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) 1982, 1986, 1988, 1993 35 * The Regents of the University of California. All rights reserved. 36 * 37 * Redistribution and use in source and binary forms, with or without 38 * modification, are permitted provided that the following conditions 39 * are met: 40 * 1. Redistributions of source code must retain the above copyright 41 * notice, this list of conditions and the following disclaimer. 42 * 2. Redistributions in binary form must reproduce the above copyright 43 * notice, this list of conditions and the following disclaimer in the 44 * documentation and/or other materials provided with the distribution. 45 * 3. All advertising materials mentioning features or use of this software 46 * must display the following acknowledgement: 47 * This product includes software developed by the University of 48 * California, Berkeley and its contributors. 49 * 4. Neither the name of the University nor the names of its contributors 50 * may be used to endorse or promote products derived from this software 51 * without specific prior written permission. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 63 * SUCH DAMAGE. 64 * 65 * @(#)if_ether.c 8.1 (Berkeley) 6/10/93 66 * $FreeBSD: src/sys/netinet/if_ether.c,v 1.64.2.23 2003/04/11 07:23:15 fjoe Exp $ 67 * $DragonFly: src/sys/netinet/if_ether.c,v 1.43 2007/09/08 12:35:04 sephe Exp $ 68 */ 69 70 /* 71 * Ethernet address resolution protocol. 72 * TODO: 73 * add "inuse/lock" bit (or ref. count) along with valid bit 74 */ 75 76 #include "opt_inet.h" 77 #include "opt_carp.h" 78 79 #include <sys/param.h> 80 #include <sys/kernel.h> 81 #include <sys/queue.h> 82 #include <sys/sysctl.h> 83 #include <sys/systm.h> 84 #include <sys/mbuf.h> 85 #include <sys/malloc.h> 86 #include <sys/socket.h> 87 #include <sys/syslog.h> 88 89 #include <net/if.h> 90 #include <net/if_dl.h> 91 #include <net/if_types.h> 92 #include <net/route.h> 93 #include <net/netisr.h> 94 #include <net/if_llc.h> 95 96 #include <netinet/in.h> 97 #include <netinet/in_var.h> 98 #include <netinet/if_ether.h> 99 100 #include <sys/thread2.h> 101 #include <sys/msgport2.h> 102 #include <net/netmsg2.h> 103 104 105 #ifdef CARP 106 #include <netinet/ip_carp.h> 107 #endif 108 109 #define SIN(s) ((struct sockaddr_in *)s) 110 #define SDL(s) ((struct sockaddr_dl *)s) 111 112 SYSCTL_DECL(_net_link_ether); 113 SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, ""); 114 115 /* timer values */ 116 static int arpt_prune = (5*60*1); /* walk list every 5 minutes */ 117 static int arpt_keep = (20*60); /* once resolved, good for 20 more minutes */ 118 static int arpt_down = 20; /* once declared down, don't send for 20 sec */ 119 120 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, prune_intvl, CTLFLAG_RW, 121 &arpt_prune, 0, ""); 122 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, max_age, CTLFLAG_RW, 123 &arpt_keep, 0, ""); 124 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, host_down_time, CTLFLAG_RW, 125 &arpt_down, 0, ""); 126 127 #define rt_expire rt_rmx.rmx_expire 128 129 struct llinfo_arp { 130 LIST_ENTRY(llinfo_arp) la_le; 131 struct rtentry *la_rt; 132 struct mbuf *la_hold; /* last packet until resolved/timeout */ 133 u_short la_preempt; /* countdown for pre-expiry arps */ 134 u_short la_asked; /* #times we QUERIED following expiration */ 135 }; 136 137 static LIST_HEAD(, llinfo_arp) llinfo_arp_list[MAXCPU]; 138 139 static int arp_maxtries = 5; 140 static int useloopback = 