2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
3 * Copyright (c) 2010-2011 Juniper Networks, Inc.
6 * Portions of this software were developed by Robert N. M. Watson under
7 * contract to Juniper Networks, Inc.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the name of the project nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $
37 * Copyright (c) 1982, 1986, 1991, 1993
38 * The Regents of the University of California. All rights reserved.
40 * Redistribution and use in source and binary forms, with or without
41 * modification, are permitted provided that the following conditions
43 * 1. Redistributions of source code must retain the above copyright
44 * notice, this list of conditions and the following disclaimer.
45 * 2. Redistributions in binary form must reproduce the above copyright
46 * notice, this list of conditions and the following disclaimer in the
47 * documentation and/or other materials provided with the distribution.
48 * 4. Neither the name of the University nor the names of its contributors
49 * may be used to endorse or promote products derived from this software
50 * without specific prior written permission.
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
64 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
67 #include <sys/cdefs.h>
68 __FBSDID("$FreeBSD$");
71 #include "opt_inet6.h"
72 #include "opt_ipsec.h"
73 #include "opt_pcbgroup.h"
75 #include <sys/param.h>
76 #include <sys/systm.h>
77 #include <sys/malloc.h>
79 #include <sys/domain.h>
80 #include <sys/protosw.h>
81 #include <sys/socket.h>
82 #include <sys/socketvar.h>
83 #include <sys/sockio.h>
84 #include <sys/errno.h>
93 #include <net/if_types.h>
94 #include <net/route.h>
96 #include <netinet/in.h>
97 #include <netinet/in_var.h>
98 #include <netinet/in_systm.h>
99 #include <netinet/tcp_var.h>
100 #include <netinet/ip6.h>
101 #include <netinet/ip_var.h>
103 #include <netinet6/ip6_var.h>
104 #include <netinet6/nd6.h>
105 #include <netinet/in_pcb.h>
106 #include <netinet6/in6_pcb.h>
107 #include <netinet6/scope6_var.h>
110 in6_pcbbind(register struct inpcb *inp, struct sockaddr *nam,
113 struct socket *so = inp->inp_socket;
114 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
115 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
117 int error, lookupflags = 0;
118 int reuseport = (so->so_options & SO_REUSEPORT);
120 INP_WLOCK_ASSERT(inp);
121 INP_HASH_WLOCK_ASSERT(pcbinfo);
123 if (TAILQ_EMPTY(&V_in6_ifaddrhead)) /* XXX broken! */
124 return (EADDRNOTAVAIL);
125 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
127 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
128 lookupflags = INPLOOKUP_WILDCARD;
130 if ((error = prison_local_ip6(cred, &inp->in6p_laddr,
131 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
134 sin6 = (struct sockaddr_in6 *)nam;
135 if (nam->sa_len != sizeof(*sin6))
140 if (nam->sa_family != AF_INET6)
141 return (EAFNOSUPPORT);
143 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
146 if ((error = prison_local_ip6(cred, &sin6->sin6_addr,
147 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
150 lport = sin6->sin6_port;
151 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
153 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
154 * allow compepte duplication of binding if
155 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
156 * and a multicast address is bound on both
157 * new and duplicated sockets.
159 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) != 0)
160 reuseport = SO_REUSEADDR|SO_REUSEPORT;
161 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
164 sin6->sin6_port = 0; /* yech... */
165 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) ==
167 (inp->inp_flags & INP_BINDANY) == 0) {
168 return (EADDRNOTAVAIL);
172 * XXX: bind to an anycast address might accidentally
173 * cause sending a packet with anycast source address.
174 * We should allow to bind to a deprecated address, since
175 * the application dares to use it.
