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)
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 == 0 ||
247 (t->inp_flags2 & INP_REUSEPORT) == 0)) {
251 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
252 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
253 struct sockaddr_in sin;
255 in6_sin6_2_sin(&sin, sin6);
256 t = in_pcblookup_local(pcbinfo, sin.sin_addr,
257 lport, lookupflags, cred);
258 if (t && t->inp_flags & INP_TIMEWAIT) {
262 if ((reuseport & tw->tw_so_options) == 0
263 && (ntohl(t->inp_laddr.s_addr) !=
264 INADDR_ANY || ((inp->inp_vflag &
266 (t->inp_vflag & INP_IPV6PROTO))))
268 } else if (t && (reuseport == 0 ||
269 (t->inp_flags2 & INP_REUSEPORT) == 0) &&
270 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
271 (t->inp_vflag & INP_IPV6PROTO) != 0))
276 inp->in6p_laddr = sin6->sin6_addr;
279 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) {
280 /* Undo an address bind that may have occurred. */
281 inp->in6p_laddr = in6addr_any;
285 inp->inp_lport = lport;
286 if (in_pcbinshash(inp) != 0) {
287 inp->in6p_laddr = in6addr_any;
296 * Transform old in6_pcbconnect() into an inner subroutine for new
297 * in6_pcbconnect(): Do some validity-checking on the remote
298 * address (in mbuf 'nam') and then determine local host address
299 * (i.e., which interface) to use to access that remote host.
301 * This preserves definition of in6_pcbconnect(), while supporting a
302 * slightly different version for T/TCP. (This is more than
303 * a bit of a kludge, but cleaning up the internal interfaces would
304 * have forced minor changes in every protocol).
307 in6_pcbladdr(register struct inpcb *inp, struct sockaddr *nam,
308 struct in6_addr *plocal_addr6)
310 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
312 struct ifnet *ifp = NULL;
313 int scope_ambiguous = 0;
314 struct in6_addr in6a;
316 INP_WLOCK_ASSERT(inp);
317 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */
319 if (nam->sa_len != sizeof (*sin6))
321 if (sin6->sin6_family != AF_INET6)
322 return (EAFNOSUPPORT);
323 if (sin6->sin6_port == 0)
324 return (EADDRNOTAVAIL);
326 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone)
328 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
331 if (!TAILQ_EMPTY(&V_in6_ifaddrhead)) {
333 * If the destination address is UNSPECIFIED addr,
334 * use the loopback addr, e.g ::1.
336 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
337 sin6->sin6_addr = in6addr_loopback;
339 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0)
342 error = in6_selectsrc(sin6, inp->in6p_outputopts,
343 inp, NULL, inp->inp_cred, &ifp, &in6a);
347 if (ifp && scope_ambiguous &&
348 (error = in6_setscope(&sin6->sin6_addr, ifp, NULL)) != 0) {
353 * Do not update this earlier, in case we return with an error.
355 * XXX: this in6_selectsrc result might replace the bound local
356 * address with the address specified by setsockopt(IPV6_PKTINFO).
357 * Is it the intended behavior?
359 *plocal_addr6 = in6a;
362 * Don't do pcblookup call here; return interface in
364 * and exit to caller, that will do the lookup.
372 * Connect from a socket to a specified address.
373 * Both address and port must be specified in argument sin.
374 * If don't have a local address for this socket yet,
378 in6_pcbconnect_mbuf(register struct inpcb *inp, struct sockaddr *nam,
379 struct ucred *cred, struct mbuf *m)
381 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
382 register struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
383 struct in6_addr addr6;
386 INP_WLOCK_ASSERT(inp);
387 INP_HASH_WLOCK_ASSERT(pcbinfo);
390 * Call inner routine, to assign local interface address.
391 * in6_pcbladdr() may automatically fill in sin6_scope_id.
