xref: /freebsd/sys/netinet6/in6_pcb.c (revision b601c69bdbe8755d26570261d7fd4c02ee4eff74)
1 /*	$FreeBSD$	*/
2 /*	$KAME: in6_pcb.c,v 1.8 2000/06/09 00:37:02 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 /*
35  * Copyright (c) 1982, 1986, 1991, 1993
36  *	The Regents of the University of California.  All rights reserved.
37  *
38  * Redistribution and use in source and binary forms, with or without
39  * modification, are permitted provided that the following conditions
40  * are met:
41  * 1. Redistributions of source code must retain the above copyright
42  *    notice, this list of conditions and the following disclaimer.
43  * 2. Redistributions in binary form must reproduce the above copyright
44  *    notice, this list of conditions and the following disclaimer in the
45  *    documentation and/or other materials provided with the distribution.
46  * 3. All advertising materials mentioning features or use of this software
47  *    must display the following acknowledgement:
48  *	This product includes software developed by the University of
49  *	California, Berkeley and its contributors.
50  * 4. Neither the name of the University nor the names of its contributors
51  *    may be used to endorse or promote products derived from this software
52  *    without specific prior written permission.
53  *
54  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
64  * SUCH DAMAGE.
65  *
66  *	@(#)in_pcb.c	8.2 (Berkeley) 1/4/94
67  */
68 
69 #include "opt_ipsec.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/proc.h>
83 #include <sys/jail.h>
84 
85 #include <vm/vm_zone.h>
86 
87 #include <net/if.h>
88 #include <net/if_types.h>
89 #include <net/route.h>
90 
91 #include <netinet/in.h>
92 #include <netinet/in_var.h>
93 #include <netinet/in_systm.h>
94 #include <netinet/ip6.h>
95 #include <netinet/ip_var.h>
96 #include <netinet6/ip6_var.h>
97 #include <netinet6/nd6.h>
98 #include <netinet/in_pcb.h>
99 #include <netinet6/in6_pcb.h>
100 
101 #include "faith.h"
102 
103 #ifdef IPSEC
104 #include <netinet6/ipsec.h>
105 #include <netinet6/ah.h>
106 #include <netinet6/ipsec6.h>
107 #include <netinet6/ah6.h>
108 #include <netkey/key.h>
109 #endif /* IPSEC */
110 
111 struct	in6_addr zeroin6_addr;
112 
113 int
114 in6_pcbbind(inp, nam, p)
115 	register struct inpcb *inp;
116 	struct sockaddr *nam;
117 	struct proc *p;
118 {
119 	struct socket *so = inp->inp_socket;
120 	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
121 	struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
122 	u_short	lport = 0;
123 	int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
124 
125 	if (!in6_ifaddr) /* 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 		wild = 1;
131 	if (nam) {
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 		/* KAME hack: embed scopeid */
142 		if (in6_embedscope(&sin6->sin6_addr, sin6, inp, NULL) != 0)
143 			return EINVAL;
144 		/* this must be cleared for ifa_ifwithaddr() */
145 		sin6->sin6_scope_id = 0;
146 
147 		lport = sin6->sin6_port;
148 		if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
149 			/*
150 			 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
151 			 * allow compepte duplication of binding if
152 			 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
153 			 * and a multicast address is bound on both
154 			 * new and duplicated sockets.
155 			 */
156 			if (so->so_options & SO_REUSEADDR)
157 				reuseport = SO_REUSEADDR|SO_REUSEPORT;
158 		} else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
159 			struct ifaddr *ia = NULL;
160 
161 			sin6->sin6_port = 0;		/* yech... */
162 			if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0)
163 				return(EADDRNOTAVAIL);
164 
165 			/*
166 			 * XXX: bind to an anycast address might accidentally
167 			 * cause sending a packet with anycast source address.
