1 /*-
2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the project nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $
30 */
31
32 /*-
33 * Copyright (c) 1982, 1986, 1991, 1993
34 * The Regents of the University of California. All rights reserved.
35 *
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
38 * are met:
39 * 1. Redistributions of source code must retain the above copyright
40 * notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 * notice, this list of conditions and the following disclaimer in the
43 * documentation and/or other materials provided with the distribution.
44 * 4. Neither the name of the University nor the names of its contributors
45 * may be used to endorse or promote products derived from this software
46 * without specific prior written permission.
47 *
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58 * SUCH DAMAGE.
59 *
60 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
61 */
62
63 #include <sys/cdefs.h>
64 __FBSDID("$FreeBSD: releng/7.4/sys/netinet6/in6_pcb.c 203063 2010-01-27 10:20:10Z bz $");
65
66 #include "opt_inet.h"
67 #include "opt_inet6.h"
68 #include "opt_ipsec.h"
69 #include "opt_mac.h"
70
71 #include <sys/param.h>
72 #include <sys/systm.h>
73 #include <sys/malloc.h>
74 #include <sys/mbuf.h>
75 #include <sys/domain.h>
76 #include <sys/protosw.h>
77 #include <sys/socket.h>
78 #include <sys/socketvar.h>
79 #include <sys/sockio.h>
80 #include <sys/errno.h>
81 #include <sys/time.h>
82 #include <sys/priv.h>
83 #include <sys/proc.h>
84 #include <sys/jail.h>
85
86 #include <vm/uma.h>
87
88 #include <net/if.h>
89 #include <net/if_types.h>
90 #include <net/route.h>
91
92 #include <netinet/in.h>
93 #include <netinet/in_var.h>
94 #include <netinet/in_systm.h>
95 #include <netinet/tcp_var.h>
96 #include <netinet/ip6.h>
97 #include <netinet/ip_var.h>
98 #include <netinet6/ip6_var.h>
99 #include <netinet6/nd6.h>
100 #include <netinet/in_pcb.h>
101 #include <netinet6/in6_pcb.h>
102 #include <netinet6/scope6_var.h>
103
104 #include <security/mac/mac_framework.h>
105
106 struct in6_addr zeroin6_addr;
107
108 int
109 in6_pcbbind(register struct inpcb *inp, struct sockaddr *nam,
110 struct ucred *cred)
111 {
112 struct socket *so = inp->inp_socket;
113 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
114 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
115 u_short lport = 0;
116 int error, wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
117
118 INP_INFO_WLOCK_ASSERT(pcbinfo);
119 INP_WLOCK_ASSERT(inp);
120
121 if (!in6_ifaddr) /* XXX broken! */
122 return (EADDRNOTAVAIL);
123 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
124 return (EINVAL);
125 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
126 wild = INPLOOKUP_WILDCARD;
127 if (nam == NULL) {
128 if ((error = prison_local_ip6(cred, &inp->in6p_laddr,
129 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
130 return (error);
131 } else {
132 sin6 = (struct sockaddr_in6 *)nam;
133 if (nam->sa_len != sizeof(*sin6))
134 return (EINVAL);
135 /*
136 * family check.
137 */
138 if (nam->sa_family != AF_INET6)
139 return (EAFNOSUPPORT);
140
141 if ((error = sa6_embedscope(sin6, ip6_use_defzone)) != 0)
142 return(error);
143
144 if ((error = prison_local_ip6(cred, &sin6->sin6_addr,
145 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
146 return (error);
147
148 lport = sin6->sin6_port;
149 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
150 /*
151 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
152 * allow compepte duplication of binding if
153 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
154 * and a multicast address is bound on both
155 * new and duplicated sockets.
156 */
157 if (so->so_options & SO_REUSEADDR)
158 reuseport = SO_REUSEADDR|SO_REUSEPORT;
159 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
160 struct ifaddr *ia = NULL;
161
162 sin6->sin6_port = 0; /* yech... */
163 if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0)
164 return (EADDRNOTAVAIL);
165
166 /*
167 * XXX: bind to an anycast address might accidentally
168 * cause sending a packet with anycast source address.
169 * We should allow to bind to a deprecated address, since
170 * the application dares to use it.
