timeout/untimeout ==> callout_*
[dragonfly.git] / sys / netinet / if_ether.c
CommitLineData
984263bc
MD
1/*
2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * @(#)if_ether.c 8.1 (Berkeley) 6/10/93
34 * $FreeBSD: src/sys/netinet/if_ether.c,v 1.64.2.23 2003/04/11 07:23:15 fjoe Exp $
c401f0fd 35 * $DragonFly: src/sys/netinet/if_ether.c,v 1.14 2004/07/17 09:43:05 joerg Exp $
984263bc
MD
36 */
37
38/*
39 * Ethernet address resolution protocol.
40 * TODO:
41 * add "inuse/lock" bit (or ref. count) along with valid bit
42 */
43
44#include "opt_inet.h"
45#include "opt_bdg.h"
46
47#include <sys/param.h>
48#include <sys/kernel.h>
49#include <sys/queue.h>
50#include <sys/sysctl.h>
51#include <sys/systm.h>
52#include <sys/mbuf.h>
53#include <sys/malloc.h>
54#include <sys/socket.h>
55#include <sys/syslog.h>
56
a80cf23b
JH
57#include <sys/thread2.h>
58#include <sys/msgport2.h>
59
984263bc
MD
60#include <net/if.h>
61#include <net/if_dl.h>
62#include <net/if_types.h>
63#include <net/route.h>
64#include <net/netisr.h>
65#include <net/if_llc.h>
66#ifdef BRIDGE
67#include <net/ethernet.h>
d2438d69 68#include <net/bridge/bridge.h>
984263bc
MD
69#endif
70
71#include <netinet/in.h>
72#include <netinet/in_var.h>
73#include <netinet/if_ether.h>
74
75#include <net/if_arc.h>
76#include <net/iso88025.h>
77
78#define SIN(s) ((struct sockaddr_in *)s)
79#define SDL(s) ((struct sockaddr_dl *)s)
80
81SYSCTL_DECL(_net_link_ether);
82SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, "");
83
84/* timer values */
85static int arpt_prune = (5*60*1); /* walk list every 5 minutes */
86static int arpt_keep = (20*60); /* once resolved, good for 20 more minutes */
87static int arpt_down = 20; /* once declared down, don't send for 20 sec */
88
89SYSCTL_INT(_net_link_ether_inet, OID_AUTO, prune_intvl, CTLFLAG_RW,
90 &arpt_prune, 0, "");
91SYSCTL_INT(_net_link_ether_inet, OID_AUTO, max_age, CTLFLAG_RW,
92 &arpt_keep, 0, "");
93SYSCTL_INT(_net_link_ether_inet, OID_AUTO, host_down_time, CTLFLAG_RW,
94 &arpt_down, 0, "");
95
96#define rt_expire rt_rmx.rmx_expire
97
98struct llinfo_arp {
99 LIST_ENTRY(llinfo_arp) la_le;
100 struct rtentry *la_rt;
101 struct mbuf *la_hold; /* last packet until resolved/timeout */
102 u_short la_preempt; /* #times we QUERIED before entry expiration */
103 u_short la_asked; /* #times we QUERIED following expiration */
104#define la_timer la_rt->rt_rmx.rmx_expire /* deletion time in seconds */
105};
106
107static LIST_HEAD(, llinfo_arp) llinfo_arp;
108
984263bc
MD
109static int arp_inuse, arp_allocated, arpinit_done;
110
111static int arp_maxtries = 5;
112static int useloopback = 1; /* use loopback interface for local traffic */
113static int arp_proxyall = 0;
114
115SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW,
116 &arp_maxtries, 0, "");
117SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW,
118 &useloopback, 0, "");
119SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW,
120 &arp_proxyall, 0, "");
121
42a7fc75
RG
122static void arp_rtrequest (int, struct rtentry *, struct rt_addrinfo *);
123static void arprequest (struct ifnet *,
124 struct in_addr *, struct in_addr *, u_char *);
b76bed62 125static int arpintr(struct netmsg *);
42a7fc75
RG
126static void arptfree (struct llinfo_arp *);
127static void arptimer (void *);
984263bc 128static struct llinfo_arp
42a7fc75 129 *arplookup (u_long, int, int);
984263bc 130#ifdef INET
42a7fc75 131static void in_arpinput (struct mbuf *);
984263bc
MD
132#endif
133
984263bc
MD
134/*
135 * Timeout routine. Age arp_tab entries periodically.
