2 * Copyright (c) 2004, 2005 The DragonFly Project. All rights reserved.
4 * This code is derived from software contributed to The DragonFly Project
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of The DragonFly Project nor the names of its
16 * contributors may be used to endorse or promote products derived
17 * from this software without specific, prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
22 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
23 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
24 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
25 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
26 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
27 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
28 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
29 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * Copyright (c) 1982, 1986, 1988, 1993
35 * The Regents of the University of California. All rights reserved.
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
40 * 1. Redistributions of source code must retain the above copyright
41 * notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 * notice, this list of conditions and the following disclaimer in the
44 * documentation and/or other materials provided with the distribution.
45 * 3. All advertising materials mentioning features or use of this software
46 * must display the following acknowledgement:
47 * This product includes software developed by the University of
48 * California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 * may be used to endorse or promote products derived from this software
51 * without specific prior written permission.
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
65 * @(#)if_ether.c 8.1 (Berkeley) 6/10/93
66 * $FreeBSD: src/sys/netinet/if_ether.c,v 1.64.2.23 2003/04/11 07:23:15 fjoe Exp $
67 * $DragonFly: src/sys/netinet/if_ether.c,v 1.38 2007/05/24 20:51:22 dillon Exp $
71 * Ethernet address resolution protocol.
73 * add "inuse/lock" bit (or ref. count) along with valid bit
78 #include <sys/param.h>
79 #include <sys/kernel.h>
80 #include <sys/queue.h>
81 #include <sys/sysctl.h>
82 #include <sys/systm.h>
84 #include <sys/malloc.h>
85 #include <sys/socket.h>
86 #include <sys/syslog.h>
89 #include <net/if_dl.h>
90 #include <net/if_types.h>
91 #include <net/route.h>
92 #include <net/netisr.h>
93 #include <net/if_llc.h>
95 #include <netinet/in.h>
96 #include <netinet/in_var.h>
97 #include <netinet/if_ether.h>
99 #include <net/if_arc.h>
100 #include <net/iso88025.h>
102 #include <sys/thread2.h>
103 #include <sys/msgport2.h>
104 #include <net/netmsg2.h>
107 #define SIN(s) ((struct sockaddr_in *)s)
108 #define SDL(s) ((struct sockaddr_dl *)s)
110 SYSCTL_DECL(_net_link_ether);
111 SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, "");
114 static int arpt_prune = (5*60*1); /* walk list every 5 minutes */
115 static int arpt_keep = (20*60); /* once resolved, good for 20 more minutes */
116 static int arpt_down = 20; /* once declared down, don't send for 20 sec */
118 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, prune_intvl, CTLFLAG_RW,
120 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, max_age, CTLFLAG_RW,
122 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, host_down_time, CTLFLAG_RW,
125 #define rt_expire rt_rmx.rmx_expire
128 LIST_ENTRY(llinfo_arp) la_le;
129 struct rtentry *la_rt;
130 struct mbuf *la_hold; /* last packet until resolved/timeout */
131 u_short la_preempt; /* countdown for pre-expiry arps */
132 u_short la_asked; /* #times we QUERIED following expiration */
135 static LIST_HEAD(, llinfo_arp) llinfo_arp_list[MAXCPU];
137 static int arp_maxtries = 5;
138 static int useloopback = 1; /* use loopback interface for local traffic */
139 static int arp_proxyall = 0;
141 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW,
142 &arp_maxtries, 0, "");
143 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW,
144 &useloopback, 0, "");
145 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW,
146 &arp_proxyall, 0, "");
148 static void arp_rtrequest (int, struct rtentry *, struct rt_addrinfo *);
149 static void arprequest (struct ifnet *,
150 struct in_addr *, struct in_addr *, u_char *);
151 static void arpintr(struct netmsg *);
152 static void arptfree (struct llinfo_arp *);
153 static void arptimer (void *);
154 static struct llinfo_arp
155 *arplookup (in_addr_t addr, boolean_t create, boolean_t proxy);
157 static void in_arpinput (struct mbuf *);
160 static struct callout arptimer_ch[MAXCPU];
163 * Timeout routine. Age arp_tab entries periodically.
167 arptimer(void *ignored_arg)
169 struct llinfo_arp *la, *nla;
172 LIST_FOREACH_MUTABLE(la, &llinfo_arp_list[mycpuid], la_le, nla) {
173 if (la->la_rt->rt_expire && la->la_rt->rt_expire <= time_second)
176 callout_reset(&arptimer_ch[mycpuid], arpt_prune * hz, arptimer, NULL);
181 * Parallel to llc_rtrequest.
184 arp_rtrequest(int req, struct rtentry *rt, struct rt_addrinfo *info)
186 struct sockaddr *gate = rt->rt_gateway;
187 struct llinfo_arp *la = rt->rt_llinfo;
189 struct sockaddr_dl null_sdl = { sizeof null_sdl, AF_LINK };
190 static boolean_t arpinit_done[MAXCPU];
192 if (!arpinit_done[mycpuid]) {
193 arpinit_done[mycpuid] = TRUE;
194 callout_init(&arptimer_ch[mycpuid]);
195 callout_reset(&arptimer_ch[mycpuid], hz, arptimer, NULL);
197 if (rt->rt_flags & RTF_GATEWAY)
203 * XXX: If this is a manually added route to interface
204 * such as older version of routed or gated might provide,
205 * restore cloning bit.
207 if (!(rt->rt_flags & RTF_HOST) &&
208 SIN(rt_mask(rt))->sin_addr.s_addr != 0xffffffff)
209 rt->rt_flags |= RTF_CLONING;
210 if (rt->rt_flags & RTF_CLONING) {
212 * Case 1: This route should come from a route to iface.
214 rt_setgate(rt, rt_key(rt),
215 (struct sockaddr *)&null_sdl);
216 gate = rt->rt_gateway;
217 SDL(gate)->sdl_type = rt->rt_ifp->if_type;
218 SDL(gate)->sdl_index = rt->rt_ifp->if_index;
219 rt->rt_expire = time_second;
222 /* Announce a new entry if requested. */
223 if (rt->rt_flags & RTF_ANNOUNCE)
224 arprequest(rt->rt_ifp,
225 &SIN(rt_key(rt))->sin_addr,
226 &SIN(rt_key(rt))->sin_addr,
230 if (gate->sa_family != AF_LINK ||
231 gate->sa_len < sizeof(struct sockaddr_dl)) {
232 log(LOG_DEBUG, "arp_rtrequest: bad gateway value\n");
235 SDL(gate)->sdl_type = rt->rt_ifp->if_type;
236 SDL(gate)->sdl_index = rt->rt_ifp->if_index;
238 break; /* This happens on a route change */
240 * Case 2: This route may come from cloning, or a manual route
241 * add with a LL address.
243 R_Malloc(la, struct llinfo_arp *, sizeof *la);
246 log(LOG_DEBUG, "arp_rtrequest: malloc failed\n");
249 bzero(la, sizeof *la);
251 rt->rt_flags |= RTF_LLINFO;
252 LIST_INSERT_HEAD(&llinfo_arp_list[mycpuid], la, la_le);
256 * This keeps the multicast addresses from showing up
257 * in `arp -a' listings as unresolved. It's not actually
258 * functional. Then the same for broadcast.
260 if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr)) &&
261 rt->rt_ifp->if_type != IFT_ARCNET) {
262 ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt))->sin_addr,
264 SDL(gate)->sdl_alen = 6;
267 if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) {
268 memcpy(LLADDR(SDL(gate)), rt->rt_ifp->if_broadcastaddr,
269 rt->rt_ifp->if_addrlen);
270 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen;
275 if (SIN(rt_key(rt))->sin_addr.s_addr ==
276 (IA_SIN(rt->rt_ifa))->sin_addr.s_addr) {
278 * This test used to be
279 * if (loif.if_flags & IFF_UP)
280 * It allowed local traffic to be forced
281 * through the hardware by configuring the
282 * loopback down. However, it causes problems
283 * during network configuration for boards
284 * that can't receive packets they send. It
285 * is now necessary to clear "useloopback" and
286 * remove the route to force traffic out to
290 bcopy(IF_LLADDR(rt->rt_ifp), LLADDR(SDL(gate)),
291 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen);
300 LIST_REMOVE(la, la_le);
301 rt->rt_llinfo = NULL;
302 rt->rt_flags &= ~RTF_LLINFO;
303 if (la->la_hold != NULL)
304 m_freem(la->la_hold);
310 * Broadcast an ARP request. Caller specifies:
311 * - arp header source ip address
312 * - arp header target ip address
313 * - arp header source ethernet address
316 arprequest(struct ifnet *ifp, struct in_addr *sip, struct in_addr *tip,
320 struct ether_header *eh;
321 struct arc_header *arh;
324 static u_char llcx[] = { 0x82, 0x40, LLC_SNAP_LSAP, LLC_SNAP_LSAP,
325 LLC_UI, 0x00, 0x00, 0x00, 0x08, 0x06 };
328 if ((m = m_gethdr(MB_DONTWAIT, MT_DATA)) == NULL)
330 m->m_pkthdr.rcvif = (struct ifnet *)NULL;
332 switch (ifp->if_type) {
334 ar_hrd = htons(ARPHRD_ARCNET);
336 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
337 m->m_pkthdr.len = m->m_len;
338 MH_ALIGN(m, m->m_len);
340 arh = (struct arc_header *)sa.sa_data;
341 arh->arc_dhost = ifp->if_broadcastaddr[0];
342 arh->arc_type = ARCTYPE_ARP;
344 ah = mtod(m, struct arphdr *);
348 ar_hrd = htons(ARPHRD_IEEE802);
350 m->m_len = (sizeof llcx) +
351 arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
352 m->m_pkthdr.len = m->m_len;
353 MH_ALIGN(m, m->m_len);
355 memcpy(mtod(m, caddr_t), llcx, sizeof llcx);
356 memcpy(sa.sa_data, ifp->if_broadcastaddr, ifp->if_addrlen);
357 memcpy(sa.sa_data + 6, enaddr, 6);
358 sa.sa_data[6] |= TR_RII;
359 sa.sa_data[12] = TR_AC;
360 sa.sa_data[13] = TR_LLC_FRAME;
362 ah = (struct arphdr *)(mtod(m, char *) + sizeof llcx);
367 * This may not be correct for types not explicitly
368 * listed, but this is our best guess
371 ar_hrd = htons(ARPHRD_ETHER);
373 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
374 m->m_pkthdr.len = m->m_len;
375 MH_ALIGN(m, m->m_len);
377 eh = (struct ether_header *)sa.sa_data;
378 /* if_output() will not swap */
379 eh->ether_type = htons(ETHERTYPE_ARP);
380 memcpy(eh->ether_dhost, ifp->if_broadcastaddr, ifp->if_addrlen);
382 ah = mtod(m, struct arphdr *);
387 ah->ar_pro = htons(ETHERTYPE_IP);
388 ah->ar_hln = ifp->if_addrlen; /* hardware address length */
389 ah->ar_pln = sizeof(struct in_addr); /* protocol address length */
390 ah->ar_op = htons(ARPOP_REQUEST);
391 memcpy(ar_sha(ah), enaddr, ah->ar_hln);
392 memset(ar_tha(ah), 0, ah->ar_hln);
393 memcpy(ar_spa(ah), sip, ah->ar_pln);
394 memcpy(ar_tpa(ah), tip, ah->ar_pln);
396 sa.sa_family = AF_UNSPEC;
397 sa.sa_len = sizeof sa;
398 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)NULL);
402 * Resolve an IP address into an ethernet address. If success,
403 * desten is filled in. If there is no entry in arptab,
404 * set one up and broadcast a request for the IP address.
405 * Hold onto this mbuf and resend it once the address
406 * is finally resolved. A return value of 1 indicates
407 * that desten has been filled in and the packet should be sent
408 * normally; a 0 return indicates that the packet has been
409 * taken over here, either now or for later transmission.
416 struct sockaddr *dst,
420 struct llinfo_arp *la = NULL;
421 struct sockaddr_dl *sdl;
423 if (m->m_flags & M_BCAST) { /* broadcast */
424 memcpy(desten, ifp->if_broadcastaddr, ifp->if_addrlen);
427 if (m->m_flags & M_MCAST && ifp->if_type != IFT_ARCNET) {/* multicast */
428 ETHER_MAP_IP_MULTICAST(&SIN(dst)->sin_addr, desten);
432 if (rt_llroute(dst, rt0, &rt) != 0) {
439 la = arplookup(SIN(dst)->sin_addr.s_addr, TRUE, FALSE);
443 if (la == NULL || rt == NULL) {
444 log(LOG_DEBUG, "arpresolve: can't allocate llinfo for %s%s%s\n",
445 inet_ntoa(SIN(dst)->sin_addr), la ? "la" : " ",
450 sdl = SDL(rt->rt_gateway);
452 * Check the address family and length is valid, the address
453 * is resolved; otherwise, try to resolve.
455 if ((rt->rt_expire == 0 || rt->rt_expire > time_second) &&
456 sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0) {
458 * If entry has an expiry time and it is approaching,
459 * see if we need to send an ARP request within this
460 * arpt_down interval.
462 if ((rt->rt_expire != 0) &&
463 (time_second + la->la_preempt > rt->rt_expire)) {
465 &SIN(rt->rt_ifa->ifa_addr)->sin_addr,
471 bcopy(LLADDR(sdl), desten, sdl->sdl_alen);
475 * If ARP is disabled on this interface, stop.
477 * Probably should not allocate empty llinfo struct if we are
478 * not going to be sending out an arp request.
480 if (ifp->if_flags & IFF_NOARP) {
485 * There is an arptab entry, but no ethernet address
486 * response yet. Replace the held mbuf with this
489 if (la->la_hold != NULL)
490 m_freem(la->la_hold);
492 if (rt->rt_expire || ((rt->rt_flags & RTF_STATIC) && !sdl->sdl_alen)) {
493 rt->rt_flags &= ~RTF_REJECT;
494 if (la->la_asked == 0 || rt->rt_expire != time_second) {
495 rt->rt_expire = time_second;
496 if (la->la_asked++ < arp_maxtries) {
498 &SIN(rt->rt_ifa->ifa_addr)->sin_addr,
502 rt->rt_flags |= RTF_REJECT;
503 rt->rt_expire += arpt_down;
505 la->la_preempt = arp_maxtries;
514 * Common length and type checks are done here,
515 * then the protocol-specific routine is called.
518 arpintr(struct netmsg *msg)
520 struct mbuf *m = ((struct netmsg_packet *)msg)->nm_packet;
524 if (m->m_len < sizeof(struct arphdr) &&
525 ((m = m_pullup(m, sizeof(struct arphdr))) == NULL)) {
526 log(LOG_ERR, "arp: runt packet -- m_pullup failed\n");
529 ar = mtod(m, struct arphdr *);
531 ar_hrd = ntohs(ar->ar_hrd);
532 if (ar_hrd != ARPHRD_ETHER &&
533 ar_hrd != ARPHRD_IEEE802 &&
534 ar_hrd != ARPHRD_ARCNET) {
536 "arp: unknown hardware address format (0x%2D)\n",
537 (unsigned char *)&ar->ar_hrd, "");
541 if (m->m_pkthdr.len < arphdr_len(ar) &&
542 (m = m_pullup(m, arphdr_len(ar))) == NULL) {
543 log(LOG_ERR, "arp: runt packet\n");
547 switch (ntohs(ar->ar_pro)) {
558 /* msg was embedded in the mbuf, do not reply! */
563 * ARP for Internet protocols on 10 Mb/s Ethernet.
564 * Algorithm is that given in RFC 826.
565 * In addition, a sanity check is performed on the sender
566 * protocol address, to catch impersonators.
567 * We no longer handle negotiations for use of trailer protocol:
568 * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent
569 * along with IP replies if we wanted trailers sent to us,
570 * and also sent them in response to IP replies.
571 * This allowed either end to announce the desire to receive
573 * We no longer reply to requests for ETHERTYPE_TRAIL protocol either,
574 * but formerly didn't normally send requests.
576 static int log_arp_wrong_iface = 1;
577 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_wrong_iface, CTLFLAG_RW,
578 &log_arp_wrong_iface, 0,
579 "log arp packets arriving on the wrong interface");
582 arp_update_oncpu(struct mbuf *m, in_addr_t saddr, boolean_t create,
585 struct arphdr *ah = mtod(m, struct arphdr *);
586 struct ifnet *ifp = m->m_pkthdr.rcvif;
587 struct llinfo_arp *la;
588 struct sockaddr_dl *sdl;
592 la = arplookup(saddr, create, FALSE);
593 if (la && (rt = la->la_rt) && (sdl = SDL(rt->rt_gateway))) {
594 struct in_addr isaddr = { saddr };
596 /* the following is not an error when doing bridging */
597 if (rt->rt_ifp != ifp) {
598 if (dologging && log_arp_wrong_iface && cpu == 0) {
601 "but got reply from %*D on %s\n",
603 rt->rt_ifp->if_xname,
604 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
610 bcmp(ar_sha(ah), LLADDR(sdl), sdl->sdl_alen)) {
611 if (rt->rt_expire != 0) {
612 if (dologging && cpu == 0) {
614 "arp: %s moved from %*D to %*D on %s\n",
616 ifp->if_addrlen, (u_char *)LLADDR(sdl),
617 ":", ifp->if_addrlen,
618 (u_char *)ar_sha(ah), ":",
622 if (dologging && cpu == 0) {
624 "arp: %*D attempts to modify permanent entry for %s on %s\n",
625 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
626 inet_ntoa(isaddr), ifp->if_xname);
632 * sanity check for the address length.
633 * XXX this does not work for protocols with variable address
636 if (dologging && sdl->sdl_alen && sdl->sdl_alen != ah->ar_hln &&
640 "arp from %*D: new addr len %d, was %d",
641 ifp->if_addrlen, (u_char *) ar_sha(ah), ":",
642 ah->ar_hln, sdl->sdl_alen);
644 if (ifp->if_addrlen != ah->ar_hln) {
645 if (dologging && cpu == 0) {
647 "arp from %*D: addr len: new %d, i/f %d (ignored)",
648 ifp->if_addrlen, (u_char *) ar_sha(ah), ":",
649 ah->ar_hln, ifp->if_addrlen);
653 memcpy(LLADDR(sdl), ar_sha(ah), sdl->sdl_alen = ah->ar_hln);
655 * If we receive an arp from a token-ring station over
656 * a token-ring nic then try to save the source
659 if (ifp->if_type == IFT_ISO88025) {
660 struct iso88025_header *th =
661 (struct iso88025_header *)m->m_pkthdr.header;
662 struct iso88025_sockaddr_dl_data *trld =
666 rif_len = TR_RCF_RIFLEN(th->rcf);
667 if ((th->iso88025_shost[0] & TR_RII) &&
669 trld->trld_rcf = th->rcf;
670 trld->trld_rcf ^= htons(TR_RCF_DIR);
671 memcpy(trld->trld_route, th->rd, rif_len - 2);
672 trld->trld_rcf &= ~htons(TR_RCF_BCST_MASK);
674 * Set up source routing information for
677 m->m_data -= rif_len;
679 m->m_pkthdr.len += rif_len;
681 th->iso88025_shost[0] &= ~TR_RII;
686 m->m_pkthdr.len += 8;
687 th->rcf = trld->trld_rcf;
689 if (rt->rt_expire != 0)
690 rt->rt_expire = time_second + arpt_keep;
691 rt->rt_flags &= ~RTF_REJECT;
693 la->la_preempt = arp_maxtries;
696 * This particular cpu might have been holding an mbuf
697 * pending ARP resolution. If so, transmit the mbuf now.
699 if (la->la_hold != NULL) {
700 m_adj(la->la_hold, sizeof(struct ether_header));
701 lwkt_serialize_enter(ifp->if_serializer);
702 (*ifp->if_output)(ifp, la->la_hold, rt_key(rt), rt);
703 lwkt_serialize_exit(ifp->if_serializer);
711 struct netmsg_arp_update {
712 struct netmsg netmsg;
718 static void arp_update_msghandler(struct netmsg *netmsg);
723 * Called from arpintr() - this routine is run from a single cpu.
726 in_arpinput(struct mbuf *m)
729 struct ifnet *ifp = m->m_pkthdr.rcvif;
730 struct ether_header *eh;
731 struct arc_header *arh;
732 struct iso88025_header *th = (struct iso88025_header *)NULL;
735 struct in_ifaddr *ia;
737 struct in_addr isaddr, itaddr, myaddr;
739 struct netmsg_arp_update msg;
744 req_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
745 if (m->m_len < req_len && (m = m_pullup(m, req_len)) == NULL) {
746 log(LOG_ERR, "in_arp: runt packet -- m_pullup failed\n");
750 ah = mtod(m, struct arphdr *);
751 op = ntohs(ah->ar_op);
752 memcpy(&isaddr, ar_spa(ah), sizeof isaddr);
753 memcpy(&itaddr, ar_tpa(ah), sizeof itaddr);
755 * Check both target and sender IP addresses:
757 * If we receive the packet on the interface owning the address,
758 * then accept the address.
760 * For a bridge, we accept the address if the receive interface and
761 * the interface owning the address are on the same bridge.
762 * (This will change slightly when we have clusters of interfaces).
764 LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash) {
765 /* Skip all ia's which don't match */
766 if (itaddr.s_addr != ia->ia_addr.sin_addr.s_addr)
769 if (ia->ia_ifp == ifp)
772 if (ifp->if_bridge && ia->ia_ifp &&
773 ifp->if_bridge == ia->ia_ifp->if_bridge)
776 LIST_FOREACH(ia, INADDR_HASH(isaddr.s_addr), ia_hash) {
777 /* Skip all ia's which don't match */
778 if (isaddr.s_addr != ia->ia_addr.sin_addr.s_addr)
781 if (ia->ia_ifp == ifp)
784 if (ifp->if_bridge && ia->ia_ifp &&
785 ifp->if_bridge == ia->ia_ifp->if_bridge)
789 * No match, use the first inet address on the receive interface
790 * as a dummy address for the rest of the function.
792 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
793 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET) {
799 * If we got here, we didn't find any suitable interface,
800 * so drop the packet.
806 myaddr = ia->ia_addr.sin_addr;
807 if (!bcmp(ar_sha(ah), IF_LLADDR(ifp), ifp->if_addrlen)) {
808 m_freem(m); /* it's from me, ignore it. */
811 if (!bcmp(ar_sha(ah), ifp->if_broadcastaddr, ifp->if_addrlen)) {
813 "arp: link address is broadcast for IP address %s!\n",
818 if (isaddr.s_addr == myaddr.s_addr && myaddr.s_addr != 0) {
820 "arp: %*D is using my IP address %s!\n",
821 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
827 netmsg_init(&msg.netmsg, &curthread->td_msgport, 0,
828 arp_update_msghandler);
830 msg.saddr = isaddr.s_addr;
831 msg.create = (itaddr.s_addr == myaddr.s_addr);
832 lwkt_domsg(rtable_portfn(0), &msg.netmsg.nm_lmsg, 0);
834 arp_update_oncpu(m, isaddr.s_addr, (itaddr.s_addr == myaddr.s_addr),
837 if (op != ARPOP_REQUEST) {
841 if (itaddr.s_addr == myaddr.s_addr) {
842 /* I am the target */
843 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
844 memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln);
846 struct llinfo_arp *la;
848 la = arplookup(itaddr.s_addr, FALSE, SIN_PROXY);
850 struct sockaddr_in sin;
857 bzero(&sin, sizeof sin);
858 sin.sin_family = AF_INET;
859 sin.sin_len = sizeof sin;
860 sin.sin_addr = itaddr;
862 rt = rtpurelookup((struct sockaddr *)&sin);
869 * Don't send proxies for nodes on the same interface
870 * as this one came out of, or we'll get into a fight
871 * over who claims what Ether address.
873 if (rt->rt_ifp == ifp) {
877 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
878 memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln);
880 kprintf("arp: proxying for %s\n", inet_ntoa(itaddr));
883 struct sockaddr_dl *sdl;
886 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
887 sdl = SDL(rt->rt_gateway);
888 memcpy(ar_sha(ah), LLADDR(sdl), ah->ar_hln);
892 memcpy(ar_tpa(ah), ar_spa(ah), ah->ar_pln);
893 memcpy(ar_spa(ah), &itaddr, ah->ar_pln);
894 ah->ar_op = htons(ARPOP_REPLY);
895 ah->ar_pro = htons(ETHERTYPE_IP); /* let's be sure! */
896 switch (ifp->if_type) {
898 arh = (struct arc_header *)sa.sa_data;
899 arh->arc_dhost = *ar_tha(ah);
900 arh->arc_type = ARCTYPE_ARP;
903 /* Re-arrange the source/dest address */
904 memcpy(th->iso88025_dhost, th->iso88025_shost,
905 sizeof th->iso88025_dhost);
906 memcpy(th->iso88025_shost, IF_LLADDR(ifp),
907 sizeof th->iso88025_shost);
908 /* Set the source routing bit if neccesary */
909 if (th->iso88025_dhost[0] & TR_RII) {
910 th->iso88025_dhost[0] &= ~TR_RII;
911 if (TR_RCF_RIFLEN(th->rcf) > 2)
912 th->iso88025_shost[0] |= TR_RII;
914 /* Copy the addresses, ac and fc into sa_data */
915 memcpy(sa.sa_data, th->iso88025_dhost,
916 (sizeof th->iso88025_dhost) * 2);
917 sa.sa_data[(sizeof th->iso88025_dhost) * 2] = TR_AC;
918 sa.sa_data[(sizeof th->iso88025_dhost) * 2 + 1] = TR_LLC_FRAME;
923 * May not be correct for types not explictly
924 * listed, but it is our best guess.
927 eh = (struct ether_header *)sa.sa_data;
928 memcpy(eh->ether_dhost, ar_tha(ah), sizeof eh->ether_dhost);
929 eh->ether_type = htons(ETHERTYPE_ARP);
932 sa.sa_family = AF_UNSPEC;
933 sa.sa_len = sizeof sa;
934 lwkt_serialize_enter(ifp->if_serializer);
935 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0);
936 lwkt_serialize_exit(ifp->if_serializer);
944 arp_update_msghandler(struct netmsg *netmsg)
946 struct netmsg_arp_update *msg = (struct netmsg_arp_update *)netmsg;
949 arp_update_oncpu(msg->m, msg->saddr, msg->create, FALSE);
951 nextcpu = mycpuid + 1;
952 if (nextcpu < ncpus) {
953 lwkt_forwardmsg(rtable_portfn(nextcpu), &msg->netmsg.nm_lmsg);
955 lwkt_replymsg(&msg->netmsg.nm_lmsg, 0);
964 * Free an arp entry. If the arp entry is actively referenced or represents
965 * a static entry we only clear it back to an unresolved state, otherwise
966 * we destroy the entry entirely.
968 * Note that static entries are created when route add ... -interface is used
969 * to create an interface route to a (direct) destination.
972 arptfree(struct llinfo_arp *la)
974 struct rtentry *rt = la->la_rt;
975 struct sockaddr_dl *sdl;
979 sdl = SDL(rt->rt_gateway);
981 ((rt->rt_refcnt > 0 && sdl->sdl_family == AF_LINK) ||
982 (rt->rt_flags & RTF_STATIC))) {
984 la->la_preempt = la->la_asked = 0;
985 rt->rt_flags &= ~RTF_REJECT;
988 rtrequest(RTM_DELETE, rt_key(rt), NULL, rt_mask(rt), 0, NULL);
992 * Lookup or enter a new address in arptab.
994 static struct llinfo_arp *
995 arplookup(in_addr_t addr, boolean_t create, boolean_t proxy)
998 struct sockaddr_inarp sin = { sizeof sin, AF_INET };
999 const char *why = NULL;
1001 sin.sin_addr.s_addr = addr;
1002 sin.sin_other = proxy ? SIN_PROXY : 0;
1004 rt = rtlookup((struct sockaddr *)&sin);
1006 rt = rtpurelookup((struct sockaddr *)&sin);
1011 if (rt->rt_flags & RTF_GATEWAY)
1012 why = "host is not on local network";
1013 else if (!(rt->rt_flags & RTF_LLINFO))
1014 why = "could not allocate llinfo";
1015 else if (rt->rt_gateway->sa_family != AF_LINK)
1016 why = "gateway route is not ours";
1020 log(LOG_DEBUG, "arplookup %s failed: %s\n",
1021 inet_ntoa(sin.sin_addr), why);
1023 if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_WASCLONED)) {
1024 /* No references to this route. Purge it. */
1025 rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway,
1026 rt_mask(rt), rt->rt_flags, NULL);
1030 return (rt->rt_llinfo);
1034 arp_ifinit(struct ifnet *ifp, struct ifaddr *ifa)
1036 if (IA_SIN(ifa)->sin_addr.s_addr != INADDR_ANY)
1037 arprequest(ifp, &IA_SIN(ifa)->sin_addr, &IA_SIN(ifa)->sin_addr,
1039 ifa->ifa_rtrequest = arp_rtrequest;
1040 ifa->ifa_flags |= RTF_CLONING;
1048 for (cpu = 0; cpu < ncpus2; cpu++)
1049 LIST_INIT(&llinfo_arp_list[cpu]);
1050 netisr_register(NETISR_ARP, cpu0_portfn, arpintr);
1053 SYSINIT(arp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, arp_init, 0);