if_ether.c revision 136960
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 136960 2004-10-26 03:31:58Z bms $ 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 43#include <sys/param.h> 44#include <sys/kernel.h> 45#include <sys/queue.h> 46#include <sys/sysctl.h> 47#include <sys/systm.h> 48#include <sys/mac.h> 49#include <sys/mbuf.h> 50#include <sys/malloc.h> 51#include <sys/socket.h> 52#include <sys/syslog.h> 53 54#include <net/if.h> 55#include <net/if_dl.h> 56#include <net/if_types.h> 57#include <net/route.h> 58#include <net/netisr.h> 59#include <net/if_llc.h> 60#include <net/ethernet.h> 61#include <net/bridge.h> 62 63#include <netinet/in.h> 64#include <netinet/in_var.h> 65#include <netinet/if_ether.h> 66 67#include <net/if_arc.h> 68#include <net/iso88025.h> 69 70#define SIN(s) ((struct sockaddr_in *)s) 71#define SDL(s) ((struct sockaddr_dl *)s) 72 73SYSCTL_DECL(_net_link_ether); 74SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, ""); 75 76/* timer values */ 77static int arpt_prune = (5*60*1); /* walk list every 5 minutes */ 78static int arpt_keep = (20*60); /* once resolved, good for 20 more minutes */ 79static int arpt_down = 20; /* once declared down, don't send for 20 sec */ 80 81SYSCTL_INT(_net_link_ether_inet, OID_AUTO, prune_intvl, CTLFLAG_RW, 82 &arpt_prune, 0, ""); 83SYSCTL_INT(_net_link_ether_inet, OID_AUTO, max_age, CTLFLAG_RW, 84 &arpt_keep, 0, ""); 85SYSCTL_INT(_net_link_ether_inet, OID_AUTO, host_down_time, CTLFLAG_RW, 86 &arpt_down, 0, ""); 87 88#define rt_expire rt_rmx.rmx_expire 89 90struct llinfo_arp { 91 LIST_ENTRY(llinfo_arp) la_le; 92 struct rtentry *la_rt; 93 struct mbuf *la_hold; /* last packet until resolved/timeout */ 94 u_short la_preempt; /* countdown for pre-expiry arps */ 95 u_short la_asked; /* #times we QUERIED following expiration */ 96#define la_timer la_rt->rt_rmx.rmx_expire /* deletion time in seconds */ 97}; 98 99static LIST_HEAD(, llinfo_arp) llinfo_arp; 100 101static struct ifqueue arpintrq; 102static int arp_allocated; 103 104static int arp_maxtries = 5; 105static int useloopback = 1; /* use loopback interface for local traffic */ 106static int arp_proxyall = 0; 107static struct callout arp_callout; 108 109SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW, 110 &arp_maxtries, 0, ""); 111SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW, 112 &useloopback, 0, ""); 113SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW, 114 &arp_proxyall, 0, ""); 115 116static void arp_init(void); 117static void arp_rtrequest(int, struct rtentry *, struct rt_addrinfo *); 118static void arprequest(struct ifnet *, 119 struct in_addr *, struct in_addr *, u_char *); 120static void arpintr(struct mbuf *); 121static void arptfree(struct llinfo_arp *); 122static void arptimer(void *); 123static struct llinfo_arp 124 *arplookup(u_long, int, int); 125#ifdef INET 126static void in_arpinput(struct mbuf *); 127#endif 128 129/* 130 * Timeout routine. Age arp_tab entries periodically. 