2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * Copyright (c) 2010-2011 Juniper Networks, Inc.
8 * Portions of this software were developed by Robert N. M. Watson under
9 * contract to Juniper Networks, Inc.
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. Neither the name of the project nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
23 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $
39 * Copyright (c) 1982, 1986, 1991, 1993
40 * The Regents of the University of California. All rights reserved.
42 * Redistribution and use in source and binary forms, with or without
43 * modification, are permitted provided that the following conditions
45 * 1. Redistributions of source code must retain the above copyright
46 * notice, this list of conditions and the following disclaimer.
47 * 2. Redistributions in binary form must reproduce the above copyright
48 * notice, this list of conditions and the following disclaimer in the
49 * documentation and/or other materials provided with the distribution.
50 * 3. 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.
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
66 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
69 #include <sys/cdefs.h>
70 __FBSDID("$FreeBSD$");
73 #include "opt_inet6.h"
74 #include "opt_ipsec.h"
75 #include "opt_pcbgroup.h"
78 #include <sys/param.h>
79 #include <sys/systm.h>
80 #include <sys/malloc.h>
82 #include <sys/domain.h>
83 #include <sys/protosw.h>
84 #include <sys/socket.h>
85 #include <sys/socketvar.h>
86 #include <sys/sockio.h>
87 #include <sys/errno.h>
96 #include <net/if_var.h>
97 #include <net/if_llatbl.h>
98 #include <net/if_types.h>
99 #include <net/route.h>
101 #include <netinet/in.h>
102 #include <netinet/in_var.h>
103 #include <netinet/in_systm.h>
104 #include <netinet/tcp_var.h>
105 #include <netinet/ip6.h>
106 #include <netinet/ip_var.h>
108 #include <netinet6/ip6_var.h>
109 #include <netinet6/nd6.h>
110 #include <netinet/in_pcb.h>
111 #include <netinet6/in6_pcb.h>
112 #include <netinet6/scope6_var.h>
114 static struct inpcb *in6_pcblookup_hash_locked(struct inpcbinfo *,
115 struct in6_addr *, u_int, struct in6_addr *, u_int, int, struct ifnet *);
118 in6_pcbbind(struct inpcb *inp, struct sockaddr *nam,
121 struct socket *so = inp->inp_socket;
122 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
123 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
125 int error, lookupflags = 0;
126 int reuseport = (so->so_options & SO_REUSEPORT);
128 INP_WLOCK_ASSERT(inp);
129 INP_HASH_WLOCK_ASSERT(pcbinfo);
131 if (TAILQ_EMPTY(&V_in6_ifaddrhead)) /* XXX broken! */
132 return (EADDRNOTAVAIL);
133 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
135 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
136 lookupflags = INPLOOKUP_WILDCARD;
138 if ((error = prison_local_ip6(cred, &inp->in6p_laddr,
139 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
142 sin6 = (struct sockaddr_in6 *)nam;
143 if (nam->sa_len != sizeof(*sin6))
148 if (nam->sa_family != AF_INET6)
149 return (EAFNOSUPPORT);
151 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
154 if ((error = prison_local_ip6(cred, &sin6->sin6_addr,
155 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
158 lport = sin6->sin6_port;
159 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
161 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
162 * allow compepte duplication of binding if
163 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
164 * and a multicast address is bound on both
165 * new and duplicated sockets.
167 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) != 0)
168 reuseport = SO_REUSEADDR|SO_REUSEPORT;
169 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
172 sin6->sin6_port = 0; /* yech... */
173 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) ==
175 (inp->inp_flags & INP_BINDANY) == 0) {
176 return (EADDRNOTAVAIL);
180 * XXX: bind to an anycast address might accidentally
181 * cause sending a packet with anycast source address.
182 * We should allow to bind to a deprecated address, since
183 * the application dares to use it.
