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
63 #include "opt_inet.h"
64 #include "opt_inet6.h"
65
66 #include <sys/param.h>
67 #include <sys/eventhandler.h>
68 #include <sys/errno.h>
69 #include <sys/jail.h>
70 #include <sys/malloc.h>
71 #include <sys/socket.h>
72 #include <sys/socketvar.h>
73 #include <sys/sockio.h>
74 #include <sys/systm.h>
75 #include <sys/priv.h>
76 #include <sys/proc.h>
77 #include <sys/time.h>
78 #include <sys/kernel.h>
79 #include <sys/lock.h>
80 #include <sys/rmlock.h>
81 #include <sys/sysctl.h>
82 #include <sys/syslog.h>
83
84 #include <net/if.h>
85 #include <net/if_var.h>
86 #include <net/if_private.h>
87 #include <net/if_types.h>
88 #include <net/if_bridgevar.h>
89 #include <net/route.h>
90 #include <net/route/route_ctl.h>
91 #include <net/route/nhop.h>
92 #include <net/if_dl.h>
93 #include <net/vnet.h>
94
95 #include <netinet/in.h>
96 #include <netinet/in_var.h>
97 #include <net/if_llatbl.h>
98 #include <netinet/if_ether.h>
99 #include <netinet/in_systm.h>
100 #include <netinet/ip.h>
101 #include <netinet/in_pcb.h>
102 #include <netinet/ip_carp.h>
103 #include <netinet/icmp6.h>
104
105 #include <netinet/ip6.h>
106 #include <netinet6/ip6_var.h>
107 #include <netinet6/nd6.h>
108 #include <netinet6/mld6_var.h>
109 #include <netinet6/ip6_mroute.h>
110 #include <netinet6/in6_ifattach.h>
111 #include <netinet6/scope6_var.h>
112 #include <netinet6/in6_fib.h>
113 #include <netinet6/in6_pcb.h>
114
115 #ifdef MAC
116 #include <security/mac/mac_framework.h>
117 #endif
118
119 /*
120 * struct in6_ifreq and struct ifreq must be type punnable for common members
121 * of ifr_ifru to allow accessors to be shared.
122 */
123 _Static_assert(offsetof(struct in6_ifreq, ifr_ifru) ==
124 offsetof(struct ifreq, ifr_ifru),
125 "struct in6_ifreq and struct ifreq are not type punnable");
126
127 VNET_DEFINE_STATIC(int, icmp6_nodeinfo_oldmcprefix) = 1;
128 #define V_icmp6_nodeinfo_oldmcprefix VNET(icmp6_nodeinfo_oldmcprefix)
129 SYSCTL_INT(_net_inet6_icmp6, ICMPV6CTL_NODEINFO_OLDMCPREFIX,
130 nodeinfo_oldmcprefix, CTLFLAG_VNET | CTLFLAG_RW,
131 &VNET_NAME(icmp6_nodeinfo_oldmcprefix), 0,
132 "Join old IPv6 NI group address in draft-ietf-ipngwg-icmp-name-lookup "
133 "for compatibility with KAME implementation");
134
135 VNET_DEFINE_STATIC(int, nd6_useloopback) = 1;
136 #define V_nd6_useloopback VNET(nd6_useloopback)
137 SYSCTL_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_USELOOPBACK, nd6_useloopback,
138 CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(nd6_useloopback), 0,
139 "Create a loopback route when configuring an IPv6 address");
140
141 /*
142 * Definitions of some costant IP6 addresses.
143 */
144 const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT;
145 const struct in6_addr in6addr_loopback = IN6ADDR_LOOPBACK_INIT;
146 const struct in6_addr in6addr_nodelocal_allnodes =
147 IN6ADDR_NODELOCAL_ALLNODES_INIT;
148 const struct in6_addr in6addr_linklocal_allnodes =
149 IN6ADDR_LINKLOCAL_ALLNODES_INIT;
150 const struct in6_addr in6addr_linklocal_allrouters =
151 IN6ADDR_LINKLOCAL_ALLROUTERS_INIT;
152 const struct in6_addr in6addr_linklocal_allv2routers =
153 IN6ADDR_LINKLOCAL_ALLV2ROUTERS_INIT;
154
155 const struct in6_addr in6mask0 = IN6MASK0;
156 const struct in6_addr in6mask32 = IN6MASK32;
157 const struct in6_addr in6mask64 = IN6MASK64;
158 const struct in6_addr in6mask96 = IN6MASK96;
159 const struct in6_addr in6mask128 = IN6MASK128;
160
161 const struct sockaddr_in6 sa6_any =
162 { sizeof(sa6_any), AF_INET6, 0, 0, IN6ADDR_ANY_INIT, 0 };
163
164 static int in6_notify_ifa(struct ifnet *, struct in6_ifaddr *,
165 struct in6_aliasreq *, int);
166 static void in6_unlink_ifa(struct in6_ifaddr *, struct ifnet *);
167
168 static int in6_validate_ifra(struct ifnet *, struct in6_aliasreq *,
169 struct in6_ifaddr *, int);
170 static struct in6_ifaddr *in6_alloc_ifa(struct ifnet *,
171 struct in6_aliasreq *, int flags);
172 static int in6_update_ifa_internal(struct ifnet *, struct in6_aliasreq *,
173 struct in6_ifaddr *, int, int);
174 static int in6_broadcast_ifa(struct ifnet *, struct in6_aliasreq *,
175 struct in6_ifaddr *, int);
176
177 static void in6_join_proxy_ndp_mc(struct ifnet *, const struct in6_addr *);
178 static void in6_leave_proxy_ndp_mc(struct ifnet *, const struct in6_addr *);
179
180 #define ifa2ia6(ifa) ((struct in6_ifaddr *)(ifa))
181 #define ia62ifa(ia6) (&((ia6)->ia_ifa))
182
183 static struct sx in6_control_sx;
184 SX_SYSINIT(in6_control_sx, &in6_control_sx, "in6_control");
185
186 void
in6_newaddrmsg(struct in6_ifaddr * ia,int cmd)187 in6_newaddrmsg(struct in6_ifaddr *ia, int cmd)
188 {
189 struct rt_addrinfo info;
190 struct ifaddr *ifa;
191 struct sockaddr_dl gateway;
192 int fibnum;
193
194 ifa = &ia->ia_ifa;
195
196 /*
197 * Prepare info data for the host route.
198 * This code mimics one from ifa_maintain_loopback_route().
199 */
200 bzero(&info, sizeof(struct rt_addrinfo));
201 info.rti_flags = ifa->ifa_flags | RTF_HOST | RTF_STATIC | RTF_PINNED;
202 info.rti_info[RTAX_DST] = ifa->ifa_addr;
203 info.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&gateway;
204 link_init_sdl(ifa->ifa_ifp, (struct sockaddr *)&gateway, ifa->ifa_ifp->if_type);
205 if (cmd != RTM_DELETE)
206 info.rti_ifp = V_loif;
207
208 fibnum = ia62ifa(ia)->ifa_ifp->if_fib;
209
210 if (cmd == RTM_ADD) {
211 rt_addrmsg(cmd, &ia->ia_ifa, fibnum);
212 rt_routemsg_info(cmd, &info, fibnum);
213 } else if (cmd == RTM_DELETE) {
214 rt_routemsg_info(cmd, &info, fibnum);
215 rt_addrmsg(cmd, &ia->ia_ifa, fibnum);
216 }
217 }
218
219 int
in6_mask2len(struct in6_addr * mask,u_char * lim0)220 in6_mask2len(struct in6_addr *mask, u_char *lim0)
221 {
222 int x = 0, y;
223 u_char *lim = lim0, *p;
224
225 /* ignore the scope_id part */
226 if (lim0 == NULL || lim0 - (u_char *)mask > sizeof(*mask))
227 lim = (u_char *)mask + sizeof(*mask);
228 for (p = (u_char *)mask; p < lim; x++, p++) {
229 if (*p != 0xff)
230 break;
231 }
232 y = 0;
233 if (p < lim) {
234 for (y = 0; y < 8; y++) {
235 if ((*p & (0x80 >> y)) == 0)
236 break;
237 }
238 }
239
240 /*
241 * when the limit pointer is given, do a stricter check on the
242 * remaining bits.
243 */
244 if (p < lim) {
245 if (y != 0 && (*p & (0x00ff >> y)) != 0)
246 return (-1);
247 for (p = p + 1; p < lim; p++)
248 if (*p != 0)
249 return (-1);
250 }
251
252 return x * 8 + y;
253 }
254
255 #ifdef COMPAT_FREEBSD32
256 struct in6_ndifreq32 {
257 char ifname[IFNAMSIZ];
258 uint32_t ifindex;
259 };
260 #define SIOCGDEFIFACE32_IN6 _IOWR('i', 86, struct in6_ndifreq32)
261 #endif
262
263 int
in6_control_ioctl(u_long cmd,void * data,struct ifnet * ifp,struct ucred * cred)264 in6_control_ioctl(u_long cmd, void *data,
265 struct ifnet *ifp, struct ucred *cred)
266 {
267 struct in6_ifreq *ifr = (struct in6_ifreq *)data;
268 struct in6_ifaddr *ia = NULL;
269 struct in6_aliasreq tmpifra, *ifra = (struct in6_aliasreq *)data;
270 struct sockaddr_in6 *sa6;
271 int error;
272 bool control_locked = false;
273
274 /*
275 * Compat to make pre-10.x ifconfig(8) operable.
276 */
277 if (cmd == OSIOCAIFADDR_IN6) {
278 gone_in(17, "%s (pid %d) uses OSIOCAIFADDR_IN6",
279 curthread->td_proc->p_comm, curthread->td_proc->p_pid);
280 cmd = SIOCAIFADDR_IN6;
281 ifra = &tmpifra;
282 memset(ifra, 0, sizeof(*ifra));
283 memcpy(ifra, data, sizeof(struct oin6_aliasreq));
284 }
285
286 switch (cmd) {
287 case SIOCAADDRCTL_POLICY:
288 case SIOCDADDRCTL_POLICY:
289 if (cred != NULL) {
290 error = priv_check_cred(cred, PRIV_NETINET_ADDRCTRL6);
291 if (error)
292 return (error);
293 }
294 return (in6_src_ioctl(cmd, data));
295 }
296
297 if (ifp == NULL)
298 return (EOPNOTSUPP);
299
300 switch (cmd) {
301 case SIOCSNDFLUSH_IN6:
302 case SIOCSPFXFLUSH_IN6:
303 case SIOCSRTRFLUSH_IN6:
304 case SIOCSDEFIFACE_IN6:
305 case SIOCSIFINFO_FLAGS:
306 case SIOCSIFINFO_IN6:
307 if (cred != NULL) {
308 error = priv_check_cred(cred, PRIV_NETINET_ND6);
309 if (error)
310 return (error);
311 }
312 /* FALLTHROUGH */
313 case SIOCGIFINFO_IN6:
314 case SIOCGNBRINFO_IN6:
315 case SIOCGDEFIFACE_IN6:
316 return (nd6_ioctl(cmd, data, ifp));
317
318 #ifdef COMPAT_FREEBSD32
319 case SIOCGDEFIFACE32_IN6:
320 {
321 struct in6_ndifreq ndif;
322 struct in6_ndifreq32 *ndif32;
323
324 error = nd6_ioctl(SIOCGDEFIFACE_IN6, (caddr_t)&ndif,
325 ifp);
326 if (error)
327 return (error);
328 ndif32 = (struct in6_ndifreq32 *)data;
329 ndif32->ifindex = ndif.ifindex;
330 return (0);
331 }
332 #endif
333 }
334
335 switch (cmd) {
336 case SIOCSIFPREFIX_IN6:
337 case SIOCDIFPREFIX_IN6:
338 case SIOCAIFPREFIX_IN6:
339 case SIOCCIFPREFIX_IN6:
340 case SIOCSGIFPREFIX_IN6:
341 case SIOCGIFPREFIX_IN6:
342 log(LOG_NOTICE,
343 "prefix ioctls are now invalidated. "
344 "please use ifconfig.\n");
345 return (EOPNOTSUPP);
346 }
347
348 switch (cmd) {
349 case SIOCSSCOPE6:
350 if (cred != NULL) {
351 error = priv_check_cred(cred, PRIV_NETINET_SCOPE6);
352 if (error)
353 return (error);
354 }
355 /* FALLTHROUGH */
356 case SIOCGSCOPE6:
357 case SIOCGSCOPE6DEF:
358 return (scope6_ioctl(cmd, data, ifp));
359 }
360
361 /*
362 * Find address for this interface, if it exists.
363 *
364 * In netinet code, we have checked ifra_addr in SIOCSIF*ADDR operation
365 * only, and used the first interface address as the target of other
366 * operations (without checking ifra_addr). This was because netinet
367 * code/API assumed at most 1 interface address per interface.
368 * Since IPv6 allows a node to assign multiple addresses
369 * on a single interface, we almost always look and check the
370 * presence of ifra_addr, and reject invalid ones here.
371 * It also decreases duplicated code among SIOC*_IN6 operations.