136 */
137/* ARGSUSED */
138static void
139arptimer(ignored_arg)
140 void *ignored_arg;
141{
142 int s = splnet();
2256ba69 143 struct llinfo_arp *la = LIST_FIRST(&llinfo_arp);
984263bc
MD
144 struct llinfo_arp *ola;
145
146 timeout(arptimer, (caddr_t)0, arpt_prune * hz);
147 while ((ola = la) != 0) {
2256ba69 148 struct rtentry *rt = la->la_rt;
984263bc
MD
149 la = LIST_NEXT(la, la_le);
150 if (rt->rt_expire && rt->rt_expire <= time_second)
151 arptfree(ola); /* timer has expired, clear */
152 }
153 splx(s);
154}
155
156/*
157 * Parallel to llc_rtrequest.
158 */
159static void
160arp_rtrequest(req, rt, info)
161 int req;
2256ba69 162 struct rtentry *rt;
984263bc
MD
163 struct rt_addrinfo *info;
164{
2256ba69
RG
165 struct sockaddr *gate = rt->rt_gateway;
166 struct llinfo_arp *la = (struct llinfo_arp *)rt->rt_llinfo;
984263bc
MD
167 static struct sockaddr_dl null_sdl = {sizeof(null_sdl), AF_LINK};
168
169 if (!arpinit_done) {
170 arpinit_done = 1;
171 timeout(arptimer, (caddr_t)0, hz);
172 }
173 if (rt->rt_flags & RTF_GATEWAY)
174 return;
175 switch (req) {
176
177 case RTM_ADD:
178 /*
179 * XXX: If this is a manually added route to interface
180 * such as older version of routed or gated might provide,
181 * restore cloning bit.
182 */
183 if ((rt->rt_flags & RTF_HOST) == 0 &&
184 SIN(rt_mask(rt))->sin_addr.s_addr != 0xffffffff)
185 rt->rt_flags |= RTF_CLONING;
186 if (rt->rt_flags & RTF_CLONING) {
187 /*
188 * Case 1: This route should come from a route to iface.
189 */
190 rt_setgate(rt, rt_key(rt),
191 (struct sockaddr *)&null_sdl);
192 gate = rt->rt_gateway;
193 SDL(gate)->sdl_type = rt->rt_ifp->if_type;
194 SDL(gate)->sdl_index = rt->rt_ifp->if_index;
195 rt->rt_expire = time_second;
196 break;
197 }
198 /* Announce a new entry if requested. */
199 if (rt->rt_flags & RTF_ANNOUNCE)
200 arprequest(rt->rt_ifp,
201 &SIN(rt_key(rt))->sin_addr,
202 &SIN(rt_key(rt))->sin_addr,
203 (u_char *)LLADDR(SDL(gate)));
204 /*FALLTHROUGH*/
205 case RTM_RESOLVE:
206 if (gate->sa_family != AF_LINK ||
207 gate->sa_len < sizeof(null_sdl)) {
208 log(LOG_DEBUG, "arp_rtrequest: bad gateway value\n");
209 break;
210 }
211 SDL(gate)->sdl_type = rt->rt_ifp->if_type;
212 SDL(gate)->sdl_index = rt->rt_ifp->if_index;
213 if (la != 0)
214 break; /* This happens on a route change */
215 /*
216 * Case 2: This route may come from cloning, or a manual route
217 * add with a LL address.
218 */
219 R_Malloc(la, struct llinfo_arp *, sizeof(*la));
220 rt->rt_llinfo = (caddr_t)la;
221 if (la == 0) {
222 log(LOG_DEBUG, "arp_rtrequest: malloc failed\n");
223 break;
224 }
225 arp_inuse++, arp_allocated++;
226 Bzero(la, sizeof(*la));
227 la->la_rt = rt;
228 rt->rt_flags |= RTF_LLINFO;
229 LIST_INSERT_HEAD(&llinfo_arp, la, la_le);
230
231#ifdef INET
232 /*
233 * This keeps the multicast addresses from showing up
234 * in `arp -a' listings as unresolved. It's not actually
235 * functional. Then the same for broadcast.
236 */
237 if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr))
238 && rt->rt_ifp->if_type != IFT_ARCNET) {
239 ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt))->sin_addr,
240 LLADDR(SDL(gate)));
241 SDL(gate)->sdl_alen = 6;
242 rt->rt_expire = 0;
243 }
244 if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) {
c401f0fd 245 memcpy(LLADDR(SDL(gate)), rt->rt_ifp->if_broadcastaddr,
984263bc
MD
246 rt->rt_ifp->if_addrlen);
247 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen;
248 rt->rt_expire = 0;
249 }
250#endif
251
252 if (SIN(rt_key(rt))->sin_addr.s_addr ==
253 (IA_SIN(rt->rt_ifa))->sin_addr.s_addr) {
254 /*
255 * This test used to be
256 * if (loif.if_flags & IFF_UP)
257 * It allowed local traffic to be forced
258 * through the hardware by configuring the loopback down.