171 */
172 if (ia &&
173 ((struct in6_ifaddr *)ia)->ia6_flags &
174 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) {
175 return (EADDRNOTAVAIL);
176 }
177 }
178 if (lport) {
179 struct inpcb *t;
180
181 /* GROSS */
182 if (ntohs(lport) <= ipport_reservedhigh &&
183 ntohs(lport) >= ipport_reservedlow &&
184 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT,
185 0))
186 return (EACCES);
187 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) &&
188 priv_check_cred(inp->inp_cred,
189 PRIV_NETINET_REUSEPORT, 0) != 0) {
190 t = in6_pcblookup_local(pcbinfo,
191 &sin6->sin6_addr, lport,
192 INPLOOKUP_WILDCARD, cred);
193 if (t &&
194 ((t->inp_flags & INP_TIMEWAIT) == 0) &&
195 (so->so_type != SOCK_STREAM ||
196 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) &&
197 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
198 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
199 (t->inp_socket->so_options & SO_REUSEPORT)
200 == 0) && (inp->inp_cred->cr_uid !=
201 t->inp_cred->cr_uid))
202 return (EADDRINUSE);
203 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
204 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
205 struct sockaddr_in sin;
206
207 in6_sin6_2_sin(&sin, sin6);
208 t = in_pcblookup_local(pcbinfo,
209 sin.sin_addr, lport,
210 INPLOOKUP_WILDCARD, cred);
211 if (t &&
212 ((t->inp_flags &
213 INP_TIMEWAIT) == 0) &&
214 (so->so_type != SOCK_STREAM ||
215 ntohl(t->inp_faddr.s_addr) ==
216 INADDR_ANY) &&
217 (inp->inp_cred->cr_uid !=
218 t->inp_cred->cr_uid))
219 return (EADDRINUSE);
220 }
221 }
222 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr,
223 lport, wild, cred);
224 if (t && (reuseport & ((t->inp_flags & INP_TIMEWAIT) ?
225 intotw(t)->tw_so_options :
226 t->inp_socket->so_options)) == 0)
227 return (EADDRINUSE);
228 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
229 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
230 struct sockaddr_in sin;
231
232 in6_sin6_2_sin(&sin, sin6);
233 t = in_pcblookup_local(pcbinfo, sin.sin_addr,
234 lport, wild, cred);
235 if (t && t->inp_flags & INP_TIMEWAIT) {
236 if ((reuseport &
237 intotw(t)->tw_so_options) == 0 &&
238 (ntohl(t->inp_laddr.s_addr) !=
239 INADDR_ANY || ((inp->inp_vflag &
240 INP_IPV6PROTO) ==
241 (t->inp_vflag & INP_IPV6PROTO))))
242 return (EADDRINUSE);
243 }
244 else if (t &&
245 (reuseport & t->inp_socket->so_options)
246 == 0 && (ntohl(t->inp_laddr.s_addr) !=
247 INADDR_ANY || INP_SOCKAF(so) ==
248 INP_SOCKAF(t->inp_socket)))
249 return (EADDRINUSE);
250 }
251 }
252 inp->in6p_laddr = sin6->sin6_addr;
253 }
254 if (lport == 0) {
255 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0)
256 return (error);
257 } else {
258 inp->inp_lport = lport;
259 if (in_pcbinshash(inp) != 0) {
260 inp->in6p_laddr = in6addr_any;
261 inp->inp_lport = 0;
262 return (EAGAIN);
263 }
264 }
265 return (0);
266 }
267
268 /*
269 * Transform old in6_pcbconnect() into an inner subroutine for new
270 * in6_pcbconnect(): Do some validity-checking on the remote
271 * address (in mbuf 'nam') and then determine local host address
272 * (i.e., which interface) to use to access that remote host.
273 *
274 * This preserves definition of in6_pcbconnect(), while supporting a
275 * slightly different version for T/TCP. (This is more than
276 * a bit of a kludge, but cleaning up the internal interfaces would
277 * have forced minor changes in every protocol).
278 */
279 int
280 in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam,
281 struct in6_addr *plocal_addr6)
282 {
283 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
284 int error = 0;
285 struct ifnet *ifp = NULL;
286 int scope_ambiguous = 0;
287 struct in6_addr in6a;
288
289 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
290 INP_WLOCK_ASSERT(inp);
291
292 if (nam->sa_len != sizeof (*sin6))
293 return (EINVAL);
294 if (sin6->sin6_family != AF_INET6)
295 return (EAFNOSUPPORT);
296 if (sin6->sin6_port == 0)
297 return (EADDRNOTAVAIL);
298
299 if (sin6->sin6_scope_id == 0 && !ip6_use_defzone)
300 scope_ambiguous = 1;
301 if ((error = sa6_embedscope(sin6, ip6_use_defzone)) != 0)
302 return(error);
303
304 if (in6_ifaddr) {
305 /*
306 * If the destination address is UNSPECIFIED addr,
307 * use the loopback addr, e.g ::1.
308 */
309 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
310 sin6->sin6_addr = in6addr_loopback;
311 }
312 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0)
313 return (error);
314
315 error = in6_selectsrc(sin6, inp->in6p_outputopts,
316 inp, NULL, inp->inp_cred, &ifp, &in6a);
317 if (error)
318 return (error);
319
320 if (ifp && scope_ambiguous &&
321 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) {
322 return(error);
323 }
324
325 /*
326 * Do not update this earlier, in case we return with an error.
327 *
328 * XXX: this in6_selectsrc result might replace the bound local
329 * address with the address specified by setsockopt(IPV6_PKTINFO).
330 * Is it the intended behavior?
331 */
332 *plocal_addr6 = in6a;
333
334 /*
335 * Don't do pcblookup call here; return interface in
336 * plocal_addr6
337 * and exit to caller, that will do the lookup.
338 */
339
340 return (0);
341 }
342
343 /*
344 * Outer subroutine:
345 * Connect from a socket to a specified address.
346 * Both address and port must be specified in argument sin.
347 * If don't have a local address for this socket yet,
348 * then pick one.
349 */
350 int
351 in6_pcbconnect(register struct inpcb *inp, struct sockaddr *nam,
352 struct ucred *cred)
353 {
354 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
355 struct in6_addr addr6;
356 int error;
357
358 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
359 INP_WLOCK_ASSERT(inp);
360
361 /*
362 * Call inner routine, to assign local interface address.
363 * in6_pcbladdr() may automatically fill in sin6_scope_id.
364 */
365 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0)
366 return (error);
367
368 if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr,
369 sin6->sin6_port,
370 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
371 ? &addr6 : &inp->in6p_laddr,
372 inp->inp_lport, 0, NULL) != NULL) {
373 return (EADDRINUSE);
374 }
375 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
376 if (inp->inp_lport == 0) {
377 error = in6_pcbbind(inp, (struct sockaddr *)0, cred);
378 if (error)
379 return (error);
380 }
381 inp->in6p_laddr = addr6;
382 }
383 inp->in6p_faddr = sin6->sin6_addr;
384 inp->inp_fport = sin6->sin6_port;
385 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
386 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
387 if (inp->inp_flags & IN6P_AUTOFLOWLABEL)
388 inp->inp_flow |=
389 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
390
391 in_pcbrehash(inp);
392
393 return (0);
394 }
395
396 void
397 in6_pcbdisconnect(struct inpcb *inp)
398 {
399
400 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
401 INP_WLOCK_ASSERT(inp);
402
403 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
404 inp->inp_fport = 0;
405 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
406 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
407 in_pcbrehash(inp);
408 }
409
410 struct sockaddr *
411 in6_sockaddr(in_port_t port, struct in6_addr *addr_p)
412 {
413 struct sockaddr_in6 *sin6;
414
415 MALLOC(sin6, struct sockaddr_in6 *, sizeof *sin6, M_SONAME, M_WAITOK);
416 bzero(sin6, sizeof *sin6);
417 sin6->sin6_family = AF_INET6;
418 sin6->sin6_len = sizeof(*sin6);
419 sin6->sin6_port = port;
420 sin6->sin6_addr = *addr_p;
421 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
422
423 return (struct sockaddr *)sin6;
424 }
425
426 struct sockaddr *
427 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p)
428 {
429 struct sockaddr_in sin;
430 struct sockaddr_in6 *sin6_p;
431
432 bzero(&sin, sizeof sin);
433 sin.sin_family = AF_INET;
434 sin.sin_len = sizeof(sin);
435 sin.sin_port = port;
436 sin.sin_addr = *addr_p;
437
438 MALLOC(sin6_p, struct sockaddr_in6 *, sizeof *sin6_p, M_SONAME,
439 M_WAITOK);
440 in6_sin_2_v4mapsin6(&sin, sin6_p);
441
442 return (struct sockaddr *)sin6_p;
443 }
444
445 int
446 in6_getsockaddr(struct socket *so, struct sockaddr **nam)
447 {
448 register struct inpcb *inp;
449 struct in6_addr addr;
450 in_port_t port;
451
452 inp = sotoinpcb(so);
453 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL"));
454
455 INP_RLOCK(inp);
456 port = inp->inp_lport;
457 addr = inp->in6p_laddr;
458 INP_RUNLOCK(inp);
459
460 *nam = in6_sockaddr(port, &addr);
461 return 0;
462 }
463
464 int
465 in6_getpeeraddr(struct socket *so, struct sockaddr **nam)
466 {
467 struct inpcb *inp;
468 struct in6_addr addr;
469 in_port_t port;
470
471 inp = sotoinpcb(so);
472 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL"));
473
474 INP_RLOCK(inp);
475 port = inp->inp_fport;
476 addr = inp->in6p_faddr;
477 INP_RUNLOCK(inp);
478
479 *nam = in6_sockaddr(port, &addr);
480 return 0;
481 }
482
483 int
484 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
485 {
486 struct inpcb *inp;
487 int error;
488
489 inp = sotoinpcb(so);
490 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL"));
491
492 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
493 error = in_getsockaddr(so, nam);
494 if (error == 0)
495 in6_sin_2_v4mapsin6_in_sock(nam);
496 } else {
497 /* scope issues will be handled in in6_getsockaddr(). */
498 error = in6_getsockaddr(so, nam);
499 }
500
501 return error;
502 }
503
504 int
505 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
506 {
507 struct inpcb *inp;
508 int error;
509
510 inp = sotoinpcb(so);
511 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL"));
512
513 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
514 error = in_getpeeraddr(so, nam);
515 if (error == 0)
516 in6_sin_2_v4mapsin6_in_sock(nam);
517 } else
518 /* scope issues will be handled in in6_getpeeraddr(). */
519 error = in6_getpeeraddr(so, nam);
520
521 return error;
522 }
523
524 /*
525 * Pass some notification to all connections of a protocol
526 * associated with address dst. The local address and/or port numbers
527 * may be specified to limit the search. The "usual action" will be
528 * taken, depending on the ctlinput cmd. The caller must filter any
529 * cmds that are uninteresting (e.g., no error in the map).
530 * Call the protocol specific routine (if any) to report
531 * any errors for each matching socket.
532 */
533 void
534 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst,
535 u_int fport_arg, const struct sockaddr *src, u_int lport_arg,
536 int cmd, void *cmdarg,
537 struct inpcb *(*notify)(struct inpcb *, int))
538 {
539 struct inpcb *inp, *inp_temp;
540 struct sockaddr_in6 sa6_src, *sa6_dst;
541 u_short fport = fport_arg, lport = lport_arg;
542 u_int32_t flowinfo;
543 int errno;
544
545 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
546 return;
547
548 sa6_dst = (struct sockaddr_in6 *)dst;
549 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
550 return;
551
552 /*
553 * note that src can be NULL when we get notify by local fragmentation.
554 */
555 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
556 flowinfo = sa6_src.sin6_flowinfo;
557
558 /*
559 * Redirects go to all references to the destination,
560 * and use in6_rtchange to invalidate the route cache.
561 * Dead host indications: also use in6_rtchange to invalidate
562 * the cache, and deliver the error to all the sockets.
563 * Otherwise, if we have knowledge of the local port and address,
564 * deliver only to that socket.
565 */
566 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
567 fport = 0;
568 lport = 0;
569 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
570
571 if (cmd != PRC_HOSTDEAD)
572 notify = in6_rtchange;
573 }
574 errno = inet6ctlerrmap[cmd];
575 INP_INFO_WLOCK(pcbinfo);
576 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) {
577 INP_WLOCK(inp);
578 if ((inp->inp_vflag & INP_IPV6) == 0) {
579 INP_WUNLOCK(inp);
580 continue;
581 }
582
583 /*
584 * If the error designates a new path MTU for a destination
585 * and the application (associated with this socket) wanted to
586 * know the value, notify. Note that we notify for all
587 * disconnected sockets if the corresponding application
588 * wanted. This is because some UDP applications keep sending
589 * sockets disconnected.
590 * XXX: should we avoid to notify the value to TCP sockets?
591 */
592 if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 &&
593 (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
594 IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) {
595 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst,
596 (u_int32_t *)cmdarg);
597 }
598
599 /*
600 * Detect if we should notify the error. If no source and
601 * destination ports are specifed, but non-zero flowinfo and
602 * local address match, notify the error. This is the case
603 * when the error is delivered with an encrypted buffer
604 * by ESP. Otherwise, just compare addresses and ports
605 * as usual.
606 */
607 if (lport == 0 && fport == 0 && flowinfo &&
608 inp->inp_socket != NULL &&
609 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) &&
610 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
611 goto do_notify;
612 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
613 &sa6_dst->sin6_addr) ||
614 inp->inp_socket == 0 ||
615 (lport && inp->inp_lport != lport) ||
616 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
617 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
618 &sa6_src.sin6_addr)) ||
619 (fport && inp->inp_fport != fport)) {
620 INP_WUNLOCK(inp);
621 continue;
622 }
623
624 do_notify:
625 if (notify) {
626 if ((*notify)(inp, errno))
627 INP_WUNLOCK(inp);
628 } else
629 INP_WUNLOCK(inp);
630 }
631 INP_INFO_WUNLOCK(pcbinfo);
632 }
633
634 /*
635 * Lookup a PCB based on the local address and port.
636 */
637 struct inpcb *
638 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr,
639 u_short lport, int wild_okay, struct ucred *cred)
640 {
641 register struct inpcb *inp;
642 int matchwild = 3, wildcard;
643
644 INP_INFO_WLOCK_ASSERT(pcbinfo);
645
646 if (!wild_okay) {
647 struct inpcbhead *head;
648 /*
649 * Look for an unconnected (wildcard foreign addr) PCB that
650 * matches the local address and port we're looking for.
651 */
652 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
653 0, pcbinfo->ipi_hashmask)];
654 LIST_FOREACH(inp, head, inp_hash) {
655 /* XXX inp locking */
656 if ((inp->inp_vflag & INP_IPV6) == 0)
657 continue;
658 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
659 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
660 inp->inp_lport == lport) {
661 /* Found. */
662 if (cred == NULL ||
663 inp->inp_cred->cr_prison == cred->cr_prison)
664 return (inp);
665 }
666 }
667 /*
668 * Not found.
669 */
670 return (NULL);
671 } else {
672 struct inpcbporthead *porthash;
673 struct inpcbport *phd;
674 struct inpcb *match = NULL;
675 /*
676 * Best fit PCB lookup.
677 *
678 * First see if this local port is in use by looking on the
679 * port hash list.
680 */
681 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport,
682 pcbinfo->ipi_porthashmask)];
683 LIST_FOREACH(phd, porthash, phd_hash) {
684 if (phd->phd_port == lport)
685 break;
686 }
687 if (phd != NULL) {
688 /*
689 * Port is in use by one or more PCBs. Look for best
690 * fit.
691 */
692 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
693 wildcard = 0;
694 if (cred != NULL &&
695 inp->inp_cred->cr_prison != cred->cr_prison)
696 continue;
697 /* XXX inp locking */
698 if ((inp->inp_vflag & INP_IPV6) == 0)
699 continue;
700 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
701 wildcard++;
702 if (!IN6_IS_ADDR_UNSPECIFIED(
703 &inp->in6p_laddr)) {
704 if (IN6_IS_ADDR_UNSPECIFIED(laddr))
705 wildcard++;
706 else if (!IN6_ARE_ADDR_EQUAL(
707 &inp->in6p_laddr, laddr))
708 continue;
709 } else {
710 if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
711 wildcard++;
712 }
713 if (wildcard < matchwild) {
714 match = inp;
715 matchwild = wildcard;
716 if (matchwild == 0)
717 break;
718 }
719 }
720 }
721 return (match);
722 }
723 }
724
725 void
726 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
727 {
728 struct inpcb *in6p;
729 struct ip6_moptions *im6o;
730 struct in6_multi_mship *imm, *nimm;
731
732 INP_INFO_RLOCK(pcbinfo);
733 LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) {
734 INP_WLOCK(in6p);
735 im6o = in6p->in6p_moptions;
736 if ((in6p->inp_vflag & INP_IPV6) &&
737 im6o) {
738 /*
739 * Unselect the outgoing interface if it is being
740 * detached.
741 */
742 if (im6o->im6o_multicast_ifp == ifp)
743 im6o->im6o_multicast_ifp = NULL;
744
745 /*
746 * Drop multicast group membership if we joined
747 * through the interface being detached.
748 * XXX controversial - is it really legal for kernel
749 * to force this?
750 */
751 for (imm = im6o->im6o_memberships.lh_first;
752 imm != NULL; imm = nimm) {
753 nimm = imm->i6mm_chain.le_next;
754 if (imm->i6mm_maddr->in6m_ifp == ifp) {
755 LIST_REMOVE(imm, i6mm_chain);
756 in6_delmulti(imm->i6mm_maddr);
757 free(imm, M_IP6MADDR);
758 }
759 }
760 }
761 INP_WUNLOCK(in6p);
762 }
763 INP_INFO_RUNLOCK(pcbinfo);
764 }
765
766 /*
767 * Check for alternatives when higher level complains
768 * about service problems. For now, invalidate cached
769 * routing information. If the route was created dynamically
770 * (by a redirect), time to try a default gateway again.
771 */
772 void
773 in6_losing(struct inpcb *in6p)
774 {
775
776 /*
777 * We don't store route pointers in the routing table anymore
778 */
779 return;
780 }
781
782 /*
783 * After a routing change, flush old routing
784 * and allocate a (hopefully) better one.
785 */
786 struct inpcb *
787 in6_rtchange(struct inpcb *inp, int errno)
788 {
789 /*
790 * We don't store route pointers in the routing table anymore
791 */
792 return inp;
793 }
794
795 /*
796 * Lookup PCB in hash list.
797 */
798 struct inpcb *
799 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
800 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg, int wildcard,
801 struct ifnet *ifp)
802 {
803 struct inpcbhead *head;
804 struct inpcb *inp, *tmpinp;
805 u_short fport = fport_arg, lport = lport_arg;
806 int faith;
807
808 INP_INFO_LOCK_ASSERT(pcbinfo);
809
810 if (faithprefix_p != NULL)
811 faith = (*faithprefix_p)(laddr);
812 else
813 faith = 0;
814
815 /*
816 * First look for an exact match.
817 */
818 tmpinp = NULL;
819 head = &pcbinfo->ipi_hashbase[
820 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport,
821 pcbinfo->ipi_hashmask)];
822 LIST_FOREACH(inp, head, inp_hash) {
823 /* XXX inp locking */
824 if ((inp->inp_vflag & INP_IPV6) == 0)
825 continue;
826 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
827 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
828 inp->inp_fport == fport &&
829 inp->inp_lport == lport) {
830 /*
831 * XXX We should be able to directly return
832 * the inp here, without any checks.
833 * Well unless both bound with SO_REUSEPORT?
834 */
835 if (jailed(inp->inp_cred))
836 return (inp);
837 if (tmpinp == NULL)
838 tmpinp = inp;
839 }
840 }
841 if (tmpinp != NULL)
842 return (tmpinp);
843
844 /*
845 * Then look for a wildcard match, if requested.
846 */
847 if (wildcard == INPLOOKUP_WILDCARD) {
848 struct inpcb *local_wild = NULL, *local_exact = NULL;
849 struct inpcb *jail_wild = NULL;
850 int injail;
851
852 /*
853 * Order of socket selection - we always prefer jails.
854 * 1. jailed, non-wild.
855 * 2. jailed, wild.
856 * 3. non-jailed, non-wild.
857 * 4. non-jailed, wild.
858 */
859 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
860 0, pcbinfo->ipi_hashmask)];
861 LIST_FOREACH(inp, head, inp_hash) {
862 /* XXX inp locking */
863 if ((inp->inp_vflag & INP_IPV6) == 0)
864 continue;
865
866 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
867 inp->inp_lport != lport) {
868 continue;
869 }
870
871 /* XXX inp locking */
872 if (faith && (inp->inp_flags & INP_FAITH) == 0)
873 continue;
874
875 injail = jailed(inp->inp_cred);
876 if (injail) {
877 if (prison_check_ip6(inp->inp_cred,
878 laddr) != 0)
879 continue;
880 } else {
881 if (local_exact != NULL)
882 continue;
883 }
884
885 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
886 if (injail)
887 return (inp);
888 else
889 local_exact = inp;
890 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
891 if (injail)
892 jail_wild = inp;
893 else
894 local_wild = inp;
895 }
896 } /* LIST_FOREACH */
897
898 if (jail_wild != NULL)
899 return (jail_wild);
900 if (local_exact != NULL)
901 return (local_exact);
902 if (local_wild != NULL)
903 return (local_wild);
904 } /* if (wildcard == INPLOOKUP_WILDCARD) */
905
906 /*
907 * Not found.
908 */
909 return (NULL);
910 }
911
912 void
913 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
914 {
915 struct ip6_hdr *ip;
916
917 ip = mtod(m, struct ip6_hdr *);
918 bzero(sin6, sizeof(*sin6));
919 sin6->sin6_len = sizeof(*sin6);
920 sin6->sin6_family = AF_INET6;
921 sin6->sin6_addr = ip->ip6_src;
922
923 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */
924
925 return;
926 }
Cache object: b1945080d92fff094aa42b79f1bb02e2
|