186 ((struct in6_ifaddr *)ifa)->ia6_flags &
187 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) {
189 return (EADDRNOTAVAIL);
199 if (ntohs(lport) <= V_ipport_reservedhigh &&
200 ntohs(lport) >= V_ipport_reservedlow &&
201 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT,
204 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) &&
205 priv_check_cred(inp->inp_cred,
206 PRIV_NETINET_REUSEPORT, 0) != 0) {
207 t = in6_pcblookup_local(pcbinfo,
208 &sin6->sin6_addr, lport,
209 INPLOOKUP_WILDCARD, cred);
211 ((inp->inp_flags2 & INP_BINDMULTI) == 0) &&
212 ((t->inp_flags & INP_TIMEWAIT) == 0) &&
213 (so->so_type != SOCK_STREAM ||
214 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) &&
215 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
216 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
217 (t->inp_flags2 & INP_REUSEPORT) == 0) &&
218 (inp->inp_cred->cr_uid !=
219 t->inp_cred->cr_uid))
223 * If the socket is a BINDMULTI socket, then
224 * the credentials need to match and the
225 * original socket also has to have been bound
228 if (t && (! in_pcbbind_check_bindmulti(inp, t)))
232 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
233 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
234 struct sockaddr_in sin;
236 in6_sin6_2_sin(&sin, sin6);
237 t = in_pcblookup_local(pcbinfo,
239 INPLOOKUP_WILDCARD, cred);
241 ((inp->inp_flags2 & INP_BINDMULTI) == 0) &&
243 INP_TIMEWAIT) == 0) &&
244 (so->so_type != SOCK_STREAM ||
245 ntohl(t->inp_faddr.s_addr) ==
247 (inp->inp_cred->cr_uid !=
248 t->inp_cred->cr_uid))
251 if (t && (! in_pcbbind_check_bindmulti(inp, t)))
256 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr,
257 lport, lookupflags, cred);
258 if (t && (t->inp_flags & INP_TIMEWAIT)) {
260 * XXXRW: If an incpb has had its timewait
261 * state recycled, we treat the address as
262 * being in use (for now). This is better
263 * than a panic, but not desirable.
267 (reuseport & tw->tw_so_options) == 0)
269 } else if (t && (reuseport & inp_so_options(t)) == 0) {
273 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
274 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
275 struct sockaddr_in sin;
277 in6_sin6_2_sin(&sin, sin6);
278 t = in_pcblookup_local(pcbinfo, sin.sin_addr,
279 lport, lookupflags, cred);
280 if (t && t->inp_flags & INP_TIMEWAIT) {
284 if ((reuseport & tw->tw_so_options) == 0
285 && (ntohl(t->inp_laddr.s_addr) !=
286 INADDR_ANY || ((inp->inp_vflag &
288 (t->inp_vflag & INP_IPV6PROTO))))
291 (reuseport & inp_so_options(t)) == 0 &&
292 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
293 (t->inp_vflag & INP_IPV6PROTO) != 0))
298 inp->in6p_laddr = sin6->sin6_addr;
301 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) {
302 /* Undo an address bind that may have occurred. */
303 inp->in6p_laddr = in6addr_any;
307 inp->inp_lport = lport;
308 if (in_pcbinshash(inp) != 0) {
309 inp->in6p_laddr = in6addr_any;
318 * Transform old in6_pcbconnect() into an inner subroutine for new
319 * in6_pcbconnect(): Do some validity-checking on the remote
320 * address (in mbuf 'nam') and then determine local host address
321 * (i.e., which interface) to use to access that remote host.
323 * This preserves definition of in6_pcbconnect(), while supporting a
324 * slightly different version for T/TCP. (This is more than
325 * a bit of a kludge, but cleaning up the internal interfaces would
326 * have forced minor changes in every protocol).
329 in6_pcbladdr(struct inpcb *inp, struct sockaddr *nam,
330 struct in6_addr *plocal_addr6)
332 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
334 int scope_ambiguous = 0;
335 struct in6_addr in6a;
337 INP_WLOCK_ASSERT(inp);
338 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */
340 if (nam->sa_len != sizeof (*sin6))
342 if (sin6->sin6_family != AF_INET6)
343 return (EAFNOSUPPORT);
344 if (sin6->sin6_port == 0)
345 return (EADDRNOTAVAIL);
347 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone)
349 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
352 if (!TAILQ_EMPTY(&V_in6_ifaddrhead)) {
354 * If the destination address is UNSPECIFIED addr,
355 * use the loopback addr, e.g ::1.
