr,uint32_t zoneid)1463 in6ifa_ifwithaddr(const struct in6_addr *addr, uint32_t zoneid)
1464 {
1465 	struct rm_priotracker in6_ifa_tracker;
1466 	struct in6_ifaddr *ia;
1467 
1468 	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1469 	LIST_FOREACH(ia, IN6ADDR_HASH(addr), ia6_hash) {
1470 		if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), addr)) {
1471 			if (zoneid != 0 &&
1472 			    zoneid != ia->ia_addr.sin6_scope_id)
1473 				continue;
1474 			ifa_ref(&ia->ia_ifa);
1475 			break;
1476 		}
1477 	}
1478 	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1479 	return (ia);
1480 }
1481 
1482 /*
1483  * find the internet address corresponding to a given interface and address.
1484  * ifaddr is returned referenced.
1485  */
1486 struct in6_ifaddr *
in6ifa_ifpwithaddr(struct ifnet * ifp,const struct in6_addr * addr)1487 in6ifa_ifpwithaddr(struct ifnet *ifp, const struct in6_addr *addr)
1488 {
1489 	struct ifaddr *ifa;
1490 
1491 	IF_ADDR_RLOCK(ifp);
1492 	TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1493 		if (ifa->ifa_addr->sa_family != AF_INET6)
1494 			continue;
1495 		if (IN6_ARE_ADDR_EQUAL(addr, IFA_IN6(ifa))) {
1496 			ifa_ref(ifa);
1497 			break;
1498 		}
1499 	}
1500 	IF_ADDR_RUNLOCK(ifp);
1501 
1502 	return ((struct in6_ifaddr *)ifa);
1503 }
1504 
1505 /*
1506  * Find a link-local scoped address on ifp and return it if any.
1507  */
1508 struct in6_ifaddr *
in6ifa_llaonifp(struct ifnet * ifp)1509 in6ifa_llaonifp(struct ifnet *ifp)
1510 {
1511 	struct sockaddr_in6 *sin6;
1512 	struct ifaddr *ifa;
1513 
1514 	if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)
1515 		return (NULL);
1516 	IF_ADDR_RLOCK(ifp);
1517 	TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1518 		if (ifa->ifa_addr->sa_family != AF_INET6)
1519 			continue;
1520 		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
1521 		if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr) ||
1522 		    IN6_IS_ADDR_MC_INTFACELOCAL(&sin6->sin6_addr) ||
1523 		    IN6_IS_ADDR_MC_NODELOCAL(&sin6->sin6_addr))
1524 			break;
1525 	}
1526 	IF_ADDR_RUNLOCK(ifp);
1527 
1528 	return ((struct in6_ifaddr *)ifa);
1529 }
1530 
1531 /*
1532  * Convert IP6 address to printable (loggable) representation. Caller
1533  * has to make sure that ip6buf is at least INET6_ADDRSTRLEN long.
1534  */
1535 static char digits[] = "0123456789abcdef";
1536 char *
ip6_sprintf(char * ip6buf,const struct in6_addr * addr)1537 ip6_sprintf(char *ip6buf, const struct in6_addr *addr)
1538 {
1539 	int i, cnt = 0, maxcnt = 0, idx = 0, index = 0;
1540 	char *cp;
1541 	const u_int16_t *a = (const u_int16_t *)addr;
1542 	const u_int8_t *d;
1543 	int dcolon = 0, zero = 0;
1544 
1545 	cp = ip6buf;
1546 
1547 	for (i = 0; i < 8; i++) {
1548 		if (*(a + i) == 0) {
1549 			cnt++;
1550 			if (cnt == 1)
1551 				idx = i;
1552 		}
1553 		else if (maxcnt < cnt) {
1554 			maxcnt = cnt;
1555 			index = idx;
1556 			cnt = 0;
1557 		}
1558 	}
1559 	if (maxcnt < cnt) {
1560 		maxcnt = cnt;
1561 		index = idx;
1562 	}
1563 
1564 	for (i = 0; i < 8; i++) {
1565 		if (dcolon == 1) {
1566 			if (*a == 0) {
1567 				if (i == 7)
1568 					*cp++ = ':';
1569 				a++;
1570 				continue;
1571 			} else
1572 				dcolon = 2;
1573 		}
1574 		if (*a == 0) {
1575 			if (dcolon == 0 && *(a + 1) == 0 && i == index) {
1576 				if (i == 0)
1577 					*cp++ = ':';
1578 				*cp++ = ':';
1579 				dcolon = 1;
1580 			} else {
1581 				*cp++ = '0';
1582 				*cp++ = ':';
1583 			}
1584 			a++;
1585 			continue;
1586 		}
1587 		d = (const u_char *)a;
1588 		/* Try to eliminate leading zeros in printout like in :0001. */
1589 		zero = 1;
1590 		*cp = digits[*d >> 4];
1591 		if (*cp != '0') {
1592 			zero = 0;
1593 			cp++;
1594 		}
1595 		*cp = digits[*d++ & 0xf];
1596 		if (zero == 0 || (*cp != '0')) {
1597 			zero = 0;
1598 			cp++;
1599 		}
1600 		*cp = digits[*d >> 4];
1601 		if (zero == 0 || (*cp != '0')) {
1602 			zero = 0;
1603 			cp++;
1604 		}
1605 		*cp++ = digits[*d & 0xf];
1606 		*cp++ = ':';
1607 		a++;
1608 	}
1609 	*--cp = '\0';
1610 	return (ip6buf);
1611 }
1612 
1613 int
in6_localaddr(struct in6_addr * in6)1614 in6_localaddr(struct in6_addr *in6)
1615 {
1616 	struct rm_priotracker in6_ifa_tracker;
1617 	struct in6_ifaddr *ia;
1618 
1619 	if (IN6_IS_ADDR_LOOPBACK(in6) || IN6_IS_ADDR_LINKLOCAL(in6))
1620 		return 1;
1621 
1622 	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1623 	TAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) {
1624 		if (IN6_ARE_MASKED_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr,
1625 		    &ia->ia_prefixmask.sin6_addr)) {
1626 			IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1627 			return 1;
1628 		}
1629 	}
1630 	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1631 
1632 	return (0);
1633 }
1634 
1635 /*
1636  * Return 1 if an internet address is for the local host and configured
1637  * on one of its interfaces.