393 if ((error = in6_pcbladdr(inp, nam, &addr6)) != 0)
396 if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr,
398 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
399 ? &addr6 : &inp->in6p_laddr,
400 inp->inp_lport, 0, NULL) != NULL) {
403 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
404 if (inp->inp_lport == 0) {
405 error = in6_pcbbind(inp, (struct sockaddr *)0, cred);
409 inp->in6p_laddr = addr6;
411 inp->in6p_faddr = sin6->sin6_addr;
412 inp->inp_fport = sin6->sin6_port;
413 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
414 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
415 if (inp->inp_flags & IN6P_AUTOFLOWLABEL)
417 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
419 in_pcbrehash_mbuf(inp, m);
425 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred)
428 return (in6_pcbconnect_mbuf(inp, nam, cred, NULL));
432 in6_pcbdisconnect(struct inpcb *inp)
435 INP_WLOCK_ASSERT(inp);
436 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
438 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
440 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
441 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
446 in6_sockaddr(in_port_t port, struct in6_addr *addr_p)
448 struct sockaddr_in6 *sin6;
450 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK);
451 bzero(sin6, sizeof *sin6);
452 sin6->sin6_family = AF_INET6;
453 sin6->sin6_len = sizeof(*sin6);
454 sin6->sin6_port = port;
455 sin6->sin6_addr = *addr_p;
456 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
458 return (struct sockaddr *)sin6;
462 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p)
464 struct sockaddr_in sin;
465 struct sockaddr_in6 *sin6_p;
467 bzero(&sin, sizeof sin);
468 sin.sin_family = AF_INET;
469 sin.sin_len = sizeof(sin);
471 sin.sin_addr = *addr_p;
473 sin6_p = malloc(sizeof *sin6_p, M_SONAME,
475 in6_sin_2_v4mapsin6(&sin, sin6_p);
477 return (struct sockaddr *)sin6_p;
481 in6_getsockaddr(struct socket *so, struct sockaddr **nam)
483 register struct inpcb *inp;
484 struct in6_addr addr;
488 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL"));
491 port = inp->inp_lport;
492 addr = inp->in6p_laddr;
495 *nam = in6_sockaddr(port, &addr);
500 in6_getpeeraddr(struct socket *so, struct sockaddr **nam)
503 struct in6_addr addr;
507 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL"));
510 port = inp->inp_fport;
511 addr = inp->in6p_faddr;
514 *nam = in6_sockaddr(port, &addr);
519 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
525 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL"));
528 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
529 error = in_getsockaddr(so, nam);
531 in6_sin_2_v4mapsin6_in_sock(nam);
535 /* scope issues will be handled in in6_getsockaddr(). */
536 error = in6_getsockaddr(so, nam);
543 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
549 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL"));
552 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
553 error = in_getpeeraddr(so, nam);
555 in6_sin_2_v4mapsin6_in_sock(nam);
558 /* scope issues will be handled in in6_getpeeraddr(). */
559 error = in6_getpeeraddr(so, nam);
565 * Pass some notification to all connections of a protocol
566 * associated with address dst. The local address and/or port numbers
567 * may be specified to limit the search. The "usual action" will be
568 * taken, depending on the ctlinput cmd. The caller must filter any
569 * cmds that are uninteresting (e.g., no error in the map).
570 * Call the protocol specific routine (if any) to report
571 * any errors for each matching socket.
574 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst,
575 u_int fport_arg, const struct sockaddr *src, u_int lport_arg,
576 int cmd, void *cmdarg,
577 struct inpcb *(*notify)(struct inpcb *, int))
579 struct inpcb *inp, *inp_temp;
580 struct sockaddr_in6 sa6_src, *sa6_dst;
581 u_short fport = fport_arg, lport = lport_arg;
585 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
588 sa6_dst = (struct sockaddr_in6 *)dst;
589 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
593 * note that src can be NULL when we get notify by local fragmentation.
595 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
596 flowinfo = sa6_src.sin6_flowinfo;
599 * Redirects go to all references to the destination,
600 * and use in6_rtchange to invalidate the route cache.
601 * Dead host indications: also use in6_rtchange to invalidate
602 * the cache, and deliver the error to all the sockets.
603 * Otherwise, if we have knowledge of the local port and address,
604 * deliver only to that socket.
606 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
609 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
611 if (cmd != PRC_HOSTDEAD)
612 notify = in6_rtchange;
614 errno = inet6ctlerrmap[cmd];
615 INP_INFO_WLOCK(pcbinfo);
616 LIST_FOREACH_SAFE(inp, pcbinfo->ipi_listhead, inp_list, inp_temp) {
618 if ((inp->inp_vflag & INP_IPV6) == 0) {
624 * If the error designates a new path MTU for a destination
625 * and the application (associated with this socket) wanted to
626 * know the value, notify. Note that we notify for all
627 * disconnected sockets if the corresponding application
628 * wanted. This is because some UDP applications keep sending
629 * sockets disconnected.