357 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
358 sin6->sin6_addr = in6addr_loopback;
360 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0)
363 error = in6_selectsrc_socket(sin6, inp->in6p_outputopts,
364 inp, inp->inp_cred, scope_ambiguous, &in6a, NULL);
369 * Do not update this earlier, in case we return with an error.
371 * XXX: this in6_selectsrc_socket result might replace the bound local
372 * address with the address specified by setsockopt(IPV6_PKTINFO).
373 * Is it the intended behavior?
375 *plocal_addr6 = in6a;
378 * Don't do pcblookup call here; return interface in
380 * and exit to caller, that will do the lookup.
388 * Connect from a socket to a specified address.
389 * Both address and port must be specified in argument sin.
390 * If don't have a local address for this socket yet,
394 in6_pcbconnect_mbuf(struct inpcb *inp, struct sockaddr *nam,
395 struct ucred *cred, struct mbuf *m)
397 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
398 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
399 struct in6_addr addr6;
402 INP_WLOCK_ASSERT(inp);
403 INP_HASH_WLOCK_ASSERT(pcbinfo);
406 * Call inner routine, to assign local interface address.
407 * in6_pcbladdr() may automatically fill in sin6_scope_id.
409 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0)
412 if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr,
414 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
415 ? &addr6 : &inp->in6p_laddr,
416 inp->inp_lport, 0, NULL) != NULL) {
419 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
420 if (inp->inp_lport == 0) {
421 error = in6_pcbbind(inp, (struct sockaddr *)0, cred);
425 inp->in6p_laddr = addr6;
427 inp->in6p_faddr = sin6->sin6_addr;
428 inp->inp_fport = sin6->sin6_port;
429 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
430 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
431 if (inp->inp_flags & IN6P_AUTOFLOWLABEL)
433 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
435 in_pcbrehash_mbuf(inp, m);
441 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred)
444 return (in6_pcbconnect_mbuf(inp, nam, cred, NULL));
448 in6_pcbdisconnect(struct inpcb *inp)
451 INP_WLOCK_ASSERT(inp);
452 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
454 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
456 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
457 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
462 in6_sockaddr(in_port_t port, struct in6_addr *addr_p)
464 struct sockaddr_in6 *sin6;
466 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK);
467 bzero(sin6, sizeof *sin6);
468 sin6->sin6_family = AF_INET6;
469 sin6->sin6_len = sizeof(*sin6);
470 sin6->sin6_port = port;
471 sin6->sin6_addr = *addr_p;
472 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
474 return (struct sockaddr *)sin6;
478 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p)
480 struct sockaddr_in sin;
481 struct sockaddr_in6 *sin6_p;
483 bzero(&sin, sizeof sin);
484 sin.sin_family = AF_INET;
485 sin.sin_len = sizeof(sin);
487 sin.sin_addr = *addr_p;
489 sin6_p = malloc(sizeof *sin6_p, M_SONAME,
491 in6_sin_2_v4mapsin6(&sin, sin6_p);
493 return (struct sockaddr *)sin6_p;
497 in6_getsockaddr(struct socket *so, struct sockaddr **nam)
500 struct in6_addr addr;
504 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL"));
507 port = inp->inp_lport;
508 addr = inp->in6p_laddr;
511 *nam = in6_sockaddr(port, &addr);
516 in6_getpeeraddr(struct socket *so, struct sockaddr **nam)
519 struct in6_addr addr;
523 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL"));
526 port = inp->inp_fport;
527 addr = inp->in6p_faddr;
530 *nam = in6_sockaddr(port, &addr);
535 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
541 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL"));
544 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
545 error = in_getsockaddr(so, nam);
547 in6_sin_2_v4mapsin6_in_sock(nam);
551 /* scope issues will be handled in in6_getsockaddr(). */
552 error = in6_getsockaddr(so, nam);
559 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
565 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL"));
568 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
569 error = in_getpeeraddr(so, nam);
571 in6_sin_2_v4mapsin6_in_sock(nam);
574 /* scope issues will be handled in in6_getpeeraddr(). */
575 error = in6_getpeeraddr(so, nam);
581 * Pass some notification to all connections of a protocol
582 * associated with address dst. The local address and/or port numbers
583 * may be specified to limit the search. The "usual action" will be
584 * taken, depending on the ctlinput cmd. The caller must filter any
585 * cmds that are uninteresting (e.g., no error in the map).
