Document that only sockaddr_storage sized parameters can hold
[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
33/*
34 * Copyright (c) 2004, 2005 Jeffrey M. Hsu. All rights reserved.
35 *
36 * License terms: all terms for the DragonFly license above plus the following:
37 *
38 * 4. All advertising materials mentioning features or use of this software
39 * must display the following acknowledgement:
40 *
41 * This product includes software developed by Jeffrey M. Hsu
42 * for the DragonFly Project.
43 *
44 * This requirement may be waived with permission from Jeffrey Hsu.
45 * Permission will be granted to any DragonFly user for free.
46 * This requirement will sunset and may be removed on Jan 31, 2006,
47 * after which the standard DragonFly license (as shown above) will
48 * apply.
49 */
50
984263bc
MD
51/*
52 * Copyright (c) 1982, 1986, 1988, 1993
53 * The Regents of the University of California. All rights reserved.
54 *
55 * Redistribution and use in source and binary forms, with or without
56 * modification, are permitted provided that the following conditions
57 * are met:
58 * 1. Redistributions of source code must retain the above copyright
59 * notice, this list of conditions and the following disclaimer.
60 * 2. Redistributions in binary form must reproduce the above copyright
61 * notice, this list of conditions and the following disclaimer in the
62 * documentation and/or other materials provided with the distribution.
63 * 3. All advertising materials mentioning features or use of this software
64 * must display the following acknowledgement:
65 * This product includes software developed by the University of
66 * California, Berkeley and its contributors.
67 * 4. Neither the name of the University nor the names of its contributors
68 * may be used to endorse or promote products derived from this software
69 * without specific prior written permission.
70 *
71 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
72 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
73 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
74 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
75 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
76 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
77 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
78 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
79 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
80 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
81 * SUCH DAMAGE.
82 *
83 * @(#)if_ether.c 8.1 (Berkeley) 6/10/93
84 * $FreeBSD: src/sys/netinet/if_ether.c,v 1.64.2.23 2003/04/11 07:23:15 fjoe Exp $
6aad077d 85 * $DragonFly: src/sys/netinet/if_ether.c,v 1.32 2006/05/20 06:32:40 dillon Exp $
984263bc
MD
86 */
87
88/*
89 * Ethernet address resolution protocol.
90 * TODO:
91 * add "inuse/lock" bit (or ref. count) along with valid bit
92 */
93
94#include "opt_inet.h"
95#include "opt_bdg.h"
96
97#include <sys/param.h>
98#include <sys/kernel.h>
99#include <sys/queue.h>
100#include <sys/sysctl.h>
101#include <sys/systm.h>
102#include <sys/mbuf.h>
103#include <sys/malloc.h>
104#include <sys/socket.h>
105#include <sys/syslog.h>
106
a80cf23b
JH
107#include <sys/thread2.h>
108#include <sys/msgport2.h>
109
984263bc
MD
110#include <net/if.h>
111#include <net/if_dl.h>
112#include <net/if_types.h>
113#include <net/route.h>
114#include <net/netisr.h>
115#include <net/if_llc.h>
116#ifdef BRIDGE
117#include <net/ethernet.h>
7b3af414 118#include <net/oldbridge/bridge.h>
984263bc
MD
119#endif
120
121#include <netinet/in.h>
122#include <netinet/in_var.h>
123#include <netinet/if_ether.h>
124
125#include <net/if_arc.h>
126#include <net/iso88025.h>
127
128#define SIN(s) ((struct sockaddr_in *)s)
129#define SDL(s) ((struct sockaddr_dl *)s)
130
131SYSCTL_DECL(_net_link_ether);
132SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, "");
133
134/* timer values */
135static int arpt_prune = (5*60*1); /* walk list every 5 minutes */
136static int arpt_keep = (20*60); /* once resolved, good for 20 more minutes */
137static int arpt_down = 20; /* once declared down, don't send for 20 sec */
138
139SYSCTL_INT(_net_link_ether_inet, OID_AUTO, prune_intvl, CTLFLAG_RW,
140 &arpt_prune, 0, "");
f23061d4 141SYSCTL_INT(_net_link_ether_inet, OID_AUTO, max_age, CTLFLAG_RW,
984263bc
MD
142 &arpt_keep, 0, "");
143SYSCTL_INT(_net_link_ether_inet, OID_AUTO, host_down_time, CTLFLAG_RW,
144 &arpt_down, 0, "");
145
f23061d4 146#define rt_expire rt_rmx.rmx_expire
984263bc
MD
147
148struct llinfo_arp {
149 LIST_ENTRY(llinfo_arp) la_le;
150 struct rtentry *la_rt;
151 struct mbuf *la_hold; /* last packet until resolved/timeout */
a916e847 152 u_short la_preempt; /* countdown for pre-expiry arps */
984263bc 153 u_short la_asked; /* #times we QUERIED following expiration */
984263bc
MD
154};
155
ecdefdda 156static LIST_HEAD(, llinfo_arp) llinfo_arp_list[MAXCPU];
984263bc 157
984263bc
MD
158static int arp_maxtries = 5;
159static int useloopback = 1; /* use loopback interface for local traffic */
160static int arp_proxyall = 0;
161
162SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW,
163 &arp_maxtries, 0, "");
164SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW,
165 &useloopback, 0, "");
166SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW,
167 &arp_proxyall, 0, "");
168
42a7fc75
RG
169static void arp_rtrequest (int, struct rtentry *, struct rt_addrinfo *);
170static void arprequest (struct ifnet *,
171 struct in_addr *, struct in_addr *, u_char *);
b76bed62 172static int arpintr(struct netmsg *);
42a7fc75
RG
173static void arptfree (struct llinfo_arp *);
174static void arptimer (void *);
984263bc 175static struct llinfo_arp
f23061d4 176 *arplookup (in_addr_t addr, boolean_t create, boolean_t proxy);
984263bc 177#ifdef INET
42a7fc75 178static void in_arpinput (struct mbuf *);
984263bc
MD
179#endif
180
ecdefdda 181static struct callout arptimer_ch[MAXCPU];
ac311540 182
984263bc
MD
183/*
184 * Timeout routine. Age arp_tab entries periodically.