372 */
373 switch (cmd) {
374 case SIOCAIFADDR_IN6:
375 case SIOCSIFPHYADDR_IN6:
376 sa6 = &ifra->ifra_addr;
377 break;
378 case SIOCSIFADDR_IN6:
379 case SIOCGIFADDR_IN6:
380 case SIOCSIFDSTADDR_IN6:
381 case SIOCSIFNETMASK_IN6:
382 case SIOCGIFDSTADDR_IN6:
383 case SIOCGIFNETMASK_IN6:
384 case SIOCDIFADDR_IN6:
385 case SIOCGIFPSRCADDR_IN6:
386 case SIOCGIFPDSTADDR_IN6:
387 case SIOCGIFAFLAG_IN6:
388 case SIOCSNDFLUSH_IN6:
389 case SIOCSPFXFLUSH_IN6:
390 case SIOCSRTRFLUSH_IN6:
391 case SIOCGIFALIFETIME_IN6:
392 case SIOCGIFSTAT_IN6:
393 case SIOCGIFSTAT_ICMP6:
394 sa6 = &ifr->ifr_addr;
395 break;
396 case SIOCSIFADDR:
397 case SIOCSIFBRDADDR:
398 case SIOCSIFDSTADDR:
399 case SIOCSIFNETMASK:
400 /*
401 * Although we should pass any non-INET6 ioctl requests
402 * down to driver, we filter some legacy INET requests.
403 * Drivers trust SIOCSIFADDR et al to come from an already
404 * privileged layer, and do not perform any credentials
405 * checks or input validation.
406 */
407 return (EINVAL);
408 default:
409 sa6 = NULL;
410 break;
411 }
412 if (sa6 && sa6->sin6_family == AF_INET6) {
413 if (sa6->sin6_scope_id != 0)
414 error = sa6_embedscope(sa6, 0);
415 else
416 error = in6_setscope(&sa6->sin6_addr, ifp, NULL);
417 if (error != 0)
418 return (error);
419 if (cred != NULL && (error = prison_check_ip6(cred,
420 &sa6->sin6_addr)) != 0)
421 return (error);
422 sx_xlock(&in6_control_sx);
423 control_locked = true;
424 ia = in6ifa_ifpwithaddr(ifp, &sa6->sin6_addr);
425 } else
426 ia = NULL;
427
428 switch (cmd) {
429 case SIOCSIFADDR_IN6:
430 case SIOCSIFDSTADDR_IN6:
431 case SIOCSIFNETMASK_IN6:
432 /*
433 * Since IPv6 allows a node to assign multiple addresses
434 * on a single interface, SIOCSIFxxx ioctls are deprecated.
435 */
436 /* we decided to obsolete this command (20000704) */
437 error = EINVAL;
438 goto out;
439
440 case SIOCDIFADDR_IN6:
441 /*
442 * for IPv4, we look for existing in_ifaddr here to allow
443 * "ifconfig if0 delete" to remove the first IPv4 address on
444 * the interface. For IPv6, as the spec allows multiple
445 * interface address from the day one, we consider "remove the
446 * first one" semantics to be not preferable.
447 */
448 if (ia == NULL) {
449 error = EADDRNOTAVAIL;
450 goto out;
451 }
452 /* FALLTHROUGH */
453 case SIOCAIFADDR_IN6:
454 /*
455 * We always require users to specify a valid IPv6 address for
456 * the corresponding operation.
457 */
458 if (ifra->ifra_addr.sin6_family != AF_INET6 ||
459 ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6)) {
460 error = EAFNOSUPPORT;
461 goto out;
462 }
463
464 if (cred != NULL) {
465 error = priv_check_cred(cred, (cmd == SIOCDIFADDR_IN6) ?
466 PRIV_NET_DELIFADDR : PRIV_NET_ADDIFADDR);
467 if (error)
468 goto out;
469 }
470 /* FALLTHROUGH */
471 case SIOCGIFSTAT_IN6:
472 case SIOCGIFSTAT_ICMP6:
473 if (ifp->if_inet6 == NULL) {
474 error = EPFNOSUPPORT;
475 goto out;
476 }
477 break;
478
479 case SIOCGIFADDR_IN6:
480 /* This interface is basically deprecated. use SIOCGIFCONF. */
481 /* FALLTHROUGH */
482 case SIOCGIFAFLAG_IN6:
483 case SIOCGIFNETMASK_IN6:
484 case SIOCGIFDSTADDR_IN6:
485 case SIOCGIFALIFETIME_IN6:
486 /* must think again about its semantics */
487 if (ia == NULL) {
488 error = EADDRNOTAVAIL;
489 goto out;
490 }
491 break;
492 }
493
494 switch (cmd) {
495 case SIOCGIFADDR_IN6:
496 ifr->ifr_addr = ia->ia_addr;
497 if ((error = sa6_recoverscope(&ifr->ifr_addr)) != 0)
498 goto out;
499 break;
500
501 case SIOCGIFDSTADDR_IN6:
502 if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
503 error = EINVAL;
504 goto out;
505 }
506 ifr->ifr_dstaddr = ia->ia_dstaddr;
507 if ((error = sa6_recoverscope(&ifr->ifr_dstaddr)) != 0)
508 goto out;
509 break;
510
511 case SIOCGIFNETMASK_IN6:
512 ifr->ifr_addr = ia->ia_prefixmask;
513 break;
514
515 case SIOCGIFAFLAG_IN6:
516 ifr->ifr_ifru.ifru_flags6 = ia->ia6_flags;
517 break;
518
519 case SIOCGIFSTAT_IN6:
520 COUNTER_ARRAY_COPY(ifp->if_inet6->in6_ifstat,
521 &ifr->ifr_ifru.ifru_stat,
522 sizeof(struct in6_ifstat) / sizeof(uint64_t));
523 break;
524
525 case SIOCGIFSTAT_ICMP6:
526 COUNTER_ARRAY_COPY(ifp->if_inet6->icmp6_ifstat,
527 &ifr->ifr_ifru.ifru_icmp6stat,
528 sizeof(struct icmp6_ifstat) / sizeof(uint64_t));
529 break;
530
531 case SIOCGIFALIFETIME_IN6:
532 ifr->ifr_ifru.ifru_lifetime = ia->ia6_lifetime;
533 if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
534 time_t maxexpire;
535 struct in6_addrlifetime *retlt =
536 &ifr->ifr_ifru.ifru_lifetime;
537
538 /*
539 * XXX: adjust expiration time assuming time_t is
540 * signed.
541 */
542 maxexpire = (-1) &
543 ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
544 if (ia->ia6_lifetime.ia6t_vltime <
545 maxexpire - ia->ia6_updatetime) {
546 retlt->ia6t_expire = ia->ia6_updatetime +
547 ia->ia6_lifetime.ia6t_vltime;
548 } else
549 retlt->ia6t_expire = maxexpire;
550 }
551 if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
552 time_t maxexpire;
553 struct in6_addrlifetime *retlt =
554 &ifr->ifr_ifru.ifru_lifetime;
555
556 /*
557 * XXX: adjust expiration time assuming time_t is
558 * signed.
559 */
560 maxexpire = (-1) &
561 ~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
562 if (ia->ia6_lifetime.ia6t_pltime <
563 maxexpire - ia->ia6_updatetime) {
564 retlt->ia6t_preferred = ia->ia6_updatetime +
565 ia->ia6_lifetime.ia6t_pltime;
566 } else
567 retlt->ia6t_preferred = maxexpire;
568 }
569 break;
570
571 case SIOCAIFADDR_IN6:
572 #ifdef MAC
573 /* Check if a MAC policy disallows setting the IPv6 address. */
574 error = mac_inet6_check_add_addr(cred, &sa6->sin6_addr, ifp);
575 if (error != 0)
576 goto out;
577 #endif
578 error = in6_addifaddr(ifp, ifra, ia);
579 ia = NULL;
580 goto out;
581
582 case SIOCDIFADDR_IN6:
583 in6_purgeifaddr(ia);
584 EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa,
585 IFADDR_EVENT_DEL);
586 break;
587
588 default:
589 if (ifp->if_ioctl == NULL) {
590 error = EOPNOTSUPP;
591 goto out;
592 }
593 error = (*ifp->if_ioctl)(ifp, cmd, data);
594 goto out;
595 }
596
597 error = 0;
598 out:
599 if (control_locked)
600 sx_xunlock(&in6_control_sx);
601
602 if (ia != NULL)
603 ifa_free(&ia->ia_ifa);
604 return (error);
605 }
606
607 int
in6_control(struct socket * so,u_long cmd,void * data,struct ifnet * ifp,struct thread * td)608 in6_control(struct socket *so, u_long cmd, void *data,
609 struct ifnet *ifp, struct thread *td)
610 {
611 switch (cmd) {
612 case SIOCGETSGCNT_IN6:
613 case SIOCGETMIFCNT_IN6:
614 return (mrt6_ioctl ?
615 mrt6_ioctl(cmd, data, so->so_fibnum) : EOPNOTSUPP);
616 }
617 return (in6_control_ioctl(cmd, data, ifp, td ? td->td_ucred : NULL));
618 }
619
620 static struct in6_multi_mship *
in6_joingroup_legacy(struct ifnet * ifp,const struct in6_addr * mcaddr,int * errorp,int delay)621 in6_joingroup_legacy(struct ifnet *ifp, const struct in6_addr *mcaddr,
622 int *errorp, int delay)
623 {
624 struct in6_multi_mship *imm;
625 int error;
626
627 imm = malloc(sizeof(*imm), M_IP6MADDR, M_NOWAIT);
628 if (imm == NULL) {
629 *errorp = ENOBUFS;
630 return (NULL);
631 }
632
633 delay = (delay * MLD_FASTHZ) / hz;
634
635 error = in6_joingroup(ifp, mcaddr, NULL, &imm->i6mm_maddr, delay);
636 if (error) {
637 *errorp = error;
638 free(imm, M_IP6MADDR);
639 return (NULL);
640 }
641
642 return (imm);
643 }
644
645 static int
in6_solicited_node_maddr(struct in6_addr * maddr,struct ifnet * ifp,const struct in6_addr * base)646 in6_solicited_node_maddr(struct in6_addr *maddr,
647 struct ifnet *ifp, const struct in6_addr *base)
648 {
649 int error;
650
651 bzero(maddr, sizeof(struct in6_addr));
652 maddr->s6_addr32[0] = IPV6_ADDR_INT32_MLL;
653 maddr->s6_addr32[2] = htonl(1);
654 maddr->s6_addr32[3] = base->s6_addr32[3];
655 maddr->s6_addr8[12] = 0xff;
656 if ((error = in6_setscope(maddr, ifp, NULL)) != 0) {
657 /* XXX: should not happen */
658 log(LOG_ERR, "%s: in6_setscope failed\n", __func__);
659 }
660
661 return error;
662 }
663
664 /*
665 * Join necessary multicast groups. Factored out from in6_update_ifa().
666 * This entire work should only be done once, for the default FIB.
667 */
668 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)669 in6_update_ifa_join_mc(struct ifnet *ifp, struct in6_aliasreq *ifra,
670 struct in6_ifaddr *ia, int flags, struct in6_multi **in6m_sol)
671 {
672 char ip6buf[INET6_ADDRSTRLEN];
673 struct in6_addr mltaddr;
674 struct in6_multi_mship *imm;
675 int delay, error;
676
677 KASSERT(in6m_sol != NULL, ("%s: in6m_sol is NULL", __func__));
678
679 /* Join solicited multicast addr for new host id. */
680 if ((error = in6_solicited_node_maddr(&mltaddr, ifp,
681 &ifra->ifra_addr.sin6_addr)) != 0)
682 goto cleanup;
683 delay = error = 0;
684 if ((flags & IN6_IFAUPDATE_DADDELAY)) {
685 /*
686 * We need a random delay for DAD on the address being
687 * configured. It also means delaying transmission of the
688 * corresponding MLD report to avoid report collision.
689 * [RFC 4861, Section 6.3.7]
690 */
691 delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
692 }
693 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
694 if (imm == NULL) {
695 nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
696 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr),
697 if_name(ifp), error));
698 goto cleanup;
699 }
700 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
701 *in6m_sol = imm->i6mm_maddr;
702
703 /*
704 * Join link-local all-nodes address.
705 */
706 mltaddr = in6addr_linklocal_allnodes;
707 if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
708 goto cleanup; /* XXX: should not fail */
709
710 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0);
711 if (imm == NULL) {
712 nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
713 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr),
714 if_name(ifp), error));
715 goto cleanup;
716 }
717 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
718
719 /*
720 * Join node information group address.
721 */
722 delay = 0;
723 if ((flags & IN6_IFAUPDATE_DADDELAY)) {
724 /*
725 * The spec does not say anything about delay for this group,
726 * but the same logic should apply.
727 */
728 delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
729 }
730 if (in6_nigroup(ifp, NULL, -1, &mltaddr) == 0) {
731 /* XXX jinmei */
732 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
733 if (imm == NULL)
734 nd6log((LOG_WARNING,
735 "%s: in6_joingroup failed for %s on %s "
736 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
737 &mltaddr), if_name(ifp), error));
738 /* XXX not very fatal, go on... */
739 else
740 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
741 }
742 if (V_icmp6_nodeinfo_oldmcprefix &&
743 in6_nigroup_oldmcprefix(ifp, NULL, -1, &mltaddr) == 0) {
744 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
745 if (imm == NULL)
746 nd6log((LOG_WARNING,
747 "%s: in6_joingroup failed for %s on %s "
748 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
749 &mltaddr), if_name(ifp), error));
750 /* XXX not very fatal, go on... */
751 else
752 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
753 }
754
755 /*
756 * Join interface-local all-nodes address.
