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 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 $ 85 * $DragonFly: src/sys/netinet/if_ether.c,v 1.34 2006/07/20 16:22:48 corecode Exp $ 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 96 #include <sys/param.h> 97 #include <sys/kernel.h> 98 #include <sys/queue.h> 99 #include <sys/sysctl.h> 100 #include <sys/systm.h> 101 #include <sys/mbuf.h> 102 #include <sys/malloc.h> 103 #include <sys/socket.h> 104 #include <sys/syslog.h> 105 106 #include <sys/thread2.h> 107 #include <sys/msgport2.h> 108 109 #include <net/if.h> 110 #include <net/if_dl.h> 111 #include <net/if_types.h> 112 #include <net/route.h> 113 #include <net/netisr.h> 114 #include <net/if_llc.h> 115 116 #include <netinet/in.h> 117 #include <netinet/in_var.h> 118 #include <netinet/if_ether.h> 119 120 #include <net/if_arc.h> 121 #include <net/iso88025.h> 122 123 #define SIN(s) ((struct sockaddr_in *)s) 124 #define SDL(s) ((struct sockaddr_dl *)s) 125 126 SYSCTL_DECL(_net_link_ether); 127 SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, ""); 128 129 /* timer values */ 130 static int arpt_prune = (5*60*1); /* walk list every 5 minutes */ 131 static int arpt_keep = (20*60); /* once resolved, good for 20 more minutes */ 132 static int arpt_down = 20; /* once declared down, don't send for 20 sec */ 133 134 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, prune_intvl, CTLFLAG_RW, 135 &arpt_prune, 0, ""); 136 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, max_age, CTLFLAG_RW, 137 &arpt_keep, 0, ""); 138 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, host_down_time, CTLFLAG_RW, 139 &arpt_down, 0, ""); 140 141 #define rt_expire rt_rmx.rmx_expire 142 143 struct llinfo_arp { 144 LIST_ENTRY(llinfo_arp) la_le; 145 struct rtentry *la_rt; 146 struct mbuf *la_hold; /* last packet until resolved/timeout */ 147 u_short la_preempt; /* countdown for pre-expiry arps */ 148 u_short la_asked; /* #times we QUERIED following expiration */ 149 }; 150 151 static LIST_HEAD(, llinfo_arp) llinfo_arp_list[MAXCPU]; 152 153 static int arp_maxtries = 5; 154 static int useloopback = 1; /* use loopback interface for local traffic */ 155 static int arp_proxyall = 0; 156 157 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW, 158 &arp_maxtries, 0, ""); 159 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW, 160 &useloopback, 0, ""); 161 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW, 162 &arp_proxyall, 0, ""); 163 164 static void arp_rtrequest (int, struct rtentry *, struct rt_addrinfo *); 165 static void arprequest (struct ifnet *, 166 struct in_addr *, struct in_addr *, u_char *); 167 static int arpintr(struct netmsg *); 168 static void arptfree (struct llinfo_arp *); 169 static void arptimer (void *); 170 static struct llinfo_arp 171 *arplookup (in_addr_t addr, boolean_t create, boolean_t proxy); 172 #ifdef INET 173 static void in_arpinput (struct mbuf *); 174 #endif 175 176 static struct callout arptimer_ch[MAXCPU]; 177 178 /* 179 * Timeout routine. Age arp_tab entries periodically. 