if_ether.c revision 150351
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 * 4. Neither the name of the University nor the names of its contributors 14 * may be used to endorse or promote products derived from this software 15 * without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * @(#)if_ether.c 8.1 (Berkeley) 6/10/93 30 * $FreeBSD: head/sys/netinet/if_ether.c 150351 2005-09-19 22:54:55Z andre $ 31 */ 32 33/* 34 * Ethernet address resolution protocol. 35 * TODO: 36 * add "inuse/lock" bit (or ref. count) along with valid bit 37 */ 38 39#include "opt_inet.h" 40#include "opt_bdg.h" 41#include "opt_mac.h" 42#include "opt_carp.h" 43 44#include <sys/param.h> 45#include <sys/kernel.h> 46#include <sys/queue.h> 47#include <sys/sysctl.h> 48#include <sys/systm.h> 49#include <sys/mac.h> 50#include <sys/mbuf.h> 51#include <sys/malloc.h> 52#include <sys/socket.h> 53#include <sys/syslog.h> 54 55#include <net/if.h> 56#include <net/if_dl.h> 57#include <net/if_types.h> 58#include <net/route.h> 59#include <net/netisr.h> 60#include <net/if_llc.h> 61#include <net/ethernet.h> 62#include <net/bridge.h> 63 64#include <netinet/in.h> 65#include <netinet/in_var.h> 66#include <netinet/if_ether.h> 67 68#include <net/if_arc.h> 69#include <net/iso88025.h> 70 71#ifdef DEV_CARP 72#include <netinet/ip_carp.h> 73#endif 74 75#define SIN(s) ((struct sockaddr_in *)s) 76#define SDL(s) ((struct sockaddr_dl *)s) 77 78SYSCTL_DECL(_net_link_ether); 79SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, ""); 80 81/* timer values */ 82static int arpt_prune = (5*60*1); /* walk list every 5 minutes */ 83static int arpt_keep = (20*60); /* once resolved, good for 20 more minutes */ 84static int arpt_down = 20; /* once declared down, don't send for 20 sec */ 85 86SYSCTL_INT(_net_link_ether_inet, OID_AUTO, prune_intvl, CTLFLAG_RW, 87 &arpt_prune, 0, ""); 88SYSCTL_INT(_net_link_ether_inet, OID_AUTO, max_age, CTLFLAG_RW, 89 &arpt_keep, 0, ""); 90SYSCTL_INT(_net_link_ether_inet, OID_AUTO, host_down_time, CTLFLAG_RW, 91 &arpt_down, 0, ""); 92 93#define rt_expire rt_rmx.rmx_expire 94 95struct llinfo_arp { 96 LIST_ENTRY(llinfo_arp) la_le; 97 struct rtentry *la_rt; 98 struct mbuf *la_hold; /* last packet until resolved/timeout */ 99 u_short la_preempt; /* countdown for pre-expiry arps */ 100 u_short la_asked; /* #times we QUERIED following expiration */ 101#define la_timer la_rt->rt_rmx.rmx_expire /* deletion time in seconds */ 102}; 103 104static LIST_HEAD(, llinfo_arp) llinfo_arp; 105 106static struct ifqueue arpintrq; 107static int arp_allocated; 108 109static int arp_maxtries = 5; 110static int useloopback = 1; /* use loopback interface for local traffic */ 111static int arp_proxyall = 0; 112static struct callout arp_callout; 113 114SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW, 115 &arp_maxtries, 0, ""); 116SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW, 117 &useloopback, 0, ""); 118SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW, 119 &arp_proxyall, 0, ""); 120 121static void arp_init(void); 122static void arp_rtrequest(int, struct rtentry *, struct rt_addrinfo *); 123static void arprequest(struct ifnet *, 124 struct in_addr *, struct in_addr *, u_char *); 125static void arpintr(struct mbuf *); 126static void arptimer(void *); 127static struct rtentry 128 *arplookup(u_long, int, int); 129#ifdef INET 130static void in_arpinput(struct mbuf *); 131#endif 132 133/* 134 * Timeout routine. Age arp_tab entries periodically. 135 */ 136/* ARGSUSED */ 137static void 138arptimer(void * __unused unused) 139{ 140 struct llinfo_arp *la, *ola; 141 142 RADIX_NODE_HEAD_LOCK(rt_tables[AF_INET]); 143 LIST_FOREACH_SAFE(la, &llinfo_arp, la_le, ola) { 144 struct rtentry *rt = la->la_rt; 145 146 RT_LOCK(rt); 147 if (rt->rt_expire && rt->rt_expire <= time_uptime) { 148 struct sockaddr_dl *sdl = SDL(rt->rt_gateway); 149 150 KASSERT(sdl->sdl_family == AF_LINK, ("sdl_family %d", 151 sdl->sdl_family)); 152 if (rt->rt_refcnt > 1) { 153 sdl->sdl_alen = 0; 154 la->la_preempt = la->la_asked = 0; 155 rt->rt_flags &= ~RTF_REJECT; 156 RT_UNLOCK(rt); 157 continue; 158 } 159 RT_UNLOCK(rt); 160 /* 161 * XXX: LIST_REMOVE() is deep inside rtrequest(). 162 */ 163 rtrequest(RTM_DELETE, rt_key(rt), NULL, rt_mask(rt), 0, 164 NULL); 165 continue; 166 } 167 RT_UNLOCK(rt); 168 } 169 RADIX_NODE_HEAD_UNLOCK(rt_tables[AF_INET]); 170 171 callout_reset(&arp_callout, arpt_prune * hz, arptimer, NULL); 172} 173 174/* 175 * Parallel to llc_rtrequest. 176 */ 177static void 178arp_rtrequest(req, rt, info) 179 int req; 180 struct rtentry *rt; 181 struct rt_addrinfo *info; 182{ 183 struct sockaddr *gate; 184 struct llinfo_arp *la; 185 static struct sockaddr_dl null_sdl = {sizeof(null_sdl), AF_LINK}; 186 struct in_ifaddr *ia; 187 struct ifaddr *ifa; 188 189 RT_LOCK_ASSERT(rt); 190 191 if (rt->rt_flags & RTF_GATEWAY) 192 return; 193 gate = rt->rt_gateway; 194 la = (struct llinfo_arp *)rt->rt_llinfo; 195 switch (req) { 196 197 case RTM_ADD: 198 /* 199 * XXX: If this is a manually added route to interface 200 * such as older version of routed or gated might provide, 201 * restore cloning bit. 202 */ 203 if ((rt->rt_flags & RTF_HOST) == 0 && 204 rt_mask(rt) != NULL && 205 SIN(rt_mask(rt))->sin_addr.s_addr != 0xffffffff) 206 rt->rt_flags |= RTF_CLONING; 207 if (rt->rt_flags & RTF_CLONING) { 208 /* 209 * Case 1: This route should come from a route to iface. 210 */ 211 rt_setgate(rt, rt_key(rt), 212 (struct sockaddr *)&null_sdl); 213 gate = rt->rt_gateway; 214 SDL(gate)->sdl_type = rt->rt_ifp->if_type; 215 SDL(gate)->sdl_index = rt->rt_ifp->if_index; 216 rt->rt_expire = time_uptime; 217 break; 218 } 219 /* Announce a new entry if requested. */ 220 if (rt->rt_flags & RTF_ANNOUNCE) 221 arprequest(rt->rt_ifp, 222 &SIN(rt_key(rt))->sin_addr, 223 &SIN(rt_key(rt))->sin_addr, 224 (u_char *)LLADDR(SDL(gate))); 225 /*FALLTHROUGH*/ 226 case RTM_RESOLVE: 227 if (gate->sa_family != AF_LINK || 228 gate->sa_len < sizeof(null_sdl)) { 229 log(LOG_DEBUG, "%s: bad gateway %s%s\n", __func__, 230 inet_ntoa(SIN(rt_key(rt))->sin_addr), 231 (gate->sa_family != AF_LINK) ? 232 " (!AF_LINK)": ""); 233 break; 234 } 235 SDL(gate)->sdl_type = rt->rt_ifp->if_type; 236 SDL(gate)->sdl_index = rt->rt_ifp->if_index; 237 if (la != 0) 238 break; /* This happens on a route change */ 239 /* 240 * Case 2: This route may come from cloning, or a manual route 241 * add with a LL address. 242 */ 243 R_Zalloc(la, struct llinfo_arp *, sizeof(*la)); 244 rt->rt_llinfo = (caddr_t)la; 245 if (la == 0) { 246 log(LOG_DEBUG, "%s: malloc failed\n", __func__); 247 break; 248 } 249 arp_allocated++; 250 /* 251 * We are storing a route entry outside of radix tree. So, 252 * it can be found and accessed by other means than radix 253 * lookup. The routing code assumes that any rtentry detached 254 * from radix can be destroyed safely. To prevent this, we 255 * add an additional reference. 256 */ 257 RT_ADDREF(rt); 258 la->la_rt = rt; 259 rt->rt_flags |= RTF_LLINFO; 260 RADIX_NODE_HEAD_LOCK_ASSERT(rt_tables[AF_INET]); 261 LIST_INSERT_HEAD(&llinfo_arp, la, la_le); 262 263#ifdef INET 264 /* 265 * This keeps the multicast addresses from showing up 266 * in `arp -a' listings as unresolved. It's not actually 267 * functional. Then the same for broadcast. 268 */ 269 if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr)) && 270 rt->rt_ifp->if_type != IFT_ARCNET) { 271 ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt))->sin_addr, 272 LLADDR(SDL(gate))); 273 SDL(gate)->sdl_alen = 6; 274 rt->rt_expire = 0; 275 } 276 if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) { 277 memcpy(LLADDR(SDL(gate)), rt->rt_ifp->if_broadcastaddr, 278 rt->rt_ifp->if_addrlen); 279 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen; 280 rt->rt_expire = 0; 281 } 282#endif 283 284 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) { 285 if (ia->ia_ifp == rt->rt_ifp && 286 SIN(rt_key(rt))->sin_addr.s_addr == 287 (IA_SIN(ia))->sin_addr.s_addr) 288 break; 289 } 290 if (ia) { 291 /* 292 * This test used to be 293 * if (loif.if_flags & IFF_UP) 294 * It allowed local traffic to be forced 295 * through the hardware by configuring the loopback down. 296 * However, it causes problems during network configuration 297 * for boards that can't receive packets they send. 298 * It is now necessary to clear "useloopback" and remove 299 * the route to force traffic out to the hardware. 300 */ 301 rt->rt_expire = 0; 302 bcopy(IF_LLADDR(rt->rt_ifp), LLADDR(SDL(gate)), 303 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen); 304 if (useloopback) 305 rt->rt_ifp = loif; 306 307 /* 308 * make sure to set rt->rt_ifa to the interface 309 * address we are using, otherwise we will have trouble 310 * with source address selection. 311 */ 312 ifa = &ia->ia_ifa; 313 if (ifa != rt->rt_ifa) { 314 IFAFREE(rt->rt_ifa); 315 IFAREF(ifa); 316 rt->rt_ifa = ifa; 317 } 318 } 319 break; 320 321 case RTM_DELETE: 322 if (la == 0) 323 break; 324 RADIX_NODE_HEAD_LOCK_ASSERT(rt_tables[AF_INET]); 325 LIST_REMOVE(la, la_le); 326 RT_REMREF(rt); 327 rt->rt_llinfo = 0; 328 rt->rt_flags &= ~RTF_LLINFO; 329 if (la->la_hold) 330 m_freem(la->la_hold); 331 Free((caddr_t)la); 332 } 333} 334 335/* 336 * Broadcast an ARP request. Caller specifies: 337 * - arp header source ip address 338 * - arp header target ip address 339 * - arp header source ethernet address 340 */ 341static void 342arprequest(ifp, sip, tip, enaddr) 343 struct ifnet *ifp; 344 struct in_addr *sip, *tip; 345 u_char *enaddr; 346{ 347 struct mbuf *m; 348 struct arphdr *ah; 349 struct sockaddr sa; 350 351 if ((m = m_gethdr(M_DONTWAIT, MT_DATA)) == NULL) 352 return; 353 m->m_len = sizeof(*ah) + 2*sizeof(struct in_addr) + 354 2*ifp->if_data.ifi_addrlen; 355 m->m_pkthdr.len = m->m_len; 356 MH_ALIGN(m, m->m_len); 357 ah = mtod(m, struct arphdr *); 358 bzero((caddr_t)ah, m->m_len); 359#ifdef MAC 360 mac_create_mbuf_linklayer(ifp, m); 361#endif 362 ah->ar_pro = htons(ETHERTYPE_IP); 363 ah->ar_hln = ifp->if_addrlen; /* hardware address length */ 364 ah->ar_pln = sizeof(struct in_addr); /* protocol address length */ 365 ah->ar_op = htons(ARPOP_REQUEST); 366 bcopy((caddr_t)enaddr, (caddr_t)ar_sha(ah), ah->ar_hln); 367 bcopy((caddr_t)sip, (caddr_t)ar_spa(ah), ah->ar_pln); 368 bcopy((caddr_t)tip, (caddr_t)ar_tpa(ah), ah->ar_pln); 369 sa.sa_family = AF_ARP; 370 sa.sa_len = 2; 371 m->m_flags |= M_BCAST; 372 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0); 373 374 return; 375} 376 377/* 378 * Resolve an IP address into an ethernet address. 