630 * XXX: should we avoid to notify the value to TCP sockets?
632 if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 &&
633 (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
634 IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) {
635 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst,
636 (u_int32_t *)cmdarg);
640 * Detect if we should notify the error. If no source and
641 * destination ports are specifed, but non-zero flowinfo and
642 * local address match, notify the error. This is the case
643 * when the error is delivered with an encrypted buffer
644 * by ESP. Otherwise, just compare addresses and ports
647 if (lport == 0 && fport == 0 && flowinfo &&
648 inp->inp_socket != NULL &&
649 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) &&
650 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
652 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
653 &sa6_dst->sin6_addr) ||
654 inp->inp_socket == 0 ||
655 (lport && inp->inp_lport != lport) ||
656 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
657 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
658 &sa6_src.sin6_addr)) ||
659 (fport && inp->inp_fport != fport)) {
666 if ((*notify)(inp, errno))
671 INP_INFO_WUNLOCK(pcbinfo);
675 * Lookup a PCB based on the local address and port. Caller must hold the
676 * hash lock. No inpcb locks or references are acquired.
679 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr,
680 u_short lport, int lookupflags, struct ucred *cred)
682 register struct inpcb *inp;
683 int matchwild = 3, wildcard;
685 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
686 ("%s: invalid lookup flags %d", __func__, lookupflags));
688 INP_HASH_WLOCK_ASSERT(pcbinfo);
690 if ((lookupflags & INPLOOKUP_WILDCARD) == 0) {
691 struct inpcbhead *head;
693 * Look for an unconnected (wildcard foreign addr) PCB that
694 * matches the local address and port we're looking for.
696 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
697 0, pcbinfo->ipi_hashmask)];
698 LIST_FOREACH(inp, head, inp_hash) {
699 /* XXX inp locking */
700 if ((inp->inp_vflag & INP_IPV6) == 0)
702 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
703 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
704 inp->inp_lport == lport) {
707 prison_equal_ip6(cred->cr_prison,
708 inp->inp_cred->cr_prison))
717 struct inpcbporthead *porthash;
718 struct inpcbport *phd;
719 struct inpcb *match = NULL;
721 * Best fit PCB lookup.
723 * First see if this local port is in use by looking on the
726 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport,
727 pcbinfo->ipi_porthashmask)];
728 LIST_FOREACH(phd, porthash, phd_hash) {
729 if (phd->phd_port == lport)
734 * Port is in use by one or more PCBs. Look for best
737 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
740 !prison_equal_ip6(cred->cr_prison,
741 inp->inp_cred->cr_prison))
743 /* XXX inp locking */
744 if ((inp->inp_vflag & INP_IPV6) == 0)
746 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
748 if (!IN6_IS_ADDR_UNSPECIFIED(
750 if (IN6_IS_ADDR_UNSPECIFIED(laddr))
752 else if (!IN6_ARE_ADDR_EQUAL(
753 &inp->in6p_laddr, laddr))
756 if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
759 if (wildcard < matchwild) {
761 matchwild = wildcard;
772 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
775 struct ip6_moptions *im6o;
778 INP_INFO_RLOCK(pcbinfo);
779 LIST_FOREACH(in6p, pcbinfo->ipi_listhead, inp_list) {
781 im6o = in6p->in6p_moptions;
782 if ((in6p->inp_vflag & INP_IPV6) && im6o != NULL) {
784 * Unselect the outgoing ifp for multicast if it
787 if (im6o->im6o_multicast_ifp == ifp)
788 im6o->im6o_multicast_ifp = NULL;
790 * Drop multicast group membership if we joined
791 * through the interface being detached.
794 for (i = 0; i < im6o->im6o_num_memberships; i++) {
795 if (im6o->im6o_membership[i]->in6m_ifp ==
797 in6_mc_leave(im6o->im6o_membership[i],
800 } else if (gap != 0) {
801 im6o->im6o_membership[i - gap] =
802 im6o->im6o_membership[i];
805 im6o->im6o_num_memberships -= gap;
809 INP_INFO_RUNLOCK(pcbinfo);
813 * Check for alternatives when higher level complains
814 * about service problems. For now, invalidate cached
815 * routing information. If the route was created dynamically
816 * (by a redirect), time to try a default gateway again.
