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_route.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>
100 #include <net/route/nhop.h>
102 #include <netinet/in.h>
103 #include <netinet/in_var.h>
104 #include <netinet/in_systm.h>
105 #include <netinet/tcp_var.h>
106 #include <netinet/ip6.h>
107 #include <netinet/ip_var.h>
109 #include <netinet6/ip6_var.h>
110 #include <netinet6/nd6.h>
111 #include <netinet/in_pcb.h>
112 #include <netinet/in_pcb_var.h>
113 #include <netinet6/in6_pcb.h>
114 #include <netinet6/in6_fib.h>
115 #include <netinet6/scope6_var.h>
118 in6_pcbsetport(struct in6_addr *laddr, struct inpcb *inp, struct ucred *cred)
120 struct socket *so = inp->inp_socket;
122 int error, lookupflags = 0;
124 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
127 INP_WLOCK_ASSERT(inp);
128 INP_HASH_WLOCK_ASSERT(pcbinfo);
130 error = prison_local_ip6(cred, laddr,
131 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0));
135 /* XXX: this is redundant when called from in6_pcbbind */
136 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT|SO_REUSEPORT_LB)) == 0)
137 lookupflags = INPLOOKUP_WILDCARD;
139 inp->inp_flags |= INP_ANONPORT;
141 error = in_pcb_lport(inp, NULL, &lport, cred, lookupflags);
145 inp->inp_lport = lport;
146 if (in_pcbinshash(inp) != 0) {
147 inp->in6p_laddr = in6addr_any;
156 in6_pcbbind(struct inpcb *inp, struct sockaddr *nam,
159 struct socket *so = inp->inp_socket;
160 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
161 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
163 int error, lookupflags = 0;
164 int reuseport = (so->so_options & SO_REUSEPORT);
167 * XXX: Maybe we could let SO_REUSEPORT_LB set SO_REUSEPORT bit here
168 * so that we don't have to add to the (already messy) code below.
170 int reuseport_lb = (so->so_options & SO_REUSEPORT_LB);
172 INP_WLOCK_ASSERT(inp);
173 INP_HASH_WLOCK_ASSERT(pcbinfo);
175 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
177 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT|SO_REUSEPORT_LB)) == 0)
178 lookupflags = INPLOOKUP_WILDCARD;
180 if ((error = prison_local_ip6(cred, &inp->in6p_laddr,
181 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
184 sin6 = (struct sockaddr_in6 *)nam;
185 KASSERT(sin6->sin6_family == AF_INET6,
186 ("%s: invalid address family for %p", __func__, sin6));
187 KASSERT(sin6->sin6_len == sizeof(*sin6),
188 ("%s: invalid address length for %p", __func__, sin6));
190 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
193 if ((error = prison_local_ip6(cred, &sin6->sin6_addr,
194 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
197 lport = sin6->sin6_port;
198 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
200 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
201 * allow compepte duplication of binding if
202 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
203 * and a multicast address is bound on both
204 * new and duplicated sockets.
206 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) != 0)
207 reuseport = SO_REUSEADDR|SO_REUSEPORT;
209 * XXX: How to deal with SO_REUSEPORT_LB here?
210 * Treat same as SO_REUSEPORT for now.
212 if ((so->so_options &
213 (SO_REUSEADDR|SO_REUSEPORT_LB)) != 0)
214 reuseport_lb = SO_REUSEADDR|SO_REUSEPORT_LB;
215 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
216 struct epoch_tracker et;
219 sin6->sin6_port = 0; /* yech... */
221 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) ==
223 (inp->inp_flags & INP_BINDANY) == 0) {
225 return (EADDRNOTAVAIL);
229 * XXX: bind to an anycast address might accidentally
230 * cause sending a packet with anycast source address.
231 * We should allow to bind to a deprecated address, since
232 * the application dares to use it.
