xref: /freebsd/sys/netinet6/in6_pcb.c (revision 5129159789cc9d7bc514e4546b88e3427695002d)
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  * $FreeBSD$
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  * 3. All advertising materials mentioning features or use of this software
45  *    must display the following acknowledgement:
46  *	This product includes software developed by the University of
47  *	California, Berkeley and its contributors.
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  * $FreeBSD$
66  */
67 
68 #include "opt_key.h"
69 
70 #include <sys/param.h>
71 #include <sys/systm.h>
72 #include <sys/malloc.h>
73 #include <sys/mbuf.h>
74 #include <sys/protosw.h>
75 #include <sys/socket.h>
76 #include <sys/socketvar.h>
77 #include <sys/sockio.h>
78 #include <sys/errno.h>
79 #include <sys/time.h>
80 #include <sys/proc.h>
81 #include <sys/jail.h>
82 
83 #include <vm/vm_zone.h>
84 
85 #include <net/if.h>
86 #include <net/if_types.h>
87 #include <net/route.h>
88 
89 #include <netinet/in.h>
90 #include <netinet/in_var.h>
91 #include <netinet/in_systm.h>
92 #include <netinet6/ip6.h>
93 #include <netinet6/ip6_var.h>
94 #include <netinet6/nd6.h>
95 #include <netinet/in_pcb.h>
96 #include <netinet6/in6_pcb.h>
97 
98 #include "faith.h"
99 
100 #ifdef IPSEC
101 #include <netinet6/ipsec.h>
102 #include <netinet6/ipsec6.h>
103 #include <netkey/key.h>
104 #ifdef KEY_DEBUG
105 #include <netkey/key_debug.h>
106 #else
107 #define DPRINTF(lev,arg)
108 #define DDO(lev, stmt)
109 #define DP(x, y, z)
110 #endif /* KEY_DEBUG */
111 #endif /* IPSEC */
112 
113 struct	in6_addr zeroin6_addr;
114 
115 int
116 in6_pcbbind(inp, nam, p)
117 	register struct inpcb *inp;
118 	struct sockaddr *nam;
119 	struct proc *p;
120 {
121 	struct socket *so = inp->inp_socket;
122 	unsigned short *lastport;
123 	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
124 	struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
125 	u_short	lport = 0;
126 	int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
127 	int error;
128 
129 	if (!in6_ifaddr) /* XXX broken! */
130 		return (EADDRNOTAVAIL);
131 	if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
132 		return(EINVAL);
133 	if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
134 		wild = 1;
135 	if (nam) {
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 		/*
146 		 * If the scope of the destination is link-local, embed the
147 		 * interface index in the address.
148 		 */
149 		if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) {
150 			/* XXX boundary check is assumed to be already done. */
151 			/* XXX sin6_scope_id is weaker than advanced-api. */
152 			struct in6_pktinfo *pi;
153 			if (inp->in6p_outputopts &&
154 			    (pi = inp->in6p_outputopts->ip6po_pktinfo) &&
155 			    pi->ipi6_ifindex) {
156 				sin6->sin6_addr.s6_addr16[1]
157 					= htons(pi->ipi6_ifindex);
158 			} else if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)
159 				&& inp->in6p_moptions
160 				&& inp->in6p_moptions->im6o_multicast_ifp) {
161 				sin6->sin6_addr.s6_addr16[1] =
162 					htons(inp->in6p_moptions->im6o_multicast_ifp->if_index);
163 			} else if (sin6->sin6_scope_id) {
164 				/* boundary check */
165 				if (sin6->sin6_scope_id < 0
166 				 || if_index < sin6->sin6_scope_id) {
167 					return ENXIO;  /* XXX EINVAL? */
168 				}
169 				sin6->sin6_addr.s6_addr16[1]
170 					= htons(sin6->sin6_scope_id & 0xffff);/*XXX*/
171 				/* this must be cleared for ifa_ifwithaddr() */
172 				sin6->sin6_scope_id = 0;
173 			}
174 		}
175 
176 		lport = sin6->sin6_port;
177 		if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
178 			/*
179 			 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
180 			 * allow compepte duplication of binding if
181 			 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
182 			 * and a multicast address is bound on both
183 			 * new and duplicated sockets.
184 			 */
185 			if (so->so_options & SO_REUSEADDR)
186 				reuseport = SO_REUSEADDR|SO_REUSEPORT;
187 		} else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
188 			struct ifaddr *ia = NULL;
189 
190 			sin6->sin6_port = 0;		/* yech... */
191 			if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == 0)
192 				return(EADDRNOTAVAIL);
193 
194 			/*
195 			 * XXX: bind to an anycast address might accidentally
196 			 * cause sending a packet with anycast source address.