259 * However, it causes problems during network configuration
260 * for boards that can't receive packets they send.
261 * It is now necessary to clear "useloopback" and remove
262 * the route to force traffic out to the hardware.
263 */
264 rt->rt_expire = 0;
265 Bcopy(IF_LLADDR(rt->rt_ifp), LLADDR(SDL(gate)),
266 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen);
267 if (useloopback)
268 rt->rt_ifp = loif;
269
270 }
271 break;
272
273 case RTM_DELETE:
274 if (la == 0)
275 break;
276 arp_inuse--;
277 LIST_REMOVE(la, la_le);
278 rt->rt_llinfo = 0;
279 rt->rt_flags &= ~RTF_LLINFO;
280 if (la->la_hold)
281 m_freem(la->la_hold);
282 Free((caddr_t)la);
283 }
284}
285
286/*
287 * Broadcast an ARP request. Caller specifies:
288 * - arp header source ip address
289 * - arp header target ip address
290 * - arp header source ethernet address
291 */
292static void
293arprequest(ifp, sip, tip, enaddr)
2256ba69
RG
294 struct ifnet *ifp;
295 struct in_addr *sip, *tip;
296 u_char *enaddr;
984263bc 297{
2256ba69
RG
298 struct mbuf *m;
299 struct ether_header *eh;
300 struct arc_header *arh;
301 struct arphdr *ah;
984263bc
MD
302 struct sockaddr sa;
303 static u_char llcx[] = { 0x82, 0x40, LLC_SNAP_LSAP, LLC_SNAP_LSAP,
304 LLC_UI, 0x00, 0x00, 0x00, 0x08, 0x06 };
305 u_short ar_hrd;
306
74f1caca 307 if ((m = m_gethdr(MB_DONTWAIT, MT_DATA)) == NULL)
984263bc
MD
308 return;
309 m->m_pkthdr.rcvif = (struct ifnet *)0;
310 switch (ifp->if_type) {
311 case IFT_ARCNET:
312 ar_hrd = htons(ARPHRD_ARCNET);
313
314 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
315 m->m_pkthdr.len = m->m_len;
316 MH_ALIGN(m, m->m_len);
317
318 arh = (struct arc_header *)sa.sa_data;
c401f0fd 319 arh->arc_dhost = ifp->if_broadcastaddr[0];
984263bc
MD
320 arh->arc_type = ARCTYPE_ARP;
321
322 ah = mtod(m, struct arphdr *);
323 break;
324
325 case IFT_ISO88025:
326 ar_hrd = htons(ARPHRD_IEEE802);
327
328 m->m_len = sizeof(llcx) +
329 arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
330 m->m_pkthdr.len = m->m_len;
331 MH_ALIGN(m, m->m_len);
332
333 (void)memcpy(mtod(m, caddr_t), llcx, sizeof(llcx));
c401f0fd 334 memcpy(sa.sa_data, ifp->if_broadcastaddr, ifp->if_addrlen);
984263bc
MD
335 (void)memcpy(sa.sa_data + 6, enaddr, 6);
336 sa.sa_data[6] |= TR_RII;
337 sa.sa_data[12] = TR_AC;
338 sa.sa_data[13] = TR_LLC_FRAME;
339
340 ah = (struct arphdr *)(mtod(m, char *) + sizeof(llcx));
341 break;
342 case IFT_FDDI:
343 case IFT_ETHER:
344 /*
345 * This may not be correct for types not explicitly
346 * listed, but this is our best guess
347 */
348 default:
349 ar_hrd = htons(ARPHRD_ETHER);
350
351 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
352 m->m_pkthdr.len = m->m_len;
353 MH_ALIGN(m, m->m_len);
354
355 eh = (struct ether_header *)sa.sa_data;
356 /* if_output will not swap */
357 eh->ether_type = htons(ETHERTYPE_ARP);
c401f0fd 358 memcpy(eh->ether_dhost, ifp->if_broadcastaddr, ifp->if_addrlen);
984263bc
MD
359
360 ah = mtod(m, struct arphdr *);
361 break;
362 }
363
364 ah->ar_hrd = ar_hrd;
365 ah->ar_pro = htons(ETHERTYPE_IP);
366 ah->ar_hln = ifp->if_addrlen; /* hardware address length */
367 ah->ar_pln = sizeof(struct in_addr); /* protocol address length */
368 ah->ar_op = htons(ARPOP_REQUEST);
369 (void)memcpy(ar_sha(ah), enaddr, ah->ar_hln);
370 memset(ar_tha(ah), 0, ah->ar_hln);
371 (void)memcpy(ar_spa(ah), sip, ah->ar_pln);
372 (void)memcpy(ar_tpa(ah), tip, ah->ar_pln);
373
374 sa.sa_family = AF_UNSPEC;
375 sa.sa_len = sizeof(sa);
376 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0);
377}
378
379/*
380 * Resolve an IP address into an ethernet address. If success,
381 * desten is filled in. If there is no entry in arptab,
382 * set one up and broadcast a request for the IP address.
383 * Hold onto this mbuf and resend it once the address
384 * is finally resolved. A return value of 1 indicates
385 * that desten has been filled in and the packet should be sent
386 * normally; a 0 return indicates that the packet has been
387 * taken over here, either now or for later transmission.
388 */
389int
390arpresolve(ifp, rt, m, dst, desten, rt0)
2256ba69
RG
391 struct ifnet *ifp;
392 struct rtentry *rt;
984263bc 393 struct mbuf *m;
2256ba69
RG
394 struct sockaddr *dst;
395 u_char *desten;
984263bc
MD
396 struct rtentry *rt0;
397{
398 struct llinfo_arp *la = 0;
399 struct sockaddr_dl *sdl;
400
401 if (m->m_flags & M_BCAST) { /* broadcast */
c401f0fd 402 memcpy(desten, ifp->if_broadcastaddr, ifp->if_addrlen);
984263bc
MD
403 return (1);
404 }
405 if (m->m_flags & M_MCAST && ifp->if_type != IFT_ARCNET) {/* multicast */
406 ETHER_MAP_IP_MULTICAST(&SIN(dst)->sin_addr, desten);
407 return(1);
408 }
409 if (rt)
410 la = (struct llinfo_arp *)rt->rt_llinfo;
411 if (la == 0) {
412 la = arplookup(SIN(dst)->sin_addr.s_addr, 1, 0);
413 if (la)
414 rt = la->la_rt;
415 }
416 if (la == 0 || rt == 0) {
417 log(LOG_DEBUG, "arpresolve: can't allocate llinfo for %s%s%s\n",
418 inet_ntoa(SIN(dst)->sin_addr), la ? "la" : "",
419 rt ? "rt" : "");
420 m_freem(m);
421 return (0);
422 }
423 sdl = SDL(rt->rt_gateway);
424 /*
425 * Check the address family and length is valid, the address
426 * is resolved; otherwise, try to resolve.
427 */
428 if ((rt->rt_expire == 0 || rt->rt_expire > time_second) &&
429 sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0) {
430 /*
431 * If entry has an expiry time and it is approaching,
432 * see if we need to send an ARP request within this
433 * arpt_down interval.
434 */
435 if ((rt->rt_expire != 0) &&
436 (time_second + (arp_maxtries - la->la_preempt) * arpt_down
437 > rt->rt_expire)) {
438 arprequest(ifp,
439 &SIN(rt->rt_ifa->ifa_addr)->sin_addr,
440 &SIN(dst)->sin_addr,
441 IF_LLADDR(ifp));
442 la->la_preempt++;
443 }
444
445 bcopy(LLADDR(sdl), desten, sdl->sdl_alen);
446 return 1;
447 }
448 /*
449 * If ARP is disabled on this interface, stop.
450 * XXX
451 * Probably should not allocate empty llinfo struct if we are
452 * not going to be sending out an arp request.
453 */
454 if (ifp->if_flags & IFF_NOARP) {
455 m_freem(m);
456 return (0);
457 }
458 /*
459 * There is an arptab entry, but no ethernet address
460 * response yet. Replace the held mbuf with this
461 * latest one.
462 */
463 if (la->la_hold)
464 m_freem(la->la_hold);
465 la->la_hold = m;
466 if (rt->rt_expire) {
467 rt->rt_flags &= ~RTF_REJECT;
468 if (la->la_asked == 0 || rt->rt_expire != time_second) {
469 rt->rt_expire = time_second;
470 if (la->la_asked++ < arp_maxtries) {
471 arprequest(ifp,
472 &SIN(rt->rt_ifa->ifa_addr)->sin_addr,
473 &SIN(dst)->sin_addr,
474 IF_LLADDR(ifp));
475 } else {
476 rt->rt_flags |= RTF_REJECT;
477 rt->rt_expire += arpt_down;
478 la->la_preempt = la->la_asked = 0;
479 }
480
481 }
482 }
483 return (0);
484}
485
486/*
487 * Common length and type checks are done here,
488 * then the protocol-specific routine is called.
489 */
b76bed62 490static int
9eeaa8a9 491arpintr(struct netmsg *msg)
984263bc 492{
9eeaa8a9 493 struct mbuf *m = ((struct netmsg_packet *)msg)->nm_packet;
2256ba69 494 struct arphdr *ar;
8bde602d 495 u_short ar_hrd;
984263bc 496
8bde602d
JH
497 if (m->m_len < sizeof(struct arphdr) &&
498 ((m = m_pullup(m, sizeof(struct arphdr))) == NULL)) {
499 log(LOG_ERR, "arp: runt packet -- m_pullup failed\n");
b76bed62 500 goto out2;
8bde602d
JH
501 }
502 ar = mtod(m, struct arphdr *);
503
504 ar_hrd = ntohs(ar->ar_hrd);
505 if (ar_hrd != ARPHRD_ETHER &&
506 ar_hrd != ARPHRD_IEEE802 &&
507 ar_hrd != ARPHRD_ARCNET) {
508 log(LOG_ERR,
509 "arp: unknown hardware address format (0x%2D)\n",
510 (unsigned char *)&ar->ar_hrd, "");
b76bed62 511 goto out1;
8bde602d 512 }
984263bc 513
8bde602d
JH
514 if (m->m_pkthdr.len < arphdr_len(ar) &&
515 (m = m_pullup(m, arphdr_len(ar))) == NULL) {
516 log(LOG_ERR, "arp: runt packet\n");
b76bed62 517 goto out1;
8bde602d
JH
518 }
519
520 switch (ntohs(ar->ar_pro)) {
984263bc 521#ifdef INET
8bde602d
JH
522 case ETHERTYPE_IP:
523 in_arpinput(m);
b76bed62 524 goto out2;
984263bc 525#endif
984263bc 526 }
b76bed62 527out1:
8bde602d 528 m_freem(m);
b76bed62 529out2:
a80cf23b 530 lwkt_replymsg(&msg->nm_lmsg, 0);
b76bed62 531 return(EASYNC);
984263bc
MD
532}
533
534#ifdef INET
535/*
536 * ARP for Internet protocols on 10 Mb/s Ethernet.
537 * Algorithm is that given in RFC 826.
538 * In addition, a sanity check is performed on the sender
539 * protocol address, to catch impersonators.
540 * We no longer handle negotiations for use of trailer protocol:
541 * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent
542 * along with IP replies if we wanted trailers sent to us,
543 * and also sent them in response to IP replies.
544 * This allowed either end to announce the desire to receive
545 * trailer packets.
546 * We no longer reply to requests for ETHERTYPE_TRAIL protocol either,
547 * but formerly didn't normally send requests.
548 */
549static int log_arp_wrong_iface = 1;
550
551SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_wrong_iface, CTLFLAG_RW,
552 &log_arp_wrong_iface, 0,
553 "log arp packets arriving on the wrong interface");
554
555static void
556in_arpinput(m)
557 struct mbuf *m;
558{
2256ba69
RG
559 struct arphdr *ah;
560 struct ifnet *ifp = m->m_pkthdr.rcvif;
984263bc
MD
561 struct ether_header *eh;
562 struct arc_header *arh;
563 struct iso88025_header *th = (struct iso88025_header *)0;
564 struct iso88025_sockaddr_dl_data *trld;
2256ba69
RG
565 struct llinfo_arp *la = 0;
566 struct rtentry *rt;
984263bc
MD
567 struct ifaddr *ifa;
568 struct in_ifaddr *ia;
569 struct sockaddr_dl *sdl;
570 struct sockaddr sa;
571 struct in_addr isaddr, itaddr, myaddr;
572 int op, rif_len;
573 int req_len;
574
575 req_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
576 if (m->m_len < req_len && (m = m_pullup(m, req_len)) == NULL) {
577 log(LOG_ERR, "in_arp: runt packet -- m_pullup failed\n");
578 return;
579 }
580
581 ah = mtod(m, struct arphdr *);
582 op = ntohs(ah->ar_op);
583 (void)memcpy(&isaddr, ar_spa(ah), sizeof (isaddr));
584 (void)memcpy(&itaddr, ar_tpa(ah), sizeof (itaddr));
585#ifdef BRIDGE
586#define BRIDGE_TEST (do_bridge)
587#else
588#define BRIDGE_TEST (0) /* cc will optimise the test away */
589#endif
590 /*
591 * For a bridge, we want to check the address irrespective
592 * of the receive interface. (This will change slightly
593 * when we have clusters of interfaces).
