2 * Copyright (c) 1982, 1986, 1991, 1993
3 * The Regents of the University of California. All rights reserved.
4 * Copyright (C) 2001 WIDE Project. All rights reserved.
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 4. Neither the name of the University nor the names of its contributors
15 * may be used to endorse or promote products derived from this software
16 * without specific prior written permission.
18 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
22 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * @(#)in.c 8.4 (Berkeley) 1/9/95
33 #include <sys/cdefs.h>
34 __FBSDID("$FreeBSD$");
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/sockio.h>
41 #include <sys/malloc.h>
43 #include <sys/socket.h>
45 #include <sys/kernel.h>
47 #include <sys/sysctl.h>
48 #include <sys/vimage.h>
51 #include <net/if_llatbl.h>
52 #include <net/if_types.h>
53 #include <net/route.h>
55 #include <netinet/in.h>
56 #include <netinet/in_var.h>
57 #include <netinet/in_pcb.h>
58 #include <netinet/ip_var.h>
59 #include <netinet/vinet.h>
60 #include <netinet/igmp_var.h>
62 static int in_mask2len(struct in_addr *);
63 static void in_len2mask(struct in_addr *, int);
64 static int in_lifaddr_ioctl(struct socket *, u_long, caddr_t,
65 struct ifnet *, struct thread *);
67 static int in_addprefix(struct in_ifaddr *, int);
68 static int in_scrubprefix(struct in_ifaddr *);
69 static void in_socktrim(struct sockaddr_in *);
70 static int in_ifinit(struct ifnet *,
71 struct in_ifaddr *, struct sockaddr_in *, int);
72 static void in_purgemaddrs(struct ifnet *);
75 static int subnetsarelocal;
76 static int sameprefixcarponly;
77 extern struct inpcbinfo ripcbinfo;
80 SYSCTL_V_INT(V_NET, vnet_inet, _net_inet_ip, OID_AUTO, subnets_are_local,
81 CTLFLAG_RW, subnetsarelocal, 0,
82 "Treat all subnets as directly connected");
83 SYSCTL_V_INT(V_NET, vnet_inet, _net_inet_ip, OID_AUTO, same_prefix_carp_only,
84 CTLFLAG_RW, sameprefixcarponly, 0,
85 "Refuse to create same prefixes on different interfaces");
88 * Return 1 if an internet address is for a ``local'' host
89 * (one to which we have a connection). If subnetsarelocal
90 * is true, this includes other subnets of the local net.
91 * Otherwise, it includes only the directly-connected (sub)nets.
94 in_localaddr(struct in_addr in)
96 INIT_VNET_INET(curvnet);
97 register u_long i = ntohl(in.s_addr);
98 register struct in_ifaddr *ia;
100 if (V_subnetsarelocal) {
101 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link)
102 if ((i & ia->ia_netmask) == ia->ia_net)
105 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link)
106 if ((i & ia->ia_subnetmask) == ia->ia_subnet)
113 * Return 1 if an internet address is for the local host and configured
114 * on one of its interfaces.
117 in_localip(struct in_addr in)
119 INIT_VNET_INET(curvnet);
120 struct in_ifaddr *ia;
122 LIST_FOREACH(ia, INADDR_HASH(in.s_addr), ia_hash) {
123 if (IA_SIN(ia)->sin_addr.s_addr == in.s_addr)
130 * Determine whether an IP address is in a reserved set of addresses
131 * that may not be forwarded, or whether datagrams to that destination
135 in_canforward(struct in_addr in)
137 register u_long i = ntohl(in.s_addr);
140 if (IN_EXPERIMENTAL(i) || IN_MULTICAST(i) || IN_LINKLOCAL(i))
143 net = i & IN_CLASSA_NET;
144 if (net == 0 || net == (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))
151 * Trim a mask in a sockaddr
154 in_socktrim(struct sockaddr_in *ap)
156 register char *cplim = (char *) &ap->sin_addr;
157 register char *cp = (char *) (&ap->sin_addr + 1);
160 while (--cp >= cplim)
162 (ap)->sin_len = cp - (char *) (ap) + 1;
169 struct in_addr *mask;
175 for (x = 0; x < sizeof(*mask); x++) {
180 if (x < sizeof(*mask)) {
181 for (y = 0; y < 8; y++) {
182 if ((p[x] & (0x80 >> y)) == 0)
190 in_len2mask(struct in_addr *mask, int len)
196 bzero(mask, sizeof(*mask));
197 for (i = 0; i < len / 8; i++)
200 p[i] = (0xff00 >> (len % 8)) & 0xff;
204 * Generic internet control operations (ioctl's).
206 * ifp is NULL if not an interface-specific ioctl.
