in6_pcb.c revision 1.120
1/*	$OpenBSD: in6_pcb.c,v 1.120 2022/08/22 10:37:27 bluhm Exp $	*/
2
3/*
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 *    notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 *    notice, this list of conditions and the following disclaimer in the
14 *    documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the project nor the names of its contributors
16 *    may be used to endorse or promote products derived from this software
17 *    without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 */
31
32/*
33 *	@(#)COPYRIGHT	1.1 (NRL) 17 January 1995
34 *
35 * NRL grants permission for redistribution and use in source and binary
36 * forms, with or without modification, of the software and documentation
37 * created at NRL provided that the following conditions are met:
38 *
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 acknowledgements:
46 *	This product includes software developed by the University of
47 *	California, Berkeley and its contributors.
48 *	This product includes software developed at the Information
49 *	Technology Division, US Naval Research Laboratory.
50 * 4. Neither the name of the NRL 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 * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS
55 * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
56 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
57 * PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL NRL OR
58 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
59 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
60 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
61 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
62 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
63 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
64 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
65 *
66 * The views and conclusions contained in the software and documentation
67 * are those of the authors and should not be interpreted as representing
68 * official policies, either expressed or implied, of the US Naval
69 * Research Laboratory (NRL).
70 */
71
72/*
73 * Copyright (c) 1982, 1986, 1990, 1993, 1995
74 *	Regents of the University of California.  All rights reserved.
75 *
76 * Redistribution and use in source and binary forms, with or without
77 * modification, are permitted provided that the following conditions
78 * are met:
79 * 1. Redistributions of source code must retain the above copyright
80 *    notice, this list of conditions and the following disclaimer.
81 * 2. Redistributions in binary form must reproduce the above copyright
82 *    notice, this list of conditions and the following disclaimer in the
83 *    documentation and/or other materials provided with the distribution.
84 * 3. Neither the name of the University nor the names of its contributors
85 *    may be used to endorse or promote products derived from this software
86 *    without specific prior written permission.
87 *
88 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
89 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
90 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
91 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
92 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
93 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
94 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
95 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
96 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
97 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
98 * SUCH DAMAGE.
99 *
100 */
101
102#include "pf.h"
103#include "stoeplitz.h"
104
105#include <sys/param.h>
106#include <sys/systm.h>
107#include <sys/mbuf.h>
108#include <sys/protosw.h>
109#include <sys/socket.h>
110#include <sys/socketvar.h>
111
112#include <net/if.h>
113#include <net/if_var.h>
114#include <net/pfvar.h>
115
116#include <netinet/in.h>
117#include <netinet/ip.h>
118#include <netinet/ip_var.h>
119#include <netinet/in_pcb.h>
120
121#include <netinet6/in6_var.h>
122
123#if NSTOEPLITZ > 0
124#include <net/toeplitz.h>
125#endif
126
127const struct in6_addr zeroin6_addr;
128
129struct inpcbhead *
130in6_pcbhash(struct inpcbtable *table, int rdom,
131    const struct in6_addr *faddr, u_short fport,
132    const struct in6_addr *laddr, u_short lport)
133{
134	SIPHASH_CTX ctx;
135	u_int32_t nrdom = htonl(rdom);
136
137	SipHash24_Init(&ctx, &table->inpt_key);
138	SipHash24_Update(&ctx, &nrdom, sizeof(nrdom));
139	SipHash24_Update(&ctx, faddr, sizeof(*faddr));
140	SipHash24_Update(&ctx, &fport, sizeof(fport));
141	SipHash24_Update(&ctx, laddr, sizeof(*laddr));
142	SipHash24_Update(&ctx, &lport, sizeof(lport));
143
144	return (&table->inpt_hashtbl[SipHash24_End(&ctx) & table->inpt_mask]);
145}
146
147int
148in6_pcbaddrisavail(struct inpcb *inp, struct sockaddr_in6 *sin6, int wild,
149    struct proc *p)
150{
151	struct socket *so = inp->inp_socket;
152	struct inpcbtable *table = inp->inp_table;
153	u_short lport = sin6->sin6_port;
154	int reuseport = (so->so_options & SO_REUSEPORT);
155
156	wild |= INPLOOKUP_IPV6;
157	/* KAME hack: embed scopeid */
158	if (in6_embedscope(&sin6->sin6_addr, sin6, inp) != 0)
159		return (EINVAL);
160	/* this must be cleared for ifa_ifwithaddr() */
161	sin6->sin6_scope_id = 0;
162	/* reject IPv4 mapped address, we have no support for it */
163	if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr))
164		return (EADDRNOTAVAIL);
165
166	if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
167		/*
168		 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
169		 * allow complete duplication of binding if
170		 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
171		 * and a multicast address is bound on both
172		 * new and duplicated sockets.