131 */ 132/* ARGSUSED */ 133static void 134arptimer(ignored_arg) 135 void *ignored_arg; 136{ 137 struct llinfo_arp *la, *ola; 138 139 RADIX_NODE_HEAD_LOCK(rt_tables[AF_INET]); 140 la = LIST_FIRST(&llinfo_arp); 141 while (la != NULL) { 142 struct rtentry *rt = la->la_rt; 143 ola = la; 144 la = LIST_NEXT(la, la_le); 145 if (rt->rt_expire && rt->rt_expire <= time_second) 146 arptfree(ola); /* timer has expired, clear */ 147 } 148 RADIX_NODE_HEAD_UNLOCK(rt_tables[AF_INET]); 149 150 callout_reset(&arp_callout, arpt_prune * hz, arptimer, NULL); 151} 152 153/* 154 * Parallel to llc_rtrequest. 155 */ 156static void 157arp_rtrequest(req, rt, info) 158 int req; 159 struct rtentry *rt; 160 struct rt_addrinfo *info; 161{ 162 struct sockaddr *gate; 163 struct llinfo_arp *la; 164 static struct sockaddr_dl null_sdl = {sizeof(null_sdl), AF_LINK}; 165 166 RT_LOCK_ASSERT(rt); 167 168 if (rt->rt_flags & RTF_GATEWAY) 169 return; 170 gate = rt->rt_gateway; 171 la = (struct llinfo_arp *)rt->rt_llinfo; 172 switch (req) { 173 174 case RTM_ADD: 175 /* 176 * XXX: If this is a manually added route to interface 177 * such as older version of routed or gated might provide, 178 * restore cloning bit. 179 */ 180 if ((rt->rt_flags & RTF_HOST) == 0 && 181 rt_mask(rt) != NULL && 182 SIN(rt_mask(rt))->sin_addr.s_addr != 0xffffffff) 183 rt->rt_flags |= RTF_CLONING; 184 if (rt->rt_flags & RTF_CLONING) { 185 /* 186 * Case 1: This route should come from a route to iface. 187 */ 188 rt_setgate(rt, rt_key(rt), 189 (struct sockaddr *)&null_sdl); 190 gate = rt->rt_gateway; 191 SDL(gate)->sdl_type = rt->rt_ifp->if_type; 192 SDL(gate)->sdl_index = rt->rt_ifp->if_index; 193 rt->rt_expire = time_second; 194 break; 195 } 196 /* Announce a new entry if requested. */ 197 if (rt->rt_flags & RTF_ANNOUNCE) 198 arprequest(rt->rt_ifp, 199 &SIN(rt_key(rt))->sin_addr, 200 &SIN(rt_key(rt))->sin_addr, 201 (u_char *)LLADDR(SDL(gate))); 202 /*FALLTHROUGH*/ 203 case RTM_RESOLVE: 204 if (gate->sa_family != AF_LINK || 205 gate->sa_len < sizeof(null_sdl)) { 206 log(LOG_DEBUG, "%s: bad gateway %s%s\n", __func__, 207 inet_ntoa(SIN(rt_key(rt))->sin_addr), 208 (gate->sa_family != AF_LINK) ? 209 " (!AF_LINK)": ""); 210 break; 211 } 212 SDL(gate)->sdl_type = rt->rt_ifp->if_type; 213 SDL(gate)->sdl_index = rt->rt_ifp->if_index; 214 if (la != 0) 215 break; /* This happens on a route change */ 216 /* 217 * Case 2: This route may come from cloning, or a manual route 218 * add with a LL address. 219 */ 220 R_Zalloc(la, struct llinfo_arp *, sizeof(*la)); 221 rt->rt_llinfo = (caddr_t)la; 222 if (la == 0) { 223 log(LOG_DEBUG, "%s: malloc failed\n", __func__); 224 break; 225 } 226 arp_allocated++; 227 la->la_rt = rt; 228 rt->rt_flags |= RTF_LLINFO; 229 RADIX_NODE_HEAD_LOCK_ASSERT(rt_tables[AF_INET]); 230 LIST_INSERT_HEAD(&llinfo_arp, la, la_le); 231 232#ifdef INET 233 /* 234 * This keeps the multicast addresses from showing up 235 * in `arp -a' listings as unresolved. It's not actually 236 * functional. Then the same for broadcast. 