210 in_control(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp,
213 INIT_VNET_INET(curvnet); /* both so and ifp can be NULL here! */
214 register struct ifreq *ifr = (struct ifreq *)data;
215 register struct in_ifaddr *ia, *iap;
216 register struct ifaddr *ifa;
217 struct in_addr allhosts_addr;
219 struct in_ifaddr *oia;
220 struct in_ifinfo *ii;
221 struct in_aliasreq *ifra = (struct in_aliasreq *)data;
222 struct sockaddr_in oldaddr;
223 int error, hostIsNew, iaIsNew, maskIsNew, s;
229 allhosts_addr.s_addr = htonl(INADDR_ALLHOSTS_GROUP);
232 * Filter out ioctls we implement directly; forward the rest on to
233 * in_lifaddr_ioctl() and ifp->if_ioctl().
250 error = priv_check(td, PRIV_NET_ADDIFADDR);
256 return in_lifaddr_ioctl(so, cmd, data, ifp, td);
260 error = priv_check(td, PRIV_NET_DELIFADDR);
266 return in_lifaddr_ioctl(so, cmd, data, ifp, td);
271 return in_lifaddr_ioctl(so, cmd, data, ifp, td);
274 if (ifp == NULL || ifp->if_ioctl == NULL)
276 return ((*ifp->if_ioctl)(ifp, cmd, data));
280 return (EADDRNOTAVAIL);
283 * Security checks before we get involved in any work.
292 error = priv_check(td, PRIV_NET_ADDIFADDR);
300 error = priv_check(td, PRIV_NET_DELIFADDR);
308 * Find address for this interface, if it exists.
310 * If an alias address was specified, find that one instead of the
311 * first one on the interface, if possible.
313 dst = ((struct sockaddr_in *)&ifr->ifr_addr)->sin_addr;
314 LIST_FOREACH(iap, INADDR_HASH(dst.s_addr), ia_hash) {
315 if (iap->ia_ifp == ifp &&
316 iap->ia_addr.sin_addr.s_addr == dst.s_addr) {
317 if (td == NULL || prison_check_ip4(td->td_ucred,
325 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
327 if (iap->ia_addr.sin_family == AF_INET) {
329 prison_check_ip4(td->td_ucred,
330 &iap->ia_addr.sin_addr) != 0)
344 if (ifra->ifra_addr.sin_family == AF_INET) {
345 for (oia = ia; ia; ia = TAILQ_NEXT(ia, ia_link)) {
346 if (ia->ia_ifp == ifp &&
347 ia->ia_addr.sin_addr.s_addr ==
348 ifra->ifra_addr.sin_addr.s_addr)
351 if ((ifp->if_flags & IFF_POINTOPOINT)
352 && (cmd == SIOCAIFADDR)
353 && (ifra->ifra_dstaddr.sin_addr.s_addr
355 error = EDESTADDRREQ;
359 if (cmd == SIOCDIFADDR && ia == NULL) {
360 error = EADDRNOTAVAIL;
368 ia = (struct in_ifaddr *)
369 malloc(sizeof *ia, M_IFADDR, M_NOWAIT |
378 ifa->ifa_addr = (struct sockaddr *)&ia->ia_addr;
379 ifa->ifa_dstaddr = (struct sockaddr *)&ia->ia_dstaddr;
380 ifa->ifa_netmask = (struct sockaddr *)&ia->ia_sockmask;
383 ia->ia_sockmask.sin_len = 8;
384 ia->ia_sockmask.sin_family = AF_INET;
385 if (ifp->if_flags & IFF_BROADCAST) {
386 ia->ia_broadaddr.sin_len = sizeof(ia->ia_addr);
387 ia->ia_broadaddr.sin_family = AF_INET;
391 TAILQ_INSERT_TAIL(&ifp->if_addrhead, ifa, ifa_link);
393 TAILQ_INSERT_TAIL(&V_in_ifaddrhead, ia, ia_link);
405 error = EADDRNOTAVAIL;
412 * Most paths in this switch return directly or via out_unlock. Only
413 * paths that remove the address break in order to hit common removal
416 * XXXRW: We enter the switch with IF_ADDR_LOCK() held, but leave
417 * without it. This is a bug.
419 IF_ADDR_LOCK_ASSERT(ifp);
422 *((struct sockaddr_in *)&ifr->ifr_addr) = ia->ia_addr;
426 if ((ifp->if_flags & IFF_BROADCAST) == 0) {
430 *((struct sockaddr_in *)&ifr->ifr_dstaddr) = ia->ia_broadaddr;
434 if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
438 *((struct sockaddr_in *)&ifr->ifr_dstaddr) = ia->ia_dstaddr;
442 *((struct sockaddr_in *)&ifr->ifr_addr) = ia->ia_sockmask;
446 if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
450 oldaddr = ia->ia_dstaddr;
451 ia->ia_dstaddr = *(struct sockaddr_in *)&ifr->ifr_dstaddr;
455 * XXXRW: Locks dropped for if_ioctl and rtinit, but ia is
458 if (ifp->if_ioctl != NULL) {
459 error = (*ifp->if_ioctl)(ifp, SIOCSIFDSTADDR,
462 ia->ia_dstaddr = oldaddr;
466 if (ia->ia_flags & IFA_ROUTE) {
467 ia->ia_ifa.ifa_dstaddr = (struct sockaddr *)&oldaddr;
468 rtinit(&(ia->ia_ifa), (int)RTM_DELETE, RTF_HOST);
469 ia->ia_ifa.ifa_dstaddr =
470 (struct sockaddr *)&ia->ia_dstaddr;
471 rtinit(&(ia->ia_ifa), (int)RTM_ADD, RTF_HOST|RTF_UP);
476 if ((ifp->if_flags & IFF_BROADCAST) == 0) {
480 ia->ia_broadaddr = *(struct sockaddr_in *)&ifr->ifr_broadaddr;
487 * XXXRW: Locks dropped for in_ifinit and in_joingroup, but ia
488 * is still being used.
490 error = in_ifinit(ifp, ia,
491 (struct sockaddr_in *) &ifr->ifr_addr, 1);
492 if (error != 0 && iaIsNew)
495 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
497 (ifp->if_flags & IFF_MULTICAST) != 0) {
498 error = in_joingroup(ifp, &allhosts_addr,
499 NULL, &ii->ii_allhosts);
501 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
506 ia->ia_sockmask.sin_addr = ifra->ifra_addr.sin_addr;
507 ia->ia_subnetmask = ntohl(ia->ia_sockmask.sin_addr.s_addr);
514 if (ia->ia_addr.sin_family == AF_INET) {
515 if (ifra->ifra_addr.sin_len == 0) {
516 ifra->ifra_addr = ia->ia_addr;
518 } else if (ifra->ifra_addr.sin_addr.s_addr ==
519 ia->ia_addr.sin_addr.s_addr)
525 * XXXRW: Locks dropped for in_ifscrub and in_ifinit, but ia
526 * is still being used.
528 if (ifra->ifra_mask.sin_len) {
530 ia->ia_sockmask = ifra->ifra_mask;
531 ia->ia_sockmask.sin_family = AF_INET;
533 ntohl(ia->ia_sockmask.sin_addr.s_addr);
536 if ((ifp->if_flags & IFF_POINTOPOINT) &&
537 (ifra->ifra_dstaddr.sin_family == AF_INET)) {
539 ia->ia_dstaddr = ifra->ifra_dstaddr;
540 maskIsNew = 1; /* We lie; but the effect's the same */
542 if (ifra->ifra_addr.sin_family == AF_INET &&
543 (hostIsNew || maskIsNew))
544 error = in_ifinit(ifp, ia, &ifra->ifra_addr, 0);
545 if (error != 0 && iaIsNew)
548 if ((ifp->if_flags & IFF_BROADCAST) &&
549 (ifra->ifra_broadaddr.sin_family == AF_INET))
550 ia->ia_broadaddr = ifra->ifra_broadaddr;
552 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
554 (ifp->if_flags & IFF_MULTICAST) != 0) {
555 error = in_joingroup(ifp, &allhosts_addr,
556 NULL, &ii->ii_allhosts);
558 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
566 * XXXRW: Locks dropped for in_ifscrub and in_ifadown, but ia
567 * is still being used.
569 * in_ifscrub kills the interface route.
574 * in_ifadown gets rid of all the rest of
575 * the routes. This is not quite the right
576 * thing to do, but at least if we are running
577 * a routing process they will come back.
579 in_ifadown(&ia->ia_ifa, 1);
580 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
585 panic("in_control: unsupported ioctl");
589 * XXXRW: In a more ideal world, we would still be holding
593 TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
596 TAILQ_REMOVE(&V_in_ifaddrhead, ia, ia_link);
597 if (ia->ia_addr.sin_family == AF_INET) {
598 LIST_REMOVE(ia, ia_hash);
600 * If this is the last IPv4 address configured on this
601 * interface, leave the all-hosts group.
602 * No state-change report need be transmitted.
607 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
609 if (ii->ii_allhosts) {
610 (void)in_leavegroup_locked(ii->ii_allhosts,
612 ii->ii_allhosts = NULL;
617 IFAFREE(&ia->ia_ifa);
629 * SIOCGLIFADDR: get first address. (?!?)
630 * SIOCGLIFADDR with IFLR_PREFIX:
631 * get first address that matches the specified prefix.
632 * SIOCALIFADDR: add the specified address.
633 * SIOCALIFADDR with IFLR_PREFIX:
634 * EINVAL since we can't deduce hostid part of the address.
635 * SIOCDLIFADDR: delete the specified address.
636 * SIOCDLIFADDR with IFLR_PREFIX:
637 * delete the first address that matches the specified prefix.
639 * EINVAL on invalid parameters
640 * EADDRNOTAVAIL on prefix match failed/specified address not found
641 * other values may be returned from in_ioctl()
644 in_lifaddr_ioctl(struct socket *so, u_long cmd, caddr_t data,
645 struct ifnet *ifp, struct thread *td)
647 struct if_laddrreq *iflr = (struct if_laddrreq *)data;
651 if (data == NULL || ifp == NULL) {
652 panic("invalid argument to in_lifaddr_ioctl");
658 /* address must be specified on GET with IFLR_PREFIX */
659 if ((iflr->flags & IFLR_PREFIX) == 0)
664 /* address must be specified on ADD and DELETE */
665 if (iflr->addr.ss_family != AF_INET)
667 if (iflr->addr.ss_len != sizeof(struct sockaddr_in))
669 /* XXX need improvement */
670 if (iflr->dstaddr.ss_family
671 && iflr->dstaddr.ss_family != AF_INET)
673 if (iflr->dstaddr.ss_family
674 && iflr->dstaddr.ss_len != sizeof(struct sockaddr_in))
677 default: /*shouldn't happen*/
680 if (sizeof(struct in_addr) * 8 < iflr->prefixlen)
686 struct in_aliasreq ifra;
688 if (iflr->flags & IFLR_PREFIX)
691 /* copy args to in_aliasreq, perform ioctl(SIOCAIFADDR_IN6). */
692 bzero(&ifra, sizeof(ifra));
693 bcopy(iflr->iflr_name, ifra.ifra_name,
694 sizeof(ifra.ifra_name));
696 bcopy(&iflr->addr, &ifra.ifra_addr, iflr->addr.ss_len);
698 if (iflr->dstaddr.ss_family) { /*XXX*/
699 bcopy(&iflr->dstaddr, &ifra.ifra_dstaddr,
700 iflr->dstaddr.ss_len);
703 ifra.ifra_mask.sin_family = AF_INET;
704 ifra.ifra_mask.sin_len = sizeof(struct sockaddr_in);
705 in_len2mask(&ifra.ifra_mask.sin_addr, iflr->prefixlen);
707 return (in_control(so, SIOCAIFADDR, (caddr_t)&ifra, ifp, td));
712 struct in_ifaddr *ia;
713 struct in_addr mask, candidate, match;
714 struct sockaddr_in *sin;
716 bzero(&mask, sizeof(mask));
717 bzero(&match, sizeof(match));
718 if (iflr->flags & IFLR_PREFIX) {
719 /* lookup a prefix rather than address. */
720 in_len2mask(&mask, iflr->prefixlen);
722 sin = (struct sockaddr_in *)&iflr->addr;
723 match.s_addr = sin->sin_addr.s_addr;
724 match.s_addr &= mask.s_addr;
726 /* if you set extra bits, that's wrong */
727 if (match.s_addr != sin->sin_addr.s_addr)
731 /* on getting an address, take the 1st match */
732 /* on deleting an address, do exact match */
733 if (cmd != SIOCGLIFADDR) {
734 in_len2mask(&mask, 32);
735 sin = (struct sockaddr_in *)&iflr->addr;
736 match.s_addr = sin->sin_addr.s_addr;
740 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
741 if (ifa->ifa_addr->sa_family != AF_INET6)
743 if (match.s_addr == 0)
745 candidate.s_addr = ((struct sockaddr_in *)&ifa->ifa_addr)->sin_addr.s_addr;
746 candidate.s_addr &= mask.s_addr;
747 if (candidate.s_addr == match.s_addr)
751 return (EADDRNOTAVAIL);
752 ia = (struct in_ifaddr *)ifa;
754 if (cmd == SIOCGLIFADDR) {
755 /* fill in the if_laddrreq structure */
756 bcopy(&ia->ia_addr, &iflr->addr, ia->ia_addr.sin_len);
758 if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
759 bcopy(&ia->ia_dstaddr, &iflr->dstaddr,
760 ia->ia_dstaddr.sin_len);
762 bzero(&iflr->dstaddr, sizeof(iflr->dstaddr));
765 in_mask2len(&ia->ia_sockmask.sin_addr);
767 iflr->flags = 0; /*XXX*/
771 struct in_aliasreq ifra;
773 /* fill in_aliasreq and do ioctl(SIOCDIFADDR_IN6) */
774 bzero(&ifra, sizeof(ifra));
775 bcopy(iflr->iflr_name, ifra.ifra_name,
776 sizeof(ifra.ifra_name));
778 bcopy(&ia->ia_addr, &ifra.ifra_addr,
779 ia->ia_addr.sin_len);
780 if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
781 bcopy(&ia->ia_dstaddr, &ifra.ifra_dstaddr,
782 ia->ia_dstaddr.sin_len);
784 bcopy(&ia->ia_sockmask, &ifra.ifra_dstaddr,
785 ia->ia_sockmask.sin_len);
787 return (in_control(so, SIOCDIFADDR, (caddr_t)&ifra,
793 return (EOPNOTSUPP); /*just for safety*/
797 * Delete any existing route for an interface.
800 in_ifscrub(struct ifnet *ifp, struct in_ifaddr *ia)
807 * Initialize an interface's internet address
808 * and routing table entry.
811 in_ifinit(struct ifnet *ifp, struct in_ifaddr *ia, struct sockaddr_in *sin,
814 INIT_VNET_INET(ifp->if_vnet);
815 register u_long i = ntohl(sin->sin_addr.s_addr);
816 struct sockaddr_in oldaddr;
817 int s = splimp(), flags = RTF_UP, error = 0;
819 oldaddr = ia->ia_addr;
820 if (oldaddr.sin_family == AF_INET)
821 LIST_REMOVE(ia, ia_hash);
823 if (ia->ia_addr.sin_family == AF_INET)
824 LIST_INSERT_HEAD(INADDR_HASH(ia->ia_addr.sin_addr.s_addr),
827 * Give the interface a chance to initialize
828 * if this is its first address,
829 * and to validate the address if necessary.
831 if (ifp->if_ioctl != NULL) {
832 error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
835 /* LIST_REMOVE(ia, ia_hash) is done in in_control */
836 ia->ia_addr = oldaddr;
837 if (ia->ia_addr.sin_family == AF_INET)
838 LIST_INSERT_HEAD(INADDR_HASH(
839 ia->ia_addr.sin_addr.s_addr), ia, ia_hash);
842 * If oldaddr family is not AF_INET (e.g.
843 * interface has been just created) in_control
844 * does not call LIST_REMOVE, and we end up
845 * with bogus ia entries in hash
847 LIST_REMOVE(ia, ia_hash);
853 ia->ia_ifa.ifa_addr = (struct sockaddr *)&oldaddr;
855 ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr;
858 ia->ia_netmask = IN_CLASSA_NET;
859 else if (IN_CLASSB(i))
860 ia->ia_netmask = IN_CLASSB_NET;
862 ia->ia_netmask = IN_CLASSC_NET;
864 * The subnet mask usually includes at least the standard network part,
865 * but may may be smaller in the case of supernetting.
866 * If it is set, we believe it.
868 if (ia->ia_subnetmask == 0) {
869 ia->ia_subnetmask = ia->ia_netmask;
870 ia->ia_sockmask.sin_addr.s_addr = htonl(ia->ia_subnetmask);
872 ia->ia_netmask &= ia->ia_subnetmask;
873 ia->ia_net = i & ia->ia_netmask;
874 ia->ia_subnet = i & ia->ia_subnetmask;
875 in_socktrim(&ia->ia_sockmask);
878 * XXX: carp(4) does not have interface route
880 if (ifp->if_type == IFT_CARP)
884 * Add route for the network.
886 ia->ia_ifa.ifa_metric = ifp->if_metric;
887 if (ifp->if_flags & IFF_BROADCAST) {
888 ia->ia_broadaddr.sin_addr.s_addr =
889 htonl(ia->ia_subnet | ~ia->ia_subnetmask);
890 ia->ia_netbroadcast.s_addr =
891 htonl(ia->ia_net | ~ ia->ia_netmask);
892 } else if (ifp->if_flags & IFF_LOOPBACK) {
893 ia->ia_dstaddr = ia->ia_addr;
895 } else if (ifp->if_flags & IFF_POINTOPOINT) {
896 if (ia->ia_dstaddr.sin_family != AF_INET)
900 if ((error = in_addprefix(ia, flags)) != 0)
906 #define rtinitflags(x) \
907 ((((x)->ia_ifp->if_flags & (IFF_LOOPBACK | IFF_POINTOPOINT)) != 0) \
910 * Check if we have a route for the given prefix already or add one accordingly.
913 in_addprefix(struct in_ifaddr *target, int flags)
915 INIT_VNET_INET(curvnet);
916 struct in_ifaddr *ia;
917 struct in_addr prefix, mask, p, m;
920 if ((flags & RTF_HOST) != 0) {
921 prefix = target->ia_dstaddr.sin_addr;
924 prefix = target->ia_addr.sin_addr;
925 mask = target->ia_sockmask.sin_addr;
926 prefix.s_addr &= mask.s_addr;
929 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
930 if (rtinitflags(ia)) {
931 p = ia->ia_addr.sin_addr;
933 if (prefix.s_addr != p.s_addr)
936 p = ia->ia_addr.sin_addr;
937 m = ia->ia_sockmask.sin_addr;
938 p.s_addr &= m.s_addr;
940 if (prefix.s_addr != p.s_addr ||
941 mask.s_addr != m.s_addr)
946 * If we got a matching prefix route inserted by other
947 * interface address, we are done here.
949 if (ia->ia_flags & IFA_ROUTE) {
950 if (V_sameprefixcarponly &&
951 target->ia_ifp->if_type != IFT_CARP &&
952 ia->ia_ifp->if_type != IFT_CARP)
960 * No-one seem to have this prefix route, so we try to insert it.
962 error = rtinit(&target->ia_ifa, (int)RTM_ADD, flags);
964 target->ia_flags |= IFA_ROUTE;
968 extern void arp_ifscrub(struct ifnet *ifp, uint32_t addr);
971 * If there is no other address in the system that can serve a route to the
972 * same prefix, remove the route. Hand over the route to the new address
976 in_scrubprefix(struct in_ifaddr *target)
978 INIT_VNET_INET(curvnet);
979 struct in_ifaddr *ia;
980 struct in_addr prefix, mask, p;
983 if ((target->ia_flags & IFA_ROUTE) == 0)
986 if (rtinitflags(target))
987 prefix = target->ia_dstaddr.sin_addr;
989 prefix = target->ia_addr.sin_addr;
990 mask = target->ia_sockmask.sin_addr;
991 prefix.s_addr &= mask.s_addr;
992 /* remove arp cache */
993 arp_ifscrub(target->ia_ifp, IA_SIN(target)->sin_addr.s_addr);
996 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
998 p = ia->ia_dstaddr.sin_addr;
1000 p = ia->ia_addr.sin_addr;
1001 p.s_addr &= ia->ia_sockmask.sin_addr.s_addr;
1004 if (prefix.s_addr != p.s_addr)
1008 * If we got a matching prefix address, move IFA_ROUTE and
1009 * the route itself to it. Make sure that routing daemons
1012 * XXX: a special case for carp(4) interface
1014 if ((ia->ia_flags & IFA_ROUTE) == 0
1016 && (ia->ia_ifp->if_type != IFT_CARP)
1019 rtinit(&(target->ia_ifa), (int)RTM_DELETE,
1020 rtinitflags(target));
1021 target->ia_flags &= ~IFA_ROUTE;
1023 error = rtinit(&ia->ia_ifa, (int)RTM_ADD,
1024 rtinitflags(ia) | RTF_UP);
1026 ia->ia_flags |= IFA_ROUTE;
1032 * As no-one seem to have this prefix, we can remove the route.
1034 rtinit(&(target->ia_ifa), (int)RTM_DELETE, rtinitflags(target));
1035 target->ia_flags &= ~IFA_ROUTE;
1042 * Return 1 if the address might be a local broadcast address.
1045 in_broadcast(struct in_addr in, struct ifnet *ifp)
1047 register struct ifaddr *ifa;
1050 if (in.s_addr == INADDR_BROADCAST ||
1051 in.s_addr == INADDR_ANY)
1053 if ((ifp->if_flags & IFF_BROADCAST) == 0)
1055 t = ntohl(in.s_addr);
1057 * Look through the list of addresses for a match
1058 * with a broadcast address.
1060 #define ia ((struct in_ifaddr *)ifa)
1061 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
1062 if (ifa->ifa_addr->sa_family == AF_INET &&
1063 (in.s_addr == ia->ia_broadaddr.sin_addr.s_addr ||
1064 in.s_addr == ia->ia_netbroadcast.s_addr ||
1066 * Check for old-style (host 0) broadcast.
1068 t == ia->ia_subnet || t == ia->ia_net) &&
1070 * Check for an all one subnetmask. These
1071 * only exist when an interface gets a secondary
1074 ia->ia_subnetmask != (u_long)0xffffffff)
1081 * On interface removal, clean up IPv4 data structures hung off of the ifnet.
1084 in_ifdetach(struct ifnet *ifp)
1086 INIT_VNET_INET(ifp->if_vnet);
1088 in_pcbpurgeif0(&V_ripcbinfo, ifp);
1089 in_pcbpurgeif0(&V_udbinfo, ifp);
1090 in_purgemaddrs(ifp);
1094 * Delete all IPv4 multicast address records, and associated link-layer
1095 * multicast address records, associated with ifp.
1096 * XXX It looks like domifdetach runs AFTER the link layer cleanup.
1097 * XXX This should not race with ifma_protospec being set during
1098 * a new allocation, if it does, we have bigger problems.
1101 in_purgemaddrs(struct ifnet *ifp)
1103 LIST_HEAD(,in_multi) purgeinms;
1104 struct in_multi *inm, *tinm;
1105 struct ifmultiaddr *ifma;
1107 LIST_INIT(&purgeinms);
1111 * Extract list of in_multi associated with the detaching ifp
1112 * which the PF_INET layer is about to release.
1113 * We need to do this as IF_ADDR_LOCK() may be re-acquired
1114 * by code further down.
1117 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
1118 if (ifma->ifma_addr->sa_family != AF_INET ||
1119 ifma->ifma_protospec == NULL)
1122 KASSERT(ifma->ifma_protospec != NULL,
1123 ("%s: ifma_protospec is NULL", __func__));
1125 inm = (struct in_multi *)ifma->ifma_protospec;
1126 LIST_INSERT_HEAD(&purgeinms, inm, inm_link);
1128 IF_ADDR_UNLOCK(ifp);
1130 LIST_FOREACH_SAFE(inm, &purgeinms, inm_link, tinm) {
1131 LIST_REMOVE(inm, inm_link);
1132 inm_release_locked(inm);
1139 #include <sys/syslog.h>
1140 #include <net/if_dl.h>
1141 #include <netinet/if_ether.h>
1144 struct llentry base;
1145 struct sockaddr_in l3_addr4;
1148 static struct llentry *
1149 in_lltable_new(const struct sockaddr *l3addr, u_int flags)
1151 struct in_llentry *lle;
1153 lle = malloc(sizeof(struct in_llentry), M_LLTABLE, M_DONTWAIT | M_ZERO);
1154 if (lle == NULL) /* NB: caller generates msg */
1157 callout_init(&lle->base.la_timer, CALLOUT_MPSAFE);
1159 * For IPv4 this will trigger "arpresolve" to generate
1162 lle->base.la_expire = time_second; /* mark expired */
1163 lle->l3_addr4 = *(const struct sockaddr_in *)l3addr;
1164 lle->base.lle_refcnt = 1;
1165 LLE_LOCK_INIT(&lle->base);
1170 * Deletes an address from the address table.
1171 * This function is called by the timer functions
1172 * such as arptimer() and nd6_llinfo_timer(), and
1173 * the caller does the locking.
1176 in_lltable_free(struct lltable *llt, struct llentry *lle)
1179 LLE_LOCK_DESTROY(lle);
1180 free(lle, M_LLTABLE);
1184 in_lltable_rtcheck(struct ifnet *ifp, const struct sockaddr *l3addr)
1188 KASSERT(l3addr->sa_family == AF_INET,
1189 ("sin_family %d", l3addr->sa_family));
1191 /* XXX rtalloc1 should take a const param */
1192 rt = rtalloc1(__DECONST(struct sockaddr *, l3addr), 0, 0);
1193 if (rt == NULL || (rt->rt_flags & RTF_GATEWAY) || rt->rt_ifp != ifp) {
1194 log(LOG_INFO, "IPv4 address: \"%s\" is not on the network\n",
1195 inet_ntoa(((const struct sockaddr_in *)l3addr)->sin_addr));
1205 * Return NULL if not found or marked for deletion.
1206 * If found return lle read locked.
1208 static struct llentry *
1209 in_lltable_lookup(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
1211 const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
1212 struct ifnet *ifp = llt->llt_ifp;
1213 struct llentry *lle;
1214 struct llentries *lleh;
1217 IF_AFDATA_LOCK_ASSERT(ifp);
1218 KASSERT(l3addr->sa_family == AF_INET,
1219 ("sin_family %d", l3addr->sa_family));
1221 hashkey = sin->sin_addr.s_addr;
1222 lleh = &llt->lle_head[LLATBL_HASH(hashkey, LLTBL_HASHMASK)];
1223 LIST_FOREACH(lle, lleh, lle_next) {
1224 struct sockaddr_in *sa2 = (struct sockaddr_in *)L3_ADDR(lle);
1225 if (lle->la_flags & LLE_DELETED)
1227 if (sa2->sin_addr.s_addr == sin->sin_addr.s_addr)
1232 if (flags & LLE_DELETE)
1233 log(LOG_INFO, "interface address is missing from cache = %p in delete\n", lle);
1235 if (!(flags & LLE_CREATE))
1238 * A route that covers the given address must have
1239 * been installed 1st because we are doing a resolution,
1242 if (!(flags & LLE_IFADDR) &&
1243 in_lltable_rtcheck(ifp, l3addr) != 0)
1246 lle = in_lltable_new(l3addr, flags);
1248 log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
1251 lle->la_flags = flags & ~LLE_CREATE;
1252 if ((flags & (LLE_CREATE | LLE_IFADDR)) == (LLE_CREATE | LLE_IFADDR)) {
1253 bcopy(IF_LLADDR(ifp), &lle->ll_addr, ifp->if_addrlen);
1254 lle->la_flags |= (LLE_VALID | LLE_STATIC);
1258 lle->lle_head = lleh;
1259 LIST_INSERT_HEAD(lleh, lle, lle_next);
1260 } else if (flags & LLE_DELETE) {
1261 if (!(lle->la_flags & LLE_IFADDR) || (flags & LLE_IFADDR)) {
1263 lle->la_flags = LLE_DELETED;
1266 log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
1272 if (LLE_IS_VALID(lle)) {
1273 if (flags & LLE_EXCLUSIVE)
1283 in_lltable_dump(struct lltable *llt, struct sysctl_req *wr)
1285 #define SIN(lle) ((struct sockaddr_in *) L3_ADDR(lle))
1286 struct ifnet *ifp = llt->llt_ifp;
1287 struct llentry *lle;
1290 struct rt_msghdr rtm;
1291 struct sockaddr_inarp sin;
1292 struct sockaddr_dl sdl;
1297 * current IFNET_RLOCK() is mapped to IFNET_WLOCK()
1298 * so it is okay to use this ASSERT, change it when
1299 * IFNET lock is finalized
1301 IFNET_WLOCK_ASSERT();
1304 for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) {
1305 LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
1306 struct sockaddr_dl *sdl;
1308 /* skip deleted entries */
1309 if ((lle->la_flags & (LLE_DELETED|LLE_VALID)) != LLE_VALID)
1311 /* Skip if jailed and not a valid IP of the prison. */
1312 if (prison_if(wr->td->td_ucred, L3_ADDR(lle)) != 0)
1315 * produce a msg made of:
1317 * struct sockaddr_inarp; (IPv4)
1318 * struct sockaddr_dl;
1320 bzero(&arpc, sizeof(arpc));
1321 arpc.rtm.rtm_msglen = sizeof(arpc);
1322 arpc.rtm.rtm_version = RTM_VERSION;
1323 arpc.rtm.rtm_type = RTM_GET;
1324 arpc.rtm.rtm_flags = RTF_UP;
1325 arpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
1326 arpc.sin.sin_family = AF_INET;
1327 arpc.sin.sin_len = sizeof(arpc.sin);
1328 arpc.sin.sin_addr.s_addr = SIN(lle)->sin_addr.s_addr;
1331 if (lle->la_flags & LLE_PUB) {
1332 arpc.rtm.rtm_flags |= RTF_ANNOUNCE;
1334 if (lle->la_flags & LLE_PROXY)
1335 arpc.sin.sin_other = SIN_PROXY;
1339 sdl->sdl_family = AF_LINK;
1340 sdl->sdl_len = sizeof(*sdl);
1341 sdl->sdl_alen = ifp->if_addrlen;
1342 sdl->sdl_index = ifp->if_index;
1343 sdl->sdl_type = ifp->if_type;
1344 bcopy(&lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
1346 arpc.rtm.rtm_rmx.rmx_expire =
1347 lle->la_flags & LLE_STATIC ? 0 : lle->la_expire;
1348 arpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
1349 if (lle->la_flags & LLE_STATIC)
1350 arpc.rtm.rtm_flags |= RTF_STATIC;
1351 arpc.rtm.rtm_index = ifp->if_index;
1352 error = SYSCTL_OUT(wr, &arpc, sizeof(arpc));
1362 in_domifattach(struct ifnet *ifp)
1364 struct in_ifinfo *ii;
1365 struct lltable *llt;
1367 ii = malloc(sizeof(struct in_ifinfo), M_IFADDR, M_WAITOK|M_ZERO);
1369 llt = lltable_init(ifp, AF_INET);
1371 llt->llt_new = in_lltable_new;
1372 llt->llt_free = in_lltable_free;
1373 llt->llt_rtcheck = in_lltable_rtcheck;
1374 llt->llt_lookup = in_lltable_lookup;
1375 llt->llt_dump = in_lltable_dump;
1379 ii->ii_igmp = igmp_domifattach(ifp);
1385 in_domifdetach(struct ifnet *ifp, void *aux)
1387 struct in_ifinfo *ii = (struct in_ifinfo *)aux;
1389 igmp_domifdetach(ifp);
1390 lltable_free(ii->ii_llt);