185 */
186/* ARGSUSED */
187static void
f23061d4 188arptimer(void *ignored_arg)
984263bc 189{
2e9572df 190 struct llinfo_arp *la, *nla;
984263bc 191
1cae611f 192 crit_enter();
ecdefdda 193 LIST_FOREACH_MUTABLE(la, &llinfo_arp_list[mycpuid], la_le, nla) {
2e9572df 194 if (la->la_rt->rt_expire && la->la_rt->rt_expire <= time_second)
ecdefdda 195 arptfree(la);
984263bc 196 }
ecdefdda 197 callout_reset(&arptimer_ch[mycpuid], arpt_prune * hz, arptimer, NULL);
1cae611f 198 crit_exit();
984263bc
MD
199}
200
201/*
202 * Parallel to llc_rtrequest.
203 */
204static void
f23061d4 205arp_rtrequest(int req, struct rtentry *rt, struct rt_addrinfo *info)
984263bc 206{
2256ba69 207 struct sockaddr *gate = rt->rt_gateway;
f23061d4
JH
208 struct llinfo_arp *la = rt->rt_llinfo;
209
5fe66e68 210 struct sockaddr_dl null_sdl = { sizeof null_sdl, AF_LINK };
ecdefdda 211 static boolean_t arpinit_done[MAXCPU];
984263bc 212
ecdefdda
MD
213 if (!arpinit_done[mycpuid]) {
214 arpinit_done[mycpuid] = TRUE;
215 callout_init(&arptimer_ch[mycpuid]);
216 callout_reset(&arptimer_ch[mycpuid], hz, arptimer, NULL);
984263bc
MD
217 }
218 if (rt->rt_flags & RTF_GATEWAY)
219 return;
984263bc 220
f23061d4 221 switch (req) {
984263bc
MD
222 case RTM_ADD:
223 /*
224 * XXX: If this is a manually added route to interface
225 * such as older version of routed or gated might provide,
226 * restore cloning bit.
227 */
f23061d4 228 if (!(rt->rt_flags & RTF_HOST) &&
984263bc
MD
229 SIN(rt_mask(rt))->sin_addr.s_addr != 0xffffffff)
230 rt->rt_flags |= RTF_CLONING;
231 if (rt->rt_flags & RTF_CLONING) {
232 /*
233 * Case 1: This route should come from a route to iface.
234 */
235 rt_setgate(rt, rt_key(rt),
0c3c561c 236 (struct sockaddr *)&null_sdl);
984263bc
MD
237 gate = rt->rt_gateway;
238 SDL(gate)->sdl_type = rt->rt_ifp->if_type;
239 SDL(gate)->sdl_index = rt->rt_ifp->if_index;
240 rt->rt_expire = time_second;
241 break;
242 }
243 /* Announce a new entry if requested. */
244 if (rt->rt_flags & RTF_ANNOUNCE)
245 arprequest(rt->rt_ifp,
246 &SIN(rt_key(rt))->sin_addr,
247 &SIN(rt_key(rt))->sin_addr,
f23061d4 248 LLADDR(SDL(gate)));
984263bc
MD
249 /*FALLTHROUGH*/
250 case RTM_RESOLVE:
251 if (gate->sa_family != AF_LINK ||
f23061d4 252 gate->sa_len < sizeof(struct sockaddr_dl)) {
984263bc
MD
253 log(LOG_DEBUG, "arp_rtrequest: bad gateway value\n");
254 break;
255 }
256 SDL(gate)->sdl_type = rt->rt_ifp->if_type;
257 SDL(gate)->sdl_index = rt->rt_ifp->if_index;
f23061d4 258 if (la != NULL)
984263bc
MD
259 break; /* This happens on a route change */
260 /*
261 * Case 2: This route may come from cloning, or a manual route
262 * add with a LL address.
263 */
0c3c561c 264 R_Malloc(la, struct llinfo_arp *, sizeof *la);
f23061d4
JH
265 rt->rt_llinfo = la;
266 if (la == NULL) {
984263bc
MD
267 log(LOG_DEBUG, "arp_rtrequest: malloc failed\n");
268 break;
269 }
f23061d4 270 bzero(la, sizeof *la);
984263bc
MD
271 la->la_rt = rt;
272 rt->rt_flags |= RTF_LLINFO;
ecdefdda 273 LIST_INSERT_HEAD(&llinfo_arp_list[mycpuid], la, la_le);
984263bc
MD
274
275#ifdef INET
276 /*
277 * This keeps the multicast addresses from showing up
278 * in `arp -a' listings as unresolved. It's not actually
279 * functional. Then the same for broadcast.
