1 /* $FreeBSD: src/sys/netinet6/in6_pcb.c,v 1.10.2.9 2003/01/24 05:11:35 sam Exp $ */
2 /* $KAME: in6_pcb.c,v 1.31 2001/05/21 05:45:10 jinmei Exp $ */
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * Copyright (c) 1982, 1986, 1991, 1993
36 * The Regents of the University of California. All rights reserved.
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. Neither the name of the University nor the names of its contributors
47 * may be used to endorse or promote products derived from this software
48 * without specific prior written permission.
50 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
51 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
52 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
53 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
54 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
55 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
56 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
57 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
58 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
59 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
66 #include "opt_inet6.h"
68 #include <sys/param.h>
69 #include <sys/systm.h>
70 #include <sys/malloc.h>
72 #include <sys/domain.h>
73 #include <sys/protosw.h>
74 #include <sys/socket.h>
75 #include <sys/socketvar.h>
76 #include <sys/sockio.h>
77 #include <sys/errno.h>
83 #include <sys/msgport2.h>
85 #include <vm/vm_zone.h>
88 #include <net/if_types.h>
89 #include <net/route.h>
90 #include <net/netisr2.h>
92 #include <netinet/in.h>
93 #include <netinet/in_var.h>
94 #include <netinet/in_systm.h>
95 #include <netinet/ip6.h>
96 #include <netinet/ip_var.h>
97 #include <netinet6/ip6_var.h>
98 #include <netinet6/nd6.h>
99 #include <netinet/in_pcb.h>
100 #include <netinet6/in6_pcb.h>
102 struct in6_addr zeroin6_addr
;
105 in6_pcbbind(struct inpcb
*inp
, struct sockaddr
*nam
, struct thread
*td
)
107 struct socket
*so
= inp
->inp_socket
;
108 struct sockaddr_in6 jsin6
;
111 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
))
115 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)nam
;
116 struct inpcbinfo
*pcbinfo
;
117 struct inpcbportinfo
*portinfo
;
118 struct inpcbporthead
*porthash
;
119 int wild
= 0, reuseport
= (so
->so_options
& SO_REUSEPORT
);
120 struct ucred
*cred
= NULL
;
122 u_short lport
, lport_ho
;
124 if ((so
->so_options
& (SO_REUSEADDR
|SO_REUSEPORT
)) == 0)
126 if (td
->td_proc
!= NULL
)
127 cred
= td
->td_proc
->p_ucred
;
129 if (nam
->sa_len
!= sizeof(*sin6
))
134 if (nam
->sa_family
!= AF_INET6
)
135 return (EAFNOSUPPORT
);
137 /* Reject v4-mapped address */
138 if (IN6_IS_ADDR_V4MAPPED(&sin6
->sin6_addr
))
139 return (EADDRNOTAVAIL
);
141 if (!prison_replace_wildcards(td
, nam
))
144 /* KAME hack: embed scopeid */
145 if (in6_embedscope(&sin6
->sin6_addr
, sin6
, inp
, NULL
) != 0)
147 /* this must be cleared for ifa_ifwithaddr() */
148 sin6
->sin6_scope_id
= 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
)) {
162 struct ifaddr
*ia
= NULL
;
164 sin6
->sin6_port
= 0; /* yech... */
165 if (!prison_replace_wildcards(td
, (struct sockaddr
*)sin6
)) {
166 sin6
->sin6_addr
= kin6addr_any
;
169 if ((ia
= ifa_ifwithaddr((struct sockaddr
*)sin6
)) == NULL
)
170 return (EADDRNOTAVAIL
);
173 * XXX: bind to an anycast address might accidentally
174 * cause sending a packet with anycast source address.
175 * We should allow to bind to a deprecated address, since
176 * the application dares to use it.
179 ((struct in6_ifaddr
*)ia
)->ia6_flags
&
180 (IN6_IFF_ANYCAST
|IN6_IFF_NOTREADY
|IN6_IFF_DETACHED
))
181 return (EADDRNOTAVAIL
);
184 inp
->in6p_laddr
= sin6
->sin6_addr
;
188 lport_ho
= ntohs(lport
);
191 if (lport_ho
< IPV6PORT_RESERVED
&& cred
&&
192 caps_priv_check(cred
, SYSCAP_NONET_RESPORT
))
194 inp
->in6p_laddr
= kin6addr_any
;
199 * Locate the proper portinfo based on lport
201 pcbinfo
= inp
->inp_pcbinfo
;
203 &pcbinfo
->portinfo
[lport_ho
% pcbinfo
->portinfo_cnt
];
204 KKASSERT((lport_ho
% pcbinfo
->portinfo_cnt
) ==
208 * This has to be atomic. If the porthash is shared across
209 * multiple protocol threads (aka tcp) then the token must
212 porthash
= in_pcbporthash_head(portinfo
, lport
);
213 GET_PORTHASH_TOKEN(porthash
);
215 if (so
->so_cred
->cr_uid
!= 0 &&
216 !IN6_IS_ADDR_MULTICAST(&sin6
->sin6_addr
)) {
217 t
= in6_pcblookup_local(porthash
, &sin6
->sin6_addr
,
218 lport
, INPLOOKUP_WILDCARD
,
221 (so
->so_cred
->cr_uid
!=
222 t
->inp_socket
->so_cred
->cr_uid
)) {
223 inp
->in6p_laddr
= kin6addr_any
;
228 if (cred
&& cred
->cr_prison
&&
229 !prison_replace_wildcards(td
, nam
)) {
230 inp
->in6p_laddr
= kin6addr_any
;
231 error
= EADDRNOTAVAIL
;
234 t
= in6_pcblookup_local(porthash
, &sin6
->sin6_addr
, lport
,
236 if (t
&& (reuseport
& t
->inp_socket
->so_options
) == 0) {
237 inp
->in6p_laddr
= kin6addr_any
;
242 inp
->inp_lport
= lport
;
243 in_pcbinsporthash(porthash
, inp
);
246 REL_PORTHASH_TOKEN(porthash
);
250 jsin6
.sin6_addr
= inp
->in6p_laddr
;
251 jsin6
.sin6_family
= AF_INET6
;
252 if (!prison_replace_wildcards(td
, (struct sockaddr
*)&jsin6
)) {
253 inp
->in6p_laddr
= kin6addr_any
;
258 return in6_pcbsetlport(&inp
->in6p_laddr
, inp
, td
);
263 * Transform old in6_pcbconnect() into an inner subroutine for new
264 * in6_pcbconnect(): Do some validity-checking on the remote
265 * address (in mbuf 'nam') and then determine local host address
266 * (i.e., which interface) to use to access that remote host.
268 * This preserves definition of in6_pcbconnect(), while supporting a
269 * slightly different version for T/TCP. (This is more than
270 * a bit of a kludge, but cleaning up the internal interfaces would
271 * have forced minor changes in every protocol).
275 in6_pcbladdr(struct inpcb
*inp
, struct sockaddr
*nam
,
276 struct in6_addr
**plocal_addr6
, struct thread
*td
)
278 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)nam
;
279 struct ifnet
*ifp
= NULL
;
282 if (nam
->sa_len
!= sizeof (*sin6
))
284 if (sin6
->sin6_family
!= AF_INET6
)
285 return (EAFNOSUPPORT
);
286 if (sin6
->sin6_port
== 0)
287 return (EADDRNOTAVAIL
);
289 /* KAME hack: embed scopeid */
290 if (in6_embedscope(&sin6
->sin6_addr
, sin6
, inp
, &ifp
) != 0)
295 * If the destination address is UNSPECIFIED addr,
296 * use the loopback addr, e.g ::1.
298 if (IN6_IS_ADDR_UNSPECIFIED(&sin6
->sin6_addr
))
299 sin6
->sin6_addr
= kin6addr_loopback
;
303 * XXX: in6_selectsrc might replace the bound local address
304 * with the address specified by setsockopt(IPV6_PKTINFO).
305 * Is it the intended behavior?
307 *plocal_addr6
= in6_selectsrc(sin6
, inp
->in6p_outputopts
,
310 &inp
->in6p_laddr
, &error
, td
);
311 if (*plocal_addr6
== NULL
) {
313 error
= EADDRNOTAVAIL
;
317 * Don't do pcblookup call here; return interface in
319 * and exit to caller, that will do the lookup.