173		 */
174		if (so->so_options & (SO_REUSEADDR|SO_REUSEPORT))
175			reuseport = SO_REUSEADDR | SO_REUSEPORT;
176	} else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
177		struct ifaddr *ifa = NULL;
178
179		sin6->sin6_port = 0;  /*
180				       * Yechhhh, because of upcoming
181				       * call to ifa_ifwithaddr(), which
182				       * does bcmp's over the PORTS as
183				       * well.  (What about flow?)
184				       */
185		sin6->sin6_flowinfo = 0;
186		if (!(so->so_options & SO_BINDANY) &&
187		    (ifa = ifa_ifwithaddr(sin6tosa(sin6),
188		    inp->inp_rtableid)) == NULL)
189			return (EADDRNOTAVAIL);
190		sin6->sin6_port = lport;
191
192		/*
193		 * bind to an anycast address might accidentally
194		 * cause sending a packet with an anycast source
195		 * address, so we forbid it.
196		 *
197		 * We should allow to bind to a deprecated address,
198		 * since the application dare to use it.
199		 * But, can we assume that they are careful enough
200		 * to check if the address is deprecated or not?
201		 * Maybe, as a safeguard, we should have a setsockopt
202		 * flag to control the bind(2) behavior against
203		 * deprecated addresses (default: forbid bind(2)).
204		 */
205		if (ifa && ifatoia6(ifa)->ia6_flags & (IN6_IFF_ANYCAST|
206		    IN6_IFF_TENTATIVE|IN6_IFF_DUPLICATED|IN6_IFF_DETACHED))
207			return (EADDRNOTAVAIL);
208	}
209	if (lport) {
210		struct inpcb *t;
211		int error = 0;
212
213		if (so->so_euid && !IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
214			t = in_pcblookup_local(table, &sin6->sin6_addr, lport,
215			    INPLOOKUP_WILDCARD | INPLOOKUP_IPV6,
216			    inp->inp_rtableid);
217			if (t && (so->so_euid != t->inp_socket->so_euid))
218				error = EADDRINUSE;
219			in_pcbunref(t);
220			if (error)
221				return (error);
222		}
223		t = in_pcblookup_local(table, &sin6->sin6_addr, lport,
224		    wild, inp->inp_rtableid);
225		if (t && (reuseport & t->inp_socket->so_options) == 0)
226			error = EADDRINUSE;
227		in_pcbunref(t);
228		if (error)
229			return (error);
230	}
231	return (0);
232}
233
234/*
235 * Connect from a socket to a specified address.
236 * Both address and port must be specified in argument sin6.
237 * Eventually, flow labels will have to be dealt with here, as well.
238 *
239 * If don't have a local address for this socket yet,
240 * then pick one.
241 */
242int
243in6_pcbconnect(struct inpcb *inp, struct mbuf *nam)
244{
245	struct in6_addr *in6a = NULL;
246	struct sockaddr_in6 *sin6;
247	struct inpcb *t;
248	int error;
249	struct sockaddr_in6 tmp;
250
251	KASSERT(inp->inp_flags & INP_IPV6);
252
253	if ((error = in6_nam2sin6(nam, &sin6)))
254		return (error);
255	if (sin6->sin6_port == 0)
256		return (EADDRNOTAVAIL);
257	/* reject IPv4 mapped address, we have no support for it */
258	if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr))
259		return (EADDRNOTAVAIL);
260
261	/* protect *sin6 from overwrites */
262	tmp = *sin6;
263	sin6 = &tmp;
264
265	/* KAME hack: embed scopeid */
266	if (in6_embedscope(&sin6->sin6_addr, sin6, inp) != 0)
267		return EINVAL;
268	/* this must be cleared for ifa_ifwithaddr() */
269	sin6->sin6_scope_id = 0;
270
271	/* Source address selection. */
272	/*
273	 * XXX: in6_selectsrc might replace the bound local address
274	 * with the address specified by setsockopt(IPV6_PKTINFO).