280 */
f23061d4
JH
281 if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr)) &&
282 rt->rt_ifp->if_type != IFT_ARCNET) {
984263bc
MD
283 ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt))->sin_addr,
284 LLADDR(SDL(gate)));
285 SDL(gate)->sdl_alen = 6;
286 rt->rt_expire = 0;
287 }
288 if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) {
c401f0fd 289 memcpy(LLADDR(SDL(gate)), rt->rt_ifp->if_broadcastaddr,
984263bc
MD
290 rt->rt_ifp->if_addrlen);
291 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen;
292 rt->rt_expire = 0;
293 }
294#endif
295
296 if (SIN(rt_key(rt))->sin_addr.s_addr ==
297 (IA_SIN(rt->rt_ifa))->sin_addr.s_addr) {
f23061d4
JH
298 /*
299 * This test used to be
300 * if (loif.if_flags & IFF_UP)
301 * It allowed local traffic to be forced
302 * through the hardware by configuring the
303 * loopback down. However, it causes problems
304 * during network configuration for boards
305 * that can't receive packets they send. It
306 * is now necessary to clear "useloopback" and
307 * remove the route to force traffic out to
308 * the hardware.
309 */
984263bc 310 rt->rt_expire = 0;
2e9572df 311 bcopy(IF_LLADDR(rt->rt_ifp), LLADDR(SDL(gate)),
984263bc
MD
312 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen);
313 if (useloopback)
314 rt->rt_ifp = loif;
984263bc
MD
315 }
316 break;
317
318 case RTM_DELETE:
f23061d4 319 if (la == NULL)
984263bc 320 break;
984263bc 321 LIST_REMOVE(la, la_le);
f23061d4 322 rt->rt_llinfo = NULL;
984263bc 323 rt->rt_flags &= ~RTF_LLINFO;
f23061d4 324 if (la->la_hold != NULL)
984263bc 325 m_freem(la->la_hold);
f23061d4 326 Free(la);
984263bc
MD
327 }
328}
329
330/*
331 * Broadcast an ARP request. Caller specifies:
332 * - arp header source ip address
333 * - arp header target ip address
334 * - arp header source ethernet address
335 */
336static void
f23061d4
JH
337arprequest(struct ifnet *ifp, struct in_addr *sip, struct in_addr *tip,
338 u_char *enaddr)
984263bc 339{
2256ba69
RG
340 struct mbuf *m;
341 struct ether_header *eh;
342 struct arc_header *arh;
343 struct arphdr *ah;
984263bc 344 struct sockaddr sa;
f23061d4
JH
345 static u_char llcx[] = { 0x82, 0x40, LLC_SNAP_LSAP, LLC_SNAP_LSAP,
346 LLC_UI, 0x00, 0x00, 0x00, 0x08, 0x06 };
984263bc
MD
347 u_short ar_hrd;
348
74f1caca 349 if ((m = m_gethdr(MB_DONTWAIT, MT_DATA)) == NULL)
984263bc 350 return;
f23061d4
JH
351 m->m_pkthdr.rcvif = (struct ifnet *)NULL;
352
984263bc
MD
353 switch (ifp->if_type) {
354 case IFT_ARCNET:
355 ar_hrd = htons(ARPHRD_ARCNET);
356
357 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
358 m->m_pkthdr.len = m->m_len;
359 MH_ALIGN(m, m->m_len);
360
361 arh = (struct arc_header *)sa.sa_data;
c401f0fd 362 arh->arc_dhost = ifp->if_broadcastaddr[0];
984263bc
MD
363 arh->arc_type = ARCTYPE_ARP;
364
365 ah = mtod(m, struct arphdr *);
366 break;
367
368 case IFT_ISO88025:
369 ar_hrd = htons(ARPHRD_IEEE802);
370
0c3c561c 371 m->m_len = (sizeof llcx) +
984263bc
MD
372 arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
373 m->m_pkthdr.len = m->m_len;
374 MH_ALIGN(m, m->m_len);
375
0c3c561c 376 memcpy(mtod(m, caddr_t), llcx, sizeof llcx);
c401f0fd 377 memcpy(sa.sa_data, ifp->if_broadcastaddr, ifp->if_addrlen);
f23061d4 378 memcpy(sa.sa_data + 6, enaddr, 6);
984263bc
MD
379 sa.sa_data[6] |= TR_RII;
380 sa.sa_data[12] = TR_AC;
381 sa.sa_data[13] = TR_LLC_FRAME;
382
0c3c561c 383 ah = (struct arphdr *)(mtod(m, char *) + sizeof llcx);
984263bc
MD
384 break;
385 case IFT_FDDI:
386 case IFT_ETHER:
387 /*
388 * This may not be correct for types not explicitly
389 * listed, but this is our best guess
390 */
391 default:
392 ar_hrd = htons(ARPHRD_ETHER);
393
394 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
395 m->m_pkthdr.len = m->m_len;
396 MH_ALIGN(m, m->m_len);
397
398 eh = (struct ether_header *)sa.sa_data;
f23061d4 399 /* if_output() will not swap */
984263bc 400 eh->ether_type = htons(ETHERTYPE_ARP);
c401f0fd 401 memcpy(eh->ether_dhost, ifp->if_broadcastaddr, ifp->if_addrlen);
984263bc
MD
402
403 ah = mtod(m, struct arphdr *);
404 break;
405 }
406
407 ah->ar_hrd = ar_hrd;
408 ah->ar_pro = htons(ETHERTYPE_IP);
409 ah->ar_hln = ifp->if_addrlen; /* hardware address length */
410 ah->ar_pln = sizeof(struct in_addr); /* protocol address length */
411 ah->ar_op = htons(ARPOP_REQUEST);
f23061d4 412 memcpy(ar_sha(ah), enaddr, ah->ar_hln);
984263bc 413 memset(ar_tha(ah), 0, ah->ar_hln);
f23061d4
JH
414 memcpy(ar_spa(ah), sip, ah->ar_pln);
415 memcpy(ar_tpa(ah), tip, ah->ar_pln);
984263bc
MD
416
417 sa.sa_family = AF_UNSPEC;
0c3c561c 418 sa.sa_len = sizeof sa;
f23061d4 419 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)NULL);
984263bc
MD
420}
421
422/*
423 * Resolve an IP address into an ethernet address. If success,
424 * desten is filled in. If there is no entry in arptab,
425 * set one up and broadcast a request for the IP address.
