in6_pcb.c revision 1.107 1 /* $NetBSD: in6_pcb.c,v 1.107 2009/04/30 18:18:34 elad Exp $ */
2 /* $KAME: in6_pcb.c,v 1.84 2001/02/08 18:02:08 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. Neither the name of the University nor the names of its contributors
46 * may be used to endorse or promote products derived from this software
47 * without specific prior written permission.
48 *
49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
59 * SUCH DAMAGE.
60 *
61 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
62 */
63
64 #include <sys/cdefs.h>
65 __KERNEL_RCSID(0, "$NetBSD: in6_pcb.c,v 1.107 2009/04/30 18:18:34 elad Exp $");
66
67 #include "opt_inet.h"
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 #include <sys/kauth.h>
82 #include <sys/domain.h>
83
84 #include <net/if.h>
85 #include <net/route.h>
86
87 #include <netinet/in.h>
88 #include <netinet/in_var.h>
89 #include <netinet/in_systm.h>
90 #include <netinet/ip.h>
91 #include <netinet/in_pcb.h>
92 #include <netinet/ip6.h>
93 #include <netinet6/ip6_var.h>
94 #include <netinet6/in6_pcb.h>
95 #include <netinet6/scope6_var.h>
96 #include <netinet6/nd6.h>
97
98 #include "faith.h"
99
100 #ifdef IPSEC
101 #include <netinet6/ipsec.h>
102 #include <netkey/key.h>
103 #endif /* IPSEC */
104
105 #ifdef FAST_IPSEC
106 #include <netipsec/ipsec.h>
107 #include <netipsec/ipsec6.h>
108 #include <netipsec/key.h>
109 #endif /* FAST_IPSEC */
110
111 const struct in6_addr zeroin6_addr;
112
113 #define IN6PCBHASH_PORT(table, lport) \
114 &(table)->inpt_porthashtbl[ntohs(lport) & (table)->inpt_porthash]
115 #define IN6PCBHASH_BIND(table, laddr, lport) \
116 &(table)->inpt_bindhashtbl[ \
117 (((laddr)->s6_addr32[0] ^ (laddr)->s6_addr32[1] ^ \
118 (laddr)->s6_addr32[2] ^ (laddr)->s6_addr32[3]) + ntohs(lport)) & \
119 (table)->inpt_bindhash]
120 #define IN6PCBHASH_CONNECT(table, faddr, fport, laddr, lport) \
121 &(table)->inpt_bindhashtbl[ \
122 ((((faddr)->s6_addr32[0] ^ (faddr)->s6_addr32[1] ^ \
123 (faddr)->s6_addr32[2] ^ (faddr)->s6_addr32[3]) + ntohs(fport)) + \
124 (((laddr)->s6_addr32[0] ^ (laddr)->s6_addr32[1] ^ \
125 (laddr)->s6_addr32[2] ^ (laddr)->s6_addr32[3]) + \
126 ntohs(lport))) & (table)->inpt_bindhash]
127
128 int ip6_anonportmin = IPV6PORT_ANONMIN;
129 int ip6_anonportmax = IPV6PORT_ANONMAX;
130 int ip6_lowportmin = IPV6PORT_RESERVEDMIN;
131 int ip6_lowportmax = IPV6PORT_RESERVEDMAX;
132
133 POOL_INIT(in6pcb_pool, sizeof(struct in6pcb), 0, 0, 0, "in6pcbpl", NULL,
134 IPL_SOFTNET);
135
136 void
137 in6_pcbinit(struct inpcbtable *table, int bindhashsize, int connecthashsize)
138 {
139
140 in_pcbinit(table, bindhashsize, connecthashsize);
141 table->inpt_lastport = (u_int16_t)ip6_anonportmax;
142 }
143
144 int
145 in6_pcballoc(struct socket *so, void *v)
146 {
147 struct inpcbtable *table = v;
148 struct in6pcb *in6p;
149 int s;
150 #if defined(IPSEC) || defined(FAST_IPSEC)
151 int error;
152 #endif
153
154 s = splnet();
155 in6p = pool_get(&in6pcb_pool, PR_NOWAIT);
156 splx(s);
157 if (in6p == NULL)
158 return (ENOBUFS);
159 memset((void *)in6p, 0, sizeof(*in6p));
160 in6p->in6p_af = AF_INET6;
161 in6p->in6p_table = table;
162 in6p->in6p_socket = so;
163 in6p->in6p_hops = -1; /* use kernel default */
164 in6p->in6p_icmp6filt = NULL;
165 #if defined(IPSEC) || defined(FAST_IPSEC)
166 error = ipsec_init_pcbpolicy(so, &in6p->in6p_sp);
167 if (error != 0) {
168 s = splnet();
169 pool_put(&in6pcb_pool, in6p);
170 splx(s);
171 return error;
172 }
173 #endif /* IPSEC */
174 s = splnet();
175 CIRCLEQ_INSERT_HEAD(&table->inpt_queue, (struct inpcb_hdr*)in6p,
176 inph_queue);
177 LIST_INSERT_HEAD(IN6PCBHASH_PORT(table, in6p->in6p_lport),
178 &in6p->in6p_head, inph_lhash);
179 in6_pcbstate(in6p, IN6P_ATTACHED);
180 splx(s);
181 if (ip6_v6only)
182 in6p->in6p_flags |= IN6P_IPV6_V6ONLY;
183 so->so_pcb = (void *)in6p;
184 return (0);
185 }
186
187 /*
188 * Bind address from sin6 to in6p.