197 			 */
198 			if (ia &&
199 			    ((struct in6_ifaddr *)ia)->ia6_flags &
200 			    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
201 			     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
202 				return(EADDRNOTAVAIL);
203 			}
204 		}
205 		if (lport) {
206 			struct inpcb *t;
207 
208 			/* GROSS */
209 			if (ntohs(lport) < IPV6PORT_RESERVED && p &&
210 			    suser_xxx(0, p, PRISON_ROOT))
211 				return(EACCES);
212 			if (so->so_cred->cr_uid != 0 &&
213 			    !IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
214 				t = in6_pcblookup_local(inp->inp_pcbinfo,
215 				    &sin6->sin6_addr, lport,
216 				    INPLOOKUP_WILDCARD);
217 				if (t &&
218 				    (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
219 				     !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
220 				     (t->inp_socket->so_options &
221 				      SO_REUSEPORT) == 0) &&
222 				    (so->so_cred->cr_uid !=
223 				     t->inp_socket->so_cred->cr_uid))
224 					return (EADDRINUSE);
225 			}
226 			t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr,
227 						lport, wild);
228 			if (t && (reuseport & t->inp_socket->so_options) == 0)
229 				return(EADDRINUSE);
230 		}
231 		inp->in6p_laddr = sin6->sin6_addr;
232 	}
233 	if (lport == 0) {
234 		ushort first, last;
235 		int count;
236 
237 		inp->inp_flags |= INP_ANONPORT;
238 
239 		if (inp->inp_flags & INP_HIGHPORT) {
240 			first = ipport_hifirstauto;	/* sysctl */
241 			last  = ipport_hilastauto;
242 			lastport = &pcbinfo->lasthi;
243 		} else if (inp->inp_flags & INP_LOWPORT) {
244 			if (p && (error = suser_xxx(0, p, PRISON_ROOT)))
245 				return error;
246 			first = ipport_lowfirstauto;	/* 1023 */
247 			last  = ipport_lowlastauto;	/* 600 */
248 			lastport = &pcbinfo->lastlow;
249 		} else {
250 			first = ipport_firstauto;	/* sysctl */
251 			last  = ipport_lastauto;
252 			lastport = &pcbinfo->lastport;
253 		}
254 		/*
255 		 * Simple check to ensure all ports are not used up causing
256 		 * a deadlock here.
257 		 *
258 		 * We split the two cases (up and down) so that the direction
259 		 * is not being tested on each round of the loop.
260 		 */
261 		if (first > last) {
262 			/*
263 			 * counting down
264 			 */
265 			count = first - last;
266 
267 			do {
268 				if (count-- < 0) {	/* completely used? */
269 					/*
270 					 * Undo any address bind that may have
271 					 * occurred above.
272 					 */
273 					inp->in6p_laddr = in6addr_any;
274 					return (EAGAIN);
275 				}
276 				--*lastport;
277 				if (*lastport > first || *lastport < last)
278 					*lastport = first;
279 				lport = htons(*lastport);
280 			} while (in6_pcblookup_local(pcbinfo,
281 				 &inp->in6p_laddr, lport, wild));
282 		} else {
283 			/*
284 			 * counting up
285 			 */
286 			count = last - first;
287 
288 			do {
289 				if (count-- < 0) {	/* completely used? */
290 					/*
291 					 * Undo any address bind that may have
292 					 * occurred above.
293 					 */
294 					inp->in6p_laddr = in6addr_any;
295 					return (EAGAIN);
296 				}
297 				++*lastport;
298 				if (*lastport < first || *lastport > last)
299 					*lastport = first;
300 				lport = htons(*lastport);
301 			} while (in6_pcblookup_local(pcbinfo,
302 				 &inp->in6p_laddr, lport, wild));
303 		}
304 	}
305 	inp->inp_lport = lport;
306 	if (in_pcbinshash(inp) != 0) {
307 		inp->in6p_laddr = in6addr_any;
308 		inp->inp_lport = 0;
309 		return (EAGAIN);
310 	}
311 	inp->in6p_flowinfo = sin6 ? sin6->sin6_flowinfo : 0;	/*XXX*/
312 	return(0);
313 }
314 
315 /*
316  *   Transform old in6_pcbconnect() into an inner subroutine for new
317  *   in6_pcbconnect(): Do some validity-checking on the remote
318  *   address (in mbuf 'nam') and then determine local host address
319  *   (i.e., which interface) to use to access that remote host.
320  *
321  *   This preserves definition of in6_pcbconnect(), while supporting a
322  *   slightly different version for T/TCP.  (This is more than
323  *   a bit of a kludge, but cleaning up the internal interfaces would
324  *   have forced minor changes in every protocol).
