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_aifaddr_ioctl(u_long, caddr_t, struct ifnet *, struct thread *);
72 static int in_difaddr_ioctl(caddr_t, struct ifnet *, struct thread *);
74 static void in_socktrim(struct sockaddr_in *);
75 static void in_purgemaddrs(struct ifnet *);
77 static VNET_DEFINE(int, nosameprefix);
78 #define V_nosameprefix VNET(nosameprefix)
79 SYSCTL_INT(_net_inet_ip, OID_AUTO, no_same_prefix, CTLFLAG_VNET | CTLFLAG_RW,
80 &VNET_NAME(nosameprefix), 0,
81 "Refuse to create same prefixes on different interfaces");
83 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
84 #define V_ripcbinfo VNET(ripcbinfo)
86 static struct sx in_control_sx;
87 SX_SYSINIT(in_control_sx, &in_control_sx, "in_control");
90 * Return 1 if an internet address is for a ``local'' host
91 * (one to which we have a connection).
94 in_localaddr(struct in_addr in)
96 register u_long i = ntohl(in.s_addr);
97 register struct in_ifaddr *ia;
100 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
101 if ((i & ia->ia_subnetmask) == ia->ia_subnet) {
111 * Return 1 if an internet address is for the local host and configured
112 * on one of its interfaces.
115 in_localip(struct in_addr in)
117 struct in_ifaddr *ia;
120 LIST_FOREACH(ia, INADDR_HASH(in.s_addr), ia_hash) {
121 if (IA_SIN(ia)->sin_addr.s_addr == in.s_addr) {
131 * Return 1 if an internet address is configured on an interface.
134 in_ifhasaddr(struct ifnet *ifp, struct in_addr in)
137 struct in_ifaddr *ia;
140 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
141 if (ifa->ifa_addr->sa_family != AF_INET)
143 ia = (struct in_ifaddr *)ifa;
144 if (ia->ia_addr.sin_addr.s_addr == in.s_addr) {
145 IF_ADDR_RUNLOCK(ifp);
149 IF_ADDR_RUNLOCK(ifp);
155 * Return a reference to the interface address which is different to
156 * the supplied one but with same IP address value.
158 static struct in_ifaddr *
159 in_localip_more(struct in_ifaddr *ia)
161 in_addr_t in = IA_SIN(ia)->sin_addr.s_addr;
162 struct in_ifaddr *it;
165 LIST_FOREACH(it, INADDR_HASH(in), ia_hash) {
166 if (it != ia && IA_SIN(it)->sin_addr.s_addr == in) {
167 ifa_ref(&it->ia_ifa);
178 * Determine whether an IP address is in a reserved set of addresses
179 * that may not be forwarded, or whether datagrams to that destination
183 in_canforward(struct in_addr in)
185 register u_long i = ntohl(in.s_addr);
188 if (IN_EXPERIMENTAL(i) || IN_MULTICAST(i) || IN_LINKLOCAL(i))
191 net = i & IN_CLASSA_NET;
192 if (net == 0 || net == (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))
199 * Trim a mask in a sockaddr
202 in_socktrim(struct sockaddr_in *ap)
204 register char *cplim = (char *) &ap->sin_addr;
205 register char *cp = (char *) (&ap->sin_addr + 1);
208 while (--cp >= cplim)
210 (ap)->sin_len = cp - (char *) (ap) + 1;
216 * Generic internet control operations (ioctl's).
219 in_control(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp,
222 struct ifreq *ifr = (struct ifreq *)data;
223 struct sockaddr_in *addr = (struct sockaddr_in *)&ifr->ifr_addr;
225 struct in_ifaddr *ia;
229 return (EADDRNOTAVAIL);
232 * Filter out 4 ioctls we implement directly. Forward the rest
233 * to specific functions and ifp->if_ioctl().
