1 /*-
2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3 * Copyright (c) 2010-2011 Juniper Networks, Inc.
4 * All rights reserved.
5 *
6 * Portions of this software were developed by Robert N. M. Watson under
7 * contract to Juniper Networks, Inc.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the name of the project nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $
34 */
35
36 /*-
37 * Copyright (c) 1982, 1986, 1991, 1993
38 * The Regents of the University of California. All rights reserved.
39 *
40 * Redistribution and use in source and binary forms, with or without
41 * modification, are permitted provided that the following conditions
42 * are met:
43 * 1. Redistributions of source code must retain the above copyright
44 * notice, this list of conditions and the following disclaimer.
45 * 2. Redistributions in binary form must reproduce the above copyright
46 * notice, this list of conditions and the following disclaimer in the
47 * documentation and/or other materials provided with the distribution.
48 * 4. Neither the name of the University nor the names of its contributors
49 * may be used to endorse or promote products derived from this software
50 * without specific prior written permission.
51 *
52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * SUCH DAMAGE.
63 *
64 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
65 */
66
67 #include <sys/cdefs.h>
68 __FBSDID("$FreeBSD$");
69
70 #include "opt_inet.h"
71 #include "opt_inet6.h"
72 #include "opt_ipsec.h"
73 #include "opt_pcbgroup.h"
74
75 #include <sys/param.h>
76 #include <sys/systm.h>
77 #include <sys/malloc.h>
78 #include <sys/mbuf.h>
79 #include <sys/domain.h>
80 #include <sys/protosw.h>
81 #include <sys/socket.h>
82 #include <sys/socketvar.h>
83 #include <sys/sockio.h>
84 #include <sys/errno.h>
85 #include <sys/time.h>
86 #include <sys/priv.h>
87 #include <sys/proc.h>
88 #include <sys/jail.h>
89
90 #include <vm/uma.h>
91
92 #include <net/if.h>
93 #include <net/if_types.h>
94 #include <net/route.h>
95
96 #include <netinet/in.h>
97 #include <netinet/in_var.h>
98 #include <netinet/in_systm.h>
99 #include <netinet/tcp_var.h>
100 #include <netinet/ip6.h>
101 #include <netinet/ip_var.h>
102
103 #include <netinet6/ip6_var.h>
104 #include <netinet6/nd6.h>
105 #include <netinet/in_pcb.h>
106 #include <netinet6/in6_pcb.h>
107 #include <netinet6/scope6_var.h>
108
109 struct in6_addr zeroin6_addr;
110
111 int
112 in6_pcbbind(register struct inpcb *inp, struct sockaddr *nam,
113 struct ucred *cred)
114 {
115 struct socket *so = inp->inp_socket;
116 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
117 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
118 u_short lport = 0;
119 int error, lookupflags = 0;
120 int reuseport = (so->so_options & SO_REUSEPORT);
121
122 INP_WLOCK_ASSERT(inp);
123 INP_HASH_WLOCK_ASSERT(pcbinfo);
124
125 if (TAILQ_EMPTY(&V_in6_ifaddrhead)) /* XXX broken! */
126 return (EADDRNOTAVAIL);
127 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
128 return (EINVAL);
129 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
130 lookupflags = INPLOOKUP_WILDCARD;
131 if (nam == NULL) {
132 if ((error = prison_local_ip6(cred, &inp->in6p_laddr,
133 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
134 return (error);
135 } else {
136 sin6 = (struct sockaddr_in6 *)nam;
137 if (nam->sa_len != sizeof(*sin6))
138 return (EINVAL);
139 /*
140 * family check.
141 */
142 if (nam->sa_family != AF_INET6)
143 return (EAFNOSUPPORT);
144
145 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
146 return(error);
147
148 if ((error = prison_local_ip6(cred, &sin6->sin6_addr,
149 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
150 return (error);
151
152 lport = sin6->sin6_port;
153 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
154 /*
155 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
156 * allow compepte duplication of binding if
157 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
158 * and a multicast address is bound on both
159 * new and duplicated sockets.
160 */
161 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) != 0)
162 reuseport = SO_REUSEADDR|SO_REUSEPORT;
163 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
164 struct ifaddr *ifa;
165
166 sin6->sin6_port = 0; /* yech... */
167 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) ==
168 NULL &&
169 (inp->inp_flags & INP_BINDANY) == 0) {
170 return (EADDRNOTAVAIL);
171 }
172
173 /*
174 * XXX: bind to an anycast address might accidentally
175 * cause sending a packet with anycast source address.