323 if (inp
->in6p_route
.ro_rt
)
324 ifp
= inp
->in6p_route
.ro_rt
->rt_ifp
;
331 * Connect from a socket to a specified address.
332 * Both address and port must be specified in argument sin.
333 * If don't have a local address for this socket yet,
337 in6_pcbconnect(struct inpcb
*inp
, struct sockaddr
*nam
, struct thread
*td
)
339 struct in6_addr
*addr6
;
340 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)nam
;
343 /* Reject v4-mapped address */
344 if (IN6_IS_ADDR_V4MAPPED(&sin6
->sin6_addr
))
345 return EADDRNOTAVAIL
;
348 * Call inner routine, to assign local interface address.
349 * in6_pcbladdr() may automatically fill in sin6_scope_id.
351 if ((error
= in6_pcbladdr(inp
, nam
, &addr6
, td
)) != 0)
354 if (in6_pcblookup_hash(inp
->inp_pcbinfo
, &sin6
->sin6_addr
,
356 IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)
357 ? addr6
: &inp
->in6p_laddr
,
358 inp
->inp_lport
, 0, NULL
) != NULL
) {
361 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
362 if (inp
->inp_lport
== 0) {
363 error
= in6_pcbbind(inp
, NULL
, td
);
367 inp
->in6p_laddr
= *addr6
;
369 inp
->in6p_faddr
= sin6
->sin6_addr
;
370 inp
->inp_fport
= sin6
->sin6_port
;
371 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
372 inp
->in6p_flowinfo
&= ~IPV6_FLOWLABEL_MASK
;
373 if (inp
->in6p_flags
& IN6P_AUTOFLOWLABEL
)
374 inp
->in6p_flowinfo
|=
375 (htonl(ip6_flow_seq
++) & IPV6_FLOWLABEL_MASK
);
377 in_pcbinsconnhash(inp
);
382 in6_pcbdisconnect(struct inpcb
*inp
)
384 bzero((caddr_t
)&inp
->in6p_faddr
, sizeof(inp
->in6p_faddr
));
386 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
387 inp
->in6p_flowinfo
&= ~IPV6_FLOWLABEL_MASK
;
388 in_pcbremconnhash(inp
);
389 if (inp
->inp_socket
->so_state
& SS_NOFDREF
)
394 in6_pcbdetach(struct inpcb
*inp
)
396 struct socket
*so
= inp
->inp_socket
;
397 struct inpcbinfo
*ipi
= inp
->inp_pcbinfo
;
399 inp
->inp_gencnt
= ++ipi
->ipi_gencnt
;
402 KKASSERT((so
->so_state
& SS_ASSERTINPROG
) == 0);
403 sofree(so
); /* remove pcb ref */
405 if (inp
->in6p_options
)
406 m_freem(inp
->in6p_options
);
407 ip6_freepcbopts(inp
->in6p_outputopts
);
408 ip6_freemoptions(inp
->in6p_moptions
);
409 if (inp
->in6p_route
.ro_rt
)
410 rtfree(inp
->in6p_route
.ro_rt
);
411 /* Check and free IPv4 related resources in case of mapped addr */
412 if (inp
->inp_options
)
413 m_free(inp
->inp_options
);
414 ip_freemoptions(inp
->inp_moptions
);
420 * The socket may have an invalid PCB, i.e. NULL. For example, a TCP
421 * socket received RST.
424 in6_setsockaddr(struct socket
*so
, struct sockaddr
**nam
)
427 struct sockaddr_in6
*sin6
;
429 KASSERT(curthread
->td_type
== TD_TYPE_NETISR
, ("not in netisr"));
434 sin6
= kmalloc(sizeof *sin6
, M_SONAME
, M_WAITOK
| M_ZERO
);
435 sin6
->sin6_family
= AF_INET6
;
436 sin6
->sin6_len
= sizeof(*sin6
);
437 sin6
->sin6_port
= inp
->inp_lport
;
438 sin6
->sin6_addr
= inp
->in6p_laddr
;
439 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
440 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
442 sin6
->sin6_scope_id
= 0; /*XXX*/
443 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
444 sin6
->sin6_addr
.s6_addr16
[1] = 0;
446 *nam
= (struct sockaddr
*)sin6
;
451 in6_setsockaddr_dispatch(netmsg_t msg
)
455 error
= in6_setsockaddr(msg
->sockaddr
.base
.nm_so
, msg
->sockaddr
.nm_nam
);
456 lwkt_replymsg(&msg
->sockaddr
.base
.lmsg
, error
);
460 in6_setpeeraddr_dispatch(netmsg_t msg
)
464 error
= in6_setpeeraddr(msg
->peeraddr
.base
.nm_so
, msg
->peeraddr
.nm_nam
);
465 lwkt_replymsg(&msg
->peeraddr
.base
.lmsg
, error
);
469 * The socket may have an invalid PCB, i.e. NULL. For example, a TCP
470 * socket received RST.
473 in6_setpeeraddr(struct socket
*so
, struct sockaddr
**nam
)
476 struct sockaddr_in6
*sin6
;
478 KASSERT(curthread
->td_type
== TD_TYPE_NETISR
, ("not in netisr"));
483 sin6
= kmalloc(sizeof(*sin6
), M_SONAME
, M_WAITOK
| M_ZERO
);
484 sin6
->sin6_family
= AF_INET6
;
485 sin6
->sin6_len
= sizeof(struct sockaddr_in6
);
486 sin6
->sin6_port
= inp
->inp_fport
;
487 sin6
->sin6_addr
= inp
->in6p_faddr
;
488 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
489 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
491 sin6
->sin6_scope_id
= 0; /*XXX*/
492 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
493 sin6
->sin6_addr
.s6_addr16
[1] = 0;
495 *nam
= (struct sockaddr
*)sin6
;
500 * Pass some notification to all connections of a protocol
501 * associated with address dst. The local address and/or port numbers
502 * may be specified to limit the search. The "usual action" will be
503 * taken, depending on the ctlinput cmd. The caller must filter any
504 * cmds that are uninteresting (e.g., no error in the map).
505 * Call the protocol specific routine (if any) to report
506 * any errors for each matching socket.
509 in6_pcbnotify(struct inpcbinfo
*pcbinfo
, struct sockaddr
*dst
, in_port_t fport
,
510 const struct sockaddr
*src
, in_port_t lport
, int cmd
, int arg
,
513 struct inpcb
*inp
, *marker
;
514 struct sockaddr_in6 sa6_src
, *sa6_dst
;
517 if ((unsigned)cmd
>= PRC_NCMDS
|| dst
->sa_family
!= AF_INET6
)
520 sa6_dst
= (struct sockaddr_in6
*)dst
;
521 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst
->sin6_addr
))
525 * note that src can be NULL when we get notify by local fragmentation.
527 sa6_src
= (src
== NULL
) ? sa6_any
: *(const struct sockaddr_in6
*)src
;
528 flowinfo
= sa6_src
.sin6_flowinfo
;
531 * Redirects go to all references to the destination,
532 * and use in6_rtchange to invalidate the route cache.
533 * Dead host indications: also use in6_rtchange to invalidate
534 * the cache, and deliver the error to all the sockets.
535 * Otherwise, if we have knowledge of the local port and address,
536 * deliver only to that socket.
538 if (PRC_IS_REDIRECT(cmd
) || cmd
== PRC_HOSTDEAD
) {
541 bzero((caddr_t
)&sa6_src
.sin6_addr
, sizeof(sa6_src
.sin6_addr
));
543 if (cmd
!= PRC_HOSTDEAD
)
544 notify
= in6_rtchange
;
546 if (cmd
!= PRC_MSGSIZE
)
547 arg
= inet6ctlerrmap
[cmd
];
549 marker
= in_pcbmarker();
551 GET_PCBINFO_TOKEN(pcbinfo
);
553 LIST_INSERT_HEAD(&pcbinfo
->pcblisthead
, marker
, inp_list
);
554 while ((inp
= LIST_NEXT(marker
, inp_list
)) != NULL
) {
555 LIST_REMOVE(marker
, inp_list
);
556 LIST_INSERT_AFTER(inp
, marker
, inp_list
);
558 if (inp
->inp_flags
& INP_PLACEMARKER
)
561 if (!INP_ISIPV6(inp
))
564 * If the error designates a new path MTU for a destination
565 * and the application (associated with this socket) wanted to
566 * know the value, notify. Note that we notify for all
567 * disconnected sockets if the corresponding application
568 * wanted. This is because some UDP applications keep sending
569 * sockets disconnected.
570 * XXX: should we avoid to notify the value to TCP sockets?
572 if (cmd
== PRC_MSGSIZE
&& (inp
->inp_flags
& IN6P_MTU
) != 0 &&
573 (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) ||
574 IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, &sa6_dst
->sin6_addr
))) {
575 ip6_notify_pmtu(inp
, (struct sockaddr_in6
*)dst
, &arg
);
579 * Detect if we should notify the error. If no source and
580 * destination ports are specifed, but non-zero flowinfo and
581 * local address match, notify the error. This is the case
582 * when the error is delivered with an encrypted buffer
583 * by ESP. Otherwise, just compare addresses and ports
586 if (lport
== 0 && fport
== 0 && flowinfo
&&
587 inp
->inp_socket
!= NULL
&&
588 flowinfo
== (inp
->in6p_flowinfo
& IPV6_FLOWLABEL_MASK
) &&
589 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, &sa6_src
.sin6_addr
))
591 else if (!IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
,
592 &sa6_dst
->sin6_addr
) ||
593 inp
->inp_socket
== 0 ||
594 (lport
&& inp
->inp_lport
!= lport
) ||
595 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src
.sin6_addr
) &&
596 !IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
597 &sa6_src
.sin6_addr
)) ||
598 (fport
&& inp
->inp_fport
!= fport
))
605 LIST_REMOVE(marker
, inp_list
);
607 REL_PCBINFO_TOKEN(pcbinfo
);
611 * Lookup a PCB based on the local address and port.
614 in6_pcblookup_local(struct inpcbporthead
*porthash
,
615 const struct in6_addr
*laddr
, u_int lport_arg
, int wild_okay
,
618 struct prison
*pscan
;
621 int matchwild
= 3, wildcard
;
622 u_short lport
= lport_arg
;
623 struct inpcbport
*phd
;
624 struct inpcb
*match
= NULL
;
627 * If the porthashbase is shared across several cpus, it must
630 ASSERT_PORTHASH_TOKEN_HELD(porthash
);
633 * Best fit PCB lookup.
635 * First see if this local port is in use by looking on the
638 LIST_FOREACH(phd
, porthash
, phd_hash
) {
639 if (phd
->phd_port
== lport
)
644 pr
= cred
? cred
->cr_prison
: NULL
;
647 * Port is in use by one or more PCBs. Look for best
650 LIST_FOREACH(inp
, &phd
->phd_pcblist
, inp_portlist
) {
652 if (!INP_ISIPV6(inp
))
654 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
))
657 if (inp
->inp_socket
&& inp
->inp_socket
->so_cred
)
658 pscan
= inp
->inp_socket
->so_cred
->cr_prison
;
664 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
665 if (!IN6_IS_ADDR_UNSPECIFIED(laddr
))
668 if (IN6_IS_ADDR_UNSPECIFIED(laddr
))
670 else if (!IN6_ARE_ADDR_EQUAL(
671 &inp
->in6p_laddr
, laddr
))
674 if (wildcard
&& !wild_okay
)
676 if (wildcard
< matchwild
) {
678 matchwild
= wildcard
;
688 in6_pcbpurgeif0(struct inpcbinfo
*pcbinfo
, struct ifnet
*ifp
)
690 struct in6pcb
*in6p
, *marker
;
691 struct ip6_moptions
*im6o
;
692 struct in6_multi_mship
*imm
, *nimm
;
695 * We only need to make sure that we are in netisr0, where all
696 * multicast operation happen. We could check inpcbinfo which
697 * does not belong to netisr0 by holding the inpcbinfo's token.
698 * In this case, the pcbinfo must be able to be shared, i.e.
699 * pcbinfo->infotoken is not NULL.
702 KASSERT(pcbinfo
->cpu
== 0 || pcbinfo
->infotoken
!= NULL
,
703 ("pcbinfo could not be shared"));
706 * Get a marker for the current netisr (netisr0).
708 * It is possible that the multicast address deletion blocks,
709 * which could cause temporary token releasing. So we use
710 * inpcb marker here to get a coherent view of the inpcb list.
712 * While, on the other hand, moptions are only added and deleted
713 * in netisr0, so we would not see staled moption or miss moption
714 * even if the token was released due to the blocking multicast
717 marker
= in_pcbmarker();
719 GET_PCBINFO_TOKEN(pcbinfo
);
721 LIST_INSERT_HEAD(&pcbinfo
->pcblisthead
, marker
, inp_list
);
722 while ((in6p
= LIST_NEXT(marker
, inp_list
)) != NULL
) {
723 LIST_REMOVE(marker
, inp_list
);
724 LIST_INSERT_AFTER(in6p
, marker
, inp_list
);
726 if (in6p
->in6p_flags
& INP_PLACEMARKER
)
728 im6o
= in6p
->in6p_moptions
;
729 if (INP_ISIPV6(in6p
) && im6o
) {
731 * Unselect the outgoing interface if it is being
734 if (im6o
->im6o_multicast_ifp
== ifp
)
735 im6o
->im6o_multicast_ifp
= NULL
;
738 * Drop multicast group membership if we joined
739 * through the interface being detached.
740 * XXX controversial - is it really legal for kernel
743 for (imm
= im6o
->im6o_memberships
.lh_first
;
744 imm
!= NULL
; imm
= nimm
) {
745 nimm
= imm
->i6mm_chain
.le_next
;
746 if (imm
->i6mm_maddr
->in6m_ifp
== ifp
) {
747 LIST_REMOVE(imm
, i6mm_chain
);
748 in6_delmulti(imm
->i6mm_maddr
);
749 kfree(imm
, M_IPMADDR
);
754 LIST_REMOVE(marker
, inp_list
);
756 REL_PCBINFO_TOKEN(pcbinfo
);
760 * Check for alternatives when higher level complains
761 * about service problems. For now, invalidate cached
762 * routing information. If the route was created dynamically
763 * (by a redirect), time to try a default gateway again.
766 in6_losing(struct inpcb
*in6p
)
769 struct rt_addrinfo info
;
771 if ((rt
= in6p
->in6p_route
.ro_rt
) != NULL
) {
772 bzero((caddr_t
)&info
, sizeof(info
));
773 info
.rti_flags
= rt
->rt_flags
;
774 info
.rti_info
[RTAX_DST
] = rt_key(rt
);
775 info
.rti_info
[RTAX_GATEWAY
] = rt
->rt_gateway
;
776 info
.rti_info
[RTAX_NETMASK
] = rt_mask(rt
);
777 rt_missmsg(RTM_LOSING
, &info
, rt
->rt_flags
, 0);
778 if (rt
->rt_flags
& RTF_DYNAMIC
) {
779 rtrequest(RTM_DELETE
, rt_key(rt
), rt
->rt_gateway
,
780 rt_mask(rt
), rt
->rt_flags
, NULL
);
782 in6p
->in6p_route
.ro_rt
= NULL
;
785 * A new route can be allocated
786 * the next time output is attempted.
792 * After a routing change, flush old routing
793 * and allocate a (hopefully) better one.
796 in6_rtchange(struct inpcb
*inp
, int error
)
798 if (inp
->in6p_route
.ro_rt
) {
799 rtfree(inp
->in6p_route
.ro_rt
);
800 inp
->in6p_route
.ro_rt
= 0;
802 * A new route can be allocated the next time
803 * output is attempted.
809 * Lookup PCB in hash list.
812 in6_pcblookup_hash(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
813 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
,
814 int wildcard
, struct ifnet
*ifp
)
816 struct inpcbhead
*head
;
818 struct inpcb
*jinp
= NULL
;
819 u_short fport
= fport_arg
, lport
= lport_arg
;
822 * First look for an exact match.
824 head
= &pcbinfo
->hashbase
[INP_PCBCONNHASH(faddr
->s6_addr32
[3] /* XXX */,
826 laddr
->s6_addr32
[3], /* XXX JH */
829 LIST_FOREACH(inp
, head
, inp_hash
) {
830 if (!INP_ISIPV6(inp
))
832 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
833 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
834 inp
->inp_fport
== fport
&&
835 inp
->inp_lport
== lport
) {
839 if (inp
->inp_socket
== NULL
||
840 inp
->inp_socket
->so_cred
->cr_prison
== NULL
) {
851 struct inpcb
*local_wild
= NULL
;
852 struct inpcb
*jinp_wild
= NULL
;
853 struct inpcontainer
*ic
;
854 struct inpcontainerhead
*chead
;
855 struct sockaddr_in6 jsin6
;
858 int net_listen_ov_local
= 0;
859 int net_listen_ov_wild
= 0;
862 * Order of socket selection:
863 * 1. non-jailed, non-wild.
864 * 2. non-jailed, wild. (allow_listen_override on)
865 * 3. jailed, non-wild.
867 * 5. non-jailed, wild. (allow_listen_override off)
869 * NOTE: jailed wildcards are still restricted to the jail
872 * NOTE: (1) and (3) already handled above.
874 jsin6
.sin6_family
= AF_INET6
;
875 chead
= &pcbinfo
->wildcardhashbase
[INP_PCBWILDCARDHASH(lport
,
876 pcbinfo
->wildcardhashmask
)];
878 GET_PCBINFO_TOKEN(pcbinfo
);
879 LIST_FOREACH(ic
, chead
, ic_list
) {
881 if (inp
->inp_flags
& INP_PLACEMARKER
)
887 if (!INP_ISIPV6(inp
))
889 if (inp
->inp_lport
!= lport
)
893 * Calculate prison, setup jsin for jailed_ip()
896 jsin6
.sin6_addr
= *laddr
;
899 if (inp
->inp_socket
) {
900 cred
= inp
->inp_socket
->so_cred
;
902 pr
= cred
->cr_prison
;
906 * Assign jinp, jinp_wild, and local_wild as
907 * appropriate, track whether the jail supports
911 if (!jailed_ip(pr
, (struct sockaddr
*)&jsin6
))
913 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
)
916 if (PRISON_CAP_ISSET(pr
->pr_caps
, PRISON_CAP_NET_LISTEN_OVERRIDE
))
917 net_listen_ov_local
= 1;
919 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
922 if (PRISON_CAP_ISSET(pr
->pr_caps
, PRISON_CAP_NET_LISTEN_OVERRIDE
))
923 net_listen_ov_wild
= 1;
926 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
)) {
927 REL_PCBINFO_TOKEN(pcbinfo
);
930 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
))
934 REL_PCBINFO_TOKEN(pcbinfo
);
936 if (net_listen_ov_local
)
938 if (net_listen_ov_wild
)
954 init_sin6(struct sockaddr_in6
*sin6
, struct mbuf
*m
)
958 ip
= mtod(m
, struct ip6_hdr
*);
959 bzero(sin6
, sizeof(*sin6
));
960 sin6
->sin6_len
= sizeof(*sin6
);
961 sin6
->sin6_family
= AF_INET6
;
962 sin6
->sin6_addr
= ip
->ip6_src
;
963 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
964 sin6
->sin6_addr
.s6_addr16
[1] = 0;
965 sin6
->sin6_scope_id
=
966 (m
->m_pkthdr
.rcvif
&& IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
967 ? m
->m_pkthdr
.rcvif
->if_index
: 0;
973 in6_savefaddr(struct socket
*so
, const struct sockaddr
*faddr
)
975 struct sockaddr_in6
*sin6
;
977 KASSERT(faddr
->sa_family
== AF_INET6
,
978 ("not AF_INET6 faddr %d", faddr
->sa_family
));
980 sin6
= kmalloc(sizeof(*sin6
), M_SONAME
, M_WAITOK
| M_ZERO
);
981 sin6
->sin6_family
= AF_INET6
;
982 sin6
->sin6_len
= sizeof(*sin6
);
984 sin6
->sin6_port
= ((const struct sockaddr_in6
*)faddr
)->sin6_port
;
985 sin6
->sin6_addr
= ((const struct sockaddr_in6
*)faddr
)->sin6_addr
;
987 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
988 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
990 sin6
->sin6_scope_id
= 0; /*XXX*/
991 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
992 sin6
->sin6_addr
.s6_addr16
[1] = 0;
994 so
->so_faddr
= (struct sockaddr
*)sin6
;