242 sx_xlock(&in_control_sx);
243 error = in_difaddr_ioctl(data, ifp, td);
244 sx_xunlock(&in_control_sx);
246 case OSIOCAIFADDR: /* 9.x compat */
248 sx_xlock(&in_control_sx);
249 error = in_aifaddr_ioctl(cmd, data, ifp, td);
250 sx_xunlock(&in_control_sx);
256 /* We no longer support that old commands. */
259 if (ifp->if_ioctl == NULL)
261 return ((*ifp->if_ioctl)(ifp, cmd, data));
264 if (addr->sin_addr.s_addr != INADDR_ANY &&
265 prison_check_ip4(td->td_ucred, &addr->sin_addr) != 0)
266 return (EADDRNOTAVAIL);
269 * Find address for this interface, if it exists. If an
270 * address was specified, find that one instead of the
271 * first one on the interface, if possible.
274 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
275 if (ifa->ifa_addr->sa_family != AF_INET)
277 ia = (struct in_ifaddr *)ifa;
278 if (ia->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr)
282 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
283 if (ifa->ifa_addr->sa_family == AF_INET) {
284 ia = (struct in_ifaddr *)ifa;
285 if (prison_check_ip4(td->td_ucred,
286 &ia->ia_addr.sin_addr) == 0)
291 IF_ADDR_RUNLOCK(ifp);
292 return (EADDRNOTAVAIL);
302 if ((ifp->if_flags & IFF_BROADCAST) == 0) {
306 *addr = ia->ia_broadaddr;
310 if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
314 *addr = ia->ia_dstaddr;
318 *addr = ia->ia_sockmask;
322 IF_ADDR_RUNLOCK(ifp);
328 in_aifaddr_ioctl(u_long cmd, caddr_t data, struct ifnet *ifp, struct thread *td)
330 const struct in_aliasreq *ifra = (struct in_aliasreq *)data;
331 const struct sockaddr_in *addr = &ifra->ifra_addr;
332 const struct sockaddr_in *broadaddr = &ifra->ifra_broadaddr;
333 const struct sockaddr_in *mask = &ifra->ifra_mask;
334 const struct sockaddr_in *dstaddr = &ifra->ifra_dstaddr;
335 const int vhid = (cmd == SIOCAIFADDR) ? ifra->ifra_vhid : 0;
337 struct in_ifaddr *ia;
341 error = priv_check(td, PRIV_NET_ADDIFADDR);
346 * ifra_addr must be present and be of INET family.
347 * ifra_broadaddr/ifra_dstaddr and ifra_mask are optional.
349 if (addr->sin_len != sizeof(struct sockaddr_in) ||
350 addr->sin_family != AF_INET)
352 if (broadaddr->sin_len != 0 &&
353 (broadaddr->sin_len != sizeof(struct sockaddr_in) ||
354 broadaddr->sin_family != AF_INET))
356 if (mask->sin_len != 0 &&
357 (mask->sin_len != sizeof(struct sockaddr_in) ||
358 mask->sin_family != AF_INET))
360 if ((ifp->if_flags & IFF_POINTOPOINT) &&
361 (dstaddr->sin_len != sizeof(struct sockaddr_in) ||
362 dstaddr->sin_addr.s_addr == INADDR_ANY))
363 return (EDESTADDRREQ);
364 if (vhid > 0 && carp_attach_p == NULL)
365 return (EPROTONOSUPPORT);
368 * See whether address already exist.
373 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
374 struct in_ifaddr *it;
376 if (ifa->ifa_addr->sa_family != AF_INET)
379 it = (struct in_ifaddr *)ifa;
381 if (it->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
382 prison_check_ip4(td->td_ucred, &addr->sin_addr) == 0)
385 IF_ADDR_RUNLOCK(ifp);
388 (void )in_difaddr_ioctl(data, ifp, td);
390 ifa = ifa_alloc(sizeof(struct in_ifaddr), M_WAITOK);
391 ia = (struct in_ifaddr *)ifa;
392 ifa->ifa_addr = (struct sockaddr *)&ia->ia_addr;
393 ifa->ifa_dstaddr = (struct sockaddr *)&ia->ia_dstaddr;
394 ifa->ifa_netmask = (struct sockaddr *)&ia->ia_sockmask;
398 if (mask->sin_len != 0) {
399 ia->ia_sockmask = *mask;
400 ia->ia_subnetmask = ntohl(ia->ia_sockmask.sin_addr.s_addr);
402 in_addr_t i = ntohl(addr->sin_addr.s_addr);
405 * Be compatible with network classes, if netmask isn't
406 * supplied, guess it based on classes.
