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("$FreeBSD$");
36 #include "opt_mpath.h"
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/syslog.h>
51 #include <net/if_var.h>
52 #include <net/if_arp.h>
53 #include <net/if_dl.h>
54 #include <net/if_llatbl.h>
55 #include <net/if_types.h>
56 #include <net/route.h>
59 #include <netinet/if_ether.h>
60 #include <netinet/in.h>
61 #include <netinet/in_var.h>
62 #include <netinet/in_pcb.h>
63 #include <netinet/ip_var.h>
64 #include <netinet/ip_carp.h>
65 #include <netinet/igmp_var.h>
66 #include <netinet/udp.h>
67 #include <netinet/udp_var.h>
69 static int in_mask2len(struct in_addr *);
70 static void in_len2mask(struct in_addr *, int);
71 static int in_lifaddr_ioctl(struct socket *, u_long, caddr_t,
72 struct ifnet *, struct thread *);
74 static void in_socktrim(struct sockaddr_in *);
75 static int in_ifinit(struct ifnet *, struct in_ifaddr *,
76 struct sockaddr_in *, int, int, int);
77 static void in_purgemaddrs(struct ifnet *);
79 static VNET_DEFINE(int, nosameprefix);
80 #define V_nosameprefix VNET(nosameprefix)
81 SYSCTL_VNET_INT(_net_inet_ip, OID_AUTO, no_same_prefix, CTLFLAG_RW,
82 &VNET_NAME(nosameprefix), 0,
83 "Refuse to create same prefixes on different interfaces");
85 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
86 #define V_ripcbinfo VNET(ripcbinfo)
88 VNET_DECLARE(struct arpstat, arpstat); /* ARP statistics, see if_arp.h */
89 #define V_arpstat VNET(arpstat)
92 * Return 1 if an internet address is for a ``local'' host
93 * (one to which we have a connection).
96 in_localaddr(struct in_addr in)
98 register u_long i = ntohl(in.s_addr);
99 register struct in_ifaddr *ia;
102 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
103 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 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) {
133 * Determine whether an IP address is in a reserved set of addresses
134 * that may not be forwarded, or whether datagrams to that destination
138 in_canforward(struct in_addr in)
140 register u_long i = ntohl(in.s_addr);
143 if (IN_EXPERIMENTAL(i) || IN_MULTICAST(i) || IN_LINKLOCAL(i))
146 net = i & IN_CLASSA_NET;
147 if (net == 0 || net == (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))
154 * Trim a mask in a sockaddr
157 in_socktrim(struct sockaddr_in *ap)
159 register char *cplim = (char *) &ap->sin_addr;
160 register char *cp = (char *) (&ap->sin_addr + 1);
163 while (--cp >= cplim)
165 (ap)->sin_len = cp - (char *) (ap) + 1;
172 struct in_addr *mask;
178 for (x = 0; x < sizeof(*mask); x++) {
183 if (x < sizeof(*mask)) {
184 for (y = 0; y < 8; y++) {
185 if ((p[x] & (0x80 >> y)) == 0)
193 in_len2mask(struct in_addr *mask, int len)
199 bzero(mask, sizeof(*mask));
200 for (i = 0; i < len / 8; i++)
203 p[i] = (0xff00 >> (len % 8)) & 0xff;
207 * Generic internet control operations (ioctl's).
209 * ifp is NULL if not an interface-specific ioctl.
213 in_control(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp,
216 register struct ifreq *ifr = (struct ifreq *)data;
217 register struct in_ifaddr *ia, *iap;
218 register struct ifaddr *ifa;
219 struct in_addr allhosts_addr;
221 struct in_ifinfo *ii;
222 struct in_aliasreq *ifra = (struct in_aliasreq *)data;
223 struct sockaddr_in oldaddr;
224 int error, hostIsNew, iaIsNew, maskIsNew;
230 allhosts_addr.s_addr = htonl(INADDR_ALLHOSTS_GROUP);
233 * Filter out ioctls we implement directly; forward the rest on to
234 * in_lifaddr_ioctl() and ifp->if_ioctl().
245 * ifra_addr must be present and be of INET family.
246 * ifra_broadaddr and ifra_mask are optional.
248 if (ifra->ifra_addr.sin_len != sizeof(struct sockaddr_in) ||
249 ifra->ifra_addr.sin_family != AF_INET)
251 if (ifra->ifra_broadaddr.sin_len != 0 &&
252 (ifra->ifra_broadaddr.sin_len !=
253 sizeof(struct sockaddr_in) ||
254 ifra->ifra_broadaddr.sin_family != AF_INET))
256 if (ifra->ifra_mask.sin_len != 0 &&
257 (ifra->ifra_mask.sin_len != sizeof(struct sockaddr_in) ||
258 ifra->ifra_mask.sin_family != AF_INET))
265 if (ifr->ifr_addr.sa_family != AF_INET ||
266 ifr->ifr_addr.sa_len != sizeof(struct sockaddr_in))
272 error = priv_check(td, PRIV_NET_ADDIFADDR);
278 return in_lifaddr_ioctl(so, cmd, data, ifp, td);
282 error = priv_check(td, PRIV_NET_DELIFADDR);
288 return in_lifaddr_ioctl(so, cmd, data, ifp, td);
293 return in_lifaddr_ioctl(so, cmd, data, ifp, td);
296 if (ifp == NULL || ifp->if_ioctl == NULL)
298 return ((*ifp->if_ioctl)(ifp, cmd, data));
302 return (EADDRNOTAVAIL);
305 * Security checks before we get involved in any work.
314 error = priv_check(td, PRIV_NET_ADDIFADDR);
322 error = priv_check(td, PRIV_NET_DELIFADDR);
330 * Find address for this interface, if it exists.
332 * If an alias address was specified, find that one instead of the
333 * first one on the interface, if possible.
335 dst = ((struct sockaddr_in *)&ifr->ifr_addr)->sin_addr;
337 LIST_FOREACH(iap, INADDR_HASH(dst.s_addr), ia_hash) {
338 if (iap->ia_ifp == ifp &&
339 iap->ia_addr.sin_addr.s_addr == dst.s_addr) {
340 if (td == NULL || prison_check_ip4(td->td_ucred,
347 ifa_ref(&ia->ia_ifa);
351 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
353 if (iap->ia_addr.sin_family == AF_INET) {
355 prison_check_ip4(td->td_ucred,
356 &iap->ia_addr.sin_addr) != 0)
363 ifa_ref(&ia->ia_ifa);
373 if (ifra->ifra_addr.sin_family == AF_INET) {
374 struct in_ifaddr *oia;
377 for (oia = ia; ia; ia = TAILQ_NEXT(ia, ia_link)) {
378 if (ia->ia_ifp == ifp &&
379 ia->ia_addr.sin_addr.s_addr ==
380 ifra->ifra_addr.sin_addr.s_addr)
383 if (ia != NULL && ia != oia)
384 ifa_ref(&ia->ia_ifa);
385 if (oia != NULL && ia != oia)
386 ifa_free(&oia->ia_ifa);
388 if ((ifp->if_flags & IFF_POINTOPOINT)
389 && (cmd == SIOCAIFADDR)
390 && (ifra->ifra_dstaddr.sin_addr.s_addr
392 error = EDESTADDRREQ;
396 if (cmd == SIOCDIFADDR && ia == NULL) {
397 error = EADDRNOTAVAIL;
405 ia = (struct in_ifaddr *)
406 malloc(sizeof *ia, M_IFADDR, M_NOWAIT |
415 ifa->ifa_addr = (struct sockaddr *)&ia->ia_addr;
416 ifa->ifa_dstaddr = (struct sockaddr *)&ia->ia_dstaddr;
417 ifa->ifa_netmask = (struct sockaddr *)&ia->ia_sockmask;
419 ia->ia_sockmask.sin_len = 8;
420 ia->ia_sockmask.sin_family = AF_INET;
421 if (ifp->if_flags & IFF_BROADCAST) {
422 ia->ia_broadaddr.sin_len = sizeof(ia->ia_addr);
423 ia->ia_broadaddr.sin_family = AF_INET;
427 ifa_ref(ifa); /* if_addrhead */
429 TAILQ_INSERT_TAIL(&ifp->if_addrhead, ifa, ifa_link);
431 ifa_ref(ifa); /* in_ifaddrhead */
433 TAILQ_INSERT_TAIL(&V_in_ifaddrhead, ia, ia_link);
445 error = EADDRNOTAVAIL;
452 * Most paths in this switch return directly or via out. Only paths
453 * that remove the address break in order to hit common removal code.
457 *((struct sockaddr_in *)&ifr->ifr_addr) = ia->ia_addr;
461 if ((ifp->if_flags & IFF_BROADCAST) == 0) {
465 *((struct sockaddr_in *)&ifr->ifr_dstaddr) = ia->ia_broadaddr;
469 if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
473 *((struct sockaddr_in *)&ifr->ifr_dstaddr) = ia->ia_dstaddr;
477 *((struct sockaddr_in *)&ifr->ifr_addr) = ia->ia_sockmask;
481 if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
485 oldaddr = ia->ia_dstaddr;
486 ia->ia_dstaddr = *(struct sockaddr_in *)&ifr->ifr_dstaddr;
487 if (ifp->if_ioctl != NULL) {
488 error = (*ifp->if_ioctl)(ifp, SIOCSIFDSTADDR,
491 ia->ia_dstaddr = oldaddr;
495 if (ia->ia_flags & IFA_ROUTE) {
496 ia->ia_ifa.ifa_dstaddr = (struct sockaddr *)&oldaddr;
497 rtinit(&(ia->ia_ifa), (int)RTM_DELETE, RTF_HOST);
498 ia->ia_ifa.ifa_dstaddr =
499 (struct sockaddr *)&ia->ia_dstaddr;
500 rtinit(&(ia->ia_ifa), (int)RTM_ADD, RTF_HOST|RTF_UP);
505 if ((ifp->if_flags & IFF_BROADCAST) == 0) {
509 ia->ia_broadaddr = *(struct sockaddr_in *)&ifr->ifr_broadaddr;
513 error = in_ifinit(ifp, ia,
514 (struct sockaddr_in *) &ifr->ifr_addr, 1, 0, 0);
515 if (error != 0 && iaIsNew)
518 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
520 (ifp->if_flags & IFF_MULTICAST) != 0) {
521 error = in_joingroup(ifp, &allhosts_addr,
522 NULL, &ii->ii_allhosts);
524 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
530 ia->ia_sockmask.sin_addr = ((struct sockaddr_in *)
531 &ifr->ifr_addr)->sin_addr;
532 ia->ia_subnetmask = ntohl(ia->ia_sockmask.sin_addr.s_addr);
539 if (ifra->ifra_addr.sin_addr.s_addr ==
540 ia->ia_addr.sin_addr.s_addr)
542 if (ifra->ifra_mask.sin_len) {
545 * Need to scrub the prefix here in case
546 * the issued command is SIOCAIFADDR with
547 * the same address, but with a different
548 * prefix length. And if the prefix length
549 * is the same as before, then the call is
550 * un-necessarily executed here.
552 in_ifscrub(ifp, ia, LLE_STATIC);
553 ia->ia_sockmask = ifra->ifra_mask;
554 ia->ia_sockmask.sin_family = AF_INET;
556 ntohl(ia->ia_sockmask.sin_addr.s_addr);
559 if ((ifp->if_flags & IFF_POINTOPOINT) &&
560 (ifra->ifra_dstaddr.sin_family == AF_INET)) {
561 in_ifscrub(ifp, ia, LLE_STATIC);
562 ia->ia_dstaddr = ifra->ifra_dstaddr;
563 maskIsNew = 1; /* We lie; but the effect's the same */
565 if (hostIsNew || maskIsNew)
566 error = in_ifinit(ifp, ia, &ifra->ifra_addr, 0,
567 maskIsNew, ifra->ifra_vhid);
568 if (error != 0 && iaIsNew)
571 if ((ifp->if_flags & IFF_BROADCAST) &&
572 ifra->ifra_broadaddr.sin_len)
573 ia->ia_broadaddr = ifra->ifra_broadaddr;
575 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
577 (ifp->if_flags & IFF_MULTICAST) != 0) {
578 error = in_joingroup(ifp, &allhosts_addr,
579 NULL, &ii->ii_allhosts);
581 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
587 * in_ifscrub kills the interface route.
589 in_ifscrub(ifp, ia, LLE_STATIC);
592 * in_ifadown gets rid of all the rest of
593 * the routes. This is not quite the right
594 * thing to do, but at least if we are running
595 * a routing process they will come back.
597 in_ifadown(&ia->ia_ifa, 1);
598 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
603 panic("in_control: unsupported ioctl");
606 if (ia->ia_ifa.ifa_carp)
607 (*carp_detach_p)(&ia->ia_ifa);
610 /* Re-check that ia is still part of the list. */
611 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
612 if (ifa == &ia->ia_ifa)
617 * If we lost the race with another thread, there is no need to
618 * try it again for the next loop as there is no other exit
619 * path between here and out.
622 error = EADDRNOTAVAIL;
625 TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
627 ifa_free(&ia->ia_ifa); /* if_addrhead */
630 TAILQ_REMOVE(&V_in_ifaddrhead, ia, ia_link);
632 LIST_REMOVE(ia, ia_hash);
635 * If this is the last IPv4 address configured on this
636 * interface, leave the all-hosts group.
637 * No state-change report need be transmitted.
641 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
643 if (ii->ii_allhosts) {
644 (void)in_leavegroup_locked(ii->ii_allhosts, NULL);
645 ii->ii_allhosts = NULL;
649 ifa_free(&iap->ia_ifa);
651 ifa_free(&ia->ia_ifa); /* in_ifaddrhead */
654 ifa_free(&ia->ia_ifa);
660 * SIOCGLIFADDR: get first address. (?!?)
661 * SIOCGLIFADDR with IFLR_PREFIX:
662 * get first address that matches the specified prefix.
663 * SIOCALIFADDR: add the specified address.
664 * SIOCALIFADDR with IFLR_PREFIX:
665 * EINVAL since we can't deduce hostid part of the address.
666 * SIOCDLIFADDR: delete the specified address.
667 * SIOCDLIFADDR with IFLR_PREFIX:
668 * delete the first address that matches the specified prefix.
670 * EINVAL on invalid parameters
671 * EADDRNOTAVAIL on prefix match failed/specified address not found
672 * other values may be returned from in_ioctl()
675 in_lifaddr_ioctl(struct socket *so, u_long cmd, caddr_t data,
676 struct ifnet *ifp, struct thread *td)
678 struct if_laddrreq *iflr = (struct if_laddrreq *)data;
682 if (data == NULL || ifp == NULL) {
683 panic("invalid argument to in_lifaddr_ioctl");
689 /* address must be specified on GET with IFLR_PREFIX */
690 if ((iflr->flags & IFLR_PREFIX) == 0)
695 /* address must be specified on ADD and DELETE */
696 if (iflr->addr.ss_family != AF_INET)
698 if (iflr->addr.ss_len != sizeof(struct sockaddr_in))
700 /* XXX need improvement */
701 if (iflr->dstaddr.ss_family
702 && iflr->dstaddr.ss_family != AF_INET)
704 if (iflr->dstaddr.ss_family
705 && iflr->dstaddr.ss_len != sizeof(struct sockaddr_in))
708 default: /*shouldn't happen*/
711 if (sizeof(struct in_addr) * 8 < iflr->prefixlen)
717 struct in_aliasreq ifra;
719 if (iflr->flags & IFLR_PREFIX)
722 /* copy args to in_aliasreq, perform ioctl(SIOCAIFADDR_IN6). */
723 bzero(&ifra, sizeof(ifra));
724 bcopy(iflr->iflr_name, ifra.ifra_name,
725 sizeof(ifra.ifra_name));
727 bcopy(&iflr->addr, &ifra.ifra_addr, iflr->addr.ss_len);
729 if (iflr->dstaddr.ss_family) { /*XXX*/
730 bcopy(&iflr->dstaddr, &ifra.ifra_dstaddr,
731 iflr->dstaddr.ss_len);
734 ifra.ifra_mask.sin_family = AF_INET;
735 ifra.ifra_mask.sin_len = sizeof(struct sockaddr_in);
736 in_len2mask(&ifra.ifra_mask.sin_addr, iflr->prefixlen);
738 return (in_control(so, SIOCAIFADDR, (caddr_t)&ifra, ifp, td));
743 struct in_ifaddr *ia;
744 struct in_addr mask, candidate, match;
745 struct sockaddr_in *sin;
747 bzero(&mask, sizeof(mask));
748 bzero(&match, sizeof(match));
749 if (iflr->flags & IFLR_PREFIX) {
750 /* lookup a prefix rather than address. */
751 in_len2mask(&mask, iflr->prefixlen);
753 sin = (struct sockaddr_in *)&iflr->addr;
754 match.s_addr = sin->sin_addr.s_addr;
755 match.s_addr &= mask.s_addr;
757 /* if you set extra bits, that's wrong */
758 if (match.s_addr != sin->sin_addr.s_addr)
762 /* on getting an address, take the 1st match */
763 /* on deleting an address, do exact match */
764 if (cmd != SIOCGLIFADDR) {
765 in_len2mask(&mask, 32);
766 sin = (struct sockaddr_in *)&iflr->addr;
767 match.s_addr = sin->sin_addr.s_addr;
771 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
772 if (ifa->ifa_addr->sa_family != AF_INET6)
774 if (match.s_addr == 0)
776 candidate.s_addr = ((struct sockaddr_in *)&ifa->ifa_addr)->sin_addr.s_addr;
777 candidate.s_addr &= mask.s_addr;
778 if (candidate.s_addr == match.s_addr)
782 return (EADDRNOTAVAIL);
783 ia = (struct in_ifaddr *)ifa;
785 if (cmd == SIOCGLIFADDR) {
786 /* fill in the if_laddrreq structure */
787 bcopy(&ia->ia_addr, &iflr->addr, ia->ia_addr.sin_len);
789 if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
790 bcopy(&ia->ia_dstaddr, &iflr->dstaddr,
791 ia->ia_dstaddr.sin_len);
793 bzero(&iflr->dstaddr, sizeof(iflr->dstaddr));
796 in_mask2len(&ia->ia_sockmask.sin_addr);
798 iflr->flags = 0; /*XXX*/
802 struct in_aliasreq ifra;
804 /* fill in_aliasreq and do ioctl(SIOCDIFADDR_IN6) */
805 bzero(&ifra, sizeof(ifra));
806 bcopy(iflr->iflr_name, ifra.ifra_name,
807 sizeof(ifra.ifra_name));
809 bcopy(&ia->ia_addr, &ifra.ifra_addr,
810 ia->ia_addr.sin_len);
811 if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
812 bcopy(&ia->ia_dstaddr, &ifra.ifra_dstaddr,
813 ia->ia_dstaddr.sin_len);
815 bcopy(&ia->ia_sockmask, &ifra.ifra_dstaddr,
816 ia->ia_sockmask.sin_len);
818 return (in_control(so, SIOCDIFADDR, (caddr_t)&ifra,
824 return (EOPNOTSUPP); /*just for safety*/
828 * Delete any existing route for an interface.
831 in_ifscrub(struct ifnet *ifp, struct in_ifaddr *ia, u_int flags)
834 in_scrubprefix(ia, flags);
838 * Initialize an interface's internet address
839 * and routing table entry.
842 in_ifinit(struct ifnet *ifp, struct in_ifaddr *ia, struct sockaddr_in *sin,
843 int scrub, int masksupplied, int vhid)
845 register u_long i = ntohl(sin->sin_addr.s_addr);
846 int flags = RTF_UP, error = 0;
849 in_scrubprefix(ia, LLE_STATIC);
852 if (ia->ia_addr.sin_family == AF_INET)
853 LIST_REMOVE(ia, ia_hash);
855 LIST_INSERT_HEAD(INADDR_HASH(ia->ia_addr.sin_addr.s_addr),
860 if (carp_attach_p != NULL)
861 error = (*carp_attach_p)(&ia->ia_ifa, vhid);
863 error = EPROTONOSUPPORT;
869 * Give the interface a chance to initialize
870 * if this is its first address,
871 * and to validate the address if necessary.
873 if (ifp->if_ioctl != NULL &&
874 (error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia)) != 0)
875 /* LIST_REMOVE(ia, ia_hash) is done in in_control */
879 * Be compatible with network classes, if netmask isn't supplied,
880 * guess it based on classes.
884 ia->ia_subnetmask = IN_CLASSA_NET;
885 else if (IN_CLASSB(i))
886 ia->ia_subnetmask = IN_CLASSB_NET;
888 ia->ia_subnetmask = IN_CLASSC_NET;
889 ia->ia_sockmask.sin_addr.s_addr = htonl(ia->ia_subnetmask);
891 ia->ia_subnet = i & ia->ia_subnetmask;
892 in_socktrim(&ia->ia_sockmask);
894 * Add route for the network.
896 ia->ia_ifa.ifa_metric = ifp->if_metric;
897 if (ifp->if_flags & IFF_BROADCAST) {
898 if (ia->ia_subnetmask == IN_RFC3021_MASK)
899 ia->ia_broadaddr.sin_addr.s_addr = INADDR_BROADCAST;
901 ia->ia_broadaddr.sin_addr.s_addr =
902 htonl(ia->ia_subnet | ~ia->ia_subnetmask);
903 } else if (ifp->if_flags & IFF_LOOPBACK) {
904 ia->ia_dstaddr = ia->ia_addr;
906 } else if (ifp->if_flags & IFF_POINTOPOINT) {
907 if (ia->ia_dstaddr.sin_family != AF_INET)
911 if (!vhid && (error = in_addprefix(ia, flags)) != 0)
914 if (ia->ia_addr.sin_addr.s_addr == INADDR_ANY)
917 if (ifp->if_flags & IFF_POINTOPOINT &&
918 ia->ia_dstaddr.sin_addr.s_addr == ia->ia_addr.sin_addr.s_addr)
922 * add a loopback route to self
924 if (V_useloopback && !vhid && !(ifp->if_flags & IFF_LOOPBACK)) {
927 bzero(&ia_ro, sizeof(ia_ro));
928 *((struct sockaddr_in *)(&ia_ro.ro_dst)) = ia->ia_addr;
929 rtalloc_ign_fib(&ia_ro, 0, 0);
930 if ((ia_ro.ro_rt != NULL) && (ia_ro.ro_rt->rt_ifp != NULL) &&
931 (ia_ro.ro_rt->rt_ifp == V_loif)) {
932 RT_LOCK(ia_ro.ro_rt);
933 RT_ADDREF(ia_ro.ro_rt);
934 RTFREE_LOCKED(ia_ro.ro_rt);
936 error = ifa_add_loopback_route((struct ifaddr *)ia,
937 (struct sockaddr *)&ia->ia_addr);
939 ia->ia_flags |= IFA_RTSELF;
940 if (ia_ro.ro_rt != NULL)
947 #define rtinitflags(x) \
948 ((((x)->ia_ifp->if_flags & (IFF_LOOPBACK | IFF_POINTOPOINT)) != 0) \
952 * Generate a routing message when inserting or deleting
953 * an interface address alias.
955 static void in_addralias_rtmsg(int cmd, struct in_addr *prefix,
956 struct in_ifaddr *target)
959 struct sockaddr_in *pfx_addr;
960 struct rtentry msg_rt;
963 * This is a bit questionable because there is no
964 * additional route entry added/deleted for an address
965 * alias. Therefore this route report is inaccurate.
967 bzero(&pfx_ro, sizeof(pfx_ro));
968 pfx_addr = (struct sockaddr_in *)(&pfx_ro.ro_dst);
969 pfx_addr->sin_len = sizeof(*pfx_addr);
970 pfx_addr->sin_family = AF_INET;
971 pfx_addr->sin_addr = *prefix;
972 rtalloc_ign_fib(&pfx_ro, 0, 0);
973 if (pfx_ro.ro_rt != NULL) {
974 msg_rt = *pfx_ro.ro_rt;
977 * Point the gateway to the new interface
978 * address as if a new prefix route entry has
979 * been added through the new address alias.
980 * All other parts of the rtentry is accurate,
981 * e.g., rt_key, rt_mask, rt_ifp etc.
984 (struct sockaddr *)&target->ia_addr;
986 (struct ifaddr *)target,
988 RTFREE(pfx_ro.ro_rt);
994 * Check if we have a route for the given prefix already or add one accordingly.
997 in_addprefix(struct in_ifaddr *target, int flags)
999 struct in_ifaddr *ia;
1000 struct in_addr prefix, mask, p, m;
1003 if ((flags & RTF_HOST) != 0) {
1004 prefix = target->ia_dstaddr.sin_addr;
1007 prefix = target->ia_addr.sin_addr;
1008 mask = target->ia_sockmask.sin_addr;
1009 prefix.s_addr &= mask.s_addr;
1013 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
1014 if (rtinitflags(ia)) {
1015 p = ia->ia_dstaddr.sin_addr;
1017 if (prefix.s_addr != p.s_addr)
1020 p = ia->ia_addr.sin_addr;
1021 m = ia->ia_sockmask.sin_addr;
1022 p.s_addr &= m.s_addr;
1024 if (prefix.s_addr != p.s_addr ||
1025 mask.s_addr != m.s_addr)
1030 * If we got a matching prefix route inserted by other
1031 * interface address, we are done here.
1033 if (ia->ia_flags & IFA_ROUTE) {
1035 if (ia->ia_addr.sin_addr.s_addr ==
1036 target->ia_addr.sin_addr.s_addr) {
1037 IN_IFADDR_RUNLOCK();
1042 if (V_nosameprefix) {
1043 IN_IFADDR_RUNLOCK();
1046 in_addralias_rtmsg(RTM_ADD, &prefix, target);
1047 IN_IFADDR_RUNLOCK();
1052 IN_IFADDR_RUNLOCK();
1055 * No-one seem to have this prefix route, so we try to insert it.
1057 error = rtinit(&target->ia_ifa, (int)RTM_ADD, flags);
1059 target->ia_flags |= IFA_ROUTE;
1064 * If there is no other address in the system that can serve a route to the
1065 * same prefix, remove the route. Hand over the route to the new address
1069 in_scrubprefix(struct in_ifaddr *target, u_int flags)
1071 struct in_ifaddr *ia;
1072 struct in_addr prefix, mask, p, m;
1074 struct sockaddr_in prefix0, mask0;
1077 * Remove the loopback route to the interface address.
1078 * The "useloopback" setting is not consulted because if the
1079 * user configures an interface address, turns off this
1080 * setting, and then tries to delete that interface address,
1081 * checking the current setting of "useloopback" would leave
1082 * that interface address loopback route untouched, which
1083 * would be wrong. Therefore the interface address loopback route
1084 * deletion is unconditional.
1086 if ((target->ia_addr.sin_addr.s_addr != INADDR_ANY) &&
1087 !(target->ia_ifp->if_flags & IFF_LOOPBACK) &&
1088 (target->ia_flags & IFA_RTSELF)) {
1092 bzero(&ia_ro, sizeof(ia_ro));
1093 *((struct sockaddr_in *)(&ia_ro.ro_dst)) = target->ia_addr;
1094 rtalloc_ign_fib(&ia_ro, 0, 0);
1095 if ((ia_ro.ro_rt != NULL) && (ia_ro.ro_rt->rt_ifp != NULL) &&
1096 (ia_ro.ro_rt->rt_ifp == V_loif)) {
1097 RT_LOCK(ia_ro.ro_rt);
1098 if (ia_ro.ro_rt->rt_refcnt <= 1)
1100 else if (flags & LLE_STATIC) {
1101 RT_REMREF(ia_ro.ro_rt);
1102 target->ia_flags &= ~IFA_RTSELF;
1104 RTFREE_LOCKED(ia_ro.ro_rt);
1106 if (freeit && (flags & LLE_STATIC)) {
1107 error = ifa_del_loopback_route((struct ifaddr *)target,
1108 (struct sockaddr *)&target->ia_addr);
1110 target->ia_flags &= ~IFA_RTSELF;
1112 if ((flags & LLE_STATIC) &&
1113 !(target->ia_ifp->if_flags & IFF_NOARP))
1114 /* remove arp cache */
1115 arp_ifscrub(target->ia_ifp, IA_SIN(target)->sin_addr.s_addr);
1118 if (rtinitflags(target)) {
1119 prefix = target->ia_dstaddr.sin_addr;
1122 prefix = target->ia_addr.sin_addr;
1123 mask = target->ia_sockmask.sin_addr;
1124 prefix.s_addr &= mask.s_addr;
1127 if ((target->ia_flags & IFA_ROUTE) == 0) {
1128 in_addralias_rtmsg(RTM_DELETE, &prefix, target);
1133 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
1134 if (rtinitflags(ia)) {
1135 p = ia->ia_dstaddr.sin_addr;
1137 if (prefix.s_addr != p.s_addr)
1140 p = ia->ia_addr.sin_addr;
1141 m = ia->ia_sockmask.sin_addr;
1142 p.s_addr &= m.s_addr;
1144 if (prefix.s_addr != p.s_addr ||
1145 mask.s_addr != m.s_addr)
1149 if ((ia->ia_ifp->if_flags & IFF_UP) == 0)
1153 * If we got a matching prefix address, move IFA_ROUTE and
1154 * the route itself to it. Make sure that routing daemons
1157 if ((ia->ia_flags & IFA_ROUTE) == 0) {
1158 ifa_ref(&ia->ia_ifa);
1159 IN_IFADDR_RUNLOCK();
1160 error = rtinit(&(target->ia_ifa), (int)RTM_DELETE,
1161 rtinitflags(target));
1163 target->ia_flags &= ~IFA_ROUTE;
1165 log(LOG_INFO, "in_scrubprefix: err=%d, old prefix delete failed\n",
1167 error = rtinit(&ia->ia_ifa, (int)RTM_ADD,
1168 rtinitflags(ia) | RTF_UP);
1170 ia->ia_flags |= IFA_ROUTE;
1172 log(LOG_INFO, "in_scrubprefix: err=%d, new prefix add failed\n",
1174 ifa_free(&ia->ia_ifa);
1178 IN_IFADDR_RUNLOCK();
1181 * remove all L2 entries on the given prefix
1183 bzero(&prefix0, sizeof(prefix0));
1184 prefix0.sin_len = sizeof(prefix0);
1185 prefix0.sin_family = AF_INET;
1186 prefix0.sin_addr.s_addr = target->ia_subnet;
1187 bzero(&mask0, sizeof(mask0));
1188 mask0.sin_len = sizeof(mask0);
1189 mask0.sin_family = AF_INET;
1190 mask0.sin_addr.s_addr = target->ia_subnetmask;
1191 lltable_prefix_free(AF_INET, (struct sockaddr *)&prefix0,
1192 (struct sockaddr *)&mask0, flags);
1195 * As no-one seem to have this prefix, we can remove the route.
1197 error = rtinit(&(target->ia_ifa), (int)RTM_DELETE, rtinitflags(target));
1199 target->ia_flags &= ~IFA_ROUTE;
1201 log(LOG_INFO, "in_scrubprefix: err=%d, prefix delete failed\n", error);
1208 * Return 1 if the address might be a local broadcast address.
1211 in_broadcast(struct in_addr in, struct ifnet *ifp)
1213 register struct ifaddr *ifa;
1216 if (in.s_addr == INADDR_BROADCAST ||
1217 in.s_addr == INADDR_ANY)
1219 if ((ifp->if_flags & IFF_BROADCAST) == 0)
1221 t = ntohl(in.s_addr);
1223 * Look through the list of addresses for a match
1224 * with a broadcast address.
1226 #define ia ((struct in_ifaddr *)ifa)
1227 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
1228 if (ifa->ifa_addr->sa_family == AF_INET &&
1229 (in.s_addr == ia->ia_broadaddr.sin_addr.s_addr ||
1231 * Check for old-style (host 0) broadcast, but
1232 * taking into account that RFC 3021 obsoletes it.
1234 (ia->ia_subnetmask != IN_RFC3021_MASK &&
1235 t == ia->ia_subnet)) &&
1237 * Check for an all one subnetmask. These
1238 * only exist when an interface gets a secondary
1241 ia->ia_subnetmask != (u_long)0xffffffff)
1248 * On interface removal, clean up IPv4 data structures hung off of the ifnet.
1251 in_ifdetach(struct ifnet *ifp)
1254 in_pcbpurgeif0(&V_ripcbinfo, ifp);
1255 in_pcbpurgeif0(&V_udbinfo, ifp);
1256 in_purgemaddrs(ifp);
1260 * Delete all IPv4 multicast address records, and associated link-layer
1261 * multicast address records, associated with ifp.
1262 * XXX It looks like domifdetach runs AFTER the link layer cleanup.
1263 * XXX This should not race with ifma_protospec being set during
1264 * a new allocation, if it does, we have bigger problems.
1267 in_purgemaddrs(struct ifnet *ifp)
1269 LIST_HEAD(,in_multi) purgeinms;
1270 struct in_multi *inm, *tinm;
1271 struct ifmultiaddr *ifma;
1273 LIST_INIT(&purgeinms);
1277 * Extract list of in_multi associated with the detaching ifp
1278 * which the PF_INET layer is about to release.
1279 * We need to do this as IF_ADDR_LOCK() may be re-acquired
1280 * by code further down.
1283 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
1284 if (ifma->ifma_addr->sa_family != AF_INET ||
1285 ifma->ifma_protospec == NULL)
1288 KASSERT(ifma->ifma_protospec != NULL,
1289 ("%s: ifma_protospec is NULL", __func__));
1291 inm = (struct in_multi *)ifma->ifma_protospec;
1292 LIST_INSERT_HEAD(&purgeinms, inm, inm_link);
1294 IF_ADDR_UNLOCK(ifp);
1296 LIST_FOREACH_SAFE(inm, &purgeinms, inm_link, tinm) {
1297 LIST_REMOVE(inm, inm_link);
1298 inm_release_locked(inm);
1306 struct llentry base;
1307 struct sockaddr_in l3_addr4;
1310 static struct llentry *
1311 in_lltable_new(const struct sockaddr *l3addr, u_int flags)
1313 struct in_llentry *lle;
1315 lle = malloc(sizeof(struct in_llentry), M_LLTABLE, M_DONTWAIT | M_ZERO);
1316 if (lle == NULL) /* NB: caller generates msg */
1319 callout_init(&lle->base.la_timer, CALLOUT_MPSAFE);
1321 * For IPv4 this will trigger "arpresolve" to generate
1324 lle->base.la_expire = time_uptime; /* mark expired */
1325 lle->l3_addr4 = *(const struct sockaddr_in *)l3addr;
1326 lle->base.lle_refcnt = 1;
1327 LLE_LOCK_INIT(&lle->base);
1332 * Deletes an address from the address table.
1333 * This function is called by the timer functions
1334 * such as arptimer() and nd6_llinfo_timer(), and
1335 * the caller does the locking.
1338 in_lltable_free(struct lltable *llt, struct llentry *lle)
1341 LLE_LOCK_DESTROY(lle);
1342 free(lle, M_LLTABLE);
1346 #define IN_ARE_MASKED_ADDR_EQUAL(d, a, m) ( \
1347 (((ntohl((d)->sin_addr.s_addr) ^ (a)->sin_addr.s_addr) & (m)->sin_addr.s_addr)) == 0 )
1350 in_lltable_prefix_free(struct lltable *llt,
1351 const struct sockaddr *prefix,
1352 const struct sockaddr *mask,
1355 const struct sockaddr_in *pfx = (const struct sockaddr_in *)prefix;
1356 const struct sockaddr_in *msk = (const struct sockaddr_in *)mask;
1357 struct llentry *lle, *next;
1359 size_t pkts_dropped;
1361 for (i=0; i < LLTBL_HASHTBL_SIZE; i++) {
1362 LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) {
1365 * (flags & LLE_STATIC) means deleting all entries
1366 * including static ARP entries
1368 if (IN_ARE_MASKED_ADDR_EQUAL((struct sockaddr_in *)L3_ADDR(lle),
1370 ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))) {
1373 canceled = callout_drain(&lle->la_timer);
1377 pkts_dropped = llentry_free(lle);
1378 ARPSTAT_ADD(dropped, pkts_dropped);
1386 in_lltable_rtcheck(struct ifnet *ifp, u_int flags, const struct sockaddr *l3addr)
1390 KASSERT(l3addr->sa_family == AF_INET,
1391 ("sin_family %d", l3addr->sa_family));
1393 /* XXX rtalloc1 should take a const param */
1394 rt = rtalloc1(__DECONST(struct sockaddr *, l3addr), 0, 0);
1400 * If the gateway for an existing host route matches the target L3
1401 * address, which is a special route inserted by some implementation
1402 * such as MANET, and the interface is of the correct type, then
1403 * allow for ARP to proceed.
1405 if (rt->rt_flags & RTF_GATEWAY) {
1406 if (!(rt->rt_flags & RTF_HOST) || !rt->rt_ifp ||
1407 rt->rt_ifp->if_type != IFT_ETHER ||
1408 (rt->rt_ifp->if_flags &
1409 (IFF_NOARP | IFF_STATICARP)) != 0 ||
1410 memcmp(rt->rt_gateway->sa_data, l3addr->sa_data,
1411 sizeof(in_addr_t)) != 0) {
1418 * Make sure that at least the destination address is covered
1419 * by the route. This is for handling the case where 2 or more
1420 * interfaces have the same prefix. An incoming packet arrives
1421 * on one interface and the corresponding outgoing packet leaves
1422 * another interface.
1424 if (!(rt->rt_flags & RTF_HOST) && rt->rt_ifp != ifp) {
1425 const char *sa, *mask, *addr, *lim;
1428 mask = (const char *)rt_mask(rt);
1430 * Just being extra cautious to avoid some custom
1431 * code getting into trouble.
1438 sa = (const char *)rt_key(rt);
1439 addr = (const char *)l3addr;
1440 len = ((const struct sockaddr_in *)l3addr)->sin_len;
1443 for ( ; addr < lim; sa++, mask++, addr++) {
1444 if ((*sa ^ *addr) & *mask) {
1446 log(LOG_INFO, "IPv4 address: \"%s\" is not on the network\n",
1447 inet_ntoa(((const struct sockaddr_in *)l3addr)->sin_addr));
1460 * Return NULL if not found or marked for deletion.
1461 * If found return lle read locked.
1463 static struct llentry *
1464 in_lltable_lookup(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
1466 const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
1467 struct ifnet *ifp = llt->llt_ifp;
1468 struct llentry *lle;
1469 struct llentries *lleh;
1472 IF_AFDATA_LOCK_ASSERT(ifp);
1473 KASSERT(l3addr->sa_family == AF_INET,
1474 ("sin_family %d", l3addr->sa_family));
1476 hashkey = sin->sin_addr.s_addr;
1477 lleh = &llt->lle_head[LLATBL_HASH(hashkey, LLTBL_HASHMASK)];
1478 LIST_FOREACH(lle, lleh, lle_next) {
1479 struct sockaddr_in *sa2 = (struct sockaddr_in *)L3_ADDR(lle);
1480 if (lle->la_flags & LLE_DELETED)
1482 if (sa2->sin_addr.s_addr == sin->sin_addr.s_addr)
1487 if (flags & LLE_DELETE)
1488 log(LOG_INFO, "interface address is missing from cache = %p in delete\n", lle);
1490 if (!(flags & LLE_CREATE))
1493 * A route that covers the given address must have
1494 * been installed 1st because we are doing a resolution,
1497 if (!(flags & LLE_IFADDR) &&
1498 in_lltable_rtcheck(ifp, flags, l3addr) != 0)
1501 lle = in_lltable_new(l3addr, flags);
1503 log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
1506 lle->la_flags = flags & ~LLE_CREATE;
1507 if ((flags & (LLE_CREATE | LLE_IFADDR)) == (LLE_CREATE | LLE_IFADDR)) {
1508 bcopy(IF_LLADDR(ifp), &lle->ll_addr, ifp->if_addrlen);
1509 lle->la_flags |= (LLE_VALID | LLE_STATIC);
1513 lle->lle_head = lleh;
1514 LIST_INSERT_HEAD(lleh, lle, lle_next);
1515 } else if (flags & LLE_DELETE) {
1516 if (!(lle->la_flags & LLE_IFADDR) || (flags & LLE_IFADDR)) {
1518 lle->la_flags = LLE_DELETED;
1519 EVENTHANDLER_INVOKE(arp_update_event, lle);
1522 log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
1528 if (LLE_IS_VALID(lle)) {
1529 if (flags & LLE_EXCLUSIVE)
1539 in_lltable_dump(struct lltable *llt, struct sysctl_req *wr)
1541 #define SIN(lle) ((struct sockaddr_in *) L3_ADDR(lle))
1542 struct ifnet *ifp = llt->llt_ifp;
1543 struct llentry *lle;
1546 struct rt_msghdr rtm;
1547 struct sockaddr_inarp sin;
1548 struct sockaddr_dl sdl;
1552 LLTABLE_LOCK_ASSERT();
1555 for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) {
1556 LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
1557 struct sockaddr_dl *sdl;
1559 /* skip deleted entries */
1560 if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
1562 /* Skip if jailed and not a valid IP of the prison. */
1563 if (prison_if(wr->td->td_ucred, L3_ADDR(lle)) != 0)
1566 * produce a msg made of:
1568 * struct sockaddr_inarp; (IPv4)
1569 * struct sockaddr_dl;
1571 bzero(&arpc, sizeof(arpc));
1572 arpc.rtm.rtm_msglen = sizeof(arpc);
1573 arpc.rtm.rtm_version = RTM_VERSION;
1574 arpc.rtm.rtm_type = RTM_GET;
1575 arpc.rtm.rtm_flags = RTF_UP;
1576 arpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
1577 arpc.sin.sin_family = AF_INET;
1578 arpc.sin.sin_len = sizeof(arpc.sin);
1579 arpc.sin.sin_addr.s_addr = SIN(lle)->sin_addr.s_addr;
1582 if (lle->la_flags & LLE_PUB) {
1583 arpc.rtm.rtm_flags |= RTF_ANNOUNCE;
1585 if (lle->la_flags & LLE_PROXY)
1586 arpc.sin.sin_other = SIN_PROXY;
1590 sdl->sdl_family = AF_LINK;
1591 sdl->sdl_len = sizeof(*sdl);
1592 sdl->sdl_index = ifp->if_index;
1593 sdl->sdl_type = ifp->if_type;
1594 if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
1595 sdl->sdl_alen = ifp->if_addrlen;
1596 bcopy(&lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
1599 bzero(LLADDR(sdl), ifp->if_addrlen);
1602 arpc.rtm.rtm_rmx.rmx_expire =
1603 lle->la_flags & LLE_STATIC ? 0 : lle->la_expire;
1604 arpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
1605 if (lle->la_flags & LLE_STATIC)
1606 arpc.rtm.rtm_flags |= RTF_STATIC;
1607 arpc.rtm.rtm_index = ifp->if_index;
1608 error = SYSCTL_OUT(wr, &arpc, sizeof(arpc));
1618 in_domifattach(struct ifnet *ifp)
1620 struct in_ifinfo *ii;
1621 struct lltable *llt;
1623 ii = malloc(sizeof(struct in_ifinfo), M_IFADDR, M_WAITOK|M_ZERO);
1625 llt = lltable_init(ifp, AF_INET);
1627 llt->llt_free = in_lltable_free;
1628 llt->llt_prefix_free = in_lltable_prefix_free;
1629 llt->llt_lookup = in_lltable_lookup;
1630 llt->llt_dump = in_lltable_dump;
1634 ii->ii_igmp = igmp_domifattach(ifp);
1640 in_domifdetach(struct ifnet *ifp, void *aux)
1642 struct in_ifinfo *ii = (struct in_ifinfo *)aux;
1644 igmp_domifdetach(ifp);
1645 lltable_free(ii->ii_llt);