in6_pcb.c revision 1.20 1 /* $NetBSD: in6_pcb.c,v 1.20 2000/03/02 06:42:52 itojun Exp $ */
2 /* $KAME: in6_pcb.c,v 1.34 2000/02/22 14:04:18 itojun Exp $ */
3
4 /*
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32
33 /*
34 * Copyright (c) 1982, 1986, 1991, 1993
35 * The Regents of the University of California. All rights reserved.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
40 * 1. Redistributions of source code must retain the above copyright
41 * notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 * notice, this list of conditions and the following disclaimer in the
44 * documentation and/or other materials provided with the distribution.
45 * 3. All advertising materials mentioning features or use of this software
46 * must display the following acknowledgement:
47 * This product includes software developed by the University of
48 * California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 * may be used to endorse or promote products derived from this software
51 * without specific prior written permission.
52 *
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63 * SUCH DAMAGE.
64 *
65 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
66 */
67
68 #include "opt_ipsec.h"
69
70 #include <sys/param.h>
71 #include <sys/systm.h>
72 #include <sys/malloc.h>
73 #include <sys/mbuf.h>
74 #include <sys/protosw.h>
75 #include <sys/socket.h>
76 #include <sys/socketvar.h>
77 #include <sys/ioctl.h>
78 #include <sys/errno.h>
79 #include <sys/time.h>
80 #include <sys/proc.h>
81
82 #include <net/if.h>
83 #include <net/route.h>
84
85 #include <netinet/in.h>
86 #include <netinet/in_var.h>
87 #include <netinet/in_systm.h>
88 #include <netinet/ip.h>
89 #include <netinet/in_pcb.h>
90 #include <netinet/ip6.h>
91 #include <netinet6/ip6_var.h>
92 #include <netinet6/in6_pcb.h>
93 #include <netinet6/nd6.h>
94
95 #include "loop.h"
96 extern struct ifnet loif[NLOOP];
97 #include "faith.h"
98
99 #ifdef IPSEC
100 #include <netinet6/ipsec.h>
101 #include <netkey/key.h>
102 #include <netkey/key_debug.h>
103 #endif /* IPSEC */
104
105 struct in6_addr zeroin6_addr;
106
107 int
108 in6_pcballoc(so, head)
109 struct socket *so;
110 struct in6pcb *head;
111 {
112 struct in6pcb *in6p;
113
114 MALLOC(in6p, struct in6pcb *, sizeof(*in6p), M_PCB, M_NOWAIT);
115 if (in6p == NULL)
116 return(ENOBUFS);
117 bzero((caddr_t)in6p, sizeof(*in6p));
118 in6p->in6p_head = head;
119 in6p->in6p_socket = so;
120 in6p->in6p_hops = -1; /* use kernel default */
121 in6p->in6p_icmp6filt = NULL;
122 in6p->in6p_next = head->in6p_next;
123 head->in6p_next = in6p;
124 in6p->in6p_prev = head;
125 in6p->in6p_next->in6p_prev = in6p;
126 #ifndef INET6_BINDV6ONLY
127 if (ip6_bindv6only)
128 in6p->in6p_flags |= IN6P_BINDV6ONLY;
129 #else
130 in6p->in6p_flags |= IN6P_BINDV6ONLY; /*just for safety*/
131 #endif
132 so->so_pcb = (caddr_t)in6p;
133 return(0);
134 }
135
136 int
137 in6_pcbbind(in6p, nam)
138 register struct in6pcb *in6p;
139 struct mbuf *nam;
140 {
141 struct socket *so = in6p->in6p_socket;
142 struct in6pcb *head = in6p->in6p_head;
143 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
144 struct proc *p = curproc; /* XXX */
145 u_int16_t lport = 0;
146 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
147 int error;
148
149 if (in6p->in6p_lport || !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
150 return(EINVAL);
151 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0 &&
152 ((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0 ||
153 (so->so_options & SO_ACCEPTCONN) == 0))
154 wild = IN6PLOOKUP_WILDCARD;
155 if (nam) {
156 sin6 = mtod(nam, struct sockaddr_in6 *);
157 if (nam->m_len != sizeof(*sin6))
158 return(EINVAL);
159 /*
160 * We should check the family, but old programs
161 * incorrectly fail to intialize it.
162 */
163 if (sin6->sin6_family != AF_INET6)
164 return(EAFNOSUPPORT);
165
166 /*
167 * If the scope of the destination is link-local, embed the
168 * interface index in the address.