180 */ 181 /* ARGSUSED */ 182 static void 183 arptimer(void *ignored_arg) 184 { 185 struct llinfo_arp *la, *nla; 186 187 crit_enter(); 188 LIST_FOREACH_MUTABLE(la, &llinfo_arp_list[mycpuid], la_le, nla) { 189 if (la->la_rt->rt_expire && la->la_rt->rt_expire <= time_second) 190 arptfree(la); 191 } 192 callout_reset(&arptimer_ch[mycpuid], arpt_prune * hz, arptimer, NULL); 193 crit_exit(); 194 } 195 196 /* 197 * Parallel to llc_rtrequest. 198 */ 199 static void 200 arp_rtrequest(int req, struct rtentry *rt, struct rt_addrinfo *info) 201 { 202 struct sockaddr *gate = rt->rt_gateway; 203 struct llinfo_arp *la = rt->rt_llinfo; 204 205 struct sockaddr_dl null_sdl = { sizeof null_sdl, AF_LINK }; 206 static boolean_t arpinit_done[MAXCPU]; 207 208 if (!arpinit_done[mycpuid]) { 209 arpinit_done[mycpuid] = TRUE; 210 callout_init(&arptimer_ch[mycpuid]); 211 callout_reset(&arptimer_ch[mycpuid], hz, arptimer, NULL); 212 } 213 if (rt->rt_flags & RTF_GATEWAY) 214 return; 215 216 switch (req) { 217 case RTM_ADD: 218 /* 219 * XXX: If this is a manually added route to interface 220 * such as older version of routed or gated might provide, 221 * restore cloning bit. 222 */ 223 if (!(rt->rt_flags & RTF_HOST) && 224 SIN(rt_mask(rt))->sin_addr.s_addr != 0xffffffff) 225 rt->rt_flags |= RTF_CLONING; 226 if (rt->rt_flags & RTF_CLONING) { 227 /* 228 * Case 1: This route should come from a route to iface. 229 */ 230 rt_setgate(rt, rt_key(rt), 231 (struct sockaddr *)&null_sdl); 232 gate = rt->rt_gateway; 233 SDL(gate)->sdl_type = rt->rt_ifp->if_type; 234 SDL(gate)->sdl_index = rt->rt_ifp->if_index; 235 rt->rt_expire = time_second; 236 break; 237 } 238 /* Announce a new entry if requested. */ 239 if (rt->rt_flags & RTF_ANNOUNCE) 240 arprequest(rt->rt_ifp, 241 &SIN(rt_key(rt))->sin_addr, 242 &SIN(rt_key(rt))->sin_addr, 243 LLADDR(SDL(gate))); 244 /*FALLTHROUGH*/ 245 case RTM_RESOLVE: 246 if (gate->sa_family != AF_LINK || 247 gate->sa_len < sizeof(struct sockaddr_dl)) { 248 log(LOG_DEBUG, "arp_rtrequest: bad gateway value\n"); 249 break; 250 } 251 SDL(gate)->sdl_type = rt->rt_ifp->if_type; 252 SDL(gate)->sdl_index = rt->rt_ifp->if_index; 253 if (la != NULL) 254 break; /* This happens on a route change */ 255 /* 256 * Case 2: This route may come from cloning, or a manual route 257 * add with a LL address. 258 */ 259 R_Malloc(la, struct llinfo_arp *, sizeof *la); 260 rt->rt_llinfo = la; 261 if (la == NULL) { 262 log(LOG_DEBUG, "arp_rtrequest: malloc failed\n"); 263 break; 264 } 265 bzero(la, sizeof *la); 266 la->la_rt = rt; 267 rt->rt_flags |= RTF_LLINFO; 268 LIST_INSERT_HEAD(&llinfo_arp_list[mycpuid], la, la_le); 269 270 #ifdef INET 271 /* 272 * This keeps the multicast addresses from showing up 273 * in `arp -a' listings as unresolved. It's not actually 274 * functional. Then the same for broadcast. 