757 * (ff01::1%ifN, and ff01::%ifN/32)
758 */
759 mltaddr = in6addr_nodelocal_allnodes;
760 if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
761 goto cleanup; /* XXX: should not fail */
762
763 imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0);
764 if (imm == NULL) {
765 nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
766 "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
767 &mltaddr), if_name(ifp), error));
768 goto cleanup;
769 }
770 LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
771
772 cleanup:
773 return (error);
774 }
775
776 /*
777 * Update parameters of an IPv6 interface address.
778 * If necessary, a new entry is created and linked into address chains.
779 * This function is separated from in6_control().
780 */
781 int
in6_update_ifa(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia,int flags)782 in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
783 struct in6_ifaddr *ia, int flags)
784 {
785 int error, hostIsNew = 0;
786
787 if ((error = in6_validate_ifra(ifp, ifra, ia, flags)) != 0)
788 return (error);
789
790 if (ia == NULL) {
791 hostIsNew = 1;
792 if ((ia = in6_alloc_ifa(ifp, ifra, flags)) == NULL)
793 return (ENOBUFS);
794 }
795
796 error = in6_update_ifa_internal(ifp, ifra, ia, hostIsNew, flags);
797 if (error != 0) {
798 if (hostIsNew != 0) {
799 in6_unlink_ifa(ia, ifp);
800 ifa_free(&ia->ia_ifa);
801 }
802 return (error);
803 }
804
805 if (hostIsNew)
806 error = in6_broadcast_ifa(ifp, ifra, ia, flags);
807
808 return (error);
809 }
810
811 /*
812 * Fill in basic IPv6 address request info.
813 */
814 void
in6_prepare_ifra(struct in6_aliasreq * ifra,const struct in6_addr * addr,const struct in6_addr * mask)815 in6_prepare_ifra(struct in6_aliasreq *ifra, const struct in6_addr *addr,
816 const struct in6_addr *mask)
817 {
818
819 memset(ifra, 0, sizeof(struct in6_aliasreq));
820
821 ifra->ifra_addr.sin6_family = AF_INET6;
822 ifra->ifra_addr.sin6_len = sizeof(struct sockaddr_in6);
823 if (addr != NULL)
824 ifra->ifra_addr.sin6_addr = *addr;
825
826 ifra->ifra_prefixmask.sin6_family = AF_INET6;
827 ifra->ifra_prefixmask.sin6_len = sizeof(struct sockaddr_in6);
828 if (mask != NULL)
829 ifra->ifra_prefixmask.sin6_addr = *mask;
830 }
831
832 static int
in6_validate_ifra(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia,int flags)833 in6_validate_ifra(struct ifnet *ifp, struct in6_aliasreq *ifra,
834 struct in6_ifaddr *ia, int flags)
835 {
836 int plen = -1;
837 struct sockaddr_in6 dst6;
838 struct in6_addrlifetime *lt;
839 char ip6buf[INET6_ADDRSTRLEN];
840
841 /* Validate parameters */
842 if (ifp == NULL || ifra == NULL) /* this maybe redundant */
843 return (EINVAL);
844
845 /*
846 * The destination address for a p2p link must have a family
847 * of AF_UNSPEC or AF_INET6.
848 */
849 if ((ifp->if_flags & IFF_POINTOPOINT) != 0 &&
850 ifra->ifra_dstaddr.sin6_family != AF_INET6 &&
851 ifra->ifra_dstaddr.sin6_family != AF_UNSPEC)
852 return (EAFNOSUPPORT);
853
854 /*
855 * Validate address
856 */
857 if (ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6) ||
858 ifra->ifra_addr.sin6_family != AF_INET6)
859 return (EINVAL);
860
861 /*
862 * validate ifra_prefixmask. don't check sin6_family, netmask
863 * does not carry fields other than sin6_len.
864 */
865 if (ifra->ifra_prefixmask.sin6_len > sizeof(struct sockaddr_in6))
866 return (EINVAL);
867 /*
868 * Because the IPv6 address architecture is classless, we require
869 * users to specify a (non 0) prefix length (mask) for a new address.
870 * We also require the prefix (when specified) mask is valid, and thus
871 * reject a non-consecutive mask.
872 */
873 if (ia == NULL && ifra->ifra_prefixmask.sin6_len == 0)
874 return (EINVAL);
875 if (ifra->ifra_prefixmask.sin6_len != 0) {
876 plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr,
877 (u_char *)&ifra->ifra_prefixmask +
878 ifra->ifra_prefixmask.sin6_len);
879 if (plen <= 0)
880 return (EINVAL);
881 } else {
882 /*
883 * In this case, ia must not be NULL. We just use its prefix
884 * length.
885 */
886 plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL);
887 }
888 /*
889 * If the destination address on a p2p interface is specified,
890 * and the address is a scoped one, validate/set the scope
891 * zone identifier.
892 */
893 dst6 = ifra->ifra_dstaddr;
894 if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) != 0 &&
895 (dst6.sin6_family == AF_INET6)) {
896 struct in6_addr in6_tmp;
897 u_int32_t zoneid;
898
899 in6_tmp = dst6.sin6_addr;
900 if (in6_setscope(&in6_tmp, ifp, &zoneid))
901 return (EINVAL); /* XXX: should be impossible */
902
903 if (dst6.sin6_scope_id != 0) {
904 if (dst6.sin6_scope_id != zoneid)
905 return (EINVAL);
906 } else /* user omit to specify the ID. */
907 dst6.sin6_scope_id = zoneid;
908
909 /* convert into the internal form */
910 if (sa6_embedscope(&dst6, 0))
911 return (EINVAL); /* XXX: should be impossible */
912 }
913 /* Modify original ifra_dstaddr to reflect changes */
914 ifra->ifra_dstaddr = dst6;
915
916 /*
917 * The destination address can be specified only for a p2p or a
918 * loopback interface. If specified, the corresponding prefix length
919 * must be 128.
920 */
921 if (ifra->ifra_dstaddr.sin6_family == AF_INET6) {
922 if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) == 0) {
923 /* XXX: noisy message */
924 nd6log((LOG_INFO, "in6_update_ifa: a destination can "
925 "be specified for a p2p or a loopback IF only\n"));
926 return (EINVAL);
927 }
928 if (plen != 128) {
929 nd6log((LOG_INFO, "in6_update_ifa: prefixlen should "
930 "be 128 when dstaddr is specified\n"));
931 return (EINVAL);
932 }
933 }
934 /* lifetime consistency check */
935 lt = &ifra->ifra_lifetime;
936 if (lt->ia6t_pltime > lt->ia6t_vltime)
937 return (EINVAL);
938 if (lt->ia6t_vltime == 0) {
939 /*
940 * the following log might be noisy, but this is a typical
941 * configuration mistake or a tool's bug.
942 */
943 nd6log((LOG_INFO,
944 "in6_update_ifa: valid lifetime is 0 for %s\n",
945 ip6_sprintf(ip6buf, &ifra->ifra_addr.sin6_addr)));
946
947 if (ia == NULL)
948 return (0); /* there's nothing to do */
949 }
950
951 /* Check prefix mask */
952 if (ia != NULL && ifra->ifra_prefixmask.sin6_len != 0) {
953 /*
954 * We prohibit changing the prefix length of an existing
955 * address, because
956 * + such an operation should be rare in IPv6, and
957 * + the operation would confuse prefix management.
958 */
959 if (ia->ia_prefixmask.sin6_len != 0 &&
960 in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL) != plen) {
961 nd6log((LOG_INFO, "in6_validate_ifa: the prefix length "
962 "of an existing %s address should not be changed\n",
963 ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
964
965 return (EINVAL);
966 }
967 }
968
969 return (0);
970 }
971
972 /*
973 * Allocate a new ifaddr and link it into chains.
974 */
975 static struct in6_ifaddr *
in6_alloc_ifa(struct ifnet * ifp,struct in6_aliasreq * ifra,int flags)976 in6_alloc_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, int flags)
977 {
978 struct in6_ifaddr *ia;
979
980 /*
981 * When in6_alloc_ifa() is called in a process of a received
982 * RA, it is called under an interrupt context. So, we should
983 * call malloc with M_NOWAIT.
984 */
985 ia = (struct in6_ifaddr *)ifa_alloc(sizeof(*ia), M_NOWAIT);
986 if (ia == NULL)
987 return (NULL);
988 LIST_INIT(&ia->ia6_memberships);
989 /* Initialize the address and masks, and put time stamp */
990 ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr;
991 ia->ia_addr.sin6_family = AF_INET6;
992 ia->ia_addr.sin6_len = sizeof(ia->ia_addr);
993 /* XXX: Can we assign ,sin6_addr and skip the rest? */
994 ia->ia_addr = ifra->ifra_addr;
995 ia->ia6_createtime = time_uptime;
996 if ((ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) != 0) {
997 /*
998 * Some functions expect that ifa_dstaddr is not
999 * NULL for p2p interfaces.
1000 */
1001 ia->ia_ifa.ifa_dstaddr =
1002 (struct sockaddr *)&ia->ia_dstaddr;
1003 } else {
1004 ia->ia_ifa.ifa_dstaddr = NULL;
1005 }
1006
1007 /* set prefix mask if any */
1008 ia->ia_ifa.ifa_netmask = (struct sockaddr *)&ia->ia_prefixmask;
1009 if (ifra->ifra_prefixmask.sin6_len != 0) {
1010 ia->ia_prefixmask.sin6_family = AF_INET6;
1011 ia->ia_prefixmask.sin6_len = ifra->ifra_prefixmask.sin6_len;
1012 ia->ia_prefixmask.sin6_addr = ifra->ifra_prefixmask.sin6_addr;
1013 }
1014
1015 ia->ia_ifp = ifp;
1016 ifa_ref(&ia->ia_ifa); /* if_addrhead */
1017 IF_ADDR_WLOCK(ifp);
1018 CK_STAILQ_INSERT_TAIL(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
1019 IF_ADDR_WUNLOCK(ifp);
1020
1021 ifa_ref(&ia->ia_ifa); /* in6_ifaddrhead */
1022 IN6_IFADDR_WLOCK();
1023 CK_STAILQ_INSERT_TAIL(&V_in6_ifaddrhead, ia, ia_link);
1024 CK_LIST_INSERT_HEAD(IN6ADDR_HASH(&ia->ia_addr.sin6_addr), ia, ia6_hash);
1025 IN6_IFADDR_WUNLOCK();
1026
1027 return (ia);
1028 }
1029
1030 time_t
in6_expire_time(uint32_t ltime)1031 in6_expire_time(uint32_t ltime)
1032 {
1033 if (ltime == ND6_INFINITE_LIFETIME)
1034 return (0);
1035 else
1036 return (time_uptime + ltime);
1037 }
1038
1039 /*
1040 * Update/configure interface address parameters:
1041 *
1042 * 1) Update lifetime
1043 * 2) Update interface metric ad flags
1044 * 3) Notify other subsystems
1045 */
1046 static int
in6_update_ifa_internal(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia,int hostIsNew,int flags)1047 in6_update_ifa_internal(struct ifnet *ifp, struct in6_aliasreq *ifra,
1048 struct in6_ifaddr *ia, int hostIsNew, int flags)
1049 {
1050 int error;
1051
1052 /* update timestamp */
1053 ia->ia6_updatetime = time_uptime;
1054
1055 /*
1056 * Set lifetimes. We do not refer to ia6t_expire and ia6t_preferred
1057 * to see if the address is deprecated or invalidated, but initialize
1058 * these members for applications.
1059 */
1060 ia->ia6_lifetime = ifra->ifra_lifetime;
1061 ia->ia6_lifetime.ia6t_expire =
1062 in6_expire_time(ifra->ifra_lifetime.ia6t_vltime);
1063 ia->ia6_lifetime.ia6t_preferred =
1064 in6_expire_time(ifra->ifra_lifetime.ia6t_pltime);
1065
1066 /*
1067 * backward compatibility - if IN6_IFF_DEPRECATED is set from the
1068 * userland, make it deprecated.
1069 */
1070 if ((ifra->ifra_flags & IN6_IFF_DEPRECATED) != 0) {
1071 ia->ia6_lifetime.ia6t_pltime = 0;
1072 ia->ia6_lifetime.ia6t_preferred = time_uptime;
1073 }
1074
1075 /*
1076 * configure address flags.
1077 */
1078 ia->ia6_flags = ifra->ifra_flags;
1079
1080 /*
1081 * Make the address tentative before joining multicast addresses,
1082 * so that corresponding MLD responses would not have a tentative
1083 * source address.
1084 */
1085 ia->ia6_flags &= ~IN6_IFF_DUPLICATED; /* safety */
1086
1087 /*
1088 * DAD should be performed for an new address or addresses on
1089 * an interface with ND6_IFF_IFDISABLED.