178 ((struct in6_ifaddr *)ifa)->ia6_flags &
179 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) {
181 return (EADDRNOTAVAIL);
191 if (ntohs(lport) <= V_ipport_reservedhigh &&
192 ntohs(lport) >= V_ipport_reservedlow &&
193 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT,
196 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) &&
197 priv_check_cred(inp->inp_cred,
198 PRIV_NETINET_REUSEPORT, 0) != 0) {
199 t = in6_pcblookup_local(pcbinfo,
200 &sin6->sin6_addr, lport,
201 INPLOOKUP_WILDCARD, cred);
203 ((t->inp_flags & INP_TIMEWAIT) == 0) &&
204 (so->so_type != SOCK_STREAM ||
205 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) &&
206 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
207 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
208 (t->inp_flags2 & INP_REUSEPORT) == 0) &&
209 (inp->inp_cred->cr_uid !=
210 t->inp_cred->cr_uid))
213 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
214 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
215 struct sockaddr_in sin;
217 in6_sin6_2_sin(&sin, sin6);
218 t = in_pcblookup_local(pcbinfo,
220 INPLOOKUP_WILDCARD, cred);
223 INP_TIMEWAIT) == 0) &&
224 (so->so_type != SOCK_STREAM ||
225 ntohl(t->inp_faddr.s_addr) ==
227 (inp->inp_cred->cr_uid !=
228 t->inp_cred->cr_uid))
233 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr,
234 lport, lookupflags, cred);
235 if (t && (t->inp_flags & INP_TIMEWAIT)) {
237 * XXXRW: If an incpb has had its timewait
238 * state recycled, we treat the address as
239 * being in use (for now). This is better
240 * than a panic, but not desirable.
244 (reuseport & tw->tw_so_options) == 0)
246 } else if (t && (reuseport & inp_so_options(t)) == 0) {
250 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
251 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
252 struct sockaddr_in sin;
254 in6_sin6_2_sin(&sin, sin6);
255 t = in_pcblookup_local(pcbinfo, sin.sin_addr,
256 lport, lookupflags, cred);
257 if (t && t->inp_flags & INP_TIMEWAIT) {
261 if ((reuseport & tw->tw_so_options) == 0
262 && (ntohl(t->inp_laddr.s_addr) !=
263 INADDR_ANY || ((inp->inp_vflag &
265 (t->inp_vflag & INP_IPV6PROTO))))
268 (reuseport & inp_so_options(t)) == 0 &&
269 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
270 (t->inp_vflag & INP_IPV6PROTO) != 0))
275 inp->in6p_laddr = sin6->sin6_addr;
278 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) {
279 /* Undo an address bind that may have occurred. */
280 inp->in6p_laddr = in6addr_any;
284 inp->inp_lport = lport;
285 if (in_pcbinshash(inp) != 0) {
286 inp->in6p_laddr = in6addr_any;
295 * Transform old in6_pcbconnect() into an inner subroutine for new
296 * in6_pcbconnect(): Do some validity-checking on the remote
297 * address (in mbuf 'nam') and then determine local host address
298 * (i.e., which interface) to use to access that remote host.
300 * This preserves definition of in6_pcbconnect(), while supporting a
301 * slightly different version for T/TCP. (This is more than
302 * a bit of a kludge, but cleaning up the internal interfaces would
303 * have forced minor changes in every protocol).
306 in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam,
307 struct in6_addr *plocal_addr6)
309 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
311 struct ifnet *ifp = NULL;
312 int scope_ambiguous = 0;
313 struct in6_addr in6a;
315 INP_WLOCK_ASSERT(inp);
316 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */
318 if (nam->sa_len != sizeof (*sin6))
320 if (sin6->sin6_family != AF_INET6)
321 return (EAFNOSUPPORT);
322 if (sin6->sin6_port == 0)
323 return (EADDRNOTAVAIL);
325 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone)
327 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
330 if (!TAILQ_EMPTY(&V_in6_ifaddrhead)) {
332 * If the destination address is UNSPECIFIED addr,
333 * use the loopback addr, e.g ::1.
335 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
336 sin6->sin6_addr = in6addr_loopback;
338 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0)
341 error = in6_selectsrc(sin6, inp->in6p_outputopts,
342 inp, NULL, inp->inp_cred, &ifp, &in6a);
346 if (ifp && scope_ambiguous &&
347 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) {
352 * Do not update this earlier, in case we return with an error.
354 * XXX: this in6_selectsrc result might replace the bound local
355 * address with the address specified by setsockopt(IPV6_PKTINFO).
356 * Is it the intended behavior?
358 *plocal_addr6 = in6a;
361 * Don't do pcblookup call here; return interface in
363 * and exit to caller, that will do the lookup.
371 * Connect from a socket to a specified address.
372 * Both address and port must be specified in argument sin.