586 * Call the protocol specific routine (if any) to report
587 * any errors for each matching socket.
590 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst,
591 u_int fport_arg, const struct sockaddr *src, u_int lport_arg,
592 int cmd, void *cmdarg,
593 struct inpcb *(*notify)(struct inpcb *, int))
595 struct inpcb *inp, *inp_temp;
596 struct sockaddr_in6 sa6_src, *sa6_dst;
597 u_short fport = fport_arg, lport = lport_arg;
601 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
604 sa6_dst = (struct sockaddr_in6 *)dst;
605 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
609 * note that src can be NULL when we get notify by local fragmentation.
611 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
612 flowinfo = sa6_src.sin6_flowinfo;
615 * Redirects go to all references to the destination,
616 * and use in6_rtchange to invalidate the route cache.
617 * Dead host indications: also use in6_rtchange to invalidate
618 * the cache, and deliver the error to all the sockets.
619 * Otherwise, if we have knowledge of the local port and address,
620 * deliver only to that socket.
622 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
625 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
627 if (cmd != PRC_HOSTDEAD)
628 notify = in6_rtchange;
630 errno = inet6ctlerrmap[cmd];
631 INP_INFO_WLOCK(pcbinfo);
632 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) {
634 if ((inp->inp_vflag & INP_IPV6) == 0) {
640 * If the error designates a new path MTU for a destination
641 * and the application (associated with this socket) wanted to
642 * know the value, notify.
643 * XXX: should we avoid to notify the value to TCP sockets?
645 if (cmd == PRC_MSGSIZE && cmdarg != NULL)
646 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst,
647 *(u_int32_t *)cmdarg);
650 * Detect if we should notify the error. If no source and
651 * destination ports are specifed, but non-zero flowinfo and
652 * local address match, notify the error. This is the case
653 * when the error is delivered with an encrypted buffer
654 * by ESP. Otherwise, just compare addresses and ports
657 if (lport == 0 && fport == 0 && flowinfo &&
658 inp->inp_socket != NULL &&
659 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) &&
660 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
662 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
663 &sa6_dst->sin6_addr) ||
664 inp->inp_socket == 0 ||
665 (lport && inp->inp_lport != lport) ||
666 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
667 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
668 &sa6_src.sin6_addr)) ||
669 (fport && inp->inp_fport != fport)) {
676 if ((*notify)(inp, errno))
681 INP_INFO_WUNLOCK(pcbinfo);
685 * Lookup a PCB based on the local address and port. Caller must hold the
686 * hash lock. No inpcb locks or references are acquired.
689 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr,
690 u_short lport, int lookupflags, struct ucred *cred)
693 int matchwild = 3, wildcard;
695 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
696 ("%s: invalid lookup flags %d", __func__, lookupflags));
698 INP_HASH_WLOCK_ASSERT(pcbinfo);
700 if ((lookupflags & INPLOOKUP_WILDCARD) == 0) {
701 struct inpcbhead *head;
703 * Look for an unconnected (wildcard foreign addr) PCB that
704 * matches the local address and port we're looking for.
706 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(
707 INP6_PCBHASHKEY(&in6addr_any), lport, 0,
708 pcbinfo->ipi_hashmask)];
709 LIST_FOREACH(inp, head, inp_hash) {
710 /* XXX inp locking */
711 if ((inp->inp_vflag & INP_IPV6) == 0)
713 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
714 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
715 inp->inp_lport == lport) {
718 prison_equal_ip6(cred->cr_prison,
719 inp->inp_cred->cr_prison))
728 struct inpcbporthead *porthash;
729 struct inpcbport *phd;
730 struct inpcb *match = NULL;
732 * Best fit PCB lookup.
