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 1.58 2008/11/09 10:18:42 sephe Exp $
71 * Ethernet address resolution protocol.
73 * add "inuse/lock" bit (or ref. count) along with valid bit
79 #include <sys/param.h>
80 #include <sys/kernel.h>
81 #include <sys/queue.h>
82 #include <sys/sysctl.h>
83 #include <sys/systm.h>
85 #include <sys/malloc.h>
86 #include <sys/socket.h>
87 #include <sys/syslog.h>
91 #include <net/if_dl.h>
92 #include <net/if_types.h>
93 #include <net/route.h>
94 #include <net/netisr.h>
95 #include <net/if_llc.h>
97 #include <netinet/in.h>
98 #include <netinet/in_var.h>
99 #include <netinet/if_ether.h>
101 #include <sys/thread2.h>
102 #include <sys/msgport2.h>
103 #include <net/netmsg2.h>
106 #include <netinet/ip_carp.h>
109 #define SIN(s) ((struct sockaddr_in *)s)
110 #define SDL(s) ((struct sockaddr_dl *)s)
112 SYSCTL_DECL(_net_link_ether
);
113 SYSCTL_NODE(_net_link_ether
, PF_INET
, inet
, CTLFLAG_RW
, 0, "");
116 static int arpt_prune
= (5*60*1); /* walk list every 5 minutes */
117 static int arpt_keep
= (20*60); /* once resolved, good for 20 more minutes */
118 static int arpt_down
= 20; /* once declared down, don't send for 20 sec */
120 SYSCTL_INT(_net_link_ether_inet
, OID_AUTO
, prune_intvl
, CTLFLAG_RW
,
122 SYSCTL_INT(_net_link_ether_inet
, OID_AUTO
, max_age
, CTLFLAG_RW
,
124 SYSCTL_INT(_net_link_ether_inet
, OID_AUTO
, host_down_time
, CTLFLAG_RW
,
127 #define rt_expire rt_rmx.rmx_expire
130 LIST_ENTRY(llinfo_arp
) la_le
;
131 struct rtentry
*la_rt
;
132 struct mbuf
*la_hold
; /* last packet until resolved/timeout */
133 struct lwkt_port
*la_msgport
; /* last packet's msgport */
134 u_short la_preempt
; /* countdown for pre-expiry arps */
135 u_short la_asked
; /* #times we QUERIED following expiration */
138 static LIST_HEAD(, llinfo_arp
) llinfo_arp_list
[MAXCPU
];
140 static int arp_maxtries
= 5;
141 static int useloopback
= 1; /* use loopback interface for local traffic */
142 static int arp_proxyall
= 0;
144 SYSCTL_INT(_net_link_ether_inet
, OID_AUTO
, maxtries
, CTLFLAG_RW
,
145 &arp_maxtries
, 0, "");
146 SYSCTL_INT(_net_link_ether_inet
, OID_AUTO
, useloopback
, CTLFLAG_RW
,
147 &useloopback
, 0, "");
148 SYSCTL_INT(_net_link_ether_inet
, OID_AUTO
, proxyall
, CTLFLAG_RW
,
149 &arp_proxyall
, 0, "");
151 static int arp_mpsafe
= 0;
152 TUNABLE_INT("net.link.ether.inet.arp_mpsafe", &arp_mpsafe
);
154 static void arp_rtrequest(int, struct rtentry
*, struct rt_addrinfo
*);
155 static void arprequest(struct ifnet
*, struct in_addr
*, struct in_addr
*,
157 static void arprequest_async(struct ifnet
*, struct in_addr
*,
158 struct in_addr
*, const u_char
*);
159 static void arpintr(struct netmsg
*);
160 static void arptfree(struct llinfo_arp
*);
161 static void arptimer(void *);
162 static struct llinfo_arp
*
163 arplookup(in_addr_t
, boolean_t
, boolean_t
);
165 static void in_arpinput(struct mbuf
*);
168 static struct callout arptimer_ch
[MAXCPU
];
171 * Timeout routine. Age arp_tab entries periodically.
175 arptimer(void *ignored_arg
)
177 struct llinfo_arp
*la
, *nla
;
180 LIST_FOREACH_MUTABLE(la
, &llinfo_arp_list
[mycpuid
], la_le
, nla
) {
181 if (la
->la_rt
->rt_expire
&& la
->la_rt
->rt_expire
<= time_second
)
184 callout_reset(&arptimer_ch
[mycpuid
], arpt_prune
* hz
, arptimer
, NULL
);
189 * Parallel to llc_rtrequest.
192 arp_rtrequest(int req
, struct rtentry
*rt
, struct rt_addrinfo
*info
)
194 struct sockaddr
*gate
= rt
->rt_gateway
;
195 struct llinfo_arp
*la
= rt
->rt_llinfo
;
197 struct sockaddr_dl null_sdl
= { sizeof null_sdl
, AF_LINK
};
198 static boolean_t arpinit_done
[MAXCPU
];
200 if (!arpinit_done
[mycpuid
]) {
201 arpinit_done
[mycpuid
] = TRUE
;
202 callout_init(&arptimer_ch
[mycpuid
]);
203 callout_reset(&arptimer_ch
[mycpuid
], hz
, arptimer
, NULL
);
205 if (rt
->rt_flags
& RTF_GATEWAY
)
211 * XXX: If this is a manually added route to interface
212 * such as older version of routed or gated might provide,
213 * restore cloning bit.
215 if (!(rt
->rt_flags
& RTF_HOST
) &&
216 SIN(rt_mask(rt
))->sin_addr
.s_addr
!= 0xffffffff)
217 rt
->rt_flags
|= RTF_CLONING
;
218 if (rt
->rt_flags
& RTF_CLONING
) {
220 * Case 1: This route should come from a route to iface.
222 rt_setgate(rt
, rt_key(rt
),
223 (struct sockaddr
*)&null_sdl
);
224 gate
= rt
->rt_gateway
;
225 SDL(gate
)->sdl_type
= rt
->rt_ifp
->if_type
;
226 SDL(gate
)->sdl_index
= rt
->rt_ifp
->if_index
;
227 rt
->rt_expire
= time_second
;
230 /* Announce a new entry if requested. */
231 if (rt
->rt_flags
& RTF_ANNOUNCE
) {
232 arprequest_async(rt
->rt_ifp
,
233 &SIN(rt_key(rt
))->sin_addr
,
234 &SIN(rt_key(rt
))->sin_addr
,
239 if (gate
->sa_family
!= AF_LINK
||
240 gate
->sa_len
< sizeof(struct sockaddr_dl
)) {
241 log(LOG_DEBUG
, "arp_rtrequest: bad gateway value\n");
244 SDL(gate
)->sdl_type
= rt
->rt_ifp
->if_type
;
245 SDL(gate
)->sdl_index
= rt
->rt_ifp
->if_index
;
247 break; /* This happens on a route change */
249 * Case 2: This route may come from cloning, or a manual route
250 * add with a LL address.
252 R_Malloc(la
, struct llinfo_arp
*, sizeof *la
);
255 log(LOG_DEBUG
, "arp_rtrequest: malloc failed\n");
258 bzero(la
, sizeof *la
);
260 rt
->rt_flags
|= RTF_LLINFO
;
261 LIST_INSERT_HEAD(&llinfo_arp_list
[mycpuid
], la
, la_le
);
265 * This keeps the multicast addresses from showing up
266 * in `arp -a' listings as unresolved. It's not actually
267 * functional. Then the same for broadcast.
269 if (IN_MULTICAST(ntohl(SIN(rt_key(rt
))->sin_addr
.s_addr
))) {
270 ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt
))->sin_addr
,
272 SDL(gate
)->sdl_alen
= 6;
275 if (in_broadcast(SIN(rt_key(rt
))->sin_addr
, rt
->rt_ifp
)) {
276 memcpy(LLADDR(SDL(gate
)), rt
->rt_ifp
->if_broadcastaddr
,
277 rt
->rt_ifp
->if_addrlen
);
278 SDL(gate
)->sdl_alen
= rt
->rt_ifp
->if_addrlen
;
283 if (SIN(rt_key(rt
))->sin_addr
.s_addr
==
284 (IA_SIN(rt
->rt_ifa
))->sin_addr
.s_addr
) {
286 * This test used to be
287 * if (loif.if_flags & IFF_UP)
288 * It allowed local traffic to be forced
289 * through the hardware by configuring the
290 * loopback down. However, it causes problems
291 * during network configuration for boards
292 * that can't receive packets they send. It
293 * is now necessary to clear "useloopback" and
294 * remove the route to force traffic out to
298 bcopy(IF_LLADDR(rt
->rt_ifp
), LLADDR(SDL(gate
)),
299 SDL(gate
)->sdl_alen
= rt
->rt_ifp
->if_addrlen
);
308 LIST_REMOVE(la
, la_le
);
309 rt
->rt_llinfo
= NULL
;
310 rt
->rt_flags
&= ~RTF_LLINFO
;
311 if (la
->la_hold
!= NULL
)
312 m_freem(la
->la_hold
);
319 arpreq_alloc(struct ifnet
*ifp
, struct in_addr
*sip
, struct in_addr
*tip
,
320 const u_char
*enaddr
)
326 if ((m
= m_gethdr(MB_DONTWAIT
, MT_DATA
)) == NULL
)
328 m
->m_pkthdr
.rcvif
= NULL
;
330 switch (ifp
->if_type
) {
333 * This may not be correct for types not explicitly
334 * listed, but this is our best guess
337 ar_hrd
= htons(ARPHRD_ETHER
);
339 m
->m_len
= arphdr_len2(ifp
->if_addrlen
, sizeof(struct in_addr
));
340 m
->m_pkthdr
.len
= m
->m_len
;
341 MH_ALIGN(m
, m
->m_len
);
343 ah
= mtod(m
, struct arphdr
*);
348 ah
->ar_pro
= htons(ETHERTYPE_IP
);
349 ah
->ar_hln
= ifp
->if_addrlen
; /* hardware address length */
350 ah
->ar_pln
= sizeof(struct in_addr
); /* protocol address length */
351 ah
->ar_op
= htons(ARPOP_REQUEST
);
352 memcpy(ar_sha(ah
), enaddr
, ah
->ar_hln
);
353 memset(ar_tha(ah
), 0, ah
->ar_hln
);
354 memcpy(ar_spa(ah
), sip
, ah
->ar_pln
);
355 memcpy(ar_tpa(ah
), tip
, ah
->ar_pln
);
361 arpreq_send(struct ifnet
*ifp
, struct mbuf
*m
)
364 struct ether_header
*eh
;
366 switch (ifp
->if_type
) {
369 * This may not be correct for types not explicitly
370 * listed, but this is our best guess
373 eh
= (struct ether_header
*)sa
.sa_data
;
374 /* if_output() will not swap */
375 eh
->ether_type
= htons(ETHERTYPE_ARP
);
376 memcpy(eh
->ether_dhost
, ifp
->if_broadcastaddr
, ifp
->if_addrlen
);
380 sa
.sa_family
= AF_UNSPEC
;
381 sa
.sa_len
= sizeof(sa
);
382 ifp
->if_output(ifp
, m
, &sa
, NULL
);
386 arpreq_send_handler(struct netmsg
*nmsg
)
388 struct mbuf
*m
= ((struct netmsg_packet
*)nmsg
)->nm_packet
;
389 struct ifnet
*ifp
= nmsg
->nm_lmsg
.u
.ms_resultp
;
392 /* nmsg was embedded in the mbuf, do not reply! */
396 * Broadcast an ARP request. Caller specifies:
397 * - arp header source ip address
398 * - arp header target ip address
399 * - arp header source ethernet address
401 * NOTE: Caller MUST NOT hold ifp's serializer
404 arprequest(struct ifnet
*ifp
, struct in_addr
*sip
, struct in_addr
*tip
,
405 const u_char
*enaddr
)
409 m
= arpreq_alloc(ifp
, sip
, tip
, enaddr
);
416 * Same as arprequest(), except:
417 * - Caller is allowed to hold ifp's serializer
418 * - Network output is done in TDF_NETWORK kernel thread
421 arprequest_async(struct ifnet
*ifp
, struct in_addr
*sip
, struct in_addr
*tip
,
422 const u_char
*enaddr
)
425 struct netmsg_packet
*pmsg
;
427 m
= arpreq_alloc(ifp
, sip
, tip
, enaddr
);
431 pmsg
= &m
->m_hdr
.mh_netmsg
;
432 netmsg_init(&pmsg
->nm_netmsg
, &netisr_apanic_rport
, 0,
433 arpreq_send_handler
);
435 pmsg
->nm_netmsg
.nm_lmsg
.u
.ms_resultp
= ifp
;
437 lwkt_sendmsg(cpu_portfn(mycpuid
), &pmsg
->nm_netmsg
.nm_lmsg
);
441 * Resolve an IP address into an ethernet address. If success,
442 * desten is filled in. If there is no entry in arptab,
443 * set one up and broadcast a request for the IP address.
444 * Hold onto this mbuf and resend it once the address
445 * is finally resolved. A return value of 1 indicates
446 * that desten has been filled in and the packet should be sent
447 * normally; a 0 return indicates that the packet has been
448 * taken over here, either now or for later transmission.
451 arpresolve(struct ifnet
*ifp
, struct rtentry
*rt0
, struct mbuf
*m
,
452 struct sockaddr
*dst
, u_char
*desten
)
455 struct llinfo_arp
*la
= NULL
;
456 struct sockaddr_dl
*sdl
;
458 if (m
->m_flags
& M_BCAST
) { /* broadcast */
459 memcpy(desten
, ifp
->if_broadcastaddr
, ifp
->if_addrlen
);
462 if (m
->m_flags
& M_MCAST
) {/* multicast */
463 ETHER_MAP_IP_MULTICAST(&SIN(dst
)->sin_addr
, desten
);
467 if (rt_llroute(dst
, rt0
, &rt
) != 0) {
474 la
= arplookup(SIN(dst
)->sin_addr
.s_addr
, TRUE
, FALSE
);
478 if (la
== NULL
|| rt
== NULL
) {
479 log(LOG_DEBUG
, "arpresolve: can't allocate llinfo for %s%s%s\n",
480 inet_ntoa(SIN(dst
)->sin_addr
), la
? "la" : " ",
485 sdl
= SDL(rt
->rt_gateway
);
487 * Check the address family and length is valid, the address
488 * is resolved; otherwise, try to resolve.
490 if ((rt
->rt_expire
== 0 || rt
->rt_expire
> time_second
) &&
491 sdl
->sdl_family
== AF_LINK
&& sdl
->sdl_alen
!= 0) {
493 * If entry has an expiry time and it is approaching,
494 * see if we need to send an ARP request within this
495 * arpt_down interval.
497 if ((rt
->rt_expire
!= 0) &&
498 (time_second
+ la
->la_preempt
> rt
->rt_expire
)) {
500 &SIN(rt
->rt_ifa
->ifa_addr
)->sin_addr
,
506 bcopy(LLADDR(sdl
), desten
, sdl
->sdl_alen
);
510 * If ARP is disabled on this interface, stop.
512 * Probably should not allocate empty llinfo struct if we are
513 * not going to be sending out an arp request.
515 if (ifp
->if_flags
& IFF_NOARP
) {
520 * There is an arptab entry, but no ethernet address
521 * response yet. Replace the held mbuf with this
524 if (la
->la_hold
!= NULL
)
525 m_freem(la
->la_hold
);
527 if (curthread
->td_flags
& TDF_NETWORK
)
528 la
->la_msgport
= &curthread
->td_msgport
;
530 la
->la_msgport
= cpu_portfn(mycpuid
);
531 if (rt
->rt_expire
|| ((rt
->rt_flags
& RTF_STATIC
) && !sdl
->sdl_alen
)) {
532 rt
->rt_flags
&= ~RTF_REJECT
;
533 if (la
->la_asked
== 0 || rt
->rt_expire
!= time_second
) {
534 rt
->rt_expire
= time_second
;
535 if (la
->la_asked
++ < arp_maxtries
) {
537 &SIN(rt
->rt_ifa
->ifa_addr
)->sin_addr
,
541 rt
->rt_flags
|= RTF_REJECT
;
542 rt
->rt_expire
+= arpt_down
;
544 la
->la_preempt
= arp_maxtries
;
552 * Common length and type checks are done here,
553 * then the protocol-specific routine is called.
556 arpintr(struct netmsg
*msg
)
558 struct mbuf
*m
= ((struct netmsg_packet
*)msg
)->nm_packet
;
562 if (m
->m_len
< sizeof(struct arphdr
) &&
563 (m
= m_pullup(m
, sizeof(struct arphdr
))) == NULL
) {
564 log(LOG_ERR
, "arp: runt packet -- m_pullup failed\n");
567 ar
= mtod(m
, struct arphdr
*);
569 ar_hrd
= ntohs(ar
->ar_hrd
);
570 if (ar_hrd
!= ARPHRD_ETHER
&& ar_hrd
!= ARPHRD_IEEE802
) {
571 log(LOG_ERR
, "arp: unknown hardware address format (0x%2D)\n",
572 (unsigned char *)&ar
->ar_hrd
, "");
577 if (m
->m_pkthdr
.len
< arphdr_len(ar
)) {
578 if ((m
= m_pullup(m
, arphdr_len(ar
))) == NULL
) {
579 log(LOG_ERR
, "arp: runt packet\n");
582 ar
= mtod(m
, struct arphdr
*);
585 switch (ntohs(ar
->ar_pro
)) {
593 /* msg was embedded in the mbuf, do not reply! */
598 * ARP for Internet protocols on 10 Mb/s Ethernet.
599 * Algorithm is that given in RFC 826.
600 * In addition, a sanity check is performed on the sender
601 * protocol address, to catch impersonators.
602 * We no longer handle negotiations for use of trailer protocol:
603 * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent
604 * along with IP replies if we wanted trailers sent to us,
605 * and also sent them in response to IP replies.
606 * This allowed either end to announce the desire to receive
608 * We no longer reply to requests for ETHERTYPE_TRAIL protocol either,
609 * but formerly didn't normally send requests.
611 static int log_arp_wrong_iface
= 1;
612 SYSCTL_INT(_net_link_ether_inet
, OID_AUTO
, log_arp_wrong_iface
, CTLFLAG_RW
,
613 &log_arp_wrong_iface
, 0,
614 "log arp packets arriving on the wrong interface");
617 arp_hold_output(struct netmsg
*nmsg
)
619 struct mbuf
*m
= ((struct netmsg_packet
*)nmsg
)->nm_packet
;
623 rt
= nmsg
->nm_lmsg
.u
.ms_resultp
;
624 ifp
= m
->m_pkthdr
.rcvif
;
625 m
->m_pkthdr
.rcvif
= NULL
;
627 ifp
->if_output(ifp
, m
, rt_key(rt
), rt
);
629 /* Drop the reference count bumped by the sender */
632 /* nmsg was embedded in the mbuf, do not reply! */
636 arp_update_oncpu(struct mbuf
*m
, in_addr_t saddr
, boolean_t create
,
639 struct arphdr
*ah
= mtod(m
, struct arphdr
*);
640 struct ifnet
*ifp
= m
->m_pkthdr
.rcvif
;
641 struct llinfo_arp
*la
;
642 struct sockaddr_dl
*sdl
;
645 la
= arplookup(saddr
, create
, FALSE
);
646 if (la
&& (rt
= la
->la_rt
) && (sdl
= SDL(rt
->rt_gateway
))) {
647 struct in_addr isaddr
= { saddr
};
649 /* the following is not an error when doing bridging */
650 if (rt
->rt_ifp
!= ifp
) {
651 if (dologging
&& log_arp_wrong_iface
) {
654 "but got reply from %*D on %s\n",
656 rt
->rt_ifp
->if_xname
,
657 ifp
->if_addrlen
, (u_char
*)ar_sha(ah
), ":",
663 bcmp(ar_sha(ah
), LLADDR(sdl
), sdl
->sdl_alen
)) {
664 if (rt
->rt_expire
!= 0) {
667 "arp: %s moved from %*D to %*D on %s\n",
669 ifp
->if_addrlen
, (u_char
*)LLADDR(sdl
),
670 ":", ifp
->if_addrlen
,
671 (u_char
*)ar_sha(ah
), ":",
677 "arp: %*D attempts to modify "
678 "permanent entry for %s on %s\n",
679 ifp
->if_addrlen
, (u_char
*)ar_sha(ah
),
680 ":", inet_ntoa(isaddr
), ifp
->if_xname
);
686 * sanity check for the address length.
687 * XXX this does not work for protocols with variable address
690 if (dologging
&& sdl
->sdl_alen
&& sdl
->sdl_alen
!= ah
->ar_hln
) {
692 "arp from %*D: new addr len %d, was %d",
693 ifp
->if_addrlen
, (u_char
*) ar_sha(ah
), ":",
694 ah
->ar_hln
, sdl
->sdl_alen
);
696 if (ifp
->if_addrlen
!= ah
->ar_hln
) {
699 "arp from %*D: addr len: new %d, i/f %d "
701 ifp
->if_addrlen
, (u_char
*) ar_sha(ah
), ":",
702 ah
->ar_hln
, ifp
->if_addrlen
);
706 memcpy(LLADDR(sdl
), ar_sha(ah
), sdl
->sdl_alen
= ah
->ar_hln
);
707 if (rt
->rt_expire
!= 0)
708 rt
->rt_expire
= time_second
+ arpt_keep
;
709 rt
->rt_flags
&= ~RTF_REJECT
;
711 la
->la_preempt
= arp_maxtries
;
714 * This particular cpu might have been holding an mbuf
715 * pending ARP resolution. If so, transmit the mbuf now.
717 if (la
->la_hold
!= NULL
) {
718 struct mbuf
*m
= la
->la_hold
;
719 struct lwkt_port
*port
= la
->la_msgport
;
720 struct netmsg_packet
*pmsg
;
723 la
->la_msgport
= NULL
;
725 m_adj(m
, sizeof(struct ether_header
));
728 * Make sure that this rtentry will not be freed
729 * before the packet is processed on the target
730 * msgport. The reference count will be dropped
731 * in the handler associated with this packet.
735 pmsg
= &m
->m_hdr
.mh_netmsg
;
736 netmsg_init(&pmsg
->nm_netmsg
, &netisr_apanic_rport
,
737 MSGF_PRIORITY
| MSGF_MPSAFE
,
741 /* Record necessary information */
742 m
->m_pkthdr
.rcvif
= ifp
;
743 pmsg
->nm_netmsg
.nm_lmsg
.u
.ms_resultp
= rt
;
745 lwkt_sendmsg(port
, &pmsg
->nm_netmsg
.nm_lmsg
);
752 struct netmsg_arp_update
{
753 struct netmsg netmsg
;
759 static void arp_update_msghandler(struct netmsg
*);
764 * Called from arpintr() - this routine is run from a single cpu.
767 in_arpinput(struct mbuf
*m
)
770 struct ifnet
*ifp
= m
->m_pkthdr
.rcvif
;
771 struct ether_header
*eh
;
773 struct ifaddr_container
*ifac
;
774 struct in_ifaddr_container
*iac
;
775 struct in_ifaddr
*ia
;
777 struct in_addr isaddr
, itaddr
, myaddr
;
779 struct netmsg_arp_update msg
;
781 u_int8_t
*enaddr
= NULL
;
785 req_len
= arphdr_len2(ifp
->if_addrlen
, sizeof(struct in_addr
));
786 if (m
->m_len
< req_len
&& (m
= m_pullup(m
, req_len
)) == NULL
) {
787 log(LOG_ERR
, "in_arp: runt packet -- m_pullup failed\n");
791 ah
= mtod(m
, struct arphdr
*);
792 op
= ntohs(ah
->ar_op
);
793 memcpy(&isaddr
, ar_spa(ah
), sizeof isaddr
);
794 memcpy(&itaddr
, ar_tpa(ah
), sizeof itaddr
);
796 * Check both target and sender IP addresses:
798 * If we receive the packet on the interface owning the address,
799 * then accept the address.
801 * For a bridge, we accept the address if the receive interface and
802 * the interface owning the address are on the same bridge.
803 * (This will change slightly when we have clusters of interfaces).
805 LIST_FOREACH(iac
, INADDR_HASH(itaddr
.s_addr
), ia_hash
) {
808 /* Skip all ia's which don't match */
809 if (itaddr
.s_addr
!= ia
->ia_addr
.sin_addr
.s_addr
)
812 if (ia
->ia_ifp
== ifp
)
815 if (ifp
->if_bridge
&& ia
->ia_ifp
&&
816 ifp
->if_bridge
== ia
->ia_ifp
->if_bridge
)
821 * If the interface does not match, but the recieving interface
822 * is part of carp, we call carp_iamatch to see if this is a
823 * request for the virtual host ip.
824 * XXX: This is really ugly!
826 if (ifp
->if_carp
!= NULL
) {
828 if (ifp
->if_carp
!= NULL
&&
829 carp_iamatch(ifp
->if_carp
, ia
, &isaddr
, &enaddr
)) {
837 LIST_FOREACH(iac
, INADDR_HASH(isaddr
.s_addr
), ia_hash
) {
840 /* Skip all ia's which don't match */
841 if (isaddr
.s_addr
!= ia
->ia_addr
.sin_addr
.s_addr
)
844 if (ia
->ia_ifp
== ifp
)
847 if (ifp
->if_bridge
&& ia
->ia_ifp
&&
848 ifp
->if_bridge
== ia
->ia_ifp
->if_bridge
)
852 * No match, use the first inet address on the receive interface
853 * as a dummy address for the rest of the function.
855 TAILQ_FOREACH(ifac
, &ifp
->if_addrheads
[mycpuid
], ifa_link
) {
856 struct ifaddr
*ifa
= ifac
->ifa
;
858 if (ifa
->ifa_addr
&& ifa
->ifa_addr
->sa_family
== AF_INET
) {
864 * If we got here, we didn't find any suitable interface,
865 * so drop the packet.
872 enaddr
= (u_int8_t
*)IF_LLADDR(ifp
);
873 myaddr
= ia
->ia_addr
.sin_addr
;
874 if (!bcmp(ar_sha(ah
), enaddr
, ifp
->if_addrlen
)) {
875 m_freem(m
); /* it's from me, ignore it. */
878 if (!bcmp(ar_sha(ah
), ifp
->if_broadcastaddr
, ifp
->if_addrlen
)) {
880 "arp: link address is broadcast for IP address %s!\n",
885 if (isaddr
.s_addr
== myaddr
.s_addr
&& myaddr
.s_addr
!= 0) {
887 "arp: %*D is using my IP address %s!\n",
888 ifp
->if_addrlen
, (u_char
*)ar_sha(ah
), ":",
894 netmsg_init(&msg
.netmsg
, &curthread
->td_msgport
, 0,
895 arp_update_msghandler
);
897 msg
.saddr
= isaddr
.s_addr
;
898 msg
.create
= (itaddr
.s_addr
== myaddr
.s_addr
);
899 lwkt_domsg(rtable_portfn(0), &msg
.netmsg
.nm_lmsg
, 0);
901 arp_update_oncpu(m
, isaddr
.s_addr
, (itaddr
.s_addr
== myaddr
.s_addr
),
905 if (op
!= ARPOP_REQUEST
) {
909 if (itaddr
.s_addr
== myaddr
.s_addr
) {
910 /* I am the target */
911 memcpy(ar_tha(ah
), ar_sha(ah
), ah
->ar_hln
);
912 memcpy(ar_sha(ah
), enaddr
, ah
->ar_hln
);
914 struct llinfo_arp
*la
;
916 la
= arplookup(itaddr
.s_addr
, FALSE
, SIN_PROXY
);
918 struct sockaddr_in sin
;
925 bzero(&sin
, sizeof sin
);
926 sin
.sin_family
= AF_INET
;
927 sin
.sin_len
= sizeof sin
;
928 sin
.sin_addr
= itaddr
;
930 rt
= rtpurelookup((struct sockaddr
*)&sin
);
937 * Don't send proxies for nodes on the same interface
938 * as this one came out of, or we'll get into a fight
939 * over who claims what Ether address.
941 if (rt
->rt_ifp
== ifp
) {
945 memcpy(ar_tha(ah
), ar_sha(ah
), ah
->ar_hln
);
946 memcpy(ar_sha(ah
), enaddr
, ah
->ar_hln
);
948 kprintf("arp: proxying for %s\n", inet_ntoa(itaddr
));
951 struct sockaddr_dl
*sdl
;
954 memcpy(ar_tha(ah
), ar_sha(ah
), ah
->ar_hln
);
955 sdl
= SDL(rt
->rt_gateway
);
956 memcpy(ar_sha(ah
), LLADDR(sdl
), ah
->ar_hln
);
960 memcpy(ar_tpa(ah
), ar_spa(ah
), ah
->ar_pln
);
961 memcpy(ar_spa(ah
), &itaddr
, ah
->ar_pln
);
962 ah
->ar_op
= htons(ARPOP_REPLY
);
963 ah
->ar_pro
= htons(ETHERTYPE_IP
); /* let's be sure! */
964 switch (ifp
->if_type
) {
967 * May not be correct for types not explictly
968 * listed, but it is our best guess.
971 eh
= (struct ether_header
*)sa
.sa_data
;
972 memcpy(eh
->ether_dhost
, ar_tha(ah
), sizeof eh
->ether_dhost
);
973 eh
->ether_type
= htons(ETHERTYPE_ARP
);
976 sa
.sa_family
= AF_UNSPEC
;
977 sa
.sa_len
= sizeof sa
;
978 ifp
->if_output(ifp
, m
, &sa
, NULL
);
984 arp_update_msghandler(struct netmsg
*netmsg
)
986 struct netmsg_arp_update
*msg
= (struct netmsg_arp_update
*)netmsg
;
989 arp_update_oncpu(msg
->m
, msg
->saddr
, msg
->create
, mycpuid
== 0);
991 nextcpu
= mycpuid
+ 1;
993 lwkt_forwardmsg(rtable_portfn(nextcpu
), &msg
->netmsg
.nm_lmsg
);
995 lwkt_replymsg(&msg
->netmsg
.nm_lmsg
, 0);
1003 * Free an arp entry. If the arp entry is actively referenced or represents
1004 * a static entry we only clear it back to an unresolved state, otherwise
1005 * we destroy the entry entirely.
1007 * Note that static entries are created when route add ... -interface is used
1008 * to create an interface route to a (direct) destination.
1011 arptfree(struct llinfo_arp
*la
)
1013 struct rtentry
*rt
= la
->la_rt
;
1014 struct sockaddr_dl
*sdl
;
1018 sdl
= SDL(rt
->rt_gateway
);
1020 ((rt
->rt_refcnt
> 0 && sdl
->sdl_family
== AF_LINK
) ||
1021 (rt
->rt_flags
& RTF_STATIC
))) {
1023 la
->la_preempt
= la
->la_asked
= 0;
1024 rt
->rt_flags
&= ~RTF_REJECT
;
1027 rtrequest(RTM_DELETE
, rt_key(rt
), NULL
, rt_mask(rt
), 0, NULL
);
1031 * Lookup or enter a new address in arptab.
1033 static struct llinfo_arp
*
1034 arplookup(in_addr_t addr
, boolean_t create
, boolean_t proxy
)
1037 struct sockaddr_inarp sin
= { sizeof sin
, AF_INET
};
1038 const char *why
= NULL
;
1040 sin
.sin_addr
.s_addr
= addr
;
1041 sin
.sin_other
= proxy
? SIN_PROXY
: 0;
1043 rt
= rtlookup((struct sockaddr
*)&sin
);
1045 rt
= rtpurelookup((struct sockaddr
*)&sin
);
1050 if (rt
->rt_flags
& RTF_GATEWAY
)
1051 why
= "host is not on local network";
1052 else if (!(rt
->rt_flags
& RTF_LLINFO
))
1053 why
= "could not allocate llinfo";
1054 else if (rt
->rt_gateway
->sa_family
!= AF_LINK
)
1055 why
= "gateway route is not ours";
1059 log(LOG_DEBUG
, "arplookup %s failed: %s\n",
1060 inet_ntoa(sin
.sin_addr
), why
);
1062 if (rt
->rt_refcnt
<= 0 && (rt
->rt_flags
& RTF_WASCLONED
)) {
1063 /* No references to this route. Purge it. */
1064 rtrequest(RTM_DELETE
, rt_key(rt
), rt
->rt_gateway
,
1065 rt_mask(rt
), rt
->rt_flags
, NULL
);
1069 return (rt
->rt_llinfo
);
1073 arp_ifinit(struct ifnet
*ifp
, struct ifaddr
*ifa
)
1075 if (IA_SIN(ifa
)->sin_addr
.s_addr
!= INADDR_ANY
) {
1076 arprequest_async(ifp
, &IA_SIN(ifa
)->sin_addr
,
1077 &IA_SIN(ifa
)->sin_addr
, IF_LLADDR(ifp
));
1079 ifa
->ifa_rtrequest
= arp_rtrequest
;
1080 ifa
->ifa_flags
|= RTF_CLONING
;
1084 arp_ifinit2(struct ifnet
*ifp
, struct ifaddr
*ifa
, u_char
*enaddr
)
1086 if (IA_SIN(ifa
)->sin_addr
.s_addr
!= INADDR_ANY
) {
1087 arprequest_async(ifp
, &IA_SIN(ifa
)->sin_addr
,
1088 &IA_SIN(ifa
)->sin_addr
, enaddr
);
1090 ifa
->ifa_rtrequest
= arp_rtrequest
;
1091 ifa
->ifa_flags
|= RTF_CLONING
;
1100 for (cpu
= 0; cpu
< ncpus2
; cpu
++)
1101 LIST_INIT(&llinfo_arp_list
[cpu
]);
1104 flags
= NETISR_FLAG_MPSAFE
;
1105 kprintf("arp: MPSAFE\n");
1107 flags
= NETISR_FLAG_NOTMPSAFE
;
1109 netisr_register(NETISR_ARP
, cpu0_portfn
, arpintr
, flags
);
1112 SYSINIT(arp
, SI_SUB_PROTO_DOMAIN
, SI_ORDER_ANY
, arp_init
, 0);