169 */
170 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) {
171 /* XXX boundary check is assumed to be already done. */
172 /* XXX sin6_scope_id is weaker than advanced-api. */
173 struct in6_pktinfo *pi;
174 if (in6p->in6p_outputopts &&
175 (pi = in6p->in6p_outputopts->ip6po_pktinfo) &&
176 pi->ipi6_ifindex) {
177 sin6->sin6_addr.s6_addr16[1]
178 = htons(pi->ipi6_ifindex);
179 } else if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)
180 && in6p->in6p_moptions
181 && in6p->in6p_moptions->im6o_multicast_ifp) {
182 sin6->sin6_addr.s6_addr16[1] =
183 htons(in6p->in6p_moptions->im6o_multicast_ifp->if_index);
184 } else if (sin6->sin6_scope_id) {
185 /* boundary check */
186 if (sin6->sin6_scope_id < 0
187 || if_index < sin6->sin6_scope_id) {
188 return ENXIO; /* XXX EINVAL? */
189 }
190 sin6->sin6_addr.s6_addr16[1]
191 = htons(sin6->sin6_scope_id & 0xffff);/*XXX*/
192 /* this must be cleared for ifa_ifwithaddr() */
193 sin6->sin6_scope_id = 0;
194 }
195 }
196
197 lport = sin6->sin6_port;
198 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
199 /*
200 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
201 * allow compepte duplication of binding if
202 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
203 * and a multicast address is bound on both
204 * new and duplicated sockets.
205 */
206 if (so->so_options & SO_REUSEADDR)
207 reuseport = SO_REUSEADDR|SO_REUSEPORT;
208 } else if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
209 struct sockaddr_in sin;
210
211 bzero(&sin, sizeof(sin));
212 sin.sin_len = sizeof(sin);
213 sin.sin_family = AF_INET;
214 bcopy(&sin6->sin6_addr.s6_addr32[3], &sin.sin_addr,
215 sizeof(sin.sin_addr));
216 if (ifa_ifwithaddr((struct sockaddr *)&sin) == 0)
217 return EADDRNOTAVAIL;
218 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
219 struct ifaddr *ia = NULL;
220
221 sin6->sin6_port = 0; /* yech... */
222 #if defined(NFAITH) && NFAITH > 0
223 if ((in6p->in6p_flags & IN6P_FAITH) == 0
224 && (ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0)
225 #else
226 if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0)
227 #endif
228 return(EADDRNOTAVAIL);
229
230 /*
231 * XXX: bind to an anycast address might accidentally
232 * cause sending a packet with anycast source address.
233 */
234 if (ia &&
235 ((struct in6_ifaddr *)ia)->ia6_flags &
236 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
237 IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
238 return(EADDRNOTAVAIL);
239 }
240 }
241 if (lport) {
242 #ifndef IPNOPRIVPORTS
243 /* GROSS */
244 if (ntohs(lport) < IPV6PORT_RESERVED &&
245 (p == 0 ||
246 (error = suser(p->p_ucred, &p->p_acflag))))
247 return(EACCES);
248 #endif
249
250 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
251 /* should check this but we can't ... */
252 #if 0
253 struct inpcb *t;
254
255 t = in_pcblookup_bind(&tcbtable,
256 (struct in_addr *)&sin6->sin6_addr.s6_addr32[3],
257 lport);
258 if (t && (reuseport & t->inp_socket->so_options) == 0)
259 return EADDRINUSE;
260 #endif
261 } else {
262 struct in6pcb *t;
263
264 t = in6_pcblookup(head, &zeroin6_addr, 0,
265 &sin6->sin6_addr, lport, wild);
266 if (t && (reuseport & t->in6p_socket->so_options) == 0)
267 return(EADDRINUSE);
268 }
269 }
270 in6p->in6p_laddr = sin6->sin6_addr;
271 }
272
273 if (lport == 0) {
274 int e;
275 if ((e = in6_pcbsetport(&in6p->in6p_laddr, in6p)) != 0)
276 return(e);
277 }
278 else
279 in6p->in6p_lport = lport;
280
281 in6p->in6p_flowinfo = sin6 ? sin6->sin6_flowinfo : 0; /*XXX*/
282 return(0);
283 }
284
285 /*
286 * Find an empty port and set it to the specified PCB.