176 * We should allow to bind to a deprecated address, since
177 * the application dares to use it.
178 */
179 if (ifa != NULL &&
180 ((struct in6_ifaddr *)ifa)->ia6_flags &
181 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) {
182 ifa_free(ifa);
183 return (EADDRNOTAVAIL);
184 }
185 if (ifa != NULL)
186 ifa_free(ifa);
187 }
188 if (lport) {
189 struct inpcb *t;
190 struct tcptw *tw;
191
192 /* GROSS */
193 if (ntohs(lport) <= V_ipport_reservedhigh &&
194 ntohs(lport) >= V_ipport_reservedlow &&
195 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT,
196 0))
197 return (EACCES);
198 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) &&
199 priv_check_cred(inp->inp_cred,
200 PRIV_NETINET_REUSEPORT, 0) != 0) {
201 t = in6_pcblookup_local(pcbinfo,
202 &sin6->sin6_addr, lport,
203 INPLOOKUP_WILDCARD, cred);
204 if (t &&
205 ((t->inp_flags & INP_TIMEWAIT) == 0) &&
206 (so->so_type != SOCK_STREAM ||
207 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) &&
208 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
209 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
210 (t->inp_flags2 & INP_REUSEPORT) == 0) &&
211 (inp->inp_cred->cr_uid !=
212 t->inp_cred->cr_uid))
213 return (EADDRINUSE);
214 #ifdef INET
215 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
216 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
217 struct sockaddr_in sin;
218
219 in6_sin6_2_sin(&sin, sin6);
220 t = in_pcblookup_local(pcbinfo,
221 sin.sin_addr, lport,
222 INPLOOKUP_WILDCARD, cred);
223 if (t &&
224 ((t->inp_flags &
225 INP_TIMEWAIT) == 0) &&
226 (so->so_type != SOCK_STREAM ||
227 ntohl(t->inp_faddr.s_addr) ==
228 INADDR_ANY) &&
229 (inp->inp_cred->cr_uid !=
230 t->inp_cred->cr_uid))
231 return (EADDRINUSE);
232 }
233 #endif
234 }
235 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr,
236 lport, lookupflags, cred);
237 if (t && (t->inp_flags & INP_TIMEWAIT)) {
238 /*
239 * XXXRW: If an incpb has had its timewait
240 * state recycled, we treat the address as
241 * being in use (for now). This is better
242 * than a panic, but not desirable.
243 */
244 tw = intotw(t);
245 if (tw == NULL ||
246 (reuseport & tw->tw_so_options) == 0)
247 return (EADDRINUSE);
248 } else if (t && (reuseport & inp_so_options(t)) == 0) {
249 return (EADDRINUSE);
250 }
251 #ifdef INET
252 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
253 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
254 struct sockaddr_in sin;
255
256 in6_sin6_2_sin(&sin, sin6);
257 t = in_pcblookup_local(pcbinfo, sin.sin_addr,
258 lport, lookupflags, cred);
259 if (t && t->inp_flags & INP_TIMEWAIT) {
260 tw = intotw(t);
261 if (tw == NULL)
262 return (EADDRINUSE);
263 if ((reuseport & tw->tw_so_options) == 0
264 && (ntohl(t->inp_laddr.s_addr) !=
265 INADDR_ANY || ((inp->inp_vflag &
266 INP_IPV6PROTO) ==
267 (t->inp_vflag & INP_IPV6PROTO))))
268 return (EADDRINUSE);
269 } else if (t &&
270 (reuseport & inp_so_options(t)) == 0 &&
271 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
272 (t->inp_vflag & INP_IPV6PROTO) != 0))
273 return (EADDRINUSE);
274 }
275 #endif
276 }
277 inp->in6p_laddr = sin6->sin6_addr;
278 }
279 if (lport == 0) {
280 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) {
281 /* Undo an address bind that may have occurred. */
282 inp->in6p_laddr = in6addr_any;
283 return (error);
284 }
285 } else {
286 inp->inp_lport = lport;
287 if (in_pcbinshash(inp) != 0) {
288 inp->in6p_laddr = in6addr_any;
289 inp->inp_lport = 0;
290 return (EAGAIN);
291 }
292 }
293 return (0);
294 }
295
296 /*
297 * Transform old in6_pcbconnect() into an inner subroutine for new
298 * in6_pcbconnect(): Do some validity-checking on the remote
299 * address (in mbuf 'nam') and then determine local host address
300 * (i.e., which interface) to use to access that remote host.