275 */ 276 if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr)) && 277 rt->rt_ifp->if_type != IFT_ARCNET) { 278 ETHER_MAP_IP_MULTICAST(&SIN(rt_key(rt))->sin_addr, 279 LLADDR(SDL(gate))); 280 SDL(gate)->sdl_alen = 6; 281 rt->rt_expire = 0; 282 } 283 if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) { 284 memcpy(LLADDR(SDL(gate)), rt->rt_ifp->if_broadcastaddr, 285 rt->rt_ifp->if_addrlen); 286 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen; 287 rt->rt_expire = 0; 288 } 289 #endif 290 291 if (SIN(rt_key(rt))->sin_addr.s_addr == 292 (IA_SIN(rt->rt_ifa))->sin_addr.s_addr) { 293 /* 294 * This test used to be 295 * if (loif.if_flags & IFF_UP) 296 * It allowed local traffic to be forced 297 * through the hardware by configuring the 298 * loopback down. However, it causes problems 299 * during network configuration for boards 300 * that can't receive packets they send. It 301 * is now necessary to clear "useloopback" and 302 * remove the route to force traffic out to 303 * the hardware. 304 */ 305 rt->rt_expire = 0; 306 bcopy(IF_LLADDR(rt->rt_ifp), LLADDR(SDL(gate)), 307 SDL(gate)->sdl_alen = rt->rt_ifp->if_addrlen); 308 if (useloopback) 309 rt->rt_ifp = loif; 310 } 311 break; 312 313 case RTM_DELETE: 314 if (la == NULL) 315 break; 316 LIST_REMOVE(la, la_le); 317 rt->rt_llinfo = NULL; 318 rt->rt_flags &= ~RTF_LLINFO; 319 if (la->la_hold != NULL) 320 m_freem(la->la_hold); 321 Free(la); 322 } 323 } 324 325 /* 326 * Broadcast an ARP request. Caller specifies: 327 * - arp header source ip address 328 * - arp header target ip address 329 * - arp header source ethernet address 330 */ 331 static void 332 arprequest(struct ifnet *ifp, struct in_addr *sip, struct in_addr *tip, 333 u_char *enaddr) 334 { 335 struct mbuf *m; 336 struct ether_header *eh; 337 struct arc_header *arh; 338 struct arphdr *ah; 339 struct sockaddr sa; 340 static u_char llcx[] = { 0x82, 0x40, LLC_SNAP_LSAP, LLC_SNAP_LSAP, 341 LLC_UI, 0x00, 0x00, 0x00, 0x08, 0x06 }; 342 u_short ar_hrd; 343 344 if ((m = m_gethdr(MB_DONTWAIT, MT_DATA)) == NULL) 345 return; 346 m->m_pkthdr.rcvif = (struct ifnet *)NULL; 347 348 switch (ifp->if_type) { 349 case IFT_ARCNET: 350 ar_hrd = htons(ARPHRD_ARCNET); 351 352 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr)); 353 m->m_pkthdr.len = m->m_len; 354 MH_ALIGN(m, m->m_len); 355 356 arh = (struct arc_header *)sa.sa_data; 357 arh->arc_dhost = ifp->if_broadcastaddr[0]; 358 arh->arc_type = ARCTYPE_ARP; 359 360 ah = mtod(m, struct arphdr *); 361 break; 362 363 case IFT_ISO88025: 364 ar_hrd = htons(ARPHRD_IEEE802); 365 366 m->m_len = (sizeof llcx) + 367 arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr)); 368 m->m_pkthdr.len = m->m_len; 369 MH_ALIGN(m, m->m_len); 370 371 memcpy(mtod(m, caddr_t), llcx, sizeof llcx); 372 memcpy(sa.sa_data, ifp->if_broadcastaddr, ifp->if_addrlen); 373 memcpy(sa.sa_data + 6, enaddr, 6); 374 sa.sa_data[6] |= TR_RII; 375 sa.sa_data[12] = TR_AC; 376 sa.sa_data[13] = TR_LLC_FRAME; 377 378 ah = (struct arphdr *)(mtod(m, char *) + sizeof llcx); 379 break; 380 case IFT_FDDI: 381 case IFT_ETHER: 382 /* 383 * This may not be correct for types not explicitly 384 * listed, but this is our best guess 385 */ 386 default: 387 ar_hrd = htons(ARPHRD_ETHER); 388 389 m->m_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr)); 