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>
49 #include <sys/rmlock.h>
50 #include <sys/sysctl.h>
51 #include <sys/syslog.h>
55 #include <net/if_var.h>
56 #include <net/if_arp.h>
57 #include <net/if_dl.h>
58 #include <net/if_llatbl.h>
59 #include <net/if_types.h>
60 #include <net/route.h>
63 #include <netinet/if_ether.h>
64 #include <netinet/in.h>
65 #include <netinet/in_var.h>
66 #include <netinet/in_pcb.h>
67 #include <netinet/ip_var.h>
68 #include <netinet/ip_carp.h>
69 #include <netinet/igmp_var.h>
70 #include <netinet/udp.h>
71 #include <netinet/udp_var.h>
73 static int in_aifaddr_ioctl(u_long, caddr_t, struct ifnet *, struct thread *);
74 static int in_difaddr_ioctl(caddr_t, struct ifnet *, struct thread *);
76 static void in_socktrim(struct sockaddr_in *);
77 static void in_purgemaddrs(struct ifnet *);
79 static VNET_DEFINE(int, nosameprefix);
80 #define V_nosameprefix VNET(nosameprefix)
81 SYSCTL_INT(_net_inet_ip, OID_AUTO, no_same_prefix, CTLFLAG_VNET | 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 static struct sx in_control_sx;
89 SX_SYSINIT(in_control_sx, &in_control_sx, "in_control");
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 struct rm_priotracker in_ifa_tracker;
99 register u_long i = ntohl(in.s_addr);
100 register struct in_ifaddr *ia;
102 IN_IFADDR_RLOCK(&in_ifa_tracker);
103 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
104 if ((i & ia->ia_subnetmask) == ia->ia_subnet) {
105 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
109 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
114 * Return 1 if an internet address is for the local host and configured
115 * on one of its interfaces.
118 in_localip(struct in_addr in)
120 struct rm_priotracker in_ifa_tracker;
121 struct in_ifaddr *ia;
123 IN_IFADDR_RLOCK(&in_ifa_tracker);
124 LIST_FOREACH(ia, INADDR_HASH(in.s_addr), ia_hash) {
125 if (IA_SIN(ia)->sin_addr.s_addr == in.s_addr) {
126 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
130 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
135 * Return 1 if an internet address is configured on an interface.
138 in_ifhasaddr(struct ifnet *ifp, struct in_addr in)
141 struct in_ifaddr *ia;
144 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
145 if (ifa->ifa_addr->sa_family != AF_INET)
147 ia = (struct in_ifaddr *)ifa;
148 if (ia->ia_addr.sin_addr.s_addr == in.s_addr) {
149 IF_ADDR_RUNLOCK(ifp);
153 IF_ADDR_RUNLOCK(ifp);
159 * Return a reference to the interface address which is different to
160 * the supplied one but with same IP address value.
162 static struct in_ifaddr *
163 in_localip_more(struct in_ifaddr *ia)
165 struct rm_priotracker in_ifa_tracker;
166 in_addr_t in = IA_SIN(ia)->sin_addr.s_addr;
167 struct in_ifaddr *it;
169 IN_IFADDR_RLOCK(&in_ifa_tracker);
170 LIST_FOREACH(it, INADDR_HASH(in), ia_hash) {
171 if (it != ia && IA_SIN(it)->sin_addr.s_addr == in) {
172 ifa_ref(&it->ia_ifa);
173 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
177 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
183 * Determine whether an IP address is in a reserved set of addresses
184 * that may not be forwarded, or whether datagrams to that destination
188 in_canforward(struct in_addr in)
190 register u_long i = ntohl(in.s_addr);
193 if (IN_EXPERIMENTAL(i) || IN_MULTICAST(i) || IN_LINKLOCAL(i))
196 net = i & IN_CLASSA_NET;
197 if (net == 0 || net == (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))
204 * Trim a mask in a sockaddr
207 in_socktrim(struct sockaddr_in *ap)
209 register char *cplim = (char *) &ap->sin_addr;
210 register char *cp = (char *) (&ap->sin_addr + 1);
213 while (--cp >= cplim)
215 (ap)->sin_len = cp - (char *) (ap) + 1;
221 * Generic internet control operations (ioctl's).
224 in_control(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp,
227 struct ifreq *ifr = (struct ifreq *)data;
228 struct sockaddr_in *addr = (struct sockaddr_in *)&ifr->ifr_addr;
230 struct in_ifaddr *ia;
234 return (EADDRNOTAVAIL);
237 * Filter out 4 ioctls we implement directly. Forward the rest
238 * to specific functions and ifp->if_ioctl().
247 sx_xlock(&in_control_sx);
248 error = in_difaddr_ioctl(data, ifp, td);
249 sx_xunlock(&in_control_sx);
251 case OSIOCAIFADDR: /* 9.x compat */
253 sx_xlock(&in_control_sx);
254 error = in_aifaddr_ioctl(cmd, data, ifp, td);
255 sx_xunlock(&in_control_sx);
261 /* We no longer support that old commands. */
264 if (ifp->if_ioctl == NULL)
266 return ((*ifp->if_ioctl)(ifp, cmd, data));
269 if (addr->sin_addr.s_addr != INADDR_ANY &&
270 prison_check_ip4(td->td_ucred, &addr->sin_addr) != 0)
271 return (EADDRNOTAVAIL);
274 * Find address for this interface, if it exists. If an
275 * address was specified, find that one instead of the
276 * first one on the interface, if possible.
279 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
280 if (ifa->ifa_addr->sa_family != AF_INET)
282 ia = (struct in_ifaddr *)ifa;
283 if (ia->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr)
287 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
288 if (ifa->ifa_addr->sa_family == AF_INET) {
289 ia = (struct in_ifaddr *)ifa;
290 if (prison_check_ip4(td->td_ucred,
291 &ia->ia_addr.sin_addr) == 0)
296 IF_ADDR_RUNLOCK(ifp);
297 return (EADDRNOTAVAIL);
307 if ((ifp->if_flags & IFF_BROADCAST) == 0) {
311 *addr = ia->ia_broadaddr;
315 if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
319 *addr = ia->ia_dstaddr;
323 *addr = ia->ia_sockmask;
327 IF_ADDR_RUNLOCK(ifp);
333 in_aifaddr_ioctl(u_long cmd, caddr_t data, struct ifnet *ifp, struct thread *td)
335 const struct in_aliasreq *ifra = (struct in_aliasreq *)data;
336 const struct sockaddr_in *addr = &ifra->ifra_addr;
337 const struct sockaddr_in *broadaddr = &ifra->ifra_broadaddr;
338 const struct sockaddr_in *mask = &ifra->ifra_mask;
339 const struct sockaddr_in *dstaddr = &ifra->ifra_dstaddr;
340 const int vhid = (cmd == SIOCAIFADDR) ? ifra->ifra_vhid : 0;
342 struct in_ifaddr *ia;
346 error = priv_check(td, PRIV_NET_ADDIFADDR);
351 * ifra_addr must be present and be of INET family.
352 * ifra_broadaddr/ifra_dstaddr and ifra_mask are optional.
354 if (addr->sin_len != sizeof(struct sockaddr_in) ||
355 addr->sin_family != AF_INET)
357 if (broadaddr->sin_len != 0 &&
358 (broadaddr->sin_len != sizeof(struct sockaddr_in) ||
359 broadaddr->sin_family != AF_INET))
361 if (mask->sin_len != 0 &&
362 (mask->sin_len != sizeof(struct sockaddr_in) ||
363 mask->sin_family != AF_INET))
365 if ((ifp->if_flags & IFF_POINTOPOINT) &&
366 (dstaddr->sin_len != sizeof(struct sockaddr_in) ||
367 dstaddr->sin_addr.s_addr == INADDR_ANY))
368 return (EDESTADDRREQ);
369 if (vhid > 0 && carp_attach_p == NULL)
370 return (EPROTONOSUPPORT);
373 * See whether address already exist.
378 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
379 struct in_ifaddr *it;
381 if (ifa->ifa_addr->sa_family != AF_INET)
384 it = (struct in_ifaddr *)ifa;
386 if (it->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
387 prison_check_ip4(td->td_ucred, &addr->sin_addr) == 0)
390 IF_ADDR_RUNLOCK(ifp);
393 (void )in_difaddr_ioctl(data, ifp, td);
395 ifa = ifa_alloc(sizeof(struct in_ifaddr), M_WAITOK);
396 ia = (struct in_ifaddr *)ifa;
397 ifa->ifa_addr = (struct sockaddr *)&ia->ia_addr;
398 ifa->ifa_dstaddr = (struct sockaddr *)&ia->ia_dstaddr;
399 ifa->ifa_netmask = (struct sockaddr *)&ia->ia_sockmask;
403 if (mask->sin_len != 0) {
404 ia->ia_sockmask = *mask;
405 ia->ia_subnetmask = ntohl(ia->ia_sockmask.sin_addr.s_addr);
407 in_addr_t i = ntohl(addr->sin_addr.s_addr);
410 * Be compatible with network classes, if netmask isn't
411 * supplied, guess it based on classes.
414 ia->ia_subnetmask = IN_CLASSA_NET;
415 else if (IN_CLASSB(i))
416 ia->ia_subnetmask = IN_CLASSB_NET;
418 ia->ia_subnetmask = IN_CLASSC_NET;
419 ia->ia_sockmask.sin_addr.s_addr = htonl(ia->ia_subnetmask);
421 ia->ia_subnet = ntohl(addr->sin_addr.s_addr) & ia->ia_subnetmask;
422 in_socktrim(&ia->ia_sockmask);
424 if (ifp->if_flags & IFF_BROADCAST) {
425 if (broadaddr->sin_len != 0) {
426 ia->ia_broadaddr = *broadaddr;
427 } else if (ia->ia_subnetmask == IN_RFC3021_MASK) {
428 ia->ia_broadaddr.sin_addr.s_addr = INADDR_BROADCAST;
429 ia->ia_broadaddr.sin_len = sizeof(struct sockaddr_in);
430 ia->ia_broadaddr.sin_family = AF_INET;
432 ia->ia_broadaddr.sin_addr.s_addr =
433 htonl(ia->ia_subnet | ~ia->ia_subnetmask);
434 ia->ia_broadaddr.sin_len = sizeof(struct sockaddr_in);
435 ia->ia_broadaddr.sin_family = AF_INET;
439 if (ifp->if_flags & IFF_POINTOPOINT)
440 ia->ia_dstaddr = *dstaddr;
442 /* XXXGL: rtinit() needs this strange assignment. */
443 if (ifp->if_flags & IFF_LOOPBACK)
444 ia->ia_dstaddr = ia->ia_addr;
447 error = (*carp_attach_p)(&ia->ia_ifa, vhid);
452 /* if_addrhead is already referenced by ifa_alloc() */
454 TAILQ_INSERT_TAIL(&ifp->if_addrhead, ifa, ifa_link);
455 IF_ADDR_WUNLOCK(ifp);
457 ifa_ref(ifa); /* in_ifaddrhead */
459 TAILQ_INSERT_TAIL(&V_in_ifaddrhead, ia, ia_link);
460 LIST_INSERT_HEAD(INADDR_HASH(ia->ia_addr.sin_addr.s_addr), ia, ia_hash);
464 * Give the interface a chance to initialize
465 * if this is its first address,
466 * and to validate the address if necessary.
468 if (ifp->if_ioctl != NULL) {
469 error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
475 * Add route for the network.
480 if (ifp->if_flags & (IFF_LOOPBACK|IFF_POINTOPOINT))
483 error = in_addprefix(ia, flags);
489 * Add a loopback route to self.
491 if (vhid == 0 && (ifp->if_flags & IFF_LOOPBACK) == 0 &&
492 ia->ia_addr.sin_addr.s_addr != INADDR_ANY &&
493 !((ifp->if_flags & IFF_POINTOPOINT) &&
494 ia->ia_dstaddr.sin_addr.s_addr == ia->ia_addr.sin_addr.s_addr)) {
495 struct in_ifaddr *eia;
497 eia = in_localip_more(ia);
500 error = ifa_add_loopback_route((struct ifaddr *)ia,
501 (struct sockaddr *)&ia->ia_addr);
505 ifa_free(&eia->ia_ifa);
508 if (iaIsFirst && (ifp->if_flags & IFF_MULTICAST)) {
509 struct in_addr allhosts_addr;
510 struct in_ifinfo *ii;
512 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
513 allhosts_addr.s_addr = htonl(INADDR_ALLHOSTS_GROUP);
515 error = in_joingroup(ifp, &allhosts_addr, NULL,
519 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
525 (void )in_scrubprefix(ia, LLE_STATIC);
528 if (ia->ia_ifa.ifa_carp)
529 (*carp_detach_p)(&ia->ia_ifa);
532 TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
533 IF_ADDR_WUNLOCK(ifp);
534 ifa_free(&ia->ia_ifa); /* if_addrhead */
537 TAILQ_REMOVE(&V_in_ifaddrhead, ia, ia_link);
538 LIST_REMOVE(ia, ia_hash);
540 ifa_free(&ia->ia_ifa); /* in_ifaddrhead */
546 in_difaddr_ioctl(caddr_t data, struct ifnet *ifp, struct thread *td)
548 const struct ifreq *ifr = (struct ifreq *)data;
549 const struct sockaddr_in *addr = (const struct sockaddr_in *)
552 struct in_ifaddr *ia;
553 bool deleteAny, iaIsLast;
557 error = priv_check(td, PRIV_NET_DELIFADDR);
562 if (addr->sin_len != sizeof(struct sockaddr_in) ||
563 addr->sin_family != AF_INET)
571 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
572 struct in_ifaddr *it;
574 if (ifa->ifa_addr->sa_family != AF_INET)
577 it = (struct in_ifaddr *)ifa;
578 if (deleteAny && ia == NULL && (td == NULL ||
579 prison_check_ip4(td->td_ucred, &it->ia_addr.sin_addr) == 0))
582 if (it->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
583 (td == NULL || prison_check_ip4(td->td_ucred,
584 &addr->sin_addr) == 0))
592 IF_ADDR_WUNLOCK(ifp);
593 return (EADDRNOTAVAIL);
596 TAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
597 IF_ADDR_WUNLOCK(ifp);
598 ifa_free(&ia->ia_ifa); /* if_addrhead */
601 TAILQ_REMOVE(&V_in_ifaddrhead, ia, ia_link);
602 LIST_REMOVE(ia, ia_hash);
606 * in_scrubprefix() kills the interface route.
608 in_scrubprefix(ia, LLE_STATIC);
611 * in_ifadown gets rid of all the rest of
612 * the routes. This is not quite the right
613 * thing to do, but at least if we are running
614 * a routing process they will come back.
616 in_ifadown(&ia->ia_ifa, 1);
618 if (ia->ia_ifa.ifa_carp)
619 (*carp_detach_p)(&ia->ia_ifa);
622 * If this is the last IPv4 address configured on this
623 * interface, leave the all-hosts group.
624 * No state-change report need be transmitted.
626 if (iaIsLast && (ifp->if_flags & IFF_MULTICAST)) {
627 struct in_ifinfo *ii;
629 ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
631 if (ii->ii_allhosts) {
632 (void)in_leavegroup_locked(ii->ii_allhosts, NULL);
633 ii->ii_allhosts = NULL;
638 EVENTHANDLER_INVOKE(ifaddr_event, ifp);
639 ifa_free(&ia->ia_ifa); /* in_ifaddrhead */
644 #define rtinitflags(x) \
645 ((((x)->ia_ifp->if_flags & (IFF_LOOPBACK | IFF_POINTOPOINT)) != 0) \
649 * Check if we have a route for the given prefix already or add one accordingly.
652 in_addprefix(struct in_ifaddr *target, int flags)
654 struct rm_priotracker in_ifa_tracker;
655 struct in_ifaddr *ia;
656 struct in_addr prefix, mask, p, m;
659 if ((flags & RTF_HOST) != 0) {
660 prefix = target->ia_dstaddr.sin_addr;
663 prefix = target->ia_addr.sin_addr;
664 mask = target->ia_sockmask.sin_addr;
665 prefix.s_addr &= mask.s_addr;
668 IN_IFADDR_RLOCK(&in_ifa_tracker);
669 /* Look for an existing address with the same prefix, mask, and fib */
670 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
671 if (rtinitflags(ia)) {
672 p = ia->ia_dstaddr.sin_addr;
674 if (prefix.s_addr != p.s_addr)
677 p = ia->ia_addr.sin_addr;
678 m = ia->ia_sockmask.sin_addr;
679 p.s_addr &= m.s_addr;
681 if (prefix.s_addr != p.s_addr ||
682 mask.s_addr != m.s_addr)
685 if (target->ia_ifp->if_fib != ia->ia_ifp->if_fib)
689 * If we got a matching prefix route inserted by other
690 * interface address, we are done here.
692 if (ia->ia_flags & IFA_ROUTE) {
694 if (ia->ia_addr.sin_addr.s_addr ==
695 target->ia_addr.sin_addr.s_addr) {
696 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
701 if (V_nosameprefix) {
702 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
707 fibnum = V_rt_add_addr_allfibs ? RT_ALL_FIBS :
708 target->ia_ifp->if_fib;
709 rt_addrmsg(RTM_ADD, &target->ia_ifa, fibnum);
710 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
715 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
718 * No-one seem to have this prefix route, so we try to insert it.
720 error = rtinit(&target->ia_ifa, (int)RTM_ADD, flags);
722 target->ia_flags |= IFA_ROUTE;
727 * Removes either all lle entries for given @ia, or lle
728 * corresponding to @ia address.
731 in_scrubprefixlle(struct in_ifaddr *ia, int all, u_int flags)
733 struct sockaddr_in addr, mask;
734 struct sockaddr *saddr, *smask;
737 saddr = (struct sockaddr *)&addr;
738 bzero(&addr, sizeof(addr));
739 addr.sin_len = sizeof(addr);
740 addr.sin_family = AF_INET;
741 smask = (struct sockaddr *)&mask;
742 bzero(&mask, sizeof(mask));
743 mask.sin_len = sizeof(mask);
744 mask.sin_family = AF_INET;
745 mask.sin_addr.s_addr = ia->ia_subnetmask;
751 * Remove all L2 entries matching given prefix.
752 * Convert address to host representation to avoid
753 * doing this on every callback. ia_subnetmask is already
754 * stored in host representation.
756 addr.sin_addr.s_addr = ntohl(ia->ia_addr.sin_addr.s_addr);
757 lltable_prefix_free(AF_INET, saddr, smask, flags);
759 /* Remove interface address only */
760 addr.sin_addr.s_addr = ia->ia_addr.sin_addr.s_addr;
761 lltable_delete_addr(LLTABLE(ifp), LLE_IFADDR, saddr);
766 * If there is no other address in the system that can serve a route to the
767 * same prefix, remove the route. Hand over the route to the new address
771 in_scrubprefix(struct in_ifaddr *target, u_int flags)
773 struct rm_priotracker in_ifa_tracker;
774 struct in_ifaddr *ia;
775 struct in_addr prefix, mask, p, m;
779 * Remove the loopback route to the interface address.
781 if ((target->ia_addr.sin_addr.s_addr != INADDR_ANY) &&
782 !(target->ia_ifp->if_flags & IFF_LOOPBACK) &&
783 (flags & LLE_STATIC)) {
784 struct in_ifaddr *eia;
787 * XXXME: add fib-aware in_localip.
788 * We definitely don't want to switch between
789 * prefixes in different fibs.
791 eia = in_localip_more(target);
794 error = ifa_switch_loopback_route((struct ifaddr *)eia,
795 (struct sockaddr *)&target->ia_addr);
796 ifa_free(&eia->ia_ifa);
798 error = ifa_del_loopback_route((struct ifaddr *)target,
799 (struct sockaddr *)&target->ia_addr);
803 if (rtinitflags(target)) {
804 prefix = target->ia_dstaddr.sin_addr;
807 prefix = target->ia_addr.sin_addr;
808 mask = target->ia_sockmask.sin_addr;
809 prefix.s_addr &= mask.s_addr;
812 if ((target->ia_flags & IFA_ROUTE) == 0) {
815 fibnum = V_rt_add_addr_allfibs ? RT_ALL_FIBS :
816 target->ia_ifp->if_fib;
817 rt_addrmsg(RTM_DELETE, &target->ia_ifa, fibnum);
820 * Removing address from !IFF_UP interface or
821 * prefix which exists on other interface (along with route).
822 * No entries should exist here except target addr.
823 * Given that, delete this entry only.
825 in_scrubprefixlle(target, 0, flags);
829 IN_IFADDR_RLOCK(&in_ifa_tracker);
830 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
831 if (rtinitflags(ia)) {
832 p = ia->ia_dstaddr.sin_addr;
834 if (prefix.s_addr != p.s_addr)
837 p = ia->ia_addr.sin_addr;
838 m = ia->ia_sockmask.sin_addr;
839 p.s_addr &= m.s_addr;
841 if (prefix.s_addr != p.s_addr ||
842 mask.s_addr != m.s_addr)
846 if ((ia->ia_ifp->if_flags & IFF_UP) == 0)
850 * If we got a matching prefix address, move IFA_ROUTE and
851 * the route itself to it. Make sure that routing daemons
854 if ((ia->ia_flags & IFA_ROUTE) == 0) {
855 ifa_ref(&ia->ia_ifa);
856 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
857 error = rtinit(&(target->ia_ifa), (int)RTM_DELETE,
858 rtinitflags(target));
860 target->ia_flags &= ~IFA_ROUTE;
862 log(LOG_INFO, "in_scrubprefix: err=%d, old prefix delete failed\n",
864 /* Scrub all entries IFF interface is different */
865 in_scrubprefixlle(target, target->ia_ifp != ia->ia_ifp,
867 error = rtinit(&ia->ia_ifa, (int)RTM_ADD,
868 rtinitflags(ia) | RTF_UP);
870 ia->ia_flags |= IFA_ROUTE;
872 log(LOG_INFO, "in_scrubprefix: err=%d, new prefix add failed\n",
874 ifa_free(&ia->ia_ifa);
878 IN_IFADDR_RUNLOCK(&in_ifa_tracker);
881 * remove all L2 entries on the given prefix
883 in_scrubprefixlle(target, 1, flags);
886 * As no-one seem to have this prefix, we can remove the route.
888 error = rtinit(&(target->ia_ifa), (int)RTM_DELETE, rtinitflags(target));
890 target->ia_flags &= ~IFA_ROUTE;
892 log(LOG_INFO, "in_scrubprefix: err=%d, prefix delete failed\n", error);
899 * Return 1 if the address might be a local broadcast address.
902 in_broadcast(struct in_addr in, struct ifnet *ifp)
904 register struct ifaddr *ifa;
907 if (in.s_addr == INADDR_BROADCAST ||
908 in.s_addr == INADDR_ANY)
910 if ((ifp->if_flags & IFF_BROADCAST) == 0)
912 t = ntohl(in.s_addr);
914 * Look through the list of addresses for a match
915 * with a broadcast address.
917 #define ia ((struct in_ifaddr *)ifa)
918 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
919 if (ifa->ifa_addr->sa_family == AF_INET &&
920 (in.s_addr == ia->ia_broadaddr.sin_addr.s_addr ||
922 * Check for old-style (host 0) broadcast, but
923 * taking into account that RFC 3021 obsoletes it.
925 (ia->ia_subnetmask != IN_RFC3021_MASK &&
926 t == ia->ia_subnet)) &&
928 * Check for an all one subnetmask. These
929 * only exist when an interface gets a secondary
932 ia->ia_subnetmask != (u_long)0xffffffff)
939 * On interface removal, clean up IPv4 data structures hung off of the ifnet.
942 in_ifdetach(struct ifnet *ifp)
945 in_pcbpurgeif0(&V_ripcbinfo, ifp);
946 in_pcbpurgeif0(&V_udbinfo, ifp);
947 in_pcbpurgeif0(&V_ulitecbinfo, ifp);
952 * Delete all IPv4 multicast address records, and associated link-layer
953 * multicast address records, associated with ifp.
954 * XXX It looks like domifdetach runs AFTER the link layer cleanup.
955 * XXX This should not race with ifma_protospec being set during
956 * a new allocation, if it does, we have bigger problems.
959 in_purgemaddrs(struct ifnet *ifp)
961 LIST_HEAD(,in_multi) purgeinms;
962 struct in_multi *inm, *tinm;
963 struct ifmultiaddr *ifma;
965 LIST_INIT(&purgeinms);
969 * Extract list of in_multi associated with the detaching ifp
970 * which the PF_INET layer is about to release.
971 * We need to do this as IF_ADDR_LOCK() may be re-acquired
972 * by code further down.
975 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
976 if (ifma->ifma_addr->sa_family != AF_INET ||
977 ifma->ifma_protospec == NULL)
980 KASSERT(ifma->ifma_protospec != NULL,
981 ("%s: ifma_protospec is NULL", __func__));
983 inm = (struct in_multi *)ifma->ifma_protospec;
984 LIST_INSERT_HEAD(&purgeinms, inm, inm_link);
986 IF_ADDR_RUNLOCK(ifp);
988 LIST_FOREACH_SAFE(inm, &purgeinms, inm_link, tinm) {
989 LIST_REMOVE(inm, inm_link);
990 inm_release_locked(inm);
1001 #define IN_LLTBL_DEFAULT_HSIZE 32
1002 #define IN_LLTBL_HASH(k, h) \
1003 (((((((k >> 8) ^ k) >> 8) ^ k) >> 8) ^ k) & ((h) - 1))
1006 * Do actual deallocation of @lle.
1009 in_lltable_destroy_lle_unlocked(struct llentry *lle)
1012 LLE_LOCK_DESTROY(lle);
1013 LLE_REQ_DESTROY(lle);
1014 free(lle, M_LLTABLE);
1018 * Called by LLE_FREE_LOCKED when number of references
1022 in_lltable_destroy_lle(struct llentry *lle)
1026 in_lltable_destroy_lle_unlocked(lle);
1029 static struct llentry *
1030 in_lltable_new(struct in_addr addr4, u_int flags)
1032 struct in_llentry *lle;
1034 lle = malloc(sizeof(struct in_llentry), M_LLTABLE, M_NOWAIT | M_ZERO);
1035 if (lle == NULL) /* NB: caller generates msg */
1039 * For IPv4 this will trigger "arpresolve" to generate
1042 lle->base.la_expire = time_uptime; /* mark expired */
1043 lle->base.r_l3addr.addr4 = addr4;
1044 lle->base.lle_refcnt = 1;
1045 lle->base.lle_free = in_lltable_destroy_lle;
1046 LLE_LOCK_INIT(&lle->base);
1047 LLE_REQ_INIT(&lle->base);
1048 callout_init(&lle->base.lle_timer, 1);
1050 return (&lle->base);
1053 #define IN_ARE_MASKED_ADDR_EQUAL(d, a, m) ( \
1054 ((((d).s_addr ^ (a).s_addr) & (m).s_addr)) == 0 )
1057 in_lltable_match_prefix(const struct sockaddr *saddr,
1058 const struct sockaddr *smask, u_int flags, struct llentry *lle)
1060 struct in_addr addr, mask, lle_addr;
1062 addr = ((const struct sockaddr_in *)saddr)->sin_addr;
1063 mask = ((const struct sockaddr_in *)smask)->sin_addr;
1064 lle_addr.s_addr = ntohl(lle->r_l3addr.addr4.s_addr);
1066 if (IN_ARE_MASKED_ADDR_EQUAL(lle_addr, addr, mask) == 0)
1069 if (lle->la_flags & LLE_IFADDR) {
1072 * Delete LLE_IFADDR records IFF address & flag matches.
1073 * Note that addr is the interface address within prefix
1075 * Note also we should handle 'ifdown' cases without removing
1078 if (addr.s_addr == lle_addr.s_addr && (flags & LLE_STATIC) != 0)
1083 /* flags & LLE_STATIC means deleting both dynamic and static entries */
1084 if ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))
1091 in_lltable_free_entry(struct lltable *llt, struct llentry *lle)
1094 size_t pkts_dropped;
1096 LLE_WLOCK_ASSERT(lle);
1097 KASSERT(llt != NULL, ("lltable is NULL"));
1099 /* Unlink entry from table if not already */
1100 if ((lle->la_flags & LLE_LINKED) != 0) {
1102 IF_AFDATA_WLOCK_ASSERT(ifp);
1103 lltable_unlink_entry(llt, lle);
1107 if (callout_stop(&lle->lle_timer) > 0)
1110 /* Drop hold queue */
1111 pkts_dropped = llentry_free(lle);
1112 ARPSTAT_ADD(dropped, pkts_dropped);
1116 in_lltable_rtcheck(struct ifnet *ifp, u_int flags, const struct sockaddr *l3addr)
1118 struct rt_addrinfo info;
1119 struct sockaddr_in rt_key, rt_mask;
1120 struct sockaddr rt_gateway;
1123 KASSERT(l3addr->sa_family == AF_INET,
1124 ("sin_family %d", l3addr->sa_family));
1126 bzero(&rt_key, sizeof(rt_key));
1127 rt_key.sin_len = sizeof(rt_key);
1128 bzero(&rt_mask, sizeof(rt_mask));
1129 rt_mask.sin_len = sizeof(rt_mask);
1130 bzero(&rt_gateway, sizeof(rt_gateway));
1131 rt_gateway.sa_len = sizeof(rt_gateway);
1133 bzero(&info, sizeof(info));
1134 info.rti_info[RTAX_DST] = (struct sockaddr *)&rt_key;
1135 info.rti_info[RTAX_NETMASK] = (struct sockaddr *)&rt_mask;
1136 info.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&rt_gateway;
1138 if (rib_lookup_info(ifp->if_fib, l3addr, NHR_REF, 0, &info) != 0)
1141 rt_flags = info.rti_flags;
1144 * If the gateway for an existing host route matches the target L3
1145 * address, which is a special route inserted by some implementation
1146 * such as MANET, and the interface is of the correct type, then
1147 * allow for ARP to proceed.
1149 if (rt_flags & RTF_GATEWAY) {
1150 if (!(rt_flags & RTF_HOST) || !info.rti_ifp ||
1151 info.rti_ifp->if_type != IFT_ETHER ||
1152 (info.rti_ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) != 0 ||
1153 memcmp(rt_gateway.sa_data, l3addr->sa_data,
1154 sizeof(in_addr_t)) != 0) {
1155 rib_free_info(&info);
1159 rib_free_info(&info);
1162 * Make sure that at least the destination address is covered
1163 * by the route. This is for handling the case where 2 or more
1164 * interfaces have the same prefix. An incoming packet arrives
1165 * on one interface and the corresponding outgoing packet leaves
1166 * another interface.
1168 if (!(rt_flags & RTF_HOST) && info.rti_ifp != ifp) {
1169 const char *sa, *mask, *addr, *lim;
1172 mask = (const char *)&rt_mask;
1174 * Just being extra cautious to avoid some custom
1175 * code getting into trouble.
1177 if ((info.rti_addrs & RTA_NETMASK) == 0)
1180 sa = (const char *)&rt_key;
1181 addr = (const char *)l3addr;
1182 len = ((const struct sockaddr_in *)l3addr)->sin_len;
1185 for ( ; addr < lim; sa++, mask++, addr++) {
1186 if ((*sa ^ *addr) & *mask) {
1188 log(LOG_INFO, "IPv4 address: \"%s\" is not on the network\n",
1189 inet_ntoa(((const struct sockaddr_in *)l3addr)->sin_addr));
1199 static inline uint32_t
1200 in_lltable_hash_dst(const struct in_addr dst, uint32_t hsize)
1203 return (IN_LLTBL_HASH(dst.s_addr, hsize));
1207 in_lltable_hash(const struct llentry *lle, uint32_t hsize)
1210 return (in_lltable_hash_dst(lle->r_l3addr.addr4, hsize));
1214 in_lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
1216 struct sockaddr_in *sin;
1218 sin = (struct sockaddr_in *)sa;
1219 bzero(sin, sizeof(*sin));
1220 sin->sin_family = AF_INET;
1221 sin->sin_len = sizeof(*sin);
1222 sin->sin_addr = lle->r_l3addr.addr4;
1225 static inline struct llentry *
1226 in_lltable_find_dst(struct lltable *llt, struct in_addr dst)
1228 struct llentry *lle;
1229 struct llentries *lleh;
1232 hashidx = in_lltable_hash_dst(dst, llt->llt_hsize);
1233 lleh = &llt->lle_head[hashidx];
1234 LIST_FOREACH(lle, lleh, lle_next) {
1235 if (lle->la_flags & LLE_DELETED)
1237 if (lle->r_l3addr.addr4.s_addr == dst.s_addr)
1245 in_lltable_delete_entry(struct lltable *llt, struct llentry *lle)
1248 lle->la_flags |= LLE_DELETED;
1249 EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
1251 log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
1256 static struct llentry *
1257 in_lltable_alloc(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
1259 const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
1260 struct ifnet *ifp = llt->llt_ifp;
1261 struct llentry *lle;
1262 char linkhdr[LLE_MAX_LINKHDR];
1266 KASSERT(l3addr->sa_family == AF_INET,
1267 ("sin_family %d", l3addr->sa_family));
1270 * A route that covers the given address must have
1271 * been installed 1st because we are doing a resolution,
1274 if (!(flags & LLE_IFADDR) &&
1275 in_lltable_rtcheck(ifp, flags, l3addr) != 0)
1278 lle = in_lltable_new(sin->sin_addr, flags);
1280 log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
1283 lle->la_flags = flags;
1284 if (flags & LLE_STATIC)
1285 lle->r_flags |= RLLE_VALID;
1286 if ((flags & LLE_IFADDR) == LLE_IFADDR) {
1287 linkhdrsize = LLE_MAX_LINKHDR;
1288 if (lltable_calc_llheader(ifp, AF_INET, IF_LLADDR(ifp),
1289 linkhdr, &linkhdrsize, &lladdr_off) != 0) {
1290 in_lltable_destroy_lle_unlocked(lle);
1293 lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
1295 lle->la_flags |= LLE_STATIC;
1296 lle->r_flags |= (RLLE_VALID | RLLE_IFADDR);
1303 * Return NULL if not found or marked for deletion.
1304 * If found return lle read locked.
1306 static struct llentry *
1307 in_lltable_lookup(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
1309 const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
1310 struct llentry *lle;
1312 IF_AFDATA_LOCK_ASSERT(llt->llt_ifp);
1313 KASSERT(l3addr->sa_family == AF_INET,
1314 ("sin_family %d", l3addr->sa_family));
1315 lle = in_lltable_find_dst(llt, sin->sin_addr);
1320 KASSERT((flags & (LLE_UNLOCKED|LLE_EXCLUSIVE)) !=
1321 (LLE_UNLOCKED|LLE_EXCLUSIVE),("wrong lle request flags: 0x%X",
1324 if (flags & LLE_UNLOCKED)
1327 if (flags & LLE_EXCLUSIVE)
1336 in_lltable_dump_entry(struct lltable *llt, struct llentry *lle,
1337 struct sysctl_req *wr)
1339 struct ifnet *ifp = llt->llt_ifp;
1342 struct rt_msghdr rtm;
1343 struct sockaddr_in sin;
1344 struct sockaddr_dl sdl;
1346 struct sockaddr_dl *sdl;
1349 bzero(&arpc, sizeof(arpc));
1350 /* skip deleted entries */
1351 if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
1353 /* Skip if jailed and not a valid IP of the prison. */
1354 lltable_fill_sa_entry(lle,(struct sockaddr *)&arpc.sin);
1355 if (prison_if(wr->td->td_ucred,
1356 (struct sockaddr *)&arpc.sin) != 0)
1359 * produce a msg made of:
1361 * struct sockaddr_in; (IPv4)
1362 * struct sockaddr_dl;
1364 arpc.rtm.rtm_msglen = sizeof(arpc);
1365 arpc.rtm.rtm_version = RTM_VERSION;
1366 arpc.rtm.rtm_type = RTM_GET;
1367 arpc.rtm.rtm_flags = RTF_UP;
1368 arpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
1371 if (lle->la_flags & LLE_PUB)
1372 arpc.rtm.rtm_flags |= RTF_ANNOUNCE;
1375 sdl->sdl_family = AF_LINK;
1376 sdl->sdl_len = sizeof(*sdl);
1377 sdl->sdl_index = ifp->if_index;
1378 sdl->sdl_type = ifp->if_type;
1379 if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
1380 sdl->sdl_alen = ifp->if_addrlen;
1381 bcopy(lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
1384 bzero(LLADDR(sdl), ifp->if_addrlen);
1387 arpc.rtm.rtm_rmx.rmx_expire =
1388 lle->la_flags & LLE_STATIC ? 0 : lle->la_expire;
1389 arpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
1390 if (lle->la_flags & LLE_STATIC)
1391 arpc.rtm.rtm_flags |= RTF_STATIC;
1392 if (lle->la_flags & LLE_IFADDR)
1393 arpc.rtm.rtm_flags |= RTF_PINNED;
1394 arpc.rtm.rtm_index = ifp->if_index;
1395 error = SYSCTL_OUT(wr, &arpc, sizeof(arpc));
1400 static struct lltable *
1401 in_lltattach(struct ifnet *ifp)
1403 struct lltable *llt;
1405 llt = lltable_allocate_htbl(IN_LLTBL_DEFAULT_HSIZE);
1406 llt->llt_af = AF_INET;
1409 llt->llt_lookup = in_lltable_lookup;
1410 llt->llt_alloc_entry = in_lltable_alloc;
1411 llt->llt_delete_entry = in_lltable_delete_entry;
1412 llt->llt_dump_entry = in_lltable_dump_entry;
1413 llt->llt_hash = in_lltable_hash;
1414 llt->llt_fill_sa_entry = in_lltable_fill_sa_entry;
1415 llt->llt_free_entry = in_lltable_free_entry;
1416 llt->llt_match_prefix = in_lltable_match_prefix;
1423 in_domifattach(struct ifnet *ifp)
1425 struct in_ifinfo *ii;
1427 ii = malloc(sizeof(struct in_ifinfo), M_IFADDR, M_WAITOK|M_ZERO);
1429 ii->ii_llt = in_lltattach(ifp);
1430 ii->ii_igmp = igmp_domifattach(ifp);
1436 in_domifdetach(struct ifnet *ifp, void *aux)
1438 struct in_ifinfo *ii = (struct in_ifinfo *)aux;
1440 igmp_domifdetach(ifp);
1441 lltable_free(ii->ii_llt);