734 * First see if this local port is in use by looking on the
737 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport,
738 pcbinfo->ipi_porthashmask)];
739 LIST_FOREACH(phd, porthash, phd_hash) {
740 if (phd->phd_port == lport)
745 * Port is in use by one or more PCBs. Look for best
748 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
751 !prison_equal_ip6(cred->cr_prison,
752 inp->inp_cred->cr_prison))
754 /* XXX inp locking */
755 if ((inp->inp_vflag & INP_IPV6) == 0)
757 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
759 if (!IN6_IS_ADDR_UNSPECIFIED(
761 if (IN6_IS_ADDR_UNSPECIFIED(laddr))
763 else if (!IN6_ARE_ADDR_EQUAL(
764 &inp->in6p_laddr, laddr))
767 if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
770 if (wildcard < matchwild) {
772 matchwild = wildcard;
783 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
786 struct ip6_moptions *im6o;
789 INP_INFO_WLOCK(pcbinfo);
790 LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) {
792 im6o = in6p->in6p_moptions;
793 if ((in6p->inp_vflag & INP_IPV6) && im6o != NULL) {
795 * Unselect the outgoing ifp for multicast if it
798 if (im6o->im6o_multicast_ifp == ifp)
799 im6o->im6o_multicast_ifp = NULL;
801 * Drop multicast group membership if we joined
802 * through the interface being detached.
805 for (i = 0; i < im6o->im6o_num_memberships; i++) {
806 if (im6o->im6o_membership[i]->in6m_ifp ==
808 in6_leavegroup(im6o->im6o_membership[i], NULL);
810 } else if (gap != 0) {
811 im6o->im6o_membership[i - gap] =
812 im6o->im6o_membership[i];
815 im6o->im6o_num_memberships -= gap;
819 INP_INFO_WUNLOCK(pcbinfo);
823 * Check for alternatives when higher level complains
824 * about service problems. For now, invalidate cached
825 * routing information. If the route was created dynamically
826 * (by a redirect), time to try a default gateway again.
829 in6_losing(struct inpcb *in6p)
832 if (in6p->inp_route6.ro_rt) {
833 RTFREE(in6p->inp_route6.ro_rt);
834 in6p->inp_route6.ro_rt = (struct rtentry *)NULL;
836 if (in6p->inp_route.ro_lle)
837 LLE_FREE(in6p->inp_route.ro_lle); /* zeros ro_lle */
842 * After a routing change, flush old routing
843 * and allocate a (hopefully) better one.
846 in6_rtchange(struct inpcb *inp, int errno)
849 if (inp->inp_route6.ro_rt) {
850 RTFREE(inp->inp_route6.ro_rt);
851 inp->inp_route6.ro_rt = (struct rtentry *)NULL;
853 if (inp->inp_route.ro_lle)
854 LLE_FREE(inp->inp_route.ro_lle); /* zeros ro_lle */
860 * Lookup PCB in hash list, using pcbgroup tables.
862 static struct inpcb *
863 in6_pcblookup_group(struct inpcbinfo *pcbinfo, struct inpcbgroup *pcbgroup,
864 struct in6_addr *faddr, u_int fport_arg, struct in6_addr *laddr,
865 u_int lport_arg, int lookupflags, struct ifnet *ifp)
867 struct inpcbhead *head;
868 struct inpcb *inp, *tmpinp;
869 u_short fport = fport_arg, lport = lport_arg;
873 * First look for an exact match.
876 INP_GROUP_LOCK(pcbgroup);
877 head = &pcbgroup->ipg_hashbase[INP_PCBHASH(
878 INP6_PCBHASHKEY(faddr), lport, fport, pcbgroup->ipg_hashmask)];
879 LIST_FOREACH(inp, head, inp_pcbgrouphash) {
880 /* XXX inp locking */
881 if ((inp->inp_vflag & INP_IPV6) == 0)
883 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
884 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
885 inp->inp_fport == fport &&
886 inp->inp_lport == lport) {
888 * XXX We should be able to directly return
889 * the inp here, without any checks.
890 * Well unless both bound with SO_REUSEPORT?
892 if (prison_flag(inp->inp_cred, PR_IP6))
898 if (tmpinp != NULL) {
904 * Then look for a wildcard match in the pcbgroup.
906 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
907 struct inpcb *local_wild = NULL, *local_exact = NULL;
908 struct inpcb *jail_wild = NULL;
912 * Order of socket selection - we always prefer jails.
913 * 1. jailed, non-wild.
915 * 3. non-jailed, non-wild.
916 * 4. non-jailed, wild.