301 *
302 * This preserves definition of in6_pcbconnect(), while supporting a
303 * slightly different version for T/TCP. (This is more than
304 * a bit of a kludge, but cleaning up the internal interfaces would
305 * have forced minor changes in every protocol).
306 */
307 int
308 in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam,
309 struct in6_addr *plocal_addr6)
310 {
311 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
312 int error = 0;
313 struct ifnet *ifp = NULL;
314 int scope_ambiguous = 0;
315 struct in6_addr in6a;
316
317 INP_WLOCK_ASSERT(inp);
318 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */
319
320 if (nam->sa_len != sizeof (*sin6))
321 return (EINVAL);
322 if (sin6->sin6_family != AF_INET6)
323 return (EAFNOSUPPORT);
324 if (sin6->sin6_port == 0)
325 return (EADDRNOTAVAIL);
326
327 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone)
328 scope_ambiguous = 1;
329 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
330 return(error);
331
332 if (!TAILQ_EMPTY(&V_in6_ifaddrhead)) {
333 /*
334 * If the destination address is UNSPECIFIED addr,
335 * use the loopback addr, e.g ::1.
336 */
337 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
338 sin6->sin6_addr = in6addr_loopback;
339 }
340 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0)
341 return (error);
342
343 error = in6_selectsrc(sin6, inp->in6p_outputopts,
344 inp, NULL, inp->inp_cred, &ifp, &in6a);
345 if (error)
346 return (error);
347
348 if (ifp && scope_ambiguous &&
349 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) {
350 return(error);
351 }
352
353 /*
354 * Do not update this earlier, in case we return with an error.
355 *
356 * XXX: this in6_selectsrc result might replace the bound local
357 * address with the address specified by setsockopt(IPV6_PKTINFO).
358 * Is it the intended behavior?
359 */
360 *plocal_addr6 = in6a;
361
362 /*
363 * Don't do pcblookup call here; return interface in
364 * plocal_addr6
365 * and exit to caller, that will do the lookup.
366 */
367
368 return (0);
369 }
370
371 /*
372 * Outer subroutine:
373 * Connect from a socket to a specified address.
374 * Both address and port must be specified in argument sin.
375 * If don't have a local address for this socket yet,
376 * then pick one.
377 */
378 int
379 in6_pcbconnect_mbuf(register struct inpcb *inp, struct sockaddr *nam,
380 struct ucred *cred, struct mbuf *m)
381 {
382 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
383 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
384 struct in6_addr addr6;
385 int error;
386
387 INP_WLOCK_ASSERT(inp);
388 INP_HASH_WLOCK_ASSERT(pcbinfo);
389
390 /*
391 * Call inner routine, to assign local interface address.
392 * in6_pcbladdr() may automatically fill in sin6_scope_id.
393 */
394 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0)
395 return (error);
396
397 if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr,
398 sin6->sin6_port,
399 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
400 ? &addr6 : &inp->in6p_laddr,
401 inp->inp_lport, 0, NULL) != NULL) {
402 return (EADDRINUSE);
403 }
404 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
405 if (inp->inp_lport == 0) {
406 error = in6_pcbbind(inp, (struct sockaddr *)0, cred);
407 if (error)
408 return (error);
409 }
410 inp->in6p_laddr = addr6;
411 }
412 inp->in6p_faddr = sin6->sin6_addr;
413 inp->inp_fport = sin6->sin6_port;
414 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
415 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
416 if (inp->inp_flags & IN6P_AUTOFLOWLABEL)
417 inp->inp_flow |=
418 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
419
420 in_pcbrehash_mbuf(inp, m);
421
422 return (0);
423 }
424
425 int
426 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred)
427 {
428
429 return (in6_pcbconnect_mbuf(inp, nam, cred, NULL));
430 }
431
432 void
433 in6_pcbdisconnect(struct inpcb *inp)
434 {
435
436 INP_WLOCK_ASSERT(inp);
437 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
438
439 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
440 inp->inp_fport = 0;
441 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
442 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
443 in_pcbrehash(inp);
444 }
445
446 struct sockaddr *
447 in6_sockaddr(in_port_t port, struct in6_addr *addr_p)