390 m->m_pkthdr.len = m->m_len; 391 MH_ALIGN(m, m->m_len); 392 393 eh = (struct ether_header *)sa.sa_data; 394 /* if_output() will not swap */ 395 eh->ether_type = htons(ETHERTYPE_ARP); 396 memcpy(eh->ether_dhost, ifp->if_broadcastaddr, ifp->if_addrlen); 397 398 ah = mtod(m, struct arphdr *); 399 break; 400 } 401 402 ah->ar_hrd = ar_hrd; 403 ah->ar_pro = htons(ETHERTYPE_IP); 404 ah->ar_hln = ifp->if_addrlen; /* hardware address length */ 405 ah->ar_pln = sizeof(struct in_addr); /* protocol address length */ 406 ah->ar_op = htons(ARPOP_REQUEST); 407 memcpy(ar_sha(ah), enaddr, ah->ar_hln); 408 memset(ar_tha(ah), 0, ah->ar_hln); 409 memcpy(ar_spa(ah), sip, ah->ar_pln); 410 memcpy(ar_tpa(ah), tip, ah->ar_pln); 411 412 sa.sa_family = AF_UNSPEC; 413 sa.sa_len = sizeof sa; 414 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)NULL); 415 } 416 417 /* 418 * Resolve an IP address into an ethernet address. If success, 419 * desten is filled in. If there is no entry in arptab, 420 * set one up and broadcast a request for the IP address. 421 * Hold onto this mbuf and resend it once the address 422 * is finally resolved. A return value of 1 indicates 423 * that desten has been filled in and the packet should be sent 424 * normally; a 0 return indicates that the packet has been 425 * taken over here, either now or for later transmission. 426 */ 427 int 428 arpresolve( 429 struct ifnet *ifp, 430 struct rtentry *rt0, 431 struct mbuf *m, 432 struct sockaddr *dst, 433 u_char *desten) 434 { 435 struct rtentry *rt; 436 struct llinfo_arp *la = NULL; 437 struct sockaddr_dl *sdl; 438 439 if (m->m_flags & M_BCAST) { /* broadcast */ 440 memcpy(desten, ifp->if_broadcastaddr, ifp->if_addrlen); 441 return (1); 442 } 443 if (m->m_flags & M_MCAST && ifp->if_type != IFT_ARCNET) {/* multicast */ 444 ETHER_MAP_IP_MULTICAST(&SIN(dst)->sin_addr, desten); 445 return (1); 446 } 447 if (rt0 != NULL) { 448 if (rt_llroute(dst, rt0, &rt) != 0) { 449 m_freem(m); 450 return 0; 451 } 452 la = rt->rt_llinfo; 453 } 454 if (la == NULL) { 455 la = arplookup(SIN(dst)->sin_addr.s_addr, TRUE, FALSE); 456 if (la != NULL) 457 rt = la->la_rt; 458 } 459 if (la == NULL || rt == NULL) { 460 log(LOG_DEBUG, "arpresolve: can't allocate llinfo for %s%s%s\n", 461 inet_ntoa(SIN(dst)->sin_addr), la ? "la" : " ", 462 rt ? "rt" : ""); 463 m_freem(m); 464 return (0); 465 } 466 sdl = SDL(rt->rt_gateway); 467 /* 468 * Check the address family and length is valid, the address 469 * is resolved; otherwise, try to resolve. 470 */ 471 if ((rt->rt_expire == 0 || rt->rt_expire > time_second) && 472 sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0) { 473 /* 474 * If entry has an expiry time and it is approaching, 475 * see if we need to send an ARP request within this 476 * arpt_down interval. 477 */ 478 if ((rt->rt_expire != 0) && 479 (time_second + la->la_preempt > rt->rt_expire)) { 480 arprequest(ifp, 481 &SIN(rt->rt_ifa->ifa_addr)->sin_addr, 482 &SIN(dst)->sin_addr, 483 IF_LLADDR(ifp)); 484 la->la_preempt--; 485 } 486 487 bcopy(LLADDR(sdl), desten, sdl->sdl_alen); 488 return 1; 489 } 490 /* 491 * If ARP is disabled on this interface, stop. 492 * XXX 493 * Probably should not allocate empty llinfo struct if we are 494 * not going to be sending out an arp request. 