379 * On input: 380 * ifp is the interface we use 381 * dst is the next hop, 382 * rt0 is the route to the final destination (possibly useless) 383 * m is the mbuf 384 * desten is where we want the address. 385 * 386 * On success, desten is filled in and the function returns 0; 387 * If the packet must be held pending resolution, we return EWOULDBLOCK 388 * On other errors, we return the corresponding error code. 389 */ 390int 391arpresolve(struct ifnet *ifp, struct rtentry *rt0, struct mbuf *m, 392 struct sockaddr *dst, u_char *desten) 393{ 394 struct llinfo_arp *la = NULL; 395 struct rtentry *rt = NULL; 396 struct sockaddr_dl *sdl; 397 int error; 398 399 if (m->m_flags & M_BCAST) { /* broadcast */ 400 (void)memcpy(desten, ifp->if_broadcastaddr, ifp->if_addrlen); 401 return (0); 402 } 403 if (m->m_flags & M_MCAST && ifp->if_type != IFT_ARCNET) {/* multicast */ 404 ETHER_MAP_IP_MULTICAST(&SIN(dst)->sin_addr, desten); 405 return (0); 406 } 407 408 if (rt0 != NULL) { 409 error = rt_check(&rt, &rt0, dst); 410 if (error) { 411 m_freem(m); 412 return error; 413 } 414 la = (struct llinfo_arp *)rt->rt_llinfo; 415 if (la == NULL) 416 RT_UNLOCK(rt); 417 } 418 if (la == NULL) { 419 /* 420 * We enter this block in case if rt0 was NULL, 421 * or if rt found by rt_check() didn't have llinfo. 422 */ 423 rt = arplookup(SIN(dst)->sin_addr.s_addr, 1, 0); 424 if (rt == NULL) { 425 log(LOG_DEBUG, 426 "arpresolve: can't allocate route for %s\n", 427 inet_ntoa(SIN(dst)->sin_addr)); 428 m_freem(m); 429 return (EINVAL); /* XXX */ 430 } 431 la = (struct llinfo_arp *)rt->rt_llinfo; 432 if (la == NULL) { 433 RT_UNLOCK(rt); 434 log(LOG_DEBUG, 435 "arpresolve: can't allocate llinfo for %s\n", 436 inet_ntoa(SIN(dst)->sin_addr)); 437 m_freem(m); 438 return (EINVAL); /* XXX */ 439 } 440 } 441 sdl = SDL(rt->rt_gateway); 442 /* 443 * Check the address family and length is valid, the address 444 * is resolved; otherwise, try to resolve. 445 */ 446 if ((rt->rt_expire == 0 || rt->rt_expire > time_uptime) && 447 sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0) { 448 449 bcopy(LLADDR(sdl), desten, sdl->sdl_alen); 450 451 /* 452 * If entry has an expiry time and it is approaching, 453 * see if we need to send an ARP request within this 454 * arpt_down interval. 455 */ 456 if ((rt->rt_expire != 0) && 457 (time_uptime + la->la_preempt > rt->rt_expire)) { 458 struct in_addr sin = 459 SIN(rt->rt_ifa->ifa_addr)->sin_addr; 460 461 la->la_preempt--; 462 RT_UNLOCK(rt); 463 arprequest(ifp, &sin, &SIN(dst)->sin_addr, 464 IF_LLADDR(ifp)); 465 return (0); 466 } 467 468 RT_UNLOCK(rt); 469 return (0); 470 } 471 /* 472 * If ARP is disabled or static on this interface, stop. 473 * XXX 474 * Probably should not allocate empty llinfo struct if we are 475 * not going to be sending out an arp request. 476 */ 477 if (ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) { 478 RT_UNLOCK(rt); 479 m_freem(m); 480 return (EINVAL); 481 } 482 /* 483 * There is an arptab entry, but no ethernet address 484 * response yet. Replace the held mbuf with this 485 * latest one. 486 */ 487 if (la->la_hold) 488 m_freem(la->la_hold); 489 la->la_hold = m; 490 if (rt->rt_expire) { 491 rt->rt_flags &= ~RTF_REJECT; 492 if (la->la_asked == 0 || rt->rt_expire != time_uptime) { 493 rt->rt_expire = time_uptime; 494 if (la->la_asked++ < arp_maxtries) { 495 struct in_addr sin = 496 SIN(rt->rt_ifa->ifa_addr)->sin_addr; 497 498 RT_UNLOCK(rt); 499 arprequest(ifp, &sin, &SIN(dst)->sin_addr, 500 IF_LLADDR(ifp)); 501 return (EWOULDBLOCK); 502 } else { 503 rt->rt_flags |= RTF_REJECT; 504 rt->rt_expire += arpt_down; 505 la->la_asked = 0; 506 la->la_preempt = arp_maxtries; 507 } 508 509 } 510 } 511 RT_UNLOCK(rt); 512 return (EWOULDBLOCK); 513} 514 515/* 516 * Common length and type checks are done here, 517 * then the protocol-specific routine is called. 518 */ 519static void 520arpintr(struct mbuf *m) 521{ 522 struct arphdr *ar; 523 524 if (m->m_len < sizeof(struct arphdr) && 525 ((m = m_pullup(m, sizeof(struct arphdr))) == NULL)) { 526 log(LOG_ERR, "arp: runt packet -- m_pullup failed\n"); 527 return; 528 } 529 ar = mtod(m, struct arphdr *); 530 531 if (ntohs(ar->ar_hrd) != ARPHRD_ETHER && 532 ntohs(ar->ar_hrd) != ARPHRD_IEEE802 && 533 ntohs(ar->ar_hrd) != ARPHRD_ARCNET && 534 ntohs(ar->ar_hrd) != ARPHRD_IEEE1394) { 535 log(LOG_ERR, "arp: unknown hardware address format (0x%2D)\n", 536 (unsigned char *)&ar->ar_hrd, ""); 537 m_freem(m); 538 return; 539 } 540 541 if (m->m_len < arphdr_len(ar)) { 542 if ((m = m_pullup(m, arphdr_len(ar))) == NULL) { 543 log(LOG_ERR, "arp: runt packet\n"); 544 m_freem(m); 545 return; 546 } 547 ar = mtod(m, struct arphdr *); 548 } 549 550 switch (ntohs(ar->ar_pro)) { 551#ifdef INET 552 case ETHERTYPE_IP: 553 in_arpinput(m); 554 return; 555#endif 556 } 557 m_freem(m); 558} 559 560#ifdef INET 561/* 562 * ARP for Internet protocols on 10 Mb/s Ethernet. 563 * Algorithm is that given in RFC 826. 564 * In addition, a sanity check is performed on the sender 565 * protocol address, to catch impersonators. 566 * We no longer handle negotiations for use of trailer protocol: 567 * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent 568 * along with IP replies if we wanted trailers sent to us, 569 * and also sent them in response to IP replies. 570 * This allowed either end to announce the desire to receive 571 * trailer packets. 572 * We no longer reply to requests for ETHERTYPE_TRAIL protocol either, 573 * but formerly didn't normally send requests. 574 */ 575static int log_arp_wrong_iface = 1; 576static int log_arp_movements = 1; 577 578SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_wrong_iface, CTLFLAG_RW, 579 &log_arp_wrong_iface, 0, 580 "log arp packets arriving on the wrong interface"); 581SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_movements, CTLFLAG_RW, 582 &log_arp_movements, 0, 583 "log arp replies from MACs different than the one in the cache"); 584 585 586static void 587in_arpinput(m) 588 struct mbuf *m; 589{ 590 struct arphdr *ah; 591 struct ifnet *ifp = m->m_pkthdr.rcvif; 592 struct iso88025_header *th = (struct iso88025_header *)0; 593 struct iso88025_sockaddr_dl_data *trld; 594 struct llinfo_arp *la; 595 struct rtentry *rt; 596 struct ifaddr *ifa; 597 struct in_ifaddr *ia; 598 struct sockaddr_dl *sdl; 599 struct sockaddr sa; 600 struct in_addr isaddr, itaddr, myaddr; 601 struct mbuf *hold; 602 u_int8_t *enaddr = NULL; 603 int op, rif_len; 604 int req_len; 605 int bridged = 0; 606#ifdef DEV_CARP 607 int carp_match = 0; 608#endif 609 610 if (do_bridge || ifp->if_bridge) 611 bridged = 1; 612 613 req_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr)); 614 if (m->m_len < req_len && (m = m_pullup(m, req_len)) == NULL) { 615 log(LOG_ERR, "in_arp: runt packet -- m_pullup failed\n"); 616 return; 617 } 618 619 ah = mtod(m, struct arphdr *); 620 op = ntohs(ah->ar_op); 621 (void)memcpy(&isaddr, ar_spa(ah), sizeof (isaddr)); 622 (void)memcpy(&itaddr, ar_tpa(ah), sizeof (itaddr)); 623 624 /* 625 * For a bridge, we want to check the address irrespective 626 * of the receive interface. (This will change slightly 627 * when we have clusters of interfaces). 628 * If the interface does not match, but the recieving interface 629 * is part of carp, we call carp_iamatch to see if this is a 630 * request for the virtual host ip. 631 * XXX: This is really ugly! 632 */ 633 LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash) { 634 if ((bridged || (ia->ia_ifp == ifp)) && 635 itaddr.s_addr == ia->ia_addr.sin_addr.s_addr) 636 goto match; 637#ifdef DEV_CARP 638 if (ifp->if_carp != NULL && 639 carp_iamatch(ifp->if_carp, ia, &isaddr, &enaddr) && 640 itaddr.s_addr == ia->ia_addr.sin_addr.s_addr) { 641 carp_match = 1; 642 goto match; 643 } 644#endif 645 } 646 LIST_FOREACH(ia, INADDR_HASH(isaddr.s_addr), ia_hash) 647 if ((bridged || (ia->ia_ifp == ifp)) && 648 isaddr.s_addr == ia->ia_addr.sin_addr.s_addr) 649 goto match; 650 /* 651 * No match, use the first inet address on the receive interface 652 * as a dummy address for the rest of the function. 653 */ 654 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) 655 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET) { 656 ia = ifatoia(ifa); 657 goto match; 658 } 659 /* 660 * If bridging, fall back to using any inet address. 661 */ 662 if (!bridged || (ia = TAILQ_FIRST(&in_ifaddrhead)) == NULL) 663 goto drop; 664match: 665 if (!enaddr) 666 enaddr = (u_int8_t *)IF_LLADDR(ifp); 667 myaddr = ia->ia_addr.sin_addr; 668 if (!bcmp(ar_sha(ah), enaddr, ifp->if_addrlen)) 669 goto drop; /* it's from me, ignore it. */ 670 if (!bcmp(ar_sha(ah), ifp->if_broadcastaddr, ifp->if_addrlen)) { 671 log(LOG_ERR, 672 "arp: link address is broadcast for IP address %s!\n", 673 inet_ntoa(isaddr)); 674 goto drop; 675 } 676 /* 677 * Warn if another host is using the same IP address, but only if the 678 * IP address isn't 0.0.0.0, which is used for DHCP only, in which 679 * case we suppress the warning to avoid false positive complaints of 680 * potential misconfiguration. 681 */ 682 if (isaddr.s_addr == myaddr.s_addr && myaddr.s_addr != 0) { 683 log(LOG_ERR, 684 "arp: %*D is using my IP address %s!\n", 685 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 686 inet_ntoa(isaddr)); 687 itaddr = myaddr; 688 goto reply; 689 } 690 if (ifp->if_flags & IFF_STATICARP) 691 goto reply; 692 rt = arplookup(isaddr.s_addr, itaddr.s_addr == myaddr.s_addr, 0); 693 if (rt != NULL) { 694 la = (struct llinfo_arp *)rt->rt_llinfo; 695 if (la == NULL) { 696 RT_UNLOCK(rt); 697 goto reply; 698 } 699 } else 700 goto reply; 701 702 /* The following is not an error when doing bridging. */ 703 if (!bridged && rt->rt_ifp != ifp 704#ifdef DEV_CARP 705 && (ifp->if_type != IFT_CARP || !carp_match) 706#endif 707 ) { 708 if (log_arp_wrong_iface) 709 log(LOG_ERR, "arp: %s is on %s but got reply from %*D on %s\n", 710 inet_ntoa(isaddr), 711 rt->rt_ifp->if_xname, 712 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 713 ifp->if_xname); 714 RT_UNLOCK(rt); 715 goto reply; 716 } 717 sdl = SDL(rt->rt_gateway); 718 if (sdl->sdl_alen && 719 bcmp(ar_sha(ah), LLADDR(sdl), sdl->sdl_alen)) { 720 if (rt->rt_expire) { 721 if (log_arp_movements) 722 log(LOG_INFO, "arp: %s moved from %*D to %*D on %s\n", 723 inet_ntoa(isaddr), 724 ifp->if_addrlen, (u_char *)LLADDR(sdl), ":", 725 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 726 ifp->if_xname); 727 } else { 728 log(LOG_ERR, 729 "arp: %*D attempts to modify permanent entry for %s on %s\n", 730 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 731 inet_ntoa(isaddr), ifp->if_xname); 732 RT_UNLOCK(rt); 733 goto reply; 734 } 735 } 736 /* 737 * sanity check for the address length. 738 * XXX this does not work for protocols with variable address 739 * length. -is 740 */ 741 if (sdl->sdl_alen && 742 sdl->sdl_alen != ah->ar_hln) { 743 log(LOG_WARNING, 744 "arp from %*D: new addr len %d, was %d", 745 ifp->if_addrlen, (u_char *) ar_sha(ah), ":", 746 ah->ar_hln, sdl->sdl_alen); 747 } 748 if (ifp->if_addrlen != ah->ar_hln) { 749 log(LOG_WARNING, 750 "arp from %*D: addr len: new %d, i/f %d (ignored)", 751 ifp->if_addrlen, (u_char *) ar_sha(ah), ":", 752 ah->ar_hln, ifp->if_addrlen); 753 RT_UNLOCK(rt); 754 goto reply; 755 } 756 (void)memcpy(LLADDR(sdl), ar_sha(ah), 757 sdl->sdl_alen = ah->ar_hln); 758 /* 759 * If we receive an arp from a token-ring station over 760 * a token-ring nic then try to save the source 761 * routing info. 762 */ 763 if (ifp->if_type == IFT_ISO88025) { 764 th = (struct iso88025_header *)m->m_pkthdr.header; 765 trld = SDL_ISO88025(sdl); 766 rif_len = TR_RCF_RIFLEN(th->rcf); 767 if ((th->iso88025_shost[0] & TR_RII) && 768 (rif_len > 2)) { 769 trld->trld_rcf = th->rcf; 770 trld->trld_rcf ^= htons(TR_RCF_DIR); 771 memcpy(trld->trld_route, th->rd, rif_len - 2); 772 trld->trld_rcf &= ~htons(TR_RCF_BCST_MASK); 773 /* 774 * Set up source routing information for 775 * reply packet (XXX) 776 */ 777 m->m_data -= rif_len; 778 m->m_len += rif_len; 779 m->m_pkthdr.len += rif_len; 780 } else { 781 th->iso88025_shost[0] &= ~TR_RII; 782 trld->trld_rcf = 0; 783 } 784 m->m_data -= 8; 785 m->m_len += 8; 786 m->m_pkthdr.len += 8; 787 th->rcf = trld->trld_rcf; 788 } 789 if (rt->rt_expire) 790 rt->rt_expire = time_uptime + arpt_keep; 791 rt->rt_flags &= ~RTF_REJECT; 792 la->la_asked = 0; 793 la->la_preempt = arp_maxtries; 794 hold = la->la_hold; 795 la->la_hold = NULL; 796 RT_UNLOCK(rt); 797 if (hold != NULL) 798 (*ifp->if_output)(ifp, hold, rt_key(rt), rt); 799 800reply: 801 if (op != ARPOP_REQUEST) 802 goto drop; 803 if (itaddr.s_addr == myaddr.s_addr) { 804 /* I am the target */ 805 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 806 (void)memcpy(ar_sha(ah), enaddr, ah->ar_hln); 807 } else { 808 rt = arplookup(itaddr.s_addr, 0, SIN_PROXY); 809 if (rt == NULL) { 810 struct sockaddr_in sin; 811 812 if (!arp_proxyall) 813 goto drop; 814 815 bzero(&sin, sizeof sin); 816 sin.sin_family = AF_INET; 817 sin.sin_len = sizeof sin; 818 sin.sin_addr = itaddr; 819 820 rt = rtalloc1((struct sockaddr *)&sin, 0, 0UL); 821 if (!rt) 822 goto drop; 823 /* 824 * Don't send proxies for nodes on the same interface 825 * as this one came out of, or we'll get into a fight 826 * over who claims what Ether address. 827 */ 828 if (rt->rt_ifp == ifp) { 829 rtfree(rt); 830 goto drop; 831 } 832 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 833 (void)memcpy(ar_sha(ah), enaddr, ah->ar_hln); 834 rtfree(rt); 835 836 /* 837 * Also check that the node which sent the ARP packet 838 * is on the the interface we expect it to be on. This 839 * avoids ARP chaos if an interface is connected to the 840 * wrong network. 841 */ 842 sin.sin_addr = isaddr; 843 844 rt = rtalloc1((struct sockaddr *)&sin, 0, 0UL); 845 if (!rt) 846 goto drop; 847 if (rt->rt_ifp != ifp) { 848 log(LOG_INFO, "arp_proxy: ignoring request" 849 " from %s via %s, expecting %s\n", 850 inet_ntoa(isaddr), ifp->if_xname, 851 rt->rt_ifp->if_xname); 852 rtfree(rt); 853 goto drop; 854 } 855 rtfree(rt); 856 857#ifdef DEBUG_PROXY 858 printf("arp: proxying for %s\n", 859 inet_ntoa(itaddr)); 860#endif 861 } else { 862 /* 863 * Return proxied ARP replies only on the interface 864 * where this network resides. Otherwise we may 865 * conflict with the host we are proxying for. 866 */ 867 if (rt->rt_ifp != ifp) { 868 RT_UNLOCK(rt); 869 goto drop; 870 } 871 sdl = SDL(rt->rt_gateway); 872 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 873 (void)memcpy(ar_sha(ah), LLADDR(sdl), ah->ar_hln); 874 RT_UNLOCK(rt); 875 } 876 } 877 878 (void)memcpy(ar_tpa(ah), ar_spa(ah), ah->ar_pln); 879 (void)memcpy(ar_spa(ah), &itaddr, ah->ar_pln); 880 ah->ar_op = htons(ARPOP_REPLY); 881 ah->ar_pro = htons(ETHERTYPE_IP); /* let's be sure! */ 882 m->m_flags &= ~(M_BCAST|M_MCAST); /* never reply by broadcast */ 883 m->m_len = sizeof(*ah) + (2 * ah->ar_pln) + (2 * ah->ar_hln); 884 m->m_pkthdr.len = m->m_len; 885 sa.sa_family = AF_ARP; 886 sa.sa_len = 2; 887 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0); 888 return; 889 890drop: 891 m_freem(m); 892} 893#endif 894 895/* 896 * Lookup or enter a new address in arptab. 897 */ 898static struct rtentry * 899arplookup(addr, create, proxy) 900 u_long addr; 901 int create, proxy; 902{ 903 struct rtentry *rt; 904 struct sockaddr_inarp sin; 905 const char *why = 0; 906 907 bzero(&sin, sizeof(sin)); 908 sin.sin_len = sizeof(sin); 909 sin.sin_family = AF_INET; 910 sin.sin_addr.s_addr = addr; 911 if (proxy) 912 sin.sin_other = SIN_PROXY; 913 rt = rtalloc1((struct sockaddr *)&sin, create, 0UL); 914 if (rt == 0) 915 return (0); 916 917 if (rt->rt_flags & RTF_GATEWAY) 918 why = "host is not on local network"; 919 else if ((rt->rt_flags & RTF_LLINFO) == 0) 920 why = "could not allocate llinfo"; 921 else if (rt->rt_gateway->sa_family != AF_LINK) 922 why = "gateway route is not ours"; 923 924 if (why) { 925#define ISDYNCLONE(_rt) \ 926 (((_rt)->rt_flags & (RTF_STATIC | RTF_WASCLONED)) == RTF_WASCLONED) 927 if (create) 928 log(LOG_DEBUG, "arplookup %s failed: %s\n", 929 inet_ntoa(sin.sin_addr), why); 930 /* 931 * If there are no references to this Layer 2 route, 932 * and it is a cloned route, and not static, and 933 * arplookup() is creating the route, then purge 934 * it from the routing table as it is probably bogus. 935 */ 936 if (rt->rt_refcnt == 1 && ISDYNCLONE(rt)) 937 rtexpunge(rt); 938 RTFREE_LOCKED(rt); 939 return (0); 940#undef ISDYNCLONE 941 } else { 942 RT_REMREF(rt); 943 return (rt); 944 } 945} 946 947void 948arp_ifinit(ifp, ifa) 949 struct ifnet *ifp; 950 struct ifaddr *ifa; 951{ 952 if (ntohl(IA_SIN(ifa)->sin_addr.s_addr) != INADDR_ANY) 953 arprequest(ifp, &IA_SIN(ifa)->sin_addr, 954 &IA_SIN(ifa)->sin_addr, IF_LLADDR(ifp)); 955 ifa->ifa_rtrequest = arp_rtrequest; 956 ifa->ifa_flags |= RTF_CLONING; 957} 958 959void 960arp_ifinit2(ifp, ifa, enaddr) 961 struct ifnet *ifp; 962 struct ifaddr *ifa; 963 u_char *enaddr; 964{ 965 if (ntohl(IA_SIN(ifa)->sin_addr.s_addr) != INADDR_ANY) 966 arprequest(ifp, &IA_SIN(ifa)->sin_addr, 967 &IA_SIN(ifa)->sin_addr, enaddr); 968 ifa->ifa_rtrequest = arp_rtrequest; 969 ifa->ifa_flags |= RTF_CLONING; 970} 971 972static void 973arp_init(void) 974{ 975 976 arpintrq.ifq_maxlen = 50; 977 mtx_init(&arpintrq.ifq_mtx, "arp_inq", NULL, MTX_DEF); 978 LIST_INIT(&llinfo_arp); 979 callout_init(&arp_callout, CALLOUT_MPSAFE); 980 netisr_register(NETISR_ARP, arpintr, &arpintrq, NETISR_MPSAFE); 981 callout_reset(&arp_callout, hz, arptimer, NULL); 982} 983SYSINIT(arp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, arp_init, 0); 984