235 ((struct in6_ifaddr *)ifa)->ia6_flags &
236 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) {
238 return (EADDRNOTAVAIL);
247 if (ntohs(lport) <= V_ipport_reservedhigh &&
248 ntohs(lport) >= V_ipport_reservedlow &&
249 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT))
251 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) &&
252 priv_check_cred(inp->inp_cred, PRIV_NETINET_REUSEPORT) != 0) {
253 t = in6_pcblookup_local(pcbinfo,
254 &sin6->sin6_addr, lport,
255 INPLOOKUP_WILDCARD, cred);
257 ((inp->inp_flags2 & INP_BINDMULTI) == 0) &&
258 ((t->inp_flags & INP_TIMEWAIT) == 0) &&
259 (so->so_type != SOCK_STREAM ||
260 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) &&
261 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
262 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
263 (t->inp_flags2 & INP_REUSEPORT) ||
264 (t->inp_flags2 & INP_REUSEPORT_LB) == 0) &&
265 (inp->inp_cred->cr_uid !=
266 t->inp_cred->cr_uid))
270 * If the socket is a BINDMULTI socket, then
271 * the credentials need to match and the
272 * original socket also has to have been bound
275 if (t && (! in_pcbbind_check_bindmulti(inp, t)))
279 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
280 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
281 struct sockaddr_in sin;
283 in6_sin6_2_sin(&sin, sin6);
284 t = in_pcblookup_local(pcbinfo,
286 INPLOOKUP_WILDCARD, cred);
288 ((inp->inp_flags2 & INP_BINDMULTI) == 0) &&
290 INP_TIMEWAIT) == 0) &&
291 (so->so_type != SOCK_STREAM ||
292 ntohl(t->inp_faddr.s_addr) ==
294 (inp->inp_cred->cr_uid !=
295 t->inp_cred->cr_uid))
298 if (t && (! in_pcbbind_check_bindmulti(inp, t)))
303 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr,
304 lport, lookupflags, cred);
305 if (t && (t->inp_flags & INP_TIMEWAIT)) {
307 * XXXRW: If an incpb has had its timewait
308 * state recycled, we treat the address as
309 * being in use (for now). This is better
310 * than a panic, but not desirable.
314 ((reuseport & tw->tw_so_options) == 0 &&
315 (reuseport_lb & tw->tw_so_options) == 0))
317 } else if (t && (reuseport & inp_so_options(t)) == 0 &&
318 (reuseport_lb & inp_so_options(t)) == 0) {
322 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
323 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
324 struct sockaddr_in sin;
326 in6_sin6_2_sin(&sin, sin6);
327 t = in_pcblookup_local(pcbinfo, sin.sin_addr,
328 lport, lookupflags, cred);
329 if (t && t->inp_flags & INP_TIMEWAIT) {
333 if ((reuseport & tw->tw_so_options) == 0
334 && (reuseport_lb & tw->tw_so_options) == 0
335 && (ntohl(t->inp_laddr.s_addr) !=
336 INADDR_ANY || ((inp->inp_vflag &
338 (t->inp_vflag & INP_IPV6PROTO))))
341 (reuseport & inp_so_options(t)) == 0 &&
342 (reuseport_lb & inp_so_options(t)) == 0 &&
343 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
344 (t->inp_vflag & INP_IPV6PROTO) != 0)) {
350 inp->in6p_laddr = sin6->sin6_addr;
353 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) {
354 /* Undo an address bind that may have occurred. */
355 inp->in6p_laddr = in6addr_any;
359 inp->inp_lport = lport;
360 if (in_pcbinshash(inp) != 0) {
361 inp->in6p_laddr = in6addr_any;
370 * Transform old in6_pcbconnect() into an inner subroutine for new
371 * in6_pcbconnect(): Do some validity-checking on the remote
372 * address (in mbuf 'nam') and then determine local host address
373 * (i.e., which interface) to use to access that remote host.
375 * This preserves definition of in6_pcbconnect(), while supporting a
376 * slightly different version for T/TCP. (This is more than
377 * a bit of a kludge, but cleaning up the internal interfaces would
378 * have forced minor changes in every protocol).
381 in6_pcbladdr(struct inpcb *inp, struct sockaddr_in6 *sin6,
382 struct in6_addr *plocal_addr6)
385 int scope_ambiguous = 0;
386 struct in6_addr in6a;
387 struct epoch_tracker et;
389 INP_WLOCK_ASSERT(inp);
390 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */
392 if (sin6->sin6_port == 0)
393 return (EADDRNOTAVAIL);
395 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone)
397 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
400 if (!CK_STAILQ_EMPTY(&V_in6_ifaddrhead)) {
402 * If the destination address is UNSPECIFIED addr,
403 * use the loopback addr, e.g ::1.
405 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
406 sin6->sin6_addr = in6addr_loopback;
408 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0)
412 error = in6_selectsrc_socket(sin6, inp->in6p_outputopts,
413 inp, inp->inp_cred, scope_ambiguous, &in6a, NULL);
419 * Do not update this earlier, in case we return with an error.
421 * XXX: this in6_selectsrc_socket result might replace the bound local
422 * address with the address specified by setsockopt(IPV6_PKTINFO).