189 */
190 static int
191 in6_pcbbind_addr(struct in6pcb *in6p, struct sockaddr_in6 *sin6, struct lwp *l)
192 {
193 int error;
194
195 /*
196 * We should check the family, but old programs
197 * incorrectly fail to intialize it.
198 */
199 if (sin6->sin6_family != AF_INET6)
200 return (EAFNOSUPPORT);
201
202 #ifndef INET
203 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr))
204 return (EADDRNOTAVAIL);
205 #endif
206
207 if ((error = sa6_embedscope(sin6, ip6_use_defzone)) != 0)
208 return (error);
209
210 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
211 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
212 return (EINVAL);
213 if (sin6->sin6_addr.s6_addr32[3]) {
214 struct sockaddr_in sin;
215
216 memset(&sin, 0, sizeof(sin));
217 sin.sin_len = sizeof(sin);
218 sin.sin_family = AF_INET;
219 bcopy(&sin6->sin6_addr.s6_addr32[3],
220 &sin.sin_addr, sizeof(sin.sin_addr));
221 if (ifa_ifwithaddr((struct sockaddr *)&sin) == 0)
222 return EADDRNOTAVAIL;
223 }
224 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
225 struct ifaddr *ia = NULL;
226
227 if ((in6p->in6p_flags & IN6P_FAITH) == 0 &&
228 (ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0)
229 return (EADDRNOTAVAIL);
230
231 /*
232 * bind to an anycast address might accidentally
233 * cause sending a packet with an anycast source
234 * address, so we forbid it.
235 *
236 * We should allow to bind to a deprecated address,
237 * since the application dare to use it.
238 * But, can we assume that they are careful enough
239 * to check if the address is deprecated or not?
240 * Maybe, as a safeguard, we should have a setsockopt
241 * flag to control the bind(2) behavior against
242 * deprecated addresses (default: forbid bind(2)).
243 */
244 if (ia &&
245 ((struct in6_ifaddr *)ia)->ia6_flags &
246 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED))
247 return (EADDRNOTAVAIL);
248 }
249
250
251 in6p->in6p_laddr = sin6->sin6_addr;
252
253
254 return (0);
255 }
256
257 /*
258 * Bind port from sin6 to in6p.
259 */
260 static int
261 in6_pcbbind_port(struct in6pcb *in6p, struct sockaddr_in6 *sin6, struct lwp *l)
262 {
263 struct inpcbtable *table = in6p->in6p_table;
264 struct socket *so = in6p->in6p_socket;
265 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
266 int error;
267
268 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0 &&
269 ((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0 ||
270 (so->so_options & SO_ACCEPTCONN) == 0))
271 wild = 1;
272
273 if (sin6->sin6_port != 0) {
274 enum kauth_network_req req;
275
276 #ifndef IPNOPRIVPORTS
277 if (ntohs(sin6->sin6_port) < IPV6PORT_RESERVED)
278 req = KAUTH_REQ_NETWORK_BIND_PRIVPORT;
279 else
280 #endif /* IPNOPRIVPORTS */
281 req = KAUTH_REQ_NETWORK_BIND_PORT;
282
283 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_BIND,
284 req, so, sin6, NULL);
285 if (error)
286 return (error);
287 }
288
289 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
290 /*
291 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
292 * allow compepte duplication of binding if
293 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
294 * and a multicast address is bound on both
295 * new and duplicated sockets.
296 */
297 if (so->so_options & SO_REUSEADDR)
298 reuseport = SO_REUSEADDR|SO_REUSEPORT;
299 }
300
301 if (sin6->sin6_port != 0) {
302 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
303 #ifdef INET
304 struct inpcb *t;
305
306 t = in_pcblookup_port(table,
307 *(struct in_addr *)&sin6->sin6_addr.s6_addr32[3],
308 sin6->sin6_port, wild);
309 if (t && (reuseport & t->inp_socket->so_options) == 0)
310 return (EADDRINUSE);
311 #else
312 return (EADDRNOTAVAIL);
313 #endif
314 }
315
316 {
317 struct in6pcb *t;
318
319 t = in6_pcblookup_port(table, &sin6->sin6_addr,
320 sin6->sin6_port, wild);
321 if (t && (reuseport & t->in6p_socket->so_options) == 0)
322 return (EADDRINUSE);
323 }
324 }
325
326 if (sin6->sin6_port == 0) {
327 int e;
328 e = in6_pcbsetport(sin6, in6p, l);
329 if (e != 0)
330 return (e);
331 } else {
332 in6p->in6p_lport = sin6->sin6_port;
333 in6_pcbstate(in6p, IN6P_BOUND);
334 }
335
336 LIST_REMOVE(&in6p->in6p_head, inph_lhash);
337 LIST_INSERT_HEAD(IN6PCBHASH_PORT(table, in6p->in6p_lport),
338 &in6p->in6p_head, inph_lhash);
339
340 return (0);
341 }
342
343 int
344 in6_pcbbind(void *v, struct mbuf *nam, struct lwp *l)
345 {
346 struct in6pcb *in6p = v;
347 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
348 int error;
349
350 if (in6p->in6p_af != AF_INET6)
351 return (EINVAL);
352
353 /*
354 * If we already have a local port or a local address it means we're
355 * bounded.
356 */
357 if (in6p->in6p_lport || !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
358 return (EINVAL);
359
360 if (nam != NULL) {
361 /* We were provided a sockaddr_in6 to use. */
362 sin6 = mtod(nam, struct sockaddr_in6 *);
363 if (nam->m_len != sizeof(*sin6))
364 return (EINVAL);
365 } else {
366 /* We always bind to *something*, even if it's "anything". */
367 sin6 = (struct sockaddr_in6 *)
368 __UNCONST(in6p->in6p_socket->so_proto->pr_domain->dom_sa_any);
369 }
370
371 /* Bind address. */
372 error = in6_pcbbind_addr(in6p, sin6, l);
373 if (error)
374 return (error);
375
376 /* Bind port. */
377 error = in6_pcbbind_port(in6p, sin6, l);
378 if (error) {
379 /*
380 * Reset the address here to "any" so we don't "leak" the
381 * in6pcb.
382 */
383 in6p->in6p_laddr = in6addr_any;
384
385 return (error);
386 }
387
388
389 #if 0
390 in6p->in6p_flowinfo = 0; /* XXX */
391 #endif
392 return (0);
393 }
394
395 /*
396 * Connect from a socket to a specified address.
397 * Both address and port must be specified in argument sin6.
398 * If don't have a local address for this socket yet,
399 * then pick one.
400 */
401 int
402 in6_pcbconnect(void *v, struct mbuf *nam, struct lwp *l)
403 {
404 struct rtentry *rt;
405 struct in6pcb *in6p = v;
406 struct in6_addr *in6a = NULL;
407 struct sockaddr_in6 *sin6 = mtod(nam, struct sockaddr_in6 *);
408 struct ifnet *ifp = NULL; /* outgoing interface */
409 int error = 0;
410 int scope_ambiguous = 0;
411 #ifdef INET
412 struct in6_addr mapped;
413 #endif
414 struct sockaddr_in6 tmp;
415
416 (void)&in6a; /* XXX fool gcc */
417
418 if (in6p->in6p_af != AF_INET6)
419 return (EINVAL);
420
421 if (nam->m_len != sizeof(*sin6))
422 return (EINVAL);
423 if (sin6->sin6_family != AF_INET6)
424 return (EAFNOSUPPORT);
425 if (sin6->sin6_port == 0)
426 return (EADDRNOTAVAIL);
427
428 if (sin6->sin6_scope_id == 0 && !ip6_use_defzone)
429 scope_ambiguous = 1;
430 if ((error = sa6_embedscope(sin6, ip6_use_defzone)) != 0)
431 return(error);
432
433 /* sanity check for mapped address case */
434 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
435 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
436 return EINVAL;
437 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
438 in6p->in6p_laddr.s6_addr16[5] = htons(0xffff);
439 if (!IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr))
440 return EINVAL;
441 } else
442 {
443 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr))
444 return EINVAL;
445 }
446
447 /* protect *sin6 from overwrites */
448 tmp = *sin6;
449 sin6 = &tmp;
450
451 /* Source address selection. */
452 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) &&
453 in6p->in6p_laddr.s6_addr32[3] == 0) {
454 #ifdef INET
455 struct sockaddr_in sin, *sinp;
456
457 memset(&sin, 0, sizeof(sin));
458 sin.sin_len = sizeof(sin);
459 sin.sin_family = AF_INET;
460 memcpy(&sin.sin_addr, &sin6->sin6_addr.s6_addr32[3],
461 sizeof(sin.sin_addr));
462 sinp = in_selectsrc(&sin, &in6p->in6p_route,
463 in6p->in6p_socket->so_options, NULL, &error);
464 if (sinp == 0) {
465 if (error == 0)
466 error = EADDRNOTAVAIL;
467 return (error);
468 }
469 memset(&mapped, 0, sizeof(mapped));
470 mapped.s6_addr16[5] = htons(0xffff);
471 memcpy(&mapped.s6_addr32[3], &sinp->sin_addr, sizeof(sinp->sin_addr));
472 in6a = &mapped;
473 #else
474 return EADDRNOTAVAIL;
475 #endif
476 } else {
477 /*
478 * XXX: in6_selectsrc might replace the bound local address
479 * with the address specified by setsockopt(IPV6_PKTINFO).
480 * Is it the intended behavior?
481 */
482 in6a = in6_selectsrc(sin6, in6p->in6p_outputopts,
483 in6p->in6p_moptions,
484 &in6p->in6p_route,
485 &in6p->in6p_laddr, &ifp, &error);
486 if (ifp && scope_ambiguous &&
487 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) {
488 return(error);
489 }
490
491 if (in6a == 0) {
492 if (error == 0)
493 error = EADDRNOTAVAIL;
494 return (error);
495 }
496 }
497 if (ifp == NULL && (rt = rtcache_validate(&in6p->in6p_route)) != NULL)
498 ifp = rt->rt_ifp;
499
500 in6p->in6p_ip6.ip6_hlim = (u_int8_t)in6_selecthlim(in6p, ifp);
501
502 if (in6_pcblookup_connect(in6p->in6p_table, &sin6->sin6_addr,
503 sin6->sin6_port,
504 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) ? in6a : &in6p->in6p_laddr,
505 in6p->in6p_lport, 0))
506 return (EADDRINUSE);
507 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) ||
508 (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) &&
509 in6p->in6p_laddr.s6_addr32[3] == 0))
510 {
511 if (in6p->in6p_lport == 0) {
512 error = in6_pcbbind(in6p, (struct mbuf *)0, l);
513 if (error != 0)
514 return error;
515 }
516 in6p->in6p_laddr = *in6a;
517 }
518 in6p->in6p_faddr = sin6->sin6_addr;
519 in6p->in6p_fport = sin6->sin6_port;
520 in6_pcbstate(in6p, IN6P_CONNECTED);
521 in6p->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
522 if (ip6_auto_flowlabel)
523 in6p->in6p_flowinfo |=
524 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
525 #if defined(IPSEC) || defined(FAST_IPSEC)
526 if (in6p->in6p_socket->so_type == SOCK_STREAM)
527 ipsec_pcbconn(in6p->in6p_sp);
528 #endif
529 return (0);
530 }
531
532 void
533 in6_pcbdisconnect(struct in6pcb *in6p)
534 {
535 memset((void *)&in6p->in6p_faddr, 0, sizeof(in6p->in6p_faddr));
536 in6p->in6p_fport = 0;
537 in6_pcbstate(in6p, IN6P_BOUND);
538 in6p->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
539 #if defined(IPSEC) || defined(FAST_IPSEC)
540 ipsec_pcbdisconn(in6p->in6p_sp);
541 #endif
542 if (in6p->in6p_socket->so_state & SS_NOFDREF)
543 in6_pcbdetach(in6p);
544 }
545
546 void
547 in6_pcbdetach(struct in6pcb *in6p)
548 {
549 struct socket *so = in6p->in6p_socket;
550 int s;
551
552 if (in6p->in6p_af != AF_INET6)
553 return;
554
555 #if defined(IPSEC) || defined(FAST_IPSEC)
556 ipsec6_delete_pcbpolicy(in6p);
557 #endif /* IPSEC */
558 so->so_pcb = 0;
559 if (in6p->in6p_options)
560 m_freem(in6p->in6p_options);
561 if (in6p->in6p_outputopts != NULL) {
562 ip6_clearpktopts(in6p->in6p_outputopts, -1);
563 free(in6p->in6p_outputopts, M_IP6OPT);
564 }
565 rtcache_free(&in6p->in6p_route);
566 ip6_freemoptions(in6p->in6p_moptions);
567 s = splnet();
568 in6_pcbstate(in6p, IN6P_ATTACHED);
569 LIST_REMOVE(&in6p->in6p_head, inph_lhash);
570 CIRCLEQ_REMOVE(&in6p->in6p_table->inpt_queue, &in6p->in6p_head,
571 inph_queue);
572 pool_put(&in6pcb_pool, in6p);
573 splx(s);
574 sofree(so); /* drops the socket's lock */
575 mutex_enter(softnet_lock); /* reacquire it */
576 }
577
578 void
579 in6_setsockaddr(struct in6pcb *in6p, struct mbuf *nam)
580 {
581 struct sockaddr_in6 *sin6;
582
583 if (in6p->in6p_af != AF_INET6)
584 return;
585
586 nam->m_len = sizeof(*sin6);
587 sin6 = mtod(nam, struct sockaddr_in6 *);
588 sockaddr_in6_init(sin6, &in6p->in6p_laddr, in6p->in6p_lport, 0, 0);
589 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
590 }
591
592 void
593 in6_setpeeraddr(struct in6pcb *in6p, struct mbuf *nam)
594 {
595 struct sockaddr_in6 *sin6;
596
597 if (in6p->in6p_af != AF_INET6)
598 return;
599
600 nam->m_len = sizeof(*sin6);
601 sin6 = mtod(nam, struct sockaddr_in6 *);
602 sockaddr_in6_init(sin6, &in6p->in6p_faddr, in6p->in6p_fport, 0, 0);
603 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
604 }
605
606 /*
607 * Pass some notification to all connections of a protocol
608 * associated with address dst. The local address and/or port numbers
609 * may be specified to limit the search. The "usual action" will be
610 * taken, depending on the ctlinput cmd. The caller must filter any
611 * cmds that are uninteresting (e.g., no error in the map).
612 * Call the protocol specific routine (if any) to report
613 * any errors for each matching socket.
614 *
615 * Must be called at splsoftnet.
616 *
617 * Note: src (4th arg) carries the flowlabel value on the original IPv6
618 * header, in sin6_flowinfo member.
619 */
620 int
621 in6_pcbnotify(struct inpcbtable *table, const struct sockaddr *dst,
622 u_int fport_arg, const struct sockaddr *src, u_int lport_arg, int cmd,
623 void *cmdarg, void (*notify)(struct in6pcb *, int))
624 {
625 struct rtentry *rt;
626 struct in6pcb *in6p, *nin6p;
627 struct sockaddr_in6 sa6_src;
628 const struct sockaddr_in6 *sa6_dst;
629 u_int16_t fport = fport_arg, lport = lport_arg;
630 int errno;
631 int nmatch = 0;
632 u_int32_t flowinfo;
633
634 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
635 return 0;
636
637 sa6_dst = (const struct sockaddr_in6 *)dst;
638 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
639 return 0;
640
641 /*
642 * note that src can be NULL when we get notify by local fragmentation.
643 */
644 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
645 flowinfo = sa6_src.sin6_flowinfo;
646
647 /*
648 * Redirects go to all references to the destination,
649 * and use in6_rtchange to invalidate the route cache.
650 * Dead host indications: also use in6_rtchange to invalidate
651 * the cache, and deliver the error to all the sockets.
652 * Otherwise, if we have knowledge of the local port and address,
653 * deliver only to that socket.
654 */
655 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
656 fport = 0;
657 lport = 0;
658 memset((void *)&sa6_src.sin6_addr, 0, sizeof(sa6_src.sin6_addr));
659
660 if (cmd != PRC_HOSTDEAD)
661 notify = in6_rtchange;
662 }
663
664 errno = inet6ctlerrmap[cmd];
665 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue);
666 in6p != (void *)&table->inpt_queue;
667 in6p = nin6p) {
668 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue);
669
670 if (in6p->in6p_af != AF_INET6)
671 continue;
672
673 /*
674 * Under the following condition, notify of redirects
675 * to the pcb, without making address matches against inpcb.
676 * - redirect notification is arrived.
677 * - the inpcb is unconnected.
678 * - the inpcb is caching !RTF_HOST routing entry.
679 * - the ICMPv6 notification is from the gateway cached in the
680 * inpcb. i.e. ICMPv6 notification is from nexthop gateway
681 * the inpcb used very recently.
682 *
683 * This is to improve interaction between netbsd/openbsd
684 * redirect handling code, and inpcb route cache code.
685 * without the clause, !RTF_HOST routing entry (which carries
686 * gateway used by inpcb right before the ICMPv6 redirect)
687 * will be cached forever in unconnected inpcb.
688 *
689 * There still is a question regarding to what is TRT:
690 * - On bsdi/freebsd, RTF_HOST (cloned) routing entry will be
691 * generated on packet output. inpcb will always cache
692 * RTF_HOST routing entry so there's no need for the clause
693 * (ICMPv6 redirect will update RTF_HOST routing entry,
694 * and inpcb is caching it already).
695 * However, bsdi/freebsd are vulnerable to local DoS attacks
696 * due to the cloned routing entries.
697 * - Specwise, "destination cache" is mentioned in RFC2461.
698 * Jinmei says that it implies bsdi/freebsd behavior, itojun
699 * is not really convinced.
700 * - Having hiwat/lowat on # of cloned host route (redirect/
701 * pmtud) may be a good idea. netbsd/openbsd has it. see
702 * icmp6_mtudisc_update().
703 */
704 if ((PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) &&
705 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
706 (rt = rtcache_validate(&in6p->in6p_route)) != NULL &&
707 !(rt->rt_flags & RTF_HOST)) {
708 const struct sockaddr_in6 *dst6;
709
710 dst6 = (const struct sockaddr_in6 *)
711 rtcache_getdst(&in6p->in6p_route);
712 if (dst6 == NULL)
713 ;
714 else if (IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr,
715 &sa6_dst->sin6_addr))
716 goto do_notify;
717 }
718
719 /*
720 * If the error designates a new path MTU for a destination
721 * and the application (associated with this socket) wanted to
722 * know the value, notify. Note that we notify for all
723 * disconnected sockets if the corresponding application
724 * wanted. This is because some UDP applications keep sending
725 * sockets disconnected.
726 * XXX: should we avoid to notify the value to TCP sockets?
727 */
728 if (cmd == PRC_MSGSIZE && (in6p->in6p_flags & IN6P_MTU) != 0 &&
729 (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) ||
730 IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &sa6_dst->sin6_addr))) {
731 ip6_notify_pmtu(in6p, (const struct sockaddr_in6 *)dst,
732 (u_int32_t *)cmdarg);
733 }
734
735 /*
736 * Detect if we should notify the error. If no source and
737 * destination ports are specified, but non-zero flowinfo and
738 * local address match, notify the error. This is the case
739 * when the error is delivered with an encrypted buffer
740 * by ESP. Otherwise, just compare addresses and ports
741 * as usual.
742 */
743 if (lport == 0 && fport == 0 && flowinfo &&
744 in6p->in6p_socket != NULL &&
745 flowinfo == (in6p->in6p_flowinfo & IPV6_FLOWLABEL_MASK) &&
746 IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &sa6_src.sin6_addr))
747 goto do_notify;
748 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr,
749 &sa6_dst->sin6_addr) ||
750 in6p->in6p_socket == 0 ||
751 (lport && in6p->in6p_lport != lport) ||
752 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
753 !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr,
754 &sa6_src.sin6_addr)) ||
755 (fport && in6p->in6p_fport != fport))
756 continue;
757
758 do_notify:
759 if (notify)
760 (*notify)(in6p, errno);
761 nmatch++;
762 }
763 return nmatch;
764 }
765
766 void
767 in6_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp)
768 {
769 struct in6pcb *in6p, *nin6p;
770 struct ip6_moptions *im6o;
771 struct in6_multi_mship *imm, *nimm;
772
773 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue);
774 in6p != (void *)&table->inpt_queue;
775 in6p = nin6p) {
776 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue);
777 if (in6p->in6p_af != AF_INET6)
778 continue;
779
780 im6o = in6p->in6p_moptions;
781 if (im6o) {
782 /*
783 * Unselect the outgoing interface if it is being
784 * detached.
785 */
786 if (im6o->im6o_multicast_ifp == ifp)
787 im6o->im6o_multicast_ifp = NULL;
788
789 /*
790 * Drop multicast group membership if we joined
791 * through the interface being detached.
792 * XXX controversial - is it really legal for kernel
793 * to force this?
794 */
795 for (imm = im6o->im6o_memberships.lh_first;
796 imm != NULL; imm = nimm) {
797 nimm = imm->i6mm_chain.le_next;
798 if (imm->i6mm_maddr->in6m_ifp == ifp) {
799 LIST_REMOVE(imm, i6mm_chain);
800 in6_leavegroup(imm);
801 }
802 }
803 }
804 }
805 }
806
807 void
808 in6_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp)
809 {
810 struct rtentry *rt;
811 struct in6pcb *in6p, *nin6p;
812
813 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue);
814 in6p != (void *)&table->inpt_queue;
815 in6p = nin6p) {
816 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue);
817 if (in6p->in6p_af != AF_INET6)
818 continue;
819 if ((rt = rtcache_validate(&in6p->in6p_route)) != NULL &&
820 rt->rt_ifp == ifp)
821 in6_rtchange(in6p, 0);
822 }
823 }
824
825 /*
826 * Check for alternatives when higher level complains
827 * about service problems. For now, invalidate cached
828 * routing information. If the route was created dynamically
829 * (by a redirect), time to try a default gateway again.
830 */
831 void
832 in6_losing(struct in6pcb *in6p)
833 {
834 struct rtentry *rt;
835 struct rt_addrinfo info;
836
837 if (in6p->in6p_af != AF_INET6)
838 return;
839
840 if ((rt = rtcache_validate(&in6p->in6p_route)) == NULL)
841 return;
842
843 memset(&info, 0, sizeof(info));
844 info.rti_info[RTAX_DST] = rtcache_getdst(&in6p->in6p_route);
845 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
846 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
847 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
848 if (rt->rt_flags & RTF_DYNAMIC) {
849 (void)rtrequest(RTM_DELETE, rt_getkey(rt),
850 rt->rt_gateway, rt_mask(rt), rt->rt_flags, NULL);
851 }
852 /*
853 * A new route can be allocated
854 * the next time output is attempted.
855 */
856 rtcache_free(&in6p->in6p_route);
857 }
858
859 /*
860 * After a routing change, flush old routing. A new route can be
861 * allocated the next time output is attempted.
862 */
863 void
864 in6_rtchange(struct in6pcb *in6p, int errno)
865 {
866 if (in6p->in6p_af != AF_INET6)
867 return;
868
869 rtcache_free(&in6p->in6p_route);
870 /*
871 * A new route can be allocated the next time
872 * output is attempted.
873 */
874 }
875
876 struct in6pcb *
877 in6_pcblookup_port(struct inpcbtable *table, struct in6_addr *laddr6,
878 u_int lport_arg, int lookup_wildcard)
879 {
880 struct inpcbhead *head;
881 struct inpcb_hdr *inph;
882 struct in6pcb *in6p, *match = 0;
883 int matchwild = 3, wildcard;
884 u_int16_t lport = lport_arg;
885
886 head = IN6PCBHASH_PORT(table, lport);
887 LIST_FOREACH(inph, head, inph_lhash) {
888 in6p = (struct in6pcb *)inph;
889 if (in6p->in6p_af != AF_INET6)
890 continue;
891
892 if (in6p->in6p_lport != lport)
893 continue;
894 wildcard = 0;
895 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) {
896 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
897 continue;
898 }
899 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr))
900 wildcard++;
901 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) {
902 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
903 continue;
904 if (!IN6_IS_ADDR_V4MAPPED(laddr6))
905 continue;
906
907 /* duplicate of IPv4 logic */
908 wildcard = 0;
909 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr) &&
910 in6p->in6p_faddr.s6_addr32[3])
911 wildcard++;
912 if (!in6p->in6p_laddr.s6_addr32[3]) {
913 if (laddr6->s6_addr32[3])
914 wildcard++;
915 } else {
916 if (!laddr6->s6_addr32[3])
917 wildcard++;
918 else {
919 if (in6p->in6p_laddr.s6_addr32[3] !=
920 laddr6->s6_addr32[3])
921 continue;
922 }
923 }
924 } else if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) {
925 if (IN6_IS_ADDR_V4MAPPED(laddr6)) {
926 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
927 continue;
928 }
929 if (!IN6_IS_ADDR_UNSPECIFIED(laddr6))
930 wildcard++;
931 } else {
932 if (IN6_IS_ADDR_V4MAPPED(laddr6)) {
933 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
934 continue;
935 }
936 if (IN6_IS_ADDR_UNSPECIFIED(laddr6))
937 wildcard++;
938 else {
939 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr,
940 laddr6))
941 continue;
942 }
943 }
944 if (wildcard && !lookup_wildcard)
945 continue;
946 if (wildcard < matchwild) {
947 match = in6p;
948 matchwild = wildcard;
949 if (matchwild == 0)
950 break;
951 }
952 }
953 return (match);
954 }
955 #undef continue
956
957 /*
958 * WARNING: return value (rtentry) could be IPv4 one if in6pcb is connected to
959 * IPv4 mapped address.
960 */
961 struct rtentry *
962 in6_pcbrtentry(struct in6pcb *in6p)
963 {
964 struct rtentry *rt;
965 struct route *ro;
966 union {
967 const struct sockaddr *sa;
968 const struct sockaddr_in6 *sa6;
969 #ifdef INET
970 const struct sockaddr_in *sa4;
971 #endif
972 } cdst;
973
974 ro = &in6p->in6p_route;
975
976 if (in6p->in6p_af != AF_INET6)
977 return (NULL);
978
979 cdst.sa = rtcache_getdst(ro);
980 if (cdst.sa == NULL)
981 ;
982 #ifdef INET
983 else if (cdst.sa->sa_family == AF_INET) {
984 KASSERT(IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr));
985 if (cdst.sa4->sin_addr.s_addr != in6p->in6p_faddr.s6_addr32[3])
986 rtcache_free(ro);
987 }
988 #endif
989 else {
990 if (!IN6_ARE_ADDR_EQUAL(&cdst.sa6->sin6_addr,
991 &in6p->in6p_faddr))
992 rtcache_free(ro);
993 }
994 if ((rt = rtcache_validate(ro)) == NULL)
995 rt = rtcache_update(ro, 1);
996 #ifdef INET
997 if (rt == NULL && IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) {
998 union {
999 struct sockaddr dst;
1000 struct sockaddr_in dst4;
1001 } u;
1002 struct in_addr addr;
1003
1004 addr.s_addr = in6p->in6p_faddr.s6_addr32[3];
1005
1006 sockaddr_in_init(&u.dst4, &addr, 0);
1007 rtcache_setdst(ro, &u.dst);
1008
1009 rt = rtcache_init(ro);
1010 } else
1011 #endif
1012 if (rt == NULL && !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
1013 union {
1014 struct sockaddr dst;
1015 struct sockaddr_in6 dst6;
1016 } u;
1017
1018 sockaddr_in6_init(&u.dst6, &in6p->in6p_faddr, 0, 0, 0);
1019 rtcache_setdst(ro, &u.dst);
1020
1021 rt = rtcache_init(ro);
1022 }
1023 return rt;
1024 }
1025
1026 struct in6pcb *
1027 in6_pcblookup_connect(struct inpcbtable *table, const struct in6_addr *faddr6,
1028 u_int fport_arg, const struct in6_addr *laddr6, u_int lport_arg,
1029 int faith)
1030 {
1031 struct inpcbhead *head;
1032 struct inpcb_hdr *inph;
1033 struct in6pcb *in6p;
1034 u_int16_t fport = fport_arg, lport = lport_arg;
1035
1036 head = IN6PCBHASH_CONNECT(table, faddr6, fport, laddr6, lport);
1037 LIST_FOREACH(inph, head, inph_hash) {
1038 in6p = (struct in6pcb *)inph;
1039 if (in6p->in6p_af != AF_INET6)
1040 continue;
1041
1042 /* find exact match on both source and dest */
1043 if (in6p->in6p_fport != fport)
1044 continue;
1045 if (in6p->in6p_lport != lport)
1046 continue;
1047 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr))
1048 continue;
1049 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6))
1050 continue;
1051 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
1052 continue;
1053 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6))
1054 continue;
1055 if ((IN6_IS_ADDR_V4MAPPED(laddr6) ||
1056 IN6_IS_ADDR_V4MAPPED(faddr6)) &&
1057 (in6p->in6p_flags & IN6P_IPV6_V6ONLY))
1058 continue;
1059 return in6p;
1060 }
1061 return NULL;
1062 }
1063
1064 struct in6pcb *
1065 in6_pcblookup_bind(struct inpcbtable *table, const struct in6_addr *laddr6,
1066 u_int lport_arg, int faith)
1067 {
1068 struct inpcbhead *head;
1069 struct inpcb_hdr *inph;
1070 struct in6pcb *in6p;
1071 u_int16_t lport = lport_arg;
1072 #ifdef INET
1073 struct in6_addr zero_mapped;
1074 #endif
1075
1076 head = IN6PCBHASH_BIND(table, laddr6, lport);
1077 LIST_FOREACH(inph, head, inph_hash) {
1078 in6p = (struct in6pcb *)inph;
1079 if (in6p->in6p_af != AF_INET6)
1080 continue;
1081
1082 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0)
1083 continue;
1084 if (in6p->in6p_fport != 0)
1085 continue;
1086 if (in6p->in6p_lport != lport)
1087 continue;
1088 if (IN6_IS_ADDR_V4MAPPED(laddr6) &&
1089 (in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
1090 continue;
1091 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6))
1092 goto out;
1093 }
1094 #ifdef INET
1095 if (IN6_IS_ADDR_V4MAPPED(laddr6)) {
1096 memset(&zero_mapped, 0, sizeof(zero_mapped));
1097 zero_mapped.s6_addr16[5] = 0xffff;
1098 head = IN6PCBHASH_BIND(table, &zero_mapped, lport);
1099 LIST_FOREACH(inph, head, inph_hash) {
1100 in6p = (struct in6pcb *)inph;
1101 if (in6p->in6p_af != AF_INET6)
1102 continue;
1103
1104 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0)
1105 continue;
1106 if (in6p->in6p_fport != 0)
1107 continue;
1108 if (in6p->in6p_lport != lport)
1109 continue;
1110 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
1111 continue;
1112 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &zero_mapped))
1113 goto out;
1114 }
1115 }
1116 #endif
1117 head = IN6PCBHASH_BIND(table, &zeroin6_addr, lport);
1118 LIST_FOREACH(inph, head, inph_hash) {
1119 in6p = (struct in6pcb *)inph;
1120 if (in6p->in6p_af != AF_INET6)
1121 continue;
1122
1123 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0)
1124 continue;
1125 if (in6p->in6p_fport != 0)
1126 continue;
1127 if (in6p->in6p_lport != lport)
1128 continue;
1129 if (IN6_IS_ADDR_V4MAPPED(laddr6) &&
1130 (in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
1131 continue;
1132 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &zeroin6_addr))
1133 goto out;
1134 }
1135 return (NULL);
1136
1137 out:
1138 inph = &in6p->in6p_head;
1139 if (inph != LIST_FIRST(head)) {
1140 LIST_REMOVE(inph, inph_hash);
1141 LIST_INSERT_HEAD(head, inph, inph_hash);
1142 }
1143 return in6p;
1144 }
1145
1146 void
1147 in6_pcbstate(struct in6pcb *in6p, int state)
1148 {
1149
1150 if (in6p->in6p_af != AF_INET6)
1151 return;
1152
1153 if (in6p->in6p_state > IN6P_ATTACHED)
1154 LIST_REMOVE(&in6p->in6p_head, inph_hash);
1155
1156 switch (state) {
1157 case IN6P_BOUND:
1158 LIST_INSERT_HEAD(IN6PCBHASH_BIND(in6p->in6p_table,
1159 &in6p->in6p_laddr, in6p->in6p_lport), &in6p->in6p_head,
1160 inph_hash);
1161 break;
1162 case IN6P_CONNECTED:
1163 LIST_INSERT_HEAD(IN6PCBHASH_CONNECT(in6p->in6p_table,
1164 &in6p->in6p_faddr, in6p->in6p_fport,
1165 &in6p->in6p_laddr, in6p->in6p_lport), &in6p->in6p_head,
1166 inph_hash);
1167 break;
1168 }
1169
1170 in6p->in6p_state = state;
1171 }
1172