1638  */
1639 int
in6_localip(struct in6_addr * in6)1640 in6_localip(struct in6_addr *in6)
1641 {
1642 	struct rm_priotracker in6_ifa_tracker;
1643 	struct in6_ifaddr *ia;
1644 
1645 	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1646 	LIST_FOREACH(ia, IN6ADDR_HASH(in6), ia6_hash) {
1647 		if (IN6_ARE_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr)) {
1648 			IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1649 			return (1);
1650 		}
1651 	}
1652 	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1653 	return (0);
1654 }
1655 
1656 /*
1657  * Return 1 if an internet address is configured on an interface.
1658  */
1659 int
in6_ifhasaddr(struct ifnet * ifp,struct in6_addr * addr)1660 in6_ifhasaddr(struct ifnet *ifp, struct in6_addr *addr)
1661 {
1662 	struct in6_addr in6;
1663 	struct ifaddr *ifa;
1664 	struct in6_ifaddr *ia6;
1665 
1666 	in6 = *addr;
1667 	if (in6_clearscope(&in6))
1668 		return (0);
1669 	in6_setscope(&in6, ifp, NULL);
1670 
1671 	IF_ADDR_RLOCK(ifp);
1672 	TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1673 		if (ifa->ifa_addr->sa_family != AF_INET6)
1674 			continue;
1675 		ia6 = (struct in6_ifaddr *)ifa;
1676 		if (IN6_ARE_ADDR_EQUAL(&ia6->ia_addr.sin6_addr, &in6)) {
1677 			IF_ADDR_RUNLOCK(ifp);
1678 			return (1);
1679 		}
1680 	}
1681 	IF_ADDR_RUNLOCK(ifp);
1682 
1683 	return (0);
1684 }
1685 
1686 int
in6_is_addr_deprecated(struct sockaddr_in6 * sa6)1687 in6_is_addr_deprecated(struct sockaddr_in6 *sa6)
1688 {
1689 	struct rm_priotracker in6_ifa_tracker;
1690 	struct in6_ifaddr *ia;
1691 
1692 	IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1693 	LIST_FOREACH(ia, IN6ADDR_HASH(&sa6->sin6_addr), ia6_hash) {
1694 		if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), &sa6->sin6_addr)) {
1695 			if (ia->ia6_flags & IN6_IFF_DEPRECATED) {
1696 				IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1697 				return (1); /* true */
1698 			}
1699 			break;
1700 		}
1701 	}
1702 	IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1703 
1704 	return (0);		/* false */
1705 }
1706 
1707 /*
1708  * return length of part which dst and src are equal
1709  * hard coding...
1710  */
1711 int
in6_matchlen(struct in6_addr * src,struct in6_addr * dst)1712 in6_matchlen(struct in6_addr *src, struct in6_addr *dst)
1713 {
1714 	int match = 0;
1715 	u_char *s = (u_char *)src, *d = (u_char *)dst;
1716 	u_char *lim = s + 16, r;
1717 
1718 	while (s < lim)
1719 		if ((r = (*d++ ^ *s++)) != 0) {
1720 			while (r < 128) {
1721 				match++;
1722 				r <<= 1;
1723 			}
1724 			break;
1725 		} else
1726 			match += 8;
1727 	return match;
1728 }
1729 
1730 /* XXX: to be scope conscious */
1731 int
in6_are_prefix_equal(struct in6_addr * p1,struct in6_addr * p2,int len)1732 in6_are_prefix_equal(struct in6_addr *p1, struct in6_addr *p2, int len)
1733 {
1734 	int bytelen, bitlen;
1735 
1736 	/* sanity check */
1737 	if (0 > len || len > 128) {
1738 		log(LOG_ERR, "in6_are_prefix_equal: invalid prefix length(%d)\n",
1739 		    len);
1740 		return (0);
1741 	}
1742 
1743 	bytelen = len / 8;
1744 	bitlen = len % 8;
1745 
1746 	if (bcmp(&p1->s6_addr, &p2->s6_addr, bytelen))
1747 		return (0);
1748 	if (bitlen != 0 &&
1749 	    p1->s6_addr[bytelen] >> (8 - bitlen) !=
1750 	    p2->s6_addr[bytelen] >> (8 - bitlen))
1751 		return (0);
1752 
1753 	return (1);
1754 }
1755 
1756 void
in6_prefixlen2mask(struct in6_addr * maskp,int len)1757 in6_prefixlen2mask(struct in6_addr *maskp, int len)
1758 {
1759 	u_char maskarray[8] = {0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, 0xff};
1760 	int bytelen, bitlen, i;
1761 
1762 	/* sanity check */
1763 	if (0 > len || len > 128) {
1764 		log(LOG_ERR, "in6_prefixlen2mask: invalid prefix length(%d)\n",
1765 		    len);
1766 		return;
1767 	}
1768 
1769 	bzero(maskp, sizeof(*maskp));
1770 	bytelen = len / 8;
1771 	bitlen = len % 8;
1772 	for (i = 0; i < bytelen; i++)
1773 		maskp->s6_addr[i] = 0xff;
1774 	if (bitlen)
1775 		maskp->s6_addr[bytelen] = maskarray[bitlen - 1];
1776 }
1777 
1778 /*
1779  * return the best address out of the same scope. if no address was
1780  * found, return the first valid address from designated IF.