819 in6_losing(struct inpcb *in6p)
823 * We don't store route pointers in the routing table anymore
829 * After a routing change, flush old routing
830 * and allocate a (hopefully) better one.
833 in6_rtchange(struct inpcb *inp, int errno)
836 * We don't store route pointers in the routing table anymore
843 * Lookup PCB in hash list, using pcbgroup tables.
845 static struct inpcb *
846 in6_pcblookup_group(struct inpcbinfo *pcbinfo, struct inpcbgroup *pcbgroup,
847 struct in6_addr *faddr, u_int fport_arg, struct in6_addr *laddr,
848 u_int lport_arg, int lookupflags, struct ifnet *ifp)
850 struct inpcbhead *head;
851 struct inpcb *inp, *tmpinp;
852 u_short fport = fport_arg, lport = lport_arg;
855 if (faithprefix_p != NULL)
856 faith = (*faithprefix_p)(laddr);
861 * First look for an exact match.
864 INP_GROUP_LOCK(pcbgroup);
865 head = &pcbgroup->ipg_hashbase[
866 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport,
867 pcbgroup->ipg_hashmask)];
868 LIST_FOREACH(inp, head, inp_pcbgrouphash) {
869 /* XXX inp locking */
870 if ((inp->inp_vflag & INP_IPV6) == 0)
872 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
873 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
874 inp->inp_fport == fport &&
875 inp->inp_lport == lport) {
877 * XXX We should be able to directly return
878 * the inp here, without any checks.
879 * Well unless both bound with SO_REUSEPORT?
881 if (prison_flag(inp->inp_cred, PR_IP6))
887 if (tmpinp != NULL) {
893 * Then look for a wildcard match, if requested.
895 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
896 struct inpcb *local_wild = NULL, *local_exact = NULL;
897 struct inpcb *jail_wild = NULL;
901 * Order of socket selection - we always prefer jails.
902 * 1. jailed, non-wild.
904 * 3. non-jailed, non-wild.
905 * 4. non-jailed, wild.
907 head = &pcbinfo->ipi_wildbase[INP_PCBHASH(INADDR_ANY, lport,
908 0, pcbinfo->ipi_wildmask)];
909 LIST_FOREACH(inp, head, inp_pcbgroup_wild) {
910 /* XXX inp locking */
911 if ((inp->inp_vflag & INP_IPV6) == 0)
914 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
915 inp->inp_lport != lport) {
919 /* XXX inp locking */
920 if (faith && (inp->inp_flags & INP_FAITH) == 0)
923 injail = prison_flag(inp->inp_cred, PR_IP6);
925 if (prison_check_ip6(inp->inp_cred,
929 if (local_exact != NULL)
933 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
938 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
955 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
956 INP_GROUP_UNLOCK(pcbgroup);
961 INP_GROUP_UNLOCK(pcbgroup);
962 if (lookupflags & INPLOOKUP_WLOCKPCB) {
964 if (in_pcbrele_wlocked(inp))
966 } else if (lookupflags & INPLOOKUP_RLOCKPCB) {
968 if (in_pcbrele_rlocked(inp))
971 panic("%s: locking buf", __func__);
974 #endif /* PCBGROUP */
977 * Lookup PCB in hash list.
980 in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
981 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg,
982 int lookupflags, struct ifnet *ifp)
984 struct inpcbhead *head;
985 struct inpcb *inp, *tmpinp;
986 u_short fport = fport_arg, lport = lport_arg;
989 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
990 ("%s: invalid lookup flags %d", __func__, lookupflags));
992 INP_HASH_LOCK_ASSERT(pcbinfo);
994 if (faithprefix_p != NULL)
995 faith = (*faithprefix_p)(laddr);
1000 * First look for an exact match.
1003 head = &pcbinfo->ipi_hashbase[
1004 INP_PCBHASH(faddr->s6_addr32[3] /* XXX */, lport, fport,
1005 pcbinfo->ipi_hashmask)];
1006 LIST_FOREACH(inp, head, inp_hash) {
1007 /* XXX inp locking */
1008 if ((inp->inp_vflag & INP_IPV6) == 0)
1010 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1011 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1012 inp->inp_fport == fport &&
1013 inp->inp_lport == lport) {
1015 * XXX We should be able to directly return
1016 * the inp here, without any checks.