373 * If don't have a local address for this socket yet,
377 in6_pcbconnect_mbuf(register struct inpcb *inp, struct sockaddr *nam,
378 struct ucred *cred, struct mbuf *m)
380 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
381 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
382 struct in6_addr addr6;
385 INP_WLOCK_ASSERT(inp);
386 INP_HASH_WLOCK_ASSERT(pcbinfo);
389 * Call inner routine, to assign local interface address.
390 * in6_pcbladdr() may automatically fill in sin6_scope_id.
392 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0)
395 if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr,
397 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
398 ? &addr6 : &inp->in6p_laddr,
399 inp->inp_lport, 0, NULL) != NULL) {
402 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
403 if (inp->inp_lport == 0) {
404 error = in6_pcbbind(inp, (struct sockaddr *)0, cred);
408 inp->in6p_laddr = addr6;
410 inp->in6p_faddr = sin6->sin6_addr;
411 inp->inp_fport = sin6->sin6_port;
412 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
413 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
414 if (inp->inp_flags & IN6P_AUTOFLOWLABEL)
416 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
418 in_pcbrehash_mbuf(inp, m);
424 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred)
427 return (in6_pcbconnect_mbuf(inp, nam, cred, NULL));
431 in6_pcbdisconnect(struct inpcb *inp)
434 INP_WLOCK_ASSERT(inp);
435 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
437 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
439 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
440 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
445 in6_sockaddr(in_port_t port, struct in6_addr *addr_p)
447 struct sockaddr_in6 *sin6;
449 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK);
450 bzero(sin6, sizeof *sin6);
451 sin6->sin6_family = AF_INET6;
452 sin6->sin6_len = sizeof(*sin6);
453 sin6->sin6_port = port;
454 sin6->sin6_addr = *addr_p;
455 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
457 return (struct sockaddr *)sin6;
461 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p)
463 struct sockaddr_in sin;
464 struct sockaddr_in6 *sin6_p;
466 bzero(&sin, sizeof sin);
467 sin.sin_family = AF_INET;
468 sin.sin_len = sizeof(sin);
470 sin.sin_addr = *addr_p;
472 sin6_p = malloc(sizeof *sin6_p, M_SONAME,
474 in6_sin_2_v4mapsin6(&sin, sin6_p);
476 return (struct sockaddr *)sin6_p;
480 in6_getsockaddr(struct socket *so, struct sockaddr **nam)
482 register struct inpcb *inp;
483 struct in6_addr addr;
487 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL"));
490 port = inp->inp_lport;
491 addr = inp->in6p_laddr;
494 *nam = in6_sockaddr(port, &addr);
499 in6_getpeeraddr(struct socket *so, struct sockaddr **nam)
502 struct in6_addr addr;
506 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL"));
509 port = inp->inp_fport;
510 addr = inp->in6p_faddr;
513 *nam = in6_sockaddr(port, &addr);
518 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
524 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL"));
527 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
528 error = in_getsockaddr(so, nam);
530 in6_sin_2_v4mapsin6_in_sock(nam);
534 /* scope issues will be handled in in6_getsockaddr(). */
535 error = in6_getsockaddr(so, nam);
542 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
548 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL"));
551 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
552 error = in_getpeeraddr(so, nam);
554 in6_sin_2_v4mapsin6_in_sock(nam);
557 /* scope issues will be handled in in6_getpeeraddr(). */
558 error = in6_getpeeraddr(so, nam);
564 * Pass some notification to all connections of a protocol
565 * associated with address dst. The local address and/or port numbers
566 * may be specified to limit the search. The "usual action" will be
567 * taken, depending on the ctlinput cmd. The caller must filter any
568 * cmds that are uninteresting (e.g., no error in the map).
569 * Call the protocol specific routine (if any) to report
570 * any errors for each matching socket.
573 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst,
574 u_int fport_arg, const struct sockaddr *src, u_int lport_arg,
575 int cmd, void *cmdarg,
576 struct inpcb *(*notify)(struct inpcb *, int))
578 struct inpcb *inp, *inp_temp;
579 struct sockaddr_in6 sa6_src, *sa6_dst;
580 u_short fport = fport_arg, lport = lport_arg;
584 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
587 sa6_dst = (struct sockaddr_in6 *)dst;
588 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
592 * note that src can be NULL when we get notify by local fragmentation.
594 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
595 flowinfo = sa6_src.sin6_flowinfo;
598 * Redirects go to all references to the destination,
599 * and use in6_rtchange to invalidate the route cache.