325  */
326 
327 int
328 in6_pcbladdr(inp, nam, plocal_addr6)
329 	register struct inpcb *inp;
330 	struct sockaddr *nam;
331 	struct in6_addr **plocal_addr6;
332 {
333 	register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
334 	struct in6_pktinfo *pi;
335 	struct ifnet *ifp = NULL;
336 	int error = 0;
337 
338 	if (nam->sa_len != sizeof (*sin6))
339 		return (EINVAL);
340 	if (sin6->sin6_family != AF_INET6)
341 		return (EAFNOSUPPORT);
342 	if (sin6->sin6_port == 0)
343 		return (EADDRNOTAVAIL);
344 
345 	/*
346 	 * If the scope of the destination is link-local, embed the interface
347 	 * index in the address.
348 	 */
349 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) {
350 		/* XXX boundary check is assumed to be already done. */
351 		/* XXX sin6_scope_id is weaker than advanced-api. */
352 		if (inp->in6p_outputopts &&
353 		    (pi = inp->in6p_outputopts->ip6po_pktinfo) &&
354 		    pi->ipi6_ifindex) {
355 			sin6->sin6_addr.s6_addr16[1] = htons(pi->ipi6_ifindex);
356 			ifp = ifindex2ifnet[pi->ipi6_ifindex];
357 		} else if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) &&
358 			 inp->in6p_moptions &&
359 			 inp->in6p_moptions->im6o_multicast_ifp) {
360 			sin6->sin6_addr.s6_addr16[1] =
361 				htons(inp->in6p_moptions->im6o_multicast_ifp->if_index);
362 			ifp = ifindex2ifnet[inp->in6p_moptions->im6o_multicast_ifp->if_index];
363 		} else if (sin6->sin6_scope_id) {
364 			/* boundary check */
365 			if (sin6->sin6_scope_id < 0
366 			 || if_index < sin6->sin6_scope_id) {
367 				return ENXIO;  /* XXX EINVAL? */
368 			}
369 			sin6->sin6_addr.s6_addr16[1]
370 				= htons(sin6->sin6_scope_id & 0xffff);/*XXX*/
371 			ifp = ifindex2ifnet[sin6->sin6_scope_id];
372 		}
373 	}
374 
375 	if (in6_ifaddr) {
376 		/*
377 		 * If the destination address is UNSPECIFIED addr,
378 		 * use the loopback addr, e.g ::1.
379 		 */
380 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
381 			sin6->sin6_addr = in6addr_loopback;
382 	}
383 	{
384 		/*
385 		 * XXX: in6_selectsrc might replace the bound local address
386 		 * with the address specified by setsockopt(IPV6_PKTINFO).
387 		 * Is it the intended behavior?
388 		 */
389 		*plocal_addr6 = in6_selectsrc(sin6, inp->in6p_outputopts,
390 					      inp->in6p_moptions,
391 					      &inp->in6p_route,
392 					      &inp->in6p_laddr, &error);
393 		if (*plocal_addr6 == 0) {
394 			if (error == 0)
395 				error = EADDRNOTAVAIL;
396 			return(error);
397 		}
398 		/*
399 		 * Don't do pcblookup call here; return interface in
400 		 * plocal_addr6
401 		 * and exit to caller, that will do the lookup.
402 		 */
403 	}
404 
405 	if (inp->in6p_route.ro_rt)
406 		ifp = inp->in6p_route.ro_rt->rt_ifp;
407 
408 	inp->in6p_ip6_hlim = (u_int8_t)in6_selecthlim(inp, ifp);
409 
410 	return(0);
411 }
412 
413 /*
414  * Outer subroutine:
415  * Connect from a socket to a specified address.
416  * Both address and port must be specified in argument sin.
417  * If don't have a local address for this socket yet,
418  * then pick one.
419  */
420 int
421 in6_pcbconnect(inp, nam, p)
422 	register struct inpcb *inp;
423 	struct sockaddr *nam;
424 	struct proc *p;
425 {
426 	struct in6_addr *addr6;
427 	register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
428 	int error;
429 
430 	/*
431 	 *   Call inner routine, to assign local interface address.
432 	 */
433 	if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0)
434 		return(error);
435 
436 	if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr,
437 			       sin6->sin6_port,
438 			      IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
439 			      ? addr6 : &inp->in6p_laddr,
440 			      inp->inp_lport, 0, NULL) != NULL) {
441 		return (EADDRINUSE);
442 	}
443 	if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
444 		if (inp->inp_lport == 0) {
445 			error = in6_pcbbind(inp, (struct sockaddr *)0, p);
446 			if (error)
447 				return (error);
448 		}
449 		inp->in6p_laddr = *addr6;
450 	}
451 	inp->in6p_faddr = sin6->sin6_addr;
452 	inp->inp_fport = sin6->sin6_port;
453 	/*
454 	 * xxx kazu flowlabel is necessary for connect?