275	 * Is it the intended behavior?
276	 */
277	error = in6_pcbselsrc(&in6a, sin6, inp, inp->inp_outputopts6);
278	if (error)
279		return (error);
280
281	inp->inp_ipv6.ip6_hlim = (u_int8_t)in6_selecthlim(inp);
282
283	t = in6_pcbhashlookup(inp->inp_table, &sin6->sin6_addr, sin6->sin6_port,
284	    IN6_IS_ADDR_UNSPECIFIED(&inp->inp_laddr6) ? in6a : &inp->inp_laddr6,
285	    inp->inp_lport, inp->inp_rtableid);
286	if (t != NULL) {
287		in_pcbunref(t);
288		return (EADDRINUSE);
289	}
290
291	KASSERT(IN6_IS_ADDR_UNSPECIFIED(&inp->inp_laddr6) || inp->inp_lport);
292
293	if (IN6_IS_ADDR_UNSPECIFIED(&inp->inp_laddr6)) {
294		if (inp->inp_lport == 0) {
295			error = in_pcbbind(inp, NULL, curproc);
296			if (error)
297				return (error);
298			t = in6_pcbhashlookup(inp->inp_table, &sin6->sin6_addr,
299			    sin6->sin6_port, in6a, inp->inp_lport,
300			    inp->inp_rtableid);
301			if (t != NULL) {
302				inp->inp_lport = 0;
303				in_pcbunref(t);
304				return (EADDRINUSE);
305			}
306		}
307		inp->inp_laddr6 = *in6a;
308	}
309	inp->inp_faddr6 = sin6->sin6_addr;
310	inp->inp_fport = sin6->sin6_port;
311	inp->inp_flowinfo &= ~IPV6_FLOWLABEL_MASK;
312	if (ip6_auto_flowlabel)
313		inp->inp_flowinfo |=
314		    (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
315#if NSTOEPLITZ > 0
316	inp->inp_flowid = stoeplitz_ip6port(&inp->inp_faddr6,
317	    &inp->inp_laddr6, inp->inp_fport, inp->inp_lport);
318#endif
319	in_pcbrehash(inp);
320	return (0);
321}
322
323/*
324 * Get the local address/port, and put it in a sockaddr_in6.
325 * This services the getsockname(2) call.
326 */
327void
328in6_setsockaddr(struct inpcb *inp, struct mbuf *nam)
329{
330	struct sockaddr_in6 *sin6;
331
332	nam->m_len = sizeof(struct sockaddr_in6);
333	sin6 = mtod(nam,struct sockaddr_in6 *);
334
335	bzero ((caddr_t)sin6,sizeof(struct sockaddr_in6));
336	sin6->sin6_family = AF_INET6;
337	sin6->sin6_len = sizeof(struct sockaddr_in6);
338	sin6->sin6_port = inp->inp_lport;
339	sin6->sin6_addr = inp->inp_laddr6;
340	/* KAME hack: recover scopeid */
341	in6_recoverscope(sin6, &inp->inp_laddr6);
342}
343
344/*
345 * Get the foreign address/port, and put it in a sockaddr_in6.
346 * This services the getpeername(2) call.
347 */
348void
349in6_setpeeraddr(struct inpcb *inp, struct mbuf *nam)
350{
351	struct sockaddr_in6 *sin6;
352
353	nam->m_len = sizeof(struct sockaddr_in6);
354	sin6 = mtod(nam,struct sockaddr_in6 *);
355
356	bzero ((caddr_t)sin6,sizeof(struct sockaddr_in6));
357	sin6->sin6_family = AF_INET6;
358	sin6->sin6_len = sizeof(struct sockaddr_in6);
359	sin6->sin6_port = inp->inp_fport;
360	sin6->sin6_addr = inp->inp_faddr6;
361	/* KAME hack: recover scopeid */
362	in6_recoverscope(sin6, &inp->inp_faddr6);
363}
364
365/*
366 * Pass some notification to all connections of a protocol
367 * associated with address dst.  The local address and/or port numbers
368 * may be specified to limit the search.  The "usual action" will be
369 * taken, depending on the ctlinput cmd.  The caller must filter any
370 * cmds that are uninteresting (e.g., no error in the map).