1090 */
1091 if (in6if_do_dad(ifp) &&
1092 (hostIsNew || (ifp->if_inet6->nd_flags & ND6_IFF_IFDISABLED)))
1093 ia->ia6_flags |= IN6_IFF_TENTATIVE;
1094
1095 /* notify other subsystems */
1096 error = in6_notify_ifa(ifp, ia, ifra, hostIsNew);
1097
1098 return (error);
1099 }
1100
1101 /*
1102 * Do link-level ifa job:
1103 * 1) Add lle entry for added address
1104 * 2) Notifies routing socket users about new address
1105 * 3) join appropriate multicast group
1106 * 4) start DAD if enabled
1107 */
1108 static int
in6_broadcast_ifa(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia,int flags)1109 in6_broadcast_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
1110 struct in6_ifaddr *ia, int flags)
1111 {
1112 struct in6_multi *in6m_sol;
1113 int error = 0;
1114
1115 /* Add local address to lltable, if necessary (ex. on p2p link). */
1116 if ((error = nd6_add_ifa_lle(ia)) != 0) {
1117 in6_purgeaddr(&ia->ia_ifa);
1118 ifa_free(&ia->ia_ifa);
1119 return (error);
1120 }
1121
1122 /* Join necessary multicast groups. */
1123 in6m_sol = NULL;
1124 if ((ifp->if_flags & IFF_MULTICAST) != 0) {
1125 error = in6_update_ifa_join_mc(ifp, ifra, ia, flags, &in6m_sol);
1126 if (error != 0) {
1127 in6_purgeaddr(&ia->ia_ifa);
1128 ifa_free(&ia->ia_ifa);
1129 return (error);
1130 }
1131 }
1132
1133 /* Perform DAD, if the address is TENTATIVE. */
1134 if ((ia->ia6_flags & IN6_IFF_TENTATIVE)) {
1135 int delay, mindelay, maxdelay;
1136
1137 delay = 0;
1138 if ((flags & IN6_IFAUPDATE_DADDELAY)) {
1139 /*
1140 * We need to impose a delay before sending an NS
1141 * for DAD. Check if we also needed a delay for the
1142 * corresponding MLD message. If we did, the delay
1143 * should be larger than the MLD delay (this could be
1144 * relaxed a bit, but this simple logic is at least
1145 * safe).
1146 * XXX: Break data hiding guidelines and look at
1147 * state for the solicited multicast group.
1148 */
1149 mindelay = 0;
1150 if (in6m_sol != NULL &&
1151 in6m_sol->in6m_state == MLD_REPORTING_MEMBER) {
1152 mindelay = in6m_sol->in6m_timer;
1153 }
1154 maxdelay = MAX_RTR_SOLICITATION_DELAY * hz;
1155 if (maxdelay - mindelay == 0)
1156 delay = 0;
1157 else {
1158 delay =
1159 (arc4random() % (maxdelay - mindelay)) +
1160 mindelay;
1161 }
1162 }
1163 nd6_dad_start((struct ifaddr *)ia, delay);
1164 }
1165
1166 in6_newaddrmsg(ia, RTM_ADD);
1167 ifa_free(&ia->ia_ifa);
1168 return (error);
1169 }
1170
1171 /*
1172 * Adds or deletes interface route for p2p ifa.
1173 * Returns 0 on success or errno.
1174 */
1175 static int
in6_handle_dstaddr_rtrequest(int cmd,struct in6_ifaddr * ia)1176 in6_handle_dstaddr_rtrequest(int cmd, struct in6_ifaddr *ia)
1177 {
1178 struct epoch_tracker et;
1179 struct ifaddr *ifa = &ia->ia_ifa;
1180 int error;
1181
1182 /* Prepare gateway */
1183 struct sockaddr_dl_short sdl = {
1184 .sdl_family = AF_LINK,
1185 .sdl_len = sizeof(struct sockaddr_dl_short),
1186 .sdl_type = ifa->ifa_ifp->if_type,
1187 .sdl_index = ifa->ifa_ifp->if_index,
1188 };
1189
1190 struct sockaddr_in6 dst = {
1191 .sin6_family = AF_INET6,
1192 .sin6_len = sizeof(struct sockaddr_in6),
1193 .sin6_addr = ia->ia_dstaddr.sin6_addr,
1194 };
1195
1196 struct rt_addrinfo info = {
1197 .rti_ifa = ifa,
1198 .rti_ifp = ifa->ifa_ifp,
1199 .rti_flags = RTF_PINNED | RTF_HOST,
1200 .rti_info = {
1201 [RTAX_DST] = (struct sockaddr *)&dst,
1202 [RTAX_GATEWAY] = (struct sockaddr *)&sdl,
1203 },
1204 };
1205 /* Don't set additional per-gw filters on removal */
1206
1207 NET_EPOCH_ENTER(et);
1208 error = rib_handle_ifaddr_info(ifa->ifa_ifp->if_fib, cmd, &info);
1209 NET_EPOCH_EXIT(et);
1210
1211 return (error);
1212 }
1213
1214 static bool
ifa_is_p2p(struct in6_ifaddr * ia)1215 ifa_is_p2p(struct in6_ifaddr *ia)
1216 {
1217 int plen;
1218
1219 plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */
1220
1221 if ((plen == 128) && (ia->ia_dstaddr.sin6_family == AF_INET6) &&
1222 !IN6_ARE_ADDR_EQUAL(&ia->ia_addr.sin6_addr, &ia->ia_dstaddr.sin6_addr))
1223 return (true);
1224
1225 return (false);
1226 }
1227
1228 int
in6_addifaddr(struct ifnet * ifp,struct in6_aliasreq * ifra,struct in6_ifaddr * ia)1229 in6_addifaddr(struct ifnet *ifp, struct in6_aliasreq *ifra, struct in6_ifaddr *ia)
1230 {
1231 struct nd_prefixctl pr0;
1232 struct nd_prefix *pr;
1233 int carp_attached = 0;
1234 int error;
1235
1236 /*
1237 * Check if bridge wants to allow adding addrs to member interfaces.
1238 */
1239 if (ifp->if_bridge != NULL && ifp->if_type != IFT_GIF &&
1240 bridge_member_ifaddrs_p != NULL) {
1241 if (bridge_member_ifaddrs_p()) {
1242 if_printf(ifp, "WARNING: Assigning an IP address to "
1243 "an interface which is also a bridge member is "
1244 "deprecated and will be unsupported in a future "
1245 "release.\n");
1246 } else {
1247 error = EINVAL;
1248 goto out;
1249 }
1250 }
1251
1252 /*
1253 * first, make or update the interface address structure,
1254 * and link it to the list.
1255 */
1256 if ((error = in6_update_ifa(ifp, ifra, ia, 0)) != 0)
1257 goto out;
1258 if (ia != NULL) {
1259 if (ia->ia_ifa.ifa_carp)
1260 (*carp_detach_p)(&ia->ia_ifa, true);
1261 ifa_free(&ia->ia_ifa);
1262 }
1263 if ((ia = in6ifa_ifpwithaddr(ifp, &ifra->ifra_addr.sin6_addr)) == NULL) {
1264 /*
1265 * this can happen when the user specify the 0 valid
1266 * lifetime.
1267 */
1268 return (0);
1269 }
1270
1271 if (ifra->ifra_vhid > 0) {
1272 if (carp_attach_p != NULL)
1273 error = (*carp_attach_p)(&ia->ia_ifa,
1274 ifra->ifra_vhid);
1275 else
1276 error = EPROTONOSUPPORT;
1277 if (error)
1278 goto out;
1279 else
1280 carp_attached = 1;
1281 }
1282
1283 /*
1284 * then, make the prefix on-link on the interface.
1285 * XXX: we'd rather create the prefix before the address, but
1286 * we need at least one address to install the corresponding
1287 * interface route, so we configure the address first.
1288 */
1289
1290 /*
1291 * convert mask to prefix length (prefixmask has already
1292 * been validated in in6_update_ifa().
1293 */
1294 bzero(&pr0, sizeof(pr0));
1295 pr0.ndpr_ifp = ifp;
1296 pr0.ndpr_plen = ia->ia_plen =
1297 in6_mask2len(&ifra->ifra_prefixmask.sin6_addr, NULL);
1298 if (pr0.ndpr_plen == 128) {
1299 /* we don't need to install a host route. */
1300 goto aifaddr_out;
1301 }
1302 pr0.ndpr_prefix = ifra->ifra_addr;
1303 /* apply the mask for safety. */
1304 IN6_MASK_ADDR(&pr0.ndpr_prefix.sin6_addr,
1305 &ifra->ifra_prefixmask.sin6_addr);
1306
1307 /*
1308 * XXX: since we don't have an API to set prefix (not address)
1309 * lifetimes, we just use the same lifetimes as addresses.
1310 * The (temporarily) installed lifetimes can be overridden by
1311 * later advertised RAs (when accept_rtadv is non 0), which is
1312 * an intended behavior.
1313 */
1314 pr0.ndpr_raf_onlink = 1; /* should be configurable? */
1315 pr0.ndpr_raf_auto =
1316 ((ifra->ifra_flags & IN6_IFF_AUTOCONF) != 0);
1317 pr0.ndpr_vltime = ifra->ifra_lifetime.ia6t_vltime;
1318 pr0.ndpr_pltime = ifra->ifra_lifetime.ia6t_pltime;
1319
1320 /* add the prefix if not yet. */
1321 if ((pr = nd6_prefix_lookup(&pr0)) == NULL) {
1322 /*
1323 * nd6_prelist_add will install the corresponding
1324 * interface route.
1325 */
1326 if ((error = nd6_prelist_add(&pr0, &pr)) != 0) {
1327 if (carp_attached)
1328 (*carp_detach_p)(&ia->ia_ifa, false);
1329 goto out;
1330 }
1331 } else if (pr->ndpr_raf_onlink) {
1332 time_t expiry;
1333
1334 /*
1335 * If the prefix already exists, update lifetimes, but avoid
1336 * shortening them.
1337 */
1338 ND6_WLOCK();
1339 expiry = in6_expire_time(pr0.ndpr_pltime);
1340 if (pr->ndpr_preferred != 0 &&
1341 (pr->ndpr_preferred < expiry || expiry == 0)) {
1342 pr->ndpr_pltime = pr0.ndpr_pltime;
1343 pr->ndpr_preferred = expiry;
1344 }
1345 expiry = in6_expire_time(pr0.ndpr_vltime);
1346 if (pr->ndpr_expire != 0 &&
1347 (pr->ndpr_expire < expiry || expiry == 0)) {
1348 pr->ndpr_vltime = pr0.ndpr_vltime;
1349 pr->ndpr_expire = expiry;
1350 }
1351 pr->ndpr_lastupdate = time_uptime;
1352 ND6_WUNLOCK();
1353 }
1354
1355 /* relate the address to the prefix */
1356 if (ia->ia6_ndpr == NULL) {
1357 ia->ia6_ndpr = pr;
1358 pr->ndpr_addrcnt++;
1359
1360 /*
1361 * If this is the first autoconf address from the
1362 * prefix, create a temporary address as well
1363 * (when required).
1364 */
1365 if ((ia->ia6_flags & IN6_IFF_AUTOCONF) &&
1366 V_ip6_use_tempaddr && pr->ndpr_addrcnt == 1) {
1367 int e;
1368 if ((e = in6_tmpifadd(ia, 1, 0)) != 0) {
1369 log(LOG_NOTICE, "in6_control: failed "
1370 "to create a temporary address, "
1371 "errno=%d\n", e);
1372 }
1373 }
1374 }
1375 nd6_prefix_rele(pr);
1376
1377 /*
1378 * this might affect the status of autoconfigured addresses,
1379 * that is, this address might make other addresses detached.
1380 */
1381 pfxlist_onlink_check();
1382
1383 aifaddr_out:
1384 /*
1385 * Try to clear the flag when a new IPv6 address is added
1386 * onto an IFDISABLED interface and it succeeds.
1387 */
1388 if (ifp->if_inet6->nd_flags & ND6_IFF_IFDISABLED) {
1389 struct in6_ndireq nd;
1390
1391 memset(&nd, 0, sizeof(nd));
1392 nd.ndi.flags = ifp->if_inet6->nd_flags;
1393 nd.ndi.flags &= ~ND6_IFF_IFDISABLED;
1394 if (nd6_ioctl(SIOCSIFINFO_FLAGS, (caddr_t)&nd, ifp) < 0)
1395 log(LOG_NOTICE, "SIOCAIFADDR_IN6: "
1396 "SIOCSIFINFO_FLAGS for -ifdisabled "
1397 "failed.");
1398 /*
1399 * Ignore failure of clearing the flag intentionally.
1400 * The failure means address duplication was detected.
1401 */
1402 }
1403 error = 0;
1404
1405 out:
1406 if (ia != NULL)
1407 ifa_free(&ia->ia_ifa);
1408 return (error);
1409 }
1410
1411 void
in6_purgeaddr(struct ifaddr * ifa)1412 in6_purgeaddr(struct ifaddr *ifa)
1413 {
1414 struct ifnet *ifp = ifa->ifa_ifp;
1415 struct in6_ifaddr *ia = (struct in6_ifaddr *) ifa;
1416 struct in6_multi_mship *imm;
1417 int error;
1418
1419 if (ifa->ifa_carp)
1420 (*carp_detach_p)(ifa, false);
1421
1422 /*
1423 * Remove the loopback route to the interface address.
1424 * The check for the current setting of "nd6_useloopback"
1425 * is not needed.
