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 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
241 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
242 (t->inp_socket->so_options & SO_REUSEPORT) == 0) &&
243 (so->so_cred->cr_uid !=
244 t->inp_socket->so_cred->cr_uid)) {
245 inp->in6p_laddr = kin6addr_any;
250 if (cred && cred->cr_prison &&
251 !prison_replace_wildcards(td, nam)) {
252 inp->in6p_laddr = kin6addr_any;
253 error = EADDRNOTAVAIL;
256 t = in6_pcblookup_local(portinfo, &sin6->sin6_addr, lport,
258 if (t && (reuseport & t->inp_socket->so_options) == 0) {
259 inp->in6p_laddr = kin6addr_any;
264 inp->inp_lport = lport;
265 in_pcbinsporthash(portinfo, inp);
268 REL_PORT_TOKEN(portinfo);
272 jsin6.sin6_addr = inp->in6p_laddr;
273 jsin6.sin6_family = AF_INET6;
274 if (!prison_replace_wildcards(td, (struct sockaddr*)&jsin6)) {
275 inp->in6p_laddr = kin6addr_any;
280 return in6_pcbsetlport(&inp->in6p_laddr, inp, td);
285 * Transform old in6_pcbconnect() into an inner subroutine for new
286 * in6_pcbconnect(): Do some validity-checking on the remote
287 * address (in mbuf 'nam') and then determine local host address
288 * (i.e., which interface) to use to access that remote host.
290 * This preserves definition of in6_pcbconnect(), while supporting a
291 * slightly different version for T/TCP. (This is more than
292 * a bit of a kludge, but cleaning up the internal interfaces would
293 * have forced minor changes in every protocol).
297 in6_pcbladdr(struct inpcb *inp, struct sockaddr *nam,
298 struct in6_addr **plocal_addr6, struct thread *td)
300 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
301 struct ifnet *ifp = NULL;
304 if (nam->sa_len != sizeof (*sin6))
306 if (sin6->sin6_family != AF_INET6)
307 return (EAFNOSUPPORT);
308 if (sin6->sin6_port == 0)
309 return (EADDRNOTAVAIL);
311 /* KAME hack: embed scopeid */
312 if (in6_embedscope(&sin6->sin6_addr, sin6, inp, &ifp) != 0)
317 * If the destination address is UNSPECIFIED addr,
318 * use the loopback addr, e.g ::1.
320 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
321 sin6->sin6_addr = kin6addr_loopback;
325 * XXX: in6_selectsrc might replace the bound local address
326 * with the address specified by setsockopt(IPV6_PKTINFO).
327 * Is it the intended behavior?
329 *plocal_addr6 = in6_selectsrc(sin6, inp->in6p_outputopts,
332 &inp->in6p_laddr, &error, td);
333 if (*plocal_addr6 == NULL) {
335 error = EADDRNOTAVAIL;
339 * Don't do pcblookup call here; return interface in
341 * and exit to caller, that will do the lookup.
345 if (inp->in6p_route.ro_rt)
346 ifp = inp->in6p_route.ro_rt->rt_ifp;
353 * Connect from a socket to a specified address.
354 * Both address and port must be specified in argument sin.
355 * If don't have a local address for this socket yet,
359 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
361 struct in6_addr *addr6;
362 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
365 /* Reject v4-mapped address */
366 if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr))
367 return EADDRNOTAVAIL;
370 * Call inner routine, to assign local interface address.
371 * in6_pcbladdr() may automatically fill in sin6_scope_id.
373 if ((error = in6_pcbladdr(inp, nam, &addr6, td)) != 0)
376 if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr,
378 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
379 ? addr6 : &inp->in6p_laddr,
380 inp->inp_lport, 0, NULL) != NULL) {
383 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
384 if (inp->inp_lport == 0) {
385 error = in6_pcbbind(inp, NULL, td);
389 inp->in6p_laddr = *addr6;
391 inp->in6p_faddr = sin6->sin6_addr;
392 inp->inp_fport = sin6->sin6_port;
393 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
394 inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
395 if (inp->in6p_flags & IN6P_AUTOFLOWLABEL)
396 inp->in6p_flowinfo |=
397 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK);
399 in_pcbinsconnhash(inp);
405 * Return an IPv6 address, which is the most appropriate for given
406 * destination and user specified options.
407 * If necessary, this function lookups the routing table and return
408 * an entry to the caller for later use.
411 in6_selectsrc(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
412 struct ip6_moptions *mopts, struct route_in6 *ro,
413 struct in6_addr *laddr, int *errorp, struct thread *td)
415 struct sockaddr_in6 jsin6;
416 struct ucred *cred = NULL;
417 struct in6_addr *dst;
418 struct in6_ifaddr *ia6 = 0;
419 struct in6_pktinfo *pi = NULL;
422 if (td && td->td_proc && td->td_proc->p_ucred)
423 cred = td->td_proc->p_ucred;
424 if (cred && cred->cr_prison)
426 jsin6.sin6_family = AF_INET6;
427 dst = &dstsock->sin6_addr;
431 * If the source address is explicitly specified by the caller,
434 if (opts && (pi = opts->ip6po_pktinfo) &&
435 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) {
436 jsin6.sin6_addr = pi->ipi6_addr;
437 if (jailed && !jailed_ip(cred->cr_prison,
438 (struct sockaddr *)&jsin6)) {
441 return (&pi->ipi6_addr);
446 * If the source address is not specified but the socket(if any)
447 * is already bound, use the bound address.
449 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) {
450 jsin6.sin6_addr = *laddr;
451 if (jailed && !jailed_ip(cred->cr_prison,
452 (struct sockaddr *)&jsin6)) {
460 * If the caller doesn't specify the source address but
461 * the outgoing interface, use an address associated with
464 if (pi && pi->ipi6_ifindex) {
465 /* XXX boundary check is assumed to be already done. */
466 ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex],
470 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
471 if (!jailed_ip(cred->cr_prison,
472 (struct sockaddr *)&jsin6))
477 *errorp = EADDRNOTAVAIL;
480 return (&satosin6(&ia6->ia_addr)->sin6_addr);
484 * If the destination address is a link-local unicast address or
485 * a multicast address, and if the outgoing interface is specified
486 * by the sin6_scope_id filed, use an address associated with the
488 * XXX: We're now trying to define more specific semantics of
489 * sin6_scope_id field, so this part will be rewritten in
492 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) &&
493 dstsock->sin6_scope_id) {
495 * I'm not sure if boundary check for scope_id is done
498 if (dstsock->sin6_scope_id < 0 ||
499 if_index < dstsock->sin6_scope_id) {
500 *errorp = ENXIO; /* XXX: better error? */
503 ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id],
507 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
508 if (!jailed_ip(cred->cr_prison,
509 (struct sockaddr *)&jsin6))
514 *errorp = EADDRNOTAVAIL;
517 return (&satosin6(&ia6->ia_addr)->sin6_addr);
521 * If the destination address is a multicast address and
522 * the outgoing interface for the address is specified
523 * by the caller, use an address associated with the interface.
524 * There is a sanity check here; if the destination has node-local
525 * scope, the outgoing interfacde should be a loopback address.
526 * Even if the outgoing interface is not specified, we also
527 * choose a loopback interface as the outgoing interface.
529 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) {
530 struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL;
532 if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) {
537 ia6 = in6_ifawithscope(ifp, dst, cred);
539 *errorp = EADDRNOTAVAIL;
542 return (&ia6->ia_addr.sin6_addr);
547 * If the next hop address for the packet is specified
548 * by caller, use an address associated with the route
552 struct sockaddr_in6 *sin6_next;
555 if (opts && opts->ip6po_nexthop) {
556 sin6_next = satosin6(opts->ip6po_nexthop);
557 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL);
559 ia6 = in6_ifawithscope(rt->rt_ifp, dst, cred);
561 ia6 = ifatoia6(rt->rt_ifa);
564 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
565 if (!jailed_ip(cred->cr_prison,
566 (struct sockaddr *)&jsin6))
571 *errorp = EADDRNOTAVAIL;
574 return (&satosin6(&ia6->ia_addr)->sin6_addr);
579 * If route is known or can be allocated now,
580 * our src addr is taken from the i/f, else punt.
584 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, dst)) {
588 if (ro->ro_rt == NULL || ro->ro_rt->rt_ifp == NULL) {
589 struct sockaddr_in6 *dst6;
591 /* No route yet, so try to acquire one */
592 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
594 dst6->sin6_family = AF_INET6;
595 dst6->sin6_len = sizeof(struct sockaddr_in6);
596 dst6->sin6_addr = *dst;
597 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) {
599 rtpurelookup((struct sockaddr *)&ro->ro_dst);
601 rtalloc((struct route *)ro);
606 * in_pcbconnect() checks out IFF_LOOPBACK to skip using
607 * the address. But we don't know why it does so.
608 * It is necessary to ensure the scope even for lo0
609 * so doesn't check out IFF_LOOPBACK.
613 ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst, cred);
615 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
616 if (!jailed_ip(cred->cr_prison,
617 (struct sockaddr *)&jsin6))
621 if (ia6 == 0) /* xxx scope error ?*/
622 ia6 = ifatoia6(ro->ro_rt->rt_ifa);
625 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
626 if (!jailed_ip(cred->cr_prison,
627 (struct sockaddr *)&jsin6))
632 *errorp = EHOSTUNREACH; /* no route */
635 return (&satosin6(&ia6->ia_addr)->sin6_addr);
638 *errorp = EADDRNOTAVAIL;
643 * Default hop limit selection. The precedence is as follows:
644 * 1. Hoplimit valued specified via ioctl.
645 * 2. (If the outgoing interface is detected) the current
646 * hop limit of the interface specified by router advertisement.
647 * 3. The system default hoplimit.
650 in6_selecthlim(struct in6pcb *in6p, struct ifnet *ifp)
652 if (in6p && in6p->in6p_hops >= 0)
653 return (in6p->in6p_hops);
655 return (ND_IFINFO(ifp)->chlim);
657 return (ip6_defhlim);
662 in6_pcbdisconnect(struct inpcb *inp)
664 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
666 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
667 inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
668 in_pcbremconnhash(inp);
669 if (inp->inp_socket->so_state & SS_NOFDREF)
674 in6_pcbdetach(struct inpcb *inp)
676 struct socket *so = inp->inp_socket;
677 struct inpcbinfo *ipi = inp->inp_pcbinfo;
680 if (inp->in6p_sp != NULL)
681 ipsec6_delete_pcbpolicy(inp);
683 inp->inp_gencnt = ++ipi->ipi_gencnt;
686 KKASSERT((so->so_state & SS_ASSERTINPROG) == 0);
687 sofree(so); /* remove pcb ref */
689 if (inp->in6p_options)
690 m_freem(inp->in6p_options);
691 ip6_freepcbopts(inp->in6p_outputopts);
692 ip6_freemoptions(inp->in6p_moptions);
693 if (inp->in6p_route.ro_rt)
694 rtfree(inp->in6p_route.ro_rt);
695 /* Check and free IPv4 related resources in case of mapped addr */
696 if (inp->inp_options)
697 m_free(inp->inp_options);
698 ip_freemoptions(inp->inp_moptions);
704 * The calling convention of in6_setsockaddr() and in6_setpeeraddr() was
705 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
706 * in struct pr_usrreqs, so that protocols can just reference then directly
707 * without the need for a wrapper function. The socket must have a valid
708 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
709 * except through a kernel programming error, so it is acceptable to panic
710 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap
711 * because there actually /is/ a programming error somewhere... XXX)
714 in6_setsockaddr_dispatch(netmsg_t msg)
718 error = in6_setsockaddr(msg->sockaddr.base.nm_so, msg->sockaddr.nm_nam);
719 lwkt_replymsg(&msg->sockaddr.base.lmsg, error);
723 in6_setsockaddr(struct socket *so, struct sockaddr **nam)
726 struct sockaddr_in6 *sin6;
729 * Do the malloc first in case it blocks.
731 sin6 = kmalloc(sizeof *sin6, M_SONAME, M_WAITOK | M_ZERO);
732 sin6->sin6_family = AF_INET6;
733 sin6->sin6_len = sizeof(*sin6);
739 kfree(sin6, M_SONAME);
742 sin6->sin6_port = inp->inp_lport;
743 sin6->sin6_addr = inp->in6p_laddr;
745 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
746 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
748 sin6->sin6_scope_id = 0; /*XXX*/
749 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
750 sin6->sin6_addr.s6_addr16[1] = 0;
752 *nam = (struct sockaddr *)sin6;
757 in6_setpeeraddr_dispatch(netmsg_t msg)
761 error = in6_setpeeraddr(msg->peeraddr.base.nm_so, msg->peeraddr.nm_nam);
762 lwkt_replymsg(&msg->peeraddr.base.lmsg, error);
766 in6_setpeeraddr(struct socket *so, struct sockaddr **nam)
769 struct sockaddr_in6 *sin6;
772 * Do the malloc first in case it blocks.
774 sin6 = kmalloc(sizeof(*sin6), M_SONAME, M_WAITOK | M_ZERO);
775 sin6->sin6_family = AF_INET6;
776 sin6->sin6_len = sizeof(struct sockaddr_in6);
782 kfree(sin6, M_SONAME);
785 sin6->sin6_port = inp->inp_fport;
786 sin6->sin6_addr = inp->in6p_faddr;
788 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
789 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
791 sin6->sin6_scope_id = 0; /*XXX*/
792 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
793 sin6->sin6_addr.s6_addr16[1] = 0;
795 *nam = (struct sockaddr *)sin6;
800 * Pass some notification to all connections of a protocol
801 * associated with address dst. The local address and/or port numbers
802 * may be specified to limit the search. The "usual action" will be
803 * taken, depending on the ctlinput cmd. The caller must filter any
804 * cmds that are uninteresting (e.g., no error in the map).
805 * Call the protocol specific routine (if any) to report
806 * any errors for each matching socket.
809 in6_pcbnotify(struct inpcbinfo *pcbinfo, struct sockaddr *dst, in_port_t fport,
810 const struct sockaddr *src, in_port_t lport, int cmd, int arg,
813 struct inpcb *inp, *marker;
814 struct sockaddr_in6 sa6_src, *sa6_dst;
817 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
820 sa6_dst = (struct sockaddr_in6 *)dst;
821 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
825 * note that src can be NULL when we get notify by local fragmentation.
827 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
828 flowinfo = sa6_src.sin6_flowinfo;
831 * Redirects go to all references to the destination,
832 * and use in6_rtchange to invalidate the route cache.
833 * Dead host indications: also use in6_rtchange to invalidate
834 * the cache, and deliver the error to all the sockets.
835 * Otherwise, if we have knowledge of the local port and address,
836 * deliver only to that socket.
838 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
841 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
843 if (cmd != PRC_HOSTDEAD)
844 notify = in6_rtchange;
846 if (cmd != PRC_MSGSIZE)
847 arg = inet6ctlerrmap[cmd];
849 marker = in_pcbmarker(mycpuid);
851 GET_PCBINFO_TOKEN(pcbinfo);
853 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, marker, inp_list);
854 while ((inp = LIST_NEXT(marker, inp_list)) != NULL) {
855 LIST_REMOVE(marker, inp_list);
856 LIST_INSERT_AFTER(inp, marker, inp_list);
858 if (inp->inp_flags & INP_PLACEMARKER)
861 if (!INP_ISIPV6(inp))
864 * If the error designates a new path MTU for a destination
865 * and the application (associated with this socket) wanted to
866 * know the value, notify. Note that we notify for all
867 * disconnected sockets if the corresponding application
868 * wanted. This is because some UDP applications keep sending
869 * sockets disconnected.
870 * XXX: should we avoid to notify the value to TCP sockets?
872 if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 &&
873 (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
874 IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) {
875 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst, &arg);
879 * Detect if we should notify the error. If no source and
880 * destination ports are specifed, but non-zero flowinfo and
881 * local address match, notify the error. This is the case
882 * when the error is delivered with an encrypted buffer
883 * by ESP. Otherwise, just compare addresses and ports
886 if (lport == 0 && fport == 0 && flowinfo &&
887 inp->inp_socket != NULL &&
888 flowinfo == (inp->in6p_flowinfo & IPV6_FLOWLABEL_MASK) &&
889 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
891 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
892 &sa6_dst->sin6_addr) ||
893 inp->inp_socket == 0 ||
894 (lport && inp->inp_lport != lport) ||
895 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
896 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
897 &sa6_src.sin6_addr)) ||
898 (fport && inp->inp_fport != fport))
905 LIST_REMOVE(marker, inp_list);
907 REL_PCBINFO_TOKEN(pcbinfo);
911 * Lookup a PCB based on the local address and port.
914 in6_pcblookup_local(struct inpcbportinfo *portinfo,
915 const struct in6_addr *laddr, u_int lport_arg, int wild_okay,
919 int matchwild = 3, wildcard;
920 u_short lport = lport_arg;
921 struct inpcbporthead *porthash;
922 struct inpcbport *phd;
923 struct inpcb *match = NULL;
926 * If the porthashbase is shared across several cpus, it must
929 ASSERT_PORT_TOKEN_HELD(portinfo);
932 * Best fit PCB lookup.
934 * First see if this local port is in use by looking on the
937 porthash = &portinfo->porthashbase[
938 INP_PCBPORTHASH(lport, portinfo->porthashmask)];
939 LIST_FOREACH(phd, porthash, phd_hash) {
940 if (phd->phd_port == lport)
946 * Port is in use by one or more PCBs. Look for best
949 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
951 if (!INP_ISIPV6(inp))
953 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
955 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
956 if (IN6_IS_ADDR_UNSPECIFIED(laddr))
958 else if (!IN6_ARE_ADDR_EQUAL(
959 &inp->in6p_laddr, laddr))
962 if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
965 if (wildcard && !wild_okay)
967 if (wildcard < matchwild &&
970 inp->inp_socket->so_cred->cr_prison)) {
972 matchwild = wildcard;
976 matchwild = wildcard;
984 in6_pcbpurgeif0(struct inpcbinfo *pcbinfo, struct ifnet *ifp)
986 struct in6pcb *in6p, *marker;
987 struct ip6_moptions *im6o;
988 struct in6_multi_mship *imm, *nimm;
991 * We only need to make sure that we are in netisr0, where all
992 * multicast operation happen. We could check inpcbinfo which
993 * does not belong to netisr0 by holding the inpcbinfo's token.
994 * In this case, the pcbinfo must be able to be shared, i.e.
995 * pcbinfo->infotoken is not NULL.
997 KASSERT(&curthread->td_msgport == netisr_cpuport(0),
999 KASSERT(pcbinfo->cpu == 0 || pcbinfo->infotoken != NULL,
1000 ("pcbinfo could not be shared"));
1003 * Get a marker for the current netisr (netisr0).
1005 * It is possible that the multicast address deletion blocks,
1006 * which could cause temporary token releasing. So we use
1007 * inpcb marker here to get a coherent view of the inpcb list.
1009 * While, on the other hand, moptions are only added and deleted
1010 * in netisr0, so we would not see staled moption or miss moption
1011 * even if the token was released due to the blocking multicast
1014 marker = in_pcbmarker(mycpuid);
1016 GET_PCBINFO_TOKEN(pcbinfo);
1018 LIST_INSERT_HEAD(&pcbinfo->pcblisthead, marker, inp_list);
1019 while ((in6p = LIST_NEXT(marker, inp_list)) != NULL) {
1020 LIST_REMOVE(marker, inp_list);
1021 LIST_INSERT_AFTER(in6p, marker, inp_list);
1023 if (in6p->in6p_flags & INP_PLACEMARKER)
1025 im6o = in6p->in6p_moptions;
1026 if (INP_ISIPV6(in6p) && im6o) {
1028 * Unselect the outgoing interface if it is being
1031 if (im6o->im6o_multicast_ifp == ifp)
1032 im6o->im6o_multicast_ifp = NULL;
1035 * Drop multicast group membership if we joined
1036 * through the interface being detached.
1037 * XXX controversial - is it really legal for kernel
1040 for (imm = im6o->im6o_memberships.lh_first;
1041 imm != NULL; imm = nimm) {
1042 nimm = imm->i6mm_chain.le_next;
1043 if (imm->i6mm_maddr->in6m_ifp == ifp) {
1044 LIST_REMOVE(imm, i6mm_chain);
1045 in6_delmulti(imm->i6mm_maddr);
1046 kfree(imm, M_IPMADDR);
1051 LIST_REMOVE(marker, inp_list);
1053 REL_PCBINFO_TOKEN(pcbinfo);
1057 * Check for alternatives when higher level complains
1058 * about service problems. For now, invalidate cached
1059 * routing information. If the route was created dynamically
1060 * (by a redirect), time to try a default gateway again.
1063 in6_losing(struct inpcb *in6p)
1066 struct rt_addrinfo info;
1068 if ((rt = in6p->in6p_route.ro_rt) != NULL) {
1069 bzero((caddr_t)&info, sizeof(info));
1070 info.rti_flags = rt->rt_flags;
1071 info.rti_info[RTAX_DST] = rt_key(rt);
1072 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1073 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1074 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
1075 if (rt->rt_flags & RTF_DYNAMIC) {
1076 rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway,
1077 rt_mask(rt), rt->rt_flags, NULL);
1079 in6p->in6p_route.ro_rt = NULL;
1082 * A new route can be allocated
1083 * the next time output is attempted.
1089 * After a routing change, flush old routing
1090 * and allocate a (hopefully) better one.
1093 in6_rtchange(struct inpcb *inp, int error)
1095 if (inp->in6p_route.ro_rt) {
1096 rtfree(inp->in6p_route.ro_rt);
1097 inp->in6p_route.ro_rt = 0;
1099 * A new route can be allocated the next time
1100 * output is attempted.
1106 * Lookup PCB in hash list.
1109 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1110 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg,
1111 int wildcard, struct ifnet *ifp)
1113 struct inpcbhead *head;
1115 struct inpcb *jinp = NULL;
1116 u_short fport = fport_arg, lport = lport_arg;
1119 if (faithprefix_p != NULL)
1120 faith = (*faithprefix_p)(laddr);
1125 * First look for an exact match.
1127 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr->s6_addr32[3] /* XXX */,
1129 laddr->s6_addr32[3], /* XXX JH */
1131 pcbinfo->hashmask)];
1132 LIST_FOREACH(inp, head, inp_hash) {
1133 if (!INP_ISIPV6(inp))
1135 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1136 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1137 inp->inp_fport == fport &&
1138 inp->inp_lport == lport) {
1142 if (inp->inp_socket == NULL ||
1143 inp->inp_socket->so_cred->cr_prison == NULL) {
1155 struct inpcontainerhead *chead;
1156 struct inpcontainer *ic;
1157 struct inpcb *local_wild = NULL;
1158 struct inpcb *jinp_wild = NULL;
1159 struct sockaddr_in6 jsin6;
1163 * Order of socket selection:
1164 * 1. non-jailed, non-wild.
1165 * 2. non-jailed, wild.
1166 * 3. jailed, non-wild.
1169 jsin6.sin6_family = AF_INET6;
1170 chead = &pcbinfo->wildcardhashbase[INP_PCBWILDCARDHASH(lport,
1171 pcbinfo->wildcardhashmask)];
1173 GET_PCBINFO_TOKEN(pcbinfo);
1174 LIST_FOREACH(ic, chead, ic_list) {
1176 if (inp->inp_flags & INP_PLACEMARKER)
1179 if (!INP_ISIPV6(inp))
1181 if (inp->inp_socket != NULL)
1182 cred = inp->inp_socket->so_cred;
1186 if (cred != NULL && jailed(cred)) {
1190 jsin6.sin6_addr = *laddr;
1191 if (!jailed_ip(cred->cr_prison,
1192 (struct sockaddr *)&jsin6))
1196 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1197 inp->inp_lport == lport) {
1198 if (faith && (inp->inp_flags & INP_FAITH) == 0)
1200 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
1202 if (cred != NULL && jailed(cred)) {
1205 REL_PCBINFO_TOKEN(pcbinfo);
1208 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1209 if (cred != NULL && jailed(cred))
1216 REL_PCBINFO_TOKEN(pcbinfo);
1218 if (local_wild != NULL)
1219 return (local_wild);
1232 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
1236 ip = mtod(m, struct ip6_hdr *);
1237 bzero(sin6, sizeof(*sin6));
1238 sin6->sin6_len = sizeof(*sin6);
1239 sin6->sin6_family = AF_INET6;
1240 sin6->sin6_addr = ip->ip6_src;
1241 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1242 sin6->sin6_addr.s6_addr16[1] = 0;
1243 sin6->sin6_scope_id =
1244 (m->m_pkthdr.rcvif && IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1245 ? m->m_pkthdr.rcvif->if_index : 0;
1251 in6_savefaddr(struct socket *so, const struct sockaddr *faddr)
1253 struct sockaddr_in6 *sin6;
1255 KASSERT(faddr->sa_family == AF_INET6,
1256 ("not AF_INET6 faddr %d", faddr->sa_family));
1258 sin6 = kmalloc(sizeof(*sin6), M_SONAME, M_WAITOK | M_ZERO);
1259 sin6->sin6_family = AF_INET6;
1260 sin6->sin6_len = sizeof(*sin6);
1262 sin6->sin6_port = ((const struct sockaddr_in6 *)faddr)->sin6_port;
1263 sin6->sin6_addr = ((const struct sockaddr_in6 *)faddr)->sin6_addr;
1265 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1266 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
1268 sin6->sin6_scope_id = 0; /*XXX*/
1269 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1270 sin6->sin6_addr.s6_addr16[1] = 0;
1272 so->so_faddr = (struct sockaddr *)sin6;