1; /* use loopback interface for local traffic */ 141 static int arp_proxyall = 0; 142 143 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW, 144 &arp_maxtries, 0, ""); 145 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW, 146 &useloopback, 0, ""); 147 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW, 148 &arp_proxyall, 0, ""); 149 150 void arprequest_acces(struct ifnet *ifp, struct in_addr *sip, struct in_addr *tip, u_char *enaddr); 151 static void arp_rtrequest (int, struct rtentry *, struct rt_addrinfo *); 152 static void arprequest (struct ifnet *, 153 struct in_addr *, struct in_addr *, u_char *); 154 static void arpintr(struct netmsg *); 155 static void arptfree (struct llinfo_arp *); 156 static void arptimer (void *); 157 static struct llinfo_arp 158 *arplookup (in_addr_t addr, boolean_t create, boolean_t proxy); 159 #ifdef INET 160 static void in_arpinput (struct mbuf *); 161 #endif 162 163 static struct callout arptimer_ch[MAXCPU]; 164 165 /* 166 * Timeout routine. Age arp_tab entries periodically. 167 */ 168 /* ARGSUSED */ 169 static void 170 arptimer(void *ignored_arg) 171 { 172 struct llinfo_arp *la, *nla; 173 174 crit_enter(); 175 LIST_FOREACH_MUTABLE(la, &llinfo_arp_list[mycpuid], la_le, nla) { 176 if (la->la_rt->rt_expire && la->la_rt->rt_expire <= time_second) 177 arptfree(la); 178 } 179 callout_reset(&arptimer_ch[mycpuid], arpt_prune * hz, arptimer, NULL); 180 crit_exit(); 181 } 182 183 /* 184 * Parallel to llc_rtrequest. 185 */ 186 static void 187 arp_rtrequest(int req, struct rtentry *rt, struct rt_addrinfo *info) 188 { 189 struct sockaddr *gate = rt->rt_gateway; 190 struct llinfo_arp *la = rt->rt_llinfo; 191 192 struct sockaddr_dl null_sdl = { sizeof null_sdl, AF_LINK }; 193 static boolean_t arpinit_done[MAXCPU]; 194 195 if (!arpinit_done[mycpuid]) { 196 arpinit_done[mycpuid] = TRUE; 197 callout_init(&arptimer_ch[mycpuid]); 198 callout_reset(&arptimer_ch[mycpuid], hz, arptimer, NULL); 199 } 200 if (rt->rt_flags & RTF_GATEWAY) 201 return; 202 203 switch (req) { 204 case RTM_ADD: 205 /* 206 * XXX: If this is a manually added route to interface 207 * such as older version of routed or gated might provide, 208 * restore cloning bit. 209 */ 210 if (!(rt->rt_flags & RTF_HOST) && 211 SIN(rt_mask(rt))->sin_addr.s_addr != 0xffffffff) 212 rt->rt_flags |= RTF_CLONING; 213 if (rt->rt_flags & RTF_CLONING) { 214 /* 215 * Case 1: This route should come from a route to iface. 216 */ 217 rt_setgate(rt, rt_key(rt), 218 (struct sockaddr *)&null_sdl); 219 gate = rt->rt_gateway; 220 SDL(gate)->sdl_type = rt->rt_ifp->if_type; 221 SDL(gate)->sdl_index = rt->rt_ifp->if_index; 222 rt->rt_expire = time_second; 223 break; 224 } 225 /* Announce a new entry if requested. */ 226 if (rt->rt_flags & RTF_ANNOUNCE) 227 arprequest(rt->rt_ifp, 228 &SIN(rt_key(rt))->sin_addr, 229 &SIN(rt_key(rt))->sin_addr, 230 LLADDR(SDL(gate))); 231 /*FALLTHROUGH*/ 232 case RTM_RESOLVE: 233 if (gate->sa_family != AF_LINK || 234 gate->sa_len < sizeof(struct sockaddr_dl)) { 235 log(LOG_DEBUG, "arp_rtrequest: bad gateway value\n"); 236 break; 237 } 238 SDL(gate)->sdl_type = rt->rt_ifp->if_type; 239 SDL(gate)->sdl_index = rt->rt_ifp->if_index; 240 if (la != NULL) 241 break; /* This happens on a route change */ 242 /* 243 * Case 2: This route may come from cloning, or a manual route 244 * add with a LL address. 