1781  */
1782 struct in6_ifaddr *
in6_ifawithifp(struct ifnet * ifp,struct in6_addr * dst)1783 in6_ifawithifp(struct ifnet *ifp, struct in6_addr *dst)
1784 {
1785 	int dst_scope =	in6_addrscope(dst), blen = -1, tlen;
1786 	struct ifaddr *ifa;
1787 	struct in6_ifaddr *besta = NULL;
1788 	struct in6_ifaddr *dep[2];	/* last-resort: deprecated */
1789 
1790 	dep[0] = dep[1] = NULL;
1791 
1792 	/*
1793 	 * We first look for addresses in the same scope.
1794 	 * If there is one, return it.
1795 	 * If two or more, return one which matches the dst longest.
1796 	 * If none, return one of global addresses assigned other ifs.
1797 	 */
1798 	IF_ADDR_RLOCK(ifp);
1799 	TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1800 		if (ifa->ifa_addr->sa_family != AF_INET6)
1801 			continue;
1802 		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
1803 			continue; /* XXX: is there any case to allow anycast? */
1804 		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
1805 			continue; /* don't use this interface */
1806 		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
1807 			continue;
1808 		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
1809 			if (V_ip6_use_deprecated)
1810 				dep[0] = (struct in6_ifaddr *)ifa;
1811 			continue;
1812 		}
1813 
1814 		if (dst_scope == in6_addrscope(IFA_IN6(ifa))) {
1815 			/*
1816 			 * call in6_matchlen() as few as possible
1817 			 */
1818 			if (besta) {
1819 				if (blen == -1)
1820 					blen = in6_matchlen(&besta->ia_addr.sin6_addr, dst);
1821 				tlen = in6_matchlen(IFA_IN6(ifa), dst);
1822 				if (tlen > blen) {
1823 					blen = tlen;
1824 					besta = (struct in6_ifaddr *)ifa;
1825 				}
1826 			} else
1827 				besta = (struct in6_ifaddr *)ifa;
1828 		}
1829 	}
1830 	if (besta) {
1831 		ifa_ref(&besta->ia_ifa);
1832 		IF_ADDR_RUNLOCK(ifp);
1833 		return (besta);
1834 	}
1835 
1836 	TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1837 		if (ifa->ifa_addr->sa_family != AF_INET6)
1838 			continue;
1839 		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
1840 			continue; /* XXX: is there any case to allow anycast? */
1841 		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
1842 			continue; /* don't use this interface */
1843 		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
1844 			continue;
1845 		if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
1846 			if (V_ip6_use_deprecated)
1847 				dep[1] = (struct in6_ifaddr *)ifa;
1848 			continue;
1849 		}
1850 
1851 		if (ifa != NULL)
1852 			ifa_ref(ifa);
1853 		IF_ADDR_RUNLOCK(ifp);
1854 		return (struct in6_ifaddr *)ifa;
1855 	}
1856 
1857 	/* use the last-resort values, that are, deprecated addresses */
1858 	if (dep[0]) {
1859 		ifa_ref((struct ifaddr *)dep[0]);
1860 		IF_ADDR_RUNLOCK(ifp);
1861 		return dep[0];
1862 	}
1863 	if (dep[1]) {
1864 		ifa_ref((struct ifaddr *)dep[1]);
1865 		IF_ADDR_RUNLOCK(ifp);
1866 		return dep[1];
1867 	}
1868 
1869 	IF_ADDR_RUNLOCK(ifp);
1870 	return NULL;
1871 }
1872 
1873 /*
1874  * perform DAD when interface becomes IFF_UP.
1875  */
1876 void
in6_if_up(struct ifnet * ifp)1877 in6_if_up(struct ifnet *ifp)
1878 {
1879 	struct ifaddr *ifa;
1880 	struct in6_ifaddr *ia;
1881 
1882 	IF_ADDR_RLOCK(ifp);
1883 	TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1884 		if (ifa->ifa_addr->sa_family != AF_INET6)
1885 			continue;
1886 		ia = (struct in6_ifaddr *)ifa;
1887 		if (ia->ia6_flags & IN6_IFF_TENTATIVE) {
1888 			/*
1889 			 * The TENTATIVE flag was likely set by hand
1890 			 * beforehand, implicitly indicating the need for DAD.
1891 			 * We may be able to skip the random delay in this
1892 			 * case, but we impose delays just in case.
1893 			 */
1894 			nd6_dad_start(ifa,
1895 			    arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz));
1896 		}
1897 	}
1898 	IF_ADDR_RUNLOCK(ifp);
1899 
1900 	/*
1901 	 * special cases, like 6to4, are handled in in6_ifattach
1902 	 */
1903 	in6_ifattach(ifp, NULL);
1904 }
1905 
1906 int
in6if_do_dad(struct ifnet * ifp)1907 in6if_do_dad(struct ifnet *ifp)
1908 {
1909 	if ((ifp->if_flags & IFF_LOOPBACK) != 0)
1910 		return (0);
1911 
1912 	if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) ||
1913 	    (ND_IFINFO(ifp)->flags & ND6_IFF_NO_DAD))
1914 		return (0);
1915 
1916 	/*
1917 	 * Our DAD routine requires the interface up and running.
1918 	 * However, some interfaces can be up before the RUNNING
1919 	 * status.  Additionally, users may try to assign addresses
1920 	 * before the interface becomes up (or running).
1921 	 * This function returns EAGAIN in that case.
1922 	 * The caller should mark "tentative" on the address instead of
1923 	 * performing DAD immediately.
1924 	 */
1925 	if (!((ifp->if_flags & IFF_UP) &&
1926 	    (ifp->if_drv_flags & IFF_DRV_RUNNING)))
1927 		return (EAGAIN);
1928 
1929 	return (1);
1930 }
1931 
1932 /*
1933  * Calculate max IPv6 MTU through all the interfaces and store it
1934  * to in6_maxmtu.
1935  */
1936 void
in6_setmaxmtu(void)1937 in6_setmaxmtu(void)
1938 {
1939 	unsigned long maxmtu = 0;
1940 	struct ifnet *ifp;
1941 
1942 	IFNET_RLOCK_NOSLEEP();
1943 	TAILQ_FOREACH(ifp, &V_ifnet, if_link) {
1944 		/* this function can be called during ifnet initialization */
1945 		if (!ifp->if_afdata[AF_INET6])
1946 			continue;
1947 		if ((ifp->if_flags & IFF_LOOPBACK) == 0 &&
1948 		    IN6_LINKMTU(ifp) > maxmtu)
1949 			maxmtu = IN6_LINKMTU(ifp);
1950 	}
1951 	IFNET_RUNLOCK_NOSLEEP();
1952 	if (maxmtu)	/* update only when maxmtu is positive */
1953 		V_in6_maxmtu = maxmtu;
1954 }
1955 
1956 /*
1957  * Provide the length of interface identifiers to be used for the link attached
1958  * to the given interface.  The length should be defined in "IPv6 over
1959  * xxx-link" document.  Note that address architecture might also define
1960  * the length for a particular set of address prefixes, regardless of the
1961  * link type.  As clarified in rfc2462bis, those two definitions should be
1962  * consistent, and those really are as of August 2004.
1963  */
1964 int
in6_if2idlen(struct ifnet * ifp)1965 in6_if2idlen(struct ifnet *ifp)
1966 {
1967 	switch (ifp->if_type) {
1968 	case IFT_ETHER:		/* RFC2464 */
1969 	case IFT_PROPVIRTUAL:	/* XXX: no RFC. treat it as ether */
1970 	case IFT_L2VLAN:	/* ditto */
1971 	case IFT_IEEE80211:	/* ditto */
1972 	case IFT_BRIDGE:	/* bridge(4) only does Ethernet-like links */
1973 	case IFT_INFINIBAND:
1974 		return (64);
1975 	case IFT_FDDI:		/* RFC2467 */
1976 		return (64);
1977 	case IFT_ISO88025:	/* RFC2470 (IPv6 over Token Ring) */
1978 		return (64);
1979 	case IFT_PPP:		/* RFC2472 */
1980 		return (64);
1981 	case IFT_ARCNET:	/* RFC2497 */
1982 		return (64);
1983 	case IFT_FRELAY:	/* RFC2590 */
1984 		return (64);
1985 	case IFT_IEEE1394:	/* RFC3146 */
1986 		return (64);
1987 	case IFT_GIF:
1988 		return (64);	/* draft-ietf-v6ops-mech-v2-07 */
1989 	case IFT_LOOP:
1990 		return (64);	/* XXX: is this really correct? */
1991 	default:
1992 		/*
1993 		 * Unknown link type:
1994 		 * It might be controversial to use the today's common constant
1995 		 * of 64 for these cases unconditionally.  For full compliance,
1996 		 * we should return an error in this case.  On the other hand,
1997 		 * if we simply miss the standard for the link type or a new
1998 		 * standard is defined for a new link type, the IFID length
1999 		 * is very likely to be the common constant.  As a compromise,
2000 		 * we always use the constant, but make an explicit notice
2001 		 * indicating the "unknown" case.
2002 		 */
2003 		printf("in6_if2idlen: unknown link type (%d)\n", ifp->if_type);
2004 		return (64);
2005 	}
2006 }
2007 
2008 #include <sys/sysctl.h>
2009 
2010 struct in6_llentry {
2011 	struct llentry		base;
2012 };
2013 
2014 #define	IN6_LLTBL_DEFAULT_HSIZE	32
2015 #define	IN6_LLTBL_HASH(k, h) \
2016 	(((((((k >> 8) ^ k) >> 8) ^ k) >> 8) ^ k) & ((h) - 1))
2017 
2018 /*
2019  * Do actual deallocation of @lle.
2020  */
2021 static void
in6_lltable_destroy_lle_unlocked(struct llentry * lle)2022 in6_lltable_destroy_lle_unlocked(struct llentry *lle)
2023 {
2024 
2025 	LLE_LOCK_DESTROY(lle);
2026 	LLE_REQ_DESTROY(lle);
2027 	free(lle, M_LLTABLE);
2028 }
2029 
2030 /*
2031  * Called by LLE_FREE_LOCKED when number of references
2032  * drops to zero.
2033  */
2034 static void
in6_lltable_destroy_lle(struct llentry * lle)2035 in6_lltable_destroy_lle(struct llentry *lle)
2036 {
2037 
2038 	LLE_WUNLOCK(lle);
2039 	in6_lltable_destroy_lle_unlocked(lle);
2040 }
2041 
2042 static struct llentry *
in6_lltable_new(const struct in6_addr * addr6,u_int flags)2043 in6_lltable_new(const struct in6_addr *addr6, u_int flags)
2044 {
2045 	struct in6_llentry *lle;
2046 
2047 	lle = malloc(sizeof(struct in6_llentry), M_LLTABLE, M_NOWAIT | M_ZERO);
2048 	if (lle == NULL)		/* NB: caller generates msg */
2049 		return NULL;
2050 
2051 	lle->base.r_l3addr.addr6 = *addr6;
2052 	lle->base.lle_refcnt = 1;
2053 	lle->base.lle_free = in6_lltable_destroy_lle;
2054 	LLE_LOCK_INIT(&lle->base);
2055 	LLE_REQ_INIT(&lle->base);
2056 	callout_init(&lle->base.lle_timer, 1);
2057 
2058 	return (&lle->base);
2059 }
2060 
2061 static int
in6_lltable_match_prefix(const struct sockaddr * saddr,const struct sockaddr * smask,u_int flags,struct llentry * lle)2062 in6_lltable_match_prefix(const struct sockaddr *saddr,
2063     const struct sockaddr *smask, u_int flags, struct llentry *lle)
2064 {
2065 	const struct in6_addr *addr, *mask, *lle_addr;
2066 
2067 	addr = &((const struct sockaddr_in6 *)saddr)->sin6_addr;
2068 	mask = &((const struct sockaddr_in6 *)smask)->sin6_addr;
2069 	lle_addr = &lle->r_l3addr.addr6;
2070 
2071 	if (IN6_ARE_MASKED_ADDR_EQUAL(lle_addr, addr, mask) == 0)
2072 		return (0);
2073 
2074 	if (lle->la_flags & LLE_IFADDR) {
2075 
2076 		/*
2077 		 * Delete LLE_IFADDR records IFF address & flag matches.
2078 		 * Note that addr is the interface address within prefix
2079 		 * being matched.
2080 		 */
2081 		if (IN6_ARE_ADDR_EQUAL(addr, lle_addr) &&
2082 		    (flags & LLE_STATIC) != 0)
2083 			return (1);
2084 		return (0);
2085 	}
2086 
2087 	/* flags & LLE_STATIC means deleting both dynamic and static entries */
2088 	if ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))
2089 		return (1);
2090 
2091 	return (0);
2092 }
2093 
2094 static void
in6_lltable_free_entry(struct lltable * llt,struct llentry * lle)2095 in6_lltable_free_entry(struct lltable *llt, struct llentry *lle)
2096 {
2097 	struct ifnet *ifp;
2098 
2099 	LLE_WLOCK_ASSERT(lle);
2100 	KASSERT(llt != NULL, ("lltable is NULL"));
2101 
2102 	/* Unlink entry from table */
2103 	if ((lle->la_flags & LLE_LINKED) != 0) {
2104 
2105 		ifp = llt->llt_ifp;
2106 		IF_AFDATA_WLOCK_ASSERT(ifp);
2107 		lltable_unlink_entry(llt, lle);
2108 	}
2109 
2110 	llentry_free(lle);
2111 }
2112 
2113 static int
in6_lltable_rtcheck(struct ifnet * ifp,u_int flags,const struct sockaddr * l3addr)2114 in6_lltable_rtcheck(struct ifnet *ifp,
2115 		    u_int flags,
2116 		    const struct sockaddr *l3addr)
2117 {
2118 	const struct sockaddr_in6 *sin6;
2119 	struct nhop6_basic nh6;
2120 	struct in6_addr dst;
2121 	uint32_t scopeid;
2122 	int error;
2123 	char ip6buf[INET6_ADDRSTRLEN];
2124 	int fibnum;
2125 
2126 	KASSERT(l3addr->sa_family == AF_INET6,
2127 	    ("sin_family %d", l3addr->sa_family));
2128 
2129 	sin6 = (const struct sockaddr_in6 *)l3addr;
2130 	in6_splitscope(&sin6->sin6_addr, &dst, &scopeid);
2131 	fibnum = V_rt_add_addr_allfibs ? RT_DEFAULT_FIB : ifp->if_fib;
2132 	error = fib6_lookup_nh_basic(fibnum, &dst, scopeid, 0, 0, &nh6);
2133 	if (error != 0 || (nh6.nh_flags & NHF_GATEWAY) || nh6.nh_ifp != ifp) {
2134 		struct ifaddr *ifa;
2135 		/*
2136 		 * Create an ND6 cache for an IPv6 neighbor
2137 		 * that is not covered by our own prefix.
2138 		 */
2139 		ifa = ifaof_ifpforaddr(l3addr, ifp);
2140 		if (ifa != NULL) {
2141 			ifa_free(ifa);
2142 			return 0;
2143 		}
2144 		log(LOG_INFO, "IPv6 address: \"%s\" is not on the network\n",
2145 		    ip6_sprintf(ip6buf, &sin6->sin6_addr));
2146 		return EINVAL;
2147 	}
2148 	return 0;
2149 }
2150 
2151 /*
2152  * Called by the datapath to indicate that the entry was used.
2153  */
2154 static void
in6_lltable_mark_used(struct llentry * lle)2155 in6_lltable_mark_used(struct llentry *lle)
2156 {
2157 
2158 	LLE_REQ_LOCK(lle);
2159 	lle->r_skip_req = 0;
2160 
2161 	/*
2162 	 * Set the hit time so the callback function
2163 	 * can determine the remaining time before
2164 	 * transiting to the DELAY state.
2165 	 */
2166 	lle->lle_hittime = time_uptime;
2167 	LLE_REQ_UNLOCK(lle);
2168 }
2169 
2170 static inline uint32_t
in6_lltable_hash_dst(const struct in6_addr * dst,uint32_t hsize)2171 in6_lltable_hash_dst(const struct in6_addr *dst, uint32_t hsize)
2172 {
2173 
2174 	return (IN6_LLTBL_HASH(dst->s6_addr32[3], hsize));
2175 }
2176 
2177 static uint32_t
in6_lltable_hash(const struct llentry * lle,uint32_t hsize)2178 in6_lltable_hash(const struct llentry *lle, uint32_t hsize)
2179 {
2180 
2181 	return (in6_lltable_hash_dst(&lle->r_l3addr.addr6, hsize));
2182 }
2183 
2184 static void
in6_lltable_fill_sa_entry(const struct llentry * lle,struct sockaddr * sa)2185 in6_lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
2186 {
2187 	struct sockaddr_in6 *sin6;
2188 
2189 	sin6 = (struct sockaddr_in6 *)sa;
2190 	bzero(sin6, sizeof(*sin6));
2191 	sin6->sin6_family = AF_INET6;
2192 	sin6->sin6_len = sizeof(*sin6);
2193 	sin6->sin6_addr = lle->r_l3addr.addr6;
2194 }
2195 
2196 static inline struct llentry *
in6_lltable_find_dst(struct lltable * llt,const struct in6_addr * dst)2197 in6_lltable_find_dst(struct lltable *llt, const struct in6_addr *dst)
2198 {
2199 	struct llentry *lle;
2200 	struct llentries *lleh;
2201 	u_int hashidx;
2202 
2203 	hashidx = in6_lltable_hash_dst(dst, llt->llt_hsize);
2204 	lleh = &llt->lle_head[hashidx];
2205 	LIST_FOREACH(lle, lleh, lle_next) {
2206 		if (lle->la_flags & LLE_DELETED)
2207 			continue;
2208 		if (IN6_ARE_ADDR_EQUAL(&lle->r_l3addr.addr6, dst))
2209 			break;
2210 	}
2211 
2212 	return (lle);
2213 }
2214 
2215 static void
in6_lltable_delete_entry(struct lltable * llt,struct llentry * lle)2216 in6_lltable_delete_entry(struct lltable *llt, struct llentry *lle)
2217 {
2218 
2219 	lle->la_flags |= LLE_DELETED;
2220 	EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
2221 #ifdef DIAGNOSTIC
2222 	log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
2223 #endif
2224 	llentry_free(lle);
2225 }
2226 
2227 static struct llentry *
in6_lltable_alloc(struct lltable * llt,u_int flags,const struct sockaddr * l3addr)2228 in6_lltable_alloc(struct lltable *llt, u_int flags,
2229 	const struct sockaddr *l3addr)
2230 {
2231 	const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr;
2232 	struct ifnet *ifp = llt->llt_ifp;
2233 	struct llentry *lle;
2234 	char linkhdr[LLE_MAX_LINKHDR];
2235 	size_t linkhdrsize;
2236 	int lladdr_off;
2237 
2238 	KASSERT(l3addr->sa_family == AF_INET6,
2239 	    ("sin_family %d", l3addr->sa_family));
2240 
2241 	/*
2242 	 * A route that covers the given address must have
2243 	 * been installed 1st because we are doing a resolution,
2244 	 * verify this.
2245 	 */
2246 	if (!(flags & LLE_IFADDR) &&
2247 	    in6_lltable_rtcheck(ifp, flags, l3addr) != 0)
2248 		return (NULL);
2249 
2250 	lle = in6_lltable_new(&sin6->sin6_addr, flags);
2251 	if (lle == NULL) {
2252 		log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
2253 		return (NULL);
2254 	}
2255 	lle->la_flags = flags;
2256 	if ((flags & LLE_IFADDR) == LLE_IFADDR) {
2257 		linkhdrsize = LLE_MAX_LINKHDR;
2258 		if (lltable_calc_llheader(ifp, AF_INET6, IF_LLADDR(ifp),
2259 		    linkhdr, &linkhdrsize, &lladdr_off) != 0) {
2260 			in6_lltable_destroy_lle_unlocked(lle);
2261 			return (NULL);
2262 		}
2263 		lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
2264 		    lladdr_off);
2265 		lle->la_flags |= LLE_STATIC;
2266 	}
2267 
2268 	if ((lle->la_flags & LLE_STATIC) != 0)
2269 		lle->ln_state = ND6_LLINFO_REACHABLE;
2270 
2271 	return (lle);
2272 }
2273 
2274 static struct llentry *
in6_lltable_lookup(struct lltable * llt,u_int flags,const struct sockaddr * l3addr)2275 in6_lltable_lookup(struct lltable *llt, u_int flags,
2276 	const struct sockaddr *l3addr)
2277 {
2278 	const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr;
2279 	struct llentry *lle;
2280 
2281 	IF_AFDATA_LOCK_ASSERT(llt->llt_ifp);
2282 	KASSERT(l3addr->sa_family == AF_INET6,
2283 	    ("sin_family %d", l3addr->sa_family));
2284 
2285 	lle = in6_lltable_find_dst(llt, &sin6->sin6_addr);
2286 
2287 	if (lle == NULL)
2288 		return (NULL);
2289 
2290 	KASSERT((flags & (LLE_UNLOCKED|LLE_EXCLUSIVE)) !=
2291 	    (LLE_UNLOCKED|LLE_EXCLUSIVE),("wrong lle request flags: 0x%X",
2292 	    flags));
2293 
2294 	if (flags & LLE_UNLOCKED)
2295 		return (lle);
2296 
2297 	if (flags & LLE_EXCLUSIVE)
2298 		LLE_WLOCK(lle);
2299 	else
2300 		LLE_RLOCK(lle);
2301 	return (lle);
2302 }
2303 
2304 static int
in6_lltable_dump_entry(struct lltable * llt,struct llentry * lle,struct sysctl_req * wr)2305 in6_lltable_dump_entry(struct lltable *llt, struct llentry *lle,
2306     struct sysctl_req *wr)
2307 {
2308 	struct ifnet *ifp = llt->llt_ifp;
2309 	/* XXX stack use */
2310 	struct {
2311 		struct rt_msghdr	rtm;
2312 		struct sockaddr_in6	sin6;
2313 		/*
2314 		 * ndp.c assumes that sdl is word aligned
2315 		 */
2316 #ifdef __LP64__
2317 		uint32_t		pad;
2318 #endif
2319 		struct sockaddr_dl	sdl;
2320 	} ndpc;
2321 	struct sockaddr_dl *sdl;
2322 	int error;
2323 
2324 	bzero(&ndpc, sizeof(ndpc));
2325 			/* skip deleted entries */
2326 			if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
2327 				return (0);
2328 			/* Skip if jailed and not a valid IP of the prison. */
2329 			lltable_fill_sa_entry(lle,
2330 			    (struct sockaddr *)&ndpc.sin6);
2331 			if (prison_if(wr->td->td_ucred,
2332 			    (struct sockaddr *)&ndpc.sin6) != 0)
2333 				return (0);
2334 			/*
2335 			 * produce a msg made of:
2336 			 *  struct rt_msghdr;
2337 			 *  struct sockaddr_in6 (IPv6)
2338 			 *  struct sockaddr_dl;
2339 			 */
2340 			ndpc.rtm.rtm_msglen = sizeof(ndpc);
2341 			ndpc.rtm.rtm_version = RTM_VERSION;
2342 			ndpc.rtm.rtm_type = RTM_GET;
2343 			ndpc.rtm.rtm_flags = RTF_UP;
2344 			ndpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
2345 			if (V_deembed_scopeid)
2346 				sa6_recoverscope(&ndpc.sin6);
2347 
2348 			/* publish */
2349 			if (lle->la_flags & LLE_PUB)
2350 				ndpc.rtm.rtm_flags |= RTF_ANNOUNCE;
2351 
2352 			sdl = &ndpc.sdl;
2353 			sdl->sdl_family = AF_LINK;
2354 			sdl->sdl_len = sizeof(*sdl);
2355 			sdl->sdl_index = ifp->if_index;
2356 			sdl->sdl_type = ifp->if_type;
2357 			if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
2358 				sdl->sdl_alen = ifp->if_addrlen;
2359 				bcopy(lle->ll_addr, LLADDR(sdl),
2360 				    ifp->if_addrlen);
2361 			} else {
2362 				sdl->sdl_alen = 0;
2363 				bzero(LLADDR(sdl), ifp->if_addrlen);
2364 			}
2365 			if (lle->la_expire != 0)
2366 				ndpc.rtm.rtm_rmx.rmx_expire = lle->la_expire +
2367 				    lle->lle_remtime / hz +
2368 				    time_second - time_uptime;
2369 			ndpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
2370 			if (lle->la_flags & LLE_STATIC)
2371 				ndpc.rtm.rtm_flags |= RTF_STATIC;
2372 			if (lle->la_flags & LLE_IFADDR)
2373 				ndpc.rtm.rtm_flags |= RTF_PINNED;
2374 			if (lle->ln_router != 0)
2375 				ndpc.rtm.rtm_flags |= RTF_GATEWAY;
2376 			ndpc.rtm.rtm_rmx.rmx_pksent = lle->la_asked;
2377 			/* Store state in rmx_weight value */
2378 			ndpc.rtm.rtm_rmx.rmx_state = lle->ln_state;
2379 			ndpc.rtm.rtm_index = ifp->if_index;
2380 			error = SYSCTL_OUT(wr, &ndpc, sizeof(ndpc));
2381 
2382 	return (error);
2383 }
2384 
2385 static struct lltable *
in6_lltattach(struct ifnet * ifp)2386 in6_lltattach(struct ifnet *ifp)
2387 {
2388 	struct lltable *llt;
2389 
2390 	llt = lltable_allocate_htbl(IN6_LLTBL_DEFAULT_HSIZE);
2391 	llt->llt_af = AF_INET6;
2392 	llt->llt_ifp = ifp;
2393 
2394 	llt->llt_lookup = in6_lltable_lookup;
2395 	llt->llt_alloc_entry = in6_lltable_alloc;
2396 	llt->llt_delete_entry = in6_lltable_delete_entry;
2397 	llt->llt_dump_entry = in6_lltable_dump_entry;
2398 	llt->llt_hash = in6_lltable_hash;
2399 	llt->llt_fill_sa_entry = in6_lltable_fill_sa_entry;
2400 	llt->llt_free_entry = in6_lltable_free_entry;
2401 	llt->llt_match_prefix = in6_lltable_match_prefix;
2402 	llt->llt_mark_used = in6_lltable_mark_used;
2403  	lltable_link(llt);
2404 
2405 	return (llt);
2406 }
2407 
2408 void *
in6_domifattach(struct ifnet * ifp)2409 in6_domifattach(struct ifnet *ifp)
2410 {
2411 	struct in6_ifextra *ext;
2412 
2413 	/* There are not IPv6-capable interfaces. */
2414 	switch (ifp->if_type) {
2415 	case IFT_PFLOG:
2416 	case IFT_PFSYNC:
2417 	case IFT_USB:
2418 		return (NULL);
2419 	}
2420 	ext = (struct in6_ifextra *)malloc(sizeof(*ext), M_IFADDR, M_WAITOK);
2421 	bzero(ext, sizeof(*ext));
2422 
2423 	ext->in6_ifstat = malloc(sizeof(counter_u64_t) *
2424 	    sizeof(struct in6_ifstat) / sizeof(uint64_t), M_IFADDR, M_WAITOK);
2425 	COUNTER_ARRAY_ALLOC(ext->in6_ifstat,
2426 	    sizeof(struct in6_ifstat) / sizeof(uint64_t), M_WAITOK);
2427 
2428 	ext->icmp6_ifstat = malloc(sizeof(counter_u64_t) *
2429 	    sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_IFADDR,
2430 	    M_WAITOK);
2431 	COUNTER_ARRAY_ALLOC(ext->icmp6_ifstat,
2432 	    sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_WAITOK);
2433 
2434 	ext->nd_ifinfo = nd6_ifattach(ifp);
2435 	ext->scope6_id = scope6_ifattach(ifp);
2436 	ext->lltable = in6_lltattach(ifp);
2437 
2438 	ext->mld_ifinfo = mld_domifattach(ifp);
2439 
2440 	return ext;
2441 }
2442 
2443 int
in6_domifmtu(struct ifnet * ifp)2444 in6_domifmtu(struct ifnet *ifp)
2445 {
2446 	if (ifp->if_afdata[AF_INET6] == NULL)
2447 		return ifp->if_mtu;
2448 
2449 	return (IN6_LINKMTU(ifp));
2450 }
2451 
2452 void
in6_domifdetach(struct ifnet * ifp,void * aux)2453 in6_domifdetach(struct ifnet *ifp, void *aux)
2454 {
2455 	struct in6_ifextra *ext = (struct in6_ifextra *)aux;
2456 
2457 	mld_domifdetach(ifp);
2458 	scope6_ifdetach(ext->scope6_id);
2459 	nd6_ifdetach(ifp, ext->nd_ifinfo);
2460 	lltable_free(ext->lltable);
2461 	COUNTER_ARRAY_FREE(ext->in6_ifstat,
2462 	    sizeof(struct in6_ifstat) / sizeof(uint64_t));
2463 	free(ext->in6_ifstat, M_IFADDR);
2464 	COUNTER_ARRAY_FREE(ext->icmp6_ifstat,
2465 	    sizeof(struct icmp6_ifstat) / sizeof(uint64_t));
2466 	free(ext->icmp6_ifstat, M_IFADDR);
2467 	free(ext, M_IFADDR);
2468 }
2469 
2470 /*
2471  * Convert sockaddr_in6 to sockaddr_in.  Original sockaddr_in6 must be
2472  * v4 mapped addr or v4 compat addr
2473  */
2474 void
in6_sin6_2_sin(struct sockaddr_in * sin,struct sockaddr_in6 * sin6)2475 in6_sin6_2_sin(struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
2476 {
2477 
2478 	bzero(sin, sizeof(*sin));
2479 	sin->sin_len = sizeof(struct sockaddr_in);
2480 	sin->sin_family = AF_INET;
2481 	sin->sin_port = sin6->sin6_port;
2482 	sin->sin_addr.s_addr = sin6->sin6_addr.s6_addr32[3];
2483 }
2484 
2485 /* Convert sockaddr_in to sockaddr_in6 in v4 mapped addr format. */
2486 void
in6_sin_2_v4mapsin6(struct sockaddr_in * sin,struct sockaddr_in6 * sin6)2487 in6_sin_2_v4mapsin6(struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
2488 {
2489 	bzero(sin6, sizeof(*sin6));
2490 	sin6->sin6_len = sizeof(struct sockaddr_in6);
2491 	sin6->sin6_family = AF_INET6;
2492 	sin6->sin6_port = sin->sin_port;
2493 	sin6->sin6_addr.s6_addr32[0] = 0;
2494 	sin6->sin6_addr.s6_addr32[1] = 0;
2495 	sin6->sin6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
2496 	sin6->sin6_addr.s6_addr32[3] = sin->sin_addr.s_addr;
2497 }
2498 
2499 /* Convert sockaddr_in6 into sockaddr_in. */
2500 void
in6_sin6_2_sin_in_sock(struct sockaddr * nam)2501 in6_sin6_2_sin_in_sock(struct sockaddr *nam)
2502 {
2503 	struct sockaddr_in *sin_p;
2504 	struct sockaddr_in6 sin6;
2505 
2506 	/*
2507 	 * Save original sockaddr_in6 addr and convert it
2508 	 * to sockaddr_in.
2509 	 */
2510 	sin6 = *(struct sockaddr_in6 *)nam;
2511 	sin_p = (struct sockaddr_in *)nam;
2512 	in6_sin6_2_sin(sin_p, &sin6);
2513 }
2514 
2515 /* Convert sockaddr_in into sockaddr_in6 in v4 mapped addr format. */
2516 void
in6_sin_2_v4mapsin6_in_sock(struct sockaddr ** nam)2517 in6_sin_2_v4mapsin6_in_sock(struct sockaddr **nam)
2518 {
2519 	struct sockaddr_in *sin_p;
2520 	struct sockaddr_in6 *sin6_p;
2521 
2522 	sin6_p = malloc(sizeof *sin6_p, M_SONAME, M_WAITOK);
2523 	sin_p = (struct sockaddr_in *)*nam;
2524 	in6_sin_2_v4mapsin6(sin_p, sin6_p);
2525 	free(*nam, M_SONAME);
2526 	*nam = (struct sockaddr *)sin6_p;
2527 }
2528