371 * Call the protocol specific routine (if any) to report
372 * any errors for each matching socket.
373 *
374 * Also perform input-side security policy check
375 *    once PCB to be notified has been located.
376 */
377void
378in6_pcbnotify(struct inpcbtable *table, struct sockaddr_in6 *dst,
379    uint fport_arg, const struct sockaddr_in6 *src, uint lport_arg,
380    u_int rtable, int cmd, void *cmdarg, void (*notify)(struct inpcb *, int))
381{
382	SIMPLEQ_HEAD(, inpcb) inpcblist;
383	struct inpcb *inp;
384	u_short fport = fport_arg, lport = lport_arg;
385	struct sockaddr_in6 sa6_src;
386	int errno;
387	u_int32_t flowinfo;
388	u_int rdomain;
389
390	if ((unsigned)cmd >= PRC_NCMDS)
391		return;
392
393	if (IN6_IS_ADDR_UNSPECIFIED(&dst->sin6_addr))
394		return;
395	if (IN6_IS_ADDR_V4MAPPED(&dst->sin6_addr)) {
396#ifdef DIAGNOSTIC
397		printf("%s: Huh?  Thought we never got "
398		       "called with mapped!\n", __func__);
399#endif
400		return;
401	}
402
403	/*
404	 * note that src can be NULL when we get notify by local fragmentation.
405	 */
406	sa6_src = (src == NULL) ? sa6_any : *src;
407	flowinfo = sa6_src.sin6_flowinfo;
408
409	/*
410	 * Redirects go to all references to the destination,
411	 * and use in_rtchange to invalidate the route cache.
412	 * Dead host indications: also use in_rtchange to invalidate
413	 * the cache, and deliver the error to all the sockets.
414	 * Otherwise, if we have knowledge of the local port and address,
415	 * deliver only to that socket.
416	 */
417	if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
418		fport = 0;
419		lport = 0;
420		sa6_src.sin6_addr = in6addr_any;
421
422		if (cmd != PRC_HOSTDEAD)
423			notify = in_rtchange;
424	}
425	errno = inet6ctlerrmap[cmd];
426	if (notify == NULL)
427		return;
428
429	SIMPLEQ_INIT(&inpcblist);
430	rdomain = rtable_l2(rtable);
431	rw_enter_write(&table->inpt_notify);
432	mtx_enter(&table->inpt_mtx);
433	TAILQ_FOREACH(inp, &table->inpt_queue, inp_queue) {
434		if ((inp->inp_flags & INP_IPV6) == 0)
435			continue;
436
437		/*
438		 * Under the following condition, notify of redirects
439		 * to the pcb, without making address matches against inpcb.
440		 * - redirect notification is arrived.
441		 * - the inpcb is unconnected.
442		 * - the inpcb is caching !RTF_HOST routing entry.
443		 * - the ICMPv6 notification is from the gateway cached in the
444		 *   inpcb.  i.e. ICMPv6 notification is from nexthop gateway
445		 *   the inpcb used very recently.
446		 *
447		 * This is to improve interaction between netbsd/openbsd
448		 * redirect handling code, and inpcb route cache code.
449		 * without the clause, !RTF_HOST routing entry (which carries
450		 * gateway used by inpcb right before the ICMPv6 redirect)
451		 * will be cached forever in unconnected inpcb.
452		 *
453		 * There still is a question regarding to what is TRT:
454		 * - On bsdi/freebsd, RTF_HOST (cloned) routing entry will be
455		 *   generated on packet output.  inpcb will always cache
456		 *   RTF_HOST routing entry so there's no need for the clause
457		 *   (ICMPv6 redirect will update RTF_HOST routing entry,
458		 *   and inpcb is caching it already).
459		 *   However, bsdi/freebsd are vulnerable to local DoS attacks
460		 *   due to the cloned routing entries.
461		 * - Specwise, "destination cache" is mentioned in RFC2461.
462		 *   Jinmei says that it implies bsdi/freebsd behavior, itojun
463		 *   is not really convinced.
464		 * - Having hiwat/lowat on # of cloned host route (redirect/
465		 *   pmtud) may be a good idea.  netbsd/openbsd has it.  see
466		 *   icmp6_mtudisc_update().
467		 */
468		if ((PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) &&
469		    IN6_IS_ADDR_UNSPECIFIED(&inp->inp_laddr6) &&
470		    inp->inp_route.ro_rt &&
471		    !(inp->inp_route.ro_rt->rt_flags & RTF_HOST)) {
472			struct sockaddr_in6 *dst6;
473
474			dst6 = satosin6(&inp->inp_route.ro_dst);
475			if (IN6_ARE_ADDR_EQUAL(&dst6->sin6_addr,
476			    &dst->sin6_addr))
477				goto do_notify;
478		}
479
480		/*
481		 * Detect if we should notify the error. If no source and
482		 * destination ports are specified, but non-zero flowinfo and
483		 * local address match, notify the error. This is the case
484		 * when the error is delivered with an encrypted buffer
485		 * by ESP. Otherwise, just compare addresses and ports
486		 * as usual.
487		 */
488		if (lport == 0 && fport == 0 && flowinfo &&
489		    inp->inp_socket != NULL &&
490		    flowinfo == (inp->inp_flowinfo & IPV6_FLOWLABEL_MASK) &&
491		    IN6_ARE_ADDR_EQUAL(&inp->inp_laddr6, &sa6_src.sin6_addr))
492			goto do_notify;
493		else if (!IN6_ARE_ADDR_EQUAL(&inp->inp_faddr6,
494					     &dst->sin6_addr) ||
495			 rtable_l2(inp->inp_rtableid) != rdomain ||
496			 inp->inp_socket == NULL ||
497			 (lport && inp->inp_lport != lport) ||
498			 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
499			  !IN6_ARE_ADDR_EQUAL(&inp->inp_laddr6,
500					      &sa6_src.sin6_addr)) ||
501			 (fport && inp->inp_fport != fport)) {
502			continue;
503		}
504	  do_notify:
505		in_pcbref(inp);
506		SIMPLEQ_INSERT_TAIL(&inpcblist, inp, inp_notify);
507	}
508	mtx_leave(&table->inpt_mtx);
509
510	while ((inp = SIMPLEQ_FIRST(&inpcblist)) != NULL) {
511		SIMPLEQ_REMOVE_HEAD(&inpcblist, inp_notify);
512		(*notify)(inp, errno);
513		in_pcbunref(inp);
514	}
515	rw_exit_write(&table->inpt_notify);
516}
517
518struct inpcb *
519in6_pcbhashlookup(struct inpcbtable *table, const struct in6_addr *faddr,
520    u_int fport_arg, const struct in6_addr *laddr, u_int lport_arg,
521    u_int rtable)
522{
523	struct inpcbhead *head;
524	struct inpcb *inp;
525	u_int16_t fport = fport_arg, lport = lport_arg;
526	u_int rdomain;
527
528	rdomain = rtable_l2(rtable);
529	mtx_enter(&table->inpt_mtx);
530	head = in6_pcbhash(table, rdomain, faddr, fport, laddr, lport);
531	LIST_FOREACH(inp, head, inp_hash) {
532		if (!(inp->inp_flags & INP_IPV6))
533			continue;
534		if (IN6_ARE_ADDR_EQUAL(&inp->inp_faddr6, faddr) &&
535		    inp->inp_fport == fport && inp->inp_lport == lport &&
536		    IN6_ARE_ADDR_EQUAL(&inp->inp_laddr6, laddr) &&
537		    rtable_l2(inp->inp_rtableid) == rdomain) {
538			/*
539			 * Move this PCB to the head of hash chain so that
540			 * repeated accesses are quicker.  This is analogous to
541			 * the historic single-entry PCB cache.
542			 */
543			if (inp != LIST_FIRST(head)) {
544				LIST_REMOVE(inp, inp_hash);
545				LIST_INSERT_HEAD(head, inp, inp_hash);
546			}
547			break;
548		}
549	}
550	in_pcbref(inp);
551	mtx_leave(&table->inpt_mtx);
552#ifdef DIAGNOSTIC
553	if (inp == NULL && in_pcbnotifymiss) {
554		printf("%s: faddr= fport=%d laddr= lport=%d rdom=%u\n",
555		    __func__, ntohs(fport), ntohs(lport), rdomain);
556	}
557#endif
558	return (inp);
559}
560
561struct inpcb *
562in6_pcblookup_listen(struct inpcbtable *table, struct in6_addr *laddr,
563    u_int lport_arg, struct mbuf *m, u_int rtable)
564{
565	struct inpcbhead *head;
566	const struct in6_addr *key1, *key2;
567	struct inpcb *inp;
568	u_int16_t lport = lport_arg;
569	u_int rdomain;
570
571	key1 = laddr;
572	key2 = &zeroin6_addr;
573#if NPF > 0
574	if (m && m->m_pkthdr.pf.flags & PF_TAG_DIVERTED) {
575		struct pf_divert *divert;
576
577		divert = pf_find_divert(m);
578		KASSERT(divert != NULL);
579		switch (divert->type) {
580		case PF_DIVERT_TO:
581			key1 = key2 = &divert->addr.v6;
582			lport = divert->port;
583			break;
584		case PF_DIVERT_REPLY:
585			return (NULL);
586		default:
587			panic("%s: unknown divert type %d, mbuf %p, divert %p",
588			    __func__, divert->type, m, divert);
589		}
590	} else if (m && m->m_pkthdr.pf.flags & PF_TAG_TRANSLATE_LOCALHOST) {
591		/*
592		 * Redirected connections should not be treated the same
593		 * as connections directed to ::1 since localhost
594		 * can only be accessed from the host itself.
595		 */
596		key1 = &zeroin6_addr;
597		key2 = laddr;
598	}
599#endif
600
601	rdomain = rtable_l2(rtable);
602	mtx_enter(&table->inpt_mtx);
603	head = in6_pcbhash(table, rdomain, &zeroin6_addr, 0, key1, lport);
604	LIST_FOREACH(inp, head, inp_hash) {
605		if (!(inp->inp_flags & INP_IPV6))
606			continue;
607		if (inp->inp_lport == lport && inp->inp_fport == 0 &&
608		    IN6_ARE_ADDR_EQUAL(&inp->inp_laddr6, key1) &&
609		    IN6_IS_ADDR_UNSPECIFIED(&inp->inp_faddr6) &&
610		    rtable_l2(inp->inp_rtableid) == rdomain)
611			break;
612	}
613	if (inp == NULL && ! IN6_ARE_ADDR_EQUAL(key1, key2)) {
614		head = in6_pcbhash(table, rdomain,
615		    &zeroin6_addr, 0, key2, lport);
616		LIST_FOREACH(inp, head, inp_hash) {
617			if (!(inp->inp_flags & INP_IPV6))
618				continue;
619			if (inp->inp_lport == lport && inp->inp_fport == 0 &&
620			    IN6_ARE_ADDR_EQUAL(&inp->inp_laddr6, key2) &&
621			    IN6_IS_ADDR_UNSPECIFIED(&inp->inp_faddr6) &&
622			    rtable_l2(inp->inp_rtableid) == rdomain)
623				break;
624		}
625	}
626	/*
627	 * Move this PCB to the head of hash chain so that
628	 * repeated accesses are quicker.  This is analogous to
629	 * the historic single-entry PCB cache.
630	 */
631	if (inp != NULL && inp != LIST_FIRST(head)) {
632		LIST_REMOVE(inp, inp_hash);
633		LIST_INSERT_HEAD(head, inp, inp_hash);
634	}
635	in_pcbref(inp);
636	mtx_leave(&table->inpt_mtx);
637#ifdef DIAGNOSTIC
638	if (inp == NULL && in_pcbnotifymiss) {
639		printf("%s: laddr= lport=%d rdom=%u\n",
640		    __func__, ntohs(lport), rdomain);
641	}
642#endif
643	return (inp);
644}
645