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