237 */ 238 if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr)) && 239 rt->rt_ifp->if_type != IFT_ARCNET) { 240 ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt))->sin_addr, 241 LLADDR(SDL(gate))); 242 SDL(gate)->sdl_alen = 6; 243 rt->rt_expire = 0; 244 } 245 if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) { 246 memcpy(LLADDR(SDL(gate)), rt->rt_ifp->if_broadcastaddr, 247 rt->rt_ifp->if_addrlen); 248 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen; 249 rt->rt_expire = 0; 250 } 251#endif 252 253 if (SIN(rt_key(rt))->sin_addr.s_addr == 254 (IA_SIN(rt->rt_ifa))->sin_addr.s_addr) { 255 /* 256 * This test used to be 257 * if (loif.if_flags & IFF_UP) 258 * It allowed local traffic to be forced 259 * through the hardware by configuring the loopback down. 260 * However, it causes problems during network configuration 261 * for boards that can't receive packets they send. 262 * It is now necessary to clear "useloopback" and remove 263 * the route to force traffic out to the hardware. 264 */ 265 rt->rt_expire = 0; 266 bcopy(IF_LLADDR(rt->rt_ifp), LLADDR(SDL(gate)), 267 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen); 268 if (useloopback) 269 rt->rt_ifp = loif; 270 271 } 272 break; 273 274 case RTM_DELETE: 275 if (la == 0) 276 break; 277 RADIX_NODE_HEAD_LOCK_ASSERT(rt_tables[AF_INET]); 278 LIST_REMOVE(la, la_le); 279 rt->rt_llinfo = 0; 280 rt->rt_flags &= ~RTF_LLINFO; 281 if (la->la_hold) 282 m_freem(la->la_hold); 283 Free((caddr_t)la); 284 } 285} 286 287/* 288 * Broadcast an ARP request. Caller specifies: 289 * - arp header source ip address 290 * - arp header target ip address 291 * - arp header source ethernet address 292 */ 293static void 294arprequest(ifp, sip, tip, enaddr) 295 struct ifnet *ifp; 296 struct in_addr *sip, *tip; 297 u_char *enaddr; 298{ 299 struct mbuf *m; 300 struct arphdr *ah; 301 struct sockaddr sa; 302 303 if ((m = m_gethdr(M_DONTWAIT, MT_DATA)) == NULL) 304 return; 305 m->m_len = sizeof(*ah) + 2*sizeof(struct in_addr) + 306 2*ifp->if_data.ifi_addrlen; 307 m->m_pkthdr.len = m->m_len; 308 MH_ALIGN(m, m->m_len); 309 ah = mtod(m, struct arphdr *); 310 bzero((caddr_t)ah, m->m_len); 311#ifdef MAC 312 mac_create_mbuf_linklayer(ifp, m); 313#endif 314 ah->ar_pro = htons(ETHERTYPE_IP); 315 ah->ar_hln = ifp->if_addrlen; /* hardware address length */ 316 ah->ar_pln = sizeof(struct in_addr); /* protocol address length */ 317 ah->ar_op = htons(ARPOP_REQUEST); 318 bcopy((caddr_t)enaddr, (caddr_t)ar_sha(ah), ah->ar_hln); 319 bcopy((caddr_t)sip, (caddr_t)ar_spa(ah), ah->ar_pln); 320 bcopy((caddr_t)tip, (caddr_t)ar_tpa(ah), ah->ar_pln); 321 sa.sa_family = AF_ARP; 322 sa.sa_len = 2; 323 m->m_flags |= M_BCAST; 324 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0); 325 326 return; 327} 328 329/* 330 * Resolve an IP address into an ethernet address. 331 * On input: 332 * ifp is the interface we use 333 * dst is the next hop, 334 * rt0 is the route to the final destination (possibly useless) 335 * m is the mbuf 336 * desten is where we want the address. 