426 * Hold onto this mbuf and resend it once the address
427 * is finally resolved. A return value of 1 indicates
428 * that desten has been filled in and the packet should be sent
429 * normally; a 0 return indicates that the packet has been
430 * taken over here, either now or for later transmission.
431 */
432int
f23061d4
JH
433arpresolve(
434 struct ifnet *ifp,
435 struct rtentry *rt0,
436 struct mbuf *m,
437 struct sockaddr *dst,
438 u_char *desten)
984263bc 439{
f23061d4
JH
440 struct rtentry *rt;
441 struct llinfo_arp *la = NULL;
984263bc
MD
442 struct sockaddr_dl *sdl;
443
444 if (m->m_flags & M_BCAST) { /* broadcast */
c401f0fd 445 memcpy(desten, ifp->if_broadcastaddr, ifp->if_addrlen);
984263bc
MD
446 return (1);
447 }
448 if (m->m_flags & M_MCAST && ifp->if_type != IFT_ARCNET) {/* multicast */
449 ETHER_MAP_IP_MULTICAST(&SIN(dst)->sin_addr, desten);
f23061d4 450 return (1);
984263bc 451 }
f23061d4
JH
452 if (rt0 != NULL) {
453 if (rt_llroute(dst, rt0, &rt) != 0) {
454 m_freem(m);
455 return 0;
456 }
457 la = rt->rt_llinfo;
458 }
459 if (la == NULL) {
460 la = arplookup(SIN(dst)->sin_addr.s_addr, TRUE, FALSE);
461 if (la != NULL)
984263bc
MD
462 rt = la->la_rt;
463 }
f23061d4 464 if (la == NULL || rt == NULL) {
984263bc 465 log(LOG_DEBUG, "arpresolve: can't allocate llinfo for %s%s%s\n",
5fe66e68
JH
466 inet_ntoa(SIN(dst)->sin_addr), la ? "la" : " ",
467 rt ? "rt" : "");
984263bc
MD
468 m_freem(m);
469 return (0);
470 }
471 sdl = SDL(rt->rt_gateway);
472 /*
473 * Check the address family and length is valid, the address
474 * is resolved; otherwise, try to resolve.
475 */
476 if ((rt->rt_expire == 0 || rt->rt_expire > time_second) &&
477 sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0) {
478 /*
479 * If entry has an expiry time and it is approaching,
480 * see if we need to send an ARP request within this
481 * arpt_down interval.
482 */
483 if ((rt->rt_expire != 0) &&
a916e847 484 (time_second + la->la_preempt > rt->rt_expire)) {
984263bc
MD
485 arprequest(ifp,
486 &SIN(rt->rt_ifa->ifa_addr)->sin_addr,
487 &SIN(dst)->sin_addr,
488 IF_LLADDR(ifp));
a916e847 489 la->la_preempt--;
f23061d4 490 }
984263bc
MD
491
492 bcopy(LLADDR(sdl), desten, sdl->sdl_alen);
493 return 1;
494 }
495 /*
496 * If ARP is disabled on this interface, stop.
497 * XXX
498 * Probably should not allocate empty llinfo struct if we are
499 * not going to be sending out an arp request.
500 */
501 if (ifp->if_flags & IFF_NOARP) {
502 m_freem(m);
503 return (0);
504 }
505 /*
506 * There is an arptab entry, but no ethernet address
507 * response yet. Replace the held mbuf with this
508 * latest one.
509 */
f23061d4 510 if (la->la_hold != NULL)
984263bc
MD
511 m_freem(la->la_hold);
512 la->la_hold = m;
03f869ef 513 if (rt->rt_expire || ((rt->rt_flags & RTF_STATIC) && !sdl->sdl_alen)) {
984263bc
MD
514 rt->rt_flags &= ~RTF_REJECT;
515 if (la->la_asked == 0 || rt->rt_expire != time_second) {
516 rt->rt_expire = time_second;
517 if (la->la_asked++ < arp_maxtries) {
518 arprequest(ifp,
519 &SIN(rt->rt_ifa->ifa_addr)->sin_addr,
520 &SIN(dst)->sin_addr,
521 IF_LLADDR(ifp));
522 } else {
523 rt->rt_flags |= RTF_REJECT;
524 rt->rt_expire += arpt_down;
a916e847
JH
525 la->la_asked = 0;
526 la->la_preempt = arp_maxtries;
984263bc
MD
527 }
528
529 }
530 }
531 return (0);
532}
533
534/*
535 * Common length and type checks are done here,
536 * then the protocol-specific routine is called.