495 */ 496 if (ifp->if_flags & IFF_NOARP) { 497 m_freem(m); 498 return (0); 499 } 500 /* 501 * There is an arptab entry, but no ethernet address 502 * response yet. Replace the held mbuf with this 503 * latest one. 504 */ 505 if (la->la_hold != NULL) 506 m_freem(la->la_hold); 507 la->la_hold = m; 508 if (rt->rt_expire || ((rt->rt_flags & RTF_STATIC) && !sdl->sdl_alen)) { 509 rt->rt_flags &= ~RTF_REJECT; 510 if (la->la_asked == 0 || rt->rt_expire != time_second) { 511 rt->rt_expire = time_second; 512 if (la->la_asked++ < arp_maxtries) { 513 arprequest(ifp, 514 &SIN(rt->rt_ifa->ifa_addr)->sin_addr, 515 &SIN(dst)->sin_addr, 516 IF_LLADDR(ifp)); 517 } else { 518 rt->rt_flags |= RTF_REJECT; 519 rt->rt_expire += arpt_down; 520 la->la_asked = 0; 521 la->la_preempt = arp_maxtries; 522 } 523 524 } 525 } 526 return (0); 527 } 528 529 /* 530 * Common length and type checks are done here, 531 * then the protocol-specific routine is called. 532 */ 533 static int 534 arpintr(struct netmsg *msg) 535 { 536 struct mbuf *m = ((struct netmsg_packet *)msg)->nm_packet; 537 struct arphdr *ar; 538 u_short ar_hrd; 539 540 if (m->m_len < sizeof(struct arphdr) && 541 ((m = m_pullup(m, sizeof(struct arphdr))) == NULL)) { 542 log(LOG_ERR, "arp: runt packet -- m_pullup failed\n"); 543 goto out2; 544 } 545 ar = mtod(m, struct arphdr *); 546 547 ar_hrd = ntohs(ar->ar_hrd); 548 if (ar_hrd != ARPHRD_ETHER && 549 ar_hrd != ARPHRD_IEEE802 && 550 ar_hrd != ARPHRD_ARCNET) { 551 log(LOG_ERR, 552 "arp: unknown hardware address format (0x%2D)\n", 553 (unsigned char *)&ar->ar_hrd, ""); 554 goto out1; 555 } 556 557 if (m->m_pkthdr.len < arphdr_len(ar) && 558 (m = m_pullup(m, arphdr_len(ar))) == NULL) { 559 log(LOG_ERR, "arp: runt packet\n"); 560 goto out1; 561 } 562 563 switch (ntohs(ar->ar_pro)) { 564 #ifdef INET 565 case ETHERTYPE_IP: 566 in_arpinput(m); 567 goto out2; 568 #endif 569 } 570 out1: 571 m_freem(m); 572 out2: 573 /* msg was embedded in the mbuf, do not reply! */ 574 return(EASYNC); 575 } 576 577 #ifdef INET 578 /* 579 * ARP for Internet protocols on 10 Mb/s Ethernet. 580 * Algorithm is that given in RFC 826. 581 * In addition, a sanity check is performed on the sender 582 * protocol address, to catch impersonators. 583 * We no longer handle negotiations for use of trailer protocol: 584 * Formerly, ARP replied for protocol type ETHERTYPE_TRAIL sent 585 * along with IP replies if we wanted trailers sent to us, 586 * and also sent them in response to IP replies. 587 * This allowed either end to announce the desire to receive 588 * trailer packets. 589 * We no longer reply to requests for ETHERTYPE_TRAIL protocol either, 590 * but formerly didn't normally send requests. 