594 */
595 LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash)
596 if ((BRIDGE_TEST || (ia->ia_ifp == ifp)) &&
597 itaddr.s_addr == ia->ia_addr.sin_addr.s_addr)
598 goto match;
599 LIST_FOREACH(ia, INADDR_HASH(isaddr.s_addr), ia_hash)
600 if ((BRIDGE_TEST || (ia->ia_ifp == ifp)) &&
601 isaddr.s_addr == ia->ia_addr.sin_addr.s_addr)
602 goto match;
603 /*
604 * No match, use the first inet address on the receive interface
605 * as a dummy address for the rest of the function.
606 */
607 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
608 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET) {
609 ia = ifatoia(ifa);
610 goto match;
611 }
612 /*
613 * If bridging, fall back to using any inet address.
614 * This is probably incorrect, the right way being try to match
615 * addresses for interfaces in the same cluster, so if we
616 * get here we should always drop the packet.
617 */
618 if (!BRIDGE_TEST ||
619 (ia = TAILQ_FIRST(&in_ifaddrhead)) == NULL) {
620 m_freem(m);
621 return;
622 }
623match:
624 myaddr = ia->ia_addr.sin_addr;
625 if (!bcmp(ar_sha(ah), IF_LLADDR(ifp), ifp->if_addrlen)) {
626 m_freem(m); /* it's from me, ignore it. */
627 return;
628 }
c401f0fd 629 if (!bcmp(ar_sha(ah), ifp->if_broadcastaddr, ifp->if_addrlen)) {
984263bc
MD
630 log(LOG_ERR,
631 "arp: link address is broadcast for IP address %s!\n",
632 inet_ntoa(isaddr));
633 m_freem(m);
634 return;
635 }
636 if (isaddr.s_addr == myaddr.s_addr) {
637 log(LOG_ERR,
638 "arp: %*D is using my IP address %s!\n",
639 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
640 inet_ntoa(isaddr));
641 itaddr = myaddr;
642 goto reply;
643 }
644 la = arplookup(isaddr.s_addr, itaddr.s_addr == myaddr.s_addr, 0);
645 if (la && (rt = la->la_rt) && (sdl = SDL(rt->rt_gateway))) {
646 /* the following is not an error when doing bridging */
647 if (!BRIDGE_TEST && rt->rt_ifp != ifp) {
648 if (log_arp_wrong_iface)
3e4a09e7 649 log(LOG_ERR, "arp: %s is on %s but got reply from %*D on %s\n",
984263bc 650 inet_ntoa(isaddr),
3e4a09e7 651 rt->rt_ifp->if_xname,
984263bc 652 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
3e4a09e7 653 ifp->if_xname);
984263bc
MD
654 goto reply;
655 }
656 if (sdl->sdl_alen &&
657 bcmp(ar_sha(ah), LLADDR(sdl), sdl->sdl_alen)) {
658 if (rt->rt_expire)
3e4a09e7 659 log(LOG_INFO, "arp: %s moved from %*D to %*D on %s\n",
984263bc
MD
660 inet_ntoa(isaddr),
661 ifp->if_addrlen, (u_char *)LLADDR(sdl), ":",
662 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
3e4a09e7 663 ifp->if_xname);
984263bc
MD
664 else {
665 log(LOG_ERR,
3e4a09e7 666 "arp: %*D attempts to modify permanent entry for %s on %s\n",
984263bc 667 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
3e4a09e7 668 inet_ntoa(isaddr), ifp->if_xname);
984263bc
MD
669 goto reply;
670 }
671 }
672 /*
673 * sanity check for the address length.
674 * XXX this does not work for protocols with variable address
675 * length. -is
676 */
677 if (sdl->sdl_alen &&
678 sdl->sdl_alen != ah->ar_hln) {
679 log(LOG_WARNING,
680 "arp from %*D: new addr len %d, was %d",
681 ifp->if_addrlen, (u_char *) ar_sha(ah), ":",
682 ah->ar_hln, sdl->sdl_alen);
683 }
684 if (ifp->if_addrlen != ah->ar_hln) {
685 log(LOG_WARNING,
686 "arp from %*D: addr len: new %d, i/f %d (ignored)",
687 ifp->if_addrlen, (u_char *) ar_sha(ah), ":",
688 ah->ar_hln, ifp->if_addrlen);
689 goto reply;
690 }
691 (void)memcpy(LLADDR(sdl), ar_sha(ah),
692 sdl->sdl_alen = ah->ar_hln);
693 /*
694 * If we receive an arp from a token-ring station over
695 * a token-ring nic then try to save the source
696 * routing info.