409 ia->ia_subnetmask = IN_CLASSA_NET;
410 else if (IN_CLASSB(i))
411 ia->ia_subnetmask = IN_CLASSB_NET;
413 ia->ia_subnetmask = IN_CLASSC_NET;
414 ia->ia_sockmask.sin_addr.s_addr = htonl(ia->ia_subnetmask);
416 ia->ia_subnet = ntohl(addr->sin_addr.s_addr) & ia->ia_subnetmask;
417 in_socktrim(&ia->ia_sockmask);
419 if (ifp->if_flags & IFF_BROADCAST) {
420 if (broadaddr->sin_len != 0) {
421 ia->ia_broadaddr = *broadaddr;
422 } else if (ia->ia_subnetmask == IN_RFC3021_MASK) {
423 ia->ia_broadaddr.sin_addr.s_addr = INADDR_BROADCAST;
424 ia->ia_broadaddr.sin_len = sizeof(struct sockaddr_in);
425 ia->ia_broadaddr.sin_family = AF_INET;
427 ia->ia_broadaddr.sin_addr.s_addr =
428 htonl(ia->ia_subnet | ~ia->ia_subnetmask);
429 ia->ia_broadaddr.sin_len = sizeof(struct sockaddr_in);
430 ia->ia_broadaddr.sin_family = AF_INET;
434 if (ifp->if_flags & IFF_POINTOPOINT)
435 ia->ia_dstaddr = *dstaddr;
437 /* XXXGL: rtinit() needs this strange assignment. */
438 if (ifp->if_flags & IFF_LOOPBACK)
439 ia->ia_dstaddr = ia->ia_addr;
442 error = (*carp_attach_p)(&ia->ia_ifa, vhid);
447 /* if_addrhead is already referenced by ifa_alloc() */
449 TAILQ_INSERT_TAIL(&ifp->if_addrhead, ifa, ifa_link);
450 IF_ADDR_WUNLOCK(ifp);
452 ifa_ref(ifa); /* in_ifaddrhead */
454 TAILQ_INSERT_TAIL(&V_in_ifaddrhead, ia, ia_link);
455 LIST_INSERT_HEAD(INADDR_HASH(ia->ia_addr.sin_addr.s_addr), ia, ia_hash);
459 * Give the interface a chance to initialize
460 * if this is its first address,
461 * and to validate the address if necessary.
463 if (ifp->if_ioctl != NULL) {
464 error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
470 * Add route for the network.
475 if (ifp->if_flags & (IFF_LOOPBACK|IFF_POINTOPOINT))
478 error = in_addprefix(ia, flags);
484 * Add a loopback route to self.
486 if (vhid == 0 && (ifp->if_flags & IFF_LOOPBACK) == 0 &&
487 ia->ia_addr.sin_addr.s_addr != INADDR_ANY &&
488 !((ifp->if_flags & IFF_POINTOPOINT) &&
489 ia->ia_dstaddr.sin_addr.s_addr == ia->ia_addr.sin_addr.s_addr)) {
490 struct in_ifaddr *eia;
492 eia = in_localip_more(ia);
495 error = ifa_add_loopback_route((struct ifaddr *)ia,
496 (struct sockaddr *)&ia->ia_addr);
500 ifa_free(&eia->ia_ifa);
503 if (iaIsFirst && (ifp->if_flags & IFF_MULTICAST)) {
504 struct in_addr allhosts_addr;
505 struct in_ifinfo *ii;
507 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
508 allhosts_addr.s_addr = htonl(INADDR_ALLHOSTS_GROUP);
510 error = in_joingroup(ifp, &allhosts_addr, NULL,
514 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
520 (void )in_scrubprefix(ia, LLE_STATIC);
523 if (ia->ia_ifa.ifa_carp)
524 (*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); /* if_addrhead */
532 TAILQ_REMOVE(&V_in_ifaddrhead, ia, ia_link);
533 LIST_REMOVE(ia, ia_hash);
535 ifa_free(&ia->ia_ifa); /* in_ifaddrhead */
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;
569 if (ifa->ifa_addr->sa_family != AF_INET)
572 it = (struct in_ifaddr *)ifa;
573 if (deleteAny && ia == NULL && (td == NULL ||
574 prison_check_ip4(td->td_ucred, &it->ia_addr.sin_addr) == 0))
577 if (it->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
578 (td == NULL || prison_check_ip4(td->td_ucred,
579 &addr->sin_addr) == 0))
587 IF_ADDR_WUNLOCK(ifp);
588 return (EADDRNOTAVAIL);
591 TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
592 IF_ADDR_WUNLOCK(ifp);
593 ifa_free(&ia->ia_ifa); /* if_addrhead */
596 TAILQ_REMOVE(&V_in_ifaddrhead, ia, ia_link);
597 LIST_REMOVE(ia, ia_hash);
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);
634 ifa_free(&ia->ia_ifa); /* in_ifaddrhead */
639 #define rtinitflags(x) \
640 ((((x)->ia_ifp->if_flags & (IFF_LOOPBACK | IFF_POINTOPOINT)) != 0) \
644 * Check if we have a route for the given prefix already or add one accordingly.
647 in_addprefix(struct in_ifaddr *target, int flags)
649 struct in_ifaddr *ia;
650 struct in_addr prefix, mask, p, m;
653 if ((flags & RTF_HOST) != 0) {
654 prefix = target->ia_dstaddr.sin_addr;
657 prefix = target->ia_addr.sin_addr;
658 mask = target->ia_sockmask.sin_addr;
659 prefix.s_addr &= mask.s_addr;
663 /* Look for an existing address with the same prefix, mask, and fib */
664 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
665 if (rtinitflags(ia)) {
666 p = ia->ia_dstaddr.sin_addr;
668 if (prefix.s_addr != p.s_addr)
671 p = ia->ia_addr.sin_addr;
672 m = ia->ia_sockmask.sin_addr;
673 p.s_addr &= m.s_addr;
675 if (prefix.s_addr != p.s_addr ||
676 mask.s_addr != m.s_addr)
679 if (target->ia_ifp->if_fib != ia->ia_ifp->if_fib)
683 * If we got a matching prefix route inserted by other
684 * interface address, we are done here.
686 if (ia->ia_flags & IFA_ROUTE) {
688 if (ia->ia_addr.sin_addr.s_addr ==
689 target->ia_addr.sin_addr.s_addr) {
695 if (V_nosameprefix) {
701 fibnum = V_rt_add_addr_allfibs ? RT_ALL_FIBS :
702 target->ia_ifp->if_fib;
703 rt_addrmsg(RTM_ADD, &target->ia_ifa, fibnum);
712 * No-one seem to have this prefix route, so we try to insert it.
714 error = rtinit(&target->ia_ifa, (int)RTM_ADD, flags);
716 target->ia_flags |= IFA_ROUTE;
721 * If there is no other address in the system that can serve a route to the
722 * same prefix, remove the route. Hand over the route to the new address
726 in_scrubprefix(struct in_ifaddr *target, u_int flags)
728 struct in_ifaddr *ia;
729 struct in_addr prefix, mask, p, m;
731 struct sockaddr_in prefix0, mask0;
734 * Remove the loopback route to the interface address.
736 if ((target->ia_addr.sin_addr.s_addr != INADDR_ANY) &&
737 !(target->ia_ifp->if_flags & IFF_LOOPBACK) &&
738 (flags & LLE_STATIC)) {
739 struct in_ifaddr *eia;
741 eia = in_localip_more(target);
744 int fibnum = target->ia_ifp->if_fib;
746 error = ifa_switch_loopback_route((struct ifaddr *)eia,
747 (struct sockaddr *)&target->ia_addr, fibnum);
748 ifa_free(&eia->ia_ifa);
750 error = ifa_del_loopback_route((struct ifaddr *)target,
751 (struct sockaddr *)&target->ia_addr);
754 if (!(target->ia_ifp->if_flags & IFF_NOARP))
755 /* remove arp cache */
756 arp_ifscrub(target->ia_ifp,
757 IA_SIN(target)->sin_addr.s_addr);
760 if (rtinitflags(target)) {
761 prefix = target->ia_dstaddr.sin_addr;
764 prefix = target->ia_addr.sin_addr;
765 mask = target->ia_sockmask.sin_addr;
766 prefix.s_addr &= mask.s_addr;
769 if ((target->ia_flags & IFA_ROUTE) == 0) {
772 fibnum = V_rt_add_addr_allfibs ? RT_ALL_FIBS :
773 target->ia_ifp->if_fib;
774 rt_addrmsg(RTM_DELETE, &target->ia_ifa, fibnum);
779 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
780 if (rtinitflags(ia)) {
781 p = ia->ia_dstaddr.sin_addr;
783 if (prefix.s_addr != p.s_addr)
786 p = ia->ia_addr.sin_addr;
787 m = ia->ia_sockmask.sin_addr;
788 p.s_addr &= m.s_addr;
790 if (prefix.s_addr != p.s_addr ||
791 mask.s_addr != m.s_addr)
795 if ((ia->ia_ifp->if_flags & IFF_UP) == 0)
799 * If we got a matching prefix address, move IFA_ROUTE and
800 * the route itself to it. Make sure that routing daemons
803 if ((ia->ia_flags & IFA_ROUTE) == 0) {
804 ifa_ref(&ia->ia_ifa);
806 error = rtinit(&(target->ia_ifa), (int)RTM_DELETE,
807 rtinitflags(target));
809 target->ia_flags &= ~IFA_ROUTE;
811 log(LOG_INFO, "in_scrubprefix: err=%d, old prefix delete failed\n",
813 error = rtinit(&ia->ia_ifa, (int)RTM_ADD,
814 rtinitflags(ia) | RTF_UP);
816 ia->ia_flags |= IFA_ROUTE;
818 log(LOG_INFO, "in_scrubprefix: err=%d, new prefix add failed\n",
820 ifa_free(&ia->ia_ifa);
827 * remove all L2 entries on the given prefix
829 bzero(&prefix0, sizeof(prefix0));
830 prefix0.sin_len = sizeof(prefix0);
831 prefix0.sin_family = AF_INET;
832 prefix0.sin_addr.s_addr = target->ia_subnet;
833 bzero(&mask0, sizeof(mask0));
834 mask0.sin_len = sizeof(mask0);
835 mask0.sin_family = AF_INET;
836 mask0.sin_addr.s_addr = target->ia_subnetmask;
837 lltable_prefix_free(AF_INET, (struct sockaddr *)&prefix0,
838 (struct sockaddr *)&mask0, flags);
841 * As no-one seem to have this prefix, we can remove the route.
843 error = rtinit(&(target->ia_ifa), (int)RTM_DELETE, rtinitflags(target));
845 target->ia_flags &= ~IFA_ROUTE;
847 log(LOG_INFO, "in_scrubprefix: err=%d, prefix delete failed\n", error);
854 * Return 1 if the address might be a local broadcast address.
857 in_broadcast(struct in_addr in, struct ifnet *ifp)
859 register struct ifaddr *ifa;
862 if (in.s_addr == INADDR_BROADCAST ||
863 in.s_addr == INADDR_ANY)
865 if ((ifp->if_flags & IFF_BROADCAST) == 0)
867 t = ntohl(in.s_addr);
869 * Look through the list of addresses for a match
870 * with a broadcast address.
872 #define ia ((struct in_ifaddr *)ifa)
873 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
874 if (ifa->ifa_addr->sa_family == AF_INET &&
875 (in.s_addr == ia->ia_broadaddr.sin_addr.s_addr ||
877 * Check for old-style (host 0) broadcast, but
878 * taking into account that RFC 3021 obsoletes it.
880 (ia->ia_subnetmask != IN_RFC3021_MASK &&
881 t == ia->ia_subnet)) &&
883 * Check for an all one subnetmask. These
884 * only exist when an interface gets a secondary
887 ia->ia_subnetmask != (u_long)0xffffffff)
894 * On interface removal, clean up IPv4 data structures hung off of the ifnet.
897 in_ifdetach(struct ifnet *ifp)
900 in_pcbpurgeif0(&V_ripcbinfo, ifp);
901 in_pcbpurgeif0(&V_udbinfo, ifp);
902 in_pcbpurgeif0(&V_ulitecbinfo, ifp);
907 * Delete all IPv4 multicast address records, and associated link-layer
908 * multicast address records, associated with ifp.
909 * XXX It looks like domifdetach runs AFTER the link layer cleanup.
910 * XXX This should not race with ifma_protospec being set during
911 * a new allocation, if it does, we have bigger problems.
914 in_purgemaddrs(struct ifnet *ifp)
916 LIST_HEAD(,in_multi) purgeinms;
917 struct in_multi *inm, *tinm;
918 struct ifmultiaddr *ifma;
920 LIST_INIT(&purgeinms);
924 * Extract list of in_multi associated with the detaching ifp
925 * which the PF_INET layer is about to release.
926 * We need to do this as IF_ADDR_LOCK() may be re-acquired
927 * by code further down.
930 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
931 if (ifma->ifma_addr->sa_family != AF_INET ||
932 ifma->ifma_protospec == NULL)
935 KASSERT(ifma->ifma_protospec != NULL,
936 ("%s: ifma_protospec is NULL", __func__));
938 inm = (struct in_multi *)ifma->ifma_protospec;
939 LIST_INSERT_HEAD(&purgeinms, inm, inm_link);
941 IF_ADDR_RUNLOCK(ifp);
943 LIST_FOREACH_SAFE(inm, &purgeinms, inm_link, tinm) {
944 LIST_REMOVE(inm, inm_link);
945 inm_release_locked(inm);
954 struct sockaddr_in l3_addr4;
958 * Deletes an address from the address table.
959 * This function is called by the timer functions
960 * such as arptimer() and nd6_llinfo_timer(), and
961 * the caller does the locking.
964 in_lltable_free(struct lltable *llt, struct llentry *lle)
967 LLE_LOCK_DESTROY(lle);
968 free(lle, M_LLTABLE);
971 static struct llentry *
972 in_lltable_new(const struct sockaddr *l3addr, u_int flags)
974 struct in_llentry *lle;
976 lle = malloc(sizeof(struct in_llentry), M_LLTABLE, M_NOWAIT | M_ZERO);
977 if (lle == NULL) /* NB: caller generates msg */
981 * For IPv4 this will trigger "arpresolve" to generate
984 lle->base.la_expire = time_uptime; /* mark expired */
985 lle->l3_addr4 = *(const struct sockaddr_in *)l3addr;
986 lle->base.lle_refcnt = 1;
987 lle->base.lle_free = in_lltable_free;
988 LLE_LOCK_INIT(&lle->base);
989 callout_init(&lle->base.la_timer, 1);
994 #define IN_ARE_MASKED_ADDR_EQUAL(d, a, m) ( \
995 (((ntohl((d)->sin_addr.s_addr) ^ (a)->sin_addr.s_addr) & (m)->sin_addr.s_addr)) == 0 )
998 in_lltable_prefix_free(struct lltable *llt, const struct sockaddr *prefix,
999 const struct sockaddr *mask, u_int flags)
1001 const struct sockaddr_in *pfx = (const struct sockaddr_in *)prefix;
1002 const struct sockaddr_in *msk = (const struct sockaddr_in *)mask;
1003 struct llentry *lle, *next;
1005 size_t pkts_dropped;
1007 IF_AFDATA_WLOCK(llt->llt_ifp);
1008 for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) {
1009 LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) {
1011 * (flags & LLE_STATIC) means deleting all entries
1012 * including static ARP entries.
1014 if (IN_ARE_MASKED_ADDR_EQUAL(satosin(L3_ADDR(lle)),
1015 pfx, msk) && ((flags & LLE_STATIC) ||
1016 !(lle->la_flags & LLE_STATIC))) {
1018 if (callout_stop(&lle->la_timer))
1020 pkts_dropped = llentry_free(lle);
1021 ARPSTAT_ADD(dropped, pkts_dropped);
1025 IF_AFDATA_WUNLOCK(llt->llt_ifp);
1030 in_lltable_rtcheck(struct ifnet *ifp, u_int flags, const struct sockaddr *l3addr)
1034 KASSERT(l3addr->sa_family == AF_INET,
1035 ("sin_family %d", l3addr->sa_family));
1037 /* XXX rtalloc1_fib should take a const param */
1038 rt = rtalloc1_fib(__DECONST(struct sockaddr *, l3addr), 0, 0,
1045 * If the gateway for an existing host route matches the target L3
1046 * address, which is a special route inserted by some implementation
1047 * such as MANET, and the interface is of the correct type, then
1048 * allow for ARP to proceed.
1050 if (rt->rt_flags & RTF_GATEWAY) {
1051 if (!(rt->rt_flags & RTF_HOST) || !rt->rt_ifp ||
1052 rt->rt_ifp->if_type != IFT_ETHER ||
1053 (rt->rt_ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) != 0 ||
1054 memcmp(rt->rt_gateway->sa_data, l3addr->sa_data,
1055 sizeof(in_addr_t)) != 0) {
1062 * Make sure that at least the destination address is covered
1063 * by the route. This is for handling the case where 2 or more
1064 * interfaces have the same prefix. An incoming packet arrives
1065 * on one interface and the corresponding outgoing packet leaves
1066 * another interface.
1068 if (!(rt->rt_flags & RTF_HOST) && rt->rt_ifp != ifp) {
1069 const char *sa, *mask, *addr, *lim;
1072 mask = (const char *)rt_mask(rt);
1074 * Just being extra cautious to avoid some custom
1075 * code getting into trouble.
1082 sa = (const char *)rt_key(rt);
1083 addr = (const char *)l3addr;
1084 len = ((const struct sockaddr_in *)l3addr)->sin_len;
1087 for ( ; addr < lim; sa++, mask++, addr++) {
1088 if ((*sa ^ *addr) & *mask) {
1090 log(LOG_INFO, "IPv4 address: \"%s\" is not on the network\n",
1091 inet_ntoa(((const struct sockaddr_in *)l3addr)->sin_addr));
1104 * Return NULL if not found or marked for deletion.
1105 * If found return lle read locked.
1107 static struct llentry *
1108 in_lltable_lookup(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
1110 const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
1111 struct ifnet *ifp = llt->llt_ifp;
1112 struct llentry *lle;
1113 struct llentries *lleh;
1116 IF_AFDATA_LOCK_ASSERT(ifp);
1117 KASSERT(l3addr->sa_family == AF_INET,
1118 ("sin_family %d", l3addr->sa_family));
1120 hashkey = sin->sin_addr.s_addr;
1121 lleh = &llt->lle_head[LLATBL_HASH(hashkey, LLTBL_HASHMASK)];
1122 LIST_FOREACH(lle, lleh, lle_next) {
1123 struct sockaddr_in *sa2 = satosin(L3_ADDR(lle));
1124 if (lle->la_flags & LLE_DELETED)
1126 if (sa2->sin_addr.s_addr == sin->sin_addr.s_addr)
1131 if (flags & LLE_DELETE)
1132 log(LOG_INFO, "interface address is missing from cache = %p in delete\n", lle);
1134 if (!(flags & LLE_CREATE))
1136 IF_AFDATA_WLOCK_ASSERT(ifp);
1138 * A route that covers the given address must have
1139 * been installed 1st because we are doing a resolution,
1142 if (!(flags & LLE_IFADDR) &&
1143 in_lltable_rtcheck(ifp, flags, l3addr) != 0)
1146 lle = in_lltable_new(l3addr, flags);
1148 log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
1151 lle->la_flags = flags & ~LLE_CREATE;
1152 if ((flags & (LLE_CREATE | LLE_IFADDR)) == (LLE_CREATE | LLE_IFADDR)) {
1153 bcopy(IF_LLADDR(ifp), &lle->ll_addr, ifp->if_addrlen);
1154 lle->la_flags |= (LLE_VALID | LLE_STATIC);
1158 lle->lle_head = lleh;
1159 lle->la_flags |= LLE_LINKED;
1160 LIST_INSERT_HEAD(lleh, lle, lle_next);
1161 } else if (flags & LLE_DELETE) {
1162 if (!(lle->la_flags & LLE_IFADDR) || (flags & LLE_IFADDR)) {
1164 lle->la_flags |= LLE_DELETED;
1165 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
1167 log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
1169 if ((lle->la_flags &
1170 (LLE_STATIC | LLE_IFADDR)) == LLE_STATIC)
1178 if (LLE_IS_VALID(lle)) {
1179 if (flags & LLE_EXCLUSIVE)
1189 in_lltable_dump(struct lltable *llt, struct sysctl_req *wr)
1191 #define SIN(lle) ((struct sockaddr_in *) L3_ADDR(lle))
1192 struct ifnet *ifp = llt->llt_ifp;
1193 struct llentry *lle;
1196 struct rt_msghdr rtm;
1197 struct sockaddr_in sin;
1198 struct sockaddr_dl sdl;
1202 LLTABLE_LOCK_ASSERT();
1205 for (i = 0; i < LLTBL_HASHTBL_SIZE; i++) {
1206 LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
1207 struct sockaddr_dl *sdl;
1209 /* skip deleted entries */
1210 if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
1212 /* Skip if jailed and not a valid IP of the prison. */
1213 if (prison_if(wr->td->td_ucred, L3_ADDR(lle)) != 0)
1216 * produce a msg made of:
1218 * struct sockaddr_in; (IPv4)
1219 * struct sockaddr_dl;
1221 bzero(&arpc, sizeof(arpc));
1222 arpc.rtm.rtm_msglen = sizeof(arpc);
1223 arpc.rtm.rtm_version = RTM_VERSION;
1224 arpc.rtm.rtm_type = RTM_GET;
1225 arpc.rtm.rtm_flags = RTF_UP;
1226 arpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
1227 arpc.sin.sin_family = AF_INET;
1228 arpc.sin.sin_len = sizeof(arpc.sin);
1229 arpc.sin.sin_addr.s_addr = SIN(lle)->sin_addr.s_addr;
1232 if (lle->la_flags & LLE_PUB)
1233 arpc.rtm.rtm_flags |= RTF_ANNOUNCE;
1236 sdl->sdl_family = AF_LINK;
1237 sdl->sdl_len = sizeof(*sdl);
1238 sdl->sdl_index = ifp->if_index;
1239 sdl->sdl_type = ifp->if_type;
1240 if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
1241 sdl->sdl_alen = ifp->if_addrlen;
1242 bcopy(&lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
1245 bzero(LLADDR(sdl), ifp->if_addrlen);
1248 arpc.rtm.rtm_rmx.rmx_expire =
1249 lle->la_flags & LLE_STATIC ? 0 : lle->la_expire;
1250 arpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
1251 if (lle->la_flags & LLE_STATIC)
1252 arpc.rtm.rtm_flags |= RTF_STATIC;
1253 arpc.rtm.rtm_index = ifp->if_index;
1254 error = SYSCTL_OUT(wr, &arpc, sizeof(arpc));
1264 in_domifattach(struct ifnet *ifp)
1266 struct in_ifinfo *ii;
1267 struct lltable *llt;
1269 ii = malloc(sizeof(struct in_ifinfo), M_IFADDR, M_WAITOK|M_ZERO);
1271 llt = lltable_init(ifp, AF_INET);
1273 llt->llt_prefix_free = in_lltable_prefix_free;
1274 llt->llt_lookup = in_lltable_lookup;
1275 llt->llt_dump = in_lltable_dump;
1279 ii->ii_igmp = igmp_domifattach(ifp);
1285 in_domifdetach(struct ifnet *ifp, void *aux)
1287 struct in_ifinfo *ii = (struct in_ifinfo *)aux;
1289 igmp_domifdetach(ifp);
1290 lltable_free(ii->ii_llt);