1 /* $NetBSD: in6_pcb.c,v 1.99 2008/08/20 18:35:20 matt 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 1.99 2008/08/20 18:35:20 matt 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
83 #include <net/if.h>
84 #include <net/route.h>
85
86 #include <netinet/in.h>
87 #include <netinet/in_var.h>
88 #include <netinet/in_systm.h>
89 #include <netinet/ip.h>
90 #include <netinet/in_pcb.h>
91 #include <netinet/ip6.h>
92 #include <netinet6/ip6_var.h>
93 #include <netinet6/in6_pcb.h>
94 #include <netinet6/scope6_var.h>
95 #include <netinet6/nd6.h>
96
97 #include "faith.h"
98
99 #ifdef IPSEC
100 #include <netinet6/ipsec.h>
101 #include <netkey/key.h>
102 #endif /* IPSEC */
103
104 #ifdef FAST_IPSEC
105 #include <netipsec/ipsec.h>
106 #include <netipsec/ipsec6.h>
107 #include <netipsec/key.h>
108 #endif /* FAST_IPSEC */
109
110 const struct in6_addr zeroin6_addr;
111
112 #define IN6PCBHASH_PORT(table, lport) \
113 &(table)->inpt_porthashtbl[ntohs(lport) & (table)->inpt_porthash]
114 #define IN6PCBHASH_BIND(table, laddr, lport) \
115 &(table)->inpt_bindhashtbl[ \
116 (((laddr)->s6_addr32[0] ^ (laddr)->s6_addr32[1] ^ \
117 (laddr)->s6_addr32[2] ^ (laddr)->s6_addr32[3]) + ntohs(lport)) & \
118 (table)->inpt_bindhash]
119 #define IN6PCBHASH_CONNECT(table, faddr, fport, laddr, lport) \
120 &(table)->inpt_bindhashtbl[ \
121 ((((faddr)->s6_addr32[0] ^ (faddr)->s6_addr32[1] ^ \
122 (faddr)->s6_addr32[2] ^ (faddr)->s6_addr32[3]) + ntohs(fport)) + \
123 (((laddr)->s6_addr32[0] ^ (laddr)->s6_addr32[1] ^ \
124 (laddr)->s6_addr32[2] ^ (laddr)->s6_addr32[3]) + \
125 ntohs(lport))) & (table)->inpt_bindhash]
126
127 int ip6_anonportmin = IPV6PORT_ANONMIN;
128 int ip6_anonportmax = IPV6PORT_ANONMAX;
129 int ip6_lowportmin = IPV6PORT_RESERVEDMIN;
130 int ip6_lowportmax = IPV6PORT_RESERVEDMAX;
131
132 POOL_INIT(in6pcb_pool, sizeof(struct in6pcb), 0, 0, 0, "in6pcbpl", NULL,
133 IPL_SOFTNET);
134
135 void
136 in6_pcbinit(struct inpcbtable *table, int bindhashsize, int connecthashsize)
137 {
138
139 in_pcbinit(table, bindhashsize, connecthashsize);
140 table->inpt_lastport = (u_int16_t)ip6_anonportmax;
141 }
142
143 int
144 in6_pcballoc(struct socket *so, void *v)
145 {
146 struct inpcbtable *table = v;
147 struct in6pcb *in6p;
148 int s;
149 #if defined(IPSEC) || defined(FAST_IPSEC)
150 int error;
151 #endif
152
153 s = splnet();
154 in6p = pool_get(&in6pcb_pool, PR_NOWAIT);
155 splx(s);
156 if (in6p == NULL)
157 return (ENOBUFS);
158 bzero((void *)in6p, sizeof(*in6p));
159 in6p->in6p_af = AF_INET6;
160 in6p->in6p_table = table;
161 in6p->in6p_socket = so;
162 in6p->in6p_hops = -1; /* use kernel default */
163 in6p->in6p_icmp6filt = NULL;
164 #if defined(IPSEC) || defined(FAST_IPSEC)
165 error = ipsec_init_pcbpolicy(so, &in6p->in6p_sp);
166 if (error != 0) {
167 s = splnet();
168 pool_put(&in6pcb_pool, in6p);
169 splx(s);
170 return error;
171 }
172 #endif /* IPSEC */
173 s = splnet();
174 CIRCLEQ_INSERT_HEAD(&table->inpt_queue, (struct inpcb_hdr*)in6p,
175 inph_queue);
176 LIST_INSERT_HEAD(IN6PCBHASH_PORT(table, in6p->in6p_lport),
177 &in6p->in6p_head, inph_lhash);
178 in6_pcbstate(in6p, IN6P_ATTACHED);
179 splx(s);
180 if (ip6_v6only)
181 in6p->in6p_flags |= IN6P_IPV6_V6ONLY;
182 so->so_pcb = (void *)in6p;
183 return (0);
184 }
185
186 int
187 in6_pcbbind(void *v, struct mbuf *nam, struct lwp *l)
188 {
189 struct in6pcb *in6p = v;
190 struct socket *so = in6p->in6p_socket;
191 struct inpcbtable *table = in6p->in6p_table;
192 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
193 u_int16_t lport = 0;
194 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
195
196 if (in6p->in6p_af != AF_INET6)
197 return (EINVAL);
198
199 if (in6p->in6p_lport || !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
200 return (EINVAL);
201 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0 &&
202 ((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0 ||
203 (so->so_options & SO_ACCEPTCONN) == 0))
204 wild = 1;
205 if (nam) {
206 int error;
207
208 sin6 = mtod(nam, struct sockaddr_in6 *);
209 if (nam->m_len != sizeof(*sin6))
210 return (EINVAL);
211 /*
212 * We should check the family, but old programs
213 * incorrectly fail to intialize it.
214 */
215 if (sin6->sin6_family != AF_INET6)
216 return (EAFNOSUPPORT);
217
218 #ifndef INET
219 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr))
220 return (EADDRNOTAVAIL);
221 #endif
222
223 if ((error = sa6_embedscope(sin6, ip6_use_defzone)) != 0)
224 return (error);
225
226 lport = sin6->sin6_port;
227 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
228 /*
229 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
230 * allow compepte duplication of binding if
231 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
232 * and a multicast address is bound on both
233 * new and duplicated sockets.
234 */
235 if (so->so_options & SO_REUSEADDR)
236 reuseport = SO_REUSEADDR|SO_REUSEPORT;
237 } else if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
238 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
239 return (EINVAL);
240 if (sin6->sin6_addr.s6_addr32[3]) {
241 struct sockaddr_in sin;
242
243 bzero(&sin, sizeof(sin));
244 sin.sin_len = sizeof(sin);
245 sin.sin_family = AF_INET;
246 bcopy(&sin6->sin6_addr.s6_addr32[3],
247 &sin.sin_addr, sizeof(sin.sin_addr));
248 if (ifa_ifwithaddr((struct sockaddr *)&sin) == 0)
249 return EADDRNOTAVAIL;
250 }
251 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
252 struct ifaddr *ia = NULL;
253
254 sin6->sin6_port = 0; /* yech... */
255 if ((in6p->in6p_flags & IN6P_FAITH) == 0 &&
256 (ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0)
257 return (EADDRNOTAVAIL);
258
259 /*
260 * bind to an anycast address might accidentally
261 * cause sending a packet with an anycast source
262 * address, so we forbid it.
263 *
264 * We should allow to bind to a deprecated address,
265 * since the application dare to use it.
266 * But, can we assume that they are careful enough
267 * to check if the address is deprecated or not?
268 * Maybe, as a safeguard, we should have a setsockopt
269 * flag to control the bind(2) behavior against
270 * deprecated addresses (default: forbid bind(2)).
271 */
272 if (ia &&
273 ((struct in6_ifaddr *)ia)->ia6_flags &
274 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED))
275 return (EADDRNOTAVAIL);
276 }
277 if (lport) {
278 #ifndef IPNOPRIVPORTS
279 int priv;
280
281 /*
282 * NOTE: all operating systems use suser() for
283 * privilege check! do not rewrite it into SS_PRIV.
284 */
285 priv = (l && !kauth_authorize_generic(l->l_cred,
286 KAUTH_GENERIC_ISSUSER, NULL)) ? 1 : 0;
287 /* GROSS */
288 if (ntohs(lport) < IPV6PORT_RESERVED && !priv)
289 return (EACCES);
290 #endif
291
292 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
293 #ifdef INET
294 struct inpcb *t;
295
296 t = in_pcblookup_port(table,
297 *(struct in_addr *)&sin6->sin6_addr.s6_addr32[3],
298 lport, wild);
299 if (t && (reuseport & t->inp_socket->so_options) == 0)
300 return (EADDRINUSE);
301 #else
302 return (EADDRNOTAVAIL);
303 #endif
304 }
305
306 {
307 struct in6pcb *t;
308
309 t = in6_pcblookup_port(table, &sin6->sin6_addr,
310 lport, wild);
311 if (t && (reuseport & t->in6p_socket->so_options) == 0)
312 return (EADDRINUSE);
313 }
314 }
315 in6p->in6p_laddr = sin6->sin6_addr;
316 }
317
318 if (lport == 0) {
319 int e;
320 e = in6_pcbsetport(&in6p->in6p_laddr, in6p, l);
321 if (e != 0)
322 return (e);
323 } else {
324 in6p->in6p_lport = lport;
325 in6_pcbstate(in6p, IN6P_BOUND);
326 }
327
328 LIST_REMOVE(&in6p->in6p_head, inph_lhash);
329 LIST_INSERT_HEAD(IN6PCBHASH_PORT(table, in6p->in6p_lport),
330 &in6p->in6p_head, inph_lhash);
331
332 #if 0
333 in6p->in6p_flowinfo = 0; /* XXX */
334 #endif
335 return (0);
336 }
337
338 /*
339 * Connect from a socket to a specified address.
340 * Both address and port must be specified in argument sin6.
341 * If don't have a local address for this socket yet,
342 * then pick one.
343 */
344 int
345 in6_pcbconnect(void *v, struct mbuf *nam, struct lwp *l)
346 {
347 struct rtentry *rt;
348 struct in6pcb *in6p = v;
349 struct in6_addr *in6a = NULL;
350 struct sockaddr_in6 *sin6 = mtod(nam, struct sockaddr_in6 *);
351 struct ifnet *ifp = NULL; /* outgoing interface */
352 int error = 0;
353 int scope_ambiguous = 0;
354 #ifdef INET
355 struct in6_addr mapped;
356 #endif
357 struct sockaddr_in6 tmp;
358
359 (void)&in6a; /* XXX fool gcc */
360
361 if (in6p->in6p_af != AF_INET6)
362 return (EINVAL);
363
364 if (nam->m_len != sizeof(*sin6))
365 return (EINVAL);
366 if (sin6->sin6_family != AF_INET6)
367 return (EAFNOSUPPORT);
368 if (sin6->sin6_port == 0)
369 return (EADDRNOTAVAIL);
370
371 if (sin6->sin6_scope_id == 0 && !ip6_use_defzone)
372 scope_ambiguous = 1;
373 if ((error = sa6_embedscope(sin6, ip6_use_defzone)) != 0)
374 return(error);
375
376 /* sanity check for mapped address case */
377 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
378 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
379 return EINVAL;
380 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
381 in6p->in6p_laddr.s6_addr16[5] = htons(0xffff);
382 if (!IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr))
383 return EINVAL;
384 } else
385 {
386 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr))
387 return EINVAL;
388 }
389
390 /* protect *sin6 from overwrites */
391 tmp = *sin6;
392 sin6 = &tmp;
393
394 /* Source address selection. */
395 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) &&
396 in6p->in6p_laddr.s6_addr32[3] == 0) {
397 #ifdef INET
398 struct sockaddr_in sin, *sinp;
399
400 bzero(&sin, sizeof(sin));
401 sin.sin_len = sizeof(sin);
402 sin.sin_family = AF_INET;
403 bcopy(&sin6->sin6_addr.s6_addr32[3], &sin.sin_addr,
404 sizeof(sin.sin_addr));
405 sinp = in_selectsrc(&sin, &in6p->in6p_route,
406 in6p->in6p_socket->so_options, NULL, &error);
407 if (sinp == 0) {
408 if (error == 0)
409 error = EADDRNOTAVAIL;
410 return (error);
411 }
412 bzero(&mapped, sizeof(mapped));
413 mapped.s6_addr16[5] = htons(0xffff);
414 bcopy(&sinp->sin_addr, &mapped.s6_addr32[3], sizeof(sinp->sin_addr));
415 in6a = &mapped;
416 #else
417 return EADDRNOTAVAIL;
418 #endif
419 } else {
420 /*
421 * XXX: in6_selectsrc might replace the bound local address
422 * with the address specified by setsockopt(IPV6_PKTINFO).
423 * Is it the intended behavior?
424 */
425 in6a = in6_selectsrc(sin6, in6p->in6p_outputopts,
426 in6p->in6p_moptions,
427 &in6p->in6p_route,
428 &in6p->in6p_laddr, &ifp, &error);
429 if (ifp && scope_ambiguous &&
430 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) {
431 return(error);
432 }
433
434 if (in6a == 0) {
435 if (error == 0)
436 error = EADDRNOTAVAIL;
437 return (error);
438 }
439 }
440 if (ifp == NULL && (rt = rtcache_validate(&in6p->in6p_route)) != NULL)
441 ifp = rt->rt_ifp;
442
443 in6p->in6p_ip6.ip6_hlim = (u_int8_t)in6_selecthlim(in6p, ifp);
444
445 if (in6_pcblookup_connect(in6p->in6p_table, &sin6->sin6_addr,
446 sin6->sin6_port,
447 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) ? in6a : &in6p->in6p_laddr,
448 in6p->in6p_lport, 0))
449 return (EADDRINUSE);
450 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) ||
451 (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) &&
452 in6p->in6p_laddr.s6_addr32[3] == 0))
453 {
454 if (in6p->in6p_lport == 0) {
455 error = in6_pcbbind(in6p, (struct mbuf *)0, l);
456 if (error != 0)
457 return error;
458 }
459 in6p->in6p_laddr = *in6a;
460 }
461 in6p->in6p_faddr = sin6->sin6_addr;
462 in6p->in6p_fport = sin6->sin6_port;
463 in6_pcbstate(in6p, IN6P_CONNECTED);
464 in6p->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
465 if (ip6_auto_flowlabel)
466 in6p->in6p_flowinfo |=
467 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
468 #if defined(IPSEC) || defined(FAST_IPSEC)
469 if (in6p->in6p_socket->so_type == SOCK_STREAM)
470 ipsec_pcbconn(in6p->in6p_sp);
471 #endif
472 return (0);
473 }
474
475 void
476 in6_pcbdisconnect(struct in6pcb *in6p)
477 {
478 bzero((void *)&in6p->in6p_faddr, sizeof(in6p->in6p_faddr));
479 in6p->in6p_fport = 0;
480 in6_pcbstate(in6p, IN6P_BOUND);
481 in6p->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
482 #if defined(IPSEC) || defined(FAST_IPSEC)
483 ipsec_pcbdisconn(in6p->in6p_sp);
484 #endif
485 if (in6p->in6p_socket->so_state & SS_NOFDREF)
486 in6_pcbdetach(in6p);
487 }
488
489 void
490 in6_pcbdetach(struct in6pcb *in6p)
491 {
492 struct socket *so = in6p->in6p_socket;
493 int s;
494
495 if (in6p->in6p_af != AF_INET6)
496 return;
497
498 #if defined(IPSEC) || defined(FAST_IPSEC)
499 ipsec6_delete_pcbpolicy(in6p);
500 #endif /* IPSEC */
501 so->so_pcb = 0;
502 if (in6p->in6p_options)
503 m_freem(in6p->in6p_options);
504 if (in6p->in6p_outputopts != NULL) {
505 ip6_clearpktopts(in6p->in6p_outputopts, -1);
506 free(in6p->in6p_outputopts, M_IP6OPT);
507 }
508 rtcache_free(&in6p->in6p_route);
509 ip6_freemoptions(in6p->in6p_moptions);
510 s = splnet();
511 in6_pcbstate(in6p, IN6P_ATTACHED);
512 LIST_REMOVE(&in6p->in6p_head, inph_lhash);
513 CIRCLEQ_REMOVE(&in6p->in6p_table->inpt_queue, &in6p->in6p_head,
514 inph_queue);
515 pool_put(&in6pcb_pool, in6p);
516 splx(s);
517 sofree(so); /* drops the socket's lock */
518 mutex_enter(softnet_lock); /* reacquire it */
519 }
520
521 void
522 in6_setsockaddr(struct in6pcb *in6p, struct mbuf *nam)
523 {
524 struct sockaddr_in6 *sin6;
525
526 if (in6p->in6p_af != AF_INET6)
527 return;
528
529 nam->m_len = sizeof(*sin6);
530 sin6 = mtod(nam, struct sockaddr_in6 *);
531 sockaddr_in6_init(sin6, &in6p->in6p_laddr, in6p->in6p_lport, 0, 0);
532 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
533 }
534
535 void
536 in6_setpeeraddr(struct in6pcb *in6p, struct mbuf *nam)
537 {
538 struct sockaddr_in6 *sin6;
539
540 if (in6p->in6p_af != AF_INET6)
541 return;
542
543 nam->m_len = sizeof(*sin6);
544 sin6 = mtod(nam, struct sockaddr_in6 *);
545 sockaddr_in6_init(sin6, &in6p->in6p_faddr, in6p->in6p_fport, 0, 0);
546 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
547 }
548
549 /*
550 * Pass some notification to all connections of a protocol
551 * associated with address dst. The local address and/or port numbers
552 * may be specified to limit the search. The "usual action" will be
553 * taken, depending on the ctlinput cmd. The caller must filter any
554 * cmds that are uninteresting (e.g., no error in the map).
555 * Call the protocol specific routine (if any) to report
556 * any errors for each matching socket.
557 *
558 * Must be called at splsoftnet.
559 *
560 * Note: src (4th arg) carries the flowlabel value on the original IPv6
561 * header, in sin6_flowinfo member.
562 */
563 int
564 in6_pcbnotify(struct inpcbtable *table, const struct sockaddr *dst,
565 u_int fport_arg, const struct sockaddr *src, u_int lport_arg, int cmd,
566 void *cmdarg, void (*notify)(struct in6pcb *, int))
567 {
568 struct rtentry *rt;
569 struct in6pcb *in6p, *nin6p;
570 struct sockaddr_in6 sa6_src;
571 const struct sockaddr_in6 *sa6_dst;
572 u_int16_t fport = fport_arg, lport = lport_arg;
573 int errno;
574 int nmatch = 0;
575 u_int32_t flowinfo;
576
577 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
578 return 0;
579
580 sa6_dst = (const struct sockaddr_in6 *)dst;
581 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
582 return 0;
583
584 /*
585 * note that src can be NULL when we get notify by local fragmentation.
586 */
587 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
588 flowinfo = sa6_src.sin6_flowinfo;
589
590 /*
591 * Redirects go to all references to the destination,
592 * and use in6_rtchange to invalidate the route cache.
593 * Dead host indications: also use in6_rtchange to invalidate
594 * the cache, and deliver the error to all the sockets.
595 * Otherwise, if we have knowledge of the local port and address,
596 * deliver only to that socket.
597 */
598 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
599 fport = 0;
600 lport = 0;
601 bzero((void *)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
602
603 if (cmd != PRC_HOSTDEAD)
604 notify = in6_rtchange;
605 }
606
607 errno = inet6ctlerrmap[cmd];
608 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue);
609 in6p != (void *)&table->inpt_queue;
610 in6p = nin6p) {
611 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue);
612
613 if (in6p->in6p_af != AF_INET6)
614 continue;
615
616 /*
617 * Under the following condition, notify of redirects
618 * to the pcb, without making address matches against inpcb.
619 * - redirect notification is arrived.
620 * - the inpcb is unconnected.
621 * - the inpcb is caching !RTF_HOST routing entry.
622 * - the ICMPv6 notification is from the gateway cached in the
623 * inpcb. i.e. ICMPv6 notification is from nexthop gateway
624 * the inpcb used very recently.
625 *
626 * This is to improve interaction between netbsd/openbsd
627 * redirect handling code, and inpcb route cache code.
628 * without the clause, !RTF_HOST routing entry (which carries
629 * gateway used by inpcb right before the ICMPv6 redirect)
630 * will be cached forever in unconnected inpcb.
631 *
632 * There still is a question regarding to what is TRT:
633 * - On bsdi/freebsd, RTF_HOST (cloned) routing entry will be
634 * generated on packet output. inpcb will always cache
635 * RTF_HOST routing entry so there's no need for the clause
636 * (ICMPv6 redirect will update RTF_HOST routing entry,
637 * and inpcb is caching it already).
638 * However, bsdi/freebsd are vulnerable to local DoS attacks
639 * due to the cloned routing entries.
640 * - Specwise, "destination cache" is mentioned in RFC2461.
641 * Jinmei says that it implies bsdi/freebsd behavior, itojun
642 * is not really convinced.
643 * - Having hiwat/lowat on # of cloned host route (redirect/
644 * pmtud) may be a good idea. netbsd/openbsd has it. see
645 * icmp6_mtudisc_update().
646 */
647 if ((PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) &&
648 IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr) &&
649 (rt = rtcache_validate(&in6p->in6p_route)) != NULL &&
650 !(rt->rt_flags & RTF_HOST)) {
651 const struct sockaddr_in6 *dst6;
652
653 dst6 = (const struct sockaddr_in6 *)
654 rtcache_getdst(&in6p->in6p_route);
655 if (dst6 == NULL)
656 ;
657 else if (IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr,
658 &sa6_dst->sin6_addr))
659 goto do_notify;
660 }
661
662 /*
663 * If the error designates a new path MTU for a destination
664 * and the application (associated with this socket) wanted to
665 * know the value, notify. Note that we notify for all
666 * disconnected sockets if the corresponding application
667 * wanted. This is because some UDP applications keep sending
668 * sockets disconnected.
669 * XXX: should we avoid to notify the value to TCP sockets?
670 */
671 if (cmd == PRC_MSGSIZE && (in6p->in6p_flags & IN6P_MTU) != 0 &&
672 (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr) ||
673 IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, &sa6_dst->sin6_addr))) {
674 ip6_notify_pmtu(in6p, (const struct sockaddr_in6 *)dst,
675 (u_int32_t *)cmdarg);
676 }
677
678 /*
679 * Detect if we should notify the error. If no source and
680 * destination ports are specified, but non-zero flowinfo and
681 * local address match, notify the error. This is the case
682 * when the error is delivered with an encrypted buffer
683 * by ESP. Otherwise, just compare addresses and ports
684 * as usual.
685 */
686 if (lport == 0 && fport == 0 && flowinfo &&
687 in6p->in6p_socket != NULL &&
688 flowinfo == (in6p->in6p_flowinfo & IPV6_FLOWLABEL_MASK) &&
689 IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &sa6_src.sin6_addr))
690 goto do_notify;
691 else if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr,
692 &sa6_dst->sin6_addr) ||
693 in6p->in6p_socket == 0 ||
694 (lport && in6p->in6p_lport != lport) ||
695 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
696 !IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr,
697 &sa6_src.sin6_addr)) ||
698 (fport && in6p->in6p_fport != fport))
699 continue;
700
701 do_notify:
702 if (notify)
703 (*notify)(in6p, errno);
704 nmatch++;
705 }
706 return nmatch;
707 }
708
709 void
710 in6_pcbpurgeif0(struct inpcbtable *table, struct ifnet *ifp)
711 {
712 struct in6pcb *in6p, *nin6p;
713 struct ip6_moptions *im6o;
714 struct in6_multi_mship *imm, *nimm;
715
716 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue);
717 in6p != (void *)&table->inpt_queue;
718 in6p = nin6p) {
719 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue);
720 if (in6p->in6p_af != AF_INET6)
721 continue;
722
723 im6o = in6p->in6p_moptions;
724 if (im6o) {
725 /*
726 * Unselect the outgoing interface if it is being
727 * detached.
728 */
729 if (im6o->im6o_multicast_ifp == ifp)
730 im6o->im6o_multicast_ifp = NULL;
731
732 /*
733 * Drop multicast group membership if we joined
734 * through the interface being detached.
735 * XXX controversial - is it really legal for kernel
736 * to force this?
737 */
738 for (imm = im6o->im6o_memberships.lh_first;
739 imm != NULL; imm = nimm) {
740 nimm = imm->i6mm_chain.le_next;
741 if (imm->i6mm_maddr->in6m_ifp == ifp) {
742 LIST_REMOVE(imm, i6mm_chain);
743 in6_leavegroup(imm);
744 }
745 }
746 }
747 }
748 }
749
750 void
751 in6_pcbpurgeif(struct inpcbtable *table, struct ifnet *ifp)
752 {
753 struct rtentry *rt;
754 struct in6pcb *in6p, *nin6p;
755
756 for (in6p = (struct in6pcb *)CIRCLEQ_FIRST(&table->inpt_queue);
757 in6p != (void *)&table->inpt_queue;
758 in6p = nin6p) {
759 nin6p = (struct in6pcb *)CIRCLEQ_NEXT(in6p, in6p_queue);
760 if (in6p->in6p_af != AF_INET6)
761 continue;
762 if ((rt = rtcache_validate(&in6p->in6p_route)) != NULL &&
763 rt->rt_ifp == ifp)
764 in6_rtchange(in6p, 0);
765 }
766 }
767
768 /*
769 * Check for alternatives when higher level complains
770 * about service problems. For now, invalidate cached
771 * routing information. If the route was created dynamically
772 * (by a redirect), time to try a default gateway again.
773 */
774 void
775 in6_losing(struct in6pcb *in6p)
776 {
777 struct rtentry *rt;
778 struct rt_addrinfo info;
779
780 if (in6p->in6p_af != AF_INET6)
781 return;
782
783 if ((rt = rtcache_validate(&in6p->in6p_route)) == NULL)
784 return;
785
786 memset(&info, 0, sizeof(info));
787 info.rti_info[RTAX_DST] = rtcache_getdst(&in6p->in6p_route);
788 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
789 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
790 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
791 if (rt->rt_flags & RTF_DYNAMIC) {
792 (void)rtrequest(RTM_DELETE, rt_getkey(rt),
793 rt->rt_gateway, rt_mask(rt), rt->rt_flags, NULL);
794 }
795 /*
796 * A new route can be allocated
797 * the next time output is attempted.
798 */
799 rtcache_free(&in6p->in6p_route);
800 }
801
802 /*
803 * After a routing change, flush old routing. A new route can be
804 * allocated the next time output is attempted.
805 */
806 void
807 in6_rtchange(struct in6pcb *in6p, int errno)
808 {
809 if (in6p->in6p_af != AF_INET6)
810 return;
811
812 rtcache_free(&in6p->in6p_route);
813 /*
814 * A new route can be allocated the next time
815 * output is attempted.
816 */
817 }
818
819 struct in6pcb *
820 in6_pcblookup_port(struct inpcbtable *table, struct in6_addr *laddr6,
821 u_int lport_arg, int lookup_wildcard)
822 {
823 struct inpcbhead *head;
824 struct inpcb_hdr *inph;
825 struct in6pcb *in6p, *match = 0;
826 int matchwild = 3, wildcard;
827 u_int16_t lport = lport_arg;
828
829 head = IN6PCBHASH_PORT(table, lport);
830 LIST_FOREACH(inph, head, inph_lhash) {
831 in6p = (struct in6pcb *)inph;
832 if (in6p->in6p_af != AF_INET6)
833 continue;
834
835 if (in6p->in6p_lport != lport)
836 continue;
837 wildcard = 0;
838 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) {
839 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
840 continue;
841 }
842 if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr))
843 wildcard++;
844 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) {
845 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
846 continue;
847 if (!IN6_IS_ADDR_V4MAPPED(laddr6))
848 continue;
849
850 /* duplicate of IPv4 logic */
851 wildcard = 0;
852 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr) &&
853 in6p->in6p_faddr.s6_addr32[3])
854 wildcard++;
855 if (!in6p->in6p_laddr.s6_addr32[3]) {
856 if (laddr6->s6_addr32[3])
857 wildcard++;
858 } else {
859 if (!laddr6->s6_addr32[3])
860 wildcard++;
861 else {
862 if (in6p->in6p_laddr.s6_addr32[3] !=
863 laddr6->s6_addr32[3])
864 continue;
865 }
866 }
867 } else if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) {
868 if (IN6_IS_ADDR_V4MAPPED(laddr6)) {
869 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
870 continue;
871 }
872 if (!IN6_IS_ADDR_UNSPECIFIED(laddr6))
873 wildcard++;
874 } else {
875 if (IN6_IS_ADDR_V4MAPPED(laddr6)) {
876 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
877 continue;
878 }
879 if (IN6_IS_ADDR_UNSPECIFIED(laddr6))
880 wildcard++;
881 else {
882 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr,
883 laddr6))
884 continue;
885 }
886 }
887 if (wildcard && !lookup_wildcard)
888 continue;
889 if (wildcard < matchwild) {
890 match = in6p;
891 matchwild = wildcard;
892 if (matchwild == 0)
893 break;
894 }
895 }
896 return (match);
897 }
898 #undef continue
899
900 /*
901 * WARNING: return value (rtentry) could be IPv4 one if in6pcb is connected to
902 * IPv4 mapped address.
903 */
904 struct rtentry *
905 in6_pcbrtentry(struct in6pcb *in6p)
906 {
907 struct rtentry *rt;
908 struct route *ro;
909 union {
910 const struct sockaddr *sa;
911 const struct sockaddr_in6 *sa6;
912 #ifdef INET
913 const struct sockaddr_in *sa4;
914 #endif
915 } cdst;
916
917 ro = &in6p->in6p_route;
918
919 if (in6p->in6p_af != AF_INET6)
920 return (NULL);
921
922 cdst.sa = rtcache_getdst(ro);
923 if (cdst.sa == NULL)
924 ;
925 #ifdef INET
926 else if (cdst.sa->sa_family == AF_INET) {
927 KASSERT(IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr));
928 if (cdst.sa4->sin_addr.s_addr != in6p->in6p_faddr.s6_addr32[3])
929 rtcache_free(ro);
930 }
931 #endif
932 else {
933 if (!IN6_ARE_ADDR_EQUAL(&cdst.sa6->sin6_addr,
934 &in6p->in6p_faddr))
935 rtcache_free(ro);
936 }
937 if ((rt = rtcache_validate(ro)) == NULL)
938 rt = rtcache_update(ro, 1);
939 #ifdef INET
940 if (rt == NULL && IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) {
941 union {
942 struct sockaddr dst;
943 struct sockaddr_in dst4;
944 } u;
945 struct in_addr addr;
946
947 addr.s_addr = in6p->in6p_faddr.s6_addr32[3];
948
949 sockaddr_in_init(&u.dst4, &addr, 0);
950 rtcache_setdst(ro, &u.dst);
951
952 rt = rtcache_init(ro);
953 } else
954 #endif
955 if (rt == NULL && !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
956 union {
957 struct sockaddr dst;
958 struct sockaddr_in6 dst6;
959 } u;
960
961 sockaddr_in6_init(&u.dst6, &in6p->in6p_faddr, 0, 0, 0);
962 rtcache_setdst(ro, &u.dst);
963
964 rt = rtcache_init(ro);
965 }
966 return rt;
967 }
968
969 struct in6pcb *
970 in6_pcblookup_connect(struct inpcbtable *table, const struct in6_addr *faddr6,
971 u_int fport_arg, const struct in6_addr *laddr6, u_int lport_arg,
972 int faith)
973 {
974 struct inpcbhead *head;
975 struct inpcb_hdr *inph;
976 struct in6pcb *in6p;
977 u_int16_t fport = fport_arg, lport = lport_arg;
978
979 head = IN6PCBHASH_CONNECT(table, faddr6, fport, laddr6, lport);
980 LIST_FOREACH(inph, head, inph_hash) {
981 in6p = (struct in6pcb *)inph;
982 if (in6p->in6p_af != AF_INET6)
983 continue;
984
985 /* find exact match on both source and dest */
986 if (in6p->in6p_fport != fport)
987 continue;
988 if (in6p->in6p_lport != lport)
989 continue;
990 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr))
991 continue;
992 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, faddr6))
993 continue;
994 if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr))
995 continue;
996 if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6))
997 continue;
998 if ((IN6_IS_ADDR_V4MAPPED(laddr6) ||
999 IN6_IS_ADDR_V4MAPPED(faddr6)) &&
1000 (in6p->in6p_flags & IN6P_IPV6_V6ONLY))
1001 continue;
1002 return in6p;
1003 }
1004 return NULL;
1005 }
1006
1007 struct in6pcb *
1008 in6_pcblookup_bind(struct inpcbtable *table, const struct in6_addr *laddr6,
1009 u_int lport_arg, int faith)
1010 {
1011 struct inpcbhead *head;
1012 struct inpcb_hdr *inph;
1013 struct in6pcb *in6p;
1014 u_int16_t lport = lport_arg;
1015 #ifdef INET
1016 struct in6_addr zero_mapped;
1017 #endif
1018
1019 head = IN6PCBHASH_BIND(table, laddr6, lport);
1020 LIST_FOREACH(inph, head, inph_hash) {
1021 in6p = (struct in6pcb *)inph;
1022 if (in6p->in6p_af != AF_INET6)
1023 continue;
1024
1025 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0)
1026 continue;
1027 if (in6p->in6p_fport != 0)
1028 continue;
1029 if (in6p->in6p_lport != lport)
1030 continue;
1031 if (IN6_IS_ADDR_V4MAPPED(laddr6) &&
1032 (in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
1033 continue;
1034 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, laddr6))
1035 goto out;
1036 }
1037 #ifdef INET
1038 if (IN6_IS_ADDR_V4MAPPED(laddr6)) {
1039 memset(&zero_mapped, 0, sizeof(zero_mapped));
1040 zero_mapped.s6_addr16[5] = 0xffff;
1041 head = IN6PCBHASH_BIND(table, &zero_mapped, lport);
1042 LIST_FOREACH(inph, head, inph_hash) {
1043 in6p = (struct in6pcb *)inph;
1044 if (in6p->in6p_af != AF_INET6)
1045 continue;
1046
1047 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0)
1048 continue;
1049 if (in6p->in6p_fport != 0)
1050 continue;
1051 if (in6p->in6p_lport != lport)
1052 continue;
1053 if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
1054 continue;
1055 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &zero_mapped))
1056 goto out;
1057 }
1058 }
1059 #endif
1060 head = IN6PCBHASH_BIND(table, &zeroin6_addr, lport);
1061 LIST_FOREACH(inph, head, inph_hash) {
1062 in6p = (struct in6pcb *)inph;
1063 if (in6p->in6p_af != AF_INET6)
1064 continue;
1065
1066 if (faith && (in6p->in6p_flags & IN6P_FAITH) == 0)
1067 continue;
1068 if (in6p->in6p_fport != 0)
1069 continue;
1070 if (in6p->in6p_lport != lport)
1071 continue;
1072 if (IN6_IS_ADDR_V4MAPPED(laddr6) &&
1073 (in6p->in6p_flags & IN6P_IPV6_V6ONLY) != 0)
1074 continue;
1075 if (IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &zeroin6_addr))
1076 goto out;
1077 }
1078 return (NULL);
1079
1080 out:
1081 inph = &in6p->in6p_head;
1082 if (inph != LIST_FIRST(head)) {
1083 LIST_REMOVE(inph, inph_hash);
1084 LIST_INSERT_HEAD(head, inph, inph_hash);
1085 }
1086 return in6p;
1087 }
1088
1089 void
1090 in6_pcbstate(struct in6pcb *in6p, int state)
1091 {
1092
1093 if (in6p->in6p_af != AF_INET6)
1094 return;
1095
1096 if (in6p->in6p_state > IN6P_ATTACHED)
1097 LIST_REMOVE(&in6p->in6p_head, inph_hash);
1098
1099 switch (state) {
1100 case IN6P_BOUND:
1101 LIST_INSERT_HEAD(IN6PCBHASH_BIND(in6p->in6p_table,
1102 &in6p->in6p_laddr, in6p->in6p_lport), &in6p->in6p_head,
1103 inph_hash);
1104 break;
1105 case IN6P_CONNECTED:
1106 LIST_INSERT_HEAD(IN6PCBHASH_CONNECT(in6p->in6p_table,
1107 &in6p->in6p_faddr, in6p->in6p_fport,
1108 &in6p->in6p_laddr, in6p->in6p_lport), &in6p->in6p_head,
1109 inph_hash);
1110 break;
1111 }
1112
1113 in6p->in6p_state = state;
1114 }
Cache object: 633e4a9f714aa0e94f9977fb75eef2c2
|