245 */ 246 R_Malloc(la, struct llinfo_arp *, sizeof *la); 247 rt->rt_llinfo = la; 248 if (la == NULL) { 249 log(LOG_DEBUG, "arp_rtrequest: malloc failed\n"); 250 break; 251 } 252 bzero(la, sizeof *la); 253 la->la_rt = rt; 254 rt->rt_flags |= RTF_LLINFO; 255 LIST_INSERT_HEAD(&llinfo_arp_list[mycpuid], la, la_le); 256 257 #ifdef INET 258 /* 259 * This keeps the multicast addresses from showing up 260 * in `arp -a' listings as unresolved. It's not actually 261 * functional. Then the same for broadcast. 262 */ 263 if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr))) { 264 ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt))->sin_addr, 265 LLADDR(SDL(gate))); 266 SDL(gate)->sdl_alen = 6; 267 rt->rt_expire = 0; 268 } 269 if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) { 270 memcpy(LLADDR(SDL(gate)), rt->rt_ifp->if_broadcastaddr, 271 rt->rt_ifp->if_addrlen); 272 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen; 273 rt->rt_expire = 0; 274 } 275 #endif 276 277 if (SIN(rt_key(rt))->sin_addr.s_addr == 278 (IA_SIN(rt->rt_ifa))->sin_addr.s_addr) { 279 /* 280 * This test used to be 281 * if (loif.if_flags & IFF_UP) 282 * It allowed local traffic to be forced 283 * through the hardware by configuring the 284 * loopback down. However, it causes problems 285 * during network configuration for boards 286 * that can't receive packets they send. It 287 * is now necessary to clear "useloopback" and 288 * remove the route to force traffic out to 289 * the hardware. 290 */ 291 rt->rt_expire = 0; 292 bcopy(IF_LLADDR(rt->rt_ifp), LLADDR(SDL(gate)), 293 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen); 294 if (useloopback) 295 rt->rt_ifp = loif; 296 } 297 break; 298 299 case RTM_DELETE: 300 if (la == NULL) 301 break; 302 LIST_REMOVE(la, la_le); 303 rt->rt_llinfo = NULL; 304 rt->rt_flags &= ~RTF_LLINFO; 305 if (la->la_hold != NULL) 306 m_freem(la->la_hold); 307 Free(la); 308 } 309 } 310 311 /* 312 * Broadcast an ARP request. Caller specifies: 313 * - arp header source ip address 314 * - arp header target ip address 315 * - arp header source ethernet address 316 */ 317 static void 318 arprequest(struct ifnet *ifp, struct in_addr *sip, struct in_addr *tip, 319 u_char *enaddr) 320 { 321 struct mbuf *m; 322 struct ether_header *eh; 323 struct arphdr *ah; 324 struct sockaddr sa; 325 u_short ar_hrd; 326 327 if ((m = m_gethdr(MB_DONTWAIT, MT_DATA)) == NULL) 328 return; 329 m->m_pkthdr.rcvif = (struct ifnet *)NULL; 330 331 switch (ifp->if_type) { 332 case IFT_ETHER: 333 /* 334 * This may not be correct for types not explicitly 335 * listed, but this is our best guess 336 */ 337 default: 338 ar_hrd = htons(ARPHRD_ETHER); 339 340 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr)); 341 m->m_pkthdr.len = m->m_len; 342 MH_ALIGN(m, m->m_len); 343 344 eh = (struct ether_header *)sa.sa_data; 345 /* if_output() will not swap */ 346 eh->ether_type = htons(ETHERTYPE_ARP); 347 memcpy(eh->ether_dhost, ifp->if_broadcastaddr, ifp->if_addrlen); 348 349 ah = mtod(m, struct arphdr *); 350 break; 351 } 352 353 ah->ar_hrd = ar_hrd; 354 ah->ar_pro = htons(ETHERTYPE_IP); 355 ah->ar_hln = ifp->if_addrlen; /* hardware address length */ 356 ah->ar_pln = sizeof(struct in_addr); /* protocol address length */ 357 ah->ar_op = htons(ARPOP_REQUEST); 358 memcpy(ar_sha(ah), enaddr, ah->ar_hln); 359 memset(ar_tha(ah), 0, ah->ar_hln); 360 memcpy(ar_spa(ah), sip, ah->ar_pln); 361 memcpy(ar_tpa(ah), tip, ah->ar_pln); 362 363 sa.sa_family = AF_UNSPEC; 364 sa.sa_len = sizeof sa; 365 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)NULL); 366 } 367 368 /* 369 * Resolve an IP address into an ethernet address. If success, 370 * desten is filled in. If there is no entry in arptab, 371 * set one up and broadcast a request for the IP address. 372 * Hold onto this mbuf and resend it once the address 373 * is finally resolved. A return value of 1 indicates 374 * that desten has been filled in and the packet should be sent 375 * normally; a 0 return indicates that the packet has been 376 * taken over here, either now or for later transmission. 377 */ 378 int 379 arpresolve( 380 struct ifnet *ifp, 381 struct rtentry *rt0, 382 struct mbuf *m, 383 struct sockaddr *dst, 384 u_char *desten) 385 { 386 struct rtentry *rt; 387 struct llinfo_arp *la = NULL; 388 struct sockaddr_dl *sdl; 389 390 if (m->m_flags & M_BCAST) { /* broadcast */ 391 memcpy(desten, ifp->if_broadcastaddr, ifp->if_addrlen); 392 return (1); 393 } 394 if (m->m_flags & M_MCAST) {/* multicast */ 395 ETHER_MAP_IP_MULTICAST(&SIN(dst)->sin_addr, desten); 396 return (1); 397 } 398 if (rt0 != NULL) { 399 if (rt_llroute(dst, rt0, &rt) != 0) { 400 m_freem(m); 401 return 0; 402 } 403 la = rt->rt_llinfo; 404 } 405 if (la == NULL) { 406 la = arplookup(SIN(dst)->sin_addr.s_addr, TRUE, FALSE); 407 if (la != NULL) 408 rt = la->la_rt; 409 } 410 if (la == NULL || rt == NULL) { 411 log(LOG_DEBUG, "arpresolve: can't allocate llinfo for %s%s%s\n", 412 inet_ntoa(SIN(dst)->sin_addr), la ? "la" : " ", 413 rt ? "rt" : ""); 414 m_freem(m); 415 return (0); 416 } 417 sdl = SDL(rt->rt_gateway); 418 /* 419 * Check the address family and length is valid, the address 420 * is resolved; otherwise, try to resolve. 421 */ 422 if ((rt->rt_expire == 0 || rt->rt_expire > time_second) && 423 sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0) { 424 /* 425 * If entry has an expiry time and it is approaching, 426 * see if we need to send an ARP request within this 427 * arpt_down interval. 428 */ 429 if ((rt->rt_expire != 0) && 430 (time_second + la->la_preempt > rt->rt_expire)) { 431 arprequest(ifp, 432 &SIN(rt->rt_ifa->ifa_addr)->sin_addr, 433 &SIN(dst)->sin_addr, 434 IF_LLADDR(ifp)); 435 la->la_preempt--; 436 } 437 438 bcopy(LLADDR(sdl), desten, sdl->sdl_alen); 439 return 1; 440 } 441 /* 442 * If ARP is disabled on this interface, stop. 443 * XXX 444 * Probably should not allocate empty llinfo struct if we are 445 * not going to be sending out an arp request. 446 */ 447 if (ifp->if_flags & IFF_NOARP) { 448 m_freem(m); 449 return (0); 450 } 451 /* 452 * There is an arptab entry, but no ethernet address 453 * response yet. Replace the held mbuf with this 454 * latest one. 455 */ 456 if (la->la_hold != NULL) 457 m_freem(la->la_hold); 458 la->la_hold = m; 459 if (rt->rt_expire || ((rt->rt_flags & RTF_STATIC) && !sdl->sdl_alen)) { 460 rt->rt_flags &= ~RTF_REJECT; 461 if (la->la_asked == 0 || rt->rt_expire != time_second) { 462 rt->rt_expire = time_second; 463 if (la->la_asked++ < arp_maxtries) { 464 arprequest(ifp, 465 &SIN(rt->rt_ifa->ifa_addr)->sin_addr, 466 &SIN(dst)->sin_addr, 467 IF_LLADDR(ifp)); 468 } else { 469 rt->rt_flags |= RTF_REJECT; 470 rt->rt_expire += arpt_down; 471 la->la_asked = 0; 472 la->la_preempt = arp_maxtries; 473 } 474 475 } 476 } 477 return (0); 478 } 479 480 /* 481 * Common length and type checks are done here, 482 * then the protocol-specific routine is called. 