423 * Is it the intended behavior?
425 *plocal_addr6 = in6a;
428 * Don't do pcblookup call here; return interface in
430 * and exit to caller, that will do the lookup.
438 * Connect from a socket to a specified address.
439 * Both address and port must be specified in argument sin.
440 * If don't have a local address for this socket yet,
444 in6_pcbconnect_mbuf(struct inpcb *inp, struct sockaddr *nam,
445 struct ucred *cred, struct mbuf *m, bool rehash)
447 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
448 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
449 struct sockaddr_in6 laddr6;
452 KASSERT(sin6->sin6_family == AF_INET6,
453 ("%s: invalid address family for %p", __func__, sin6));
454 KASSERT(sin6->sin6_len == sizeof(*sin6),
455 ("%s: invalid address length for %p", __func__, sin6));
457 bzero(&laddr6, sizeof(laddr6));
458 laddr6.sin6_family = AF_INET6;
460 INP_WLOCK_ASSERT(inp);
461 INP_HASH_WLOCK_ASSERT(pcbinfo);
464 if (CALC_FLOWID_OUTBOUND) {
465 uint32_t hash_type, hash_val;
467 hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
468 &sin6->sin6_addr, 0, sin6->sin6_port,
469 inp->inp_socket->so_proto->pr_protocol, &hash_type);
470 inp->inp_flowid = hash_val;
471 inp->inp_flowtype = hash_type;
475 * Call inner routine, to assign local interface address.
476 * in6_pcbladdr() may automatically fill in sin6_scope_id.
478 if ((error = in6_pcbladdr(inp, sin6, &laddr6.sin6_addr)) != 0)
481 if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr,
483 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
484 ? &laddr6.sin6_addr : &inp->in6p_laddr,
485 inp->inp_lport, 0, NULL, M_NODOM) != NULL) {
488 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
489 if (inp->inp_lport == 0) {
491 * rehash was required to be true in the past for
492 * this case; retain that convention. However,
493 * we now call in_pcb_lport_dest rather than
494 * in6_pcbbind; the former does not insert into
495 * the hash table, the latter does. Change rehash
496 * to false to do the in_pcbinshash below.
498 KASSERT(rehash == true,
499 ("Rehashing required for unbound inps"));
501 error = in_pcb_lport_dest(inp,
502 (struct sockaddr *) &laddr6, &inp->inp_lport,
503 (struct sockaddr *) sin6, sin6->sin6_port, cred,
508 inp->in6p_laddr = laddr6.sin6_addr;
510 inp->in6p_faddr = sin6->sin6_addr;
511 inp->inp_fport = sin6->sin6_port;
512 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
513 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
514 if (inp->inp_flags & IN6P_AUTOFLOWLABEL)
516 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
528 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred)
531 return (in6_pcbconnect_mbuf(inp, nam, cred, NULL, true));
535 in6_pcbdisconnect(struct inpcb *inp)
538 INP_WLOCK_ASSERT(inp);
539 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
541 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
543 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
544 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
549 in6_sockaddr(in_port_t port, struct in6_addr *addr_p)
551 struct sockaddr_in6 *sin6;
553 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK);
554 bzero(sin6, sizeof *sin6);
555 sin6->sin6_family = AF_INET6;
556 sin6->sin6_len = sizeof(*sin6);
557 sin6->sin6_port = port;
558 sin6->sin6_addr = *addr_p;
559 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
561 return (struct sockaddr *)sin6;
565 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p)
567 struct sockaddr_in sin;
568 struct sockaddr_in6 *sin6_p;
570 bzero(&sin, sizeof sin);
571 sin.sin_family = AF_INET;
572 sin.sin_len = sizeof(sin);
574 sin.sin_addr = *addr_p;
576 sin6_p = malloc(sizeof *sin6_p, M_SONAME,
578 in6_sin_2_v4mapsin6(&sin, sin6_p);
580 return (struct sockaddr *)sin6_p;
584 in6_getsockaddr(struct socket *so, struct sockaddr **nam)
587 struct in6_addr addr;
591 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL"));
594 port = inp->inp_lport;
595 addr = inp->in6p_laddr;
598 *nam = in6_sockaddr(port, &addr);
603 in6_getpeeraddr(struct socket *so, struct sockaddr **nam)
606 struct in6_addr addr;
610 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL"));
613 port = inp->inp_fport;
614 addr = inp->in6p_faddr;
617 *nam = in6_sockaddr(port, &addr);
622 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
628 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL"));
631 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
632 error = in_getsockaddr(so, nam);
634 in6_sin_2_v4mapsin6_in_sock(nam);
638 /* scope issues will be handled in in6_getsockaddr(). */
639 error = in6_getsockaddr(so, nam);
646 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
652 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL"));
655 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
656 error = in_getpeeraddr(so, nam);
658 in6_sin_2_v4mapsin6_in_sock(nam);
661 /* scope issues will be handled in in6_getpeeraddr(). */
662 error = in6_getpeeraddr(so, nam);
668 * Pass some notification to all connections of a protocol
669 * associated with address dst. The local address and/or port numbers
670 * may be specified to limit the search. The "usual action" will be
671 * taken, depending on the ctlinput cmd. The caller must filter any
672 * cmds that are uninteresting (e.g., no error in the map).