537 */
b76bed62 538static int
9eeaa8a9 539arpintr(struct netmsg *msg)
984263bc 540{
9eeaa8a9 541 struct mbuf *m = ((struct netmsg_packet *)msg)->nm_packet;
2256ba69 542 struct arphdr *ar;
8bde602d 543 u_short ar_hrd;
984263bc 544
8bde602d
JH
545 if (m->m_len < sizeof(struct arphdr) &&
546 ((m = m_pullup(m, sizeof(struct arphdr))) == NULL)) {
547 log(LOG_ERR, "arp: runt packet -- m_pullup failed\n");
b76bed62 548 goto out2;
8bde602d
JH
549 }
550 ar = mtod(m, struct arphdr *);
551
552 ar_hrd = ntohs(ar->ar_hrd);
553 if (ar_hrd != ARPHRD_ETHER &&
554 ar_hrd != ARPHRD_IEEE802 &&
555 ar_hrd != ARPHRD_ARCNET) {
556 log(LOG_ERR,
557 "arp: unknown hardware address format (0x%2D)\n",
558 (unsigned char *)&ar->ar_hrd, "");
b76bed62 559 goto out1;
8bde602d 560 }
984263bc 561
8bde602d
JH
562 if (m->m_pkthdr.len < arphdr_len(ar) &&
563 (m = m_pullup(m, arphdr_len(ar))) == NULL) {
564 log(LOG_ERR, "arp: runt packet\n");
b76bed62 565 goto out1;
8bde602d
JH
566 }
567
568 switch (ntohs(ar->ar_pro)) {
984263bc 569#ifdef INET
8bde602d
JH
570 case ETHERTYPE_IP:
571 in_arpinput(m);
b76bed62 572 goto out2;
984263bc 573#endif
984263bc 574 }
b76bed62 575out1:
8bde602d 576 m_freem(m);
b76bed62 577out2:
6aad077d 578 /* msg was embedded in the mbuf, do not reply! */
b76bed62 579 return(EASYNC);
984263bc
MD
580}
581
582#ifdef INET
583/*
584 * ARP for Internet protocols on 10 Mb/s Ethernet.
585 * Algorithm is that given in RFC 826.
586 * In addition, a sanity check is performed on the sender
587 * protocol address, to catch impersonators.
588 * We no longer handle negotiations for use of trailer protocol:
589 * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent
590 * along with IP replies if we wanted trailers sent to us,
591 * and also sent them in response to IP replies.
592 * This allowed either end to announce the desire to receive
593 * trailer packets.
594 * We no longer reply to requests for ETHERTYPE_TRAIL protocol either,
595 * but formerly didn't normally send requests.
596 */
597static int log_arp_wrong_iface = 1;
984263bc
MD
598SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_wrong_iface, CTLFLAG_RW,
599 &log_arp_wrong_iface, 0,
600 "log arp packets arriving on the wrong interface");
601
ecdefdda
MD
602#ifdef BRIDGE
603#define BRIDGE_TEST (do_bridge)
604#else
605#define BRIDGE_TEST (0) /* cc will optimise the test away */
606#endif
607
608static void
609arp_update_oncpu(struct mbuf *m, in_addr_t saddr, boolean_t create,
610 boolean_t dologging)
611{
612 struct arphdr *ah = mtod(m, struct arphdr *);
613 struct ifnet *ifp = m->m_pkthdr.rcvif;
614 struct llinfo_arp *la;
615 struct sockaddr_dl *sdl;
616 struct rtentry *rt;
617 int cpu = mycpuid;
618
619 la = arplookup(saddr, create, FALSE);
620 if (la && (rt = la->la_rt) && (sdl = SDL(rt->rt_gateway))) {
621 struct in_addr isaddr = { saddr };
622
623 /* the following is not an error when doing bridging */
624 if (!BRIDGE_TEST && rt->rt_ifp != ifp) {
625 if (dologging && log_arp_wrong_iface && cpu == 0) {
626 log(LOG_ERR,
627 "arp: %s is on %s "
628 "but got reply from %*D on %s\n",
629 inet_ntoa(isaddr),
630 rt->rt_ifp->if_xname,
631 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
632 ifp->if_xname);
633 }
634 return;
635 }
636 if (sdl->sdl_alen &&
637 bcmp(ar_sha(ah), LLADDR(sdl), sdl->sdl_alen)) {
638 if (rt->rt_expire != 0) {
639 if (dologging && cpu == 0) {
640 log(LOG_INFO,
641 "arp: %s moved from %*D to %*D on %s\n",
642 inet_ntoa(isaddr),
643 ifp->if_addrlen, (u_char *)LLADDR(sdl),
644 ":", ifp->if_addrlen,
645 (u_char *)ar_sha(ah), ":",
646 ifp->if_xname);
647 }
648 } else {
649 if (dologging && cpu == 0) {
650 log(LOG_ERR,
651 "arp: %*D attempts to modify permanent entry for %s on %s\n",
652 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
653 inet_ntoa(isaddr), ifp->if_xname);
654 }
655 return;
656 }
657 }
658 /*
659 * sanity check for the address length.
660 * XXX this does not work for protocols with variable address
661 * length. -is
662 */
663 if (dologging && sdl->sdl_alen && sdl->sdl_alen != ah->ar_hln &&
664 cpu == 0)
665 {
666 log(LOG_WARNING,
667 "arp from %*D: new addr len %d, was %d",
668 ifp->if_addrlen, (u_char *) ar_sha(ah), ":",
669 ah->ar_hln, sdl->sdl_alen);
670 }
671 if (ifp->if_addrlen != ah->ar_hln) {
672 if (dologging && cpu == 0) {
673 log(LOG_WARNING,
674 "arp from %*D: addr len: new %d, i/f %d (ignored)",
675 ifp->if_addrlen, (u_char *) ar_sha(ah), ":",
676 ah->ar_hln, ifp->if_addrlen);
677 }
678 return;
679 }
680 memcpy(LLADDR(sdl), ar_sha(ah), sdl->sdl_alen = ah->ar_hln);
681 /*
682 * If we receive an arp from a token-ring station over
683 * a token-ring nic then try to save the source
684 * routing info.