287 */
288 int
289 in6_pcbsetport(laddr, in6p)
290 struct in6_addr *laddr;
291 struct in6pcb *in6p;
292 {
293 struct socket *so = in6p->in6p_socket;
294 struct in6pcb *head = in6p->in6p_head;
295 u_int16_t last_port, lport = 0;
296 int wild = 0;
297 void *t;
298 u_int16_t min, max;
299 #ifndef IPNOPRIVPORTS
300 struct proc *p = curproc; /*XXX*/
301 #endif
302
303 /* XXX: this is redundant when called from in6_pcbbind */
304 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0 &&
305 ((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0 ||
306 (so->so_options & SO_ACCEPTCONN) == 0))
307 wild = IN6PLOOKUP_WILDCARD;
308
309 if (in6p->in6p_flags & IN6P_LOWPORT) {
310 #ifndef IPNOPRIVPORTS
311 if (p == 0 || (suser(p->p_ucred, &p->p_acflag) != 0))
312 return (EACCES);
313 #endif
314 min = IPV6PORT_RESERVEDMIN;
315 max = IPV6PORT_RESERVEDMAX;
316 } else {
317 min = IPV6PORT_ANONMIN;
318 max = IPV6PORT_ANONMAX;
319 }
320
321 /* value out of range */
322 if (head->in6p_lport < min)
323 head->in6p_lport = min;
324 else if (head->in6p_lport > max)
325 head->in6p_lport = min;
326 last_port = head->in6p_lport;
327 goto startover; /*to randomize*/
328 for (;;) {
329 lport = htons(head->in6p_lport);
330 if (IN6_IS_ADDR_V4MAPPED(laddr)) {
331 #if 0
332 t = in_pcblookup_bind(&tcbtable,
333 (struct in_addr *)&in6p->in6p_laddr.s6_addr32[3],
334 lport);
335 #else
336 t = NULL;
337 #endif
338 } else {
339 t = in6_pcblookup(head, &zeroin6_addr, 0, laddr,
340 lport, wild);
341 }
342 if (t == 0)
343 break;
344 startover:
345 if (head->in6p_lport >= max)
346 head->in6p_lport = min;
347 else
348 head->in6p_lport++;
349 if (head->in6p_lport == last_port)
350 return (EADDRINUSE);
351 }
352
353 in6p->in6p_lport = lport;
354 return(0); /* success */
355 }
356
357 /*
358 * Connect from a socket to a specified address.
359 * Both address and port must be specified in argument sin6.
360 * If don't have a local address for this socket yet,
361 * then pick one.
362 */
363 int
364 in6_pcbconnect(in6p, nam)
365 struct in6pcb *in6p;
366 struct mbuf *nam;
367 {
368 struct in6_addr *in6a = NULL;
369 struct sockaddr_in6 *sin6 = mtod(nam, struct sockaddr_in6 *);
370 struct in6_pktinfo *pi;
371 struct ifnet *ifp = NULL; /* outgoing interface */
372 int error = 0;
373 struct in6_addr mapped;
374
375 (void)&in6a; /* XXX fool gcc */
376
377 if (nam->m_len != sizeof(*sin6))
378 return(EINVAL);
379 if (sin6->sin6_family != AF_INET6)
380 return(EAFNOSUPPORT);
381 if (sin6->sin6_port == 0)
382 return(EADDRNOTAVAIL);
383
384 /* sanity check for mapped address case */
385 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
386 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
387 in6p->in6p_laddr.s6_addr16[5] = htons(0xffff);
388 if (!IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr))
389 return EINVAL;
390 } else {
391 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr))
392 return EINVAL;
393 }
394
395 /*
396 * If the scope of the destination is link-local, embed the interface
397 * index in the address.
398 */
399 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) {
400 /* XXX boundary check is assumed to be already done. */
401 /* XXX sin6_scope_id is weaker than advanced-api. */
402 if (in6p->in6p_outputopts &&
403 (pi = in6p->in6p_outputopts->ip6po_pktinfo) &&
404 pi->ipi6_ifindex) {
405 sin6->sin6_addr.s6_addr16[1] = htons(pi->ipi6_ifindex);
406 ifp = ifindex2ifnet[pi->ipi6_ifindex];
407 }
408 else if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) &&
409 in6p->in6p_moptions &&
410 in6p->in6p_moptions->im6o_multicast_ifp) {
411 sin6->sin6_addr.s6_addr16[1] =
412 htons(in6p->in6p_moptions->im6o_multicast_ifp->if_index);
413 ifp = ifindex2ifnet[in6p->in6p_moptions->im6o_multicast_ifp->if_index];
414 } else if (sin6->sin6_scope_id) {
415 /* boundary check */
416 if (sin6->sin6_scope_id < 0
417 || if_index < sin6->sin6_scope_id) {
418 return ENXIO; /* XXX EINVAL? */
419 }
420 sin6->sin6_addr.s6_addr16[1]
421 = htons(sin6->sin6_scope_id & 0xffff);/*XXX*/
422 ifp = ifindex2ifnet[sin6->sin6_scope_id];
423 }
424 }
425
426 /* Source address selection. */
427 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)
428 && in6p->in6p_laddr.s6_addr32[3] == 0) {
429 struct sockaddr_in sin, *sinp;
430
431 bzero(&sin, sizeof(sin));
432 sin.sin_len = sizeof(sin);
433 sin.sin_family = AF_INET;
434 bcopy(&sin6->sin6_addr.s6_addr32[3], &sin.sin_addr,
435 sizeof(sin.sin_addr));
436 sinp = in_selectsrc(&sin, (struct route *)&in6p->in6p_route,
437 in6p->in6p_socket->so_options, NULL, &error);
438 if (sinp == 0) {
439 if (error == 0)
440 error = EADDRNOTAVAIL;
441 return(error);
442 }
443 bzero(&mapped, sizeof(mapped));
444 mapped.s6_addr16[5] = htons(0xffff);
445 bcopy(&sinp->sin_addr, &mapped.s6_addr32[3], sizeof(sinp->sin_addr));
446 in6a = &mapped;
447 } else {
448 /*
449 * XXX: in6_selectsrc might replace the bound local address
450 * with the address specified by setsockopt(IPV6_PKTINFO).
451 * Is it the intended behavior?
452 */
453 in6a = in6_selectsrc(sin6, in6p->in6p_outputopts,
454 in6p->in6p_moptions,
455 &in6p->in6p_route,
456 &in6p->in6p_laddr, &error);
457 if (in6a == 0) {
458 if (error == 0)
459 error = EADDRNOTAVAIL;
460 return(error);
461 }
462 }
463 if (in6p->in6p_route.ro_rt)
464 ifp = in6p->in6p_route.ro_rt->rt_ifp;
465
466 in6p->in6p_ip6.ip6_hlim = (u_int8_t)in6_selecthlim(in6p, ifp);
467
468 if (in6_pcblookup(in6p->in6p_head,
469 &sin6->sin6_addr,
470 sin6->sin6_port,
471 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) ?
472 in6a : &in6p->in6p_laddr,
473 in6p->in6p_lport,
474 0))
475 return(EADDRINUSE);
476 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)
477 || (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)
478 && in6p->in6p_laddr.s6_addr32[3] == 0)) {
479 if (in6p->in6p_lport == 0)
480 (void)in6_pcbbind(in6p, (struct mbuf *)0);
481 in6p->in6p_laddr = *in6a;
482 }
483 in6p->in6p_faddr = sin6->sin6_addr;
484 in6p->in6p_fport = sin6->sin6_port;
485 /*
486 * xxx kazu flowlabel is necessary for connect?
487 * but if this line is missing, the garbage value remains.
488 */
489 in6p->in6p_flowinfo = sin6->sin6_flowinfo;
490 return(0);
491 }
492
493 /*
494 * Return an IPv6 address, which is the most appropriate for given
495 * destination and user specified options.
496 * If necessary, this function lookups the routing table and return
497 * an entry to the caller for later use.
498 */
499 struct in6_addr *
500 in6_selectsrc(dstsock, opts, mopts, ro, laddr, errorp)
501 struct sockaddr_in6 *dstsock;
502 struct ip6_pktopts *opts;
503 struct ip6_moptions *mopts;
504 struct route_in6 *ro;
505 struct in6_addr *laddr;
506 int *errorp;
507 {
508 struct in6_addr *dst;
509 struct in6_ifaddr *ia6 = 0;
510 struct in6_pktinfo *pi = NULL;
511
512 dst = &dstsock->sin6_addr;
513 *errorp = 0;
514
515 /*
516 * If the source address is explicitly specified by the caller,
517 * use it.
518 */
519 if (opts && (pi = opts->ip6po_pktinfo) &&
520 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr))
521 return(&pi->ipi6_addr);
522
523 /*
524 * If the source address is not specified but the socket(if any)
525 * is already bound, use the bound address.
526 */
527 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr))
528 return(laddr);
529
530 /*
531 * If the caller doesn't specify the source address but
532 * the outgoing interface, use an address associated with
533 * the interface.
534 */
535 if (pi && pi->ipi6_ifindex) {
536 /* XXX boundary check is assumed to be already done. */
537 ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex],
538 dst);
539 if (ia6 == 0) {
540 *errorp = EADDRNOTAVAIL;
541 return(0);
542 }
543 return(&satosin6(&ia6->ia_addr)->sin6_addr);
544 }
545
546 /*
547 * If the destination address is a link-local unicast address or
548 * a multicast address, and if the outgoing interface is specified
549 * by the sin6_scope_id filed, use an address associated with the
550 * interface.
551 * XXX: We're now trying to define more specific semantics of
552 * sin6_scope_id field, so this part will be rewritten in
553 * the near future.
554 */
555 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) &&
556 dstsock->sin6_scope_id) {
557 /*
558 * I'm not sure if boundary check for scope_id is done
559 * somewhere...
560 */
561 if (dstsock->sin6_scope_id < 0 ||
562 if_index < dstsock->sin6_scope_id) {
563 *errorp = ENXIO; /* XXX: better error? */
564 return(0);
565 }
566 ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id],
567 dst);
568 if (ia6 == 0) {
569 *errorp = EADDRNOTAVAIL;
570 return(0);
571 }
572 return(&satosin6(&ia6->ia_addr)->sin6_addr);
573 }
574
575 /*
576 * If the destination address is a multicast address and
577 * the outgoing interface for the address is specified
578 * by the caller, use an address associated with the interface.
579 * There is a sanity check here; if the destination has node-local
580 * scope, the outgoing interfacde should be a loopback address.
581 * Even if the outgoing interface is not specified, we also
582 * choose a loopback interface as the outgoing interface.
583 */
584 if (IN6_IS_ADDR_MULTICAST(dst)) {
585 struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL;
586
587 if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) {
588 ifp = &loif[0];
589 }
590
591 if (ifp) {
592 ia6 = in6_ifawithscope(ifp, dst);
593 if (ia6 == 0) {
594 *errorp = EADDRNOTAVAIL;
595 return(0);
596 }
597 return(&satosin6(&ia6->ia_addr)->sin6_addr);
598 }
599 }
600
601 /*
602 * If the next hop address for the packet is specified
603 * by caller, use an address associated with the route
604 * to the next hop.
605 */
606 {
607 struct sockaddr_in6 *sin6_next;
608 struct rtentry *rt;
609
610 if (opts && opts->ip6po_nexthop) {
611 sin6_next = satosin6(opts->ip6po_nexthop);
612 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL);
613 if (rt) {
614 ia6 = in6_ifawithscope(rt->rt_ifp, dst);
615 if (ia6 == 0)
616 ia6 = ifatoia6(rt->rt_ifa);
617 }
618 if (ia6 == 0) {
619 *errorp = EADDRNOTAVAIL;
620 return(0);
621 }
622 return(&satosin6(&ia6->ia_addr)->sin6_addr);
623 }
624 }
625
626 /*
627 * If route is known or can be allocated now,
628 * our src addr is taken from the i/f, else punt.
629 */
630 if (ro) {
631 if (ro->ro_rt &&
632 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, dst)) {
633 RTFREE(ro->ro_rt);
634 ro->ro_rt = (struct rtentry *)0;
635 }
636 if (ro->ro_rt == (struct rtentry *)0 ||
637 ro->ro_rt->rt_ifp == (struct ifnet *)0) {
638 /* No route yet, so try to acquire one */
639 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
640 ro->ro_dst.sin6_family = AF_INET6;
641 ro->ro_dst.sin6_len = sizeof(struct sockaddr_in6);
642 ro->ro_dst.sin6_addr = *dst;
643 if (IN6_IS_ADDR_MULTICAST(dst)) {
644 ro->ro_rt = rtalloc1(&((struct route *)ro)
645 ->ro_dst, 0);
646 } else {
647 rtalloc((struct route *)ro);
648 }
649
650 }
651
652 /*
653 * in_pcbconnect() checks out IFF_LOOPBACK to skip using
654 * the address. But we don't know why it does so.
655 * It is necessary to ensure the scope even for lo0
656 * so doesn't check out IFF_LOOPBACK.
657 */
658
659 if (ro->ro_rt) {
660 ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst);
661 if (ia6 == 0) /* xxx scope error ?*/
662 ia6 = ifatoia6(ro->ro_rt->rt_ifa);
663 }
664 #if 0
665 /*
666 * xxx The followings are necessary? (kazu)
667 * I don't think so.
668 * It's for SO_DONTROUTE option in IPv4.(jinmei)
669 */
670 if (ia6 == 0) {
671 struct sockaddr_in6 sin6 = {sizeof(sin6), AF_INET6, 0};
672
673 sin6->sin6_addr = *dst;
674
675 ia6 = ifatoia6(ifa_ifwithdstaddr(sin6tosa(&sin6)));
676 if (ia6 == 0)
677 ia6 = ifatoia6(ifa_ifwithnet(sin6tosa(&sin6)));
678 if (ia6 == 0)
679 return(0);
680 return(&satosin6(&ia6->ia_addr)->sin6_addr);
681 }
682 #endif /* 0 */
683 if (ia6 == 0) {
684 *errorp = EHOSTUNREACH; /* no route */
685 return(0);
686 }
687 return(&satosin6(&ia6->ia_addr)->sin6_addr);
688 }
689
690 *errorp = EADDRNOTAVAIL;
691 return(0);
692 }
693
694 /*
695 * Default hop limit selection. The precedence is as follows:
696 * 1. Hoplimit valued specified via ioctl.
697 * 2. (If the outgoing interface is detected) the current
698 * hop limit of the interface specified by router advertisement.
699 * 3. The system default hoplimit.
700 */
701 int
702 in6_selecthlim(in6p, ifp)
703 struct in6pcb *in6p;
704 struct ifnet *ifp;
705 {
706 if (in6p && in6p->in6p_hops >= 0)
707 return(in6p->in6p_hops);
708 else if (ifp)
709 return(nd_ifinfo[ifp->if_index].chlim);
710 else
711 return(ip6_defhlim);
712 }
713
714 void
715 in6_pcbdisconnect(in6p)
716 struct in6pcb *in6p;
717 {
718 bzero((caddr_t)&in6p->in6p_faddr, sizeof(in6p->in6p_faddr));
719 in6p->in6p_fport = 0;
720 if (in6p->in6p_socket->so_state & SS_NOFDREF)
721 in6_pcbdetach(in6p);
722 }
723
724 void
725 in6_pcbdetach(in6p)
726 struct in6pcb *in6p;
727 {
728 struct socket *so = in6p->in6p_socket;
729
730 #ifdef IPSEC
731 ipsec6_delete_pcbpolicy(in6p);
732 #endif /* IPSEC */
733 sotoin6pcb(so) = 0;
734 sofree(so);
735 if (in6p->in6p_options)
736 m_freem(in6p->in6p_options);
737 if (in6p->in6p_outputopts) {
738 if (in6p->in6p_outputopts->ip6po_rthdr &&
739 in6p->in6p_outputopts->ip6po_route.ro_rt)
740 RTFREE(in6p->in6p_outputopts->ip6po_route.ro_rt);
741 if (in6p->in6p_outputopts->ip6po_m)
742 (void)m_free(in6p->in6p_outputopts->ip6po_m);
743 free(in6p->in6p_outputopts, M_IP6OPT);
744 }
745 if (in6p->in6p_route.ro_rt)
746 rtfree(in6p->in6p_route.ro_rt);
747 ip6_freemoptions(in6p->in6p_moptions);
748 in6p->in6p_next->in6p_prev = in6p->in6p_prev;
749 in6p->in6p_prev->in6p_next = in6p->in6p_next;
750 in6p->in6p_prev = NULL;
751 FREE(in6p, M_PCB);
752 }
753
754 void
755 in6_setsockaddr(in6p, nam)
756 struct in6pcb *in6p;
757 struct mbuf *nam;
758 {
759 struct sockaddr_in6 *sin6;
760
761 nam->m_len = sizeof(*sin6);
762 sin6 = mtod(nam, struct sockaddr_in6 *);
763 bzero((caddr_t)sin6, sizeof(*sin6));
764 sin6->sin6_family = AF_INET6;
765 sin6->sin6_len = sizeof(struct sockaddr_in6);
766 sin6->sin6_port = in6p->in6p_lport;
767 sin6->sin6_addr = in6p->in6p_laddr;
768 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
769 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
770 else
771 sin6->sin6_scope_id = 0; /*XXX*/
772 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
773 sin6->sin6_addr.s6_addr16[1] = 0;
774 }
775
776 void
777 in6_setpeeraddr(in6p, nam)
778 struct in6pcb *in6p;
779 struct mbuf *nam;
780 {
781 struct sockaddr_in6 *sin6;
782
783 nam->m_len = sizeof(*sin6);
784 sin6 = mtod(nam, struct sockaddr_in6 *);
785 bzero((caddr_t)sin6, sizeof(*sin6));
786 sin6->sin6_family = AF_INET6;
787 sin6->sin6_len = sizeof(struct sockaddr_in6);
788 sin6->sin6_port = in6p->in6p_fport;
789 sin6->sin6_addr = in6p->in6p_faddr;
790 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
791 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
792 else
793 sin6->sin6_scope_id = 0; /*XXX*/
794 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
795 sin6->sin6_addr.s6_addr16[1] = 0;
796 }
797
798 /*
799 * Pass some notification to all connections of a protocol
800 * associated with address dst. The local address and/or port numbers
801 * may be specified to limit the search. The "usual action" will be
802 * taken, depending on the ctlinput cmd. The caller must filter any
803 * cmds that are uninteresting (e.g., no error in the map).
804 * Call the protocol specific routine (if any) to report
805 * any errors for each matching socket.
806 *
807 * Must be called at splsoftnet.
808 */
809 int
810 in6_pcbnotify(head, dst, fport_arg, laddr6, lport_arg, cmd, notify)
811 struct in6pcb *head;
812 struct sockaddr *dst;
813 u_int fport_arg, lport_arg;
814 struct in6_addr *laddr6;
815 int cmd;
816 void (*notify) __P((struct in6pcb *, int));
817 {
818 struct in6pcb *in6p, *nin6p;
819 struct in6_addr faddr6;
820 u_int16_t fport = fport_arg, lport = lport_arg;
821 int errno;
822 int nmatch = 0;
823 void (*notify2) __P((struct in6pcb *, int));
824
825 notify2 = NULL;
826
827 if ((unsigned)cmd > PRC_NCMDS || dst->sa_family != AF_INET6)
828 return 0;
829 faddr6 = ((struct sockaddr_in6 *)dst)->sin6_addr;
830 if (IN6_IS_ADDR_UNSPECIFIED(&faddr6))
831 return 0;
832
833 /*
834 * Redirects go to all references to the destination,
835 * and use in6_rtchange to invalidate the route cache.
836 * Dead host indications: also use in6_rtchange to invalidate
837 * the cache, and deliver the error to all the sockets.
838 * Otherwise, if we have knowledge of the local port and address,
839 * deliver only to that socket.
840 */
841 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
842 fport = 0;
843 lport = 0;
844 bzero((caddr_t)laddr6, sizeof(*laddr6));
845
846 /*
847 * Keep the old notify function to store a soft error
848 * in each PCB.
849 */
850 if (cmd == PRC_HOSTDEAD && notify != in6_rtchange)
851 notify2 = notify;
852
853 notify = in6_rtchange;
854 }
855
856 if (notify == NULL)
857 return 0;
858
859 errno = inet6ctlerrmap[cmd];
860 for (in6p = head->in6p_next; in6p != head; in6p = nin6p) {
861 nin6p = in6p->in6p_next;
862
863 if (notify == in6_rtchange) {
864 /*
865 * Since a non-connected PCB might have a cached route,
866 * we always call in6_rtchange without matching
867 * the PCB to the src/dst pair.
868 *
869 * XXX: we assume in6_rtchange does not free the PCB.
870 */
871 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_route.ro_dst.sin6_addr,
872 &faddr6))
873 in6_rtchange(in6p, errno);
874
875 if (notify2 == NULL)
876 continue;
877
878 notify = notify2;
879 }
880
881 /* at this point, we can assume that NOTIFY is not NULL. */
882
883 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &faddr6) ||
884 in6p->in6p_socket == 0 ||
885 (lport && in6p->in6p_lport != lport) ||
886 (!IN6_IS_ADDR_UNSPECIFIED(laddr6) &&
887 !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6)) ||
888 (fport && in6p->in6p_fport != fport))
889 continue;
890
891 (*notify)(in6p, errno);
892 nmatch++;
893 }
894 return nmatch;
895 }
896
897 void
898 in6_pcbpurgeif(head, ifp)
899 struct in6pcb *head;
900 struct ifnet *ifp;
901 {
902 struct in6pcb *in6p, *nin6p;
903
904 for (in6p = head->in6p_next; in6p != head; in6p = nin6p) {
905 nin6p = in6p->in6p_next;
906 if (in6p->in6p_route.ro_rt != NULL &&
907 in6p->in6p_route.ro_rt->rt_ifp == ifp)
908 in6_rtchange(in6p, 0);
909 }
910 }
911
912 /*
913 * Check for alternatives when higher level complains
914 * about service problems. For now, invalidate cached
915 * routing information. If the route was created dynamically
916 * (by a redirect), time to try a default gateway again.
917 */
918 void
919 in6_losing(in6p)
920 struct in6pcb *in6p;
921 {
922 struct rtentry *rt;
923 struct rt_addrinfo info;
924
925 if ((rt = in6p->in6p_route.ro_rt) != NULL) {
926 in6p->in6p_route.ro_rt = 0;
927 bzero((caddr_t)&info, sizeof(info));
928 info.rti_info[RTAX_DST] =
929 (struct sockaddr *)&in6p->in6p_route.ro_dst;
930 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
931 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
932 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
933 if (rt->rt_flags & RTF_DYNAMIC)
934 (void)rtrequest(RTM_DELETE, rt_key(rt),
935 rt->rt_gateway, rt_mask(rt), rt->rt_flags,
936 (struct rtentry **)0);
937 else
938 /*
939 * A new route can be allocated
940 * the next time output is attempted.
941 */
942 rtfree(rt);
943 }
944 }
945
946 /*
947 * After a routing change, flush old routing
948 * and allocate a (hopefully) better one.
949 */
950 void
951 in6_rtchange(in6p, errno)
952 struct in6pcb *in6p;
953 int errno;
954 {
955 if (in6p->in6p_route.ro_rt) {
956 rtfree(in6p->in6p_route.ro_rt);
957 in6p->in6p_route.ro_rt = 0;
958 /*
959 * A new route can be allocated the next time
960 * output is attempted.