697 */
698 if (ifp->if_type == IFT_ISO88025) {
699 th = (struct iso88025_header *)m->m_pkthdr.header;
700 trld = SDL_ISO88025(sdl);
701 rif_len = TR_RCF_RIFLEN(th->rcf);
702 if ((th->iso88025_shost[0] & TR_RII) &&
703 (rif_len > 2)) {
704 trld->trld_rcf = th->rcf;
705 trld->trld_rcf ^= htons(TR_RCF_DIR);
706 memcpy(trld->trld_route, th->rd, rif_len - 2);
707 trld->trld_rcf &= ~htons(TR_RCF_BCST_MASK);
708 /*
709 * Set up source routing information for
710 * reply packet (XXX)
711 */
712 m->m_data -= rif_len;
713 m->m_len += rif_len;
714 m->m_pkthdr.len += rif_len;
715 } else {
716 th->iso88025_shost[0] &= ~TR_RII;
717 trld->trld_rcf = 0;
718 }
719 m->m_data -= 8;
720 m->m_len += 8;
721 m->m_pkthdr.len += 8;
722 th->rcf = trld->trld_rcf;
723 }
724 if (rt->rt_expire)
725 rt->rt_expire = time_second + arpt_keep;
726 rt->rt_flags &= ~RTF_REJECT;
727 la->la_preempt = la->la_asked = 0;
728 if (la->la_hold) {
729 (*ifp->if_output)(ifp, la->la_hold,
730 rt_key(rt), rt);
731 la->la_hold = 0;
732 }
733 }
734reply:
735 if (op != ARPOP_REQUEST) {
736 m_freem(m);
737 return;
738 }
739 if (itaddr.s_addr == myaddr.s_addr) {
740 /* I am the target */
741 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
742 (void)memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln);
743 } else {
744 la = arplookup(itaddr.s_addr, 0, SIN_PROXY);
745 if (la == NULL) {
746 struct sockaddr_in sin;
747
748 if (!arp_proxyall) {
749 m_freem(m);
750 return;
751 }
752
753 bzero(&sin, sizeof sin);
754 sin.sin_family = AF_INET;
755 sin.sin_len = sizeof sin;
756 sin.sin_addr = itaddr;
757
758 rt = rtalloc1((struct sockaddr *)&sin, 0, 0UL);
759 if (!rt) {
760 m_freem(m);
761 return;
762 }
763 /*
764 * Don't send proxies for nodes on the same interface
765 * as this one came out of, or we'll get into a fight
766 * over who claims what Ether address.
767 */
768 if (rt->rt_ifp == ifp) {
769 rtfree(rt);
770 m_freem(m);
771 return;
772 }
773 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
774 (void)memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln);
775 rtfree(rt);
776#ifdef DEBUG_PROXY
777 printf("arp: proxying for %s\n",
778 inet_ntoa(itaddr));
779#endif
780 } else {
781 rt = la->la_rt;
782 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
783 sdl = SDL(rt->rt_gateway);
784 (void)memcpy(ar_sha(ah), LLADDR(sdl), ah->ar_hln);
785 }
786 }
787
788 (void)memcpy(ar_tpa(ah), ar_spa(ah), ah->ar_pln);
789 (void)memcpy(ar_spa(ah), &itaddr, ah->ar_pln);
790 ah->ar_op = htons(ARPOP_REPLY);
791 ah->ar_pro = htons(ETHERTYPE_IP); /* let's be sure! */
792 switch (ifp->if_type) {
793 case IFT_ARCNET:
794 arh = (struct arc_header *)sa.sa_data;
795 arh->arc_dhost = *ar_tha(ah);
796 arh->arc_type = ARCTYPE_ARP;
797 break;
798
799 case IFT_ISO88025:
800 /* Re-arrange the source/dest address */
801 memcpy(th->iso88025_dhost, th->iso88025_shost,
802 sizeof(th->iso88025_dhost));
803 memcpy(th->iso88025_shost, IF_LLADDR(ifp),
804 sizeof(th->iso88025_shost));
805 /* Set the source routing bit if neccesary */
806 if (th->iso88025_dhost[0] & TR_RII) {
807 th->iso88025_dhost[0] &= ~TR_RII;
808 if (TR_RCF_RIFLEN(th->rcf) > 2)
809 th->iso88025_shost[0] |= TR_RII;
810 }
811 /* Copy the addresses, ac and fc into sa_data */
812 memcpy(sa.sa_data, th->iso88025_dhost,
813 sizeof(th->iso88025_dhost) * 2);
814 sa.sa_data[(sizeof(th->iso88025_dhost) * 2)] = TR_AC;
815 sa.sa_data[(sizeof(th->iso88025_dhost) * 2) + 1] = TR_LLC_FRAME;
816 break;
817 case IFT_ETHER:
818 case IFT_FDDI:
819 /*
820 * May not be correct for types not explictly
821 * listed, but it is our best guess.
822 */
823 default:
824 eh = (struct ether_header *)sa.sa_data;
825 (void)memcpy(eh->ether_dhost, ar_tha(ah),
826 sizeof(eh->ether_dhost));
827 eh->ether_type = htons(ETHERTYPE_ARP);
828 break;
829 }
830 sa.sa_family = AF_UNSPEC;
831 sa.sa_len = sizeof(sa);
832 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0);
833 return;
834}
835#endif
836
837/*
838 * Free an arp entry.
839 */
840static void
841arptfree(la)
2256ba69 842 struct llinfo_arp *la;
984263bc 843{
2256ba69
RG
844 struct rtentry *rt = la->la_rt;
845 struct sockaddr_dl *sdl;
984263bc
MD
846 if (rt == 0)
847 panic("arptfree");
848 if (rt->rt_refcnt > 0 && (sdl = SDL(rt->rt_gateway)) &&
849 sdl->sdl_family == AF_LINK) {
850 sdl->sdl_alen = 0;
851 la->la_preempt = la->la_asked = 0;
852 rt->rt_flags &= ~RTF_REJECT;
853 return;
854 }
855 rtrequest(RTM_DELETE, rt_key(rt), (struct sockaddr *)0, rt_mask(rt),
856 0, (struct rtentry **)0);
857}
858/*
859 * Lookup or enter a new address in arptab.
860 */
861static struct llinfo_arp *
862arplookup(addr, create, proxy)
863 u_long addr;
864 int create, proxy;
865{
2256ba69 866 struct rtentry *rt;
984263bc
MD
867 static struct sockaddr_inarp sin = {sizeof(sin), AF_INET };
868 const char *why = 0;
869
870 sin.sin_addr.s_addr = addr;
871 sin.sin_other = proxy ? SIN_PROXY : 0;
872 rt = rtalloc1((struct sockaddr *)&sin, create, 0UL);
873 if (rt == 0)
874 return (0);
875 rt->rt_refcnt--;
876
877 if (rt->rt_flags & RTF_GATEWAY)
878 why = "host is not on local network";
879 else if ((rt->rt_flags & RTF_LLINFO) == 0)
880 why = "could not allocate llinfo";
881 else if (rt->rt_gateway->sa_family != AF_LINK)
882 why = "gateway route is not ours";
883
8fd3e627
MD
884 if (why) {
885 if (create) {
886 log(LOG_DEBUG, "arplookup %s failed: %s\n",
887 inet_ntoa(sin.sin_addr), why);
888 }
889
890 /* if there are no references to this route, purge it */
891 if (rt->rt_refcnt <= 0 &&
892 (rt->rt_flags & RTF_WASCLONED) == RTF_WASCLONED) {
893 rtrequest(RTM_DELETE,
894 (struct sockaddr *)rt_key(rt), rt->rt_gateway,
895 rt_mask(rt), rt->rt_flags, 0);
896 }
897 return (0);
984263bc
MD
898 }
899 return ((struct llinfo_arp *)rt->rt_llinfo);
900}
901
902void
903arp_ifinit(ifp, ifa)
904 struct ifnet *ifp;
905 struct ifaddr *ifa;
906{
907 if (ntohl(IA_SIN(ifa)->sin_addr.s_addr) != INADDR_ANY)
908 arprequest(ifp, &IA_SIN(ifa)->sin_addr,
909 &IA_SIN(ifa)->sin_addr, IF_LLADDR(ifp));
910 ifa->ifa_rtrequest = arp_rtrequest;
911 ifa->ifa_flags |= RTF_CLONING;
912}
913
914static void
915arp_init(void)
916{
917 LIST_INIT(&llinfo_arp);
bf82f9b7 918 netisr_register(NETISR_ARP, cpu0_portfn, arpintr);
984263bc
MD
919}
920
921SYSINIT(arp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, arp_init, 0);