455 	 * but if this line is missing, the garbage value remains.
456 	 */
457 	inp->in6p_flowinfo = sin6->sin6_flowinfo;
458 
459 	in_pcbrehash(inp);
460 	return (0);
461 }
462 
463 /*
464  * Return an IPv6 address, which is the most appropriate for given
465  * destination and user specified options.
466  * If necessary, this function lookups the routing table and return
467  * an entry to the caller for later use.
468  */
469 struct in6_addr *
470 in6_selectsrc(dstsock, opts, mopts, ro, laddr, errorp)
471 	struct sockaddr_in6 *dstsock;
472 	struct ip6_pktopts *opts;
473 	struct ip6_moptions *mopts;
474 	struct route_in6 *ro;
475 	struct in6_addr *laddr;
476 	int *errorp;
477 {
478 	struct in6_addr *dst;
479 	struct in6_ifaddr *ia6 = 0;
480 	struct in6_pktinfo *pi = NULL;
481 
482 	dst = &dstsock->sin6_addr;
483 	*errorp = 0;
484 
485 	/*
486 	 * If the source address is explicitly specified by the caller,
487 	 * use it.
488 	 */
489 	if (opts && (pi = opts->ip6po_pktinfo) &&
490 	    !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr))
491 		return(&pi->ipi6_addr);
492 
493 	/*
494 	 * If the source address is not specified but the socket(if any)
495 	 * is already bound, use the bound address.
496 	 */
497 	if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr))
498 		return(laddr);
499 
500 	/*
501 	 * If the caller doesn't specify the source address but
502 	 * the outgoing interface, use an address associated with
503 	 * the interface.
504 	 */
505 	if (pi && pi->ipi6_ifindex) {
506 		/* XXX boundary check is assumed to be already done. */
507 		ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex],
508 				       dst);
509 		if (ia6 == 0) {
510 			*errorp = EADDRNOTAVAIL;
511 			return(0);
512 		}
513 		return(&satosin6(&ia6->ia_addr)->sin6_addr);
514 	}
515 
516 	/*
517 	 * If the destination address is a link-local unicast address or
518 	 * a multicast address, and if the outgoing interface is specified
519 	 * by the sin6_scope_id filed, use an address associated with the
520 	 * interface.
521 	 * XXX: We're now trying to define more specific semantics of
522 	 *      sin6_scope_id field, so this part will be rewritten in
523 	 *      the near future.
524 	 */
525 	if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) &&
526 	    dstsock->sin6_scope_id) {
527 		/*
528 		 * I'm not sure if boundary check for scope_id is done
529 		 * somewhere...
530 		 */
531 		if (dstsock->sin6_scope_id < 0 ||
532 		    if_index < dstsock->sin6_scope_id) {
533 			*errorp = ENXIO; /* XXX: better error? */
534 			return(0);
535 		}
536 		ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id],
537 				       dst);
538 		if (ia6 == 0) {
539 			*errorp = EADDRNOTAVAIL;
540 			return(0);
541 		}
542 		return(&satosin6(&ia6->ia_addr)->sin6_addr);
543 	}
544 
545 	/*
546 	 * If the destination address is a multicast address and
547 	 * the outgoing interface for the address is specified
548 	 * by the caller, use an address associated with the interface.
549 	 * There is a sanity check here; if the destination has node-local
550 	 * scope, the outgoing interfacde should be a loopback address.
551 	 * Even if the outgoing interface is not specified, we also
552 	 * choose a loopback interface as the outgoing interface.
553 	 */
554 	if (IN6_IS_ADDR_MULTICAST(dst)) {
555 		struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL;
556 
557 		if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) {
558 			ifp = &loif[0];
559 		}
560 
561 		if (ifp) {
562 			ia6 = in6_ifawithscope(ifp, dst);
563 			if (ia6 == 0) {
564 				*errorp = EADDRNOTAVAIL;
565 				return(0);
566 			}
567 			return(&ia6->ia_addr.sin6_addr);
568 		}
569 	}
570 
571 	/*
572 	 * If the next hop address for the packet is specified
573 	 * by caller, use an address associated with the route
574 	 * to the next hop.
575 	 */
576 	{
577 		struct sockaddr_in6 *sin6_next;
578 		struct rtentry *rt;
579 
580 		if (opts && opts->ip6po_nexthop) {
581 			sin6_next = satosin6(opts->ip6po_nexthop);
582 			rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL);
583 			if (rt) {
584 				ia6 = in6_ifawithscope(rt->rt_ifp, dst);
585 				if (ia6 == 0)
586 					ia6 = ifatoia6(rt->rt_ifa);
587 			}
588 			if (ia6 == 0) {
589 				*errorp = EADDRNOTAVAIL;
590 				return(0);
591 			}
592 			return(&satosin6(&ia6->ia_addr)->sin6_addr);
593 		}
594 	}
595 
596 	/*
597 	 * If route is known or can be allocated now,
598 	 * our src addr is taken from the i/f, else punt.