1426 */
1427 if (ia->ia_flags & IFA_RTSELF) {
1428 error = ifa_del_loopback_route((struct ifaddr *)ia,
1429 (struct sockaddr *)&ia->ia_addr);
1430 if (error == 0)
1431 ia->ia_flags &= ~IFA_RTSELF;
1432 }
1433
1434 /* make sure there are no queued ND6 */
1435 nd6_queue_stop(ifa);
1436
1437 /* stop DAD processing */
1438 nd6_dad_stop(ifa);
1439
1440 /* Leave multicast groups. */
1441 while ((imm = LIST_FIRST(&ia->ia6_memberships)) != NULL) {
1442 LIST_REMOVE(imm, i6mm_chain);
1443 if (imm->i6mm_maddr != NULL)
1444 in6_leavegroup(imm->i6mm_maddr, NULL);
1445 free(imm, M_IP6MADDR);
1446 }
1447 /* Check if we need to remove p2p route */
1448 if ((ia->ia_flags & IFA_ROUTE) && ifa_is_p2p(ia)) {
1449 error = in6_handle_dstaddr_rtrequest(RTM_DELETE, ia);
1450 if (error != 0)
1451 log(LOG_INFO, "%s: err=%d, destination address delete "
1452 "failed\n", __func__, error);
1453 ia->ia_flags &= ~IFA_ROUTE;
1454 }
1455
1456 in6_newaddrmsg(ia, RTM_DELETE);
1457 in6_unlink_ifa(ia, ifp);
1458 }
1459
1460 /*
1461 * Removes @ia from the corresponding interfaces and unlinks corresponding
1462 * prefix if no addresses are using it anymore.
1463 */
1464 void
in6_purgeifaddr(struct in6_ifaddr * ia)1465 in6_purgeifaddr(struct in6_ifaddr *ia)
1466 {
1467 struct nd_prefix *pr;
1468
1469 /*
1470 * If the address being deleted is the only one that owns
1471 * the corresponding prefix, expire the prefix as well.
1472 * XXX: theoretically, we don't have to worry about such
1473 * relationship, since we separate the address management
1474 * and the prefix management. We do this, however, to provide
1475 * as much backward compatibility as possible in terms of
1476 * the ioctl operation.
1477 * Note that in6_purgeaddr() will decrement ndpr_addrcnt.
1478 */
1479 pr = ia->ia6_ndpr;
1480 in6_purgeaddr(&ia->ia_ifa);
1481 if (pr != NULL && pr->ndpr_addrcnt == 0) {
1482 ND6_WLOCK();
1483 nd6_prefix_unlink(pr, NULL);
1484 ND6_WUNLOCK();
1485 nd6_prefix_del(pr);
1486 }
1487 }
1488
1489
1490 static void
in6_unlink_ifa(struct in6_ifaddr * ia,struct ifnet * ifp)1491 in6_unlink_ifa(struct in6_ifaddr *ia, struct ifnet *ifp)
1492 {
1493 char ip6buf[INET6_ADDRSTRLEN];
1494 int remove_lle;
1495
1496 IF_ADDR_WLOCK(ifp);
1497 CK_STAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifaddr, ifa_link);
1498 IF_ADDR_WUNLOCK(ifp);
1499 ifa_free(&ia->ia_ifa); /* if_addrhead */
1500
1501 /*
1502 * Defer the release of what might be the last reference to the
1503 * in6_ifaddr so that it can't be freed before the remainder of the
1504 * cleanup.
1505 */
1506 IN6_IFADDR_WLOCK();
1507 CK_STAILQ_REMOVE(&V_in6_ifaddrhead, ia, in6_ifaddr, ia_link);
1508 CK_LIST_REMOVE(ia, ia6_hash);
1509 IN6_IFADDR_WUNLOCK();
1510
1511 /*
1512 * Release the reference to the base prefix. There should be a
1513 * positive reference.
1514 */
1515 remove_lle = 0;
1516 if (ia->ia6_ndpr != NULL) {
1517 ia->ia6_ndpr->ndpr_addrcnt--;
1518 /* Do not delete lles within prefix if refcont != 0 */
1519 if (ia->ia6_ndpr->ndpr_addrcnt == 0)
1520 remove_lle = 1;
1521 ia->ia6_ndpr = NULL;
1522 } else if (ia->ia_plen < 128) {
1523 nd6log((LOG_NOTICE,
1524 "in6_unlink_ifa: autoconf'ed address "
1525 "%s has no prefix\n", ip6_sprintf(ip6buf, IA6_IN6(ia))));
1526 }
1527
1528 nd6_rem_ifa_lle(ia, remove_lle);
1529
1530 /*
1531 * Also, if the address being removed is autoconf'ed, call
1532 * pfxlist_onlink_check() since the release might affect the status of
1533 * other (detached) addresses.
1534 */
1535 if ((ia->ia6_flags & IN6_IFF_AUTOCONF)) {
1536 pfxlist_onlink_check();
1537 }
1538 ifa_free(&ia->ia_ifa); /* in6_ifaddrhead */
1539 }
1540
1541 /*
1542 * Notifies other subsystems about address change/arrival:
1543 * 1) Notifies device handler on the first IPv6 address assignment
1544 * 2) Handle routing table changes for P2P links and route
1545 * 3) Handle routing table changes for address host route
1546 */
1547 static int
in6_notify_ifa(struct ifnet * ifp,struct in6_ifaddr * ia,struct in6_aliasreq * ifra,int hostIsNew)1548 in6_notify_ifa(struct ifnet *ifp, struct in6_ifaddr *ia,
1549 struct in6_aliasreq *ifra, int hostIsNew)
1550 {
1551 int error = 0, ifacount = 0;
1552 struct ifaddr *ifa;
1553 struct sockaddr_in6 *pdst;
1554 char ip6buf[INET6_ADDRSTRLEN];
1555
1556 /*
1557 * Give the interface a chance to initialize
1558 * if this is its first address,
1559 */
1560 if (hostIsNew != 0) {
1561 struct epoch_tracker et;
1562
1563 NET_EPOCH_ENTER(et);
1564 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1565 if (ifa->ifa_addr->sa_family != AF_INET6)
1566 continue;
1567 ifacount++;
1568 }
1569 NET_EPOCH_EXIT(et);
1570 }
1571
1572 if (ifacount <= 1 && ifp->if_ioctl) {
1573 error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
1574 if (error)
1575 goto done;
1576 }
1577
1578 /*
1579 * If a new destination address is specified, scrub the old one and
1580 * install the new destination. Note that the interface must be
1581 * p2p or loopback.
1582 */
1583 pdst = &ifra->ifra_dstaddr;
1584 if (pdst->sin6_family == AF_INET6 &&
1585 !IN6_ARE_ADDR_EQUAL(&pdst->sin6_addr, &ia->ia_dstaddr.sin6_addr)) {
1586 if ((ia->ia_flags & IFA_ROUTE) != 0 &&
1587 (in6_handle_dstaddr_rtrequest(RTM_DELETE, ia) != 0)) {
1588 nd6log((LOG_ERR, "in6_update_ifa_internal: failed to "
1589 "remove a route to the old destination: %s\n",
1590 ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
1591 /* proceed anyway... */
1592 } else
1593 ia->ia_flags &= ~IFA_ROUTE;
1594 ia->ia_dstaddr = *pdst;
1595 }
1596
1597 /*
1598 * If a new destination address is specified for a point-to-point
1599 * interface, install a route to the destination as an interface
1600 * direct route.
1601 * XXX: the logic below rejects assigning multiple addresses on a p2p
1602 * interface that share the same destination.
1603 */
1604 if (!(ia->ia_flags & IFA_ROUTE) && ifa_is_p2p(ia)) {
1605 error = in6_handle_dstaddr_rtrequest(RTM_ADD, ia);
1606 if (error)
1607 goto done;
1608 ia->ia_flags |= IFA_ROUTE;
1609 }
1610
1611 /*
1612 * add a loopback route to self if not exists
1613 */
1614 if (!(ia->ia_flags & IFA_RTSELF) && V_nd6_useloopback) {
1615 error = ifa_add_loopback_route((struct ifaddr *)ia,
1616 (struct sockaddr *)&ia->ia_addr);
1617 if (error == 0)
1618 ia->ia_flags |= IFA_RTSELF;
1619 }
1620 done:
1621 WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
1622 "Invoking IPv6 network device address event may sleep");
1623
1624 ifa_ref(&ia->ia_ifa);
1625 EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa,
1626 IFADDR_EVENT_ADD);
1627 ifa_free(&ia->ia_ifa);
1628
1629 return (error);
1630 }
1631
1632 /*
1633 * Find an IPv6 interface link-local address specific to an interface.
1634 * ifaddr is returned referenced.
1635 */
1636 struct in6_ifaddr *
in6ifa_ifpforlinklocal(struct ifnet * ifp,int ignoreflags)1637 in6ifa_ifpforlinklocal(struct ifnet *ifp, int ignoreflags)
1638 {
1639 struct ifaddr *ifa;
1640
1641 NET_EPOCH_ASSERT();
1642
1643 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1644 if (ifa->ifa_addr->sa_family != AF_INET6)
1645 continue;
1646 if (IN6_IS_ADDR_LINKLOCAL(IFA_IN6(ifa))) {
1647 if ((((struct in6_ifaddr *)ifa)->ia6_flags &
1648 ignoreflags) != 0)
1649 continue;
1650 ifa_ref(ifa);
1651 break;
1652 }
1653 }
1654
1655 return ((struct in6_ifaddr *)ifa);
1656 }
1657
1658 /*
1659 * find the interface address corresponding to a given IPv6 address.
1660 * ifaddr is returned referenced if @referenced flag is set.
1661 */
1662 struct in6_ifaddr *
in6ifa_ifwithaddr(const struct in6_addr * addr,uint32_t zoneid,bool referenced)1663 in6ifa_ifwithaddr(const struct in6_addr *addr, uint32_t zoneid, bool referenced)
1664 {
1665 struct rm_priotracker in6_ifa_tracker;
1666 struct in6_ifaddr *ia;
1667
1668 IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1669 CK_LIST_FOREACH(ia, IN6ADDR_HASH(addr), ia6_hash) {
1670 if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), addr)) {
1671 if (zoneid != 0 &&
1672 zoneid != ia->ia_addr.sin6_scope_id)
1673 continue;
1674 if (referenced)
1675 ifa_ref(&ia->ia_ifa);
1676 break;
1677 }
1678 }
1679 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1680 return (ia);
1681 }
1682
1683 /*
1684 * find the internet address corresponding to a given interface and address.
1685 * ifaddr is returned referenced.
1686 */
1687 struct in6_ifaddr *
in6ifa_ifpwithaddr(struct ifnet * ifp,const struct in6_addr * addr)1688 in6ifa_ifpwithaddr(struct ifnet *ifp, const struct in6_addr *addr)
1689 {
1690 struct epoch_tracker et;
1691 struct ifaddr *ifa;
1692
1693 NET_EPOCH_ENTER(et);
1694 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1695 if (ifa->ifa_addr->sa_family != AF_INET6)
1696 continue;
1697 if (IN6_ARE_ADDR_EQUAL(addr, IFA_IN6(ifa))) {
1698 ifa_ref(ifa);
1699 break;
1700 }
1701 }
1702 NET_EPOCH_EXIT(et);
1703
1704 return ((struct in6_ifaddr *)ifa);
1705 }
1706
1707 /*
1708 * Find a link-local scoped address on ifp and return it if any.
1709 */
1710 struct in6_ifaddr *
in6ifa_llaonifp(struct ifnet * ifp)1711 in6ifa_llaonifp(struct ifnet *ifp)
1712 {
1713 struct epoch_tracker et;
1714 struct sockaddr_in6 *sin6;
1715 struct ifaddr *ifa;
1716
1717 if (ifp->if_inet6->nd_flags & ND6_IFF_IFDISABLED)
1718 return (NULL);
1719 NET_EPOCH_ENTER(et);
1720 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1721 if (ifa->ifa_addr->sa_family != AF_INET6)
1722 continue;
1723 sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
1724 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr) ||
1725 IN6_IS_ADDR_MC_INTFACELOCAL(&sin6->sin6_addr) ||
1726 IN6_IS_ADDR_MC_NODELOCAL(&sin6->sin6_addr))
1727 break;
1728 }
1729 NET_EPOCH_EXIT(et);
1730
1731 return ((struct in6_ifaddr *)ifa);
1732 }
1733
1734 /*
1735 * Convert IP6 address to printable (loggable) representation. Caller
1736 * has to make sure that ip6buf is at least INET6_ADDRSTRLEN long.