448 {
449 struct sockaddr_in6 *sin6;
450
451 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK);
452 bzero(sin6, sizeof *sin6);
453 sin6->sin6_family = AF_INET6;
454 sin6->sin6_len = sizeof(*sin6);
455 sin6->sin6_port = port;
456 sin6->sin6_addr = *addr_p;
457 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
458
459 return (struct sockaddr *)sin6;
460 }
461
462 struct sockaddr *
463 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p)
464 {
465 struct sockaddr_in sin;
466 struct sockaddr_in6 *sin6_p;
467
468 bzero(&sin, sizeof sin);
469 sin.sin_family = AF_INET;
470 sin.sin_len = sizeof(sin);
471 sin.sin_port = port;
472 sin.sin_addr = *addr_p;
473
474 sin6_p = malloc(sizeof *sin6_p, M_SONAME,
475 M_WAITOK);
476 in6_sin_2_v4mapsin6(&sin, sin6_p);
477
478 return (struct sockaddr *)sin6_p;
479 }
480
481 int
482 in6_getsockaddr(struct socket *so, struct sockaddr **nam)
483 {
484 register struct inpcb *inp;
485 struct in6_addr addr;
486 in_port_t port;
487
488 inp = sotoinpcb(so);
489 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL"));
490
491 INP_RLOCK(inp);
492 port = inp->inp_lport;
493 addr = inp->in6p_laddr;
494 INP_RUNLOCK(inp);
495
496 *nam = in6_sockaddr(port, &addr);
497 return 0;
498 }
499
500 int
501 in6_getpeeraddr(struct socket *so, struct sockaddr **nam)
502 {
503 struct inpcb *inp;
504 struct in6_addr addr;
505 in_port_t port;
506
507 inp = sotoinpcb(so);
508 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL"));
509
510 INP_RLOCK(inp);
511 port = inp->inp_fport;
512 addr = inp->in6p_faddr;
513 INP_RUNLOCK(inp);
514
515 *nam = in6_sockaddr(port, &addr);
516 return 0;
517 }
518
519 int
520 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
521 {
522 struct inpcb *inp;
523 int error;
524
525 inp = sotoinpcb(so);
526 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL"));
527
528 #ifdef INET
529 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
530 error = in_getsockaddr(so, nam);
531 if (error == 0)
532 in6_sin_2_v4mapsin6_in_sock(nam);
533 } else
534 #endif
535 {
536 /* scope issues will be handled in in6_getsockaddr(). */
537 error = in6_getsockaddr(so, nam);
538 }
539
540 return error;
541 }
542
543 int
544 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
545 {
546 struct inpcb *inp;
547 int error;
548
549 inp = sotoinpcb(so);
550 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL"));
551
552 #ifdef INET
553 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
554 error = in_getpeeraddr(so, nam);
555 if (error == 0)
556 in6_sin_2_v4mapsin6_in_sock(nam);
557 } else
558 #endif
559 /* scope issues will be handled in in6_getpeeraddr(). */
560 error = in6_getpeeraddr(so, nam);
561
562 return error;
563 }
564
565 /*
566 * Pass some notification to all connections of a protocol
567 * associated with address dst. The local address and/or port numbers
568 * may be specified to limit the search. The "usual action" will be
569 * taken, depending on the ctlinput cmd. The caller must filter any
570 * cmds that are uninteresting (e.g., no error in the map).
571 * Call the protocol specific routine (if any) to report
572 * any errors for each matching socket.
573 */
574 void
575 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst,
576 u_int fport_arg, const struct sockaddr *src, u_int lport_arg,
577 int cmd, void *cmdarg,
578 struct inpcb *(*notify)(struct inpcb *, int))
579 {
580 struct inpcb *inp, *inp_temp;
581 struct sockaddr_in6 sa6_src, *sa6_dst;
582 u_short fport = fport_arg, lport = lport_arg;
583 u_int32_t flowinfo;
584 int errno;
585
586 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
587 return;
588
589 sa6_dst = (struct sockaddr_in6 *)dst;
590 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
591 return;
592
593 /*
594 * note that src can be NULL when we get notify by local fragmentation.
595 */
596 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
597 flowinfo = sa6_src.sin6_flowinfo;
598
599 /*
600 * Redirects go to all references to the destination,
601 * and use in6_rtchange to invalidate the route cache.
602 * Dead host indications: also use in6_rtchange to invalidate
603 * the cache, and deliver the error to all the sockets.
604 * Otherwise, if we have knowledge of the local port and address,
605 * deliver only to that socket.