961 */
962 }
963 }
964
965 struct in6pcb *
966 in6_pcblookup(head, faddr6, fport_arg, laddr6, lport_arg, flags)
967 struct in6pcb *head;
968 struct in6_addr *faddr6, *laddr6;
969 u_int fport_arg, lport_arg;
970 int flags;
971 {
972 struct in6pcb *in6p, *match = 0;
973 int matchwild = 3, wildcard;
974 u_int16_t fport = fport_arg, lport = lport_arg;
975
976 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) {
977 if (in6p->in6p_lport != lport)
978 continue;
979 wildcard = 0;
980 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) {
981 if (IN6_IS_ADDR_UNSPECIFIED(laddr6))
982 wildcard++;
983 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6))
984 continue;
985 }
986 #ifndef TCP6
987 else if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)
988 && in6p->in6p_laddr.s6_addr32[3] == 0) {
989 if (!IN6_IS_ADDR_V4MAPPED(laddr6))
990 continue;
991 if (laddr6->s6_addr32[3] == 0)
992 ;
993 else
994 wildcard++;
995 }
996 #endif
997 else {
998 if (IN6_IS_ADDR_V4MAPPED(laddr6)) {
999 #if !defined(TCP6) && !defined(INET6_BINDV6ONLY)
1000 if (in6p->in6p_flags & IN6P_BINDV6ONLY)
1001 continue;
1002 else
1003 wildcard++;
1004 #else
1005 continue;
1006 #endif
1007 } else if (!IN6_IS_ADDR_UNSPECIFIED(laddr6))
1008 wildcard++;
1009 }
1010 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
1011 if (IN6_IS_ADDR_UNSPECIFIED(faddr6))
1012 wildcard++;
1013 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6)
1014 || in6p->in6p_fport != fport)
1015 continue;
1016 }
1017 #ifndef TCP6
1018 else if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)
1019 && in6p->in6p_faddr.s6_addr32[3] == 0) {
1020 if (!IN6_IS_ADDR_V4MAPPED(faddr6))
1021 continue;
1022 if (faddr6->s6_addr32[3] == 0)
1023 ;
1024 else
1025 wildcard++;
1026 }
1027 #endif
1028 else {
1029 if (IN6_IS_ADDR_V4MAPPED(faddr6)) {
1030 #if !defined(TCP6) && !defined(INET6_BINDV6ONLY)
1031 if (in6p->in6p_flags & IN6P_BINDV6ONLY)
1032 continue;
1033 else
1034 wildcard++;
1035 #else
1036 continue;
1037 #endif
1038 } else if (!IN6_IS_ADDR_UNSPECIFIED(faddr6))
1039 wildcard++;
1040 }
1041
1042 if (wildcard && (flags & IN6PLOOKUP_WILDCARD) == 0)
1043 continue;
1044 if (wildcard < matchwild) {
1045 match = in6p;
1046 matchwild = wildcard;
1047 if (matchwild == 0)
1048 break;
1049 }
1050 }
1051 return(match);
1052 }
1053
1054 #ifndef TCP6
1055 struct rtentry *
1056 in6_pcbrtentry(in6p)
1057 struct in6pcb *in6p;
1058 {
1059 struct route_in6 *ro;
1060
1061 ro = &in6p->in6p_route;
1062
1063 if (ro->ro_rt == NULL) {
1064 /*
1065 * No route yet, so try to acquire one.
1066 */
1067 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
1068 bzero(&ro->ro_dst, sizeof(ro->ro_dst));
1069 ro->ro_dst.sin6_family = AF_INET6;
1070 ro->ro_dst.sin6_len = sizeof(struct sockaddr_in6);
1071 satosin6(&ro->ro_dst)->sin6_addr = in6p->in6p_faddr;
1072 rtalloc((struct route *)ro);
1073 }
1074 }
1075 return (ro->ro_rt);
1076 }
1077
1078 struct in6pcb *
1079 in6_pcblookup_connect(head, faddr6, fport_arg, laddr6, lport_arg, faith)
1080 struct in6pcb *head;
1081 struct in6_addr *faddr6, *laddr6;
1082 u_int fport_arg, lport_arg;
1083 int faith;
1084 {
1085 struct in6pcb *in6p;
1086 u_int16_t fport = fport_arg, lport = lport_arg;
1087
1088 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) {
1089 #if defined(NFAITH) && NFAITH > 0
1090 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0)
1091 continue;
1092 #endif
1093 /* find exact match on both source and dest */
1094 if (in6p->in6p_fport != fport)
1095 continue;
1096 if (in6p->in6p_lport != lport)
1097 continue;
1098 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr))
1099 continue;
1100 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6))
1101 continue;
1102 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
1103 continue;
1104 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6))
1105 continue;
1106 return in6p;
1107 }
1108 return NULL;
1109 }
1110
1111 struct in6pcb *
1112 in6_pcblookup_bind(head, laddr6, lport_arg, faith)
1113 struct in6pcb *head;
1114 struct in6_addr *laddr6;
1115 u_int lport_arg;
1116 int faith;
1117 {
1118 struct in6pcb *in6p, *match;
1119 u_int16_t lport = lport_arg;
1120
1121 match = NULL;
1122 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) {
1123 /*
1124 * find destination match. exact match is preferred
1125 * against wildcard match.
1126 */
1127 #if defined(NFAITH) && NFAITH > 0
1128 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0)
1129 continue;
1130 #endif
1131 if (in6p->in6p_fport != 0)
1132 continue;
1133 if (in6p->in6p_lport != lport)
1134 continue;
1135 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) {
1136 if (IN6_IS_ADDR_V4MAPPED(laddr6)) {
1137 #ifndef INET6_BINDV6ONLY
1138 if (in6p->in6p_flags & IN6P_BINDV6ONLY)
1139 continue;
1140 else
1141 match = in6p;
1142 #else
1143 continue;
1144 #endif
1145 } else
1146 match = in6p;
1147 }
1148 else if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) &&
1149 in6p->in6p_laddr.s6_addr32[3] == 0) {
1150 if (IN6_IS_ADDR_V4MAPPED(laddr6) &&
1151 laddr6->s6_addr32[3] != 0)
1152 match = in6p;
1153 }
1154 else if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6))
1155 return in6p;
1156 }
1157 return match;
1158 }
1159 #endif
1160