673 * Call the protocol specific routine (if any) to report
674 * any errors for each matching socket.
677 inp_match6(const struct inpcb *inp, void *v __unused)
680 return ((inp->inp_vflag & INP_IPV6) != 0);
683 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst,
684 u_int fport_arg, const struct sockaddr *src, u_int lport_arg,
685 int cmd, void *cmdarg,
686 struct inpcb *(*notify)(struct inpcb *, int))
688 struct inpcb_iterator inpi = INP_ITERATOR(pcbinfo, INPLOOKUP_WLOCKPCB,
691 struct sockaddr_in6 sa6_src, *sa6_dst;
692 u_short fport = fport_arg, lport = lport_arg;
696 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
699 sa6_dst = (struct sockaddr_in6 *)dst;
700 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
704 * note that src can be NULL when we get notify by local fragmentation.
706 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
707 flowinfo = sa6_src.sin6_flowinfo;
710 * Redirects go to all references to the destination,
711 * and use in6_rtchange to invalidate the route cache.
712 * Dead host indications: also use in6_rtchange to invalidate
713 * the cache, and deliver the error to all the sockets.
714 * Otherwise, if we have knowledge of the local port and address,
715 * deliver only to that socket.
717 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
720 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
722 if (cmd != PRC_HOSTDEAD)
723 notify = in6_rtchange;
725 errno = inet6ctlerrmap[cmd];
726 while ((inp = inp_next(&inpi)) != NULL) {
727 INP_WLOCK_ASSERT(inp);
729 * If the error designates a new path MTU for a destination
730 * and the application (associated with this socket) wanted to
731 * know the value, notify.
732 * XXX: should we avoid to notify the value to TCP sockets?
734 if (cmd == PRC_MSGSIZE && cmdarg != NULL)
735 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst,
736 *(u_int32_t *)cmdarg);
739 * Detect if we should notify the error. If no source and
740 * destination ports are specified, but non-zero flowinfo and
741 * local address match, notify the error. This is the case
742 * when the error is delivered with an encrypted buffer
743 * by ESP. Otherwise, just compare addresses and ports
746 if (lport == 0 && fport == 0 && flowinfo &&
747 inp->inp_socket != NULL &&
748 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) &&
749 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
751 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
752 &sa6_dst->sin6_addr) ||
753 inp->inp_socket == 0 ||
754 (lport && inp->inp_lport != lport) ||
755 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
756 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
757 &sa6_src.sin6_addr)) ||
758 (fport && inp->inp_fport != fport)) {
764 (*notify)(inp, errno);
769 * Lookup a PCB based on the local address and port. Caller must hold the
770 * hash lock. No inpcb locks or references are acquired.
773 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr,
774 u_short lport, int lookupflags, struct ucred *cred)
777 int matchwild = 3, wildcard;
779 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
780 ("%s: invalid lookup flags %d", __func__, lookupflags));
782 INP_HASH_LOCK_ASSERT(pcbinfo);
784 if ((lookupflags & INPLOOKUP_WILDCARD) == 0) {
785 struct inpcbhead *head;
787 * Look for an unconnected (wildcard foreign addr) PCB that
788 * matches the local address and port we're looking for.
790 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(
791 INP6_PCBHASHKEY(&in6addr_any), lport, 0,
792 pcbinfo->ipi_hashmask)];
793 CK_LIST_FOREACH(inp, head, inp_hash) {
794 /* XXX inp locking */
795 if ((inp->inp_vflag & INP_IPV6) == 0)
797 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
798 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
799 inp->inp_lport == lport) {
802 prison_equal_ip6(cred->cr_prison,
803 inp->inp_cred->cr_prison))
812 struct inpcbporthead *porthash;
813 struct inpcbport *phd;
814 struct inpcb *match = NULL;
816 * Best fit PCB lookup.
818 * First see if this local port is in use by looking on the
821 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport,
822 pcbinfo->ipi_porthashmask)];
823 CK_LIST_FOREACH(phd, porthash, phd_hash) {
824 if (phd->phd_port == lport)
829 * Port is in use by one or more PCBs. Look for best
832 CK_LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
835 !prison_equal_ip6(cred->cr_prison,
836 inp->inp_cred->cr_prison))
838 /* XXX inp locking */
839 if ((inp->inp_vflag & INP_IPV6) == 0)
841 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
843 if (!IN6_IS_ADDR_UNSPECIFIED(
845 if (IN6_IS_ADDR_UNSPECIFIED(laddr))
847 else if (!IN6_ARE_ADDR_EQUAL(
848 &inp->in6p_laddr, laddr))
851 if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
854 if (wildcard < matchwild) {
856 matchwild = wildcard;
867 in6_multi_match(const struct inpcb *inp, void *v __unused)
870 if ((inp->inp_vflag & INP_IPV6) && inp->in6p_moptions != NULL)
877 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
879 struct inpcb_iterator inpi = INP_ITERATOR(pcbinfo, INPLOOKUP_RLOCKPCB,
880 in6_multi_match, NULL);
882 struct in6_multi *inm;
883 struct in6_mfilter *imf;
884 struct ip6_moptions *im6o;
886 IN6_MULTI_LOCK_ASSERT();
888 while ((inp = inp_next(&inpi)) != NULL) {
889 INP_RLOCK_ASSERT(inp);
891 im6o = inp->in6p_moptions;
893 * Unselect the outgoing ifp for multicast if it
896 if (im6o->im6o_multicast_ifp == ifp)
897 im6o->im6o_multicast_ifp = NULL;
899 * Drop multicast group membership if we joined
900 * through the interface being detached.
903 IP6_MFILTER_FOREACH(imf, &im6o->im6o_head) {
904 if ((inm = imf->im6f_in6m) == NULL)
906 if (inm->in6m_ifp != ifp)
908 ip6_mfilter_remove(&im6o->im6o_head, imf);
909 in6_leavegroup_locked(inm, NULL);
910 ip6_mfilter_free(imf);
917 * Check for alternatives when higher level complains
918 * about service problems. For now, invalidate cached
919 * routing information. If the route was created dynamically
920 * (by a redirect), time to try a default gateway again.
923 in6_losing(struct inpcb *inp)
926 RO_INVALIDATE_CACHE(&inp->inp_route6);
930 * After a routing change, flush old routing
931 * and allocate a (hopefully) better one.
934 in6_rtchange(struct inpcb *inp, int errno __unused)
937 RO_INVALIDATE_CACHE(&inp->inp_route6);
941 static struct inpcb *
942 in6_pcblookup_lbgroup(const struct inpcbinfo *pcbinfo,
943 const struct in6_addr *laddr, uint16_t lport, const struct in6_addr *faddr,
944 uint16_t fport, int lookupflags, uint8_t numa_domain)
946 struct inpcb *local_wild, *numa_wild;
947 const struct inpcblbgrouphead *hdr;
948 struct inpcblbgroup *grp;
951 INP_HASH_LOCK_ASSERT(pcbinfo);
953 hdr = &pcbinfo->ipi_lbgrouphashbase[
954 INP_PCBPORTHASH(lport, pcbinfo->ipi_lbgrouphashmask)];
957 * Order of socket selection:
959 * 2. wild (if lookupflags contains INPLOOKUP_WILDCARD).
962 * - Load balanced group does not contain jailed sockets.
963 * - Load balanced does not contain IPv4 mapped INET6 wild sockets.
967 CK_LIST_FOREACH(grp, hdr, il_list) {
969 if (!(grp->il_vflag & INP_IPV6))
972 if (grp->il_lport != lport)
975 idx = INP_PCBLBGROUP_PKTHASH(INP6_PCBHASHKEY(faddr), lport,
976 fport) % grp->il_inpcnt;
977 if (IN6_ARE_ADDR_EQUAL(&grp->il6_laddr, laddr)) {
978 if (numa_domain == M_NODOM ||
979 grp->il_numa_domain == numa_domain) {
980 return (grp->il_inp[idx]);
983 numa_wild = grp->il_inp[idx];
985 if (IN6_IS_ADDR_UNSPECIFIED(&grp->il6_laddr) &&
986 (lookupflags & INPLOOKUP_WILDCARD) != 0 &&
987 (local_wild == NULL || numa_domain == M_NODOM ||
988 grp->il_numa_domain == numa_domain)) {
989 local_wild = grp->il_inp[idx];
992 if (numa_wild != NULL)
998 * Lookup PCB in hash list. Used in in_pcb.c as well as here.
1001 in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1002 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg,
1003 int lookupflags, struct ifnet *ifp, uint8_t numa_domain)
1005 struct inpcbhead *head;
1006 struct inpcb *inp, *tmpinp;
1007 u_short fport = fport_arg, lport = lport_arg;
1009 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
1010 ("%s: invalid lookup flags %d", __func__, lookupflags));
1012 INP_HASH_LOCK_ASSERT(pcbinfo);
1015 * First look for an exact match.
1018 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(
1019 INP6_PCBHASHKEY(faddr), lport, fport, pcbinfo->ipi_hashmask)];
1020 CK_LIST_FOREACH(inp, head, inp_hash) {
1021 /* XXX inp locking */
1022 if ((inp->inp_vflag & INP_IPV6) == 0)
1024 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1025 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1026 inp->inp_fport == fport &&
1027 inp->inp_lport == lport) {
1029 * XXX We should be able to directly return
1030 * the inp here, without any checks.
1031 * Well unless both bound with SO_REUSEPORT?
1033 if (prison_flag(inp->inp_cred, PR_IP6))
1043 * Then look in lb group (for wildcard match).
1045 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
1046 inp = in6_pcblookup_lbgroup(pcbinfo, laddr, lport, faddr,
1047 fport, lookupflags, numa_domain);
1053 * Then look for a wildcard match, if requested.
1055 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
1056 struct inpcb *local_wild = NULL, *local_exact = NULL;
1057 struct inpcb *jail_wild = NULL;
1061 * Order of socket selection - we always prefer jails.
1062 * 1. jailed, non-wild.
1064 * 3. non-jailed, non-wild.
1065 * 4. non-jailed, wild.
1067 head = &pcbinfo->ipi_hashbase[INP_PCBHASH(
1068 INP6_PCBHASHKEY(&in6addr_any), lport, 0,
1069 pcbinfo->ipi_hashmask)];
1070 CK_LIST_FOREACH(inp, head, inp_hash) {
1071 /* XXX inp locking */
1072 if ((inp->inp_vflag & INP_IPV6) == 0)
1075 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
1076 inp->inp_lport != lport) {
1080 injail = prison_flag(inp->inp_cred, PR_IP6);
1082 if (prison_check_ip6(inp->inp_cred,
1086 if (local_exact != NULL)
1090 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
1095 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1101 } /* LIST_FOREACH */
1103 if (jail_wild != NULL)
1105 if (local_exact != NULL)
1106 return (local_exact);
1107 if (local_wild != NULL)
1108 return (local_wild);
1109 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
1118 * Lookup PCB in hash list, using pcbinfo tables. This variation locks the
1119 * hash list lock, and will return the inpcb locked (i.e., requires
1120 * INPLOOKUP_LOCKPCB).
1122 static struct inpcb *
1123 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1124 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1125 struct ifnet *ifp, uint8_t numa_domain)
1129 smr_enter(pcbinfo->ipi_smr);
1130 inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport,
1131 lookupflags & INPLOOKUP_WILDCARD, ifp, numa_domain);
1133 if (__predict_false(inp_smr_lock(inp,
1134 (lookupflags & INPLOOKUP_LOCKMASK)) == false))
1137 smr_exit(pcbinfo->ipi_smr);
1143 * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf
1144 * from which a pre-calculated hash value may be extracted.
1147 in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport,
1148 struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp)
1151 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1152 ("%s: invalid lookup flags %d", __func__, lookupflags));
1153 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1154 ("%s: LOCKPCB not set", __func__));
1156 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1157 lookupflags, ifp, M_NODOM));
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 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1167 ("%s: invalid lookup flags %d", __func__, lookupflags));
1168 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1169 ("%s: LOCKPCB not set", __func__));
1171 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1172 lookupflags, ifp, m->m_pkthdr.numa_domain));
1176 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m, int srcordst)
1180 ip = mtod(m, struct ip6_hdr *);
1181 bzero(sin6, sizeof(*sin6));
1182 sin6->sin6_len = sizeof(*sin6);
1183 sin6->sin6_family = AF_INET6;
1184 sin6->sin6_addr = srcordst ? ip->ip6_dst : ip->ip6_src;
1186 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */