Add ifa_listmask field in ifaddr_container; currently it is mainly used
[dragonfly.git] / sys / netinet / if_ether.c
CommitLineData
f3ed2586
JH
1/*
2 * Copyright (c) 2004, 2005 The DragonFly Project. All rights reserved.
3 *
4 * This code is derived from software contributed to The DragonFly Project
5 * by Jeffrey M. Hsu.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
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.
18 *
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
30 * SUCH DAMAGE.
31 */
32
984263bc
MD
33/*
34 * Copyright (c) 1982, 1986, 1988, 1993
35 * The Regents of the University of California. All rights reserved.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
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.
52 *
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
63 * SUCH DAMAGE.
64 *
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 $
57dff79c 67 * $DragonFly: src/sys/netinet/if_ether.c,v 1.47 2008/05/28 12:11:13 sephe Exp $
984263bc
MD
68 */
69
70/*
71 * Ethernet address resolution protocol.
72 * TODO:
73 * add "inuse/lock" bit (or ref. count) along with valid bit
74 */
75
76#include "opt_inet.h"
0d16ba1d 77#include "opt_carp.h"
984263bc
MD
78
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>
84#include <sys/mbuf.h>
85#include <sys/malloc.h>
86#include <sys/socket.h>
87#include <sys/syslog.h>
88
89#include <net/if.h>
90#include <net/if_dl.h>
91#include <net/if_types.h>
92#include <net/route.h>
93#include <net/netisr.h>
94#include <net/if_llc.h>
984263bc
MD
95
96#include <netinet/in.h>
97#include <netinet/in_var.h>
98#include <netinet/if_ether.h>
99
4599cf19
MD
100#include <sys/thread2.h>
101#include <sys/msgport2.h>
102#include <net/netmsg2.h>
103
104
0d16ba1d
MD
105#ifdef CARP
106#include <netinet/ip_carp.h>
107#endif
108
984263bc
MD
109#define SIN(s) ((struct sockaddr_in *)s)
110#define SDL(s) ((struct sockaddr_dl *)s)
111
112SYSCTL_DECL(_net_link_ether);
113SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, "");
114
115/* timer values */
116static int arpt_prune = (5*60*1); /* walk list every 5 minutes */
117static int arpt_keep = (20*60); /* once resolved, good for 20 more minutes */
118static int arpt_down = 20; /* once declared down, don't send for 20 sec */
119
120SYSCTL_INT(_net_link_ether_inet, OID_AUTO, prune_intvl, CTLFLAG_RW,
121 &arpt_prune, 0, "");
f23061d4 122SYSCTL_INT(_net_link_ether_inet, OID_AUTO, max_age, CTLFLAG_RW,
984263bc
MD
123 &arpt_keep, 0, "");
124SYSCTL_INT(_net_link_ether_inet, OID_AUTO, host_down_time, CTLFLAG_RW,
125 &arpt_down, 0, "");
126
f23061d4 127#define rt_expire rt_rmx.rmx_expire
984263bc
MD
128
129struct llinfo_arp {
130 LIST_ENTRY(llinfo_arp) la_le;
131 struct rtentry *la_rt;
132 struct mbuf *la_hold; /* last packet until resolved/timeout */
a916e847 133 u_short la_preempt; /* countdown for pre-expiry arps */
984263bc 134 u_short la_asked; /* #times we QUERIED following expiration */
984263bc
MD
135};
136
ecdefdda 137static LIST_HEAD(, llinfo_arp) llinfo_arp_list[MAXCPU];
984263bc 138
984263bc
MD
139static int arp_maxtries = 5;
140static int useloopback = 1; /* use loopback interface for local traffic */
141static int arp_proxyall = 0;
142
143SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW,
144 &arp_maxtries, 0, "");
145SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW,
146 &useloopback, 0, "");
147SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW,
148 &arp_proxyall, 0, "");
149
0d16ba1d 150void arprequest_acces(struct ifnet *ifp, struct in_addr *sip, struct in_addr *tip, u_char *enaddr);
42a7fc75
RG
151static void arp_rtrequest (int, struct rtentry *, struct rt_addrinfo *);
152static void arprequest (struct ifnet *,
153 struct in_addr *, struct in_addr *, u_char *);
4599cf19 154static void arpintr(struct netmsg *);
42a7fc75
RG
155static void arptfree (struct llinfo_arp *);
156static void arptimer (void *);
984263bc 157static struct llinfo_arp
f23061d4 158 *arplookup (in_addr_t addr, boolean_t create, boolean_t proxy);
984263bc 159#ifdef INET
42a7fc75 160static void in_arpinput (struct mbuf *);
984263bc
MD
161#endif
162
ecdefdda 163static struct callout arptimer_ch[MAXCPU];
ac311540 164
984263bc
MD
165/*
166 * Timeout routine. Age arp_tab entries periodically.
167 */
168/* ARGSUSED */
169static void
f23061d4 170arptimer(void *ignored_arg)
984263bc 171{
2e9572df 172 struct llinfo_arp *la, *nla;
984263bc 173
1cae611f 174 crit_enter();
ecdefdda 175 LIST_FOREACH_MUTABLE(la, &llinfo_arp_list[mycpuid], la_le, nla) {
2e9572df 176 if (la->la_rt->rt_expire && la->la_rt->rt_expire <= time_second)
ecdefdda 177 arptfree(la);
984263bc 178 }
ecdefdda 179 callout_reset(&arptimer_ch[mycpuid], arpt_prune * hz, arptimer, NULL);
1cae611f 180 crit_exit();
984263bc
MD
181}
182
183/*
184 * Parallel to llc_rtrequest.
185 */
186static void
f23061d4 187arp_rtrequest(int req, struct rtentry *rt, struct rt_addrinfo *info)
984263bc 188{
2256ba69 189 struct sockaddr *gate = rt->rt_gateway;
f23061d4
JH
190 struct llinfo_arp *la = rt->rt_llinfo;
191
5fe66e68 192 struct sockaddr_dl null_sdl = { sizeof null_sdl, AF_LINK };
ecdefdda 193 static boolean_t arpinit_done[MAXCPU];
984263bc 194
ecdefdda
MD
195 if (!arpinit_done[mycpuid]) {
196 arpinit_done[mycpuid] = TRUE;
197 callout_init(&arptimer_ch[mycpuid]);
198 callout_reset(&arptimer_ch[mycpuid], hz, arptimer, NULL);
984263bc
MD
199 }
200 if (rt->rt_flags & RTF_GATEWAY)
201 return;
984263bc 202
f23061d4 203 switch (req) {
984263bc
MD
204 case RTM_ADD:
205 /*
206 * XXX: If this is a manually added route to interface
207 * such as older version of routed or gated might provide,
208 * restore cloning bit.
209 */
f23061d4 210 if (!(rt->rt_flags & RTF_HOST) &&
984263bc
MD
211 SIN(rt_mask(rt))->sin_addr.s_addr != 0xffffffff)
212 rt->rt_flags |= RTF_CLONING;
213 if (rt->rt_flags & RTF_CLONING) {
214 /*
215 * Case 1: This route should come from a route to iface.
216 */
217 rt_setgate(rt, rt_key(rt),
0c3c561c 218 (struct sockaddr *)&null_sdl);
984263bc
MD
219 gate = rt->rt_gateway;
220 SDL(gate)->sdl_type = rt->rt_ifp->if_type;
221 SDL(gate)->sdl_index = rt->rt_ifp->if_index;
222 rt->rt_expire = time_second;
223 break;
224 }
225 /* Announce a new entry if requested. */
9c94ab97 226 if (rt->rt_flags & RTF_ANNOUNCE) {
984263bc
MD
227 arprequest(rt->rt_ifp,
228 &SIN(rt_key(rt))->sin_addr,
229 &SIN(rt_key(rt))->sin_addr,
f23061d4 230 LLADDR(SDL(gate)));
9c94ab97 231 }
984263bc
MD
232 /*FALLTHROUGH*/
233 case RTM_RESOLVE:
234 if (gate->sa_family != AF_LINK ||
f23061d4 235 gate->sa_len < sizeof(struct sockaddr_dl)) {
984263bc
MD
236 log(LOG_DEBUG, "arp_rtrequest: bad gateway value\n");
237 break;
238 }
239 SDL(gate)->sdl_type = rt->rt_ifp->if_type;
240 SDL(gate)->sdl_index = rt->rt_ifp->if_index;
f23061d4 241 if (la != NULL)
984263bc
MD
242 break; /* This happens on a route change */
243 /*
244 * Case 2: This route may come from cloning, or a manual route
245 * add with a LL address.
246 */
0c3c561c 247 R_Malloc(la, struct llinfo_arp *, sizeof *la);
f23061d4
JH
248 rt->rt_llinfo = la;
249 if (la == NULL) {
984263bc
MD
250 log(LOG_DEBUG, "arp_rtrequest: malloc failed\n");
251 break;
252 }
f23061d4 253 bzero(la, sizeof *la);
984263bc
MD
254 la->la_rt = rt;
255 rt->rt_flags |= RTF_LLINFO;
ecdefdda 256 LIST_INSERT_HEAD(&llinfo_arp_list[mycpuid], la, la_le);
984263bc
MD
257
258#ifdef INET
259 /*
260 * This keeps the multicast addresses from showing up
261 * in `arp -a' listings as unresolved. It's not actually
262 * functional. Then the same for broadcast.
263 */
ce6da440 264 if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr))) {
984263bc
MD
265 ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt))->sin_addr,
266 LLADDR(SDL(gate)));
267 SDL(gate)->sdl_alen = 6;
268 rt->rt_expire = 0;
269 }
270 if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) {
c401f0fd 271 memcpy(LLADDR(SDL(gate)), rt->rt_ifp->if_broadcastaddr,
984263bc
MD
272 rt->rt_ifp->if_addrlen);
273 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen;
274 rt->rt_expire = 0;
275 }
276#endif
277
278 if (SIN(rt_key(rt))->sin_addr.s_addr ==
279 (IA_SIN(rt->rt_ifa))->sin_addr.s_addr) {
f23061d4
JH
280 /*
281 * This test used to be
282 * if (loif.if_flags & IFF_UP)
283 * It allowed local traffic to be forced
284 * through the hardware by configuring the
285 * loopback down. However, it causes problems
286 * during network configuration for boards
287 * that can't receive packets they send. It
288 * is now necessary to clear "useloopback" and
289 * remove the route to force traffic out to
290 * the hardware.
291 */
984263bc 292 rt->rt_expire = 0;
2e9572df 293 bcopy(IF_LLADDR(rt->rt_ifp), LLADDR(SDL(gate)),
984263bc
MD
294 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen);
295 if (useloopback)
296 rt->rt_ifp = loif;
984263bc
MD
297 }
298 break;
299
300 case RTM_DELETE:
f23061d4 301 if (la == NULL)
984263bc 302 break;
984263bc 303 LIST_REMOVE(la, la_le);
f23061d4 304 rt->rt_llinfo = NULL;
984263bc 305 rt->rt_flags &= ~RTF_LLINFO;
f23061d4 306 if (la->la_hold != NULL)
984263bc 307 m_freem(la->la_hold);
f23061d4 308 Free(la);
984263bc
MD
309 }
310}
311
312/*
313 * Broadcast an ARP request. Caller specifies:
314 * - arp header source ip address
315 * - arp header target ip address
316 * - arp header source ethernet address
317 */
318static void
f23061d4
JH
319arprequest(struct ifnet *ifp, struct in_addr *sip, struct in_addr *tip,
320 u_char *enaddr)
984263bc 321{
2256ba69
RG
322 struct mbuf *m;
323 struct ether_header *eh;
2256ba69 324 struct arphdr *ah;
984263bc 325 struct sockaddr sa;
984263bc
MD
326 u_short ar_hrd;
327
74f1caca 328 if ((m = m_gethdr(MB_DONTWAIT, MT_DATA)) == NULL)
984263bc 329 return;
f23061d4
JH
330 m->m_pkthdr.rcvif = (struct ifnet *)NULL;
331
984263bc 332 switch (ifp->if_type) {
984263bc
MD
333 case IFT_ETHER:
334 /*
335 * This may not be correct for types not explicitly
336 * listed, but this is our best guess
337 */
338 default:
339 ar_hrd = htons(ARPHRD_ETHER);
340
341 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
342 m->m_pkthdr.len = m->m_len;
343 MH_ALIGN(m, m->m_len);
344
345 eh = (struct ether_header *)sa.sa_data;
f23061d4 346 /* if_output() will not swap */
984263bc 347 eh->ether_type = htons(ETHERTYPE_ARP);
c401f0fd 348 memcpy(eh->ether_dhost, ifp->if_broadcastaddr, ifp->if_addrlen);
984263bc
MD
349
350 ah = mtod(m, struct arphdr *);
351 break;
352 }
353
354 ah->ar_hrd = ar_hrd;
355 ah->ar_pro = htons(ETHERTYPE_IP);
356 ah->ar_hln = ifp->if_addrlen; /* hardware address length */
357 ah->ar_pln = sizeof(struct in_addr); /* protocol address length */
358 ah->ar_op = htons(ARPOP_REQUEST);
f23061d4 359 memcpy(ar_sha(ah), enaddr, ah->ar_hln);
984263bc 360 memset(ar_tha(ah), 0, ah->ar_hln);
f23061d4
JH
361 memcpy(ar_spa(ah), sip, ah->ar_pln);
362 memcpy(ar_tpa(ah), tip, ah->ar_pln);
984263bc
MD
363
364 sa.sa_family = AF_UNSPEC;
0c3c561c 365 sa.sa_len = sizeof sa;
9db4b353 366 ifp->if_output(ifp, m, &sa, (struct rtentry *)NULL);
984263bc
MD
367}
368
369/*
370 * Resolve an IP address into an ethernet address. If success,
371 * desten is filled in. If there is no entry in arptab,
372 * set one up and broadcast a request for the IP address.
373 * Hold onto this mbuf and resend it once the address
374 * is finally resolved. A return value of 1 indicates
375 * that desten has been filled in and the packet should be sent
376 * normally; a 0 return indicates that the packet has been
377 * taken over here, either now or for later transmission.
378 */
379int
f23061d4
JH
380arpresolve(
381 struct ifnet *ifp,
382 struct rtentry *rt0,
383 struct mbuf *m,
384 struct sockaddr *dst,
385 u_char *desten)
984263bc 386{
f23061d4
JH
387 struct rtentry *rt;
388 struct llinfo_arp *la = NULL;
984263bc
MD
389 struct sockaddr_dl *sdl;
390
391 if (m->m_flags & M_BCAST) { /* broadcast */
c401f0fd 392 memcpy(desten, ifp->if_broadcastaddr, ifp->if_addrlen);
984263bc
MD
393 return (1);
394 }
ce6da440 395 if (m->m_flags & M_MCAST) {/* multicast */
984263bc 396 ETHER_MAP_IP_MULTICAST(&SIN(dst)->sin_addr, desten);
f23061d4 397 return (1);
984263bc 398 }
f23061d4
JH
399 if (rt0 != NULL) {
400 if (rt_llroute(dst, rt0, &rt) != 0) {
401 m_freem(m);
402 return 0;
403 }
404 la = rt->rt_llinfo;
405 }
406 if (la == NULL) {
407 la = arplookup(SIN(dst)->sin_addr.s_addr, TRUE, FALSE);
408 if (la != NULL)
984263bc
MD
409 rt = la->la_rt;
410 }
f23061d4 411 if (la == NULL || rt == NULL) {
984263bc 412 log(LOG_DEBUG, "arpresolve: can't allocate llinfo for %s%s%s\n",
5fe66e68
JH
413 inet_ntoa(SIN(dst)->sin_addr), la ? "la" : " ",
414 rt ? "rt" : "");
984263bc
MD
415 m_freem(m);
416 return (0);
417 }
418 sdl = SDL(rt->rt_gateway);
419 /*
420 * Check the address family and length is valid, the address
421 * is resolved; otherwise, try to resolve.
422 */
423 if ((rt->rt_expire == 0 || rt->rt_expire > time_second) &&
424 sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0) {
425 /*
426 * If entry has an expiry time and it is approaching,
427 * see if we need to send an ARP request within this
428 * arpt_down interval.
429 */
430 if ((rt->rt_expire != 0) &&
a916e847 431 (time_second + la->la_preempt > rt->rt_expire)) {
984263bc
MD
432 arprequest(ifp,
433 &SIN(rt->rt_ifa->ifa_addr)->sin_addr,
434 &SIN(dst)->sin_addr,
435 IF_LLADDR(ifp));
a916e847 436 la->la_preempt--;
f23061d4 437 }
984263bc
MD
438
439 bcopy(LLADDR(sdl), desten, sdl->sdl_alen);
440 return 1;
441 }
442 /*
443 * If ARP is disabled on this interface, stop.
444 * XXX
445 * Probably should not allocate empty llinfo struct if we are
446 * not going to be sending out an arp request.
447 */
448 if (ifp->if_flags & IFF_NOARP) {
449 m_freem(m);
450 return (0);
451 }
452 /*
453 * There is an arptab entry, but no ethernet address
454 * response yet. Replace the held mbuf with this
455 * latest one.
456 */
f23061d4 457 if (la->la_hold != NULL)
984263bc
MD
458 m_freem(la->la_hold);
459 la->la_hold = m;
03f869ef 460 if (rt->rt_expire || ((rt->rt_flags & RTF_STATIC) && !sdl->sdl_alen)) {
984263bc
MD
461 rt->rt_flags &= ~RTF_REJECT;
462 if (la->la_asked == 0 || rt->rt_expire != time_second) {
463 rt->rt_expire = time_second;
464 if (la->la_asked++ < arp_maxtries) {
465 arprequest(ifp,
466 &SIN(rt->rt_ifa->ifa_addr)->sin_addr,
467 &SIN(dst)->sin_addr,
468 IF_LLADDR(ifp));
469 } else {
470 rt->rt_flags |= RTF_REJECT;
471 rt->rt_expire += arpt_down;
a916e847
JH
472 la->la_asked = 0;
473 la->la_preempt = arp_maxtries;
984263bc
MD
474 }
475
476 }
477 }
478 return (0);
479}
480
481/*
482 * Common length and type checks are done here,
483 * then the protocol-specific routine is called.
484 */
4599cf19 485static void
9eeaa8a9 486arpintr(struct netmsg *msg)
984263bc 487{
9eeaa8a9 488 struct mbuf *m = ((struct netmsg_packet *)msg)->nm_packet;
2256ba69 489 struct arphdr *ar;
8bde602d 490 u_short ar_hrd;
984263bc 491
8bde602d
JH
492 if (m->m_len < sizeof(struct arphdr) &&
493 ((m = m_pullup(m, sizeof(struct arphdr))) == NULL)) {
494 log(LOG_ERR, "arp: runt packet -- m_pullup failed\n");
b76bed62 495 goto out2;
8bde602d
JH
496 }
497 ar = mtod(m, struct arphdr *);
498
499 ar_hrd = ntohs(ar->ar_hrd);
500 if (ar_hrd != ARPHRD_ETHER &&
ce6da440 501 ar_hrd != ARPHRD_IEEE802) {
8bde602d
JH
502 log(LOG_ERR,
503 "arp: unknown hardware address format (0x%2D)\n",
504 (unsigned char *)&ar->ar_hrd, "");
b76bed62 505 goto out1;
8bde602d 506 }
984263bc 507
8bde602d
JH
508 if (m->m_pkthdr.len < arphdr_len(ar) &&
509 (m = m_pullup(m, arphdr_len(ar))) == NULL) {
510 log(LOG_ERR, "arp: runt packet\n");
b76bed62 511 goto out1;
8bde602d
JH
512 }
513
514 switch (ntohs(ar->ar_pro)) {
984263bc 515#ifdef INET
8bde602d
JH
516 case ETHERTYPE_IP:
517 in_arpinput(m);
b76bed62 518 goto out2;
984263bc 519#endif
984263bc 520 }
b76bed62 521out1:
8bde602d 522 m_freem(m);
b76bed62 523out2:
4599cf19 524 ;
6aad077d 525 /* msg was embedded in the mbuf, do not reply! */
984263bc
MD
526}
527
528#ifdef INET
529/*
530 * ARP for Internet protocols on 10 Mb/s Ethernet.
531 * Algorithm is that given in RFC 826.
532 * In addition, a sanity check is performed on the sender
533 * protocol address, to catch impersonators.
534 * We no longer handle negotiations for use of trailer protocol:
535 * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent
536 * along with IP replies if we wanted trailers sent to us,
537 * and also sent them in response to IP replies.
538 * This allowed either end to announce the desire to receive
539 * trailer packets.
540 * We no longer reply to requests for ETHERTYPE_TRAIL protocol either,
541 * but formerly didn't normally send requests.
542 */
543static int log_arp_wrong_iface = 1;
984263bc
MD
544SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_wrong_iface, CTLFLAG_RW,
545 &log_arp_wrong_iface, 0,
546 "log arp packets arriving on the wrong interface");
547
ecdefdda
MD
548static void
549arp_update_oncpu(struct mbuf *m, in_addr_t saddr, boolean_t create,
550 boolean_t dologging)
551{
552 struct arphdr *ah = mtod(m, struct arphdr *);
553 struct ifnet *ifp = m->m_pkthdr.rcvif;
554 struct llinfo_arp *la;
555 struct sockaddr_dl *sdl;
556 struct rtentry *rt;
557 int cpu = mycpuid;
558
559 la = arplookup(saddr, create, FALSE);
560 if (la && (rt = la->la_rt) && (sdl = SDL(rt->rt_gateway))) {
561 struct in_addr isaddr = { saddr };
562
563 /* the following is not an error when doing bridging */
a8d45119 564 if (rt->rt_ifp != ifp) {
ecdefdda
MD
565 if (dologging && log_arp_wrong_iface && cpu == 0) {
566 log(LOG_ERR,
567 "arp: %s is on %s "
568 "but got reply from %*D on %s\n",
569 inet_ntoa(isaddr),
570 rt->rt_ifp->if_xname,
571 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
572 ifp->if_xname);
573 }
574 return;
575 }
576 if (sdl->sdl_alen &&
577 bcmp(ar_sha(ah), LLADDR(sdl), sdl->sdl_alen)) {
578 if (rt->rt_expire != 0) {
579 if (dologging && cpu == 0) {
580 log(LOG_INFO,
581 "arp: %s moved from %*D to %*D on %s\n",
582 inet_ntoa(isaddr),
583 ifp->if_addrlen, (u_char *)LLADDR(sdl),
584 ":", ifp->if_addrlen,
585 (u_char *)ar_sha(ah), ":",
586 ifp->if_xname);
587 }
588 } else {
589 if (dologging && cpu == 0) {
590 log(LOG_ERR,
591 "arp: %*D attempts to modify permanent entry for %s on %s\n",
592 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
593 inet_ntoa(isaddr), ifp->if_xname);
594 }
595 return;
596 }
597 }
598 /*
599 * sanity check for the address length.
600 * XXX this does not work for protocols with variable address
601 * length. -is
602 */
603 if (dologging && sdl->sdl_alen && sdl->sdl_alen != ah->ar_hln &&
604 cpu == 0)
605 {
606 log(LOG_WARNING,
607 "arp from %*D: new addr len %d, was %d",
608 ifp->if_addrlen, (u_char *) ar_sha(ah), ":",
609 ah->ar_hln, sdl->sdl_alen);
610 }
611 if (ifp->if_addrlen != ah->ar_hln) {
612 if (dologging && cpu == 0) {
613 log(LOG_WARNING,
614 "arp from %*D: addr len: new %d, i/f %d (ignored)",
615 ifp->if_addrlen, (u_char *) ar_sha(ah), ":",
616 ah->ar_hln, ifp->if_addrlen);
617 }
618 return;
619 }
620 memcpy(LLADDR(sdl), ar_sha(ah), sdl->sdl_alen = ah->ar_hln);
ecdefdda
MD
621 if (rt->rt_expire != 0)
622 rt->rt_expire = time_second + arpt_keep;
623 rt->rt_flags &= ~RTF_REJECT;
624 la->la_asked = 0;
625 la->la_preempt = arp_maxtries;
626
627 /*
628 * This particular cpu might have been holding an mbuf
629 * pending ARP resolution. If so, transmit the mbuf now.
630 */
631 if (la->la_hold != NULL) {
632 m_adj(la->la_hold, sizeof(struct ether_header));
9db4b353 633 ifp->if_output(ifp, la->la_hold, rt_key(rt), rt);
ecdefdda
MD
634 la->la_hold = NULL;
635 }
636 }
637}
638
639#ifdef SMP
640
641struct netmsg_arp_update {
4599cf19 642 struct netmsg netmsg;
ecdefdda
MD
643 struct mbuf *m;
644 in_addr_t saddr;
645 boolean_t create;
646};
647
4599cf19 648static void arp_update_msghandler(struct netmsg *netmsg);
ecdefdda
MD
649
650#endif
651
652/*
653 * Called from arpintr() - this routine is run from a single cpu.
654 */
984263bc 655static void
f23061d4 656in_arpinput(struct mbuf *m)
984263bc 657{
2256ba69
RG
658 struct arphdr *ah;
659 struct ifnet *ifp = m->m_pkthdr.rcvif;
984263bc 660 struct ether_header *eh;
2256ba69 661 struct rtentry *rt;
b2632176 662 struct ifaddr_container *ifac;
984263bc 663 struct in_ifaddr *ia;
984263bc
MD
664 struct sockaddr sa;
665 struct in_addr isaddr, itaddr, myaddr;
ecdefdda
MD
666#ifdef SMP
667 struct netmsg_arp_update msg;
668#endif
0d16ba1d 669 u_int8_t *enaddr = NULL;
ecdefdda 670 int op;
984263bc
MD
671 int req_len;
672
673 req_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
674 if (m->m_len < req_len && (m = m_pullup(m, req_len)) == NULL) {
675 log(LOG_ERR, "in_arp: runt packet -- m_pullup failed\n");
676 return;
677 }
678
679 ah = mtod(m, struct arphdr *);
680 op = ntohs(ah->ar_op);
f23061d4
JH
681 memcpy(&isaddr, ar_spa(ah), sizeof isaddr);
682 memcpy(&itaddr, ar_tpa(ah), sizeof itaddr);
984263bc 683 /*
ed5bc924
SS
684 * Check both target and sender IP addresses:
685 *
686 * If we receive the packet on the interface owning the address,
687 * then accept the address.
688 *
689 * For a bridge, we accept the address if the receive interface and
690 * the interface owning the address are on the same bridge.
691 * (This will change slightly when we have clusters of interfaces).
984263bc 692 */
ac93838f 693 LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash) {
ed5bc924
SS
694 /* Skip all ia's which don't match */
695 if (itaddr.s_addr != ia->ia_addr.sin_addr.s_addr)
696 continue;
697
698 if (ia->ia_ifp == ifp)
699 goto match;
700
ac93838f 701 if (ifp->if_bridge && ia->ia_ifp &&
ed5bc924 702 ifp->if_bridge == ia->ia_ifp->if_bridge)
ac93838f 703 goto match;
0d16ba1d
MD
704
705#ifdef CARP
706 /*
707 * If the interface does not match, but the recieving interface
708 * is part of carp, we call carp_iamatch to see if this is a
709 * request for the virtual host ip.
710 * XXX: This is really ugly!
711 */
712 if (ifp->if_carp != NULL &&
713 carp_iamatch(ifp->if_carp, ia, &isaddr, &enaddr) &&
714 itaddr.s_addr == ia->ia_addr.sin_addr.s_addr)
715 goto match;
716#endif
ac93838f
SS
717 }
718 LIST_FOREACH(ia, INADDR_HASH(isaddr.s_addr), ia_hash) {
ed5bc924
SS
719 /* Skip all ia's which don't match */
720 if (isaddr.s_addr != ia->ia_addr.sin_addr.s_addr)
721 continue;
722
723 if (ia->ia_ifp == ifp)
724 goto match;
725
ac93838f 726 if (ifp->if_bridge && ia->ia_ifp &&
ed5bc924 727 ifp->if_bridge == ia->ia_ifp->if_bridge)
ac93838f 728 goto match;
ac93838f 729 }
984263bc
MD
730 /*
731 * No match, use the first inet address on the receive interface
732 * as a dummy address for the rest of the function.
733 */
b2632176
SZ
734 TAILQ_FOREACH(ifac, &ifp->if_addrheads[mycpuid], ifa_link) {
735 struct ifaddr *ifa = ifac->ifa;
736
984263bc
MD
737 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET) {
738 ia = ifatoia(ifa);
739 goto match;
740 }
984263bc 741 }
ed5bc924
SS
742 /*
743 * If we got here, we didn't find any suitable interface,
744 * so drop the packet.
745 */
746 m_freem(m);
747 return;
748
984263bc 749match:
0d16ba1d
MD
750 if (!enaddr)
751 enaddr = (u_int8_t *)IF_LLADDR(ifp);
984263bc 752 myaddr = ia->ia_addr.sin_addr;
0d16ba1d 753 if (!bcmp(ar_sha(ah), enaddr, ifp->if_addrlen)) {
984263bc
MD
754 m_freem(m); /* it's from me, ignore it. */
755 return;
756 }
c401f0fd 757 if (!bcmp(ar_sha(ah), ifp->if_broadcastaddr, ifp->if_addrlen)) {
984263bc
MD
758 log(LOG_ERR,
759 "arp: link address is broadcast for IP address %s!\n",
760 inet_ntoa(isaddr));
761 m_freem(m);
762 return;
763 }
c3749f74 764 if (isaddr.s_addr == myaddr.s_addr && myaddr.s_addr != 0) {
984263bc
MD
765 log(LOG_ERR,
766 "arp: %*D is using my IP address %s!\n",
767 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
768 inet_ntoa(isaddr));
769 itaddr = myaddr;
770 goto reply;
771 }
ecdefdda 772#ifdef SMP
4599cf19
MD
773 netmsg_init(&msg.netmsg, &curthread->td_msgport, 0,
774 arp_update_msghandler);
ecdefdda
MD
775 msg.m = m;
776 msg.saddr = isaddr.s_addr;
777 msg.create = (itaddr.s_addr == myaddr.s_addr);
a22c590e 778 lwkt_domsg(rtable_portfn(0), &msg.netmsg.nm_lmsg, 0);
ecdefdda
MD
779#endif
780 arp_update_oncpu(m, isaddr.s_addr, (itaddr.s_addr == myaddr.s_addr),
781 TRUE);
984263bc
MD
782reply:
783 if (op != ARPOP_REQUEST) {
784 m_freem(m);
785 return;
786 }
787 if (itaddr.s_addr == myaddr.s_addr) {
788 /* I am the target */
f23061d4 789 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
0d16ba1d 790 memcpy(ar_sha(ah), enaddr, ah->ar_hln);
984263bc 791 } else {
ecdefdda
MD
792 struct llinfo_arp *la;
793
f23061d4 794 la = arplookup(itaddr.s_addr, FALSE, SIN_PROXY);
984263bc
MD
795 if (la == NULL) {
796 struct sockaddr_in sin;
797
798 if (!arp_proxyall) {
799 m_freem(m);
800 return;
801 }
802
803 bzero(&sin, sizeof sin);
804 sin.sin_family = AF_INET;
805 sin.sin_len = sizeof sin;
806 sin.sin_addr = itaddr;
807
f3ed2586 808 rt = rtpurelookup((struct sockaddr *)&sin);
0c3c561c 809 if (rt == NULL) {
984263bc
MD
810 m_freem(m);
811 return;
812 }
be23faf1 813 --rt->rt_refcnt;
984263bc
MD
814 /*
815 * Don't send proxies for nodes on the same interface
816 * as this one came out of, or we'll get into a fight
817 * over who claims what Ether address.
818 */
819 if (rt->rt_ifp == ifp) {
984263bc
MD
820 m_freem(m);
821 return;
822 }
f23061d4 823 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
0d16ba1d 824 memcpy(ar_sha(ah), enaddr, ah->ar_hln);
984263bc 825#ifdef DEBUG_PROXY
a6ec04bc 826 kprintf("arp: proxying for %s\n", inet_ntoa(itaddr));
984263bc
MD
827#endif
828 } else {
ecdefdda
MD
829 struct sockaddr_dl *sdl;
830
984263bc 831 rt = la->la_rt;
f23061d4 832 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
984263bc 833 sdl = SDL(rt->rt_gateway);
f23061d4 834 memcpy(ar_sha(ah), LLADDR(sdl), ah->ar_hln);
984263bc
MD
835 }
836 }
837
f23061d4
JH
838 memcpy(ar_tpa(ah), ar_spa(ah), ah->ar_pln);
839 memcpy(ar_spa(ah), &itaddr, ah->ar_pln);
984263bc
MD
840 ah->ar_op = htons(ARPOP_REPLY);
841 ah->ar_pro = htons(ETHERTYPE_IP); /* let's be sure! */
842 switch (ifp->if_type) {
984263bc 843 case IFT_ETHER:
984263bc
MD
844 /*
845 * May not be correct for types not explictly
846 * listed, but it is our best guess.
847 */
848 default:
849 eh = (struct ether_header *)sa.sa_data;
0c3c561c 850 memcpy(eh->ether_dhost, ar_tha(ah), sizeof eh->ether_dhost);
984263bc
MD
851 eh->ether_type = htons(ETHERTYPE_ARP);
852 break;
853 }
854 sa.sa_family = AF_UNSPEC;
0c3c561c 855 sa.sa_len = sizeof sa;
9db4b353 856 ifp->if_output(ifp, m, &sa, (struct rtentry *)0);
984263bc
MD
857 return;
858}
ecdefdda
MD
859
860#ifdef SMP
861
862static
4599cf19
MD
863void
864arp_update_msghandler(struct netmsg *netmsg)
ecdefdda 865{
4599cf19 866 struct netmsg_arp_update *msg = (struct netmsg_arp_update *)netmsg;
ecdefdda
MD
867 int nextcpu;
868
869 arp_update_oncpu(msg->m, msg->saddr, msg->create, FALSE);
870
871 nextcpu = mycpuid + 1;
872 if (nextcpu < ncpus) {
4599cf19 873 lwkt_forwardmsg(rtable_portfn(nextcpu), &msg->netmsg.nm_lmsg);
ecdefdda 874 } else {
4599cf19 875 lwkt_replymsg(&msg->netmsg.nm_lmsg, 0);
ecdefdda 876 }
ecdefdda
MD
877}
878
879#endif
880
984263bc
MD
881#endif
882
883/*
03f869ef
MD
884 * Free an arp entry. If the arp entry is actively referenced or represents
885 * a static entry we only clear it back to an unresolved state, otherwise
886 * we destroy the entry entirely.
887 *
888 * Note that static entries are created when route add ... -interface is used
889 * to create an interface route to a (direct) destination.
984263bc
MD
890 */
891static void
5fe66e68 892arptfree(struct llinfo_arp *la)
984263bc 893{
2256ba69
RG
894 struct rtentry *rt = la->la_rt;
895 struct sockaddr_dl *sdl;
f23061d4 896
03f869ef 897 if (rt == NULL)
984263bc 898 panic("arptfree");
03f869ef 899 sdl = SDL(rt->rt_gateway);
5fe66e68
JH
900 if (sdl != NULL &&
901 ((rt->rt_refcnt > 0 && sdl->sdl_family == AF_LINK) ||
902 (rt->rt_flags & RTF_STATIC))) {
984263bc
MD
903 sdl->sdl_alen = 0;
904 la->la_preempt = la->la_asked = 0;
905 rt->rt_flags &= ~RTF_REJECT;
906 return;
907 }
f23061d4 908 rtrequest(RTM_DELETE, rt_key(rt), NULL, rt_mask(rt), 0, NULL);
984263bc 909}
0c3c561c 910
984263bc
MD
911/*
912 * Lookup or enter a new address in arptab.
913 */
914static struct llinfo_arp *
f23061d4 915arplookup(in_addr_t addr, boolean_t create, boolean_t proxy)
984263bc 916{
2256ba69 917 struct rtentry *rt;
f3ed2586 918 struct sockaddr_inarp sin = { sizeof sin, AF_INET };
f23061d4 919 const char *why = NULL;
984263bc
MD
920
921 sin.sin_addr.s_addr = addr;
922 sin.sin_other = proxy ? SIN_PROXY : 0;
f3ed2586
JH
923 if (create)
924 rt = rtlookup((struct sockaddr *)&sin);
925 else
926 rt = rtpurelookup((struct sockaddr *)&sin);
f23061d4
JH
927 if (rt == NULL)
928 return (NULL);
984263bc
MD
929 rt->rt_refcnt--;
930
931 if (rt->rt_flags & RTF_GATEWAY)
932 why = "host is not on local network";
f23061d4 933 else if (!(rt->rt_flags & RTF_LLINFO))
984263bc
MD
934 why = "could not allocate llinfo";
935 else if (rt->rt_gateway->sa_family != AF_LINK)
936 why = "gateway route is not ours";
937
8fd3e627
MD
938 if (why) {
939 if (create) {
940 log(LOG_DEBUG, "arplookup %s failed: %s\n",
941 inet_ntoa(sin.sin_addr), why);
942 }
f23061d4
JH
943 if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_WASCLONED)) {
944 /* No references to this route. Purge it. */
945 rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway,
be23faf1 946 rt_mask(rt), rt->rt_flags, NULL);
8fd3e627 947 }
f23061d4 948 return (NULL);
984263bc 949 }
f23061d4 950 return (rt->rt_llinfo);
984263bc
MD
951}
952
953void
f23061d4 954arp_ifinit(struct ifnet *ifp, struct ifaddr *ifa)
984263bc 955{
57dff79c
SZ
956 ASSERT_SERIALIZED(ifp->if_serializer);
957
958 if (IA_SIN(ifa)->sin_addr.s_addr != INADDR_ANY) {
959 lwkt_serialize_exit(ifp->if_serializer);
f23061d4
JH
960 arprequest(ifp, &IA_SIN(ifa)->sin_addr, &IA_SIN(ifa)->sin_addr,
961 IF_LLADDR(ifp));
57dff79c
SZ
962 lwkt_serialize_enter(ifp->if_serializer);
963 }
984263bc
MD
964 ifa->ifa_rtrequest = arp_rtrequest;
965 ifa->ifa_flags |= RTF_CLONING;
966}
967
0d16ba1d
MD
968void
969arp_ifinit2(struct ifnet *ifp, struct ifaddr *ifa, u_char *enaddr)
970{
57dff79c
SZ
971 ASSERT_NOT_SERIALIZED(ifp->if_serializer);
972
0d16ba1d
MD
973 if (IA_SIN(ifa)->sin_addr.s_addr != INADDR_ANY)
974 arprequest(ifp, &IA_SIN(ifa)->sin_addr, &IA_SIN(ifa)->sin_addr,
975 enaddr);
976 ifa->ifa_rtrequest = arp_rtrequest;
977 ifa->ifa_flags |= RTF_CLONING;
978}
979
984263bc
MD
980static void
981arp_init(void)
982{
ecdefdda
MD
983 int cpu;
984
985 for (cpu = 0; cpu < ncpus2; cpu++)
986 LIST_INIT(&llinfo_arp_list[cpu]);
bf82f9b7 987 netisr_register(NETISR_ARP, cpu0_portfn, arpintr);
984263bc
MD
988}
989
990SYSINIT(arp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, arp_init, 0);