600 * Dead host indications: also use in6_rtchange to invalidate
601 * the cache, and deliver the error to all the sockets.
602 * Otherwise, if we have knowledge of the local port and address,
603 * deliver only to that socket.
605 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
608 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
610 if (cmd != PRC_HOSTDEAD)
611 notify = in6_rtchange;
613 errno = inet6ctlerrmap[cmd];
614 INP_INFO_WLOCK(pcbinfo);
615 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) {
617 if ((inp->inp_vflag & INP_IPV6) == 0) {
623 * If the error designates a new path MTU for a destination
624 * and the application (associated with this socket) wanted to
625 * know the value, notify.
626 * XXX: should we avoid to notify the value to TCP sockets?
628 if (cmd == PRC_MSGSIZE && cmdarg != NULL)
629 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst,
630 *(u_int32_t *)cmdarg);
633 * Detect if we should notify the error. If no source and
634 * destination ports are specifed, but non-zero flowinfo and
635 * local address match, notify the error. This is the case
636 * when the error is delivered with an encrypted buffer
637 * by ESP. Otherwise, just compare addresses and ports
640 if (lport == 0 && fport == 0 && flowinfo &&
641 inp->inp_socket != NULL &&
642 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) &&
643 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
645 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
646 &sa6_dst->sin6_addr) ||
647 inp->inp_socket == 0 ||
648 (lport && inp->inp_lport != lport) ||
649 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
650 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
651 &sa6_src.sin6_addr)) ||
652 (fport && inp->inp_fport != fport)) {
659 if ((*notify)(inp, errno))
664 INP_INFO_WUNLOCK(pcbinfo);
668 * Lookup a PCB based on the local address and port. Caller must hold the
669 * hash lock. No inpcb locks or references are acquired.
672 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr,
673 u_short lport, int lookupflags, struct ucred *cred)
675 register struct inpcb *inp;
676 int matchwild = 3, wildcard;
678 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
679 ("%s: invalid lookup flags %d", __func__, lookupflags));
681 INP_HASH_WLOCK_ASSERT(pcbinfo);
683 if ((lookupflags & INPLOOKUP_WILDCARD) == 0) {
684 struct inpcbhead *head;
686 * Look for an unconnected (wildcard foreign addr) PCB that
687 * matches the local address and port we're looking for.
689 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
690 0, pcbinfo->ipi_hashmask)];
691 LIST_FOREACH(inp, head, inp_hash) {
692 /* XXX inp locking */
693 if ((inp->inp_vflag & INP_IPV6) == 0)
695 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
696 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
697 inp->inp_lport == lport) {
700 prison_equal_ip6(cred->cr_prison,
701 inp->inp_cred->cr_prison))
710 struct inpcbporthead *porthash;
711 struct inpcbport *phd;
712 struct inpcb *match = NULL;
714 * Best fit PCB lookup.
716 * First see if this local port is in use by looking on the
719 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport,
720 pcbinfo->ipi_porthashmask)];
721 LIST_FOREACH(phd, porthash, phd_hash) {
722 if (phd->phd_port == lport)
727 * Port is in use by one or more PCBs. Look for best
730 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
733 !prison_equal_ip6(cred->cr_prison,
734 inp->inp_cred->cr_prison))
736 /* XXX inp locking */
737 if ((inp->inp_vflag & INP_IPV6) == 0)
739 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
741 if (!IN6_IS_ADDR_UNSPECIFIED(
743 if (IN6_IS_ADDR_UNSPECIFIED(laddr))
745 else if (!IN6_ARE_ADDR_EQUAL(
746 &inp->in6p_laddr, laddr))
749 if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
752 if (wildcard < matchwild) {
754 matchwild = wildcard;
765 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
768 struct ip6_moptions *im6o;
771 INP_INFO_RLOCK(pcbinfo);
772 LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) {
774 im6o = in6p->in6p_moptions;
775 if ((in6p->inp_vflag & INP_IPV6) && im6o != NULL) {
777 * Unselect the outgoing ifp for multicast if it
780 if (im6o->im6o_multicast_ifp == ifp)
781 im6o->im6o_multicast_ifp = NULL;
783 * Drop multicast group membership if we joined
784 * through the interface being detached.
787 for (i = 0; i < im6o->im6o_num_memberships; i++) {
788 if (im6o->im6o_membership[i]->in6m_ifp ==
790 in6_mc_leave(im6o->im6o_membership[i],
793 } else if (gap != 0) {
794 im6o->im6o_membership[i - gap] =
795 im6o->im6o_membership[i];
798 im6o->im6o_num_memberships -= gap;
802 INP_INFO_RUNLOCK(pcbinfo);
806 * Check for alternatives when higher level complains
807 * about service problems. For now, invalidate cached
808 * routing information. If the route was created dynamically
809 * (by a redirect), time to try a default gateway again.
812 in6_losing(struct inpcb *in6p)
816 * We don't store route pointers in the routing table anymore
822 * After a routing change, flush old routing
823 * and allocate a (hopefully) better one.
826 in6_rtchange(struct inpcb *inp, int errno)
829 * We don't store route pointers in the routing table anymore
836 * Lookup PCB in hash list, using pcbgroup tables.
838 static struct inpcb *
839 in6_pcblookup_group(struct inpcbinfo *pcbinfo, struct inpcbgroup *pcbgroup,
840 struct in6_addr *faddr, u_int fport_arg, struct in6_addr *laddr,
841 u_int lport_arg, int lookupflags, struct ifnet *ifp)
843 struct inpcbhead *head;
844 struct inpcb *inp, *tmpinp;
845 u_short fport = fport_arg, lport = lport_arg;
848 if (faithprefix_p != NULL)
849 faith = (*faithprefix_p)(laddr);
854 * First look for an exact match.
857 INP_GROUP_LOCK(pcbgroup);
858 head = &pcbgroup->ipg_hashbase[
859 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport,
860 pcbgroup->ipg_hashmask)];
861 LIST_FOREACH(inp, head, inp_pcbgrouphash) {
862 /* XXX inp locking */
863 if ((inp->inp_vflag & INP_IPV6) == 0)
865 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
866 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
867 inp->inp_fport == fport &&
868 inp->inp_lport == lport) {
870 * XXX We should be able to directly return
871 * the inp here, without any checks.
872 * Well unless both bound with SO_REUSEPORT?
874 if (prison_flag(inp->inp_cred, PR_IP6))
880 if (tmpinp != NULL) {
886 * Then look for a wildcard match, if requested.
888 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
889 struct inpcb *local_wild = NULL, *local_exact = NULL;
890 struct inpcb *jail_wild = NULL;
894 * Order of socket selection - we always prefer jails.
895 * 1. jailed, non-wild.
897 * 3. non-jailed, non-wild.
898 * 4. non-jailed, wild.
900 head = &pcbinfo->ipi_wildbase[INP_PCBHASH(INADDR_ANY, lport,
901 0, pcbinfo->ipi_wildmask)];
902 LIST_FOREACH(inp, head, inp_pcbgroup_wild) {
903 /* XXX inp locking */
904 if ((inp->inp_vflag & INP_IPV6) == 0)
907 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
908 inp->inp_lport != lport) {
912 /* XXX inp locking */
913 if (faith && (inp->inp_flags & INP_FAITH) == 0)
916 injail = prison_flag(inp->inp_cred, PR_IP6);
918 if (prison_check_ip6(inp->inp_cred,
922 if (local_exact != NULL)
926 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
931 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
948 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
949 INP_GROUP_UNLOCK(pcbgroup);
954 INP_GROUP_UNLOCK(pcbgroup);
955 if (lookupflags & INPLOOKUP_WLOCKPCB) {
957 if (in_pcbrele_wlocked(inp))
959 } else if (lookupflags & INPLOOKUP_RLOCKPCB) {
961 if (in_pcbrele_rlocked(inp))
964 panic("%s: locking buf", __func__);
967 #endif /* PCBGROUP */
970 * Lookup PCB in hash list.
973 in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
974 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg,
975 int lookupflags, struct ifnet *ifp)
977 struct inpcbhead *head;
978 struct inpcb *inp, *tmpinp;
979 u_short fport = fport_arg, lport = lport_arg;
982 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
983 ("%s: invalid lookup flags %d", __func__, lookupflags));
985 INP_HASH_LOCK_ASSERT(pcbinfo);
987 if (faithprefix_p != NULL)
988 faith = (*faithprefix_p)(laddr);
993 * First look for an exact match.
996 head = &pcbinfo->ipi_hashbase[
997 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport,
998 pcbinfo->ipi_hashmask)];
999 LIST_FOREACH(inp, head, inp_hash) {
1000 /* XXX inp locking */
1001 if ((inp->inp_vflag & INP_IPV6) == 0)
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) {
1008 * XXX We should be able to directly return
1009 * the inp here, without any checks.
1010 * Well unless both bound with SO_REUSEPORT?
1012 if (prison_flag(inp->inp_cred, PR_IP6))
1022 * Then look for a wildcard match, if requested.
1024 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
1025 struct inpcb *local_wild = NULL, *local_exact = NULL;
1026 struct inpcb *jail_wild = NULL;
1030 * Order of socket selection - we always prefer jails.
1031 * 1. jailed, non-wild.
1033 * 3. non-jailed, non-wild.
1034 * 4. non-jailed, wild.
1036 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
1037 0, pcbinfo->ipi_hashmask)];
1038 LIST_FOREACH(inp, head, inp_hash) {
1039 /* XXX inp locking */
1040 if ((inp->inp_vflag & INP_IPV6) == 0)
1043 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
1044 inp->inp_lport != lport) {
1048 /* XXX inp locking */
1049 if (faith && (inp->inp_flags & INP_FAITH) == 0)
1052 injail = prison_flag(inp->inp_cred, PR_IP6);
1054 if (prison_check_ip6(inp->inp_cred,
1058 if (local_exact != NULL)
1062 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
1067 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1073 } /* LIST_FOREACH */
1075 if (jail_wild != NULL)
1077 if (local_exact != NULL)
1078 return (local_exact);
1079 if (local_wild != NULL)
1080 return (local_wild);
1081 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
1090 * Lookup PCB in hash list, using pcbinfo tables. This variation locks the
1091 * hash list lock, and will return the inpcb locked (i.e., requires
1092 * INPLOOKUP_LOCKPCB).
1094 static struct inpcb *
1095 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1096 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1101 INP_HASH_RLOCK(pcbinfo);
1102 inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport,
1103 (lookupflags & ~(INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)), ifp);
1106 INP_HASH_RUNLOCK(pcbinfo);
1107 if (lookupflags & INPLOOKUP_WLOCKPCB) {
1109 if (in_pcbrele_wlocked(inp))
1111 } else if (lookupflags & INPLOOKUP_RLOCKPCB) {
1113 if (in_pcbrele_rlocked(inp))
1116 panic("%s: locking bug", __func__);
1118 INP_HASH_RUNLOCK(pcbinfo);
1123 * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf
1124 * from which a pre-calculated hash value may be extracted.
1126 * Possibly more of this logic should be in in6_pcbgroup.c.
1129 in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport,
1130 struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp)
1132 #if defined(PCBGROUP)
1133 struct inpcbgroup *pcbgroup;
1136 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1137 ("%s: invalid lookup flags %d", __func__, lookupflags));
1138 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1139 ("%s: LOCKPCB not set", __func__));
1141 #if defined(PCBGROUP)
1142 if (in_pcbgroup_enabled(pcbinfo)) {
1143 pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr,
1145 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport,
1146 laddr, lport, lookupflags, ifp));
1149 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1154 in6_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1155 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1156 struct ifnet *ifp, struct mbuf *m)
1159 struct inpcbgroup *pcbgroup;
1162 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1163 ("%s: invalid lookup flags %d", __func__, lookupflags));
1164 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1165 ("%s: LOCKPCB not set", __func__));
1168 if (in_pcbgroup_enabled(pcbinfo)) {
1169 pcbgroup = in6_pcbgroup_byhash(pcbinfo, M_HASHTYPE_GET(m),
1170 m->m_pkthdr.flowid);
1171 if (pcbgroup != NULL)
1172 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr,
1173 fport, laddr, lport, lookupflags, ifp));
1174 pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr,
1176 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport,
1177 laddr, lport, lookupflags, ifp));
1180 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1185 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
1189 ip = mtod(m, struct ip6_hdr *);
1190 bzero(sin6, sizeof(*sin6));
1191 sin6->sin6_len = sizeof(*sin6);
1192 sin6->sin6_family = AF_INET6;
1193 sin6->sin6_addr = ip->ip6_src;
1195 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */