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"
79 #include <sys/param.h>
80 #include <sys/systm.h>
81 #include <sys/malloc.h>
83 #include <sys/domain.h>
84 #include <sys/protosw.h>
85 #include <sys/socket.h>
86 #include <sys/socketvar.h>
87 #include <sys/sockio.h>
88 #include <sys/errno.h>
97 #include <net/if_var.h>
98 #include <net/if_llatbl.h>
99 #include <net/if_types.h>
100 #include <net/route.h>
101 #include <net/route/nhop.h>
103 #include <netinet/in.h>
104 #include <netinet/in_var.h>
105 #include <netinet/in_systm.h>
106 #include <netinet/tcp_var.h>
107 #include <netinet/ip6.h>
108 #include <netinet/ip_var.h>
110 #include <netinet6/ip6_var.h>
111 #include <netinet6/nd6.h>
112 #include <netinet/in_pcb.h>
113 #include <netinet/in_pcb_var.h>
114 #include <netinet6/in6_pcb.h>
115 #include <netinet6/in6_fib.h>
116 #include <netinet6/scope6_var.h>
119 in6_pcbsetport(struct in6_addr *laddr, struct inpcb *inp, struct ucred *cred)
121 struct socket *so = inp->inp_socket;
123 int error, lookupflags = 0;
125 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
128 INP_WLOCK_ASSERT(inp);
129 INP_HASH_WLOCK_ASSERT(pcbinfo);
131 error = prison_local_ip6(cred, laddr,
132 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0));
136 /* XXX: this is redundant when called from in6_pcbbind */
137 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT|SO_REUSEPORT_LB)) == 0)
138 lookupflags = INPLOOKUP_WILDCARD;
140 inp->inp_flags |= INP_ANONPORT;
142 error = in_pcb_lport(inp, NULL, &lport, cred, lookupflags);
146 inp->inp_lport = lport;
147 if (in_pcbinshash(inp) != 0) {
148 inp->in6p_laddr = in6addr_any;
157 in6_pcbbind(struct inpcb *inp, struct sockaddr *nam,
160 struct socket *so = inp->inp_socket;
161 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)NULL;
162 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
164 int error, lookupflags = 0;
165 int reuseport = (so->so_options & SO_REUSEPORT);
168 * XXX: Maybe we could let SO_REUSEPORT_LB set SO_REUSEPORT bit here
169 * so that we don't have to add to the (already messy) code below.
171 int reuseport_lb = (so->so_options & SO_REUSEPORT_LB);
173 INP_WLOCK_ASSERT(inp);
174 INP_HASH_WLOCK_ASSERT(pcbinfo);
176 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
178 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT|SO_REUSEPORT_LB)) == 0)
179 lookupflags = INPLOOKUP_WILDCARD;
181 if ((error = prison_local_ip6(cred, &inp->in6p_laddr,
182 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
185 sin6 = (struct sockaddr_in6 *)nam;
186 KASSERT(sin6->sin6_family == AF_INET6,
187 ("%s: invalid address family for %p", __func__, sin6));
188 KASSERT(sin6->sin6_len == sizeof(*sin6),
189 ("%s: invalid address length for %p", __func__, sin6));
191 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
194 if ((error = prison_local_ip6(cred, &sin6->sin6_addr,
195 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
198 lport = sin6->sin6_port;
199 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
201 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
202 * allow compepte duplication of binding if
203 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
204 * and a multicast address is bound on both
205 * new and duplicated sockets.
207 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) != 0)
208 reuseport = SO_REUSEADDR|SO_REUSEPORT;
210 * XXX: How to deal with SO_REUSEPORT_LB here?
211 * Treat same as SO_REUSEPORT for now.
213 if ((so->so_options &
214 (SO_REUSEADDR|SO_REUSEPORT_LB)) != 0)
215 reuseport_lb = SO_REUSEADDR|SO_REUSEPORT_LB;
216 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
217 struct epoch_tracker et;
220 sin6->sin6_port = 0; /* yech... */
222 if ((ifa = ifa_ifwithaddr((struct sockaddr *)sin6)) ==
224 (inp->inp_flags & INP_BINDANY) == 0) {
226 return (EADDRNOTAVAIL);
230 * XXX: bind to an anycast address might accidentally
231 * cause sending a packet with anycast source address.
232 * We should allow to bind to a deprecated address, since
233 * the application dares to use it.
236 ((struct in6_ifaddr *)ifa)->ia6_flags &
237 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED)) {
239 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 (so->so_type != SOCK_STREAM ||
259 IN6_IS_ADDR_UNSPECIFIED(&t->in6p_faddr)) &&
260 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
261 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
262 (t->inp_flags2 & INP_REUSEPORT) ||
263 (t->inp_flags2 & INP_REUSEPORT_LB) == 0) &&
264 (inp->inp_cred->cr_uid !=
265 t->inp_cred->cr_uid))
269 * If the socket is a BINDMULTI socket, then
270 * the credentials need to match and the
271 * original socket also has to have been bound
274 if (t && (! in_pcbbind_check_bindmulti(inp, t)))
278 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
279 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
280 struct sockaddr_in sin;
282 in6_sin6_2_sin(&sin, sin6);
283 t = in_pcblookup_local(pcbinfo,
285 INPLOOKUP_WILDCARD, cred);
287 ((inp->inp_flags2 & INP_BINDMULTI) == 0) &&
288 (so->so_type != SOCK_STREAM ||
289 ntohl(t->inp_faddr.s_addr) ==
291 (inp->inp_cred->cr_uid !=
292 t->inp_cred->cr_uid))
295 if (t && (! in_pcbbind_check_bindmulti(inp, t)))
300 t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr,
301 lport, lookupflags, cred);
302 if (t && (reuseport & inp_so_options(t)) == 0 &&
303 (reuseport_lb & inp_so_options(t)) == 0) {
307 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
308 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
309 struct sockaddr_in sin;
311 in6_sin6_2_sin(&sin, sin6);
312 t = in_pcblookup_local(pcbinfo, sin.sin_addr,
313 lport, lookupflags, cred);
315 (reuseport & inp_so_options(t)) == 0 &&
316 (reuseport_lb & inp_so_options(t)) == 0 &&
317 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
318 (t->inp_vflag & INP_IPV6PROTO) != 0)) {
324 inp->in6p_laddr = sin6->sin6_addr;
327 if ((error = in6_pcbsetport(&inp->in6p_laddr, inp, cred)) != 0) {
328 /* Undo an address bind that may have occurred. */
329 inp->in6p_laddr = in6addr_any;
333 inp->inp_lport = lport;
334 if (in_pcbinshash(inp) != 0) {
335 inp->in6p_laddr = in6addr_any;
344 * Transform old in6_pcbconnect() into an inner subroutine for new
345 * in6_pcbconnect(): Do some validity-checking on the remote
346 * address (in mbuf 'nam') and then determine local host address
347 * (i.e., which interface) to use to access that remote host.
349 * This preserves definition of in6_pcbconnect(), while supporting a
350 * slightly different version for T/TCP. (This is more than
351 * a bit of a kludge, but cleaning up the internal interfaces would
352 * have forced minor changes in every protocol).
355 in6_pcbladdr(struct inpcb *inp, struct sockaddr_in6 *sin6,
356 struct in6_addr *plocal_addr6)
359 int scope_ambiguous = 0;
360 struct in6_addr in6a;
361 struct epoch_tracker et;
363 INP_WLOCK_ASSERT(inp);
364 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo); /* XXXRW: why? */
366 if (sin6->sin6_port == 0)
367 return (EADDRNOTAVAIL);
369 if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone)
371 if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
374 if (!CK_STAILQ_EMPTY(&V_in6_ifaddrhead)) {
376 * If the destination address is UNSPECIFIED addr,
377 * use the loopback addr, e.g ::1.
379 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
380 sin6->sin6_addr = in6addr_loopback;
382 if ((error = prison_remote_ip6(inp->inp_cred, &sin6->sin6_addr)) != 0)
386 error = in6_selectsrc_socket(sin6, inp->in6p_outputopts,
387 inp, inp->inp_cred, scope_ambiguous, &in6a, NULL);
393 * Do not update this earlier, in case we return with an error.
395 * XXX: this in6_selectsrc_socket result might replace the bound local
396 * address with the address specified by setsockopt(IPV6_PKTINFO).
397 * Is it the intended behavior?
399 *plocal_addr6 = in6a;
402 * Don't do pcblookup call here; return interface in
404 * and exit to caller, that will do the lookup.
412 * Connect from a socket to a specified address.
413 * Both address and port must be specified in argument sin.
414 * If don't have a local address for this socket yet,
418 in6_pcbconnect_mbuf(struct inpcb *inp, struct sockaddr *nam,
419 struct ucred *cred, struct mbuf *m, bool rehash)
421 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
422 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
423 struct sockaddr_in6 laddr6;
426 KASSERT(sin6->sin6_family == AF_INET6,
427 ("%s: invalid address family for %p", __func__, sin6));
428 KASSERT(sin6->sin6_len == sizeof(*sin6),
429 ("%s: invalid address length for %p", __func__, sin6));
431 bzero(&laddr6, sizeof(laddr6));
432 laddr6.sin6_family = AF_INET6;
434 INP_WLOCK_ASSERT(inp);
435 INP_HASH_WLOCK_ASSERT(pcbinfo);
438 if (CALC_FLOWID_OUTBOUND) {
439 uint32_t hash_type, hash_val;
441 hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
442 &sin6->sin6_addr, 0, sin6->sin6_port,
443 inp->inp_socket->so_proto->pr_protocol, &hash_type);
444 inp->inp_flowid = hash_val;
445 inp->inp_flowtype = hash_type;
449 * Call inner routine, to assign local interface address.
450 * in6_pcbladdr() may automatically fill in sin6_scope_id.
452 if ((error = in6_pcbladdr(inp, sin6, &laddr6.sin6_addr)) != 0)
455 if (in6_pcblookup_hash_locked(pcbinfo, &sin6->sin6_addr,
457 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
458 ? &laddr6.sin6_addr : &inp->in6p_laddr,
459 inp->inp_lport, 0, NULL, M_NODOM) != NULL) {
462 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
463 if (inp->inp_lport == 0) {
465 * rehash was required to be true in the past for
466 * this case; retain that convention. However,
467 * we now call in_pcb_lport_dest rather than
468 * in6_pcbbind; the former does not insert into
469 * the hash table, the latter does. Change rehash
470 * to false to do the in_pcbinshash below.
472 KASSERT(rehash == true,
473 ("Rehashing required for unbound inps"));
475 error = in_pcb_lport_dest(inp,
476 (struct sockaddr *) &laddr6, &inp->inp_lport,
477 (struct sockaddr *) sin6, sin6->sin6_port, cred,
482 inp->in6p_laddr = laddr6.sin6_addr;
484 inp->in6p_faddr = sin6->sin6_addr;
485 inp->inp_fport = sin6->sin6_port;
486 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
487 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
488 if (inp->inp_flags & IN6P_AUTOFLOWLABEL)
490 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK);
502 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct ucred *cred)
505 return (in6_pcbconnect_mbuf(inp, nam, cred, NULL, true));
509 in6_pcbdisconnect(struct inpcb *inp)
512 INP_WLOCK_ASSERT(inp);
513 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
515 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
517 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
518 inp->inp_flow &= ~IPV6_FLOWLABEL_MASK;
523 in6_sockaddr(in_port_t port, struct in6_addr *addr_p)
525 struct sockaddr_in6 *sin6;
527 sin6 = malloc(sizeof *sin6, M_SONAME, M_WAITOK);
528 bzero(sin6, sizeof *sin6);
529 sin6->sin6_family = AF_INET6;
530 sin6->sin6_len = sizeof(*sin6);
531 sin6->sin6_port = port;
532 sin6->sin6_addr = *addr_p;
533 (void)sa6_recoverscope(sin6); /* XXX: should catch errors */
535 return (struct sockaddr *)sin6;
539 in6_v4mapsin6_sockaddr(in_port_t port, struct in_addr *addr_p)
541 struct sockaddr_in sin;
542 struct sockaddr_in6 *sin6_p;
544 bzero(&sin, sizeof sin);
545 sin.sin_family = AF_INET;
546 sin.sin_len = sizeof(sin);
548 sin.sin_addr = *addr_p;
550 sin6_p = malloc(sizeof *sin6_p, M_SONAME,
552 in6_sin_2_v4mapsin6(&sin, sin6_p);
554 return (struct sockaddr *)sin6_p;
558 in6_getsockaddr(struct socket *so, struct sockaddr **nam)
561 struct in6_addr addr;
565 KASSERT(inp != NULL, ("in6_getsockaddr: inp == NULL"));
568 port = inp->inp_lport;
569 addr = inp->in6p_laddr;
572 *nam = in6_sockaddr(port, &addr);
577 in6_getpeeraddr(struct socket *so, struct sockaddr **nam)
580 struct in6_addr addr;
584 KASSERT(inp != NULL, ("in6_getpeeraddr: inp == NULL"));
587 port = inp->inp_fport;
588 addr = inp->in6p_faddr;
591 *nam = in6_sockaddr(port, &addr);
596 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
602 KASSERT(inp != NULL, ("in6_mapped_sockaddr: inp == NULL"));
605 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
606 error = in_getsockaddr(so, nam);
608 in6_sin_2_v4mapsin6_in_sock(nam);
612 /* scope issues will be handled in in6_getsockaddr(). */
613 error = in6_getsockaddr(so, nam);
620 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
626 KASSERT(inp != NULL, ("in6_mapped_peeraddr: inp == NULL"));
629 if ((inp->inp_vflag & (INP_IPV4 | INP_IPV6)) == INP_IPV4) {
630 error = in_getpeeraddr(so, nam);
632 in6_sin_2_v4mapsin6_in_sock(nam);
635 /* scope issues will be handled in in6_getpeeraddr(). */
636 error = in6_getpeeraddr(so, nam);
642 * Pass some notification to all connections of a protocol
643 * associated with address dst. The local address and/or port numbers
644 * may be specified to limit the search. The "usual action" will be
645 * taken, depending on the ctlinput cmd. The caller must filter any
646 * cmds that are uninteresting (e.g., no error in the map).
647 * Call the protocol specific routine (if any) to report
648 * any errors for each matching socket.
651 inp_match6(const struct inpcb *inp, void *v __unused)
654 return ((inp->inp_vflag & INP_IPV6) != 0);
658 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr_in6 *sa6_dst,
659 u_int fport_arg, const struct sockaddr_in6 *src, u_int lport_arg,
660 int errno, void *cmdarg,
661 struct inpcb *(*notify)(struct inpcb *, int))
663 struct inpcb_iterator inpi = INP_ITERATOR(pcbinfo, INPLOOKUP_WLOCKPCB,
666 struct sockaddr_in6 sa6_src;
667 u_short fport = fport_arg, lport = lport_arg;
670 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
674 * note that src can be NULL when we get notify by local fragmentation.
676 sa6_src = (src == NULL) ? sa6_any : *src;
677 flowinfo = sa6_src.sin6_flowinfo;
679 while ((inp = inp_next(&inpi)) != NULL) {
680 INP_WLOCK_ASSERT(inp);
682 * If the error designates a new path MTU for a destination
683 * and the application (associated with this socket) wanted to
684 * know the value, notify.
685 * XXX: should we avoid to notify the value to TCP sockets?
687 if (errno == EMSGSIZE && cmdarg != NULL)
688 ip6_notify_pmtu(inp, sa6_dst, *(uint32_t *)cmdarg);
691 * Detect if we should notify the error. If no source and
692 * destination ports are specified, but non-zero flowinfo and
693 * local address match, notify the error. This is the case
694 * when the error is delivered with an encrypted buffer
695 * by ESP. Otherwise, just compare addresses and ports
698 if (lport == 0 && fport == 0 && flowinfo &&
699 inp->inp_socket != NULL &&
700 flowinfo == (inp->inp_flow & IPV6_FLOWLABEL_MASK) &&
701 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
703 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
704 &sa6_dst->sin6_addr) ||
705 inp->inp_socket == 0 ||
706 (lport && inp->inp_lport != lport) ||
707 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
708 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
709 &sa6_src.sin6_addr)) ||
710 (fport && inp->inp_fport != fport)) {
716 (*notify)(inp, errno);
721 * Lookup a PCB based on the local address and port. Caller must hold the
722 * hash lock. No inpcb locks or references are acquired.
725 in6_pcblookup_local(struct inpcbinfo *pcbinfo, struct in6_addr *laddr,
726 u_short lport, int lookupflags, struct ucred *cred)
729 int matchwild = 3, wildcard;
731 KASSERT((lookupflags & ~(INPLOOKUP_WILDCARD)) == 0,
732 ("%s: invalid lookup flags %d", __func__, lookupflags));
734 INP_HASH_LOCK_ASSERT(pcbinfo);
736 if ((lookupflags & INPLOOKUP_WILDCARD) == 0) {
737 struct inpcbhead *head;
739 * Look for an unconnected (wildcard foreign addr) PCB that
740 * matches the local address and port we're looking for.
742 head = &pcbinfo->ipi_hashbase[INP_PCBHASH_WILD(lport,
743 pcbinfo->ipi_hashmask)];
744 CK_LIST_FOREACH(inp, head, inp_hash) {
745 /* XXX inp locking */
746 if ((inp->inp_vflag & INP_IPV6) == 0)
748 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
749 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
750 inp->inp_lport == lport) {
752 if (prison_equal_ip6(cred->cr_prison,
753 inp->inp_cred->cr_prison))
762 struct inpcbporthead *porthash;
763 struct inpcbport *phd;
764 struct inpcb *match = NULL;
766 * Best fit PCB lookup.
768 * First see if this local port is in use by looking on the
771 porthash = &pcbinfo->ipi_porthashbase[INP_PCBPORTHASH(lport,
772 pcbinfo->ipi_porthashmask)];
773 CK_LIST_FOREACH(phd, porthash, phd_hash) {
774 if (phd->phd_port == lport)
779 * Port is in use by one or more PCBs. Look for best
782 CK_LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
784 if (!prison_equal_ip6(cred->cr_prison,
785 inp->inp_cred->cr_prison))
787 /* XXX inp locking */
788 if ((inp->inp_vflag & INP_IPV6) == 0)
790 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
792 if (!IN6_IS_ADDR_UNSPECIFIED(
794 if (IN6_IS_ADDR_UNSPECIFIED(laddr))
796 else if (!IN6_ARE_ADDR_EQUAL(
797 &inp->in6p_laddr, laddr))
800 if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
803 if (wildcard < matchwild) {
805 matchwild = wildcard;
816 in6_multi_match(const struct inpcb *inp, void *v __unused)
819 if ((inp->inp_vflag & INP_IPV6) && inp->in6p_moptions != NULL)
826 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
828 struct inpcb_iterator inpi = INP_ITERATOR(pcbinfo, INPLOOKUP_RLOCKPCB,
829 in6_multi_match, NULL);
831 struct in6_multi *inm;
832 struct in6_mfilter *imf;
833 struct ip6_moptions *im6o;
835 IN6_MULTI_LOCK_ASSERT();
837 while ((inp = inp_next(&inpi)) != NULL) {
838 INP_RLOCK_ASSERT(inp);
840 im6o = inp->in6p_moptions;
842 * Unselect the outgoing ifp for multicast if it
845 if (im6o->im6o_multicast_ifp == ifp)
846 im6o->im6o_multicast_ifp = NULL;
848 * Drop multicast group membership if we joined
849 * through the interface being detached.
852 IP6_MFILTER_FOREACH(imf, &im6o->im6o_head) {
853 if ((inm = imf->im6f_in6m) == NULL)
855 if (inm->in6m_ifp != ifp)
857 ip6_mfilter_remove(&im6o->im6o_head, imf);
858 in6_leavegroup_locked(inm, NULL);
859 ip6_mfilter_free(imf);
866 * Check for alternatives when higher level complains
867 * about service problems. For now, invalidate cached
868 * routing information. If the route was created dynamically
869 * (by a redirect), time to try a default gateway again.
872 in6_losing(struct inpcb *inp)
875 RO_INVALIDATE_CACHE(&inp->inp_route6);
879 * After a routing change, flush old routing
880 * and allocate a (hopefully) better one.
883 in6_rtchange(struct inpcb *inp, int errno __unused)
886 RO_INVALIDATE_CACHE(&inp->inp_route6);
891 in6_pcblookup_lb_numa_match(const struct inpcblbgroup *grp, int domain)
893 return (domain == M_NODOM || domain == grp->il_numa_domain);
896 static struct inpcb *
897 in6_pcblookup_lbgroup(const struct inpcbinfo *pcbinfo,
898 const struct in6_addr *laddr, uint16_t lport, const struct in6_addr *faddr,
899 uint16_t fport, int lookupflags, uint8_t domain)
901 const struct inpcblbgrouphead *hdr;
902 struct inpcblbgroup *grp;
903 struct inpcblbgroup *jail_exact, *jail_wild, *local_exact, *local_wild;
905 INP_HASH_LOCK_ASSERT(pcbinfo);
907 hdr = &pcbinfo->ipi_lbgrouphashbase[
908 INP_PCBPORTHASH(lport, pcbinfo->ipi_lbgrouphashmask)];
911 * Search for an LB group match based on the following criteria:
912 * - prefer jailed groups to non-jailed groups
913 * - prefer exact source address matches to wildcard matches
914 * - prefer groups bound to the specified NUMA domain
916 jail_exact = jail_wild = local_exact = local_wild = NULL;
917 CK_LIST_FOREACH(grp, hdr, il_list) {
921 if (!(grp->il_vflag & INP_IPV6))
924 if (grp->il_lport != lport)
927 injail = prison_flag(grp->il_cred, PR_IP6) != 0;
928 if (injail && prison_check_ip6_locked(grp->il_cred->cr_prison,
932 if (IN6_ARE_ADDR_EQUAL(&grp->il6_laddr, laddr)) {
935 if (in6_pcblookup_lb_numa_match(grp, domain))
936 /* This is a perfect match. */
938 } else if (local_exact == NULL ||
939 in6_pcblookup_lb_numa_match(grp, domain)) {
942 } else if (IN6_IS_ADDR_UNSPECIFIED(&grp->il6_laddr) &&
943 (lookupflags & INPLOOKUP_WILDCARD) != 0) {
945 if (jail_wild == NULL ||
946 in6_pcblookup_lb_numa_match(grp, domain))
948 } else if (local_wild == NULL ||
949 in6_pcblookup_lb_numa_match(grp, domain)) {
955 if (jail_exact != NULL)
957 else if (jail_wild != NULL)
959 else if (local_exact != NULL)
966 return (grp->il_inp[INP6_PCBLBGROUP_PKTHASH(faddr, lport, fport) %
971 * Lookup PCB in hash list. Used in in_pcb.c as well as here.
974 in6_pcblookup_hash_locked(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
975 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg,
976 int lookupflags, struct ifnet *ifp, uint8_t numa_domain)
978 struct inpcbhead *head;
979 struct inpcb *inp, *tmpinp;
980 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);
988 * First look for an exact match.
991 head = &pcbinfo->ipi_hashbase[INP6_PCBHASH(faddr, lport, fport,
992 pcbinfo->ipi_hashmask)];
993 CK_LIST_FOREACH(inp, head, inp_hash) {
994 /* XXX inp locking */
995 if ((inp->inp_vflag & INP_IPV6) == 0)
997 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
998 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
999 inp->inp_fport == fport &&
1000 inp->inp_lport == lport) {
1002 * XXX We should be able to directly return
1003 * the inp here, without any checks.
1004 * Well unless both bound with SO_REUSEPORT?
1006 if (prison_flag(inp->inp_cred, PR_IP6))
1016 * Then look for a wildcard match, if requested.
1018 if ((lookupflags & INPLOOKUP_WILDCARD) != 0) {
1019 struct inpcb *local_wild = NULL, *local_exact = NULL;
1020 struct inpcb *jail_wild = NULL;
1024 * First see if an LB group matches the request before scanning
1025 * all sockets on this port.
1027 inp = in6_pcblookup_lbgroup(pcbinfo, laddr, lport, faddr,
1028 fport, lookupflags, numa_domain);
1033 * Order of socket selection - we always prefer jails.
1034 * 1. jailed, non-wild.
1036 * 3. non-jailed, non-wild.
1037 * 4. non-jailed, wild.
1039 head = &pcbinfo->ipi_hashbase[INP_PCBHASH_WILD(lport,
1040 pcbinfo->ipi_hashmask)];
1041 CK_LIST_FOREACH(inp, head, inp_hash) {
1042 /* XXX inp locking */
1043 if ((inp->inp_vflag & INP_IPV6) == 0)
1046 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
1047 inp->inp_lport != lport) {
1051 injail = prison_flag(inp->inp_cred, PR_IP6);
1053 if (prison_check_ip6_locked(
1054 inp->inp_cred->cr_prison, laddr) != 0)
1057 if (local_exact != NULL)
1061 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr)) {
1066 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1072 } /* LIST_FOREACH */
1074 if (jail_wild != NULL)
1076 if (local_exact != NULL)
1077 return (local_exact);
1078 if (local_wild != NULL)
1079 return (local_wild);
1080 } /* if ((lookupflags & INPLOOKUP_WILDCARD) != 0) */
1089 * Lookup PCB in hash list, using pcbinfo tables. This variation locks the
1090 * hash list lock, and will return the inpcb locked (i.e., requires
1091 * INPLOOKUP_LOCKPCB).
1093 static struct inpcb *
1094 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1095 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1096 struct ifnet *ifp, uint8_t numa_domain)
1100 smr_enter(pcbinfo->ipi_smr);
1101 inp = in6_pcblookup_hash_locked(pcbinfo, faddr, fport, laddr, lport,
1102 lookupflags & INPLOOKUP_WILDCARD, ifp, numa_domain);
1104 if (__predict_false(inp_smr_lock(inp,
1105 (lookupflags & INPLOOKUP_LOCKMASK)) == false))
1108 smr_exit(pcbinfo->ipi_smr);
1114 * Public inpcb lookup routines, accepting a 4-tuple, and optionally, an mbuf
1115 * from which a pre-calculated hash value may be extracted.
1118 in6_pcblookup(struct inpcbinfo *pcbinfo, struct in6_addr *faddr, u_int fport,
1119 struct in6_addr *laddr, u_int lport, int lookupflags, struct ifnet *ifp)
1122 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1123 ("%s: invalid lookup flags %d", __func__, lookupflags));
1124 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1125 ("%s: LOCKPCB not set", __func__));
1127 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1128 lookupflags, ifp, M_NODOM));
1132 in6_pcblookup_mbuf(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1133 u_int fport, struct in6_addr *laddr, u_int lport, int lookupflags,
1134 struct ifnet *ifp, struct mbuf *m)
1137 KASSERT((lookupflags & ~INPLOOKUP_MASK) == 0,
1138 ("%s: invalid lookup flags %d", __func__, lookupflags));
1139 KASSERT((lookupflags & (INPLOOKUP_RLOCKPCB | INPLOOKUP_WLOCKPCB)) != 0,
1140 ("%s: LOCKPCB not set", __func__));
1142 return (in6_pcblookup_hash(pcbinfo, faddr, fport, laddr, lport,
1143 lookupflags, ifp, m->m_pkthdr.numa_domain));
1147 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m, int srcordst)
1151 ip = mtod(m, struct ip6_hdr *);
1152 bzero(sin6, sizeof(*sin6));
1153 sin6->sin6_len = sizeof(*sin6);
1154 sin6->sin6_family = AF_INET6;
1155 sin6->sin6_addr = srcordst ? ip->ip6_dst : ip->ip6_src;
1157 (void)sa6_recoverscope(sin6); /* XXX: should catch errors... */