1737 */
1738 static char digits[] = "0123456789abcdef";
1739 char *
ip6_sprintf(char * ip6buf,const struct in6_addr * addr)1740 ip6_sprintf(char *ip6buf, const struct in6_addr *addr)
1741 {
1742 int i, cnt = 0, maxcnt = 0, idx = 0, index = 0;
1743 char *cp;
1744 const u_int16_t *a = (const u_int16_t *)addr;
1745 const u_int8_t *d;
1746 int dcolon = 0, zero = 0;
1747
1748 cp = ip6buf;
1749
1750 for (i = 0; i < 8; i++) {
1751 if (*(a + i) == 0) {
1752 cnt++;
1753 if (cnt == 1)
1754 idx = i;
1755 }
1756 else if (maxcnt < cnt) {
1757 maxcnt = cnt;
1758 index = idx;
1759 cnt = 0;
1760 }
1761 }
1762 if (maxcnt < cnt) {
1763 maxcnt = cnt;
1764 index = idx;
1765 }
1766
1767 for (i = 0; i < 8; i++) {
1768 if (dcolon == 1) {
1769 if (*a == 0) {
1770 if (i == 7)
1771 *cp++ = ':';
1772 a++;
1773 continue;
1774 } else
1775 dcolon = 2;
1776 }
1777 if (*a == 0) {
1778 if (dcolon == 0 && *(a + 1) == 0 && i == index) {
1779 if (i == 0)
1780 *cp++ = ':';
1781 *cp++ = ':';
1782 dcolon = 1;
1783 } else {
1784 *cp++ = '0';
1785 *cp++ = ':';
1786 }
1787 a++;
1788 continue;
1789 }
1790 d = (const u_char *)a;
1791 /* Try to eliminate leading zeros in printout like in :0001. */
1792 zero = 1;
1793 *cp = digits[*d >> 4];
1794 if (*cp != '0') {
1795 zero = 0;
1796 cp++;
1797 }
1798 *cp = digits[*d++ & 0xf];
1799 if (zero == 0 || (*cp != '0')) {
1800 zero = 0;
1801 cp++;
1802 }
1803 *cp = digits[*d >> 4];
1804 if (zero == 0 || (*cp != '0')) {
1805 zero = 0;
1806 cp++;
1807 }
1808 *cp++ = digits[*d & 0xf];
1809 *cp++ = ':';
1810 a++;
1811 }
1812 *--cp = '\0';
1813 return (ip6buf);
1814 }
1815
1816 int
in6_localaddr(struct in6_addr * in6)1817 in6_localaddr(struct in6_addr *in6)
1818 {
1819 struct rm_priotracker in6_ifa_tracker;
1820 struct in6_ifaddr *ia;
1821
1822 if (IN6_IS_ADDR_LOOPBACK(in6) || IN6_IS_ADDR_LINKLOCAL(in6))
1823 return 1;
1824
1825 IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1826 CK_STAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) {
1827 if (IN6_ARE_MASKED_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr,
1828 &ia->ia_prefixmask.sin6_addr)) {
1829 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1830 return 1;
1831 }
1832 }
1833 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1834
1835 return (0);
1836 }
1837
1838 /*
1839 * Return 1 if an internet address is for the local host and configured
1840 * on one of its interfaces.
1841 */
1842 int
in6_localip(struct in6_addr * in6)1843 in6_localip(struct in6_addr *in6)
1844 {
1845 struct rm_priotracker in6_ifa_tracker;
1846 struct in6_ifaddr *ia;
1847
1848 IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1849 CK_LIST_FOREACH(ia, IN6ADDR_HASH(in6), ia6_hash) {
1850 if (IN6_ARE_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr)) {
1851 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1852 return (1);
1853 }
1854 }
1855 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1856 return (0);
1857 }
1858
1859 /*
1860 * Like in6_localip(), but FIB-aware and carp(4)-aware.
1861 */
1862 bool
in6_localip_fib(struct in6_addr * in6,uint16_t fib)1863 in6_localip_fib(struct in6_addr *in6, uint16_t fib)
1864 {
1865 struct rm_priotracker in6_ifa_tracker;
1866 struct in6_ifaddr *ia;
1867
1868 IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1869 CK_LIST_FOREACH(ia, IN6ADDR_HASH(in6), ia6_hash) {
1870 if (IN6_ARE_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr) &&
1871 (ia->ia_ifa.ifa_carp == NULL ||
1872 carp_master_p(&ia->ia_ifa)) &&
1873 ia->ia_ifa.ifa_ifp->if_fib == fib) {
1874 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1875 return (true);
1876 }
1877 }
1878 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1879 return (false);
1880 }
1881
1882 /*
1883 * Return 1 if an internet address is configured on an interface.
1884 */
1885 int
in6_ifhasaddr(struct ifnet * ifp,struct in6_addr * addr)1886 in6_ifhasaddr(struct ifnet *ifp, struct in6_addr *addr)
1887 {
1888 struct in6_addr in6;
1889 struct ifaddr *ifa;
1890 struct in6_ifaddr *ia6;
1891
1892 NET_EPOCH_ASSERT();
1893
1894 in6 = *addr;
1895 if (in6_clearscope(&in6))
1896 return (0);
1897 in6_setscope(&in6, ifp, NULL);
1898
1899 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1900 if (ifa->ifa_addr->sa_family != AF_INET6)
1901 continue;
1902 ia6 = (struct in6_ifaddr *)ifa;
1903 if (IN6_ARE_ADDR_EQUAL(&ia6->ia_addr.sin6_addr, &in6))
1904 return (1);
1905 }
1906
1907 return (0);
1908 }
1909
1910 int
in6_is_addr_deprecated(struct sockaddr_in6 * sa6)1911 in6_is_addr_deprecated(struct sockaddr_in6 *sa6)
1912 {
1913 struct rm_priotracker in6_ifa_tracker;
1914 struct in6_ifaddr *ia;
1915
1916 IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1917 CK_LIST_FOREACH(ia, IN6ADDR_HASH(&sa6->sin6_addr), ia6_hash) {
1918 if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), &sa6->sin6_addr)) {
1919 if (ia->ia6_flags & IN6_IFF_DEPRECATED) {
1920 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1921 return (1); /* true */
1922 }
1923 break;
1924 }
1925 }
1926 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1927
1928 return (0); /* false */
1929 }
1930
1931 /*
1932 * return length of part which dst and src are equal
1933 * hard coding...
1934 */
1935 int
in6_matchlen(struct in6_addr * src,struct in6_addr * dst)1936 in6_matchlen(struct in6_addr *src, struct in6_addr *dst)
1937 {
1938 int match = 0;
1939 u_char *s = (u_char *)src, *d = (u_char *)dst;
1940 u_char *lim = s + 16, r;
1941
1942 while (s < lim)
1943 if ((r = (*d++ ^ *s++)) != 0) {
1944 while (r < 128) {
1945 match++;
1946 r <<= 1;
1947 }
1948 break;
1949 } else
1950 match += 8;
1951 return match;
1952 }
1953
1954 /* XXX: to be scope conscious */
1955 int
in6_are_prefix_equal(struct in6_addr * p1,struct in6_addr * p2,int len)1956 in6_are_prefix_equal(struct in6_addr *p1, struct in6_addr *p2, int len)
1957 {
1958 int bytelen, bitlen;
1959
1960 /* sanity check */
1961 if (0 > len || len > 128) {
1962 log(LOG_ERR, "in6_are_prefix_equal: invalid prefix length(%d)\n",
1963 len);
1964 return (0);
1965 }
1966
1967 bytelen = len / 8;
1968 bitlen = len % 8;
1969
1970 if (bcmp(&p1->s6_addr, &p2->s6_addr, bytelen))
1971 return (0);
1972 if (bitlen != 0 &&
1973 p1->s6_addr[bytelen] >> (8 - bitlen) !=
1974 p2->s6_addr[bytelen] >> (8 - bitlen))
1975 return (0);
1976
1977 return (1);
1978 }
1979
1980 void
in6_prefixlen2mask(struct in6_addr * maskp,int len)1981 in6_prefixlen2mask(struct in6_addr *maskp, int len)
1982 {
1983 u_char maskarray[8] = {0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, 0xff};
1984 int bytelen, bitlen, i;
1985
1986 /* sanity check */
1987 if (0 > len || len > 128) {
1988 log(LOG_ERR, "in6_prefixlen2mask: invalid prefix length(%d)\n",
1989 len);
1990 return;
1991 }
1992
1993 bzero(maskp, sizeof(*maskp));
1994 bytelen = len / 8;
1995 bitlen = len % 8;
1996 for (i = 0; i < bytelen; i++)
1997 maskp->s6_addr[i] = 0xff;
1998 if (bitlen)
1999 maskp->s6_addr[bytelen] = maskarray[bitlen - 1];
2000 }
2001
2002 /*
2003 * return the best address out of the same scope. if no address was
2004 * found, return the first valid address from designated IF.
2005 */
2006 struct in6_ifaddr *
in6_ifawithifp(struct ifnet * ifp,struct in6_addr * dst)2007 in6_ifawithifp(struct ifnet *ifp, struct in6_addr *dst)
2008 {
2009 int dst_scope = in6_addrscope(dst), blen = -1, tlen;
2010 struct ifaddr *ifa;
2011 struct in6_ifaddr *besta = NULL;
2012 struct in6_ifaddr *dep[2]; /* last-resort: deprecated */
2013
2014 NET_EPOCH_ASSERT();
2015
2016 dep[0] = dep[1] = NULL;
2017
2018 /*
2019 * We first look for addresses in the same scope.
2020 * If there is one, return it.
2021 * If two or more, return one which matches the dst longest.
2022 * If none, return one of global addresses assigned other ifs.
2023 */
2024 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
2025 if (ifa->ifa_addr->sa_family != AF_INET6)
2026 continue;
2027 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
2028 continue; /* XXX: is there any case to allow anycast? */
2029 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
2030 continue; /* don't use this interface */
2031 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
2032 continue;
2033 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
2034 if (V_ip6_use_deprecated)
2035 dep[0] = (struct in6_ifaddr *)ifa;
2036 continue;
2037 }
2038
2039 if (dst_scope == in6_addrscope(IFA_IN6(ifa))) {
2040 /*
2041 * call in6_matchlen() as few as possible
2042 */
2043 if (besta) {
2044 if (blen == -1)
2045 blen = in6_matchlen(&besta->ia_addr.sin6_addr, dst);
2046 tlen = in6_matchlen(IFA_IN6(ifa), dst);
2047 if (tlen > blen) {
2048 blen = tlen;
2049 besta = (struct in6_ifaddr *)ifa;
2050 }
2051 } else
2052 besta = (struct in6_ifaddr *)ifa;
2053 }
2054 }
2055 if (besta)
2056 return (besta);
2057
2058 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
2059 if (ifa->ifa_addr->sa_family != AF_INET6)
2060 continue;
2061 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
2062 continue; /* XXX: is there any case to allow anycast? */
2063 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
2064 continue; /* don't use this interface */
2065 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
2066 continue;
2067 if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
2068 if (V_ip6_use_deprecated)
2069 dep[1] = (struct in6_ifaddr *)ifa;
2070 continue;
2071 }
2072
2073 return (struct in6_ifaddr *)ifa;
2074 }
2075
2076 /* use the last-resort values, that are, deprecated addresses */
2077 if (dep[0])
2078 return dep[0];
2079 if (dep[1])
2080 return dep[1];
2081
2082 return NULL;
2083 }
2084
2085 /*
2086 * perform DAD when interface becomes IFF_UP.
2087 */
2088 void
in6_if_up(struct ifnet * ifp)2089 in6_if_up(struct ifnet *ifp)
2090 {
2091 struct epoch_tracker et;
2092 struct ifaddr *ifa;
2093 struct in6_ifaddr *ia;
2094
2095 NET_EPOCH_ENTER(et);
2096 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
2097 if (ifa->ifa_addr->sa_family != AF_INET6)
2098 continue;
2099 ia = (struct in6_ifaddr *)ifa;
2100 if (ia->ia6_flags & IN6_IFF_TENTATIVE) {
2101 /*
2102 * The TENTATIVE flag was likely set by hand
2103 * beforehand, implicitly indicating the need for DAD.
2104 * We may be able to skip the random delay in this
2105 * case, but we impose delays just in case.
2106 */
2107 nd6_dad_start(ifa,
2108 arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz));
2109 }
2110 }
2111 NET_EPOCH_EXIT(et);
2112
2113 /*
2114 * special cases, like 6to4, are handled in in6_ifattach
2115 */
2116 in6_ifattach(ifp, NULL);
2117 }
2118
2119 static void
in6_ifevent(void * arg __unused,struct ifnet * ifp,int event)2120 in6_ifevent(void *arg __unused, struct ifnet *ifp, int event)
2121 {
2122 if (event == IFNET_EVENT_UP)
2123 in6_if_up(ifp);
2124 }
2125
2126 static void
in6_init(void * arg __unused)2127 in6_init(void *arg __unused)
2128 {
2129 EVENTHANDLER_REGISTER(ifnet_event, in6_ifevent, NULL, EVENTHANDLER_PRI_ANY);
2130 }
2131 SYSINIT(in6_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, in6_init, NULL);
2132
2133 int
in6if_do_dad(struct ifnet * ifp)2134 in6if_do_dad(struct ifnet *ifp)
2135 {
2136
2137 if ((ifp->if_flags & IFF_LOOPBACK) != 0)
2138 return (0);
2139 if ((ifp->if_flags & IFF_MULTICAST) == 0)
2140 return (0);
2141 if ((ifp->if_inet6->nd_flags &
2142 (ND6_IFF_IFDISABLED | ND6_IFF_NO_DAD)) != 0)
2143 return (0);
2144 return (1);
2145 }
2146
2147 /*
2148 * Provide the length of interface identifiers to be used for the link attached
2149 * to the given interface. The length should be defined in "IPv6 over
2150 * xxx-link" document. Note that address architecture might also define
2151 * the length for a particular set of address prefixes, regardless of the
2152 * link type. As clarified in rfc2462bis, those two definitions should be
2153 * consistent, and those really are as of August 2004.