1017 * Well unless both bound with SO_REUSEPORT?
1019 if (prison_flag(inp->inp_cred, PR_IP6))
1029 * Then look for a wildcard match, if requested.
1031 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
1032 struct inpcb *local_wild = NULL, *local_exact = NULL;
1033 struct inpcb *jail_wild = NULL;
1037 * Order of socket selection - we always prefer jails.
1038 * 1. jailed, non-wild.
1040 * 3. non-jailed, non-wild.
1041 * 4. non-jailed, wild.
1043 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(INADDR_ANY, lport,
1044 0, pcbinfo->ipi_hashmask)];
1045 LIST_FOREACH(inp, head, inp_hash) {
1046 /* XXX inp locking */
1047 if ((inp->inp_vflag & INP_IPV6) == 0)
1050 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
1051 inp->inp_lport != lport) {
1055 /* XXX inp locking */
1056 if (faith && (inp->inp_flags & INP_FAITH) == 0)
1059 injail = prison_flag(inp->inp_cred, PR_IP6);
1061 if (prison_check_ip6(inp->inp_cred,
1065 if (local_exact != NULL)
1069 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
1074 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1080 } /* LIST_FOREACH */
1082 if (jail_wild != NULL)
1084 if (local_exact != NULL)
1085 return (local_exact);
1086 if (local_wild != NULL)
1087 return (local_wild);
1088 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
1097 * Lookup PCB in hash list, using pcbinfo tables. This variation locks the
1098 * hash list lock, and will return the inpcb locked (i.e., requires
1099 * INPLOOKUP_LOCKPCB).
1101 static struct inpcb *
1102 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1103 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1108 INP_HASH_RLOCK(pcbinfo);
1109 inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport,
1110 (lookupflags & ~(INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)), ifp);
1113 INP_HASH_RUNLOCK(pcbinfo);
1114 if (lookupflags & INPLOOKUP_WLOCKPCB) {
1116 if (in_pcbrele_wlocked(inp))
1118 } else if (lookupflags & INPLOOKUP_RLOCKPCB) {
1120 if (in_pcbrele_rlocked(inp))
1123 panic("%s: locking bug", __func__);
1125 INP_HASH_RUNLOCK(pcbinfo);
1130 * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf
1131 * from which a pre-calculated hash value may be extracted.
1133 * Possibly more of this logic should be in in6_pcbgroup.c.
1136 in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport,
1137 struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp)
1139 #if defined(PCBGROUP)
1140 struct inpcbgroup *pcbgroup;
1143 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1144 ("%s: invalid lookup flags %d", __func__, lookupflags));
1145 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1146 ("%s: LOCKPCB not set", __func__));
1148 #if defined(PCBGROUP)
1149 if (in_pcbgroup_enabled(pcbinfo)) {
1150 pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr,
1152 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport,
1153 laddr, lport, lookupflags, ifp));
1156 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1161 in6_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1162 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1163 struct ifnet *ifp, struct mbuf *m)
1166 struct inpcbgroup *pcbgroup;
1169 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1170 ("%s: invalid lookup flags %d", __func__, lookupflags));
1171 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1172 ("%s: LOCKPCB not set", __func__));
1175 if (in_pcbgroup_enabled(pcbinfo)) {
1176 pcbgroup = in6_pcbgroup_byhash(pcbinfo, M_HASHTYPE_GET(m),
1177 m->m_pkthdr.flowid);
1178 if (pcbgroup != NULL)
1179 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr,
1180 fport, laddr, lport, lookupflags, ifp));
1181 pcbgroup = in6_pcbgroup_bytuple(pcbinfo, laddr, lport, faddr,
1183 return (in6_pcblookup_group(pcbinfo, pcbgroup, faddr, fport,
1184 laddr, lport, lookupflags, ifp));
1187 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1192 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
1196 ip = mtod(m, struct ip6_hdr *);
1197 bzero(sin6, sizeof(*sin6));
1198 sin6->sin6_len = sizeof(*sin6);
1199 sin6->sin6_family = AF_INET6;
1200 sin6->sin6_addr = ip->ip6_src;
1202 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */