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>
93 #include <netinet/in.h>
94 #include <netinet/in_var.h>
95 #include <netinet/in_systm.h>
96 #include <netinet/ip6.h>
97 #include <netinet/ip_var.h>
98 #include <netinet6/ip6_var.h>
99 #include <netinet6/nd6.h>
100 #include <netinet/in_pcb.h>
101 #include <netinet6/in6_pcb.h>
104 #include <netinet6/ipsec.h>
106 #include <netinet6/ipsec6.h>
108 #include <netinet6/ah.h>
110 #include <netinet6/ah6.h>
112 #include <netproto/key/key.h>
116 #include <netproto/ipsec/ipsec.h>
117 #include <netproto/ipsec/ipsec6.h>
118 #include <netproto/ipsec/key.h>
120 #endif /* FAST_IPSEC */
122 struct in6_addr zeroin6_addr;
125 in6_pcbbind(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
127 struct socket *so = inp->inp_socket;
128 struct sockaddr_in6 jsin6;
131 if (!in6_ifaddr) /* XXX broken! */
132 return (EADDRNOTAVAIL);
133 if (inp->inp_lport || !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
137 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
138 struct inpcbinfo *pcbinfo;
139 struct inpcbportinfo *portinfo;
140 int wild = 0, reuseport = (so->so_options & SO_REUSEPORT);
141 struct ucred *cred = NULL;
143 u_short lport, lport_ho;
145 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
147 if (td->td_proc != NULL)
148 cred = td->td_proc->p_ucred;
150 if (nam->sa_len != sizeof(*sin6))
155 if (nam->sa_family != AF_INET6)
156 return (EAFNOSUPPORT);
158 if (!prison_replace_wildcards(td, nam))
161 /* KAME hack: embed scopeid */
162 if (in6_embedscope(&sin6->sin6_addr, sin6, inp, NULL) != 0)
164 /* this must be cleared for ifa_ifwithaddr() */
165 sin6->sin6_scope_id = 0;
167 lport = sin6->sin6_port;
168 if (IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
170 * Treat SO_REUSEADDR as SO_REUSEPORT for multicast;
171 * allow compepte duplication of binding if
172 * SO_REUSEPORT is set, or if SO_REUSEADDR is set
173 * and a multicast address is bound on both
174 * new and duplicated sockets.
176 if (so->so_options & SO_REUSEADDR)
177 reuseport = SO_REUSEADDR|SO_REUSEPORT;
178 } else if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
179 struct ifaddr *ia = NULL;
181 sin6->sin6_port = 0; /* yech... */
182 if (!prison_replace_wildcards(td, (struct sockaddr *)sin6)) {
183 sin6->sin6_addr = kin6addr_any;
186 if ((ia = ifa_ifwithaddr((struct sockaddr *)sin6)) == NULL)
187 return (EADDRNOTAVAIL);
190 * XXX: bind to an anycast address might accidentally
191 * cause sending a packet with anycast source address.
192 * We should allow to bind to a deprecated address, since
193 * the application dare to use it.
196 ((struct in6_ifaddr *)ia)->ia6_flags &
197 (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|IN6_IFF_DETACHED))
198 return (EADDRNOTAVAIL);
201 inp->in6p_laddr = sin6->sin6_addr;
205 lport_ho = ntohs(lport);
208 if (lport_ho < IPV6PORT_RESERVED && cred &&
209 priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 0)) {
210 inp->in6p_laddr = kin6addr_any;
215 * Locate the proper portinfo based on lport
217 pcbinfo = inp->inp_pcbinfo;
219 &pcbinfo->portinfo[lport_ho & pcbinfo->portinfo_mask];
220 KKASSERT((lport_ho & pcbinfo->portinfo_mask) ==
224 * This has to be atomic. If the porthash is shared across
225 * multiple protocol threads (aka tcp) then the token must
228 GET_PORT_TOKEN(portinfo);
230 if (so->so_cred->cr_uid != 0 &&
231 !IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) {
232 t = in6_pcblookup_local(portinfo,
233 &sin6->sin6_addr, lport, INPLOOKUP_WILDCARD, cred);
235 (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) ||
236 !IN6_IS_ADDR_UNSPECIFIED(&t->in6p_laddr) ||
237 (t->inp_socket->so_options & SO_REUSEPORT) == 0) &&
238 (so->so_cred->cr_uid !=
239 t->inp_socket->so_cred->cr_uid)) {
240 inp->in6p_laddr = kin6addr_any;
244 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
245 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
246 struct sockaddr_in sin;
248 in6_sin6_2_sin(&sin, sin6);
249 t = in_pcblookup_local(portinfo,
251 INPLOOKUP_WILDCARD, cred);
253 (so->so_cred->cr_uid !=
254 t->inp_socket->so_cred->cr_uid) &&
255 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
257 INP_SOCKAF(t->inp_socket))) {
258 inp->in6p_laddr = kin6addr_any;
264 if (cred && cred->cr_prison &&
265 !prison_replace_wildcards(td, nam)) {
266 inp->in6p_laddr = kin6addr_any;
267 error = EADDRNOTAVAIL;
270 t = in6_pcblookup_local(portinfo, &sin6->sin6_addr, lport,
272 if (t && (reuseport & t->inp_socket->so_options) == 0) {
273 inp->in6p_laddr = kin6addr_any;
277 if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0 &&
278 IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
279 struct sockaddr_in sin;
281 in6_sin6_2_sin(&sin, sin6);
282 t = in_pcblookup_local(portinfo, sin.sin_addr, lport,
284 if (t && (reuseport & t->inp_socket->so_options) == 0 &&
285 (ntohl(t->inp_laddr.s_addr) != INADDR_ANY ||
286 INP_SOCKAF(so) == INP_SOCKAF(t->inp_socket))) {
287 inp->in6p_laddr = kin6addr_any;
293 inp->inp_lport = lport;
294 in_pcbinsporthash(portinfo, inp);
297 REL_PORT_TOKEN(portinfo);
301 jsin6.sin6_addr = inp->in6p_laddr;
302 jsin6.sin6_family = AF_INET6;
303 if (!prison_replace_wildcards(td, (struct sockaddr*)&jsin6)) {
304 inp->in6p_laddr = kin6addr_any;
309 return in6_pcbsetlport(&inp->in6p_laddr, inp, td);
314 * Transform old in6_pcbconnect() into an inner subroutine for new
315 * in6_pcbconnect(): Do some validity-checking on the remote
316 * address (in mbuf 'nam') and then determine local host address
317 * (i.e., which interface) to use to access that remote host.
319 * This preserves definition of in6_pcbconnect(), while supporting a
320 * slightly different version for T/TCP. (This is more than
321 * a bit of a kludge, but cleaning up the internal interfaces would
322 * have forced minor changes in every protocol).
326 in6_pcbladdr(struct inpcb *inp, struct sockaddr *nam,
327 struct in6_addr **plocal_addr6, struct thread *td)
329 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
330 struct ifnet *ifp = NULL;
333 if (nam->sa_len != sizeof (*sin6))
335 if (sin6->sin6_family != AF_INET6)
336 return (EAFNOSUPPORT);
337 if (sin6->sin6_port == 0)
338 return (EADDRNOTAVAIL);
340 /* KAME hack: embed scopeid */
341 if (in6_embedscope(&sin6->sin6_addr, sin6, inp, &ifp) != 0)
346 * If the destination address is UNSPECIFIED addr,
347 * use the loopback addr, e.g ::1.
349 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
350 sin6->sin6_addr = kin6addr_loopback;
354 * XXX: in6_selectsrc might replace the bound local address
355 * with the address specified by setsockopt(IPV6_PKTINFO).
356 * Is it the intended behavior?
358 *plocal_addr6 = in6_selectsrc(sin6, inp->in6p_outputopts,
361 &inp->in6p_laddr, &error, td);
362 if (*plocal_addr6 == NULL) {
364 error = EADDRNOTAVAIL;
368 * Don't do pcblookup call here; return interface in
370 * and exit to caller, that will do the lookup.
374 if (inp->in6p_route.ro_rt)
375 ifp = inp->in6p_route.ro_rt->rt_ifp;
382 * Connect from a socket to a specified address.
383 * Both address and port must be specified in argument sin.
384 * If don't have a local address for this socket yet,
388 in6_pcbconnect(struct inpcb *inp, struct sockaddr *nam, struct thread *td)
390 struct in6_addr *addr6;
391 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
395 * Call inner routine, to assign local interface address.
396 * in6_pcbladdr() may automatically fill in sin6_scope_id.
398 if ((error = in6_pcbladdr(inp, nam, &addr6, td)) != 0)
401 if (in6_pcblookup_hash(inp->inp_pcbinfo, &sin6->sin6_addr,
403 IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
404 ? addr6 : &inp->in6p_laddr,
405 inp->inp_lport, 0, NULL) != NULL) {
408 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
409 if (inp->inp_lport == 0) {
410 error = in6_pcbbind(inp, NULL, td);
414 inp->in6p_laddr = *addr6;
416 inp->in6p_faddr = sin6->sin6_addr;
417 inp->inp_fport = sin6->sin6_port;
418 /* update flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
419 inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
420 if (inp->in6p_flags & IN6P_AUTOFLOWLABEL)
421 inp->in6p_flowinfo |=
422 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK);
424 in_pcbinsconnhash(inp);
430 * Return an IPv6 address, which is the most appropriate for given
431 * destination and user specified options.
432 * If necessary, this function lookups the routing table and return
433 * an entry to the caller for later use.
436 in6_selectsrc(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
437 struct ip6_moptions *mopts, struct route_in6 *ro,
438 struct in6_addr *laddr, int *errorp, struct thread *td)
440 struct sockaddr_in6 jsin6;
441 struct ucred *cred = NULL;
442 struct in6_addr *dst;
443 struct in6_ifaddr *ia6 = 0;
444 struct in6_pktinfo *pi = NULL;
447 if (td && td->td_proc && td->td_proc->p_ucred)
448 cred = td->td_proc->p_ucred;
449 if (cred && cred->cr_prison)
451 jsin6.sin6_family = AF_INET6;
452 dst = &dstsock->sin6_addr;
456 * If the source address is explicitly specified by the caller,
459 if (opts && (pi = opts->ip6po_pktinfo) &&
460 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) {
461 jsin6.sin6_addr = pi->ipi6_addr;
462 if (jailed && !jailed_ip(cred->cr_prison,
463 (struct sockaddr *)&jsin6)) {
466 return (&pi->ipi6_addr);
471 * If the source address is not specified but the socket(if any)
472 * is already bound, use the bound address.
474 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) {
475 jsin6.sin6_addr = *laddr;
476 if (jailed && !jailed_ip(cred->cr_prison,
477 (struct sockaddr *)&jsin6)) {
485 * If the caller doesn't specify the source address but
486 * the outgoing interface, use an address associated with
489 if (pi && pi->ipi6_ifindex) {
490 /* XXX boundary check is assumed to be already done. */
491 ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex],
495 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
496 if (!jailed_ip(cred->cr_prison,
497 (struct sockaddr *)&jsin6))
502 *errorp = EADDRNOTAVAIL;
505 return (&satosin6(&ia6->ia_addr)->sin6_addr);
509 * If the destination address is a link-local unicast address or
510 * a multicast address, and if the outgoing interface is specified
511 * by the sin6_scope_id filed, use an address associated with the
513 * XXX: We're now trying to define more specific semantics of
514 * sin6_scope_id field, so this part will be rewritten in
517 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) &&
518 dstsock->sin6_scope_id) {
520 * I'm not sure if boundary check for scope_id is done
523 if (dstsock->sin6_scope_id < 0 ||
524 if_index < dstsock->sin6_scope_id) {
525 *errorp = ENXIO; /* XXX: better error? */
528 ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id],
532 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
533 if (!jailed_ip(cred->cr_prison,
534 (struct sockaddr *)&jsin6))
539 *errorp = EADDRNOTAVAIL;
542 return (&satosin6(&ia6->ia_addr)->sin6_addr);
546 * If the destination address is a multicast address and
547 * the outgoing interface for the address is specified
548 * by the caller, use an address associated with the interface.
549 * There is a sanity check here; if the destination has node-local
550 * scope, the outgoing interfacde should be a loopback address.
551 * Even if the outgoing interface is not specified, we also
552 * choose a loopback interface as the outgoing interface.
554 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) {
555 struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL;
557 if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) {
562 ia6 = in6_ifawithscope(ifp, dst, cred);
564 *errorp = EADDRNOTAVAIL;
567 return (&ia6->ia_addr.sin6_addr);
572 * If the next hop address for the packet is specified
573 * by caller, use an address associated with the route
577 struct sockaddr_in6 *sin6_next;
580 if (opts && opts->ip6po_nexthop) {
581 sin6_next = satosin6(opts->ip6po_nexthop);
582 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL);
584 ia6 = in6_ifawithscope(rt->rt_ifp, dst, cred);
586 ia6 = ifatoia6(rt->rt_ifa);
589 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
590 if (!jailed_ip(cred->cr_prison,
591 (struct sockaddr *)&jsin6))
596 *errorp = EADDRNOTAVAIL;
599 return (&satosin6(&ia6->ia_addr)->sin6_addr);
604 * If route is known or can be allocated now,
605 * our src addr is taken from the i/f, else punt.
609 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr, dst)) {
613 if (ro->ro_rt == NULL || ro->ro_rt->rt_ifp == NULL) {
614 struct sockaddr_in6 *dst6;
616 /* No route yet, so try to acquire one */
617 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
619 dst6->sin6_family = AF_INET6;
620 dst6->sin6_len = sizeof(struct sockaddr_in6);
621 dst6->sin6_addr = *dst;
622 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) {
624 rtpurelookup((struct sockaddr *)&ro->ro_dst);
626 rtalloc((struct route *)ro);
631 * in_pcbconnect() checks out IFF_LOOPBACK to skip using
632 * the address. But we don't know why it does so.
633 * It is necessary to ensure the scope even for lo0
634 * so doesn't check out IFF_LOOPBACK.
638 ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst, cred);
640 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
641 if (!jailed_ip(cred->cr_prison,
642 (struct sockaddr *)&jsin6))
646 if (ia6 == 0) /* xxx scope error ?*/
647 ia6 = ifatoia6(ro->ro_rt->rt_ifa);
650 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
651 if (!jailed_ip(cred->cr_prison,
652 (struct sockaddr *)&jsin6))
657 *errorp = EHOSTUNREACH; /* no route */
660 return (&satosin6(&ia6->ia_addr)->sin6_addr);
663 *errorp = EADDRNOTAVAIL;
668 * Default hop limit selection. The precedence is as follows:
669 * 1. Hoplimit valued specified via ioctl.
670 * 2. (If the outgoing interface is detected) the current
671 * hop limit of the interface specified by router advertisement.
672 * 3. The system default hoplimit.
675 in6_selecthlim(struct in6pcb *in6p, struct ifnet *ifp)
677 if (in6p && in6p->in6p_hops >= 0)
678 return (in6p->in6p_hops);
680 return (ND_IFINFO(ifp)->chlim);
682 return (ip6_defhlim);
687 in6_pcbdisconnect(struct inpcb *inp)
689 bzero((caddr_t)&inp->in6p_faddr, sizeof(inp->in6p_faddr));
691 /* clear flowinfo - draft-itojun-ipv6-flowlabel-api-00 */
692 inp->in6p_flowinfo &= ~IPV6_FLOWLABEL_MASK;
693 in_pcbremconnhash(inp);
694 if (inp->inp_socket->so_state & SS_NOFDREF)
699 in6_pcbdetach(struct inpcb *inp)
701 struct socket *so = inp->inp_socket;
702 struct inpcbinfo *ipi = inp->inp_pcbinfo;
705 if (inp->in6p_sp != NULL)
706 ipsec6_delete_pcbpolicy(inp);
708 inp->inp_gencnt = ++ipi->ipi_gencnt;
711 KKASSERT((so->so_state & SS_ASSERTINPROG) == 0);
712 sofree(so); /* remove pcb ref */
714 if (inp->in6p_options)
715 m_freem(inp->in6p_options);
716 ip6_freepcbopts(inp->in6p_outputopts);
717 ip6_freemoptions(inp->in6p_moptions);
718 if (inp->in6p_route.ro_rt)
719 rtfree(inp->in6p_route.ro_rt);
720 /* Check and free IPv4 related resources in case of mapped addr */
721 if (inp->inp_options)
722 m_free(inp->inp_options);
723 ip_freemoptions(inp->inp_moptions);
730 * The calling convention of in6_setsockaddr() and in6_setpeeraddr() was
731 * modified to match the pru_sockaddr() and pru_peeraddr() entry points
732 * in struct pr_usrreqs, so that protocols can just reference then directly
733 * without the need for a wrapper function. The socket must have a valid
734 * (i.e., non-nil) PCB, but it should be impossible to get an invalid one
735 * except through a kernel programming error, so it is acceptable to panic
736 * (or in this case trap) if the PCB is invalid. (Actually, we don't trap
737 * because there actually /is/ a programming error somewhere... XXX)
740 in6_setsockaddr_dispatch(netmsg_t msg)
744 error = in6_setsockaddr(msg->sockaddr.base.nm_so, msg->sockaddr.nm_nam);
745 lwkt_replymsg(&msg->sockaddr.base.lmsg, error);
749 in6_setsockaddr(struct socket *so, struct sockaddr **nam)
752 struct sockaddr_in6 *sin6;
755 * Do the malloc first in case it blocks.
757 sin6 = kmalloc(sizeof *sin6, M_SONAME, M_WAITOK | M_ZERO);
758 sin6->sin6_family = AF_INET6;
759 sin6->sin6_len = sizeof(*sin6);
765 kfree(sin6, M_SONAME);
768 sin6->sin6_port = inp->inp_lport;
769 sin6->sin6_addr = inp->in6p_laddr;
771 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
772 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
774 sin6->sin6_scope_id = 0; /*XXX*/
775 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
776 sin6->sin6_addr.s6_addr16[1] = 0;
778 *nam = (struct sockaddr *)sin6;
783 in6_setpeeraddr_dispatch(netmsg_t msg)
787 error = in6_setpeeraddr(msg->peeraddr.base.nm_so, msg->peeraddr.nm_nam);
788 lwkt_replymsg(&msg->peeraddr.base.lmsg, error);
792 in6_setpeeraddr(struct socket *so, struct sockaddr **nam)
795 struct sockaddr_in6 *sin6;
798 * Do the malloc first in case it blocks.
800 sin6 = kmalloc(sizeof(*sin6), M_SONAME, M_WAITOK | M_ZERO);
801 sin6->sin6_family = AF_INET6;
802 sin6->sin6_len = sizeof(struct sockaddr_in6);
808 kfree(sin6, M_SONAME);
811 sin6->sin6_port = inp->inp_fport;
812 sin6->sin6_addr = inp->in6p_faddr;
814 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
815 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
817 sin6->sin6_scope_id = 0; /*XXX*/
818 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
819 sin6->sin6_addr.s6_addr16[1] = 0;
821 *nam = (struct sockaddr *)sin6;
826 in6_mapped_sockaddr_dispatch(netmsg_t msg)
830 error = in6_mapped_sockaddr(msg->sockaddr.base.nm_so,
831 msg->sockaddr.nm_nam);
832 lwkt_replymsg(&msg->sockaddr.base.lmsg, error);
836 in6_mapped_sockaddr(struct socket *so, struct sockaddr **nam)
838 struct inpcb *inp = so->so_pcb;
843 if (inp->inp_vflag & INP_IPV4) {
844 error = in_setsockaddr(so, nam);
846 in6_sin_2_v4mapsin6_in_sock(nam);
848 /* scope issues will be handled in in6_setsockaddr(). */
849 error = in6_setsockaddr(so, nam);
855 in6_mapped_peeraddr(struct socket *so, struct sockaddr **nam)
857 struct inpcb *inp = so->so_pcb;
862 if (inp->inp_vflag & INP_IPV4) {
863 error = in_setpeeraddr(so, nam);
865 in6_sin_2_v4mapsin6_in_sock(nam);
867 /* scope issues will be handled in in6_setpeeraddr(). */
868 error = in6_setpeeraddr(so, nam);
874 in6_mapped_peeraddr_dispatch(netmsg_t msg)
878 error = in6_mapped_peeraddr(msg->base.nm_so, msg->peeraddr.nm_nam);
879 lwkt_replymsg(&msg->base.lmsg, error);
883 * Pass some notification to all connections of a protocol
884 * associated with address dst. The local address and/or port numbers
885 * may be specified to limit the search. The "usual action" will be
886 * taken, depending on the ctlinput cmd. The caller must filter any
887 * cmds that are uninteresting (e.g., no error in the map).
888 * Call the protocol specific routine (if any) to report
889 * any errors for each matching socket.
891 * Must be called under crit_enter().
894 in6_pcbnotify(struct inpcbhead *head, struct sockaddr *dst, in_port_t fport,
895 const struct sockaddr *src, in_port_t lport, int cmd, int arg,
896 void (*notify) (struct inpcb *, int))
898 struct inpcb *inp, *ninp;
899 struct sockaddr_in6 sa6_src, *sa6_dst;
902 if ((unsigned)cmd >= PRC_NCMDS || dst->sa_family != AF_INET6)
905 sa6_dst = (struct sockaddr_in6 *)dst;
906 if (IN6_IS_ADDR_UNSPECIFIED(&sa6_dst->sin6_addr))
910 * note that src can be NULL when we get notify by local fragmentation.
912 sa6_src = (src == NULL) ? sa6_any : *(const struct sockaddr_in6 *)src;
913 flowinfo = sa6_src.sin6_flowinfo;
916 * Redirects go to all references to the destination,
917 * and use in6_rtchange to invalidate the route cache.
918 * Dead host indications: also use in6_rtchange to invalidate
919 * the cache, and deliver the error to all the sockets.
920 * Otherwise, if we have knowledge of the local port and address,
921 * deliver only to that socket.
923 if (PRC_IS_REDIRECT(cmd) || cmd == PRC_HOSTDEAD) {
926 bzero((caddr_t)&sa6_src.sin6_addr, sizeof(sa6_src.sin6_addr));
928 if (cmd != PRC_HOSTDEAD)
929 notify = in6_rtchange;
931 if (cmd != PRC_MSGSIZE)
932 arg = inet6ctlerrmap[cmd];
934 for (inp = LIST_FIRST(head); inp != NULL; inp = ninp) {
935 ninp = LIST_NEXT(inp, inp_list);
937 if (inp->inp_flags & INP_PLACEMARKER)
940 if ((inp->inp_vflag & INP_IPV6) == 0)
943 * If the error designates a new path MTU for a destination
944 * and the application (associated with this socket) wanted to
945 * know the value, notify. Note that we notify for all
946 * disconnected sockets if the corresponding application
947 * wanted. This is because some UDP applications keep sending
948 * sockets disconnected.
949 * XXX: should we avoid to notify the value to TCP sockets?
951 if (cmd == PRC_MSGSIZE && (inp->inp_flags & IN6P_MTU) != 0 &&
952 (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) ||
953 IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &sa6_dst->sin6_addr))) {
954 ip6_notify_pmtu(inp, (struct sockaddr_in6 *)dst, &arg);
958 * Detect if we should notify the error. If no source and
959 * destination ports are specifed, but non-zero flowinfo and
960 * local address match, notify the error. This is the case
961 * when the error is delivered with an encrypted buffer
962 * by ESP. Otherwise, just compare addresses and ports
965 if (lport == 0 && fport == 0 && flowinfo &&
966 inp->inp_socket != NULL &&
967 flowinfo == (inp->in6p_flowinfo & IPV6_FLOWLABEL_MASK) &&
968 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &sa6_src.sin6_addr))
970 else if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
971 &sa6_dst->sin6_addr) ||
972 inp->inp_socket == 0 ||
973 (lport && inp->inp_lport != lport) ||
974 (!IN6_IS_ADDR_UNSPECIFIED(&sa6_src.sin6_addr) &&
975 !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
976 &sa6_src.sin6_addr)) ||
977 (fport && inp->inp_fport != fport))
988 * Lookup a PCB based on the local address and port.
991 in6_pcblookup_local(struct inpcbportinfo *portinfo,
992 const struct in6_addr *laddr, u_int lport_arg, int wild_okay,
996 int matchwild = 3, wildcard;
997 u_short lport = lport_arg;
998 struct inpcbporthead *porthash;
999 struct inpcbport *phd;
1000 struct inpcb *match = NULL;
1003 * If the porthashbase is shared across several cpus, it must
1006 ASSERT_PORT_TOKEN_HELD(portinfo);
1009 * Best fit PCB lookup.
1011 * First see if this local port is in use by looking on the
1014 porthash = &portinfo->porthashbase[
1015 INP_PCBPORTHASH(lport, portinfo->porthashmask)];
1016 LIST_FOREACH(phd, porthash, phd_hash) {
1017 if (phd->phd_port == lport)
1023 * Port is in use by one or more PCBs. Look for best
1026 LIST_FOREACH(inp, &phd->phd_pcblist, inp_portlist) {
1028 if ((inp->inp_vflag & INP_IPV6) == 0)
1030 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr))
1032 if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1033 if (IN6_IS_ADDR_UNSPECIFIED(laddr))
1035 else if (!IN6_ARE_ADDR_EQUAL(
1036 &inp->in6p_laddr, laddr))
1039 if (!IN6_IS_ADDR_UNSPECIFIED(laddr))
1042 if (wildcard && !wild_okay)
1044 if (wildcard < matchwild &&
1047 inp->inp_socket->so_cred->cr_prison)) {
1049 matchwild = wildcard;
1053 matchwild = wildcard;
1061 in6_pcbpurgeif0(struct in6pcb *head, struct ifnet *ifp)
1063 struct in6pcb *in6p;
1064 struct ip6_moptions *im6o;
1065 struct in6_multi_mship *imm, *nimm;
1067 for (in6p = head; in6p != NULL; in6p = LIST_NEXT(in6p, inp_list)) {
1068 if (in6p->in6p_flags & INP_PLACEMARKER)
1070 im6o = in6p->in6p_moptions;
1071 if ((in6p->inp_vflag & INP_IPV6) &&
1074 * Unselect the outgoing interface if it is being
1077 if (im6o->im6o_multicast_ifp == ifp)
1078 im6o->im6o_multicast_ifp = NULL;
1081 * Drop multicast group membership if we joined
1082 * through the interface being detached.
1083 * XXX controversial - is it really legal for kernel
1086 for (imm = im6o->im6o_memberships.lh_first;
1087 imm != NULL; imm = nimm) {
1088 nimm = imm->i6mm_chain.le_next;
1089 if (imm->i6mm_maddr->in6m_ifp == ifp) {
1090 LIST_REMOVE(imm, i6mm_chain);
1091 in6_delmulti(imm->i6mm_maddr);
1092 kfree(imm, M_IPMADDR);
1100 * Check for alternatives when higher level complains
1101 * about service problems. For now, invalidate cached
1102 * routing information. If the route was created dynamically
1103 * (by a redirect), time to try a default gateway again.
1106 in6_losing(struct inpcb *in6p)
1109 struct rt_addrinfo info;
1111 if ((rt = in6p->in6p_route.ro_rt) != NULL) {
1112 bzero((caddr_t)&info, sizeof(info));
1113 info.rti_flags = rt->rt_flags;
1114 info.rti_info[RTAX_DST] = rt_key(rt);
1115 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1116 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1117 rt_missmsg(RTM_LOSING, &info, rt->rt_flags, 0);
1118 if (rt->rt_flags & RTF_DYNAMIC) {
1119 rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway,
1120 rt_mask(rt), rt->rt_flags, NULL);
1122 in6p->in6p_route.ro_rt = NULL;
1125 * A new route can be allocated
1126 * the next time output is attempted.
1132 * After a routing change, flush old routing
1133 * and allocate a (hopefully) better one.
1136 in6_rtchange(struct inpcb *inp, int error)
1138 if (inp->in6p_route.ro_rt) {
1139 rtfree(inp->in6p_route.ro_rt);
1140 inp->in6p_route.ro_rt = 0;
1142 * A new route can be allocated the next time
1143 * output is attempted.
1149 * Lookup PCB in hash list.
1152 in6_pcblookup_hash(struct inpcbinfo *pcbinfo, struct in6_addr *faddr,
1153 u_int fport_arg, struct in6_addr *laddr, u_int lport_arg,
1154 int wildcard, struct ifnet *ifp)
1156 struct inpcbhead *head;
1158 struct inpcb *jinp = NULL;
1159 u_short fport = fport_arg, lport = lport_arg;
1162 if (faithprefix_p != NULL)
1163 faith = (*faithprefix_p)(laddr);
1168 * First look for an exact match.
1170 head = &pcbinfo->hashbase[INP_PCBCONNHASH(faddr->s6_addr32[3] /* XXX */,
1172 laddr->s6_addr32[3], /* XXX JH */
1174 pcbinfo->hashmask)];
1175 LIST_FOREACH(inp, head, inp_hash) {
1176 if ((inp->inp_vflag & INP_IPV6) == 0)
1178 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, faddr) &&
1179 IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, laddr) &&
1180 inp->inp_fport == fport &&
1181 inp->inp_lport == lport) {
1185 if (inp->inp_socket == NULL ||
1186 inp->inp_socket->so_cred->cr_prison == NULL) {
1197 struct inpcontainerhead *chead;
1198 struct inpcontainer *ic;
1199 struct inpcb *local_wild = NULL;
1200 struct inpcb *jinp_wild = NULL;
1201 struct sockaddr_in6 jsin6;
1205 * Order of socket selection:
1206 * 1. non-jailed, non-wild.
1207 * 2. non-jailed, wild.
1208 * 3. jailed, non-wild.
1211 jsin6.sin6_family = AF_INET6;
1212 chead = &pcbinfo->wildcardhashbase[INP_PCBWILDCARDHASH(lport,
1213 pcbinfo->wildcardhashmask)];
1214 LIST_FOREACH(ic, chead, ic_list) {
1217 if (!(inp->inp_vflag & INP_IPV6))
1219 if (inp->inp_socket != NULL)
1220 cred = inp->inp_socket->so_cred;
1224 if (cred != NULL && jailed(cred)) {
1228 jsin6.sin6_addr = *laddr;
1229 if (!jailed_ip(cred->cr_prison,
1230 (struct sockaddr *)&jsin6))
1234 if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
1235 inp->inp_lport == lport) {
1236 if (faith && (inp->inp_flags & INP_FAITH) == 0)
1238 if (IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
1240 if (cred != NULL && jailed(cred))
1244 } else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
1245 if (cred != NULL && jailed(cred))
1252 if (local_wild != NULL)
1253 return (local_wild);
1266 init_sin6(struct sockaddr_in6 *sin6, struct mbuf *m)
1270 ip = mtod(m, struct ip6_hdr *);
1271 bzero(sin6, sizeof(*sin6));
1272 sin6->sin6_len = sizeof(*sin6);
1273 sin6->sin6_family = AF_INET6;
1274 sin6->sin6_addr = ip->ip6_src;
1275 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1276 sin6->sin6_addr.s6_addr16[1] = 0;
1277 sin6->sin6_scope_id =
1278 (m->m_pkthdr.rcvif && IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1279 ? m->m_pkthdr.rcvif->if_index : 0;
1285 in6_savefaddr(struct socket *so, const struct sockaddr *faddr)
1287 struct sockaddr_in6 *sin6;
1289 KASSERT(faddr->sa_family == AF_INET6,
1290 ("not AF_INET6 faddr %d", faddr->sa_family));
1292 sin6 = kmalloc(sizeof(*sin6), M_SONAME, M_WAITOK | M_ZERO);
1293 sin6->sin6_family = AF_INET6;
1294 sin6->sin6_len = sizeof(*sin6);
1296 sin6->sin6_port = ((const struct sockaddr_in6 *)faddr)->sin6_port;
1297 sin6->sin6_addr = ((const struct sockaddr_in6 *)faddr)->sin6_addr;
1299 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1300 sin6->sin6_scope_id = ntohs(sin6->sin6_addr.s6_addr16[1]);
1302 sin6->sin6_scope_id = 0; /*XXX*/
1303 if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr))
1304 sin6->sin6_addr.s6_addr16[1] = 0;
1306 so->so_faddr = (struct sockaddr *)sin6;
1310 in6_mapped_savefaddr(struct socket *so, const struct sockaddr *faddr)
1312 if (faddr->sa_family == AF_INET) {
1313 in_savefaddr(so, faddr);
1314 in6_sin_2_v4mapsin6_in_sock(&so->so_faddr);
1316 in6_savefaddr(so, faddr);