2154 */
2155 int
in6_if2idlen(struct ifnet * ifp)2156 in6_if2idlen(struct ifnet *ifp)
2157 {
2158 switch (ifp->if_type) {
2159 case IFT_ETHER: /* RFC2464 */
2160 case IFT_PROPVIRTUAL: /* XXX: no RFC. treat it as ether */
2161 case IFT_L2VLAN: /* ditto */
2162 case IFT_BRIDGE: /* bridge(4) only does Ethernet-like links */
2163 case IFT_INFINIBAND:
2164 return (64);
2165 case IFT_PPP: /* RFC2472 */
2166 return (64);
2167 case IFT_FRELAY: /* RFC2590 */
2168 return (64);
2169 case IFT_IEEE1394: /* RFC3146 */
2170 return (64);
2171 case IFT_GIF:
2172 return (64); /* draft-ietf-v6ops-mech-v2-07 */
2173 case IFT_LOOP:
2174 return (64); /* XXX: is this really correct? */
2175 default:
2176 /*
2177 * Unknown link type:
2178 * It might be controversial to use the today's common constant
2179 * of 64 for these cases unconditionally. For full compliance,
2180 * we should return an error in this case. On the other hand,
2181 * if we simply miss the standard for the link type or a new
2182 * standard is defined for a new link type, the IFID length
2183 * is very likely to be the common constant. As a compromise,
2184 * we always use the constant, but make an explicit notice
2185 * indicating the "unknown" case.
2186 */
2187 printf("in6_if2idlen: unknown link type (%d)\n", ifp->if_type);
2188 return (64);
2189 }
2190 }
2191
2192 struct in6_llentry {
2193 struct llentry base;
2194 };
2195
2196 #define IN6_LLTBL_DEFAULT_HSIZE 32
2197 #define IN6_LLTBL_HASH(k, h) \
2198 (((((((k >> 8) ^ k) >> 8) ^ k) >> 8) ^ k) & ((h) - 1))
2199
2200 /*
2201 * Do actual deallocation of @lle.
2202 */
2203 static void
in6_lltable_destroy_lle_unlocked(epoch_context_t ctx)2204 in6_lltable_destroy_lle_unlocked(epoch_context_t ctx)
2205 {
2206 struct llentry *lle;
2207
2208 lle = __containerof(ctx, struct llentry, lle_epoch_ctx);
2209 LLE_LOCK_DESTROY(lle);
2210 LLE_REQ_DESTROY(lle);
2211 free(lle, M_LLTABLE);
2212 }
2213
2214 /*
2215 * Called by LLE_FREE_LOCKED when number of references
2216 * drops to zero.
2217 */
2218 static void
in6_lltable_destroy_lle(struct llentry * lle)2219 in6_lltable_destroy_lle(struct llentry *lle)
2220 {
2221
2222 LLE_WUNLOCK(lle);
2223 NET_EPOCH_CALL(in6_lltable_destroy_lle_unlocked, &lle->lle_epoch_ctx);
2224 }
2225
2226 static struct llentry *
in6_lltable_new(const struct in6_addr * addr6,u_int flags)2227 in6_lltable_new(const struct in6_addr *addr6, u_int flags)
2228 {
2229 struct in6_llentry *lle;
2230
2231 lle = malloc(sizeof(struct in6_llentry), M_LLTABLE, M_NOWAIT | M_ZERO);
2232 if (lle == NULL) /* NB: caller generates msg */
2233 return NULL;
2234
2235 lle->base.r_l3addr.addr6 = *addr6;
2236 lle->base.lle_refcnt = 1;
2237 lle->base.lle_free = in6_lltable_destroy_lle;
2238 LLE_LOCK_INIT(&lle->base);
2239 LLE_REQ_INIT(&lle->base);
2240 callout_init(&lle->base.lle_timer, 1);
2241
2242 return (&lle->base);
2243 }
2244
2245 static int
in6_lltable_match_prefix(const struct sockaddr * saddr,const struct sockaddr * smask,u_int flags,struct llentry * lle)2246 in6_lltable_match_prefix(const struct sockaddr *saddr,
2247 const struct sockaddr *smask, u_int flags, struct llentry *lle)
2248 {
2249 const struct in6_addr *addr, *mask, *lle_addr;
2250
2251 addr = &((const struct sockaddr_in6 *)saddr)->sin6_addr;
2252 mask = &((const struct sockaddr_in6 *)smask)->sin6_addr;
2253 lle_addr = &lle->r_l3addr.addr6;
2254
2255 if (IN6_ARE_MASKED_ADDR_EQUAL(lle_addr, addr, mask) == 0)
2256 return (0);
2257
2258 if (lle->la_flags & LLE_IFADDR) {
2259 /*
2260 * Delete LLE_IFADDR records IFF address & flag matches.
2261 * Note that addr is the interface address within prefix
2262 * being matched.
2263 */
2264 if (IN6_ARE_ADDR_EQUAL(addr, lle_addr) &&
2265 (flags & LLE_STATIC) != 0)
2266 return (1);
2267 return (0);
2268 }
2269
2270 /* flags & LLE_STATIC means deleting both dynamic and static entries */
2271 if ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))
2272 return (1);
2273
2274 return (0);
2275 }
2276
2277 static void
in6_lltable_free_entry(struct lltable * llt,struct llentry * lle)2278 in6_lltable_free_entry(struct lltable *llt, struct llentry *lle)
2279 {
2280
2281 LLE_WLOCK_ASSERT(lle);
2282 KASSERT(llt != NULL, ("lltable is NULL"));
2283
2284 /* Unlink entry from table */
2285 if ((lle->la_flags & LLE_LINKED) != 0) {
2286 LLTABLE_LOCK_ASSERT(llt);
2287 lltable_unlink_entry(llt, lle);
2288 }
2289
2290 llentry_free(lle);
2291 }
2292
2293 static int
in6_lltable_rtcheck(struct ifnet * ifp,u_int flags,const struct sockaddr * l3addr)2294 in6_lltable_rtcheck(struct ifnet *ifp,
2295 u_int flags,
2296 const struct sockaddr *l3addr)
2297 {
2298 const struct sockaddr_in6 *sin6;
2299 struct nhop_object *nh;
2300 struct in6_addr dst;
2301 uint32_t scopeid;
2302 char ip6buf[INET6_ADDRSTRLEN];
2303 int fibnum;
2304
2305 NET_EPOCH_ASSERT();
2306 KASSERT(l3addr->sa_family == AF_INET6,
2307 ("sin_family %d", l3addr->sa_family));
2308
2309 sin6 = (const struct sockaddr_in6 *)l3addr;
2310 in6_splitscope(&sin6->sin6_addr, &dst, &scopeid);
2311 fibnum = V_rt_add_addr_allfibs ? RT_DEFAULT_FIB : ifp->if_fib;
2312 nh = fib6_lookup(fibnum, &dst, scopeid, NHR_NONE, 0);
2313 if (nh && ((nh->nh_flags & NHF_GATEWAY) || nh->nh_ifp != ifp)) {
2314 struct ifaddr *ifa;
2315 /*
2316 * Create an ND6 cache for an IPv6 neighbor
2317 * that is not covered by our own prefix.
2318 */
2319 ifa = ifaof_ifpforaddr(l3addr, ifp);
2320 if (ifa != NULL) {
2321 return 0;
2322 }
2323 log(LOG_INFO, "IPv6 address: \"%s\" is not on the network\n",
2324 ip6_sprintf(ip6buf, &sin6->sin6_addr));
2325 return EINVAL;
2326 }
2327 return 0;
2328 }
2329
2330 static inline uint32_t
in6_lltable_hash_dst(const struct in6_addr * dst,uint32_t hsize)2331 in6_lltable_hash_dst(const struct in6_addr *dst, uint32_t hsize)
2332 {
2333
2334 return (IN6_LLTBL_HASH(dst->s6_addr32[3], hsize));
2335 }
2336
2337 static uint32_t
in6_lltable_hash(const struct llentry * lle,uint32_t hsize)2338 in6_lltable_hash(const struct llentry *lle, uint32_t hsize)
2339 {
2340
2341 return (in6_lltable_hash_dst(&lle->r_l3addr.addr6, hsize));
2342 }
2343
2344 static void
in6_lltable_fill_sa_entry(const struct llentry * lle,struct sockaddr * sa)2345 in6_lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
2346 {
2347 struct sockaddr_in6 *sin6;
2348
2349 sin6 = (struct sockaddr_in6 *)sa;
2350 bzero(sin6, sizeof(*sin6));
2351 sin6->sin6_family = AF_INET6;
2352 sin6->sin6_len = sizeof(*sin6);
2353 sin6->sin6_addr = lle->r_l3addr.addr6;
2354 }
2355
2356 static inline struct llentry *
in6_lltable_find_dst(struct lltable * llt,const struct in6_addr * dst)2357 in6_lltable_find_dst(struct lltable *llt, const struct in6_addr *dst)
2358 {
2359 struct llentry *lle;
2360 struct llentries *lleh;
2361 u_int hashidx;
2362
2363 hashidx = in6_lltable_hash_dst(dst, llt->llt_hsize);
2364 lleh = &llt->lle_head[hashidx];
2365 CK_LIST_FOREACH(lle, lleh, lle_next) {
2366 if (lle->la_flags & LLE_DELETED)
2367 continue;
2368 if (IN6_ARE_ADDR_EQUAL(&lle->r_l3addr.addr6, dst))
2369 break;
2370 }
2371
2372 return (lle);
2373 }
2374
2375 static void
in6_lltable_delete_entry(struct lltable * llt,struct llentry * lle)2376 in6_lltable_delete_entry(struct lltable *llt, struct llentry *lle)
2377 {
2378
2379 lle->la_flags |= LLE_DELETED;
2380
2381 /* Leave the solicited multicast group. */
2382 if ((lle->la_flags & LLE_PUB) != 0)
2383 in6_leave_proxy_ndp_mc(llt->llt_ifp, &lle->r_l3addr.addr6);
2384 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
2385 #ifdef DIAGNOSTIC
2386 log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
2387 #endif
2388 llentry_free(lle);
2389 }
2390
2391 static struct llentry *
in6_lltable_alloc(struct lltable * llt,u_int flags,const struct sockaddr * l3addr)2392 in6_lltable_alloc(struct lltable *llt, u_int flags,
2393 const struct sockaddr *l3addr)
2394 {
2395 const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr;
2396 struct ifnet *ifp = llt->llt_ifp;
2397 struct llentry *lle;
2398 char linkhdr[LLE_MAX_LINKHDR];
2399 size_t linkhdrsize;
2400 int lladdr_off;
2401
2402 KASSERT(l3addr->sa_family == AF_INET6,
2403 ("sin_family %d", l3addr->sa_family));
2404
2405 /*
2406 * A route that covers the given address must have
2407 * been installed 1st because we are doing a resolution,
2408 * verify this.
2409 */
2410 if (!(flags & LLE_IFADDR) &&
2411 in6_lltable_rtcheck(ifp, flags, l3addr) != 0)
2412 return (NULL);
2413
2414 lle = in6_lltable_new(&sin6->sin6_addr, flags);
2415 if (lle == NULL) {
2416 log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
2417 return (NULL);
2418 }
2419 lle->la_flags = flags;
2420 if ((flags & LLE_IFADDR) == LLE_IFADDR) {
2421 linkhdrsize = LLE_MAX_LINKHDR;
2422 if (lltable_calc_llheader(ifp, AF_INET6, IF_LLADDR(ifp),
2423 linkhdr, &linkhdrsize, &lladdr_off) != 0) {
2424 in6_lltable_free_entry(llt, lle);
2425 return (NULL);
2426 }
2427 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
2428 lladdr_off);
2429 lle->la_flags |= LLE_STATIC;
2430 }
2431
2432 if ((lle->la_flags & LLE_STATIC) != 0)
2433 lle->ln_state = ND6_LLINFO_REACHABLE;
2434
2435 return (lle);
2436 }
2437
2438 static struct llentry *
in6_lltable_lookup(struct lltable * llt,u_int flags,const struct sockaddr * l3addr)2439 in6_lltable_lookup(struct lltable *llt, u_int flags,
2440 const struct sockaddr *l3addr)
2441 {
2442 const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr;
2443 int family = flags >> 16;
2444 struct llentry *lle;
2445
2446 LLTABLE_RLOCK_ASSERT(llt);
2447 KASSERT(l3addr->sa_family == AF_INET6,
2448 ("sin_family %d", l3addr->sa_family));
2449 KASSERT((flags & (LLE_UNLOCKED | LLE_EXCLUSIVE)) !=
2450 (LLE_UNLOCKED | LLE_EXCLUSIVE),
2451 ("wrong lle request flags: %#x", flags));
2452
2453 lle = in6_lltable_find_dst(llt, &sin6->sin6_addr);
2454
2455 if (__predict_false(family != AF_INET6))
2456 lle = llentry_lookup_family(lle, family);
2457
2458 if (lle == NULL)
2459 return (NULL);
2460
2461 if (flags & LLE_UNLOCKED)
2462 return (lle);
2463
2464 if (flags & LLE_EXCLUSIVE)
2465 LLE_WLOCK(lle);
2466 else
2467 LLE_RLOCK(lle);
2468
2469 /*
2470 * If the lltable lock is not held, the LLE may have been unlinked while
2471 * we were blocked on the LLE lock. Check for this case.