599 	 */
600 	if (ro) {
601 		if (ro->ro_rt &&
602 		    !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, dst)) {
603 			RTFREE(ro->ro_rt);
604 			ro->ro_rt = (struct rtentry *)0;
605 		}
606 		if (ro->ro_rt == (struct rtentry *)0 ||
607 		    ro->ro_rt->rt_ifp == (struct ifnet *)0) {
608 			/* No route yet, so try to acquire one */
609 			bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
610 			ro->ro_dst.sin6_family = AF_INET6;
611 			ro->ro_dst.sin6_len = sizeof(struct sockaddr_in6);
612 			ro->ro_dst.sin6_addr = *dst;
613 			if (IN6_IS_ADDR_MULTICAST(dst)) {
614 				ro->ro_rt = rtalloc1(&((struct route *)ro)
615 						     ->ro_dst, 0, 0UL);
616 			} else {
617 				rtalloc((struct route *)ro);
618 			}
619 		}
620 
621 		/*
622 		 * in_pcbconnect() checks out IFF_LOOPBACK to skip using
623 		 * the address. But we don't know why it does so.
624 		 * It is necessary to ensure the scope even for lo0
625 		 * so doesn't check out IFF_LOOPBACK.
626 		 */
627 
628 		if (ro->ro_rt) {
629 			ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst);
630 			if (ia6 == 0) /* xxx scope error ?*/
631 				ia6 = ifatoia6(ro->ro_rt->rt_ifa);
632 		}
633 		if (ia6 == 0) {
634 			*errorp = EHOSTUNREACH;	/* no route */
635 			return(0);
636 		}
637 		return(&satosin6(&ia6->ia_addr)->sin6_addr);
638 	}
639 
640 	*errorp = EADDRNOTAVAIL;
641 	return(0);
642 }
643 
644 /*
645  * Default hop limit selection. The precedence is as follows:
646  * 1. Hoplimit valued specified via ioctl.
647  * 2. (If the outgoing interface is detected) the current
648  *     hop limit of the interface specified by router advertisement.
649  * 3. The system default hoplimit.
650 */
651 int
652 in6_selecthlim(in6p, ifp)
653 	struct in6pcb *in6p;
654 	struct ifnet *ifp;
655 {
656 	if (in6p && in6p->in6p_hops >= 0)
657 		return(in6p->in6p_hops);
658 	else if (ifp)
659 		return(nd_ifinfo[ifp->if_index].chlim);
660 	else
661 		return(ip6_defhlim);
662 }
663 
664 void
665 in6_pcbdisconnect(inp)
666 	struct inpcb *inp;
667 {
668 	bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
669 	inp->inp_fport = 0;
670 	in_pcbrehash(inp);
671 	if (inp->inp_socket->so_state & SS_NOFDREF)
672 		in6_pcbdetach(inp);
673 }
674 
675 void
676 in6_pcbdetach(inp)
677 	struct inpcb *inp;
678 {
679 	struct socket *so = inp->inp_socket;
680 	struct inpcbinfo *ipi = inp->inp_pcbinfo;
681 
682 #ifdef IPSEC
683 	if (sotoinpcb(so) != 0)
684 		key_freeso(so);
685 	ipsec6_delete_pcbpolicy(inp);
686 #endif /* IPSEC */
687 	inp->inp_gencnt = ++ipi->ipi_gencnt;
688 	in_pcbremlists(inp);
689 	sotoinpcb(so) = 0;
690 	sofree(so);
691 	if (inp->in6p_options)
692 		m_freem(inp->in6p_options);
693 	if (inp->in6p_outputopts) {
694 		if (inp->in6p_outputopts->ip6po_rthdr &&
695 		    inp->in6p_outputopts->ip6po_route.ro_rt)
696 			RTFREE(inp->in6p_outputopts->ip6po_route.ro_rt);
697 		if (inp->in6p_outputopts->ip6po_m)
698 			(void)m_free(inp->in6p_outputopts->ip6po_m);
699 		free(inp->in6p_outputopts, M_IP6OPT);
700 	}
701 	if (inp->in6p_route.ro_rt)
702 		rtfree(inp->in6p_route.ro_rt);
703 	ip6_freemoptions(inp->in6p_moptions);
704 	inp->inp_vflag = 0;
705 	zfreei(ipi->ipi_zone, inp);
706 }
707 
708 /*
709  * The calling convention of in6_setsockaddr() and in6_setpeeraddr() was
710  * modified to match the pru_sockaddr() and pru_peeraddr() entry points
711  * in struct pr_usrreqs, so that protocols can just reference then directly
712  * without the need for a wrapper function.  The socket must have a valid
713  * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
714  * except through a kernel programming error, so it is acceptable to panic
715  * (or in this case trap) if the PCB is invalid.  (Actually, we don't trap
716  * because there actually /is/ a programming error somewhere... XXX)
717  */
718 int
719 in6_setsockaddr(so, nam)
720 	struct socket *so;
721 	struct sockaddr **nam;
722 {
723 	int s;
724 	register struct inpcb *inp;
725 	register struct sockaddr_in6 *sin6;
726 
727 	/*
728 	 * Do the malloc first in case it blocks.
729 	 */
730 	MALLOC(sin6, struct sockaddr_in6 *, sizeof *sin6, M_SONAME, M_WAITOK);
731 	bzero(sin6, sizeof *sin6);
732 	sin6->sin6_family = AF_INET6;
733 	sin6->sin6_len = sizeof(*sin6);
734 
735 	s = splnet();
736 	inp = sotoinpcb(so);
737 	if (!inp) {
738 		splx(s);
739 		free(sin6, M_SONAME);
740 		return EINVAL;
741 	}
742 	sin6->sin6_port = inp->inp_lport;
743 	sin6->sin6_addr = inp->in6p_laddr;
744 	splx(s);
745 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
746 		sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
747 	else
748 		sin6->sin6_scope_id = 0;	/*XXX*/
749 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
750 		sin6->sin6_addr.s6_addr16[1] = 0;
751 
752 	*nam = (struct sockaddr *)sin6;
753 	return 0;
754 }
755 
756 int
757 in6_setpeeraddr(so, nam)
758 	struct socket *so;
759 	struct sockaddr **nam;
760 {
761 	int s;
762 	struct inpcb *inp;
763 	register struct sockaddr_in6 *sin6;
764 
765 	/*
766 	 * Do the malloc first in case it blocks.
767 	 */
768 	MALLOC(sin6, struct sockaddr_in6 *, sizeof(*sin6), M_SONAME, M_WAITOK);
769 	bzero((caddr_t)sin6, sizeof (*sin6));
770 	sin6->sin6_family = AF_INET6;
771 	sin6->sin6_len = sizeof(struct sockaddr_in6);
772 
773 	s = splnet();
774 	inp = sotoinpcb(so);
775 	if (!inp) {
776 		splx(s);
777 		free(sin6, M_SONAME);
778 		return EINVAL;
779 	}
780 	sin6->sin6_port = inp->inp_fport;
781 	sin6->sin6_addr = inp->in6p_faddr;
782 	splx(s);
783 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
784 		sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
785 	else
786 		sin6->sin6_scope_id = 0;	/*XXX*/
787 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
788 		sin6->sin6_addr.s6_addr16[1] = 0;
789 
790 	*nam = (struct sockaddr *)sin6;
791 	return 0;
792 }
793 
794 int
795 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
796 {
797 	struct	inpcb *inp = sotoinpcb(so);
798 	int	error;
799 
800 	if (inp == NULL)
801 		return EINVAL;
802 	if (inp->inp_vflag & INP_IPV4) {
803 		error = in_setsockaddr(so, nam);
804 		if (error == NULL)
805 			in6_sin_2_v4mapsin6_in_sock(nam);
806 	} else
807 	error = in6_setsockaddr(so, nam);
808 
809 	return error;
810 }
811 
812 int
813 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
814 {
815 	struct	inpcb *inp = sotoinpcb(so);
816 	int	error;
817 
818 	if (inp == NULL)
819 		return EINVAL;
820 	if (inp->inp_vflag & INP_IPV4) {
821 		error = in_setpeeraddr(so, nam);
822 		if (error == NULL)
823 			in6_sin_2_v4mapsin6_in_sock(nam);
824 	} else
825 	error = in6_setpeeraddr(so, nam);
826 
827 	return error;
828 }
829 
830 /*
831  * Pass some notification to all connections of a protocol
832  * associated with address dst.  The local address and/or port numbers
833  * may be specified to limit the search.  The "usual action" will be
834  * taken, depending on the ctlinput cmd.  The caller must filter any
835  * cmds that are uninteresting (e.g., no error in the map).
836  * Call the protocol specific routine (if any) to report
837  * any errors for each matching socket.
838  *
839  * Must be called at splnet.
840  */
841 void
842 in6_pcbnotify(head, dst, fport_arg, laddr6, lport_arg, cmd, notify)
843 	struct inpcbhead *head;
844 	struct sockaddr *dst;
845 	u_int fport_arg, lport_arg;
846 	struct in6_addr *laddr6;
847 	int cmd;
848 	void (*notify) __P((struct inpcb *, int));
849 {
850 	struct inpcb *inp, *oinp;
851 	struct in6_addr faddr6;
852 	u_short	fport = fport_arg, lport = lport_arg;
853 	int errno, s;
854 
855 	if ((unsigned)cmd > PRC_NCMDS || dst->sa_family != AF_INET6)
856 		return;
857 	faddr6 = ((struct sockaddr_in6 *)dst)->sin6_addr;
858 	if (IN6_IS_ADDR_UNSPECIFIED(&faddr6))
859 		return;
860 
861 	/*
862 	 * Redirects go to all references to the destination,
863 	 * and use in_rtchange to invalidate the route cache.
864 	 * Dead host indications: notify all references to the destination.
865 	 * Otherwise, if we have knowledge of the local port and address,
866 	 * deliver only to that socket.
867 	 */
868 	if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
869 		fport = 0;
870 		lport = 0;
871 		bzero((caddr_t)laddr6, sizeof(*laddr6));
872 		if (cmd != PRC_HOSTDEAD)
873 			notify = in6_rtchange;
874 	}
875 	errno = inet6ctlerrmap[cmd];
876 	s = splnet();
877 	for (inp = LIST_FIRST(head); inp != NULL;) {
878 		if ((inp->inp_vflag & INP_IPV6) == NULL) {
879 			inp = LIST_NEXT(inp, inp_list);
880 			continue;
881 		}
882 		if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &faddr6) ||
883 		   inp->inp_socket == 0 ||
884 		   (lport && inp->inp_lport != lport) ||
885 		   (!IN6_IS_ADDR_UNSPECIFIED(laddr6) &&
886 		    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr6)) ||
887 		   (fport && inp->inp_fport != fport)) {
888 			inp = LIST_NEXT(inp, inp_list);
889 			continue;
890 		}
891 		oinp = inp;
892 		inp = LIST_NEXT(inp, inp_list);
893 		if (notify)
894 			(*notify)(oinp, errno);
895 	}
896 	splx(s);
897 }
898 
899 /*
900  * Lookup a PCB based on the local address and port.
901  */
902 struct inpcb *
903 in6_pcblookup_local(pcbinfo, laddr, lport_arg, wild_okay)
904 	struct inpcbinfo *pcbinfo;
905 	struct in6_addr *laddr;
906 	u_int lport_arg;
907 	int wild_okay;
908 {
909 	register struct inpcb *inp;
910 	int matchwild = 3, wildcard;
911 	u_short lport = lport_arg;
912 
913 	if (!wild_okay) {
914 		struct inpcbhead *head;
915 		/*
916 		 * Look for an unconnected (wildcard foreign addr) PCB that
917 		 * matches the local address and port we're looking for.
918 		 */
919 		head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0,
920 						      pcbinfo->hashmask)];
921 		LIST_FOREACH(inp, head, inp_hash) {
922 			if ((inp->inp_vflag & INP_IPV6) == NULL)
923 				continue;
924 			if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
925 			    IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
926 			    inp->inp_lport == lport) {
927 				/*
928 				 * Found.
929 				 */
930 				return (inp);
931 			}
932 		}
933 		/*
934 		 * Not found.
935 		 */
936 		return (NULL);
937 	} else {
938 		struct inpcbporthead *porthash;
939 		struct inpcbport *phd;
940 		struct inpcb *match = NULL;
941 		/*
942 		 * Best fit PCB lookup.
943 		 *
944 		 * First see if this local port is in use by looking on the
945 		 * port hash list.
946 		 */
947 		porthash = &pcbinfo->porthashbase[INP_PCBPORTHASH(lport,
948 		    pcbinfo->porthashmask)];
949 		LIST_FOREACH(phd, porthash, phd_hash) {
950 			if (phd->phd_port == lport)
951 				break;
952 		}
953 		if (phd != NULL) {
954 			/*
955 			 * Port is in use by one or more PCBs. Look for best
956 			 * fit.
957 			 */
958 			LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
959 				wildcard = 0;
960 				if ((inp->inp_vflag & INP_IPV6) == NULL)
961 					continue;
962 				if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
963 					wildcard++;
964 				if (!IN6_IS_ADDR_UNSPECIFIED(
965 					&inp->in6p_laddr)) {
966 					if (IN6_IS_ADDR_UNSPECIFIED(laddr))
967 						wildcard++;
968 					else if (!IN6_ARE_ADDR_EQUAL(
969 						&inp->in6p_laddr, laddr))
970 						continue;
971 				} else {
972 					if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
973 						wildcard++;
974 				}
975 				if (wildcard < matchwild) {
976 					match = inp;
977 					matchwild = wildcard;
978 					if (matchwild == 0) {
979 						break;
980 					}
981 				}
982 			}
983 		}
984 		return (match);
985 	}
986 }
987 
988 /*
989  * Check for alternatives when higher level complains
990  * about service problems.  For now, invalidate cached
991  * routing information.  If the route was created dynamically
992  * (by a redirect), time to try a default gateway again.
993  */
994 void
995 in6_losing(in6p)
996 	struct inpcb *in6p;
997 {
998 	struct rtentry *rt;
999 	struct rt_addrinfo info;
1000 
1001 	if ((rt = in6p->in6p_route.ro_rt) != NULL) {
1002 		in6p->in6p_route.ro_rt = 0;
1003 		bzero((caddr_t)&info, sizeof(info));
1004 		info.rti_info[RTAX_DST] =
1005 			(struct sockaddr *)&in6p->in6p_route.ro_dst;
1006 		info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1007 		info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1008 		rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
1009 		if (rt->rt_flags & RTF_DYNAMIC)
1010 			(void)rtrequest(RTM_DELETE, rt_key(rt),
1011 					rt->rt_gateway, rt_mask(rt), rt->rt_flags,
1012 					(struct rtentry **)0);
1013 		else
1014 		/*
1015 		 * A new route can be allocated
1016 		 * the next time output is attempted.
1017 		 */
1018 			rtfree(rt);
1019 	}
1020 }
1021 
1022 /*
1023  * After a routing change, flush old routing
1024  * and allocate a (hopefully) better one.
1025  */
1026 void
1027 in6_rtchange(inp, errno)
1028 	struct inpcb *inp;
1029 	int errno;
1030 {
1031 	if (inp->in6p_route.ro_rt) {
1032 		rtfree(inp->in6p_route.ro_rt);
1033 		inp->in6p_route.ro_rt = 0;
1034 		/*
1035 		 * A new route can be allocated the next time
1036 		 * output is attempted.
1037 		 */
1038 	}
1039 }
1040 
1041 /*
1042  * Lookup PCB in hash list.
1043  */
1044 struct inpcb *
1045 in6_pcblookup_hash(pcbinfo, faddr, fport_arg, laddr, lport_arg, wildcard, ifp)
1046 	struct inpcbinfo *pcbinfo;
1047 	struct in6_addr *faddr, *laddr;
1048 	u_int fport_arg, lport_arg;
1049 	int wildcard;
1050 	struct ifnet *ifp;
1051 {
1052 	struct inpcbhead *head;
1053 	register struct inpcb *inp;
1054 	u_short fport = fport_arg, lport = lport_arg;
1055 
1056 	/*
1057 	 * First look for an exact match.
1058 	 */
1059 	head = &pcbinfo->hashbase[INP_PCBHASH(faddr->s6_addr32[3] /* XXX */,
1060 					      lport, fport,
1061 					      pcbinfo->hashmask)];
1062 	LIST_FOREACH(inp, head, inp_hash) {
1063 		if ((inp->inp_vflag & INP_IPV6) == NULL)
1064 			continue;
1065 		if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1066 		    IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1067 		    inp->inp_fport == fport &&
1068 		    inp->inp_lport == lport) {
1069 			/*
1070 			 * Found.
1071 			 */
1072 			return (inp);
1073 		}
1074 	}
1075 	if (wildcard) {
1076 		struct inpcb *local_wild = NULL;
1077 
1078 		head = &pcbinfo->hashbase[INP_PCBHASH(INADDR_ANY, lport, 0,
1079 						      pcbinfo->hashmask)];
1080 		LIST_FOREACH(inp, head, inp_hash) {
1081 			if ((inp->inp_vflag & INP_IPV6) == NULL)
1082 				continue;
1083 			if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1084 			    inp->inp_lport == lport) {
1085 #if defined(NFAITH) && NFAITH > 0
1086 				if (ifp && ifp->if_type == IFT_FAITH &&
1087 				    (inp->inp_flags & INP_FAITH) == 0)
1088 					continue;
1089 #endif
1090 				if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
1091 						       laddr))
1092 					return (inp);
1093 				else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
1094 					local_wild = inp;
1095 			}
1096 		}
1097 		return (local_wild);
1098 	}
1099 
1100 	/*
1101 	 * Not found.
1102 	 */
1103 	return (NULL);
1104 }
1105 
1106 void
1107 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
1108 {
1109 	struct ip6_hdr *ip;
1110 
1111 	ip = mtod(m, struct ip6_hdr *);
1112 	bzero(sin6, sizeof(*sin6));
1113 	sin6->sin6_len = sizeof(*sin6);
1114 	sin6->sin6_family = AF_INET6;
1115 	sin6->sin6_addr = ip->ip6_src;
1116 	if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1117 		sin6->sin6_addr.s6_addr16[1] = 0;
1118 	sin6->sin6_scope_id =
1119 		(m->m_pkthdr.rcvif && IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1120 		? m->m_pkthdr.rcvif->if_index : 0;
1121 
1122 	return;
1123 }
1124