483 */ 484 static void 485 arpintr(struct netmsg *msg) 486 { 487 struct mbuf *m = ((struct netmsg_packet *)msg)->nm_packet; 488 struct arphdr *ar; 489 u_short ar_hrd; 490 491 if (m->m_len < sizeof(struct arphdr) && 492 ((m = m_pullup(m, sizeof(struct arphdr))) == NULL)) { 493 log(LOG_ERR, "arp: runt packet -- m_pullup failed\n"); 494 goto out2; 495 } 496 ar = mtod(m, struct arphdr *); 497 498 ar_hrd = ntohs(ar->ar_hrd); 499 if (ar_hrd != ARPHRD_ETHER && 500 ar_hrd != ARPHRD_IEEE802) { 501 log(LOG_ERR, 502 "arp: unknown hardware address format (0x%2D)\n", 503 (unsigned char *)&ar->ar_hrd, ""); 504 goto out1; 505 } 506 507 if (m->m_pkthdr.len < arphdr_len(ar) && 508 (m = m_pullup(m, arphdr_len(ar))) == NULL) { 509 log(LOG_ERR, "arp: runt packet\n"); 510 goto out1; 511 } 512 513 switch (ntohs(ar->ar_pro)) { 514 #ifdef INET 515 case ETHERTYPE_IP: 516 in_arpinput(m); 517 goto out2; 518 #endif 519 } 520 out1: 521 m_freem(m); 522 out2: 523 ; 524 /* msg was embedded in the mbuf, do not reply! */ 525 } 526 527 #ifdef INET 528 /* 529 * ARP for Internet protocols on 10 Mb/s Ethernet. 530 * Algorithm is that given in RFC 826. 531 * In addition, a sanity check is performed on the sender 532 * protocol address, to catch impersonators. 533 * We no longer handle negotiations for use of trailer protocol: 534 * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent 535 * along with IP replies if we wanted trailers sent to us, 536 * and also sent them in response to IP replies. 537 * This allowed either end to announce the desire to receive 538 * trailer packets. 539 * We no longer reply to requests for ETHERTYPE_TRAIL protocol either, 540 * but formerly didn't normally send requests. 541 */ 542 static int log_arp_wrong_iface = 1; 543 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_wrong_iface, CTLFLAG_RW, 544 &log_arp_wrong_iface, 0, 545 "log arp packets arriving on the wrong interface"); 546 547 static void 548 arp_update_oncpu(struct mbuf *m, in_addr_t saddr, boolean_t create, 549 boolean_t dologging) 550 { 551 struct arphdr *ah = mtod(m, struct arphdr *); 552 struct ifnet *ifp = m->m_pkthdr.rcvif; 553 struct llinfo_arp *la; 554 struct sockaddr_dl *sdl; 555 struct rtentry *rt; 556 int cpu = mycpuid; 557 558 la = arplookup(saddr, create, FALSE); 559 if (la && (rt = la->la_rt) && (sdl = SDL(rt->rt_gateway))) { 560 struct in_addr isaddr = { saddr }; 561 562 /* the following is not an error when doing bridging */ 563 if (rt->rt_ifp != ifp) { 564 if (dologging && log_arp_wrong_iface && cpu == 0) { 565 log(LOG_ERR, 566 "arp: %s is on %s " 567 "but got reply from %*D on %s\n", 568 inet_ntoa(isaddr), 569 rt->rt_ifp->if_xname, 570 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 571 ifp->if_xname); 572 } 573 return; 574 } 575 if (sdl->sdl_alen && 576 bcmp(ar_sha(ah), LLADDR(sdl), sdl->sdl_alen)) { 577 if (rt->rt_expire != 0) { 578 if (dologging && cpu == 0) { 579 log(LOG_INFO, 580 "arp: %s moved from %*D to %*D on %s\n", 581 inet_ntoa(isaddr), 582 ifp->if_addrlen, (u_char *)LLADDR(sdl), 583 ":", ifp->if_addrlen, 584 (u_char *)ar_sha(ah), ":", 585 ifp->if_xname); 586 } 587 } else { 588 if (dologging && cpu == 0) { 589 log(LOG_ERR, 590 "arp: %*D attempts to modify permanent entry for %s on %s\n", 591 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 592 inet_ntoa(isaddr), ifp->if_xname); 593 } 594 return; 595 } 596 } 597 /* 598 * sanity check for the address length. 599 * XXX this does not work for protocols with variable address 600 * length. -is 601 */ 602 if (dologging && sdl->sdl_alen && sdl->sdl_alen != ah->ar_hln && 603 cpu == 0) 604 { 605 log(LOG_WARNING, 606 "arp from %*D: new addr len %d, was %d", 607 ifp->if_addrlen, (u_char *) ar_sha(ah), ":", 608 ah->ar_hln, sdl->sdl_alen); 609 } 610 if (ifp->if_addrlen != ah->ar_hln) { 611 if (dologging && cpu == 0) { 612 log(LOG_WARNING, 613 "arp from %*D: addr len: new %d, i/f %d (ignored)", 614 ifp->if_addrlen, (u_char *) ar_sha(ah), ":", 615 ah->ar_hln, ifp->if_addrlen); 616 } 617 return; 618 } 619 memcpy(LLADDR(sdl), ar_sha(ah), sdl->sdl_alen = ah->ar_hln); 620 if (rt->rt_expire != 0) 621 rt->rt_expire = time_second + arpt_keep; 622 rt->rt_flags &= ~RTF_REJECT; 623 la->la_asked = 0; 624 la->la_preempt = arp_maxtries; 625 626 /* 627 * This particular cpu might have been holding an mbuf 628 * pending ARP resolution. If so, transmit the mbuf now. 629 */ 630 if (la->la_hold != NULL) { 631 m_adj(la->la_hold, sizeof(struct ether_header)); 632 lwkt_serialize_enter(ifp->if_serializer); 633 (*ifp->if_output)(ifp, la->la_hold, rt_key(rt), rt); 634 lwkt_serialize_exit(ifp->if_serializer); 635 la->la_hold = NULL; 636 } 637 } 638 } 639 640 #ifdef SMP 641 642 struct netmsg_arp_update { 643 struct netmsg netmsg; 644 struct mbuf *m; 645 in_addr_t saddr; 646 boolean_t create; 647 }; 648 649 static void arp_update_msghandler(struct netmsg *netmsg); 650 651 #endif 652 653 /* 654 * Called from arpintr() - this routine is run from a single cpu. 655 */ 656 static void 657 in_arpinput(struct mbuf *m) 658 { 659 struct arphdr *ah; 660 struct ifnet *ifp = m->m_pkthdr.rcvif; 661 struct ether_header *eh; 662 struct rtentry *rt; 663 struct ifaddr *ifa; 664 struct in_ifaddr *ia; 665 struct sockaddr sa; 666 struct in_addr isaddr, itaddr, myaddr; 667 #ifdef SMP 668 struct netmsg_arp_update msg; 669 #endif 670 u_int8_t *enaddr = NULL; 671 int op; 672 int req_len; 673 674 req_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr)); 675 if (m->m_len < req_len && (m = m_pullup(m, req_len)) == NULL) { 676 log(LOG_ERR, "in_arp: runt packet -- m_pullup failed\n"); 677 return; 678 } 679 680 ah = mtod(m, struct arphdr *); 681 op = ntohs(ah->ar_op); 682 memcpy(&isaddr, ar_spa(ah), sizeof isaddr); 683 memcpy(&itaddr, ar_tpa(ah), sizeof itaddr); 684 /* 685 * Check both target and sender IP addresses: 686 * 687 * If we receive the packet on the interface owning the address, 688 * then accept the address. 689 * 690 * For a bridge, we accept the address if the receive interface and 691 * the interface owning the address are on the same bridge. 692 * (This will change slightly when we have clusters of interfaces). 693 */ 694 LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash) { 695 /* Skip all ia's which don't match */ 696 if (itaddr.s_addr != ia->ia_addr.sin_addr.s_addr) 697 continue; 698 699 if (ia->ia_ifp == ifp) 700 goto match; 701 702 if (ifp->if_bridge && ia->ia_ifp && 703 ifp->if_bridge == ia->ia_ifp->if_bridge) 704 goto match; 705 706 #ifdef CARP 707 /* 708 * If the interface does not match, but the recieving interface 709 * is part of carp, we call carp_iamatch to see if this is a 710 * request for the virtual host ip. 711 * XXX: This is really ugly! 712 */ 713 if (ifp->if_carp != NULL && 714 carp_iamatch(ifp->if_carp, ia, &isaddr, &enaddr) && 715 itaddr.s_addr == ia->ia_addr.sin_addr.s_addr) 716 goto match; 717 #endif 718 } 719 LIST_FOREACH(ia, INADDR_HASH(isaddr.s_addr), ia_hash) { 720 /* Skip all ia's which don't match */ 721 if (isaddr.s_addr != ia->ia_addr.sin_addr.s_addr) 722 continue; 723 724 if (ia->ia_ifp == ifp) 725 goto match; 726 727 if (ifp->if_bridge && ia->ia_ifp && 728 ifp->if_bridge == ia->ia_ifp->if_bridge) 729 goto match; 730 } 731 /* 732 * No match, use the first inet address on the receive interface 733 * as a dummy address for the rest of the function. 734 */ 735 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 736 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET) { 737 ia = ifatoia(ifa); 738 goto match; 739 } 740 } 741 /* 742 * If we got here, we didn't find any suitable interface, 743 * so drop the packet. 744 */ 745 m_freem(m); 746 return; 747 748 match: 749 if (!enaddr) 750 enaddr = (u_int8_t *)IF_LLADDR(ifp); 751 myaddr = ia->ia_addr.sin_addr; 752 if (!bcmp(ar_sha(ah), enaddr, ifp->if_addrlen)) { 753 m_freem(m); /* it's from me, ignore it. */ 754 return; 755 } 756 if (!bcmp(ar_sha(ah), ifp->if_broadcastaddr, ifp->if_addrlen)) { 757 log(LOG_ERR, 758 "arp: link address is broadcast for IP address %s!\n", 759 inet_ntoa(isaddr)); 760 m_freem(m); 761 return; 762 } 763 if (isaddr.s_addr == myaddr.s_addr && myaddr.s_addr != 0) { 764 log(LOG_ERR, 765 "arp: %*D is using my IP address %s!\n", 766 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 767 inet_ntoa(isaddr)); 768 itaddr = myaddr; 769 goto reply; 770 } 771 #ifdef SMP 772 netmsg_init(&msg.netmsg, &curthread->td_msgport, 0, 773 arp_update_msghandler); 774 msg.m = m; 775 msg.saddr = isaddr.s_addr; 776 msg.create = (itaddr.s_addr == myaddr.s_addr); 777 lwkt_domsg(rtable_portfn(0), &msg.netmsg.nm_lmsg, 0); 778 #endif 779 arp_update_oncpu(m, isaddr.s_addr, (itaddr.s_addr == myaddr.s_addr), 780 TRUE); 781 reply: 782 if (op != ARPOP_REQUEST) { 783 m_freem(m); 784 return; 785 } 786 if (itaddr.s_addr == myaddr.s_addr) { 787 /* I am the target */ 788 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 789 memcpy(ar_sha(ah), enaddr, ah->ar_hln); 790 } else { 791 struct llinfo_arp *la; 792 793 la = arplookup(itaddr.s_addr, FALSE, SIN_PROXY); 794 if (la == NULL) { 795 struct sockaddr_in sin; 796 797 if (!arp_proxyall) { 798 m_freem(m); 799 return; 800 } 801 802 bzero(&sin, sizeof sin); 803 sin.sin_family = AF_INET; 804 sin.sin_len = sizeof sin; 805 sin.sin_addr = itaddr; 806 807 rt = rtpurelookup((struct sockaddr *)&sin); 808 if (rt == NULL) { 809 m_freem(m); 810 return; 811 } 812 --rt->rt_refcnt; 813 /* 814 * Don't send proxies for nodes on the same interface 815 * as this one came out of, or we'll get into a fight 816 * over who claims what Ether address. 817 */ 818 if (rt->rt_ifp == ifp) { 819 m_freem(m); 820 return; 821 } 822 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 823 memcpy(ar_sha(ah), enaddr, ah->ar_hln); 824 #ifdef DEBUG_PROXY 825 kprintf("arp: proxying for %s\n", inet_ntoa(itaddr)); 826 #endif 827 } else { 828 struct sockaddr_dl *sdl; 829 830 rt = la->la_rt; 831 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 832 sdl = SDL(rt->rt_gateway); 833 memcpy(ar_sha(ah), LLADDR(sdl), ah->ar_hln); 834 } 835 } 836 837 memcpy(ar_tpa(ah), ar_spa(ah), ah->ar_pln); 838 memcpy(ar_spa(ah), &itaddr, ah->ar_pln); 839 ah->ar_op = htons(ARPOP_REPLY); 840 ah->ar_pro = htons(ETHERTYPE_IP); /* let's be sure! */ 841 switch (ifp->if_type) { 842 case IFT_ETHER: 843 /* 844 * May not be correct for types not explictly 845 * listed, but it is our best guess. 846 */ 847 default: 848 eh = (struct ether_header *)sa.sa_data; 849 memcpy(eh->ether_dhost, ar_tha(ah), sizeof eh->ether_dhost); 850 eh->ether_type = htons(ETHERTYPE_ARP); 851 break; 852 } 853 sa.sa_family = AF_UNSPEC; 854 sa.sa_len = sizeof sa; 855 lwkt_serialize_enter(ifp->if_serializer); 856 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0); 857 lwkt_serialize_exit(ifp->if_serializer); 858 return; 859 } 860 861 #ifdef SMP 862 863 static 864 void 865 arp_update_msghandler(struct netmsg *netmsg) 866 { 867 struct netmsg_arp_update *msg = (struct netmsg_arp_update *)netmsg; 868 int nextcpu; 869 870 arp_update_oncpu(msg->m, msg->saddr, msg->create, FALSE); 871 872 nextcpu = mycpuid + 1; 873 if (nextcpu < ncpus) { 874 lwkt_forwardmsg(rtable_portfn(nextcpu), &msg->netmsg.nm_lmsg); 875 } else { 876 lwkt_replymsg(&msg->netmsg.nm_lmsg, 0); 877 } 878 } 879 880 #endif 881 882 #endif 883 884 /* 885 * Free an arp entry. If the arp entry is actively referenced or represents 886 * a static entry we only clear it back to an unresolved state, otherwise 887 * we destroy the entry entirely. 888 * 889 * Note that static entries are created when route add ... -interface is used 890 * to create an interface route to a (direct) destination. 891 */ 892 static void 893 arptfree(struct llinfo_arp *la) 894 { 895 struct rtentry *rt = la->la_rt; 896 struct sockaddr_dl *sdl; 897 898 if (rt == NULL) 899 panic("arptfree"); 900 sdl = SDL(rt->rt_gateway); 901 if (sdl != NULL && 902 ((rt->rt_refcnt > 0 && sdl->sdl_family == AF_LINK) || 903 (rt->rt_flags & RTF_STATIC))) { 904 sdl->sdl_alen = 0; 905 la->la_preempt = la->la_asked = 0; 906 rt->rt_flags &= ~RTF_REJECT; 907 return; 908 } 909 rtrequest(RTM_DELETE, rt_key(rt), NULL, rt_mask(rt), 0, NULL); 910 } 911 912 /* 913 * Lookup or enter a new address in arptab. 914 */ 915 static struct llinfo_arp * 916 arplookup(in_addr_t addr, boolean_t create, boolean_t proxy) 917 { 918 struct rtentry *rt; 919 struct sockaddr_inarp sin = { sizeof sin, AF_INET }; 920 const char *why = NULL; 921 922 sin.sin_addr.s_addr = addr; 923 sin.sin_other = proxy ? SIN_PROXY : 0; 924 if (create) 925 rt = rtlookup((struct sockaddr *)&sin); 926 else 927 rt = rtpurelookup((struct sockaddr *)&sin); 928 if (rt == NULL) 929 return (NULL); 930 rt->rt_refcnt--; 931 932 if (rt->rt_flags & RTF_GATEWAY) 933 why = "host is not on local network"; 934 else if (!(rt->rt_flags & RTF_LLINFO)) 935 why = "could not allocate llinfo"; 936 else if (rt->rt_gateway->sa_family != AF_LINK) 937 why = "gateway route is not ours"; 938 939 if (why) { 940 if (create) { 941 log(LOG_DEBUG, "arplookup %s failed: %s\n", 942 inet_ntoa(sin.sin_addr), why); 943 } 944 if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_WASCLONED)) { 945 /* No references to this route. Purge it. */ 946 rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway, 947 rt_mask(rt), rt->rt_flags, NULL); 948 } 949 return (NULL); 950 } 951 return (rt->rt_llinfo); 952 } 953 954 void 955 arp_ifinit(struct ifnet *ifp, struct ifaddr *ifa) 956 { 957 if (IA_SIN(ifa)->sin_addr.s_addr != INADDR_ANY) 958 arprequest(ifp, &IA_SIN(ifa)->sin_addr, &IA_SIN(ifa)->sin_addr, 959 IF_LLADDR(ifp)); 960 ifa->ifa_rtrequest = arp_rtrequest; 961 ifa->ifa_flags |= RTF_CLONING; 962 } 963 964 void 965 arp_ifinit2(struct ifnet *ifp, struct ifaddr *ifa, u_char *enaddr) 966 { 967 if (IA_SIN(ifa)->sin_addr.s_addr != INADDR_ANY) 968 arprequest(ifp, &IA_SIN(ifa)->sin_addr, &IA_SIN(ifa)->sin_addr, 969 enaddr); 970 ifa->ifa_rtrequest = arp_rtrequest; 971 ifa->ifa_flags |= RTF_CLONING; 972 } 973 974 static void 975 arp_init(void) 976 { 977 int cpu; 978 979 for (cpu = 0; cpu < ncpus2; cpu++) 980 LIST_INIT(&llinfo_arp_list[cpu]); 981 netisr_register(NETISR_ARP, cpu0_portfn, arpintr); 982 } 983 984 SYSINIT(arp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, arp_init, 0); 985