2472 */
2473 if (__predict_false((lle->la_flags & LLE_LINKED) == 0)) {
2474 if (flags & LLE_EXCLUSIVE)
2475 LLE_WUNLOCK(lle);
2476 else
2477 LLE_RUNLOCK(lle);
2478 return (NULL);
2479 }
2480 return (lle);
2481 }
2482
2483 static int
in6_lltable_dump_entry(struct lltable * llt,struct llentry * lle,struct sysctl_req * wr)2484 in6_lltable_dump_entry(struct lltable *llt, struct llentry *lle,
2485 struct sysctl_req *wr)
2486 {
2487 struct ifnet *ifp = llt->llt_ifp;
2488 /* XXX stack use */
2489 struct {
2490 struct rt_msghdr rtm;
2491 struct sockaddr_in6 sin6;
2492 /*
2493 * ndp.c assumes that sdl is word aligned
2494 */
2495 #ifdef __LP64__
2496 uint32_t pad;
2497 #endif
2498 struct sockaddr_dl sdl;
2499 } ndpc;
2500 struct sockaddr_dl *sdl;
2501 int error;
2502
2503 bzero(&ndpc, sizeof(ndpc));
2504 /* skip deleted entries */
2505 if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
2506 return (0);
2507 /* Skip if jailed and not a valid IP of the prison. */
2508 lltable_fill_sa_entry(lle, (struct sockaddr *)&ndpc.sin6);
2509 if (prison_if(wr->td->td_ucred, (struct sockaddr *)&ndpc.sin6) != 0)
2510 return (0);
2511 /*
2512 * produce a msg made of:
2513 * struct rt_msghdr;
2514 * struct sockaddr_in6 (IPv6)
2515 * struct sockaddr_dl;
2516 */
2517 ndpc.rtm.rtm_msglen = sizeof(ndpc);
2518 ndpc.rtm.rtm_version = RTM_VERSION;
2519 ndpc.rtm.rtm_type = RTM_GET;
2520 ndpc.rtm.rtm_flags = RTF_UP;
2521 ndpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
2522 sa6_recoverscope(&ndpc.sin6);
2523
2524 /* publish */
2525 if (lle->la_flags & LLE_PUB)
2526 ndpc.rtm.rtm_flags |= RTF_ANNOUNCE;
2527
2528 sdl = &ndpc.sdl;
2529 sdl->sdl_family = AF_LINK;
2530 sdl->sdl_len = sizeof(*sdl);
2531 sdl->sdl_index = ifp->if_index;
2532 sdl->sdl_type = ifp->if_type;
2533 if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
2534 sdl->sdl_alen = ifp->if_addrlen;
2535 bcopy(lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
2536 } else {
2537 sdl->sdl_alen = 0;
2538 bzero(LLADDR(sdl), ifp->if_addrlen);
2539 }
2540 if (lle->la_expire != 0)
2541 ndpc.rtm.rtm_rmx.rmx_expire = lle->la_expire +
2542 lle->lle_remtime / hz + time_second - time_uptime;
2543 ndpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
2544 if (lle->la_flags & LLE_STATIC)
2545 ndpc.rtm.rtm_flags |= RTF_STATIC;
2546 if (lle->la_flags & LLE_IFADDR)
2547 ndpc.rtm.rtm_flags |= RTF_PINNED;
2548 if (lle->ln_router != 0)
2549 ndpc.rtm.rtm_flags |= RTF_GATEWAY;
2550 ndpc.rtm.rtm_rmx.rmx_pksent = lle->la_asked;
2551 /* Store state in rmx_weight value */
2552 ndpc.rtm.rtm_rmx.rmx_state = lle->ln_state;
2553 ndpc.rtm.rtm_index = ifp->if_index;
2554 error = SYSCTL_OUT(wr, &ndpc, sizeof(ndpc));
2555
2556 return (error);
2557 }
2558
2559 static void
in6_lltable_post_resolved(struct lltable * llt,struct llentry * lle)2560 in6_lltable_post_resolved(struct lltable *llt, struct llentry *lle)
2561 {
2562 /* Join the solicited multicast group for dst. */
2563 if ((lle->la_flags & LLE_PUB) == LLE_PUB)
2564 in6_join_proxy_ndp_mc(llt->llt_ifp, &lle->r_l3addr.addr6);
2565 }
2566
2567 static struct lltable *
in6_lltattach(struct ifnet * ifp)2568 in6_lltattach(struct ifnet *ifp)
2569 {
2570 struct lltable *llt;
2571
2572 llt = lltable_allocate_htbl(IN6_LLTBL_DEFAULT_HSIZE);
2573 llt->llt_af = AF_INET6;
2574 llt->llt_ifp = ifp;
2575
2576 llt->llt_lookup = in6_lltable_lookup;
2577 llt->llt_alloc_entry = in6_lltable_alloc;
2578 llt->llt_delete_entry = in6_lltable_delete_entry;
2579 llt->llt_dump_entry = in6_lltable_dump_entry;
2580 llt->llt_hash = in6_lltable_hash;
2581 llt->llt_fill_sa_entry = in6_lltable_fill_sa_entry;
2582 llt->llt_free_entry = in6_lltable_free_entry;
2583 llt->llt_match_prefix = in6_lltable_match_prefix;
2584 llt->llt_mark_used = llentry_mark_used;
2585 llt->llt_post_resolved = in6_lltable_post_resolved;
2586 lltable_link(llt);
2587
2588 return (llt);
2589 }
2590
2591 struct lltable *
in6_lltable_get(struct ifnet * ifp)2592 in6_lltable_get(struct ifnet *ifp)
2593 {
2594 if (ifp->if_inet6 == NULL)
2595 return (NULL);
2596
2597 return (ifp->if_inet6->lltable);
2598 }
2599
2600 void
in6_ifarrival(void * arg __unused,struct ifnet * ifp)2601 in6_ifarrival(void *arg __unused, struct ifnet *ifp)
2602 {
2603 struct in6_ifextra *ext;
2604
2605 /* There are not IPv6-capable interfaces. */
2606 switch (ifp->if_type) {
2607 case IFT_PFSYNC:
2608 ifp->if_inet6 = NULL;
2609 return;
2610 }
2611 ext = ifp->if_inet6 = malloc(sizeof(*ext), M_IFADDR, M_WAITOK | M_ZERO);
2612 COUNTER_ARRAY_ALLOC(ext->in6_ifstat,
2613 sizeof(struct in6_ifstat) / sizeof(uint64_t), M_WAITOK);
2614 COUNTER_ARRAY_ALLOC(ext->icmp6_ifstat,
2615 sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_WAITOK);
2616 nd6_ifattach(ifp);
2617 mld_domifattach(ifp);
2618 scope6_ifattach(ifp);
2619
2620 ext->lltable = in6_lltattach(ifp);
2621 }
2622 EVENTHANDLER_DEFINE(ifnet_arrival_event, in6_ifarrival, NULL,
2623 EVENTHANDLER_PRI_ANY);
2624
2625 uint32_t
in6_ifmtu(const struct ifnet * ifp)2626 in6_ifmtu(const struct ifnet *ifp)
2627 {
2628 const uint32_t
2629 linkmtu = ifp->if_inet6->nd_linkmtu,
2630 maxmtu = ifp->if_inet6->nd_maxmtu,
2631 ifmtu = ifp->if_mtu;
2632
2633 if (linkmtu > 0 && linkmtu < ifmtu)
2634 return (linkmtu);
2635 if (maxmtu > 0 && maxmtu < ifmtu)
2636 return (maxmtu);
2637 return (ifmtu);
2638 }
2639
2640 /*
2641 * Convert sockaddr_in6 to sockaddr_in. Original sockaddr_in6 must be
2642 * v4 mapped addr or v4 compat addr
2643 */
2644 void
in6_sin6_2_sin(struct sockaddr_in * sin,const struct sockaddr_in6 * sin6)2645 in6_sin6_2_sin(struct sockaddr_in *sin, const struct sockaddr_in6 *sin6)
2646 {
2647
2648 bzero(sin, sizeof(*sin));
2649 sin->sin_len = sizeof(struct sockaddr_in);
2650 sin->sin_family = AF_INET;
2651 sin->sin_port = sin6->sin6_port;
2652 sin->sin_addr.s_addr = sin6->sin6_addr.s6_addr32[3];
2653 }
2654
2655 /* Convert sockaddr_in to sockaddr_in6 in v4 mapped addr format. */
2656 void
in6_sin_2_v4mapsin6(const struct sockaddr_in * sin,struct sockaddr_in6 * sin6)2657 in6_sin_2_v4mapsin6(const struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
2658 {
2659 bzero(sin6, sizeof(*sin6));
2660 sin6->sin6_len = sizeof(struct sockaddr_in6);
2661 sin6->sin6_family = AF_INET6;
2662 sin6->sin6_port = sin->sin_port;
2663 sin6->sin6_addr.s6_addr32[0] = 0;
2664 sin6->sin6_addr.s6_addr32[1] = 0;
2665 sin6->sin6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
2666 sin6->sin6_addr.s6_addr32[3] = sin->sin_addr.s_addr;
2667 }
2668
2669 /* Convert sockaddr_in6 into sockaddr_in. */
2670 void
in6_sin6_2_sin_in_sock(struct sockaddr * nam)2671 in6_sin6_2_sin_in_sock(struct sockaddr *nam)
2672 {
2673 struct sockaddr_in *sin_p;
2674 struct sockaddr_in6 sin6;
2675
2676 /*
2677 * Save original sockaddr_in6 addr and convert it
2678 * to sockaddr_in.
2679 */
2680 sin6 = *(struct sockaddr_in6 *)nam;
2681 sin_p = (struct sockaddr_in *)nam;
2682 in6_sin6_2_sin(sin_p, &sin6);
2683 }
2684
2685 /*
2686 * Join/leave the solicited multicast groups for proxy NDP entries.
2687 */
2688 static void
in6_join_proxy_ndp_mc(struct ifnet * ifp,const struct in6_addr * dst)2689 in6_join_proxy_ndp_mc(struct ifnet *ifp, const struct in6_addr *dst)
2690 {
2691 struct in6_multi *inm;
2692 struct in6_addr mltaddr;
2693 char ip6buf[INET6_ADDRSTRLEN];
2694 int error;
2695
2696 if (in6_solicited_node_maddr(&mltaddr, ifp, dst) != 0)
2697 return; /* error logged in in6_solicited_node_maddr. */
2698
2699 error = in6_joingroup(ifp, &mltaddr, NULL, &inm, 0);
2700 if (error != 0) {
2701 nd6log((LOG_WARNING,
2702 "%s: in6_joingroup failed for %s on %s (errno=%d)\n",
2703 __func__, ip6_sprintf(ip6buf, &mltaddr), if_name(ifp),
2704 error));
2705 }
2706 }
2707
2708 static void
in6_leave_proxy_ndp_mc(struct ifnet * ifp,const struct in6_addr * dst)2709 in6_leave_proxy_ndp_mc(struct ifnet *ifp, const struct in6_addr *dst)
2710 {
2711 struct epoch_tracker et;
2712 struct in6_multi *inm;
2713 struct in6_addr mltaddr;
2714 char ip6buf[INET6_ADDRSTRLEN];
2715
2716 if (in6_solicited_node_maddr(&mltaddr, ifp, dst) != 0)
2717 return; /* error logged in in6_solicited_node_maddr. */
2718
2719 NET_EPOCH_ENTER(et);
2720 inm = in6m_lookup(ifp, &mltaddr);
2721 NET_EPOCH_EXIT(et);
2722 if (inm != NULL)
2723 in6_leavegroup(inm, NULL);
2724 else
2725 nd6log((LOG_WARNING, "%s: in6m_lookup failed for %s on %s\n",
2726 __func__, ip6_sprintf(ip6buf, &mltaddr), if_name(ifp)));
2727 }
2728
2729 static bool
in6_lle_match_pub(struct lltable * llt,struct llentry * lle,void * farg)2730 in6_lle_match_pub(struct lltable *llt, struct llentry *lle, void *farg)
2731 {
2732 return ((lle->la_flags & LLE_PUB) != 0);
2733 }
2734
2735 void
in6_purge_proxy_ndp(struct ifnet * ifp)2736 in6_purge_proxy_ndp(struct ifnet *ifp)
2737 {
2738 struct lltable *llt;
2739 bool need_purge;
2740
2741 if (ifp->if_inet6 == NULL)
2742 return;
2743
2744 llt = LLTABLE6(ifp);
2745 LLTABLE_LOCK(llt);
2746 need_purge = ((llt->llt_flags & LLT_ADDEDPROXY) != 0);
2747 LLTABLE_UNLOCK(llt);
2748
2749 /*
2750 * Ever added proxy ndp entries, leave solicited node multicast
2751 * before deleting the llentry.
2752 */
2753 if (need_purge)
2754 lltable_delete_conditional(llt, in6_lle_match_pub, NULL);
2755 }
2756