918 head = &pcbgroup->ipg_hashbase[
919 INP_PCBHASH(INADDR_ANY, lport, 0, pcbgroup->ipg_hashmask)];
920 LIST_FOREACH(inp, head, inp_pcbgrouphash) {
921 /* XXX inp locking */
922 if ((inp->inp_vflag & INP_IPV6) == 0)
925 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
926 inp->inp_lport != lport) {
930 injail = prison_flag(inp->inp_cred, PR_IP6);
932 if (prison_check_ip6(inp->inp_cred,
936 if (local_exact != NULL)
940 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
945 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
965 * Then look for a wildcard match, if requested.
967 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
968 struct inpcb *local_wild = NULL, *local_exact = NULL;
969 struct inpcb *jail_wild = NULL;
973 * Order of socket selection - we always prefer jails.
974 * 1. jailed, non-wild.
976 * 3. non-jailed, non-wild.
977 * 4. non-jailed, wild.
979 head = &pcbinfo->ipi_wildbase[INP_PCBHASH(
980 INP6_PCBHASHKEY(&in6addr_any), lport, 0,
981 pcbinfo->ipi_wildmask)];
982 LIST_FOREACH(inp, head, inp_pcbgroup_wild) {
983 /* XXX inp locking */
984 if ((inp->inp_vflag & INP_IPV6) == 0)
987 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
988 inp->inp_lport != lport) {
992 injail = prison_flag(inp->inp_cred, PR_IP6);
994 if (prison_check_ip6(inp->inp_cred,
998 if (local_exact != NULL)
1002 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
1007 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1013 } /* LIST_FOREACH */
1024 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
1025 INP_GROUP_UNLOCK(pcbgroup);
1029 if (lookupflags & INPLOOKUP_WLOCKPCB)
1030 locked = INP_TRY_WLOCK(inp);
1031 else if (lookupflags & INPLOOKUP_RLOCKPCB)
1032 locked = INP_TRY_RLOCK(inp);
1034 panic("%s: locking buf", __func__);
1037 INP_GROUP_UNLOCK(pcbgroup);
1039 if (lookupflags & INPLOOKUP_WLOCKPCB) {
1041 if (in_pcbrele_wlocked(inp))
1045 if (in_pcbrele_rlocked(inp))
1050 if (lookupflags & INPLOOKUP_WLOCKPCB)
1051 INP_WLOCK_ASSERT(inp);
1053 INP_RLOCK_ASSERT(inp);
1057 #endif /* PCBGROUP */
1060 * Lookup PCB in hash list.
1062 static struct inpcb *
1063 in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1064 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg,
1065 int lookupflags, struct ifnet *ifp)
1067 struct inpcbhead *head;
1068 struct inpcb *inp, *tmpinp;
1069 u_short fport = fport_arg, lport = lport_arg;
1071 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
1072 ("%s: invalid lookup flags %d", __func__, lookupflags));
1074 INP_HASH_LOCK_ASSERT(pcbinfo);
1077 * First look for an exact match.
1080 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(
1081 INP6_PCBHASHKEY(faddr), lport, fport, pcbinfo->ipi_hashmask)];
1082 LIST_FOREACH(inp, head, inp_hash) {
1083 /* XXX inp locking */
1084 if ((inp->inp_vflag & INP_IPV6) == 0)
1086 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1087 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1088 inp->inp_fport == fport &&
1089 inp->inp_lport == lport) {
1091 * XXX We should be able to directly return
1092 * the inp here, without any checks.
1093 * Well unless both bound with SO_REUSEPORT?
1095 if (prison_flag(inp->inp_cred, PR_IP6))
1105 * Then look for a wildcard match, if requested.
1107 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
1108 struct inpcb *local_wild = NULL, *local_exact = NULL;
1109 struct inpcb *jail_wild = NULL;
1113 * Order of socket selection - we always prefer jails.
1114 * 1. jailed, non-wild.
1116 * 3. non-jailed, non-wild.
1117 * 4. non-jailed, wild.
