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"
67 #include "opt_ipsec.h"
69 #include <sys/param.h>
70 #include <sys/systm.h>
71 #include <sys/malloc.h>
73 #include <sys/domain.h>
74 #include <sys/protosw.h>
75 #include <sys/socket.h>
76 #include <sys/socketvar.h>
77 #include <sys/sockio.h>
78 #include <sys/errno.h>
84 #include <sys/thread2.h>
85 #include <sys/msgport2.h>
87 #include <vm/vm_zone.h>
90 #include <net/if_types.h>
91 #include <net/route.h>
92 #include <net/netisr2.h>
94 #include <netinet/in.h>
95 #include <netinet/in_var.h>
96 #include <netinet/in_systm.h>
97 #include <netinet/ip6.h>
98 #include <netinet/ip_var.h>
99 #include <netinet6/ip6_var.h>
100 #include <netinet6/nd6.h>
101 #include <netinet/in_pcb.h>
102 #include <netinet6/in6_pcb.h>
105 #include <netinet6/ipsec.h>
107 #include <netinet6/ipsec6.h>
109 #include <netinet6/ah.h>
111 #include <netinet6/ah6.h>
113 #include <netproto/key/key.h>
117 #include <netproto/ipsec/ipsec.h>
118 #include <netproto/ipsec/ipsec6.h>
119 #include <netproto/ipsec/key.h>
121 #endif /* FAST_IPSEC */
123 struct in6_addr zeroin6_addr
;
126 in6_pcbbind(struct inpcb
*inp
, struct sockaddr
*nam
, struct thread
*td
)
128 struct socket
*so
= inp
->inp_socket
;
129 struct sockaddr_in6 jsin6
;
132 if (!in6_ifaddr
) /* XXX broken! */
133 return (EADDRNOTAVAIL
);
134 if (inp
->inp_lport
|| !IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
))
138 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)nam
;
139 struct inpcbinfo
*pcbinfo
;
140 struct inpcbportinfo
*portinfo
;
141 int wild
= 0, reuseport
= (so
->so_options
& SO_REUSEPORT
);
142 struct ucred
*cred
= NULL
;
144 u_short lport
, lport_ho
;
146 if ((so
->so_options
& (SO_REUSEADDR
|SO_REUSEPORT
)) == 0)
148 if (td
->td_proc
!= NULL
)
149 cred
= td
->td_proc
->p_ucred
;
151 if (nam
->sa_len
!= sizeof(*sin6
))
156 if (nam
->sa_family
!= AF_INET6
)
157 return (EAFNOSUPPORT
);
159 /* Reject v4-mapped address */
160 if (IN6_IS_ADDR_V4MAPPED(&sin6
->sin6_addr
))
161 return (EADDRNOTAVAIL
);
163 if (!prison_replace_wildcards(td
, nam
))
166 /* KAME hack: embed scopeid */
167 if (in6_embedscope(&sin6
->sin6_addr
, sin6
, inp
, NULL
) != 0)
169 /* this must be cleared for ifa_ifwithaddr() */
170 sin6
->sin6_scope_id
= 0;
172 lport
= sin6
->sin6_port
;
173 if (IN6_IS_ADDR_MULTICAST(&sin6
->sin6_addr
)) {
175 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
176 * allow compepte duplication of binding if
177 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
178 * and a multicast address is bound on both
179 * new and duplicated sockets.
181 if (so
->so_options
& SO_REUSEADDR
)
182 reuseport
= SO_REUSEADDR
|SO_REUSEPORT
;
183 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6
->sin6_addr
)) {
184 struct ifaddr
*ia
= NULL
;
186 sin6
->sin6_port
= 0; /* yech... */
187 if (!prison_replace_wildcards(td
, (struct sockaddr
*)sin6
)) {
188 sin6
->sin6_addr
= kin6addr_any
;
191 if ((ia
= ifa_ifwithaddr((struct sockaddr
*)sin6
)) == NULL
)
192 return (EADDRNOTAVAIL
);
195 * XXX: bind to an anycast address might accidentally
196 * cause sending a packet with anycast source address.
197 * We should allow to bind to a deprecated address, since
198 * the application dare to use it.
201 ((struct in6_ifaddr
*)ia
)->ia6_flags
&
202 (IN6_IFF_ANYCAST
|IN6_IFF_NOTREADY
|IN6_IFF_DETACHED
))
203 return (EADDRNOTAVAIL
);
206 inp
->in6p_laddr
= sin6
->sin6_addr
;
210 lport_ho
= ntohs(lport
);
213 if (lport_ho
< IPV6PORT_RESERVED
&& cred
&&
214 priv_check_cred(cred
, PRIV_NETINET_RESERVEDPORT
, 0)) {
215 inp
->in6p_laddr
= kin6addr_any
;
220 * Locate the proper portinfo based on lport
222 pcbinfo
= inp
->inp_pcbinfo
;
224 &pcbinfo
->portinfo
[lport_ho
& pcbinfo
->portinfo_mask
];
225 KKASSERT((lport_ho
& pcbinfo
->portinfo_mask
) ==
229 * This has to be atomic. If the porthash is shared across
230 * multiple protocol threads (aka tcp) then the token must
233 GET_PORT_TOKEN(portinfo
);
235 if (so
->so_cred
->cr_uid
!= 0 &&
236 !IN6_IS_ADDR_MULTICAST(&sin6
->sin6_addr
)) {
237 t
= in6_pcblookup_local(portinfo
,
238 &sin6
->sin6_addr
, lport
, INPLOOKUP_WILDCARD
, cred
);
240 (so
->so_cred
->cr_uid
!=
241 t
->inp_socket
->so_cred
->cr_uid
)) {
242 inp
->in6p_laddr
= kin6addr_any
;
247 if (cred
&& cred
->cr_prison
&&
248 !prison_replace_wildcards(td
, nam
)) {
249 inp
->in6p_laddr
= kin6addr_any
;
250 error
= EADDRNOTAVAIL
;
253 t
= in6_pcblookup_local(portinfo
, &sin6
->sin6_addr
, lport
,
255 if (t
&& (reuseport
& t
->inp_socket
->so_options
) == 0) {
256 inp
->in6p_laddr
= kin6addr_any
;
261 inp
->inp_lport
= lport
;
262 in_pcbinsporthash(portinfo
, inp
);
265 REL_PORT_TOKEN(portinfo
);
269 jsin6
.sin6_addr
= inp
->in6p_laddr
;
270 jsin6
.sin6_family
= AF_INET6
;
271 if (!prison_replace_wildcards(td
, (struct sockaddr
*)&jsin6
)) {
272 inp
->in6p_laddr
= kin6addr_any
;
277 return in6_pcbsetlport(&inp
->in6p_laddr
, inp
, td
);
282 * Transform old in6_pcbconnect() into an inner subroutine for new
283 * in6_pcbconnect(): Do some validity-checking on the remote
284 * address (in mbuf 'nam') and then determine local host address
285 * (i.e., which interface) to use to access that remote host.
287 * This preserves definition of in6_pcbconnect(), while supporting a
288 * slightly different version for T/TCP. (This is more than
289 * a bit of a kludge, but cleaning up the internal interfaces would
290 * have forced minor changes in every protocol).
294 in6_pcbladdr(struct inpcb
*inp
, struct sockaddr
*nam
,
295 struct in6_addr
**plocal_addr6
, struct thread
*td
)
297 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)nam
;
298 struct ifnet
*ifp
= NULL
;
301 if (nam
->sa_len
!= sizeof (*sin6
))
303 if (sin6
->sin6_family
!= AF_INET6
)
304 return (EAFNOSUPPORT
);
305 if (sin6
->sin6_port
== 0)
306 return (EADDRNOTAVAIL
);
308 /* KAME hack: embed scopeid */
309 if (in6_embedscope(&sin6
->sin6_addr
, sin6
, inp
, &ifp
) != 0)
314 * If the destination address is UNSPECIFIED addr,
315 * use the loopback addr, e.g ::1.
317 if (IN6_IS_ADDR_UNSPECIFIED(&sin6
->sin6_addr
))
318 sin6
->sin6_addr
= kin6addr_loopback
;
322 * XXX: in6_selectsrc might replace the bound local address
323 * with the address specified by setsockopt(IPV6_PKTINFO).
324 * Is it the intended behavior?
326 *plocal_addr6
= in6_selectsrc(sin6
, inp
->in6p_outputopts
,
329 &inp
->in6p_laddr
, &error
, td
);
330 if (*plocal_addr6
== NULL
) {
332 error
= EADDRNOTAVAIL
;
336 * Don't do pcblookup call here; return interface in
338 * and exit to caller, that will do the lookup.
342 if (inp
->in6p_route
.ro_rt
)
343 ifp
= inp
->in6p_route
.ro_rt
->rt_ifp
;
350 * Connect from a socket to a specified address.
351 * Both address and port must be specified in argument sin.
352 * If don't have a local address for this socket yet,
356 in6_pcbconnect(struct inpcb
*inp
, struct sockaddr
*nam
, struct thread
*td
)
358 struct in6_addr
*addr6
;
359 struct sockaddr_in6
*sin6
= (struct sockaddr_in6
*)nam
;
362 /* Reject v4-mapped address */
363 if (IN6_IS_ADDR_V4MAPPED(&sin6
->sin6_addr
))
364 return EADDRNOTAVAIL
;
367 * Call inner routine, to assign local interface address.
368 * in6_pcbladdr() may automatically fill in sin6_scope_id.
370 if ((error
= in6_pcbladdr(inp
, nam
, &addr6
, td
)) != 0)
373 if (in6_pcblookup_hash(inp
->inp_pcbinfo
, &sin6
->sin6_addr
,
375 IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)
376 ? addr6
: &inp
->in6p_laddr
,
377 inp
->inp_lport
, 0, NULL
) != NULL
) {
380 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
381 if (inp
->inp_lport
== 0) {
382 error
= in6_pcbbind(inp
, NULL
, td
);
386 inp
->in6p_laddr
= *addr6
;
388 inp
->in6p_faddr
= sin6
->sin6_addr
;
389 inp
->inp_fport
= sin6
->sin6_port
;
390 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
391 inp
->in6p_flowinfo
&= ~IPV6_FLOWLABEL_MASK
;
392 if (inp
->in6p_flags
& IN6P_AUTOFLOWLABEL
)
393 inp
->in6p_flowinfo
|=
394 (htonl(ip6_flow_seq
++) & IPV6_FLOWLABEL_MASK
);
396 in_pcbinsconnhash(inp
);
402 * Return an IPv6 address, which is the most appropriate for given
403 * destination and user specified options.
404 * If necessary, this function lookups the routing table and return
405 * an entry to the caller for later use.
408 in6_selectsrc(struct sockaddr_in6
*dstsock
, struct ip6_pktopts
*opts
,
409 struct ip6_moptions
*mopts
, struct route_in6
*ro
,
410 struct in6_addr
*laddr
, int *errorp
, struct thread
*td
)
412 struct sockaddr_in6 jsin6
;
413 struct ucred
*cred
= NULL
;
414 struct in6_addr
*dst
;
415 struct in6_ifaddr
*ia6
= 0;
416 struct in6_pktinfo
*pi
= NULL
;
419 if (td
&& td
->td_proc
&& td
->td_proc
->p_ucred
)
420 cred
= td
->td_proc
->p_ucred
;
421 if (cred
&& cred
->cr_prison
)
423 jsin6
.sin6_family
= AF_INET6
;
424 dst
= &dstsock
->sin6_addr
;
428 * If the source address is explicitly specified by the caller,
431 if (opts
&& (pi
= opts
->ip6po_pktinfo
) &&
432 !IN6_IS_ADDR_UNSPECIFIED(&pi
->ipi6_addr
)) {
433 jsin6
.sin6_addr
= pi
->ipi6_addr
;
434 if (jailed
&& !jailed_ip(cred
->cr_prison
,
435 (struct sockaddr
*)&jsin6
)) {
438 return (&pi
->ipi6_addr
);
443 * If the source address is not specified but the socket(if any)
444 * is already bound, use the bound address.
446 if (laddr
&& !IN6_IS_ADDR_UNSPECIFIED(laddr
)) {
447 jsin6
.sin6_addr
= *laddr
;
448 if (jailed
&& !jailed_ip(cred
->cr_prison
,
449 (struct sockaddr
*)&jsin6
)) {
457 * If the caller doesn't specify the source address but
458 * the outgoing interface, use an address associated with
461 if (pi
&& pi
->ipi6_ifindex
) {
462 /* XXX boundary check is assumed to be already done. */
463 ia6
= in6_ifawithscope(ifindex2ifnet
[pi
->ipi6_ifindex
],
467 jsin6
.sin6_addr
= (&ia6
->ia_addr
)->sin6_addr
;
468 if (!jailed_ip(cred
->cr_prison
,
469 (struct sockaddr
*)&jsin6
))
474 *errorp
= EADDRNOTAVAIL
;
477 return (&satosin6(&ia6
->ia_addr
)->sin6_addr
);
481 * If the destination address is a link-local unicast address or
482 * a multicast address, and if the outgoing interface is specified
483 * by the sin6_scope_id filed, use an address associated with the
485 * XXX: We're now trying to define more specific semantics of
486 * sin6_scope_id field, so this part will be rewritten in
489 if ((IN6_IS_ADDR_LINKLOCAL(dst
) || IN6_IS_ADDR_MULTICAST(dst
)) &&
490 dstsock
->sin6_scope_id
) {
492 * I'm not sure if boundary check for scope_id is done
495 if (dstsock
->sin6_scope_id
< 0 ||
496 if_index
< dstsock
->sin6_scope_id
) {
497 *errorp
= ENXIO
; /* XXX: better error? */
500 ia6
= in6_ifawithscope(ifindex2ifnet
[dstsock
->sin6_scope_id
],
504 jsin6
.sin6_addr
= (&ia6
->ia_addr
)->sin6_addr
;
505 if (!jailed_ip(cred
->cr_prison
,
506 (struct sockaddr
*)&jsin6
))
511 *errorp
= EADDRNOTAVAIL
;
514 return (&satosin6(&ia6
->ia_addr
)->sin6_addr
);
518 * If the destination address is a multicast address and
519 * the outgoing interface for the address is specified
520 * by the caller, use an address associated with the interface.
521 * There is a sanity check here; if the destination has node-local
522 * scope, the outgoing interfacde should be a loopback address.
523 * Even if the outgoing interface is not specified, we also
524 * choose a loopback interface as the outgoing interface.
526 if (!jailed
&& IN6_IS_ADDR_MULTICAST(dst
)) {
527 struct ifnet
*ifp
= mopts
? mopts
->im6o_multicast_ifp
: NULL
;
529 if (ifp
== NULL
&& IN6_IS_ADDR_MC_NODELOCAL(dst
)) {
534 ia6
= in6_ifawithscope(ifp
, dst
, cred
);
536 *errorp
= EADDRNOTAVAIL
;
539 return (&ia6
->ia_addr
.sin6_addr
);
544 * If the next hop address for the packet is specified
545 * by caller, use an address associated with the route
549 struct sockaddr_in6
*sin6_next
;
552 if (opts
&& opts
->ip6po_nexthop
) {
553 sin6_next
= satosin6(opts
->ip6po_nexthop
);
554 rt
= nd6_lookup(&sin6_next
->sin6_addr
, 1, NULL
);
556 ia6
= in6_ifawithscope(rt
->rt_ifp
, dst
, cred
);
558 ia6
= ifatoia6(rt
->rt_ifa
);
561 jsin6
.sin6_addr
= (&ia6
->ia_addr
)->sin6_addr
;
562 if (!jailed_ip(cred
->cr_prison
,
563 (struct sockaddr
*)&jsin6
))
568 *errorp
= EADDRNOTAVAIL
;
571 return (&satosin6(&ia6
->ia_addr
)->sin6_addr
);
576 * If route is known or can be allocated now,
577 * our src addr is taken from the i/f, else punt.
581 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro
->ro_dst
)->sin6_addr
, dst
)) {
585 if (ro
->ro_rt
== NULL
|| ro
->ro_rt
->rt_ifp
== NULL
) {
586 struct sockaddr_in6
*dst6
;
588 /* No route yet, so try to acquire one */
589 bzero(&ro
->ro_dst
, sizeof(struct sockaddr_in6
));
591 dst6
->sin6_family
= AF_INET6
;
592 dst6
->sin6_len
= sizeof(struct sockaddr_in6
);
593 dst6
->sin6_addr
= *dst
;
594 if (!jailed
&& IN6_IS_ADDR_MULTICAST(dst
)) {
596 rtpurelookup((struct sockaddr
*)&ro
->ro_dst
);
598 rtalloc((struct route
*)ro
);
603 * in_pcbconnect() checks out IFF_LOOPBACK to skip using
604 * the address. But we don't know why it does so.
605 * It is necessary to ensure the scope even for lo0
606 * so doesn't check out IFF_LOOPBACK.
610 ia6
= in6_ifawithscope(ro
->ro_rt
->rt_ifa
->ifa_ifp
, dst
, cred
);
612 jsin6
.sin6_addr
= (&ia6
->ia_addr
)->sin6_addr
;
613 if (!jailed_ip(cred
->cr_prison
,
614 (struct sockaddr
*)&jsin6
))
618 if (ia6
== 0) /* xxx scope error ?*/
619 ia6
= ifatoia6(ro
->ro_rt
->rt_ifa
);
622 jsin6
.sin6_addr
= (&ia6
->ia_addr
)->sin6_addr
;
623 if (!jailed_ip(cred
->cr_prison
,
624 (struct sockaddr
*)&jsin6
))
629 *errorp
= EHOSTUNREACH
; /* no route */
632 return (&satosin6(&ia6
->ia_addr
)->sin6_addr
);
635 *errorp
= EADDRNOTAVAIL
;
640 * Default hop limit selection. The precedence is as follows:
641 * 1. Hoplimit valued specified via ioctl.
642 * 2. (If the outgoing interface is detected) the current
643 * hop limit of the interface specified by router advertisement.
644 * 3. The system default hoplimit.
647 in6_selecthlim(struct in6pcb
*in6p
, struct ifnet
*ifp
)
651 if (in6p
&& in6p
->in6p_hops
>= 0) {
652 return (in6p
->in6p_hops
);
654 hlim
= ND_IFINFO(ifp
)->chlim
;
655 if (hlim
< ip6_minhlim
)
666 in6_pcbdisconnect(struct inpcb
*inp
)
668 bzero((caddr_t
)&inp
->in6p_faddr
, sizeof(inp
->in6p_faddr
));
670 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
671 inp
->in6p_flowinfo
&= ~IPV6_FLOWLABEL_MASK
;
672 in_pcbremconnhash(inp
);
673 if (inp
->inp_socket
->so_state
& SS_NOFDREF
)
678 in6_pcbdetach(struct inpcb
*inp
)
680 struct socket
*so
= inp
->inp_socket
;
681 struct inpcbinfo
*ipi
= inp
->inp_pcbinfo
;
684 if (inp
->in6p_sp
!= NULL
)
685 ipsec6_delete_pcbpolicy(inp
);
687 inp
->inp_gencnt
= ++ipi
->ipi_gencnt
;
690 KKASSERT((so
->so_state
& SS_ASSERTINPROG
) == 0);
691 sofree(so
); /* remove pcb ref */
693 if (inp
->in6p_options
)
694 m_freem(inp
->in6p_options
);
695 ip6_freepcbopts(inp
->in6p_outputopts
);
696 ip6_freemoptions(inp
->in6p_moptions
);
697 if (inp
->in6p_route
.ro_rt
)
698 rtfree(inp
->in6p_route
.ro_rt
);
699 /* Check and free IPv4 related resources in case of mapped addr */
700 if (inp
->inp_options
)
701 m_free(inp
->inp_options
);
702 ip_freemoptions(inp
->inp_moptions
);
708 * The calling convention of in6_setsockaddr() and in6_setpeeraddr() was
709 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
710 * in struct pr_usrreqs, so that protocols can just reference then directly
711 * without the need for a wrapper function. The socket must have a valid
712 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
713 * except through a kernel programming error, so it is acceptable to panic
714 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap
715 * because there actually /is/ a programming error somewhere... XXX)
718 in6_setsockaddr_dispatch(netmsg_t msg
)
722 error
= in6_setsockaddr(msg
->sockaddr
.base
.nm_so
, msg
->sockaddr
.nm_nam
);
723 lwkt_replymsg(&msg
->sockaddr
.base
.lmsg
, error
);
727 in6_setsockaddr(struct socket
*so
, struct sockaddr
**nam
)
730 struct sockaddr_in6
*sin6
;
733 * Do the malloc first in case it blocks.
735 sin6
= kmalloc(sizeof *sin6
, M_SONAME
, M_WAITOK
| M_ZERO
);
736 sin6
->sin6_family
= AF_INET6
;
737 sin6
->sin6_len
= sizeof(*sin6
);
743 kfree(sin6
, M_SONAME
);
746 sin6
->sin6_port
= inp
->inp_lport
;
747 sin6
->sin6_addr
= inp
->in6p_laddr
;
749 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
750 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
752 sin6
->sin6_scope_id
= 0; /*XXX*/
753 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
754 sin6
->sin6_addr
.s6_addr16
[1] = 0;
756 *nam
= (struct sockaddr
*)sin6
;
761 in6_setpeeraddr_dispatch(netmsg_t msg
)
765 error
= in6_setpeeraddr(msg
->peeraddr
.base
.nm_so
, msg
->peeraddr
.nm_nam
);
766 lwkt_replymsg(&msg
->peeraddr
.base
.lmsg
, error
);
770 in6_setpeeraddr(struct socket
*so
, struct sockaddr
**nam
)
773 struct sockaddr_in6
*sin6
;
776 * Do the malloc first in case it blocks.
778 sin6
= kmalloc(sizeof(*sin6
), M_SONAME
, M_WAITOK
| M_ZERO
);
779 sin6
->sin6_family
= AF_INET6
;
780 sin6
->sin6_len
= sizeof(struct sockaddr_in6
);
786 kfree(sin6
, M_SONAME
);
789 sin6
->sin6_port
= inp
->inp_fport
;
790 sin6
->sin6_addr
= inp
->in6p_faddr
;
792 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
793 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
795 sin6
->sin6_scope_id
= 0; /*XXX*/
796 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
797 sin6
->sin6_addr
.s6_addr16
[1] = 0;
799 *nam
= (struct sockaddr
*)sin6
;
804 * Pass some notification to all connections of a protocol
805 * associated with address dst. The local address and/or port numbers
806 * may be specified to limit the search. The "usual action" will be
807 * taken, depending on the ctlinput cmd. The caller must filter any
808 * cmds that are uninteresting (e.g., no error in the map).
809 * Call the protocol specific routine (if any) to report
810 * any errors for each matching socket.
813 in6_pcbnotify(struct inpcbinfo
*pcbinfo
, struct sockaddr
*dst
, in_port_t fport
,
814 const struct sockaddr
*src
, in_port_t lport
, int cmd
, int arg
,
817 struct inpcb
*inp
, *marker
;
818 struct sockaddr_in6 sa6_src
, *sa6_dst
;
821 if ((unsigned)cmd
>= PRC_NCMDS
|| dst
->sa_family
!= AF_INET6
)
824 sa6_dst
= (struct sockaddr_in6
*)dst
;
825 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst
->sin6_addr
))
829 * note that src can be NULL when we get notify by local fragmentation.
831 sa6_src
= (src
== NULL
) ? sa6_any
: *(const struct sockaddr_in6
*)src
;
832 flowinfo
= sa6_src
.sin6_flowinfo
;
835 * Redirects go to all references to the destination,
836 * and use in6_rtchange to invalidate the route cache.
837 * Dead host indications: also use in6_rtchange to invalidate
838 * the cache, and deliver the error to all the sockets.
839 * Otherwise, if we have knowledge of the local port and address,
840 * deliver only to that socket.
842 if (PRC_IS_REDIRECT(cmd
) || cmd
== PRC_HOSTDEAD
) {
845 bzero((caddr_t
)&sa6_src
.sin6_addr
, sizeof(sa6_src
.sin6_addr
));
847 if (cmd
!= PRC_HOSTDEAD
)
848 notify
= in6_rtchange
;
850 if (cmd
!= PRC_MSGSIZE
)
851 arg
= inet6ctlerrmap
[cmd
];
853 marker
= in_pcbmarker(mycpuid
);
855 GET_PCBINFO_TOKEN(pcbinfo
);
857 LIST_INSERT_HEAD(&pcbinfo
->pcblisthead
, marker
, inp_list
);
858 while ((inp
= LIST_NEXT(marker
, inp_list
)) != NULL
) {
859 LIST_REMOVE(marker
, inp_list
);
860 LIST_INSERT_AFTER(inp
, marker
, inp_list
);
862 if (inp
->inp_flags
& INP_PLACEMARKER
)
865 if (!INP_ISIPV6(inp
))
868 * If the error designates a new path MTU for a destination
869 * and the application (associated with this socket) wanted to
870 * know the value, notify. Note that we notify for all
871 * disconnected sockets if the corresponding application
872 * wanted. This is because some UDP applications keep sending
873 * sockets disconnected.
874 * XXX: should we avoid to notify the value to TCP sockets?
876 if (cmd
== PRC_MSGSIZE
&& (inp
->inp_flags
& IN6P_MTU
) != 0 &&
877 (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) ||
878 IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, &sa6_dst
->sin6_addr
))) {
879 ip6_notify_pmtu(inp
, (struct sockaddr_in6
*)dst
, &arg
);
883 * Detect if we should notify the error. If no source and
884 * destination ports are specifed, but non-zero flowinfo and
885 * local address match, notify the error. This is the case
886 * when the error is delivered with an encrypted buffer
887 * by ESP. Otherwise, just compare addresses and ports
890 if (lport
== 0 && fport
== 0 && flowinfo
&&
891 inp
->inp_socket
!= NULL
&&
892 flowinfo
== (inp
->in6p_flowinfo
& IPV6_FLOWLABEL_MASK
) &&
893 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, &sa6_src
.sin6_addr
))
895 else if (!IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
,
896 &sa6_dst
->sin6_addr
) ||
897 inp
->inp_socket
== 0 ||
898 (lport
&& inp
->inp_lport
!= lport
) ||
899 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src
.sin6_addr
) &&
900 !IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
901 &sa6_src
.sin6_addr
)) ||
902 (fport
&& inp
->inp_fport
!= fport
))
909 LIST_REMOVE(marker
, inp_list
);
911 REL_PCBINFO_TOKEN(pcbinfo
);
915 * Lookup a PCB based on the local address and port.
918 in6_pcblookup_local(struct inpcbportinfo
*portinfo
,
919 const struct in6_addr
*laddr
, u_int lport_arg
, int wild_okay
,
923 int matchwild
= 3, wildcard
;
924 u_short lport
= lport_arg
;
925 struct inpcbporthead
*porthash
;
926 struct inpcbport
*phd
;
927 struct inpcb
*match
= NULL
;
930 * If the porthashbase is shared across several cpus, it must
933 ASSERT_PORT_TOKEN_HELD(portinfo
);
936 * Best fit PCB lookup.
938 * First see if this local port is in use by looking on the
941 porthash
= &portinfo
->porthashbase
[
942 INP_PCBPORTHASH(lport
, portinfo
->porthashmask
)];
943 LIST_FOREACH(phd
, porthash
, phd_hash
) {
944 if (phd
->phd_port
== lport
)
950 * Port is in use by one or more PCBs. Look for best
953 LIST_FOREACH(inp
, &phd
->phd_pcblist
, inp_portlist
) {
955 if (!INP_ISIPV6(inp
))
957 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
))
959 if (!IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
960 if (IN6_IS_ADDR_UNSPECIFIED(laddr
))
962 else if (!IN6_ARE_ADDR_EQUAL(
963 &inp
->in6p_laddr
, laddr
))
966 if (!IN6_IS_ADDR_UNSPECIFIED(laddr
))
969 if (wildcard
&& !wild_okay
)
971 if (wildcard
< matchwild
&&
974 inp
->inp_socket
->so_cred
->cr_prison
)) {
976 matchwild
= wildcard
;
980 matchwild
= wildcard
;
988 in6_pcbpurgeif0(struct inpcbinfo
*pcbinfo
, struct ifnet
*ifp
)
990 struct in6pcb
*in6p
, *marker
;
991 struct ip6_moptions
*im6o
;
992 struct in6_multi_mship
*imm
, *nimm
;
995 * We only need to make sure that we are in netisr0, where all
996 * multicast operation happen. We could check inpcbinfo which
997 * does not belong to netisr0 by holding the inpcbinfo's token.
998 * In this case, the pcbinfo must be able to be shared, i.e.
999 * pcbinfo->infotoken is not NULL.
1001 KASSERT(&curthread
->td_msgport
== netisr_cpuport(0),
1002 ("not in netisr0"));
1003 KASSERT(pcbinfo
->cpu
== 0 || pcbinfo
->infotoken
!= NULL
,
1004 ("pcbinfo could not be shared"));
1007 * Get a marker for the current netisr (netisr0).
1009 * It is possible that the multicast address deletion blocks,
1010 * which could cause temporary token releasing. So we use
1011 * inpcb marker here to get a coherent view of the inpcb list.
1013 * While, on the other hand, moptions are only added and deleted
1014 * in netisr0, so we would not see staled moption or miss moption
1015 * even if the token was released due to the blocking multicast
1018 marker
= in_pcbmarker(mycpuid
);
1020 GET_PCBINFO_TOKEN(pcbinfo
);
1022 LIST_INSERT_HEAD(&pcbinfo
->pcblisthead
, marker
, inp_list
);
1023 while ((in6p
= LIST_NEXT(marker
, inp_list
)) != NULL
) {
1024 LIST_REMOVE(marker
, inp_list
);
1025 LIST_INSERT_AFTER(in6p
, marker
, inp_list
);
1027 if (in6p
->in6p_flags
& INP_PLACEMARKER
)
1029 im6o
= in6p
->in6p_moptions
;
1030 if (INP_ISIPV6(in6p
) && im6o
) {
1032 * Unselect the outgoing interface if it is being
1035 if (im6o
->im6o_multicast_ifp
== ifp
)
1036 im6o
->im6o_multicast_ifp
= NULL
;
1039 * Drop multicast group membership if we joined
1040 * through the interface being detached.
1041 * XXX controversial - is it really legal for kernel
1044 for (imm
= im6o
->im6o_memberships
.lh_first
;
1045 imm
!= NULL
; imm
= nimm
) {
1046 nimm
= imm
->i6mm_chain
.le_next
;
1047 if (imm
->i6mm_maddr
->in6m_ifp
== ifp
) {
1048 LIST_REMOVE(imm
, i6mm_chain
);
1049 in6_delmulti(imm
->i6mm_maddr
);
1050 kfree(imm
, M_IPMADDR
);
1055 LIST_REMOVE(marker
, inp_list
);
1057 REL_PCBINFO_TOKEN(pcbinfo
);
1061 * Check for alternatives when higher level complains
1062 * about service problems. For now, invalidate cached
1063 * routing information. If the route was created dynamically
1064 * (by a redirect), time to try a default gateway again.
1067 in6_losing(struct inpcb
*in6p
)
1070 struct rt_addrinfo info
;
1072 if ((rt
= in6p
->in6p_route
.ro_rt
) != NULL
) {
1073 bzero((caddr_t
)&info
, sizeof(info
));
1074 info
.rti_flags
= rt
->rt_flags
;
1075 info
.rti_info
[RTAX_DST
] = rt_key(rt
);
1076 info
.rti_info
[RTAX_GATEWAY
] = rt
->rt_gateway
;
1077 info
.rti_info
[RTAX_NETMASK
] = rt_mask(rt
);
1078 rt_missmsg(RTM_LOSING
, &info
, rt
->rt_flags
, 0);
1079 if (rt
->rt_flags
& RTF_DYNAMIC
) {
1080 rtrequest(RTM_DELETE
, rt_key(rt
), rt
->rt_gateway
,
1081 rt_mask(rt
), rt
->rt_flags
, NULL
);
1083 in6p
->in6p_route
.ro_rt
= NULL
;
1086 * A new route can be allocated
1087 * the next time output is attempted.
1093 * After a routing change, flush old routing
1094 * and allocate a (hopefully) better one.
1097 in6_rtchange(struct inpcb
*inp
, int error
)
1099 if (inp
->in6p_route
.ro_rt
) {
1100 rtfree(inp
->in6p_route
.ro_rt
);
1101 inp
->in6p_route
.ro_rt
= 0;
1103 * A new route can be allocated the next time
1104 * output is attempted.
1110 * Lookup PCB in hash list.
1113 in6_pcblookup_hash(struct inpcbinfo
*pcbinfo
, struct in6_addr
*faddr
,
1114 u_int fport_arg
, struct in6_addr
*laddr
, u_int lport_arg
,
1115 int wildcard
, struct ifnet
*ifp
)
1117 struct inpcbhead
*head
;
1119 struct inpcb
*jinp
= NULL
;
1120 u_short fport
= fport_arg
, lport
= lport_arg
;
1123 if (faithprefix_p
!= NULL
)
1124 faith
= (*faithprefix_p
)(laddr
);
1129 * First look for an exact match.
1131 head
= &pcbinfo
->hashbase
[INP_PCBCONNHASH(faddr
->s6_addr32
[3] /* XXX */,
1133 laddr
->s6_addr32
[3], /* XXX JH */
1135 pcbinfo
->hashmask
)];
1136 LIST_FOREACH(inp
, head
, inp_hash
) {
1137 if (!INP_ISIPV6(inp
))
1139 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_faddr
, faddr
) &&
1140 IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
, laddr
) &&
1141 inp
->inp_fport
== fport
&&
1142 inp
->inp_lport
== lport
) {
1146 if (inp
->inp_socket
== NULL
||
1147 inp
->inp_socket
->so_cred
->cr_prison
== NULL
) {
1159 struct inpcontainerhead
*chead
;
1160 struct inpcontainer
*ic
;
1161 struct inpcb
*local_wild
= NULL
;
1162 struct inpcb
*jinp_wild
= NULL
;
1163 struct sockaddr_in6 jsin6
;
1167 * Order of socket selection:
1168 * 1. non-jailed, non-wild.
1169 * 2. non-jailed, wild.
1170 * 3. jailed, non-wild.
1173 jsin6
.sin6_family
= AF_INET6
;
1174 chead
= &pcbinfo
->wildcardhashbase
[INP_PCBWILDCARDHASH(lport
,
1175 pcbinfo
->wildcardhashmask
)];
1177 GET_PCBINFO_TOKEN(pcbinfo
);
1178 LIST_FOREACH(ic
, chead
, ic_list
) {
1180 if (inp
->inp_flags
& INP_PLACEMARKER
)
1183 if (!INP_ISIPV6(inp
))
1185 if (inp
->inp_socket
!= NULL
)
1186 cred
= inp
->inp_socket
->so_cred
;
1190 if (cred
!= NULL
&& jailed(cred
)) {
1194 jsin6
.sin6_addr
= *laddr
;
1195 if (!jailed_ip(cred
->cr_prison
,
1196 (struct sockaddr
*)&jsin6
))
1200 if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_faddr
) &&
1201 inp
->inp_lport
== lport
) {
1202 if (faith
&& (inp
->inp_flags
& INP_FAITH
) == 0)
1204 if (IN6_ARE_ADDR_EQUAL(&inp
->in6p_laddr
,
1206 if (cred
!= NULL
&& jailed(cred
)) {
1209 REL_PCBINFO_TOKEN(pcbinfo
);
1212 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp
->in6p_laddr
)) {
1213 if (cred
!= NULL
&& jailed(cred
))
1220 REL_PCBINFO_TOKEN(pcbinfo
);
1222 if (local_wild
!= NULL
)
1223 return (local_wild
);
1236 init_sin6(struct sockaddr_in6
*sin6
, struct mbuf
*m
)
1240 ip
= mtod(m
, struct ip6_hdr
*);
1241 bzero(sin6
, sizeof(*sin6
));
1242 sin6
->sin6_len
= sizeof(*sin6
);
1243 sin6
->sin6_family
= AF_INET6
;
1244 sin6
->sin6_addr
= ip
->ip6_src
;
1245 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
1246 sin6
->sin6_addr
.s6_addr16
[1] = 0;
1247 sin6
->sin6_scope_id
=
1248 (m
->m_pkthdr
.rcvif
&& IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
1249 ? m
->m_pkthdr
.rcvif
->if_index
: 0;
1255 in6_savefaddr(struct socket
*so
, const struct sockaddr
*faddr
)
1257 struct sockaddr_in6
*sin6
;
1259 KASSERT(faddr
->sa_family
== AF_INET6
,
1260 ("not AF_INET6 faddr %d", faddr
->sa_family
));
1262 sin6
= kmalloc(sizeof(*sin6
), M_SONAME
, M_WAITOK
| M_ZERO
);
1263 sin6
->sin6_family
= AF_INET6
;
1264 sin6
->sin6_len
= sizeof(*sin6
);
1266 sin6
->sin6_port
= ((const struct sockaddr_in6
*)faddr
)->sin6_port
;
1267 sin6
->sin6_addr
= ((const struct sockaddr_in6
*)faddr
)->sin6_addr
;
1269 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
1270 sin6
->sin6_scope_id
= ntohs(sin6
->sin6_addr
.s6_addr16
[1]);
1272 sin6
->sin6_scope_id
= 0; /*XXX*/
1273 if (IN6_IS_SCOPE_LINKLOCAL(&sin6
->sin6_addr
))
1274 sin6
->sin6_addr
.s6_addr16
[1] = 0;
1276 so
->so_faddr
= (struct sockaddr
*)sin6
;