168 			 */
169 			if (ia &&
170 			    ((struct in6_ifaddr *)ia)->ia6_flags &
171 			    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
172 			     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
173 				return(EADDRNOTAVAIL);
174 			}
175 		}
176 		if (lport) {
177 			struct inpcb *t;
178 
179 			/* GROSS */
180 			if (ntohs(lport) < IPV6PORT_RESERVED && p &&
181 			    suser_xxx(0, p, PRISON_ROOT))
182 				return(EACCES);
183 			if (so->so_cred->cr_uid != 0 &&
184 			    !IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
185 				t = in6_pcblookup_local(pcbinfo,
186 				    &sin6->sin6_addr, lport,
187 				    INPLOOKUP_WILDCARD);
188 				if (t &&
189 				    (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
190 				     !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
191 				     (t->inp_socket->so_options &
192 				      SO_REUSEPORT) == 0) &&
193 				    (so->so_cred->cr_uid !=
194 				     t->inp_socket->so_cred->cr_uid))
195 					return (EADDRINUSE);
196 				if (ip6_mapped_addr_on != 0 &&
197 				    IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
198 					struct sockaddr_in sin;
199 
200 					in6_sin6_2_sin(&sin, sin6);
201 					t = in_pcblookup_local(pcbinfo,
202 						sin.sin_addr, lport,
203 						INPLOOKUP_WILDCARD);
204 					if (t &&
205 					    (so->so_cred->cr_uid !=
206 					     t->inp_socket->so_cred->cr_uid) &&
207 					    (ntohl(t->inp_laddr.s_addr) !=
208 					     INADDR_ANY ||
209 					     INP_SOCKAF(so) ==
210 					     INP_SOCKAF(t->inp_socket)))
211 						return (EADDRINUSE);
212 				}
213 			}
214 			t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr,
215 						lport, wild);
216 			if (t && (reuseport & t->inp_socket->so_options) == 0)
217 				return(EADDRINUSE);
218 			if (ip6_mapped_addr_on != 0 &&
219 			    IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
220 				struct sockaddr_in sin;
221 
222 				in6_sin6_2_sin(&sin, sin6);
223 				t = in_pcblookup_local(pcbinfo, sin.sin_addr,
224 						       lport, wild);
225 				if (t &&
226 				    (reuseport & t->inp_socket->so_options)
227 				    == 0 &&
228 				    (ntohl(t->inp_laddr.s_addr)
229 				     != INADDR_ANY ||
230 				     INP_SOCKAF(so) ==
231 				     INP_SOCKAF(t->inp_socket)))
232 					return (EADDRINUSE);
233 			}
234 		}
235 		inp->in6p_laddr = sin6->sin6_addr;
236 	}
237 	if (lport == 0) {
238 		int e;
239 		if ((e = in6_pcbsetport(&inp->in6p_laddr, inp, p)) != 0)
240 			return(e);
241 	}
242 	else {
243 		inp->inp_lport = lport;
244 		if (in_pcbinshash(inp) != 0) {
245 			inp->in6p_laddr = in6addr_any;
246 			inp->inp_lport = 0;
247 			return (EAGAIN);
248 		}
249 	}
250 	inp->in6p_flowinfo = sin6 ? sin6->sin6_flowinfo : 0;	/*XXX*/
251 	return(0);
252 }
253 
254 /*
255  *   Transform old in6_pcbconnect() into an inner subroutine for new
256  *   in6_pcbconnect(): Do some validity-checking on the remote
257  *   address (in mbuf 'nam') and then determine local host address
258  *   (i.e., which interface) to use to access that remote host.
259  *
260  *   This preserves definition of in6_pcbconnect(), while supporting a
261  *   slightly different version for T/TCP.  (This is more than
262  *   a bit of a kludge, but cleaning up the internal interfaces would
263  *   have forced minor changes in every protocol).
264  */
265 
266 int
267 in6_pcbladdr(inp, nam, plocal_addr6)
268 	register struct inpcb *inp;
269 	struct sockaddr *nam;
270 	struct in6_addr **plocal_addr6;
271 {
272 	register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
273 	struct in6_pktinfo *pi;
274 	struct ifnet *ifp = NULL;
275 	int error = 0;
276 
277 	if (nam->sa_len != sizeof (*sin6))
278 		return (EINVAL);
279 	if (sin6->sin6_family != AF_INET6)
280 		return (EAFNOSUPPORT);
281 	if (sin6->sin6_port == 0)
282 		return (EADDRNOTAVAIL);
283 
284 	/* KAME hack: embed scopeid */
285 	if (in6_embedscope(&sin6->sin6_addr, sin6, inp, &ifp) != 0)
286 		return EINVAL;
287 
288 	if (in6_ifaddr) {
289 		/*
290 		 * If the destination address is UNSPECIFIED addr,
291 		 * use the loopback addr, e.g ::1.
292 		 */
293 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
294 			sin6->sin6_addr = in6addr_loopback;
295 	}
296 	{
297 		/*
298 		 * XXX: in6_selectsrc might replace the bound local address
299 		 * with the address specified by setsockopt(IPV6_PKTINFO).
300 		 * Is it the intended behavior?
301 		 */
302 		*plocal_addr6 = in6_selectsrc(sin6, inp->in6p_outputopts,
303 					      inp->in6p_moptions,
304 					      &inp->in6p_route,
305 					      &inp->in6p_laddr, &error);
306 		if (*plocal_addr6 == 0) {
307 			if (error == 0)
308 				error = EADDRNOTAVAIL;
309 			return(error);
310 		}
311 		/*
312 		 * Don't do pcblookup call here; return interface in
313 		 * plocal_addr6
314 		 * and exit to caller, that will do the lookup.
315 		 */
316 	}
317 
318 	if (inp->in6p_route.ro_rt)
319 		ifp = inp->in6p_route.ro_rt->rt_ifp;
320 
321 	return(0);
322 }
323 
324 /*
325  * Outer subroutine:
326  * Connect from a socket to a specified address.
327  * Both address and port must be specified in argument sin.
328  * If don't have a local address for this socket yet,
329  * then pick one.
330  */
331 int
332 in6_pcbconnect(inp, nam, p)
333 	register struct inpcb *inp;
334 	struct sockaddr *nam;
335 	struct proc *p;
336 {
337 	struct in6_addr *addr6;
338 	register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
339 	int error;
340 
341 	/*
342 	 *   Call inner routine, to assign local interface address.
343 	 */
344 	if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0)
345 		return(error);
346 
347 	if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr,
348 			       sin6->sin6_port,
349 			      IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
350 			      ? addr6 : &inp->in6p_laddr,
351 			      inp->inp_lport, 0, NULL) != NULL) {
352 		return (EADDRINUSE);
353 	}
354 	if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
355 		if (inp->inp_lport == 0) {
356 			error = in6_pcbbind(inp, (struct sockaddr *)0, p);
357 			if (error)
358 				return (error);
359 		}
360 		inp->in6p_laddr = *addr6;
361 	}
362 	inp->in6p_faddr = sin6->sin6_addr;
363 	inp->inp_fport = sin6->sin6_port;
364 	/*
365 	 * xxx kazu flowlabel is necessary for connect?
366 	 * but if this line is missing, the garbage value remains.
367 	 */
368 	inp->in6p_flowinfo = sin6->sin6_flowinfo;
369 	if ((inp->in6p_flowinfo & IPV6_FLOWLABEL_MASK) == 0 &&
370 	    ip6_auto_flowlabel != 0)
371 		inp->in6p_flowinfo |=
372 			(htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK);
373 
374 	in_pcbrehash(inp);
375 	return (0);
376 }
377 
378 #if 0
379 /*
380  * Return an IPv6 address, which is the most appropriate for given
381  * destination and user specified options.
382  * If necessary, this function lookups the routing table and return
383  * an entry to the caller for later use.
384  */
385 struct in6_addr *
386 in6_selectsrc(dstsock, opts, mopts, ro, laddr, errorp)
387 	struct sockaddr_in6 *dstsock;
388 	struct ip6_pktopts *opts;
389 	struct ip6_moptions *mopts;
390 	struct route_in6 *ro;
391 	struct in6_addr *laddr;
392 	int *errorp;
393 {
394 	struct in6_addr *dst;
395 	struct in6_ifaddr *ia6 = 0;
396 	struct in6_pktinfo *pi = NULL;
397 
398 	dst = &dstsock->sin6_addr;
399 	*errorp = 0;
400 
401 	/*
402 	 * If the source address is explicitly specified by the caller,
403 	 * use it.
404 	 */
405 	if (opts && (pi = opts->ip6po_pktinfo) &&
406 	    !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr))
407 		return(&pi->ipi6_addr);
408 
409 	/*
410 	 * If the source address is not specified but the socket(if any)
411 	 * is already bound, use the bound address.
412 	 */
413 	if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr))
414 		return(laddr);
415 
416 	/*
417 	 * If the caller doesn't specify the source address but
418 	 * the outgoing interface, use an address associated with
419 	 * the interface.
420 	 */
421 	if (pi && pi->ipi6_ifindex) {
422 		/* XXX boundary check is assumed to be already done. */
423 		ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex],
424 				       dst);
425 		if (ia6 == 0) {
426 			*errorp = EADDRNOTAVAIL;
427 			return(0);
428 		}
429 		return(&satosin6(&ia6->ia_addr)->sin6_addr);
430 	}
431 
432 	/*
433 	 * If the destination address is a link-local unicast address or
434 	 * a multicast address, and if the outgoing interface is specified
435 	 * by the sin6_scope_id filed, use an address associated with the
436 	 * interface.
437 	 * XXX: We're now trying to define more specific semantics of
438 	 *      sin6_scope_id field, so this part will be rewritten in
439 	 *      the near future.
440 	 */
441 	if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) &&
442 	    dstsock->sin6_scope_id) {
443 		/*
444 		 * I'm not sure if boundary check for scope_id is done
445 		 * somewhere...
446 		 */
447 		if (dstsock->sin6_scope_id < 0 ||
448 		    if_index < dstsock->sin6_scope_id) {
449 			*errorp = ENXIO; /* XXX: better error? */
450 			return(0);
451 		}
452 		ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id],
453 				       dst);
454 		if (ia6 == 0) {
455 			*errorp = EADDRNOTAVAIL;
456 			return(0);
457 		}
458 		return(&satosin6(&ia6->ia_addr)->sin6_addr);
459 	}
460 
461 	/*
462 	 * If the destination address is a multicast address and
463 	 * the outgoing interface for the address is specified
464 	 * by the caller, use an address associated with the interface.
465 	 * There is a sanity check here; if the destination has node-local
466 	 * scope, the outgoing interfacde should be a loopback address.
467 	 * Even if the outgoing interface is not specified, we also
468 	 * choose a loopback interface as the outgoing interface.
469 	 */
470 	if (IN6_IS_ADDR_MULTICAST(dst)) {
471 		struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL;
472 
473 		if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) {
474 			ifp = &loif[0];
475 		}
476 
477 		if (ifp) {
478 			ia6 = in6_ifawithscope(ifp, dst);
479 			if (ia6 == 0) {
480 				*errorp = EADDRNOTAVAIL;
481 				return(0);
482 			}
483 			return(&ia6->ia_addr.sin6_addr);
484 		}
485 	}
486 
487 	/*
488 	 * If the next hop address for the packet is specified
489 	 * by caller, use an address associated with the route
490 	 * to the next hop.
491 	 */
492 	{
493 		struct sockaddr_in6 *sin6_next;
494 		struct rtentry *rt;
495 
496 		if (opts && opts->ip6po_nexthop) {
497 			sin6_next = satosin6(opts->ip6po_nexthop);
498 			rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL);
499 			if (rt) {
500 				ia6 = in6_ifawithscope(rt->rt_ifp, dst);
501 				if (ia6 == 0)
502 					ia6 = ifatoia6(rt->rt_ifa);
503 			}
504 			if (ia6 == 0) {
505 				*errorp = EADDRNOTAVAIL;
506 				return(0);
507 			}
508 			return(&satosin6(&ia6->ia_addr)->sin6_addr);
509 		}
510 	}
511 
512 	/*
513 	 * If route is known or can be allocated now,
514 	 * our src addr is taken from the i/f, else punt.
515 	 */
516 	if (ro) {
517 		if (ro->ro_rt &&
518 		    !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, dst)) {
519 			RTFREE(ro->ro_rt);
520 			ro->ro_rt = (struct rtentry *)0;
521 		}
522 		if (ro->ro_rt == (struct rtentry *)0 ||
523 		    ro->ro_rt->rt_ifp == (struct ifnet *)0) {
524 			/* No route yet, so try to acquire one */
525 			bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
526 			ro->ro_dst.sin6_family = AF_INET6;
527 			ro->ro_dst.sin6_len = sizeof(struct sockaddr_in6);
528 			ro->ro_dst.sin6_addr = *dst;
529 			if (IN6_IS_ADDR_MULTICAST(dst)) {
530 				ro->ro_rt = rtalloc1(&((struct route *)ro)
531 						     ->ro_dst, 0, 0UL);
532 			} else {
533 				rtalloc((struct route *)ro);
534 			}
535 		}
536 
537 		/*
538 		 * in_pcbconnect() checks out IFF_LOOPBACK to skip using
539 		 * the address. But we don't know why it does so.
540 		 * It is necessary to ensure the scope even for lo0
541 		 * so doesn't check out IFF_LOOPBACK.
542 		 */
543 
544 		if (ro->ro_rt) {
545 			ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst);
546 			if (ia6 == 0) /* xxx scope error ?*/
547 				ia6 = ifatoia6(ro->ro_rt->rt_ifa);
548 		}
549 		if (ia6 == 0) {
550 			*errorp = EHOSTUNREACH;	/* no route */
551 			return(0);
552 		}
553 		return(&satosin6(&ia6->ia_addr)->sin6_addr);
554 	}
555 
556 	*errorp = EADDRNOTAVAIL;
557 	return(0);
558 }
559 
560 /*
561  * Default hop limit selection. The precedence is as follows:
562  * 1. Hoplimit valued specified via ioctl.
563  * 2. (If the outgoing interface is detected) the current
564  *     hop limit of the interface specified by router advertisement.
565  * 3. The system default hoplimit.
566 */
567 int
568 in6_selecthlim(in6p, ifp)
569 	struct in6pcb *in6p;
570 	struct ifnet *ifp;
571 {
572 	if (in6p && in6p->in6p_hops >= 0)
573 		return(in6p->in6p_hops);
574 	else if (ifp)
575 		return(nd_ifinfo[ifp->if_index].chlim);
576 	else
577 		return(ip6_defhlim);
578 }
579 #endif
580 
581 void
582 in6_pcbdisconnect(inp)
583 	struct inpcb *inp;
584 {
585 	bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
586 	inp->inp_fport = 0;
587 	in_pcbrehash(inp);
588 	if (inp->inp_socket->so_state & SS_NOFDREF)
589 		in6_pcbdetach(inp);
590 }
591 
592 void
593 in6_pcbdetach(inp)
594 	struct inpcb *inp;
595 {
596 	struct socket *so = inp->inp_socket;
597 	struct inpcbinfo *ipi = inp->inp_pcbinfo;
598 
599 #ifdef IPSEC
600 	if (inp->in6p_sp != NULL)
601 		ipsec6_delete_pcbpolicy(inp);
602 #endif /* IPSEC */
603 	inp->inp_gencnt = ++ipi->ipi_gencnt;
604 	in_pcbremlists(inp);
605 	sotoinpcb(so) = 0;
606 	sofree(so);
607 	if (inp->in6p_options)
608 		m_freem(inp->in6p_options);
609 	if (inp->in6p_outputopts) {
610 		if (inp->in6p_outputopts->ip6po_rthdr &&
611 		    inp->in6p_outputopts->ip6po_route.ro_rt)
612 			RTFREE(inp->in6p_outputopts->ip6po_route.ro_rt);
613 		if (inp->in6p_outputopts->ip6po_m)
614 			(void)m_free(inp->in6p_outputopts->ip6po_m);
615 		free(inp->in6p_outputopts, M_IP6OPT);
616 	}
617 	if (inp->in6p_route.ro_rt)
618 		rtfree(inp->in6p_route.ro_rt);
619 	ip6_freemoptions(inp->in6p_moptions);
620 
621 	/* Check and free IPv4 related resources in case of mapped addr */
622 	if (inp->inp_options)
623 		(void)m_free(inp->inp_options);
624 	ip_freemoptions(inp->inp_moptions);
625 
626 	inp->inp_vflag = 0;
627 	zfreei(ipi->ipi_zone, inp);
628 }
629 
630 /*
631  * The calling convention of in6_setsockaddr() and in6_setpeeraddr() was
632  * modified to match the pru_sockaddr() and pru_peeraddr() entry points
633  * in struct pr_usrreqs, so that protocols can just reference then directly
634  * without the need for a wrapper function.  The socket must have a valid
635  * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
636  * except through a kernel programming error, so it is acceptable to panic
637  * (or in this case trap) if the PCB is invalid.  (Actually, we don't trap
638  * because there actually /is/ a programming error somewhere... XXX)
639  */
640 int
641 in6_setsockaddr(so, nam)
642 	struct socket *so;
643 	struct sockaddr **nam;
644 {
645 	int s;
646 	register struct inpcb *inp;
647 	register struct sockaddr_in6 *sin6;
648 
649 	/*
650 	 * Do the malloc first in case it blocks.
651 	 */
652 	MALLOC(sin6, struct sockaddr_in6 *, sizeof *sin6, M_SONAME, M_WAITOK);
653 	bzero(sin6, sizeof *sin6);
654 	sin6->sin6_family = AF_INET6;
655 	sin6->sin6_len = sizeof(*sin6);
656 
657 	s = splnet();
658 	inp = sotoinpcb(so);
659 	if (!inp) {
660 		splx(s);
661 		free(sin6, M_SONAME);
662 		return EINVAL;
663 	}
664 	sin6->sin6_port = inp->inp_lport;
665 	sin6->sin6_addr = inp->in6p_laddr;
666 	splx(s);
667 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
668 		sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
669 	else
670 		sin6->sin6_scope_id = 0;	/*XXX*/
671 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
672 		sin6->sin6_addr.s6_addr16[1] = 0;
673 
674 	*nam = (struct sockaddr *)sin6;
675 	return 0;
676 }
677 
678 int
679 in6_setpeeraddr(so, nam)
680 	struct socket *so;
681 	struct sockaddr **nam;
682 {
683 	int s;
684 	struct inpcb *inp;
685 	register struct sockaddr_in6 *sin6;
686 
687 	/*
688 	 * Do the malloc first in case it blocks.
689 	 */
690 	MALLOC(sin6, struct sockaddr_in6 *, sizeof(*sin6), M_SONAME, M_WAITOK);
691 	bzero((caddr_t)sin6, sizeof (*sin6));
692 	sin6->sin6_family = AF_INET6;
693 	sin6->sin6_len = sizeof(struct sockaddr_in6);
694 
695 	s = splnet();
696 	inp = sotoinpcb(so);
697 	if (!inp) {
698 		splx(s);
699 		free(sin6, M_SONAME);
700 		return EINVAL;
701 	}
702 	sin6->sin6_port = inp->inp_fport;
703 	sin6->sin6_addr = inp->in6p_faddr;
704 	splx(s);
705 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
706 		sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
707 	else
708 		sin6->sin6_scope_id = 0;	/*XXX*/
709 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
710 		sin6->sin6_addr.s6_addr16[1] = 0;
711 
712 	*nam = (struct sockaddr *)sin6;
713 	return 0;
714 }
715 
716 int
717 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
718 {
719 	struct	inpcb *inp = sotoinpcb(so);
720 	int	error;
721 
722 	if (inp == NULL)
723 		return EINVAL;
724 	if (inp->inp_vflag & INP_IPV4) {
725 		error = in_setsockaddr(so, nam);
726 		if (error == 0)
727 			in6_sin_2_v4mapsin6_in_sock(nam);
728 	} else
729 	error = in6_setsockaddr(so, nam);
730 
731 	return error;
732 }
733 
734 int
735 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
736 {
737 	struct	inpcb *inp = sotoinpcb(so);
738 	int	error;
739 
740 	if (inp == NULL)
741 		return EINVAL;
742 	if (inp->inp_vflag & INP_IPV4) {
743 		error = in_setpeeraddr(so, nam);
744 		if (error == 0)
745 			in6_sin_2_v4mapsin6_in_sock(nam);
746 	} else
747 	error = in6_setpeeraddr(so, nam);
748 
749 	return error;
750 }
751 
752 /*
753  * Pass some notification to all connections of a protocol
754  * associated with address dst.  The local address and/or port numbers
755  * may be specified to limit the search.  The "usual action" will be
756  * taken, depending on the ctlinput cmd.  The caller must filter any
757  * cmds that are uninteresting (e.g., no error in the map).
758  * Call the protocol specific routine (if any) to report
759  * any errors for each matching socket.
760  *
761  * Must be called at splnet.
762  */
763 void
764 in6_pcbnotify(head, dst, fport_arg, laddr6, lport_arg, cmd, notify)
765 	struct inpcbhead *head;
766 	struct sockaddr *dst;
767 	u_int fport_arg, lport_arg;
768 	struct in6_addr *laddr6;
769 	int cmd;
770 	void (*notify) __P((struct inpcb *, int));
771 {
772 	struct inpcb *inp, *ninp;
773 	struct in6_addr faddr6;
774 	u_short	fport = fport_arg, lport = lport_arg;
775 	int errno, s;
776 	int do_rtchange = (notify == in6_rtchange);
777 
778 	if ((unsigned)cmd > PRC_NCMDS || dst->sa_family != AF_INET6)
779 		return;
780 	faddr6 = ((struct sockaddr_in6 *)dst)->sin6_addr;
781 	if (IN6_IS_ADDR_UNSPECIFIED(&faddr6))
782 		return;
783 
784 	/*
785 	 * Redirects go to all references to the destination,
786 	 * and use in6_rtchange to invalidate the route cache.
787 	 * Dead host indications: also use in6_rtchange to invalidate
788 	 * the cache, and deliver the error to all the sockets.
789 	 * Otherwise, if we have knowledge of the local port and address,
790 	 * deliver only to that socket.
791 	 */
792 	if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
793 		fport = 0;
794 		lport = 0;
795 		bzero((caddr_t)laddr6, sizeof(*laddr6));
796 
797 		do_rtchange = 1;
798 	}
799 	errno = inet6ctlerrmap[cmd];
800 	s = splnet();
801  	for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) {
802  		ninp = LIST_NEXT(inp, inp_list);
803 
804  		if ((inp->inp_vflag & INP_IPV6) == NULL)
805 			continue;
806 
807  		if (do_rtchange) {
808  			/*
809  			 * Since a non-connected PCB might have a cached route,
810  			 * we always call in6_rtchange without matching
811  			 * the PCB to the src/dst pair.
812  			 *
813  			 * XXX: we assume in6_rtchange does not free the PCB.
814  			 */
815  			if (IN6_ARE_ADDR_EQUAL(&inp->in6p_route.ro_dst.sin6_addr,
816  					       &faddr6))
817  				in6_rtchange(inp, errno);
818 
819  			if (notify == in6_rtchange)
820  				continue; /* there's nothing to do any more */
821   		}
822 
823 		if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &faddr6) ||
824 		   inp->inp_socket == 0 ||
825 		   (lport && inp->inp_lport != lport) ||
826 		   (!IN6_IS_ADDR_UNSPECIFIED(laddr6) &&
827 		    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr6)) ||
828 		    (fport && inp->inp_fport != fport))
829 			continue;
830 
831 		if (notify)
832  			(*notify)(inp, errno);
833 	}
834 	splx(s);
835 }
836 
837 /*
838  * Lookup a PCB based on the local address and port.
839  */
840 struct inpcb *
841 in6_pcblookup_local(pcbinfo, laddr, lport_arg, wild_okay)
842 	struct inpcbinfo *pcbinfo;
843 	struct in6_addr *laddr;
844 	u_int lport_arg;
845 	int wild_okay;
846 {
847 	register struct inpcb *inp;
848 	int matchwild = 3, wildcard;
849 	u_short lport = lport_arg;
850 
851 	if (!wild_okay) {
852 		struct inpcbhead *head;
853 		/*
854 		 * Look for an unconnected (wildcard foreign addr) PCB that
855 		 * matches the local address and port we're looking for.
856 		 */
857 		head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0,
858 						      pcbinfo->hashmask)];
859 		LIST_FOREACH(inp, head, inp_hash) {
860 			if ((inp->inp_vflag & INP_IPV6) == 0)
861 				continue;
862 			if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
863 			    IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
864 			    inp->inp_lport == lport) {
865 				/*
866 				 * Found.
867 				 */
868 				return (inp);
869 			}
870 		}
871 		/*
872 		 * Not found.
873 		 */
874 		return (NULL);
875 	} else {
876 		struct inpcbporthead *porthash;
877 		struct inpcbport *phd;
878 		struct inpcb *match = NULL;
879 		/*
880 		 * Best fit PCB lookup.
881 		 *
882 		 * First see if this local port is in use by looking on the
883 		 * port hash list.
884 		 */
885 		porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport,
886 		    pcbinfo->porthashmask)];
887 		LIST_FOREACH(phd, porthash, phd_hash) {
888 			if (phd->phd_port == lport)
889 				break;
890 		}
891 		if (phd != NULL) {
892 			/*
893 			 * Port is in use by one or more PCBs. Look for best
894 			 * fit.
895 			 */
896 			LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
897 				wildcard = 0;
898 				if ((inp->inp_vflag & INP_IPV6) == 0)
899 					continue;
900 				if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
901 					wildcard++;
902 				if (!IN6_IS_ADDR_UNSPECIFIED(
903 					&inp->in6p_laddr)) {
904 					if (IN6_IS_ADDR_UNSPECIFIED(laddr))
905 						wildcard++;
906 					else if (!IN6_ARE_ADDR_EQUAL(
907 						&inp->in6p_laddr, laddr))
908 						continue;
909 				} else {
910 					if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
911 						wildcard++;
912 				}
913 				if (wildcard < matchwild) {
914 					match = inp;
915 					matchwild = wildcard;
916 					if (matchwild == 0) {
917 						break;
918 					}
919 				}
920 			}
921 		}
922 		return (match);
923 	}
924 }
925 
926 /*
927  * Check for alternatives when higher level complains
928  * about service problems.  For now, invalidate cached
929  * routing information.  If the route was created dynamically
930  * (by a redirect), time to try a default gateway again.
931  */
932 void
933 in6_losing(in6p)
934 	struct inpcb *in6p;
935 {
936 	struct rtentry *rt;
937 	struct rt_addrinfo info;
938 
939 	if ((rt = in6p->in6p_route.ro_rt) != NULL) {
940 		in6p->in6p_route.ro_rt = 0;
941 		bzero((caddr_t)&info, sizeof(info));
942 		info.rti_info[RTAX_DST] =
943 			(struct sockaddr *)&in6p->in6p_route.ro_dst;
944 		info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
945 		info.rti_info[RTAX_NETMASK] = rt_mask(rt);
946 		rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
947 		if (rt->rt_flags & RTF_DYNAMIC)
948 			(void)rtrequest(RTM_DELETE, rt_key(rt),
949 					rt->rt_gateway, rt_mask(rt), rt->rt_flags,
950 					(struct rtentry **)0);
951 		else
952 		/*
953 		 * A new route can be allocated
954 		 * the next time output is attempted.
955 		 */
956 			rtfree(rt);
957 	}
958 }
959 
960 /*
961  * After a routing change, flush old routing
962  * and allocate a (hopefully) better one.
963  */
964 void
965 in6_rtchange(inp, errno)
966 	struct inpcb *inp;
967 	int errno;
968 {
969 	if (inp->in6p_route.ro_rt) {
970 		rtfree(inp->in6p_route.ro_rt);
971 		inp->in6p_route.ro_rt = 0;
972 		/*
973 		 * A new route can be allocated the next time
974 		 * output is attempted.
975 		 */
976 	}
977 }
978 
979 /*
980  * Lookup PCB in hash list.
981  */
982 struct inpcb *
983 in6_pcblookup_hash(pcbinfo, faddr, fport_arg, laddr, lport_arg, wildcard, ifp)
984 	struct inpcbinfo *pcbinfo;
985 	struct in6_addr *faddr, *laddr;
986 	u_int fport_arg, lport_arg;
987 	int wildcard;
988 	struct ifnet *ifp;
989 {
990 	struct inpcbhead *head;
991 	register struct inpcb *inp;
992 	u_short fport = fport_arg, lport = lport_arg;
993 
994 	/*
995 	 * First look for an exact match.
996 	 */
997 	head = &pcbinfo->hashbase[INP_PCBHASH(faddr->s6_addr32[3] /* XXX */,
998 					      lport, fport,
999 					      pcbinfo->hashmask)];
1000 	LIST_FOREACH(inp, head, inp_hash) {
1001 		if ((inp->inp_vflag & INP_IPV6) == 0)
1002 			continue;
1003 		if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1004 		    IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1005 		    inp->inp_fport == fport &&
1006 		    inp->inp_lport == lport) {
1007 			/*
1008 			 * Found.
1009 			 */
1010 			return (inp);
1011 		}
1012 	}
1013 	if (wildcard) {
1014 		struct inpcb *local_wild = NULL;
1015 
1016 		head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0,
1017 						      pcbinfo->hashmask)];
1018 		LIST_FOREACH(inp, head, inp_hash) {
1019 			if ((inp->inp_vflag & INP_IPV6) == 0)
1020 				continue;
1021 			if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1022 			    inp->inp_lport == lport) {
1023 #if defined(NFAITH) && NFAITH > 0
1024 				if (ifp && ifp->if_type == IFT_FAITH &&
1025 				    (inp->inp_flags & INP_FAITH) == 0)
1026 					continue;
1027 #endif
1028 				if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
1029 						       laddr))
1030 					return (inp);
1031 				else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
1032 					local_wild = inp;
1033 			}
1034 		}
1035 		return (local_wild);
1036 	}
1037 
1038 	/*
1039 	 * Not found.
1040 	 */
1041 	return (NULL);
1042 }
1043 
1044 void
1045 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
1046 {
1047 	struct ip6_hdr *ip;
1048 
1049 	ip = mtod(m, struct ip6_hdr *);
1050 	bzero(sin6, sizeof(*sin6));
1051 	sin6->sin6_len = sizeof(*sin6);
1052 	sin6->sin6_family = AF_INET6;
1053 	sin6->sin6_addr = ip->ip6_src;
1054 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1055 		sin6->sin6_addr.s6_addr16[1] = 0;
1056 	sin6->sin6_scope_id =
1057 		(m->m_pkthdr.rcvif && IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1058 		? m->m_pkthdr.rcvif->if_index : 0;
1059 
1060 	return;
1061 }
1062