685 */
686 if (ifp->if_type == IFT_ISO88025) {
687 struct iso88025_header *th =
688 (struct iso88025_header *)m->m_pkthdr.header;
689 struct iso88025_sockaddr_dl_data *trld =
690 SDL_ISO88025(sdl);
691 int rif_len;
692
693 rif_len = TR_RCF_RIFLEN(th->rcf);
694 if ((th->iso88025_shost[0] & TR_RII) &&
695 (rif_len > 2)) {
696 trld->trld_rcf = th->rcf;
697 trld->trld_rcf ^= htons(TR_RCF_DIR);
698 memcpy(trld->trld_route, th->rd, rif_len - 2);
699 trld->trld_rcf &= ~htons(TR_RCF_BCST_MASK);
700 /*
701 * Set up source routing information for
702 * reply packet (XXX)
703 */
704 m->m_data -= rif_len;
705 m->m_len += rif_len;
706 m->m_pkthdr.len += rif_len;
707 } else {
708 th->iso88025_shost[0] &= ~TR_RII;
709 trld->trld_rcf = 0;
710 }
711 m->m_data -= 8;
712 m->m_len += 8;
713 m->m_pkthdr.len += 8;
714 th->rcf = trld->trld_rcf;
715 }
716 if (rt->rt_expire != 0)
717 rt->rt_expire = time_second + arpt_keep;
718 rt->rt_flags &= ~RTF_REJECT;
719 la->la_asked = 0;
720 la->la_preempt = arp_maxtries;
721
722 /*
723 * This particular cpu might have been holding an mbuf
724 * pending ARP resolution. If so, transmit the mbuf now.
725 */
726 if (la->la_hold != NULL) {
727 m_adj(la->la_hold, sizeof(struct ether_header));
728 lwkt_serialize_enter(ifp->if_serializer);
729 (*ifp->if_output)(ifp, la->la_hold, rt_key(rt), rt);
730 lwkt_serialize_exit(ifp->if_serializer);
731 la->la_hold = NULL;
732 }
733 }
734}
735
736#ifdef SMP
737
738struct netmsg_arp_update {
739 struct lwkt_msg lmsg;
740 struct mbuf *m;
741 in_addr_t saddr;
742 boolean_t create;
743};
744
745static int arp_update_msghandler(struct lwkt_msg *lmsg);
746
747#endif
748
749/*
750 * Called from arpintr() - this routine is run from a single cpu.
751 */
984263bc 752static void
f23061d4 753in_arpinput(struct mbuf *m)
984263bc 754{
2256ba69
RG
755 struct arphdr *ah;
756 struct ifnet *ifp = m->m_pkthdr.rcvif;
984263bc
MD
757 struct ether_header *eh;
758 struct arc_header *arh;
5fe66e68 759 struct iso88025_header *th = (struct iso88025_header *)NULL;
2256ba69 760 struct rtentry *rt;
984263bc
MD
761 struct ifaddr *ifa;
762 struct in_ifaddr *ia;
984263bc
MD
763 struct sockaddr sa;
764 struct in_addr isaddr, itaddr, myaddr;
ecdefdda
MD
765#ifdef SMP
766 struct netmsg_arp_update msg;
767#endif
768 int op;
984263bc
MD
769 int req_len;
770
771 req_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr));
772 if (m->m_len < req_len && (m = m_pullup(m, req_len)) == NULL) {
773 log(LOG_ERR, "in_arp: runt packet -- m_pullup failed\n");
774 return;
775 }
776
777 ah = mtod(m, struct arphdr *);
778 op = ntohs(ah->ar_op);
f23061d4
JH
779 memcpy(&isaddr, ar_spa(ah), sizeof isaddr);
780 memcpy(&itaddr, ar_tpa(ah), sizeof itaddr);
984263bc
MD
781 /*
782 * For a bridge, we want to check the address irrespective
783 * of the receive interface. (This will change slightly
784 * when we have clusters of interfaces).
785 */
ac93838f
SS
786 LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash) {
787 /*
788 * Old style bridging OBSOLETE
789 */
984263bc 790 if ((BRIDGE_TEST || (ia->ia_ifp == ifp)) &&
ac93838f 791 itaddr.s_addr == ia->ia_addr.sin_addr.s_addr) {
984263bc 792 goto match;
ac93838f
SS
793 }
794
795 /*
796 * New style bridging
797 */
798 if (ifp->if_bridge && ia->ia_ifp &&
799 ifp->if_bridge == ia->ia_ifp->if_bridge &&
800 itaddr.s_addr == ia->ia_addr.sin_addr.s_addr) {
801 goto match;
802 }
803 }
804 LIST_FOREACH(ia, INADDR_HASH(isaddr.s_addr), ia_hash) {
805 /*
806 * Old style bridging OBSOLETE
807 */
984263bc 808 if ((BRIDGE_TEST || (ia->ia_ifp == ifp)) &&
ac93838f 809 isaddr.s_addr == ia->ia_addr.sin_addr.s_addr) {
984263bc 810 goto match;
ac93838f
SS
811 }
812
813 /*
814 * New style bridging
815 */
816 if (ifp->if_bridge && ia->ia_ifp &&
817 ifp->if_bridge == ia->ia_ifp->if_bridge &&
818 isaddr.s_addr == ia->ia_addr.sin_addr.s_addr) {
819 goto match;
820 }
821 }
984263bc
MD
822 /*
823 * No match, use the first inet address on the receive interface
824 * as a dummy address for the rest of the function.
825 */
826 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
827 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET) {
828 ia = ifatoia(ifa);
829 goto match;
830 }
831 /*
832 * If bridging, fall back to using any inet address.
833 * This is probably incorrect, the right way being try to match
834 * addresses for interfaces in the same cluster, so if we
835 * get here we should always drop the packet.
836 */
837 if (!BRIDGE_TEST ||
838 (ia = TAILQ_FIRST(&in_ifaddrhead)) == NULL) {
839 m_freem(m);
840 return;
841 }
842match:
843 myaddr = ia->ia_addr.sin_addr;
844 if (!bcmp(ar_sha(ah), IF_LLADDR(ifp), ifp->if_addrlen)) {
845 m_freem(m); /* it's from me, ignore it. */
846 return;
847 }
c401f0fd 848 if (!bcmp(ar_sha(ah), ifp->if_broadcastaddr, ifp->if_addrlen)) {
984263bc
MD
849 log(LOG_ERR,
850 "arp: link address is broadcast for IP address %s!\n",
851 inet_ntoa(isaddr));
852 m_freem(m);
853 return;
854 }
c3749f74 855 if (isaddr.s_addr == myaddr.s_addr && myaddr.s_addr != 0) {
984263bc
MD
856 log(LOG_ERR,
857 "arp: %*D is using my IP address %s!\n",
858 ifp->if_addrlen, (u_char *)ar_sha(ah), ":",
859 inet_ntoa(isaddr));
860 itaddr = myaddr;
861 goto reply;
862 }
ecdefdda
MD
863#ifdef SMP
864 lwkt_initmsg(&msg.lmsg, &curthread->td_msgport, 0,
865 lwkt_cmd_func(arp_update_msghandler), lwkt_cmd_op_none);
866 msg.m = m;
867 msg.saddr = isaddr.s_addr;
868 msg.create = (itaddr.s_addr == myaddr.s_addr);
869 lwkt_domsg(rtable_portfn(0), &msg.lmsg);
870#endif
871 arp_update_oncpu(m, isaddr.s_addr, (itaddr.s_addr == myaddr.s_addr),
872 TRUE);
984263bc
MD
873reply:
874 if (op != ARPOP_REQUEST) {
875 m_freem(m);
876 return;
877 }
878 if (itaddr.s_addr == myaddr.s_addr) {
879 /* I am the target */
f23061d4
JH
880 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
881 memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln);
984263bc 882 } else {
ecdefdda
MD
883 struct llinfo_arp *la;
884
f23061d4 885 la = arplookup(itaddr.s_addr, FALSE, SIN_PROXY);
984263bc
MD
886 if (la == NULL) {
887 struct sockaddr_in sin;
888
889 if (!arp_proxyall) {
890 m_freem(m);
891 return;
892 }
893
894 bzero(&sin, sizeof sin);
895 sin.sin_family = AF_INET;
896 sin.sin_len = sizeof sin;
897 sin.sin_addr = itaddr;
898
f3ed2586 899 rt = rtpurelookup((struct sockaddr *)&sin);
0c3c561c 900 if (rt == NULL) {
984263bc
MD
901 m_freem(m);
902 return;
903 }
be23faf1 904 --rt->rt_refcnt;
984263bc
MD
905 /*
906 * Don't send proxies for nodes on the same interface
907 * as this one came out of, or we'll get into a fight
908 * over who claims what Ether address.
909 */
910 if (rt->rt_ifp == ifp) {
984263bc
MD
911 m_freem(m);
912 return;
913 }
f23061d4
JH
914 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
915 memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln);
984263bc 916#ifdef DEBUG_PROXY
0c3c561c 917 printf("arp: proxying for %s\n", inet_ntoa(itaddr));
984263bc
MD
918#endif
919 } else {
ecdefdda
MD
920 struct sockaddr_dl *sdl;
921
984263bc 922 rt = la->la_rt;
f23061d4 923 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln);
984263bc 924 sdl = SDL(rt->rt_gateway);
f23061d4 925 memcpy(ar_sha(ah), LLADDR(sdl), ah->ar_hln);
984263bc
MD
926 }
927 }
928
f23061d4
JH
929 memcpy(ar_tpa(ah), ar_spa(ah), ah->ar_pln);
930 memcpy(ar_spa(ah), &itaddr, ah->ar_pln);
984263bc
MD
931 ah->ar_op = htons(ARPOP_REPLY);
932 ah->ar_pro = htons(ETHERTYPE_IP); /* let's be sure! */
933 switch (ifp->if_type) {
934 case IFT_ARCNET:
935 arh = (struct arc_header *)sa.sa_data;
936 arh->arc_dhost = *ar_tha(ah);
937 arh->arc_type = ARCTYPE_ARP;
938 break;
984263bc
MD
939 case IFT_ISO88025:
940 /* Re-arrange the source/dest address */
941 memcpy(th->iso88025_dhost, th->iso88025_shost,
0c3c561c 942 sizeof th->iso88025_dhost);
984263bc 943 memcpy(th->iso88025_shost, IF_LLADDR(ifp),
0c3c561c 944 sizeof th->iso88025_shost);
984263bc
MD
945 /* Set the source routing bit if neccesary */
946 if (th->iso88025_dhost[0] & TR_RII) {
947 th->iso88025_dhost[0] &= ~TR_RII;
948 if (TR_RCF_RIFLEN(th->rcf) > 2)
949 th->iso88025_shost[0] |= TR_RII;
950 }
951 /* Copy the addresses, ac and fc into sa_data */
952 memcpy(sa.sa_data, th->iso88025_dhost,
0c3c561c
JH
953 (sizeof th->iso88025_dhost) * 2);
954 sa.sa_data[(sizeof th->iso88025_dhost) * 2] = TR_AC;
955 sa.sa_data[(sizeof th->iso88025_dhost) * 2 + 1] = TR_LLC_FRAME;
984263bc
MD
956 break;
957 case IFT_ETHER:
958 case IFT_FDDI:
959 /*
960 * May not be correct for types not explictly
961 * listed, but it is our best guess.
962 */
963 default:
964 eh = (struct ether_header *)sa.sa_data;
0c3c561c 965 memcpy(eh->ether_dhost, ar_tha(ah), sizeof eh->ether_dhost);
984263bc
MD
966 eh->ether_type = htons(ETHERTYPE_ARP);
967 break;
968 }
969 sa.sa_family = AF_UNSPEC;
0c3c561c 970 sa.sa_len = sizeof sa;
78195a76 971 lwkt_serialize_enter(ifp->if_serializer);
984263bc 972 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0);
78195a76 973 lwkt_serialize_exit(ifp->if_serializer);
984263bc
MD
974 return;
975}
ecdefdda
MD
976
977#ifdef SMP
978
979static
980int
981arp_update_msghandler(struct lwkt_msg *lmsg)
982{
983 struct netmsg_arp_update *msg = (struct netmsg_arp_update *)lmsg;
984 int nextcpu;
985
986 arp_update_oncpu(msg->m, msg->saddr, msg->create, FALSE);
987
988 nextcpu = mycpuid + 1;
989 if (nextcpu < ncpus) {
990 lwkt_forwardmsg(rtable_portfn(nextcpu), &msg->lmsg);
991 } else {
992 lwkt_replymsg(&msg->lmsg, 0);
993 }
994 return (0);
995}
996
997#endif
998
984263bc
MD
999#endif
1000
1001/*
03f869ef
MD
1002 * Free an arp entry. If the arp entry is actively referenced or represents
1003 * a static entry we only clear it back to an unresolved state, otherwise
1004 * we destroy the entry entirely.
1005 *
1006 * Note that static entries are created when route add ... -interface is used
1007 * to create an interface route to a (direct) destination.
984263bc
MD
1008 */
1009static void
5fe66e68 1010arptfree(struct llinfo_arp *la)
984263bc 1011{
2256ba69
RG
1012 struct rtentry *rt = la->la_rt;
1013 struct sockaddr_dl *sdl;
f23061d4 1014
03f869ef 1015 if (rt == NULL)
984263bc 1016 panic("arptfree");
03f869ef 1017 sdl = SDL(rt->rt_gateway);
5fe66e68
JH
1018 if (sdl != NULL &&
1019 ((rt->rt_refcnt > 0 && sdl->sdl_family == AF_LINK) ||
1020 (rt->rt_flags & RTF_STATIC))) {
984263bc
MD
1021 sdl->sdl_alen = 0;
1022 la->la_preempt = la->la_asked = 0;
1023 rt->rt_flags &= ~RTF_REJECT;
1024 return;
1025 }
f23061d4 1026 rtrequest(RTM_DELETE, rt_key(rt), NULL, rt_mask(rt), 0, NULL);
984263bc 1027}
0c3c561c 1028
984263bc
MD
1029/*
1030 * Lookup or enter a new address in arptab.
1031 */
1032static struct llinfo_arp *
f23061d4 1033arplookup(in_addr_t addr, boolean_t create, boolean_t proxy)
984263bc 1034{
2256ba69 1035 struct rtentry *rt;
f3ed2586 1036 struct sockaddr_inarp sin = { sizeof sin, AF_INET };
f23061d4 1037 const char *why = NULL;
984263bc
MD
1038
1039 sin.sin_addr.s_addr = addr;
1040 sin.sin_other = proxy ? SIN_PROXY : 0;
f3ed2586
JH
1041 if (create)
1042 rt = rtlookup((struct sockaddr *)&sin);
1043 else
1044 rt = rtpurelookup((struct sockaddr *)&sin);
f23061d4
JH
1045 if (rt == NULL)
1046 return (NULL);
984263bc
MD
1047 rt->rt_refcnt--;
1048
1049 if (rt->rt_flags & RTF_GATEWAY)
1050 why = "host is not on local network";
f23061d4 1051 else if (!(rt->rt_flags & RTF_LLINFO))
984263bc
MD
1052 why = "could not allocate llinfo";
1053 else if (rt->rt_gateway->sa_family != AF_LINK)
1054 why = "gateway route is not ours";
1055
8fd3e627
MD
1056 if (why) {
1057 if (create) {
1058 log(LOG_DEBUG, "arplookup %s failed: %s\n",
1059 inet_ntoa(sin.sin_addr), why);
1060 }
f23061d4
JH
1061 if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_WASCLONED)) {
1062 /* No references to this route. Purge it. */
1063 rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway,
be23faf1 1064 rt_mask(rt), rt->rt_flags, NULL);
8fd3e627 1065 }
f23061d4 1066 return (NULL);
984263bc 1067 }
f23061d4 1068 return (rt->rt_llinfo);
984263bc
MD
1069}
1070
1071void
f23061d4 1072arp_ifinit(struct ifnet *ifp, struct ifaddr *ifa)
984263bc 1073{
f23061d4
JH
1074 if (IA_SIN(ifa)->sin_addr.s_addr != INADDR_ANY)
1075 arprequest(ifp, &IA_SIN(ifa)->sin_addr, &IA_SIN(ifa)->sin_addr,
1076 IF_LLADDR(ifp));
984263bc
MD
1077 ifa->ifa_rtrequest = arp_rtrequest;
1078 ifa->ifa_flags |= RTF_CLONING;
1079}
1080
1081static void
1082arp_init(void)
1083{
ecdefdda
MD
1084 int cpu;
1085
1086 for (cpu = 0; cpu < ncpus2; cpu++)
1087 LIST_INIT(&llinfo_arp_list[cpu]);
bf82f9b7 1088 netisr_register(NETISR_ARP, cpu0_portfn, arpintr);
984263bc
MD
1089}
1090
1091SYSINIT(arp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, arp_init, 0);