606 */
607 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
608 fport = 0;
609 lport = 0;
610 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
611
612 if (cmd != PRC_HOSTDEAD)
613 notify = in6_rtchange;
614 }
615 errno = inet6ctlerrmap[cmd];
616 INP_INFO_WLOCK(pcbinfo);
617 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) {
618 INP_WLOCK(inp);
619 if ((inp->inp_vflag & INP_IPV6) == 0) {
620 INP_WUNLOCK(inp);
621 continue;
622 }
623
624 /*
625 * If the error designates a new path MTU for a destination
626 * and the application (associated with this socket) wanted to
627 * know the value, notify.
628 * XXX: should we avoid to notify the value to TCP sockets?
629 */
630 if (cmd == PRC_MSGSIZE && cmdarg != NULL)
631 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst,
632 *(u_int32_t *)cmdarg);
633
634 /*
635 * Detect if we should notify the error. If no source and
636 * destination ports are specifed, but non-zero flowinfo and
637 * local address match, notify the error. This is the case
638 * when the error is delivered with an encrypted buffer
639 * by ESP. Otherwise, just compare addresses and ports
640 * as usual.
641 */
642 if (lport == 0 && fport == 0 && flowinfo &&
643 inp->inp_socket != NULL &&
644 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) &&
645 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
646 goto do_notify;
647 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
648 &sa6_dst->sin6_addr) ||
649 inp->inp_socket == 0 ||
650 (lport && inp->inp_lport != lport) ||
651 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
652 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
653 &sa6_src.sin6_addr)) ||
654 (fport && inp->inp_fport != fport)) {
655 INP_WUNLOCK(inp);
656 continue;
657 }
658
659 do_notify:
660 if (notify) {
661 if ((*notify)(inp, errno))
662 INP_WUNLOCK(inp);
663 } else
664 INP_WUNLOCK(inp);
665 }
666 INP_INFO_WUNLOCK(pcbinfo);
667 }
668
669 /*
670 * Lookup a PCB based on the local address and port. Caller must hold the
671 * hash lock. No inpcb locks or references are acquired.
672 */
673 struct inpcb *
674 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr,
675 u_short lport, int lookupflags, struct ucred *cred)
676 {
677 register struct inpcb *inp;
678 int matchwild = 3, wildcard;
679
680 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
681 ("%s: invalid lookup flags %d", __func__, lookupflags));
682
683 INP_HASH_WLOCK_ASSERT(pcbinfo);
684
685 if ((lookupflags & INPLOOKUP_WILDCARD) == 0) {
686 struct inpcbhead *head;
687 /*
688 * Look for an unconnected (wildcard foreign addr) PCB that
689 * matches the local address and port we're looking for.
690 */
691 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
692 0, pcbinfo->ipi_hashmask)];
693 LIST_FOREACH(inp, head, inp_hash) {
694 /* XXX inp locking */
695 if ((inp->inp_vflag & INP_IPV6) == 0)
696 continue;
697 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
698 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
699 inp->inp_lport == lport) {
700 /* Found. */
701 if (cred == NULL ||
702 prison_equal_ip6(cred->cr_prison,
703 inp->inp_cred->cr_prison))
704 return (inp);
705 }
706 }
707 /*
708 * Not found.
709 */
710 return (NULL);
711 } else {
712 struct inpcbporthead *porthash;
713 struct inpcbport *phd;
714 struct inpcb *match = NULL;
715 /*
716 * Best fit PCB lookup.
717 *
718 * First see if this local port is in use by looking on the
719 * port hash list.
720 */
721 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport,
722 pcbinfo->ipi_porthashmask)];
723 LIST_FOREACH(phd, porthash, phd_hash) {
724 if (phd->phd_port == lport)
725 break;
726 }
727 if (phd != NULL) {
728 /*
729 * Port is in use by one or more PCBs. Look for best
730 * fit.
731 */
732 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
733 wildcard = 0;
734 if (cred != NULL &&
735 !prison_equal_ip6(cred->cr_prison,
736 inp->inp_cred->cr_prison))
737 continue;
738 /* XXX inp locking */
739 if ((inp->inp_vflag & INP_IPV6) == 0)
740 continue;
741 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
742 wildcard++;
743 if (!IN6_IS_ADDR_UNSPECIFIED(
744 &inp->in6p_laddr)) {
745 if (IN6_IS_ADDR_UNSPECIFIED(laddr))
746 wildcard++;
747 else if (!IN6_ARE_ADDR_EQUAL(
748 &inp->in6p_laddr, laddr))
749 continue;
750 } else {
751 if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
752 wildcard++;
753 }
754 if (wildcard < matchwild) {
755 match = inp;
756 matchwild = wildcard;
757 if (matchwild == 0)
758 break;
759 }
760 }
761 }
762 return (match);
763 }
764 }
765
766 void
767 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
768 {
769 struct inpcb *in6p;
770 struct ip6_moptions *im6o;
771 int i, gap;
772
773 INP_INFO_RLOCK(pcbinfo);
774 LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) {
775 INP_WLOCK(in6p);
776 im6o = in6p->in6p_moptions;
777 if ((in6p->inp_vflag & INP_IPV6) && im6o != NULL) {
778 /*
779 * Unselect the outgoing ifp for multicast if it
780 * is being detached.
781 */
782 if (im6o->im6o_multicast_ifp == ifp)
783 im6o->im6o_multicast_ifp = NULL;
784 /*
785 * Drop multicast group membership if we joined
786 * through the interface being detached.
787 */
788 gap = 0;
789 for (i = 0; i < im6o->im6o_num_memberships; i++) {
790 if (im6o->im6o_membership[i]->in6m_ifp ==
791 ifp) {
792 in6_mc_leave(im6o->im6o_membership[i],
793 NULL);
794 gap++;
795 } else if (gap != 0) {
796 im6o->im6o_membership[i - gap] =
797 im6o->im6o_membership[i];
798 }
799 }
800 im6o->im6o_num_memberships -= gap;
801 }
802 INP_WUNLOCK(in6p);
803 }
804 INP_INFO_RUNLOCK(pcbinfo);
805 }
806
807 /*
808 * Check for alternatives when higher level complains
809 * about service problems. For now, invalidate cached
810 * routing information. If the route was created dynamically
811 * (by a redirect), time to try a default gateway again.
812 */
813 void
814 in6_losing(struct inpcb *in6p)
815 {
816
817 /*
818 * We don't store route pointers in the routing table anymore
819 */
820 return;
821 }
822
823 /*
824 * After a routing change, flush old routing
825 * and allocate a (hopefully) better one.
826 */
827 struct inpcb *
828 in6_rtchange(struct inpcb *inp, int errno)
829 {
830 /*
831 * We don't store route pointers in the routing table anymore
832 */
833 return inp;
834 }
835
836 #ifdef PCBGROUP
837 /*
838 * Lookup PCB in hash list, using pcbgroup tables.
839 */
840 static struct inpcb *
841 in6_pcblookup_group(struct inpcbinfo *pcbinfo, struct inpcbgroup *pcbgroup,
842 struct in6_addr *faddr, u_int fport_arg, struct in6_addr *laddr,
843 u_int lport_arg, int lookupflags, struct ifnet *ifp)
844 {
845 struct inpcbhead *head;
846 struct inpcb *inp, *tmpinp;
847 u_short fport = fport_arg, lport = lport_arg;
848 int faith;
849
850 if (faithprefix_p != NULL)
851 faith = (*faithprefix_p)(laddr);
852 else
853 faith = 0;
854
855 /*
856 * First look for an exact match.
857 */
858 tmpinp = NULL;
859 INP_GROUP_LOCK(pcbgroup);
860 head = &pcbgroup->ipg_hashbase[
861 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport,
862 pcbgroup->ipg_hashmask)];
863 LIST_FOREACH(inp, head, inp_pcbgrouphash) {
864 /* XXX inp locking */
865 if ((inp->inp_vflag & INP_IPV6) == 0)
866 continue;
867 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
868 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
869 inp->inp_fport == fport &&
870 inp->inp_lport == lport) {
871 /*
872 * XXX We should be able to directly return
873 * the inp here, without any checks.
874 * Well unless both bound with SO_REUSEPORT?
875 */
876 if (prison_flag(inp->inp_cred, PR_IP6))
877 goto found;
878 if (tmpinp == NULL)
879 tmpinp = inp;
880 }
881 }
882 if (tmpinp != NULL) {
883 inp = tmpinp;
884 goto found;
885 }
886
887 /*
888 * Then look for a wildcard match, if requested.
889 */
890 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
891 struct inpcb *local_wild = NULL, *local_exact = NULL;
892 struct inpcb *jail_wild = NULL;
893 int injail;
894
895 /*
896 * Order of socket selection - we always prefer jails.
897 * 1. jailed, non-wild.
898 * 2. jailed, wild.
899 * 3. non-jailed, non-wild.
900 * 4. non-jailed, wild.
901 */
902 head = &pcbinfo->ipi_wildbase[INP_PCBHASH(INADDR_ANY, lport,
903 0, pcbinfo->ipi_wildmask)];
904 LIST_FOREACH(inp, head, inp_pcbgroup_wild) {
905 /* XXX inp locking */
906 if ((inp->inp_vflag & INP_IPV6) == 0)
907 continue;
908
909 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
910 inp->inp_lport != lport) {
911 continue;
912 }
913
914 /* XXX inp locking */
915 if (faith && (inp->inp_flags & INP_FAITH) == 0)
916 continue;
917
918 injail = prison_flag(inp->inp_cred, PR_IP6);
919 if (injail) {
920 if (prison_check_ip6(inp->inp_cred,
921 laddr) != 0)
922 continue;
923 } else {
924 if (local_exact != NULL)
925 continue;
926 }
927
928 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
929 if (injail)
930 goto found;
931 else
932 local_exact = inp;
933 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
934 if (injail)
935 jail_wild = inp;
936 else
937 local_wild = inp;
938 }
939 } /* LIST_FOREACH */
940
941 inp = jail_wild;
942 if (inp == NULL)
943 inp = jail_wild;
944 if (inp == NULL)
945 inp = local_exact;
946 if (inp == NULL)
947 inp = local_wild;
948 if (inp != NULL)
949 goto found;
950 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
951 INP_GROUP_UNLOCK(pcbgroup);
952 return (NULL);
953
954 found:
955 in_pcbref(inp);
956 INP_GROUP_UNLOCK(pcbgroup);
957 if (lookupflags & INPLOOKUP_WLOCKPCB) {
958 INP_WLOCK(inp);
959 if (in_pcbrele_wlocked(inp))
960 return (NULL);
961 } else if (lookupflags & INPLOOKUP_RLOCKPCB) {
962 INP_RLOCK(inp);
963 if (in_pcbrele_rlocked(inp))
964 return (NULL);
965 } else
966 panic("%s: locking buf", __func__);
967 return (inp);
968 }
969 #endif /* PCBGROUP */
970
971 /*
972 * Lookup PCB in hash list.
973 */
974 struct inpcb *
975 in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
976 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg,
977 int lookupflags, struct ifnet *ifp)
978 {
979 struct inpcbhead *head;
980 struct inpcb *inp, *tmpinp;
981 u_short fport = fport_arg, lport = lport_arg;
982 int faith;
983
984 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
985 ("%s: invalid lookup flags %d", __func__, lookupflags));
986
987 INP_HASH_LOCK_ASSERT(pcbinfo);
988
989 if (faithprefix_p != NULL)
990 faith = (*faithprefix_p)(laddr);
991 else
992 faith = 0;
993
994 /*
995 * First look for an exact match.
996 */
997 tmpinp = NULL;
998 head = &pcbinfo->ipi_hashbase[
999 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport,
1000 pcbinfo->ipi_hashmask)];
1001 LIST_FOREACH(inp, head, inp_hash) {
1002 /* XXX inp locking */
1003 if ((inp->inp_vflag & INP_IPV6) == 0)
1004 continue;
1005 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1006 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1007 inp->inp_fport == fport &&
1008 inp->inp_lport == lport) {
1009 /*
1010 * XXX We should be able to directly return
1011 * the inp here, without any checks.
1012 * Well unless both bound with SO_REUSEPORT?
1013 */
1014 if (prison_flag(inp->inp_cred, PR_IP6))
1015 return (inp);
1016 if (tmpinp == NULL)
1017 tmpinp = inp;
1018 }
1019 }
1020 if (tmpinp != NULL)
1021 return (tmpinp);
1022
1023 /*
1024 * Then look for a wildcard match, if requested.
1025 */
1026 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
1027 struct inpcb *local_wild = NULL, *local_exact = NULL;
1028 struct inpcb *jail_wild = NULL;
1029 int injail;
1030
1031 /*
1032 * Order of socket selection - we always prefer jails.
1033 * 1. jailed, non-wild.
1034 * 2. jailed, wild.
1035 * 3. non-jailed, non-wild.
1036 * 4. non-jailed, wild.
1037 */
1038 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
1039 0, pcbinfo->ipi_hashmask)];
1040 LIST_FOREACH(inp, head, inp_hash) {
1041 /* XXX inp locking */
1042 if ((inp->inp_vflag & INP_IPV6) == 0)
1043 continue;
1044
1045 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
1046 inp->inp_lport != lport) {
1047 continue;
1048 }
1049
1050 /* XXX inp locking */
1051 if (faith && (inp->inp_flags & INP_FAITH) == 0)
1052 continue;
1053
1054 injail = prison_flag(inp->inp_cred, PR_IP6);
1055 if (injail) {
1056 if (prison_check_ip6(inp->inp_cred,
1057 laddr) != 0)
1058 continue;
1059 } else {
1060 if (local_exact != NULL)
1061 continue;
1062 }
1063
1064 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
1065 if (injail)
1066 return (inp);
1067 else
1068 local_exact = inp;
1069 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1070 if (injail)
1071 jail_wild = inp;
1072 else
1073 local_wild = inp;
1074 }
1075 } /* LIST_FOREACH */
1076
1077 if (jail_wild != NULL)
1078 return (jail_wild);
1079 if (local_exact != NULL)
1080 return (local_exact);
1081 if (local_wild != NULL)
1082 return (local_wild);
1083 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
1084
1085 /*
1086 * Not found.
1087 */
1088 return (NULL);
1089 }
1090
1091 /*
1092 * Lookup PCB in hash list, using pcbinfo tables. This variation locks the
1093 * hash list lock, and will return the inpcb locked (i.e., requires
1094 * INPLOOKUP_LOCKPCB).
1095 */
1096 static struct inpcb *
1097 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1098 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1099 struct ifnet *ifp)
1100 {
1101 struct inpcb *inp;
1102
1103 INP_HASH_RLOCK(pcbinfo);
1104 inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport,
1105 (lookupflags & ~(INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)), ifp);
1106 if (inp != NULL) {
1107 in_pcbref(inp);
1108 INP_HASH_RUNLOCK(pcbinfo);
1109 if (lookupflags & INPLOOKUP_WLOCKPCB) {
1110 INP_WLOCK(inp);
1111 if (in_pcbrele_wlocked(inp))
1112 return (NULL);
1113 } else if (lookupflags & INPLOOKUP_RLOCKPCB) {
1114 INP_RLOCK(inp);
1115 if (in_pcbrele_rlocked(inp))
1116 return (NULL);
1117 } else
1118 panic("%s: locking bug", __func__);
1119 } else
1120 INP_HASH_RUNLOCK(pcbinfo);
1121 return (inp);
1122 }
1123
1124 /*
1125 * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf
1126 * from which a pre-calculated hash value may be extracted.
1127 *
1128 * Possibly more of this logic should be in in6_pcbgroup.c.
1129 */
1130 struct inpcb *
1131 in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport,
1132 struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp)
1133 {
1134 #if defined(PCBGROUP)
1135 struct inpcbgroup *pcbgroup;
1136 #endif
1137
1138 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1139 ("%s: invalid lookup flags %d", __func__, lookupflags));
1140 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1141 ("%s: LOCKPCB not set", __func__));
1142
1143 #if defined(PCBGROUP)
1144 if (in_pcbgroup_enabled(pcbinfo)) {
1145 pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr,
1146 fport);
1147 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport,
1148 laddr, lport, lookupflags, ifp));
1149 }
1150 #endif
1151 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1152 lookupflags, ifp));
1153 }
1154
1155 struct inpcb *
1156 in6_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1157 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1158 struct ifnet *ifp, struct mbuf *m)
1159 {
1160 #ifdef PCBGROUP
1161 struct inpcbgroup *pcbgroup;
1162 #endif
1163
1164 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1165 ("%s: invalid lookup flags %d", __func__, lookupflags));
1166 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1167 ("%s: LOCKPCB not set", __func__));
1168
1169 #ifdef PCBGROUP
1170 if (in_pcbgroup_enabled(pcbinfo)) {
1171 pcbgroup = in6_pcbgroup_byhash(pcbinfo, M_HASHTYPE_GET(m),
1172 m->m_pkthdr.flowid);
1173 if (pcbgroup != NULL)
1174 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr,
1175 fport, laddr, lport, lookupflags, ifp));
1176 pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr,
1177 fport);
1178 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport,
1179 laddr, lport, lookupflags, ifp));
1180 }
1181 #endif
1182 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1183 lookupflags, ifp));
1184 }
1185
1186 void
1187 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
1188 {
1189 struct ip6_hdr *ip;
1190
1191 ip = mtod(m, struct ip6_hdr *);
1192 bzero(sin6, sizeof(*sin6));
1193 sin6->sin6_len = sizeof(*sin6);
1194 sin6->sin6_family = AF_INET6;
1195 sin6->sin6_addr = ip->ip6_src;
1196
1197 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */
1198
1199 return;
1200 }
Cache object: 8ed0ada358edf644f5a098a69cc811d7
|