1119 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(
1120 INP6_PCBHASHKEY(&in6addr_any), lport, 0,
1121 pcbinfo->ipi_hashmask)];
1122 LIST_FOREACH(inp, head, inp_hash) {
1123 /* XXX inp locking */
1124 if ((inp->inp_vflag & INP_IPV6) == 0)
1127 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
1128 inp->inp_lport != lport) {
1132 injail = prison_flag(inp->inp_cred, PR_IP6);
1134 if (prison_check_ip6(inp->inp_cred,
1138 if (local_exact != NULL)
1142 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
1147 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1153 } /* LIST_FOREACH */
1155 if (jail_wild != NULL)
1157 if (local_exact != NULL)
1158 return (local_exact);
1159 if (local_wild != NULL)
1160 return (local_wild);
1161 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
1170 * Lookup PCB in hash list, using pcbinfo tables. This variation locks the
1171 * hash list lock, and will return the inpcb locked (i.e., requires
1172 * INPLOOKUP_LOCKPCB).
1174 static struct inpcb *
1175 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1176 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1182 INP_HASH_RLOCK(pcbinfo);
1183 inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport,
1184 (lookupflags & ~(INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)), ifp);
1186 if (lookupflags & INPLOOKUP_WLOCKPCB)
1187 locked = INP_TRY_WLOCK(inp);
1188 else if (lookupflags & INPLOOKUP_RLOCKPCB)
1189 locked = INP_TRY_RLOCK(inp);
1191 panic("%s: locking bug", __func__);
1194 INP_HASH_RUNLOCK(pcbinfo);
1196 if (lookupflags & INPLOOKUP_WLOCKPCB) {
1198 if (in_pcbrele_wlocked(inp))
1202 if (in_pcbrele_rlocked(inp))
1207 if (lookupflags & INPLOOKUP_WLOCKPCB)
1208 INP_WLOCK_ASSERT(inp);
1210 INP_RLOCK_ASSERT(inp);
1213 INP_HASH_RUNLOCK(pcbinfo);
1218 * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf
1219 * from which a pre-calculated hash value may be extracted.
1221 * Possibly more of this logic should be in in6_pcbgroup.c.
1224 in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport,
1225 struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp)
1227 #if defined(PCBGROUP) && !defined(RSS)
1228 struct inpcbgroup *pcbgroup;
1231 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1232 ("%s: invalid lookup flags %d", __func__, lookupflags));
1233 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1234 ("%s: LOCKPCB not set", __func__));
1237 * When not using RSS, use connection groups in preference to the
1238 * reservation table when looking up 4-tuples. When using RSS, just
1239 * use the reservation table, due to the cost of the Toeplitz hash
1242 * XXXRW: This policy belongs in the pcbgroup code, as in principle
1243 * we could be doing RSS with a non-Toeplitz hash that is affordable
1246 #if defined(PCBGROUP) && !defined(RSS)
1247 if (in_pcbgroup_enabled(pcbinfo)) {
1248 pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr,
1250 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport,
1251 laddr, lport, lookupflags, ifp));
1254 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1259 in6_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1260 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1261 struct ifnet *ifp, struct mbuf *m)
1264 struct inpcbgroup *pcbgroup;
1267 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1268 ("%s: invalid lookup flags %d", __func__, lookupflags));
1269 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1270 ("%s: LOCKPCB not set", __func__));
1274 * If we can use a hardware-generated hash to look up the connection
1275 * group, use that connection group to find the inpcb. Otherwise
1276 * fall back on a software hash -- or the reservation table if we're
1279 * XXXRW: As above, that policy belongs in the pcbgroup code.
1281 if (in_pcbgroup_enabled(pcbinfo) &&
1282 M_HASHTYPE_TEST(m, M_HASHTYPE_NONE) == 0) {
1283 pcbgroup = in6_pcbgroup_byhash(pcbinfo, M_HASHTYPE_GET(m),
1284 m->m_pkthdr.flowid);
1285 if (pcbgroup != NULL)
1286 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr,
1287 fport, laddr, lport, lookupflags, ifp));
1289 pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr,
1291 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport,
1292 laddr, lport, lookupflags, ifp));
1296 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1301 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m, int srcordst)
1305 ip = mtod(m, struct ip6_hdr *);
1306 bzero(sin6, sizeof(*sin6));
1307 sin6->sin6_len = sizeof(*sin6);
1308 sin6->sin6_family = AF_INET6;
1309 sin6->sin6_addr = srcordst ? ip->ip6_dst : ip->ip6_src;
1311 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */