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/eventhandler.h>
40 #include <sys/systm.h>
41 #include <sys/sockio.h>
42 #include <sys/malloc.h>
44 #include <sys/socket.h>
46 #include <sys/kernel.h>
48 #include <sys/sysctl.h>
49 #include <sys/syslog.h>
53 #include <net/if_var.h>
54 #include <net/if_arp.h>
55 #include <net/if_dl.h>
56 #include <net/if_llatbl.h>
57 #include <net/if_types.h>
58 #include <net/route.h>
61 #include <netinet/if_ether.h>
62 #include <netinet/in.h>
63 #include <netinet/in_var.h>
64 #include <netinet/in_pcb.h>
65 #include <netinet/ip_var.h>
66 #include <netinet/ip_carp.h>
67 #include <netinet/igmp_var.h>
68 #include <netinet/udp.h>
69 #include <netinet/udp_var.h>
71 static int in_mask2len(struct in_addr *);
72 static void in_len2mask(struct in_addr *, int);
73 static int in_lifaddr_ioctl(struct socket *, u_long, caddr_t,
74 struct ifnet *, struct thread *);
75 static int in_aifaddr_ioctl(caddr_t, struct ifnet *, struct thread *);
76 static int in_difaddr_ioctl(caddr_t, struct ifnet *, struct thread *);
78 static void in_socktrim(struct sockaddr_in *);
79 static void in_purgemaddrs(struct ifnet *);
81 static VNET_DEFINE(int, nosameprefix);
82 #define V_nosameprefix VNET(nosameprefix)
83 SYSCTL_VNET_INT(_net_inet_ip, OID_AUTO, no_same_prefix, CTLFLAG_RW,
84 &VNET_NAME(nosameprefix), 0,
85 "Refuse to create same prefixes on different interfaces");
87 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
88 #define V_ripcbinfo VNET(ripcbinfo)
90 static struct sx in_control_sx;
91 SX_SYSINIT(in_control_sx, &in_control_sx, "in_control");
94 * Return 1 if an internet address is for a ``local'' host
95 * (one to which we have a connection).
98 in_localaddr(struct in_addr in)
100 register u_long i = ntohl(in.s_addr);
101 register struct in_ifaddr *ia;
104 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
105 if ((i & ia->ia_subnetmask) == ia->ia_subnet) {
115 * Return 1 if an internet address is for the local host and configured
116 * on one of its interfaces.
119 in_localip(struct in_addr in)
121 struct in_ifaddr *ia;
124 LIST_FOREACH(ia, INADDR_HASH(in.s_addr), ia_hash) {
125 if (IA_SIN(ia)->sin_addr.s_addr == in.s_addr) {
135 * Return a reference to the interface address which is different to
136 * the supplied one but with same IP address value.
138 static struct in_ifaddr *
139 in_localip_more(struct in_ifaddr *ia)
141 in_addr_t in = IA_SIN(ia)->sin_addr.s_addr;
142 struct in_ifaddr *it;
145 LIST_FOREACH(it, INADDR_HASH(in), ia_hash) {
146 if (it != ia && IA_SIN(it)->sin_addr.s_addr == in) {
147 ifa_ref(&it->ia_ifa);
158 * Determine whether an IP address is in a reserved set of addresses
159 * that may not be forwarded, or whether datagrams to that destination
163 in_canforward(struct in_addr in)
165 register u_long i = ntohl(in.s_addr);
168 if (IN_EXPERIMENTAL(i) || IN_MULTICAST(i) || IN_LINKLOCAL(i))
171 net = i & IN_CLASSA_NET;
172 if (net == 0 || net == (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))
179 * Trim a mask in a sockaddr
182 in_socktrim(struct sockaddr_in *ap)
184 register char *cplim = (char *) &ap->sin_addr;
185 register char *cp = (char *) (&ap->sin_addr + 1);
188 while (--cp >= cplim)
190 (ap)->sin_len = cp - (char *) (ap) + 1;
197 struct in_addr *mask;
203 for (x = 0; x < sizeof(*mask); x++) {
208 if (x < sizeof(*mask)) {
209 for (y = 0; y < 8; y++) {
210 if ((p[x] & (0x80 >> y)) == 0)
218 in_len2mask(struct in_addr *mask, int len)
224 bzero(mask, sizeof(*mask));
225 for (i = 0; i < len / 8; i++)
228 p[i] = (0xff00 >> (len % 8)) & 0xff;
232 * Generic internet control operations (ioctl's).
235 in_control(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp,
238 struct ifreq *ifr = (struct ifreq *)data;
239 struct sockaddr_in *addr = (struct sockaddr_in *)&ifr->ifr_addr;
240 struct in_ifaddr *ia;
244 return (EADDRNOTAVAIL);
247 * Filter out 4 ioctls we implement directly. Forward the rest
248 * to specific functions and ifp->if_ioctl().
257 sx_xlock(&in_control_sx);
258 error = in_difaddr_ioctl(data, ifp, td);
259 sx_xunlock(&in_control_sx);
262 sx_xlock(&in_control_sx);
263 error = in_aifaddr_ioctl(data, ifp, td);
264 sx_xunlock(&in_control_sx);
269 return (in_lifaddr_ioctl(so, cmd, data, ifp, td));
274 /* We no longer support that old commands. */
277 if (ifp->if_ioctl == NULL)
279 return ((*ifp->if_ioctl)(ifp, cmd, data));
283 * Find address for this interface, if it exists.
286 LIST_FOREACH(ia, INADDR_HASH(addr->sin_addr.s_addr), ia_hash) {
287 if (ia->ia_ifp == ifp &&
288 ia->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
289 prison_check_ip4(td->td_ucred, &addr->sin_addr) == 0)
295 return (EADDRNOTAVAIL);
305 if ((ifp->if_flags & IFF_BROADCAST) == 0) {
309 *addr = ia->ia_broadaddr;
313 if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
317 *addr = ia->ia_dstaddr;
321 *addr = ia->ia_sockmask;
331 in_aifaddr_ioctl(caddr_t data, struct ifnet *ifp, struct thread *td)
333 const struct in_aliasreq *ifra = (struct in_aliasreq *)data;
334 const struct sockaddr_in *addr = &ifra->ifra_addr;
335 const struct sockaddr_in *broadaddr = &ifra->ifra_broadaddr;
336 const struct sockaddr_in *mask = &ifra->ifra_mask;
337 const struct sockaddr_in *dstaddr = &ifra->ifra_dstaddr;
338 const int vhid = ifra->ifra_vhid;
340 struct in_ifaddr *ia;
344 error = priv_check(td, PRIV_NET_ADDIFADDR);
349 * ifra_addr must be present and be of INET family.
350 * ifra_broadaddr/ifra_dstaddr and ifra_mask are optional.
352 if (addr->sin_len != sizeof(struct sockaddr_in) ||
353 addr->sin_family != AF_INET)
355 if (broadaddr->sin_len != 0 &&
356 (broadaddr->sin_len != sizeof(struct sockaddr_in) ||
357 broadaddr->sin_family != AF_INET))
359 if (mask->sin_len != 0 &&
360 (mask->sin_len != sizeof(struct sockaddr_in) ||
361 mask->sin_family != AF_INET))
363 if ((ifp->if_flags & IFF_POINTOPOINT) &&
364 (dstaddr->sin_len != sizeof(struct sockaddr_in) ||
365 dstaddr->sin_addr.s_addr == INADDR_ANY))
366 return (EDESTADDRREQ);
367 if (vhid > 0 && carp_attach_p == NULL)
368 return (EPROTONOSUPPORT);
371 * See whether address already exist.
376 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
377 struct in_ifaddr *it = ifatoia(ifa);
379 if (it->ia_addr.sin_family != AF_INET)
383 if (it->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
384 prison_check_ip4(td->td_ucred, &addr->sin_addr) == 0)
387 IF_ADDR_RUNLOCK(ifp);
390 (void )in_difaddr_ioctl(data, ifp, td);
392 ifa = ifa_alloc(sizeof(struct in_ifaddr), M_WAITOK);
393 ia = (struct in_ifaddr *)ifa;
394 ifa->ifa_addr = (struct sockaddr *)&ia->ia_addr;
395 ifa->ifa_dstaddr = (struct sockaddr *)&ia->ia_dstaddr;
396 ifa->ifa_netmask = (struct sockaddr *)&ia->ia_sockmask;
399 ia->ia_ifa.ifa_metric = ifp->if_metric;
401 if (mask->sin_len != 0) {
402 ia->ia_sockmask = *mask;
403 ia->ia_subnetmask = ntohl(ia->ia_sockmask.sin_addr.s_addr);
405 in_addr_t i = ntohl(addr->sin_addr.s_addr);
408 * Be compatible with network classes, if netmask isn't
409 * supplied, guess it based on classes.
412 ia->ia_subnetmask = IN_CLASSA_NET;
413 else if (IN_CLASSB(i))
414 ia->ia_subnetmask = IN_CLASSB_NET;
416 ia->ia_subnetmask = IN_CLASSC_NET;
417 ia->ia_sockmask.sin_addr.s_addr = htonl(ia->ia_subnetmask);
419 ia->ia_subnet = ntohl(addr->sin_addr.s_addr) & ia->ia_subnetmask;
420 in_socktrim(&ia->ia_sockmask);
422 if (ifp->if_flags & IFF_BROADCAST) {
423 if (broadaddr->sin_len != 0) {
424 ia->ia_broadaddr = *broadaddr;
425 } else if (ia->ia_subnetmask == IN_RFC3021_MASK) {
426 ia->ia_broadaddr.sin_addr.s_addr = INADDR_BROADCAST;
427 ia->ia_broadaddr.sin_len = sizeof(struct sockaddr_in);
428 ia->ia_broadaddr.sin_family = AF_INET;
430 ia->ia_broadaddr.sin_addr.s_addr =
431 htonl(ia->ia_subnet | ~ia->ia_subnetmask);
432 ia->ia_broadaddr.sin_len = sizeof(struct sockaddr_in);
433 ia->ia_broadaddr.sin_family = AF_INET;
437 if (ifp->if_flags & IFF_POINTOPOINT)
438 ia->ia_dstaddr = *dstaddr;
440 /* XXXGL: rtinit() needs this strange assignment. */
441 if (ifp->if_flags & IFF_LOOPBACK)
442 ia->ia_dstaddr = ia->ia_addr;
444 ifa_ref(ifa); /* if_addrhead */
446 TAILQ_INSERT_TAIL(&ifp->if_addrhead, ifa, ifa_link);
447 IF_ADDR_WUNLOCK(ifp);
449 ifa_ref(ifa); /* in_ifaddrhead */
451 TAILQ_INSERT_TAIL(&V_in_ifaddrhead, ia, ia_link);
452 LIST_INSERT_HEAD(INADDR_HASH(ia->ia_addr.sin_addr.s_addr), ia, ia_hash);
456 error = (*carp_attach_p)(&ia->ia_ifa, vhid);
461 * Give the interface a chance to initialize
462 * if this is its first address,
463 * and to validate the address if necessary.
465 if (ifp->if_ioctl != NULL)
466 error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
471 * Add route for the network.
476 if (ifp->if_flags & (IFF_LOOPBACK|IFF_POINTOPOINT))
479 error = in_addprefix(ia, flags);
485 * Add a loopback route to self.
487 if (vhid == 0 && (ifp->if_flags & IFF_LOOPBACK) == 0 &&
488 ia->ia_addr.sin_addr.s_addr != INADDR_ANY) {
489 struct in_ifaddr *eia;
491 eia = in_localip_more(ia);
494 error = ifa_add_loopback_route((struct ifaddr *)ia,
495 (struct sockaddr *)&ia->ia_addr);
499 ifa_free(&eia->ia_ifa);
502 if (iaIsFirst && (ifp->if_flags & IFF_MULTICAST)) {
503 struct in_addr allhosts_addr;
504 struct in_ifinfo *ii;
506 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
507 allhosts_addr.s_addr = htonl(INADDR_ALLHOSTS_GROUP);
509 error = in_joingroup(ifp, &allhosts_addr, NULL,
513 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
519 (void )in_scrubprefix(ia, LLE_STATIC);
522 if (ia->ia_ifa.ifa_carp)
523 (*carp_detach_p)(&ia->ia_ifa);
527 TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
528 IF_ADDR_WUNLOCK(ifp);
529 ifa_free(&ia->ia_ifa);
532 TAILQ_REMOVE(&V_in_ifaddrhead, ia, ia_link);
533 LIST_REMOVE(ia, ia_hash);
535 ifa_free(&ia->ia_ifa);
541 in_difaddr_ioctl(caddr_t data, struct ifnet *ifp, struct thread *td)
543 const struct ifreq *ifr = (struct ifreq *)data;
544 const struct sockaddr_in *addr = (const struct sockaddr_in *)
547 struct in_ifaddr *ia;
548 bool deleteAny, iaIsLast;
552 error = priv_check(td, PRIV_NET_DELIFADDR);
557 if (addr->sin_len != sizeof(struct sockaddr_in) ||
558 addr->sin_family != AF_INET)
566 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
567 struct in_ifaddr *it = ifatoia(ifa);
569 if (it->ia_addr.sin_family != AF_INET)
572 if (deleteAny && ia == NULL && (td == NULL ||
573 prison_check_ip4(td->td_ucred, &it->ia_addr.sin_addr) == 0))
576 if (it->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
577 (td == NULL || prison_check_ip4(td->td_ucred,
578 &addr->sin_addr) == 0))
586 IF_ADDR_WUNLOCK(ifp);
587 return (EADDRNOTAVAIL);
590 TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
591 IF_ADDR_WUNLOCK(ifp);
592 ifa_free(&ia->ia_ifa); /* if_addrhead */
595 TAILQ_REMOVE(&V_in_ifaddrhead, ia, ia_link);
596 LIST_REMOVE(ia, ia_hash);
598 ifa_free(&ia->ia_ifa); /* in_ifaddrhead */
601 * in_scrubprefix() kills the interface route.
603 in_scrubprefix(ia, LLE_STATIC);
606 * in_ifadown gets rid of all the rest of
607 * the routes. This is not quite the right
608 * thing to do, but at least if we are running
609 * a routing process they will come back.
611 in_ifadown(&ia->ia_ifa, 1);
613 if (ia->ia_ifa.ifa_carp)
614 (*carp_detach_p)(&ia->ia_ifa);
617 * If this is the last IPv4 address configured on this
618 * interface, leave the all-hosts group.
619 * No state-change report need be transmitted.
621 if (iaIsLast && (ifp->if_flags & IFF_MULTICAST)) {
622 struct in_ifinfo *ii;
624 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
626 if (ii->ii_allhosts) {
627 (void)in_leavegroup_locked(ii->ii_allhosts, NULL);
628 ii->ii_allhosts = NULL;
633 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
640 * SIOCGLIFADDR: get first address. (?!?)
641 * SIOCGLIFADDR with IFLR_PREFIX:
642 * get first address that matches the specified prefix.
643 * SIOCALIFADDR: add the specified address.
644 * SIOCALIFADDR with IFLR_PREFIX:
645 * EINVAL since we can't deduce hostid part of the address.
646 * SIOCDLIFADDR: delete the specified address.
647 * SIOCDLIFADDR with IFLR_PREFIX:
648 * delete the first address that matches the specified prefix.
650 * EINVAL on invalid parameters
651 * EADDRNOTAVAIL on prefix match failed/specified address not found
652 * other values may be returned from in_ioctl()
655 in_lifaddr_ioctl(struct socket *so, u_long cmd, caddr_t data,
656 struct ifnet *ifp, struct thread *td)
658 struct if_laddrreq *iflr = (struct if_laddrreq *)data;
665 error = priv_check(td, PRIV_NET_ADDIFADDR);
672 error = priv_check(td, PRIV_NET_DELIFADDR);
681 /* address must be specified on GET with IFLR_PREFIX */
682 if ((iflr->flags & IFLR_PREFIX) == 0)
687 /* address must be specified on ADD and DELETE */
688 if (iflr->addr.ss_family != AF_INET)
690 if (iflr->addr.ss_len != sizeof(struct sockaddr_in))
692 /* XXX need improvement */
693 if (iflr->dstaddr.ss_family
694 && iflr->dstaddr.ss_family != AF_INET)
696 if (iflr->dstaddr.ss_family
697 && iflr->dstaddr.ss_len != sizeof(struct sockaddr_in))
700 default: /*shouldn't happen*/
703 if (sizeof(struct in_addr) * 8 < iflr->prefixlen)
709 struct in_aliasreq ifra;
711 if (iflr->flags & IFLR_PREFIX)
714 /* copy args to in_aliasreq, perform ioctl(SIOCAIFADDR). */
715 bzero(&ifra, sizeof(ifra));
716 bcopy(iflr->iflr_name, ifra.ifra_name,
717 sizeof(ifra.ifra_name));
719 bcopy(&iflr->addr, &ifra.ifra_addr, iflr->addr.ss_len);
721 if (iflr->dstaddr.ss_family) { /*XXX*/
722 bcopy(&iflr->dstaddr, &ifra.ifra_dstaddr,
723 iflr->dstaddr.ss_len);
726 ifra.ifra_mask.sin_family = AF_INET;
727 ifra.ifra_mask.sin_len = sizeof(struct sockaddr_in);
728 in_len2mask(&ifra.ifra_mask.sin_addr, iflr->prefixlen);
730 return (in_control(so, SIOCAIFADDR, (caddr_t)&ifra, ifp, td));
735 struct in_ifaddr *ia;
736 struct in_addr mask, candidate, match;
737 struct sockaddr_in *sin;
739 bzero(&mask, sizeof(mask));
740 bzero(&match, sizeof(match));
741 if (iflr->flags & IFLR_PREFIX) {
742 /* lookup a prefix rather than address. */
743 in_len2mask(&mask, iflr->prefixlen);
745 sin = (struct sockaddr_in *)&iflr->addr;
746 match.s_addr = sin->sin_addr.s_addr;
747 match.s_addr &= mask.s_addr;
749 /* if you set extra bits, that's wrong */
750 if (match.s_addr != sin->sin_addr.s_addr)
754 /* on getting an address, take the 1st match */
755 /* on deleting an address, do exact match */
756 if (cmd != SIOCGLIFADDR) {
757 in_len2mask(&mask, 32);
758 sin = (struct sockaddr_in *)&iflr->addr;
759 match.s_addr = sin->sin_addr.s_addr;
764 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
765 if (ifa->ifa_addr->sa_family != AF_INET)
767 if (match.s_addr == 0)
769 sin = (struct sockaddr_in *)&ifa->ifa_addr;
770 candidate.s_addr = sin->sin_addr.s_addr;
771 candidate.s_addr &= mask.s_addr;
772 if (candidate.s_addr == match.s_addr)
777 IF_ADDR_RUNLOCK(ifp);
779 return (EADDRNOTAVAIL);
780 ia = (struct in_ifaddr *)ifa;
782 if (cmd == SIOCGLIFADDR) {
783 /* fill in the if_laddrreq structure */
784 bcopy(&ia->ia_addr, &iflr->addr, ia->ia_addr.sin_len);
786 if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
787 bcopy(&ia->ia_dstaddr, &iflr->dstaddr,
788 ia->ia_dstaddr.sin_len);
790 bzero(&iflr->dstaddr, sizeof(iflr->dstaddr));
793 in_mask2len(&ia->ia_sockmask.sin_addr);
795 iflr->flags = 0; /*XXX*/
800 struct in_aliasreq ifra;
802 /* fill in_aliasreq and do ioctl(SIOCDIFADDR) */
803 bzero(&ifra, sizeof(ifra));
804 bcopy(iflr->iflr_name, ifra.ifra_name,
805 sizeof(ifra.ifra_name));
807 bcopy(&ia->ia_addr, &ifra.ifra_addr,
808 ia->ia_addr.sin_len);
809 if ((ifp->if_flags & IFF_POINTOPOINT) != 0) {
810 bcopy(&ia->ia_dstaddr, &ifra.ifra_dstaddr,
811 ia->ia_dstaddr.sin_len);
813 bcopy(&ia->ia_sockmask, &ifra.ifra_dstaddr,
814 ia->ia_sockmask.sin_len);
817 return (in_control(so, SIOCDIFADDR, (caddr_t)&ifra,
823 return (EOPNOTSUPP); /*just for safety*/
826 #define rtinitflags(x) \
827 ((((x)->ia_ifp->if_flags & (IFF_LOOPBACK | IFF_POINTOPOINT)) != 0) \
831 * Generate a routing message when inserting or deleting
832 * an interface address alias.
834 static void in_addralias_rtmsg(int cmd, struct in_addr *prefix,
835 struct in_ifaddr *target)
838 struct sockaddr_in *pfx_addr;
839 struct rtentry msg_rt;
842 * This is a bit questionable because there is no
843 * additional route entry added/deleted for an address
844 * alias. Therefore this route report is inaccurate.
846 bzero(&pfx_ro, sizeof(pfx_ro));
847 pfx_addr = (struct sockaddr_in *)(&pfx_ro.ro_dst);
848 pfx_addr->sin_len = sizeof(*pfx_addr);
849 pfx_addr->sin_family = AF_INET;
850 pfx_addr->sin_addr = *prefix;
851 rtalloc_ign_fib(&pfx_ro, 0, 0);
852 if (pfx_ro.ro_rt != NULL) {
853 msg_rt = *pfx_ro.ro_rt;
856 * Point the gateway to the new interface
857 * address as if a new prefix route entry has
858 * been added through the new address alias.
859 * All other parts of the rtentry is accurate,
860 * e.g., rt_key, rt_mask, rt_ifp etc.
862 msg_rt.rt_gateway = (struct sockaddr *)&target->ia_addr;
863 rt_newaddrmsg(cmd, (struct ifaddr *)target, 0, &msg_rt);
864 RTFREE(pfx_ro.ro_rt);
870 * Check if we have a route for the given prefix already or add one accordingly.
873 in_addprefix(struct in_ifaddr *target, int flags)
875 struct in_ifaddr *ia;
876 struct in_addr prefix, mask, p, m;
879 if ((flags & RTF_HOST) != 0) {
880 prefix = target->ia_dstaddr.sin_addr;
883 prefix = target->ia_addr.sin_addr;
884 mask = target->ia_sockmask.sin_addr;
885 prefix.s_addr &= mask.s_addr;
889 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
890 if (rtinitflags(ia)) {
891 p = ia->ia_dstaddr.sin_addr;
893 if (prefix.s_addr != p.s_addr)
896 p = ia->ia_addr.sin_addr;
897 m = ia->ia_sockmask.sin_addr;
898 p.s_addr &= m.s_addr;
900 if (prefix.s_addr != p.s_addr ||
901 mask.s_addr != m.s_addr)
906 * If we got a matching prefix route inserted by other
907 * interface address, we are done here.
909 if (ia->ia_flags & IFA_ROUTE) {
911 if (ia->ia_addr.sin_addr.s_addr ==
912 target->ia_addr.sin_addr.s_addr) {
918 if (V_nosameprefix) {
922 in_addralias_rtmsg(RTM_ADD, &prefix, target);
931 * No-one seem to have this prefix route, so we try to insert it.
933 error = rtinit(&target->ia_ifa, (int)RTM_ADD, flags);
935 target->ia_flags |= IFA_ROUTE;
940 * If there is no other address in the system that can serve a route to the
941 * same prefix, remove the route. Hand over the route to the new address
945 in_scrubprefix(struct in_ifaddr *target, u_int flags)
947 struct in_ifaddr *ia;
948 struct in_addr prefix, mask, p, m;
950 struct sockaddr_in prefix0, mask0;
953 * Remove the loopback route to the interface address.
955 if ((target->ia_addr.sin_addr.s_addr != INADDR_ANY) &&
956 !(target->ia_ifp->if_flags & IFF_LOOPBACK) &&
957 (flags & LLE_STATIC)) {
958 struct in_ifaddr *eia;
960 eia = in_localip_more(target);
963 error = ifa_switch_loopback_route((struct ifaddr *)eia,
964 (struct sockaddr *)&target->ia_addr);
965 ifa_free(&eia->ia_ifa);
967 error = ifa_del_loopback_route((struct ifaddr *)target,
968 (struct sockaddr *)&target->ia_addr);
971 if (!(target->ia_ifp->if_flags & IFF_NOARP))
972 /* remove arp cache */
973 arp_ifscrub(target->ia_ifp,
974 IA_SIN(target)->sin_addr.s_addr);
977 if (rtinitflags(target)) {
978 prefix = target->ia_dstaddr.sin_addr;
981 prefix = target->ia_addr.sin_addr;
982 mask = target->ia_sockmask.sin_addr;
983 prefix.s_addr &= mask.s_addr;
986 if ((target->ia_flags & IFA_ROUTE) == 0) {
987 in_addralias_rtmsg(RTM_DELETE, &prefix, target);
992 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
993 if (rtinitflags(ia)) {
994 p = ia->ia_dstaddr.sin_addr;
996 if (prefix.s_addr != p.s_addr)
999 p = ia->ia_addr.sin_addr;
1000 m = ia->ia_sockmask.sin_addr;
1001 p.s_addr &= m.s_addr;
1003 if (prefix.s_addr != p.s_addr ||
1004 mask.s_addr != m.s_addr)
1008 if ((ia->ia_ifp->if_flags & IFF_UP) == 0)
1012 * If we got a matching prefix address, move IFA_ROUTE and
1013 * the route itself to it. Make sure that routing daemons
1016 if ((ia->ia_flags & IFA_ROUTE) == 0) {
1017 ifa_ref(&ia->ia_ifa);
1018 IN_IFADDR_RUNLOCK();
1019 error = rtinit(&(target->ia_ifa), (int)RTM_DELETE,
1020 rtinitflags(target));
1022 target->ia_flags &= ~IFA_ROUTE;
1024 log(LOG_INFO, "in_scrubprefix: err=%d, old prefix delete failed\n",
1026 error = rtinit(&ia->ia_ifa, (int)RTM_ADD,
1027 rtinitflags(ia) | RTF_UP);
1029 ia->ia_flags |= IFA_ROUTE;
1031 log(LOG_INFO, "in_scrubprefix: err=%d, new prefix add failed\n",
1033 ifa_free(&ia->ia_ifa);
1037 IN_IFADDR_RUNLOCK();
1040 * remove all L2 entries on the given prefix
1042 bzero(&prefix0, sizeof(prefix0));
1043 prefix0.sin_len = sizeof(prefix0);
1044 prefix0.sin_family = AF_INET;
1045 prefix0.sin_addr.s_addr = target->ia_subnet;
1046 bzero(&mask0, sizeof(mask0));
1047 mask0.sin_len = sizeof(mask0);
1048 mask0.sin_family = AF_INET;
1049 mask0.sin_addr.s_addr = target->ia_subnetmask;
1050 lltable_prefix_free(AF_INET, (struct sockaddr *)&prefix0,
1051 (struct sockaddr *)&mask0, flags);
1054 * As no-one seem to have this prefix, we can remove the route.
1056 error = rtinit(&(target->ia_ifa), (int)RTM_DELETE, rtinitflags(target));
1058 target->ia_flags &= ~IFA_ROUTE;
1060 log(LOG_INFO, "in_scrubprefix: err=%d, prefix delete failed\n", error);
1067 * Return 1 if the address might be a local broadcast address.
1070 in_broadcast(struct in_addr in, struct ifnet *ifp)
1072 register struct ifaddr *ifa;
1075 if (in.s_addr == INADDR_BROADCAST ||
1076 in.s_addr == INADDR_ANY)
1078 if ((ifp->if_flags & IFF_BROADCAST) == 0)
1080 t = ntohl(in.s_addr);
1082 * Look through the list of addresses for a match
1083 * with a broadcast address.
1085 #define ia ((struct in_ifaddr *)ifa)
1086 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
1087 if (ifa->ifa_addr->sa_family == AF_INET &&
1088 (in.s_addr == ia->ia_broadaddr.sin_addr.s_addr ||
1090 * Check for old-style (host 0) broadcast, but
1091 * taking into account that RFC 3021 obsoletes it.
1093 (ia->ia_subnetmask != IN_RFC3021_MASK &&
1094 t == ia->ia_subnet)) &&
1096 * Check for an all one subnetmask. These
1097 * only exist when an interface gets a secondary
1100 ia->ia_subnetmask != (u_long)0xffffffff)
1107 * On interface removal, clean up IPv4 data structures hung off of the ifnet.
1110 in_ifdetach(struct ifnet *ifp)
1113 in_pcbpurgeif0(&V_ripcbinfo, ifp);
1114 in_pcbpurgeif0(&V_udbinfo, ifp);
1115 in_purgemaddrs(ifp);
1119 * Delete all IPv4 multicast address records, and associated link-layer
1120 * multicast address records, associated with ifp.
1121 * XXX It looks like domifdetach runs AFTER the link layer cleanup.
1122 * XXX This should not race with ifma_protospec being set during
1123 * a new allocation, if it does, we have bigger problems.
1126 in_purgemaddrs(struct ifnet *ifp)
1128 LIST_HEAD(,in_multi) purgeinms;
1129 struct in_multi *inm, *tinm;
1130 struct ifmultiaddr *ifma;
1132 LIST_INIT(&purgeinms);
1136 * Extract list of in_multi associated with the detaching ifp
1137 * which the PF_INET layer is about to release.
1138 * We need to do this as IF_ADDR_LOCK() may be re-acquired
1139 * by code further down.
1142 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
1143 if (ifma->ifma_addr->sa_family != AF_INET ||
1144 ifma->ifma_protospec == NULL)
1147 KASSERT(ifma->ifma_protospec != NULL,
1148 ("%s: ifma_protospec is NULL", __func__));
1150 inm = (struct in_multi *)ifma->ifma_protospec;
1151 LIST_INSERT_HEAD(&purgeinms, inm, inm_link);
1153 IF_ADDR_RUNLOCK(ifp);
1155 LIST_FOREACH_SAFE(inm, &purgeinms, inm_link, tinm) {
1156 LIST_REMOVE(inm, inm_link);
1157 inm_release_locked(inm);
1165 struct llentry base;
1166 struct sockaddr_in l3_addr4;
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);
1183 static struct llentry *
1184 in_lltable_new(const struct sockaddr *l3addr, u_int flags)
1186 struct in_llentry *lle;
1188 lle = malloc(sizeof(struct in_llentry), M_LLTABLE, M_NOWAIT | M_ZERO);
1189 if (lle == NULL) /* NB: caller generates msg */
1193 * For IPv4 this will trigger "arpresolve" to generate
1196 lle->base.la_expire = time_uptime; /* mark expired */
1197 lle->l3_addr4 = *(const struct sockaddr_in *)l3addr;
1198 lle->base.lle_refcnt = 1;
1199 lle->base.lle_free = in_lltable_free;
1200 LLE_LOCK_INIT(&lle->base);
1201 callout_init_rw(&lle->base.la_timer, &lle->base.lle_lock,
1202 CALLOUT_RETURNUNLOCKED);
1204 return (&lle->base);
1207 #define IN_ARE_MASKED_ADDR_EQUAL(d, a, m) ( \
1208 (((ntohl((d)->sin_addr.s_addr) ^ (a)->sin_addr.s_addr) & (m)->sin_addr.s_addr)) == 0 )
1211 in_lltable_prefix_free(struct lltable *llt, const struct sockaddr *prefix,
1212 const struct sockaddr *mask, u_int flags)
1214 const struct sockaddr_in *pfx = (const struct sockaddr_in *)prefix;
1215 const struct sockaddr_in *msk = (const struct sockaddr_in *)mask;
1216 struct llentry *lle, *next;
1218 size_t pkts_dropped;
1220 IF_AFDATA_WLOCK(llt->llt_ifp);
1221 for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) {
1222 LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) {
1224 * (flags & LLE_STATIC) means deleting all entries
1225 * including static ARP entries.
1227 if (IN_ARE_MASKED_ADDR_EQUAL(satosin(L3_ADDR(lle)),
1228 pfx, msk) && ((flags & LLE_STATIC) ||
1229 !(lle->la_flags & LLE_STATIC))) {
1231 if (callout_stop(&lle->la_timer))
1233 pkts_dropped = llentry_free(lle);
1234 ARPSTAT_ADD(dropped, pkts_dropped);
1238 IF_AFDATA_WUNLOCK(llt->llt_ifp);
1243 in_lltable_rtcheck(struct ifnet *ifp, u_int flags, const struct sockaddr *l3addr)
1247 KASSERT(l3addr->sa_family == AF_INET,
1248 ("sin_family %d", l3addr->sa_family));
1250 /* XXX rtalloc1 should take a const param */
1251 rt = rtalloc1(__DECONST(struct sockaddr *, l3addr), 0, 0);
1257 * If the gateway for an existing host route matches the target L3
1258 * address, which is a special route inserted by some implementation
1259 * such as MANET, and the interface is of the correct type, then
1260 * allow for ARP to proceed.
1262 if (rt->rt_flags & RTF_GATEWAY) {
1263 if (!(rt->rt_flags & RTF_HOST) || !rt->rt_ifp ||
1264 rt->rt_ifp->if_type != IFT_ETHER ||
1265 (rt->rt_ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) != 0 ||
1266 memcmp(rt->rt_gateway->sa_data, l3addr->sa_data,
1267 sizeof(in_addr_t)) != 0) {
1274 * Make sure that at least the destination address is covered
1275 * by the route. This is for handling the case where 2 or more
1276 * interfaces have the same prefix. An incoming packet arrives
1277 * on one interface and the corresponding outgoing packet leaves
1278 * another interface.
1280 if (!(rt->rt_flags & RTF_HOST) && rt->rt_ifp != ifp) {
1281 const char *sa, *mask, *addr, *lim;
1284 mask = (const char *)rt_mask(rt);
1286 * Just being extra cautious to avoid some custom
1287 * code getting into trouble.
1294 sa = (const char *)rt_key(rt);
1295 addr = (const char *)l3addr;
1296 len = ((const struct sockaddr_in *)l3addr)->sin_len;
1299 for ( ; addr < lim; sa++, mask++, addr++) {
1300 if ((*sa ^ *addr) & *mask) {
1302 log(LOG_INFO, "IPv4 address: \"%s\" is not on the network\n",
1303 inet_ntoa(((const struct sockaddr_in *)l3addr)->sin_addr));
1316 * Return NULL if not found or marked for deletion.
1317 * If found return lle read locked.
1319 static struct llentry *
1320 in_lltable_lookup(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
1322 const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
1323 struct ifnet *ifp = llt->llt_ifp;
1324 struct llentry *lle;
1325 struct llentries *lleh;
1328 IF_AFDATA_LOCK_ASSERT(ifp);
1329 KASSERT(l3addr->sa_family == AF_INET,
1330 ("sin_family %d", l3addr->sa_family));
1332 hashkey = sin->sin_addr.s_addr;
1333 lleh = &llt->lle_head[LLATBL_HASH(hashkey, LLTBL_HASHMASK)];
1334 LIST_FOREACH(lle, lleh, lle_next) {
1335 struct sockaddr_in *sa2 = satosin(L3_ADDR(lle));
1336 if (lle->la_flags & LLE_DELETED)
1338 if (sa2->sin_addr.s_addr == sin->sin_addr.s_addr)
1343 if (flags & LLE_DELETE)
1344 log(LOG_INFO, "interface address is missing from cache = %p in delete\n", lle);
1346 if (!(flags & LLE_CREATE))
1349 * A route that covers the given address must have
1350 * been installed 1st because we are doing a resolution,
1353 if (!(flags & LLE_IFADDR) &&
1354 in_lltable_rtcheck(ifp, flags, l3addr) != 0)
1357 lle = in_lltable_new(l3addr, flags);
1359 log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
1362 lle->la_flags = flags & ~LLE_CREATE;
1363 if ((flags & (LLE_CREATE | LLE_IFADDR)) == (LLE_CREATE | LLE_IFADDR)) {
1364 bcopy(IF_LLADDR(ifp), &lle->ll_addr, ifp->if_addrlen);
1365 lle->la_flags |= (LLE_VALID | LLE_STATIC);
1369 lle->lle_head = lleh;
1370 lle->la_flags |= LLE_LINKED;
1371 LIST_INSERT_HEAD(lleh, lle, lle_next);
1372 } else if (flags & LLE_DELETE) {
1373 if (!(lle->la_flags & LLE_IFADDR) || (flags & LLE_IFADDR)) {
1375 lle->la_flags |= LLE_DELETED;
1376 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
1378 log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
1380 if ((lle->la_flags &
1381 (LLE_STATIC | LLE_IFADDR)) == LLE_STATIC)
1389 if (LLE_IS_VALID(lle)) {
1390 if (flags & LLE_EXCLUSIVE)
1400 in_lltable_dump(struct lltable *llt, struct sysctl_req *wr)
1402 #define SIN(lle) ((struct sockaddr_in *) L3_ADDR(lle))
1403 struct ifnet *ifp = llt->llt_ifp;
1404 struct llentry *lle;
1407 struct rt_msghdr rtm;
1408 struct sockaddr_in sin;
1409 struct sockaddr_dl sdl;
1413 LLTABLE_LOCK_ASSERT();
1416 for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) {
1417 LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
1418 struct sockaddr_dl *sdl;
1420 /* skip deleted entries */
1421 if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
1423 /* Skip if jailed and not a valid IP of the prison. */
1424 if (prison_if(wr->td->td_ucred, L3_ADDR(lle)) != 0)
1427 * produce a msg made of:
1429 * struct sockaddr_in; (IPv4)
1430 * struct sockaddr_dl;
1432 bzero(&arpc, sizeof(arpc));
1433 arpc.rtm.rtm_msglen = sizeof(arpc);
1434 arpc.rtm.rtm_version = RTM_VERSION;
1435 arpc.rtm.rtm_type = RTM_GET;
1436 arpc.rtm.rtm_flags = RTF_UP;
1437 arpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
1438 arpc.sin.sin_family = AF_INET;
1439 arpc.sin.sin_len = sizeof(arpc.sin);
1440 arpc.sin.sin_addr.s_addr = SIN(lle)->sin_addr.s_addr;
1443 if (lle->la_flags & LLE_PUB)
1444 arpc.rtm.rtm_flags |= RTF_ANNOUNCE;
1447 sdl->sdl_family = AF_LINK;
1448 sdl->sdl_len = sizeof(*sdl);
1449 sdl->sdl_index = ifp->if_index;
1450 sdl->sdl_type = ifp->if_type;
1451 if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
1452 sdl->sdl_alen = ifp->if_addrlen;
1453 bcopy(&lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
1456 bzero(LLADDR(sdl), ifp->if_addrlen);
1459 arpc.rtm.rtm_rmx.rmx_expire =
1460 lle->la_flags & LLE_STATIC ? 0 : lle->la_expire;
1461 arpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
1462 if (lle->la_flags & LLE_STATIC)
1463 arpc.rtm.rtm_flags |= RTF_STATIC;
1464 arpc.rtm.rtm_index = ifp->if_index;
1465 error = SYSCTL_OUT(wr, &arpc, sizeof(arpc));
1475 in_domifattach(struct ifnet *ifp)
1477 struct in_ifinfo *ii;
1478 struct lltable *llt;
1480 ii = malloc(sizeof(struct in_ifinfo), M_IFADDR, M_WAITOK|M_ZERO);
1482 llt = lltable_init(ifp, AF_INET);
1484 llt->llt_prefix_free = in_lltable_prefix_free;
1485 llt->llt_lookup = in_lltable_lookup;
1486 llt->llt_dump = in_lltable_dump;
1490 ii->ii_igmp = igmp_domifattach(ifp);
1496 in_domifdetach(struct ifnet *ifp, void *aux)
1498 struct in_ifinfo *ii = (struct in_ifinfo *)aux;
1500 igmp_domifdetach(ifp);
1501 lltable_free(ii->ii_llt);