337 * 338 * On success, desten is filled in and the function returns 0; 339 * If the packet must be held pending resolution, we return EWOULDBLOCK 340 * On other errors, we return the corresponding error code. 341 */ 342int 343arpresolve(struct ifnet *ifp, struct rtentry *rt0, struct mbuf *m, 344 struct sockaddr *dst, u_char *desten) 345{ 346 struct llinfo_arp *la = 0; 347 struct sockaddr_dl *sdl; 348 int error; 349 struct rtentry *rt; 350 351 error = rt_check(&rt, &rt0, dst); 352 if (error) { 353 m_freem(m); 354 return error; 355 } 356 357 if (m->m_flags & M_BCAST) { /* broadcast */ 358 (void)memcpy(desten, ifp->if_broadcastaddr, ifp->if_addrlen); 359 return (0); 360 } 361 if (m->m_flags & M_MCAST && ifp->if_type != IFT_ARCNET) {/* multicast */ 362 ETHER_MAP_IP_MULTICAST(&SIN(dst)->sin_addr, desten); 363 return (0); 364 } 365 if (rt) 366 la = (struct llinfo_arp *)rt->rt_llinfo; 367 if (la == 0) { 368 la = arplookup(SIN(dst)->sin_addr.s_addr, 1, 0); 369 if (la) 370 rt = la->la_rt; 371 } 372 if (la == 0 || rt == 0) { 373 log(LOG_DEBUG, "arpresolve: can't allocate llinfo for %s%s%s\n", 374 inet_ntoa(SIN(dst)->sin_addr), la ? "la" : "", 375 rt ? "rt" : ""); 376 m_freem(m); 377 return (EINVAL); /* XXX */ 378 } 379 sdl = SDL(rt->rt_gateway); 380 /* 381 * Check the address family and length is valid, the address 382 * is resolved; otherwise, try to resolve. 383 */ 384 if ((rt->rt_expire == 0 || rt->rt_expire > time_second) && 385 sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0) { 386 /* 387 * If entry has an expiry time and it is approaching, 388 * see if we need to send an ARP request within this 389 * arpt_down interval. 390 */ 391 if ((rt->rt_expire != 0) && 392 (time_second + la->la_preempt > rt->rt_expire)) { 393 arprequest(ifp, 394 &SIN(rt->rt_ifa->ifa_addr)->sin_addr, 395 &SIN(dst)->sin_addr, 396 IF_LLADDR(ifp)); 397 la->la_preempt--; 398 } 399 400 bcopy(LLADDR(sdl), desten, sdl->sdl_alen); 401 return (0); 402 } 403 /* 404 * If ARP is disabled or static on this interface, stop. 405 * XXX 406 * Probably should not allocate empty llinfo struct if we are 407 * not going to be sending out an arp request. 408 */ 409 if (ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) { 410 m_freem(m); 411 return (EINVAL); 412 } 413 /* 414 * There is an arptab entry, but no ethernet address 415 * response yet. Replace the held mbuf with this 416 * latest one. 417 */ 418 if (la->la_hold) 419 m_freem(la->la_hold); 420 la->la_hold = m; 421 if (rt->rt_expire) { 422 RT_LOCK(rt); 423 rt->rt_flags &= ~RTF_REJECT; 424 if (la->la_asked == 0 || rt->rt_expire != time_second) { 425 rt->rt_expire = time_second; 426 if (la->la_asked++ < arp_maxtries) { 427 arprequest(ifp, 428 &SIN(rt->rt_ifa->ifa_addr)->sin_addr, 429 &SIN(dst)->sin_addr, 430 IF_LLADDR(ifp)); 431 } else { 432 rt->rt_flags |= RTF_REJECT; 433 rt->rt_expire += arpt_down; 434 la->la_asked = 0; 435 la->la_preempt = arp_maxtries; 436 } 437 438 } 439 RT_UNLOCK(rt); 440 } 441 return (EWOULDBLOCK); 442} 443 444/* 445 * Common length and type checks are done here, 446 * then the protocol-specific routine is called. 447 */ 448static void 449arpintr(struct mbuf *m) 450{ 451 struct arphdr *ar; 452 453 if (m->m_len < sizeof(struct arphdr) && 454 ((m = m_pullup(m, sizeof(struct arphdr))) == NULL)) { 455 log(LOG_ERR, "arp: runt packet -- m_pullup failed\n"); 456 return; 457 } 458 ar = mtod(m, struct arphdr *); 459 460 if (ntohs(ar->ar_hrd) != ARPHRD_ETHER && 461 ntohs(ar->ar_hrd) != ARPHRD_IEEE802 && 462 ntohs(ar->ar_hrd) != ARPHRD_ARCNET && 463 ntohs(ar->ar_hrd) != ARPHRD_IEEE1394) { 464 log(LOG_ERR, "arp: unknown hardware address format (0x%2D)\n", 465 (unsigned char *)&ar->ar_hrd, ""); 466 m_freem(m); 467 return; 468 } 469 470 if (m->m_len < arphdr_len(ar)) { 471 if ((m = m_pullup(m, arphdr_len(ar))) == NULL) { 472 log(LOG_ERR, "arp: runt packet\n"); 473 m_freem(m); 474 return; 475 } 476 ar = mtod(m, struct arphdr *); 477 } 478 479 switch (ntohs(ar->ar_pro)) { 480#ifdef INET 481 case ETHERTYPE_IP: 482 in_arpinput(m); 483 return; 484#endif 485 } 486 m_freem(m); 487} 488 489#ifdef INET 490/* 491 * ARP for Internet protocols on 10 Mb/s Ethernet. 492 * Algorithm is that given in RFC 826. 493 * In addition, a sanity check is performed on the sender 494 * protocol address, to catch impersonators. 495 * We no longer handle negotiations for use of trailer protocol: 496 * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent 497 * along with IP replies if we wanted trailers sent to us, 498 * and also sent them in response to IP replies. 499 * This allowed either end to announce the desire to receive 500 * trailer packets. 501 * We no longer reply to requests for ETHERTYPE_TRAIL protocol either, 502 * but formerly didn't normally send requests. 503 */ 504static int log_arp_wrong_iface = 1; 505static int log_arp_movements = 1; 506 507SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_wrong_iface, CTLFLAG_RW, 508 &log_arp_wrong_iface, 0, 509 "log arp packets arriving on the wrong interface"); 510SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_movements, CTLFLAG_RW, 511 &log_arp_movements, 0, 512 "log arp replies from MACs different than the one in the cache"); 513 514 515static void 516in_arpinput(m) 517 struct mbuf *m; 518{ 519 struct arphdr *ah; 520 struct ifnet *ifp = m->m_pkthdr.rcvif; 521 struct iso88025_header *th = (struct iso88025_header *)0; 522 struct iso88025_sockaddr_dl_data *trld; 523 struct llinfo_arp *la = 0; 524 struct rtentry *rt; 525 struct ifaddr *ifa; 526 struct in_ifaddr *ia; 527 struct sockaddr_dl *sdl; 528 struct sockaddr sa; 529 struct in_addr isaddr, itaddr, myaddr; 530 int op, rif_len; 531 int req_len; 532 533 req_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr)); 534 if (m->m_len < req_len && (m = m_pullup(m, req_len)) == NULL) { 535 log(LOG_ERR, "in_arp: runt packet -- m_pullup failed\n"); 536 return; 537 } 538 539 ah = mtod(m, struct arphdr *); 540 op = ntohs(ah->ar_op); 541 (void)memcpy(&isaddr, ar_spa(ah), sizeof (isaddr)); 542 (void)memcpy(&itaddr, ar_tpa(ah), sizeof (itaddr)); 543 544 /* 545 * For a bridge, we want to check the address irrespective 546 * of the receive interface. (This will change slightly 547 * when we have clusters of interfaces). 548 */ 549 LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash) 550 if ((do_bridge || (ia->ia_ifp == ifp)) && 551 itaddr.s_addr == ia->ia_addr.sin_addr.s_addr) 552 goto match; 553 LIST_FOREACH(ia, INADDR_HASH(isaddr.s_addr), ia_hash) 554 if ((do_bridge || (ia->ia_ifp == ifp)) && 555 isaddr.s_addr == ia->ia_addr.sin_addr.s_addr) 556 goto match; 557 /* 558 * No match, use the first inet address on the receive interface 559 * as a dummy address for the rest of the function. 560 */ 561 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) 562 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET) { 563 ia = ifatoia(ifa); 564 goto match; 565 } 566 /* 567 * If bridging, fall back to using any inet address. 568 */ 569 if (!do_bridge || (ia = TAILQ_FIRST(&in_ifaddrhead)) == NULL) 570 goto drop; 571match: 572 myaddr = ia->ia_addr.sin_addr; 573 if (!bcmp(ar_sha(ah), IF_LLADDR(ifp), ifp->if_addrlen)) 574 goto drop; /* it's from me, ignore it. */ 575 if (!bcmp(ar_sha(ah), ifp->if_broadcastaddr, ifp->if_addrlen)) { 576 log(LOG_ERR, 577 "arp: link address is broadcast for IP address %s!\n", 578 inet_ntoa(isaddr)); 579 goto drop; 580 } 581 /* 582 * Warn if another host is using the same IP address, but only if the 583 * IP address isn't 0.0.0.0, which is used for DHCP only, in which 584 * case we suppress the warning to avoid false positive complaints of 585 * potential misconfiguration. 586 */ 587 if (isaddr.s_addr == myaddr.s_addr && myaddr.s_addr != 0) { 588 log(LOG_ERR, 589 "arp: %*D is using my IP address %s!\n", 590 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 591 inet_ntoa(isaddr)); 592 itaddr = myaddr; 593 goto reply; 594 } 595 if (ifp->if_flags & IFF_STATICARP) 596 goto reply; 597 la = arplookup(isaddr.s_addr, itaddr.s_addr == myaddr.s_addr, 0); 598 if (la && (rt = la->la_rt) && (sdl = SDL(rt->rt_gateway))) { 599 /* the following is not an error when doing bridging */ 600 if (!do_bridge && rt->rt_ifp != ifp) { 601 if (log_arp_wrong_iface) 602 log(LOG_ERR, "arp: %s is on %s but got reply from %*D on %s\n", 603 inet_ntoa(isaddr), 604 rt->rt_ifp->if_xname, 605 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 606 ifp->if_xname); 607 goto reply; 608 } 609 if (sdl->sdl_alen && 610 bcmp(ar_sha(ah), LLADDR(sdl), sdl->sdl_alen)) { 611 if (rt->rt_expire) { 612 if (log_arp_movements) 613 log(LOG_INFO, "arp: %s moved from %*D to %*D on %s\n", 614 inet_ntoa(isaddr), 615 ifp->if_addrlen, (u_char *)LLADDR(sdl), ":", 616 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 617 ifp->if_xname); 618 } else { 619 log(LOG_ERR, 620 "arp: %*D attempts to modify permanent entry for %s on %s\n", 621 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 622 inet_ntoa(isaddr), ifp->if_xname); 623 goto reply; 624 } 625 } 626 /* 627 * sanity check for the address length. 628 * XXX this does not work for protocols with variable address 629 * length. -is 630 */ 631 if (sdl->sdl_alen && 632 sdl->sdl_alen != ah->ar_hln) { 633 log(LOG_WARNING, 634 "arp from %*D: new addr len %d, was %d", 635 ifp->if_addrlen, (u_char *) ar_sha(ah), ":", 636 ah->ar_hln, sdl->sdl_alen); 637 } 638 if (ifp->if_addrlen != ah->ar_hln) { 639 log(LOG_WARNING, 640 "arp from %*D: addr len: new %d, i/f %d (ignored)", 641 ifp->if_addrlen, (u_char *) ar_sha(ah), ":", 642 ah->ar_hln, ifp->if_addrlen); 643 goto reply; 644 } 645 (void)memcpy(LLADDR(sdl), ar_sha(ah), 646 sdl->sdl_alen = ah->ar_hln); 647 /* 648 * If we receive an arp from a token-ring station over 649 * a token-ring nic then try to save the source 650 * routing info. 651 */ 652 if (ifp->if_type == IFT_ISO88025) { 653 th = (struct iso88025_header *)m->m_pkthdr.header; 654 trld = SDL_ISO88025(sdl); 655 rif_len = TR_RCF_RIFLEN(th->rcf); 656 if ((th->iso88025_shost[0] & TR_RII) && 657 (rif_len > 2)) { 658 trld->trld_rcf = th->rcf; 659 trld->trld_rcf ^= htons(TR_RCF_DIR); 660 memcpy(trld->trld_route, th->rd, rif_len - 2); 661 trld->trld_rcf &= ~htons(TR_RCF_BCST_MASK); 662 /* 663 * Set up source routing information for 664 * reply packet (XXX) 665 */ 666 m->m_data -= rif_len; 667 m->m_len += rif_len; 668 m->m_pkthdr.len += rif_len; 669 } else { 670 th->iso88025_shost[0] &= ~TR_RII; 671 trld->trld_rcf = 0; 672 } 673 m->m_data -= 8; 674 m->m_len += 8; 675 m->m_pkthdr.len += 8; 676 th->rcf = trld->trld_rcf; 677 } 678 RT_LOCK(rt); 679 if (rt->rt_expire) 680 rt->rt_expire = time_second + arpt_keep; 681 rt->rt_flags &= ~RTF_REJECT; 682 RT_UNLOCK(rt); 683 la->la_asked = 0; 684 la->la_preempt = arp_maxtries; 685 if (la->la_hold) { 686 (*ifp->if_output)(ifp, la->la_hold, rt_key(rt), rt); 687 la->la_hold = 0; 688 } 689 } 690reply: 691 if (op != ARPOP_REQUEST) 692 goto drop; 693 if (itaddr.s_addr == myaddr.s_addr) { 694 /* I am the target */ 695 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 696 (void)memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln); 697 } else { 698 la = arplookup(itaddr.s_addr, 0, SIN_PROXY); 699 if (la == NULL) { 700 struct sockaddr_in sin; 701 702 if (!arp_proxyall) 703 goto drop; 704 705 bzero(&sin, sizeof sin); 706 sin.sin_family = AF_INET; 707 sin.sin_len = sizeof sin; 708 sin.sin_addr = itaddr; 709 710 rt = rtalloc1((struct sockaddr *)&sin, 0, 0UL); 711 if (!rt) 712 goto drop; 713 /* 714 * Don't send proxies for nodes on the same interface 715 * as this one came out of, or we'll get into a fight 716 * over who claims what Ether address. 717 */ 718 if (rt->rt_ifp == ifp) { 719 rtfree(rt); 720 goto drop; 721 } 722 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 723 (void)memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln); 724 rtfree(rt); 725 726 /* 727 * Also check that the node which sent the ARP packet 728 * is on the the interface we expect it to be on. This 729 * avoids ARP chaos if an interface is connected to the 730 * wrong network. 731 */ 732 sin.sin_addr = isaddr; 733 734 rt = rtalloc1((struct sockaddr *)&sin, 0, 0UL); 735 if (!rt) 736 goto drop; 737 if (rt->rt_ifp != ifp) { 738 log(LOG_INFO, "arp_proxy: ignoring request" 739 " from %s via %s, expecting %s\n", 740 inet_ntoa(isaddr), ifp->if_xname, 741 rt->rt_ifp->if_xname); 742 rtfree(rt); 743 goto drop; 744 } 745 rtfree(rt); 746 747#ifdef DEBUG_PROXY 748 printf("arp: proxying for %s\n", 749 inet_ntoa(itaddr)); 750#endif 751 } else { 752 rt = la->la_rt; 753 (void)memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 754 sdl = SDL(rt->rt_gateway); 755 (void)memcpy(ar_sha(ah), LLADDR(sdl), ah->ar_hln); 756 } 757 } 758 759 (void)memcpy(ar_tpa(ah), ar_spa(ah), ah->ar_pln); 760 (void)memcpy(ar_spa(ah), &itaddr, ah->ar_pln); 761 ah->ar_op = htons(ARPOP_REPLY); 762 ah->ar_pro = htons(ETHERTYPE_IP); /* let's be sure! */ 763 m->m_flags &= ~(M_BCAST|M_MCAST); /* never reply by broadcast */ 764 m->m_len = sizeof(*ah) + (2 * ah->ar_pln) + (2 * ah->ar_hln); 765 m->m_pkthdr.len = m->m_len; 766 sa.sa_family = AF_ARP; 767 sa.sa_len = 2; 768 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0); 769 return; 770 771drop: 772 m_freem(m); 773} 774#endif 775 776/* 777 * Free an arp entry. 778 */ 779static void 780arptfree(la) 781 struct llinfo_arp *la; 782{ 783 struct rtentry *rt = la->la_rt; 784 struct sockaddr_dl *sdl; 785 786 if (rt == 0) 787 panic("arptfree"); 788 if (rt->rt_refcnt > 0 && (sdl = SDL(rt->rt_gateway)) && 789 sdl->sdl_family == AF_LINK) { 790 sdl->sdl_alen = 0; 791 la->la_preempt = la->la_asked = 0; 792 RT_LOCK(rt); /* XXX needed or move higher? */ 793 rt->rt_flags &= ~RTF_REJECT; 794 RT_UNLOCK(rt); 795 return; 796 } 797 rtrequest(RTM_DELETE, rt_key(rt), (struct sockaddr *)0, rt_mask(rt), 798 0, (struct rtentry **)0); 799} 800/* 801 * Lookup or enter a new address in arptab. 802 */ 803static struct llinfo_arp * 804arplookup(addr, create, proxy) 805 u_long addr; 806 int create, proxy; 807{ 808 struct rtentry *rt; 809 struct sockaddr_inarp sin; 810 const char *why = 0; 811 812 bzero(&sin, sizeof(sin)); 813 sin.sin_len = sizeof(sin); 814 sin.sin_family = AF_INET; 815 sin.sin_addr.s_addr = addr; 816 if (proxy) 817 sin.sin_other = SIN_PROXY; 818 rt = rtalloc1((struct sockaddr *)&sin, create, 0UL); 819 if (rt == 0) 820 return (0); 821 822 if (rt->rt_flags & RTF_GATEWAY) 823 why = "host is not on local network"; 824 else if ((rt->rt_flags & RTF_LLINFO) == 0) 825 why = "could not allocate llinfo"; 826 else if (rt->rt_gateway->sa_family != AF_LINK) 827 why = "gateway route is not ours"; 828 829 if (why) { 830#define ISDYNCLONE(_rt) \ 831 (((_rt)->rt_flags & (RTF_STATIC | RTF_WASCLONED)) == RTF_WASCLONED) 832 if (create) 833 log(LOG_DEBUG, "arplookup %s failed: %s\n", 834 inet_ntoa(sin.sin_addr), why); 835 /* 836 * If there are no references to this Layer 2 route, 837 * and it is a cloned route, and not static, and 838 * arplookup() is creating the route, then purge 839 * it from the routing table as it is probably bogus. 840 */ 841 if (rt->rt_refcnt == 1 && ISDYNCLONE(rt)) 842 rtexpunge(rt); 843 RTFREE_LOCKED(rt); 844 return (0); 845#undef ISDYNCLONE 846 } else { 847 RT_REMREF(rt); 848 RT_UNLOCK(rt); 849 return ((struct llinfo_arp *)rt->rt_llinfo); 850 } 851} 852 853void 854arp_ifinit(ifp, ifa) 855 struct ifnet *ifp; 856 struct ifaddr *ifa; 857{ 858 if (ntohl(IA_SIN(ifa)->sin_addr.s_addr) != INADDR_ANY) 859 arprequest(ifp, &IA_SIN(ifa)->sin_addr, 860 &IA_SIN(ifa)->sin_addr, IF_LLADDR(ifp)); 861 ifa->ifa_rtrequest = arp_rtrequest; 862 ifa->ifa_flags |= RTF_CLONING; 863} 864 865static void 866arp_init(void) 867{ 868 869 arpintrq.ifq_maxlen = 50; 870 mtx_init(&arpintrq.ifq_mtx, "arp_inq", NULL, MTX_DEF); 871 LIST_INIT(&llinfo_arp); 872 callout_init(&arp_callout, CALLOUT_MPSAFE); 873 netisr_register(NETISR_ARP, arpintr, &arpintrq, NETISR_MPSAFE); 874 callout_reset(&arp_callout, hz, arptimer, NULL); 875} 876SYSINIT(arp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, arp_init, 0); 877