591 */ 592 static int log_arp_wrong_iface = 1; 593 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_wrong_iface, CTLFLAG_RW, 594 &log_arp_wrong_iface, 0, 595 "log arp packets arriving on the wrong interface"); 596 597 static void 598 arp_update_oncpu(struct mbuf *m, in_addr_t saddr, boolean_t create, 599 boolean_t dologging) 600 { 601 struct arphdr *ah = mtod(m, struct arphdr *); 602 struct ifnet *ifp = m->m_pkthdr.rcvif; 603 struct llinfo_arp *la; 604 struct sockaddr_dl *sdl; 605 struct rtentry *rt; 606 int cpu = mycpuid; 607 608 la = arplookup(saddr, create, FALSE); 609 if (la && (rt = la->la_rt) && (sdl = SDL(rt->rt_gateway))) { 610 struct in_addr isaddr = { saddr }; 611 612 /* the following is not an error when doing bridging */ 613 if (rt->rt_ifp != ifp) { 614 if (dologging && log_arp_wrong_iface && cpu == 0) { 615 log(LOG_ERR, 616 "arp: %s is on %s " 617 "but got reply from %*D on %s\n", 618 inet_ntoa(isaddr), 619 rt->rt_ifp->if_xname, 620 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 621 ifp->if_xname); 622 } 623 return; 624 } 625 if (sdl->sdl_alen && 626 bcmp(ar_sha(ah), LLADDR(sdl), sdl->sdl_alen)) { 627 if (rt->rt_expire != 0) { 628 if (dologging && cpu == 0) { 629 log(LOG_INFO, 630 "arp: %s moved from %*D to %*D on %s\n", 631 inet_ntoa(isaddr), 632 ifp->if_addrlen, (u_char *)LLADDR(sdl), 633 ":", ifp->if_addrlen, 634 (u_char *)ar_sha(ah), ":", 635 ifp->if_xname); 636 } 637 } else { 638 if (dologging && cpu == 0) { 639 log(LOG_ERR, 640 "arp: %*D attempts to modify permanent entry for %s on %s\n", 641 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 642 inet_ntoa(isaddr), ifp->if_xname); 643 } 644 return; 645 } 646 } 647 /* 648 * sanity check for the address length. 649 * XXX this does not work for protocols with variable address 650 * length. -is 651 */ 652 if (dologging && sdl->sdl_alen && sdl->sdl_alen != ah->ar_hln && 653 cpu == 0) 654 { 655 log(LOG_WARNING, 656 "arp from %*D: new addr len %d, was %d", 657 ifp->if_addrlen, (u_char *) ar_sha(ah), ":", 658 ah->ar_hln, sdl->sdl_alen); 659 } 660 if (ifp->if_addrlen != ah->ar_hln) { 661 if (dologging && cpu == 0) { 662 log(LOG_WARNING, 663 "arp from %*D: addr len: new %d, i/f %d (ignored)", 664 ifp->if_addrlen, (u_char *) ar_sha(ah), ":", 665 ah->ar_hln, ifp->if_addrlen); 666 } 667 return; 668 } 669 memcpy(LLADDR(sdl), ar_sha(ah), sdl->sdl_alen = ah->ar_hln); 670 /* 671 * If we receive an arp from a token-ring station over 672 * a token-ring nic then try to save the source 673 * routing info. 674 */ 675 if (ifp->if_type == IFT_ISO88025) { 676 struct iso88025_header *th = 677 (struct iso88025_header *)m->m_pkthdr.header; 678 struct iso88025_sockaddr_dl_data *trld = 679 SDL_ISO88025(sdl); 680 int rif_len; 681 682 rif_len = TR_RCF_RIFLEN(th->rcf); 683 if ((th->iso88025_shost[0] & TR_RII) && 684 (rif_len > 2)) { 685 trld->trld_rcf = th->rcf; 686 trld->trld_rcf ^= htons(TR_RCF_DIR); 687 memcpy(trld->trld_route, th->rd, rif_len - 2); 688 trld->trld_rcf &= ~htons(TR_RCF_BCST_MASK); 689 /* 690 * Set up source routing information for 691 * reply packet (XXX) 692 */ 693 m->m_data -= rif_len; 694 m->m_len += rif_len; 695 m->m_pkthdr.len += rif_len; 696 } else { 697 th->iso88025_shost[0] &= ~TR_RII; 698 trld->trld_rcf = 0; 699 } 700 m->m_data -= 8; 701 m->m_len += 8; 702 m->m_pkthdr.len += 8; 703 th->rcf = trld->trld_rcf; 704 } 705 if (rt->rt_expire != 0) 706 rt->rt_expire = time_second + arpt_keep; 707 rt->rt_flags &= ~RTF_REJECT; 708 la->la_asked = 0; 709 la->la_preempt = arp_maxtries; 710 711 /* 712 * This particular cpu might have been holding an mbuf 713 * pending ARP resolution. If so, transmit the mbuf now. 714 */ 715 if (la->la_hold != NULL) { 716 m_adj(la->la_hold, sizeof(struct ether_header)); 717 lwkt_serialize_enter(ifp->if_serializer); 718 (*ifp->if_output)(ifp, la->la_hold, rt_key(rt), rt); 719 lwkt_serialize_exit(ifp->if_serializer); 720 la->la_hold = NULL; 721 } 722 } 723 } 724 725 #ifdef SMP 726 727 struct netmsg_arp_update { 728 struct lwkt_msg lmsg; 729 struct mbuf *m; 730 in_addr_t saddr; 731 boolean_t create; 732 }; 733 734 static int arp_update_msghandler(struct lwkt_msg *lmsg); 735 736 #endif 737 738 /* 739 * Called from arpintr() - this routine is run from a single cpu. 740 */ 741 static void 742 in_arpinput(struct mbuf *m) 743 { 744 struct arphdr *ah; 745 struct ifnet *ifp = m->m_pkthdr.rcvif; 746 struct ether_header *eh; 747 struct arc_header *arh; 748 struct iso88025_header *th = (struct iso88025_header *)NULL; 749 struct rtentry *rt; 750 struct ifaddr *ifa; 751 struct in_ifaddr *ia; 752 struct sockaddr sa; 753 struct in_addr isaddr, itaddr, myaddr; 754 #ifdef SMP 755 struct netmsg_arp_update msg; 756 #endif 757 int op; 758 int req_len; 759 760 req_len = arphdr_len2(ifp->if_addrlen, sizeof(struct in_addr)); 761 if (m->m_len < req_len && (m = m_pullup(m, req_len)) == NULL) { 762 log(LOG_ERR, "in_arp: runt packet -- m_pullup failed\n"); 763 return; 764 } 765 766 ah = mtod(m, struct arphdr *); 767 op = ntohs(ah->ar_op); 768 memcpy(&isaddr, ar_spa(ah), sizeof isaddr); 769 memcpy(&itaddr, ar_tpa(ah), sizeof itaddr); 770 /* 771 * Check both target and sender IP addresses: 772 * 773 * If we receive the packet on the interface owning the address, 774 * then accept the address. 775 * 776 * For a bridge, we accept the address if the receive interface and 777 * the interface owning the address are on the same bridge. 778 * (This will change slightly when we have clusters of interfaces). 779 */ 780 LIST_FOREACH(ia, INADDR_HASH(itaddr.s_addr), ia_hash) { 781 /* Skip all ia's which don't match */ 782 if (itaddr.s_addr != ia->ia_addr.sin_addr.s_addr) 783 continue; 784 785 if (ia->ia_ifp == ifp) 786 goto match; 787 788 if (ifp->if_bridge && ia->ia_ifp && 789 ifp->if_bridge == ia->ia_ifp->if_bridge) 790 goto match; 791 } 792 LIST_FOREACH(ia, INADDR_HASH(isaddr.s_addr), ia_hash) { 793 /* Skip all ia's which don't match */ 794 if (isaddr.s_addr != ia->ia_addr.sin_addr.s_addr) 795 continue; 796 797 if (ia->ia_ifp == ifp) 798 goto match; 799 800 if (ifp->if_bridge && ia->ia_ifp && 801 ifp->if_bridge == ia->ia_ifp->if_bridge) 802 goto match; 803 } 804 /* 805 * No match, use the first inet address on the receive interface 806 * as a dummy address for the rest of the function. 807 */ 808 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { 809 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET) { 810 ia = ifatoia(ifa); 811 goto match; 812 } 813 } 814 /* 815 * If we got here, we didn't find any suitable interface, 816 * so drop the packet. 817 */ 818 m_freem(m); 819 return; 820 821 match: 822 myaddr = ia->ia_addr.sin_addr; 823 if (!bcmp(ar_sha(ah), IF_LLADDR(ifp), ifp->if_addrlen)) { 824 m_freem(m); /* it's from me, ignore it. */ 825 return; 826 } 827 if (!bcmp(ar_sha(ah), ifp->if_broadcastaddr, ifp->if_addrlen)) { 828 log(LOG_ERR, 829 "arp: link address is broadcast for IP address %s!\n", 830 inet_ntoa(isaddr)); 831 m_freem(m); 832 return; 833 } 834 if (isaddr.s_addr == myaddr.s_addr && myaddr.s_addr != 0) { 835 log(LOG_ERR, 836 "arp: %*D is using my IP address %s!\n", 837 ifp->if_addrlen, (u_char *)ar_sha(ah), ":", 838 inet_ntoa(isaddr)); 839 itaddr = myaddr; 840 goto reply; 841 } 842 #ifdef SMP 843 lwkt_initmsg(&msg.lmsg, &curthread->td_msgport, 0, 844 lwkt_cmd_func(arp_update_msghandler), lwkt_cmd_op_none); 845 msg.m = m; 846 msg.saddr = isaddr.s_addr; 847 msg.create = (itaddr.s_addr == myaddr.s_addr); 848 lwkt_domsg(rtable_portfn(0), &msg.lmsg); 849 #endif 850 arp_update_oncpu(m, isaddr.s_addr, (itaddr.s_addr == myaddr.s_addr), 851 TRUE); 852 reply: 853 if (op != ARPOP_REQUEST) { 854 m_freem(m); 855 return; 856 } 857 if (itaddr.s_addr == myaddr.s_addr) { 858 /* I am the target */ 859 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 860 memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln); 861 } else { 862 struct llinfo_arp *la; 863 864 la = arplookup(itaddr.s_addr, FALSE, SIN_PROXY); 865 if (la == NULL) { 866 struct sockaddr_in sin; 867 868 if (!arp_proxyall) { 869 m_freem(m); 870 return; 871 } 872 873 bzero(&sin, sizeof sin); 874 sin.sin_family = AF_INET; 875 sin.sin_len = sizeof sin; 876 sin.sin_addr = itaddr; 877 878 rt = rtpurelookup((struct sockaddr *)&sin); 879 if (rt == NULL) { 880 m_freem(m); 881 return; 882 } 883 --rt->rt_refcnt; 884 /* 885 * Don't send proxies for nodes on the same interface 886 * as this one came out of, or we'll get into a fight 887 * over who claims what Ether address. 888 */ 889 if (rt->rt_ifp == ifp) { 890 m_freem(m); 891 return; 892 } 893 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 894 memcpy(ar_sha(ah), IF_LLADDR(ifp), ah->ar_hln); 895 #ifdef DEBUG_PROXY 896 printf("arp: proxying for %s\n", inet_ntoa(itaddr)); 897 #endif 898 } else { 899 struct sockaddr_dl *sdl; 900 901 rt = la->la_rt; 902 memcpy(ar_tha(ah), ar_sha(ah), ah->ar_hln); 903 sdl = SDL(rt->rt_gateway); 904 memcpy(ar_sha(ah), LLADDR(sdl), ah->ar_hln); 905 } 906 } 907 908 memcpy(ar_tpa(ah), ar_spa(ah), ah->ar_pln); 909 memcpy(ar_spa(ah), &itaddr, ah->ar_pln); 910 ah->ar_op = htons(ARPOP_REPLY); 911 ah->ar_pro = htons(ETHERTYPE_IP); /* let's be sure! */ 912 switch (ifp->if_type) { 913 case IFT_ARCNET: 914 arh = (struct arc_header *)sa.sa_data; 915 arh->arc_dhost = *ar_tha(ah); 916 arh->arc_type = ARCTYPE_ARP; 917 break; 918 case IFT_ISO88025: 919 /* Re-arrange the source/dest address */ 920 memcpy(th->iso88025_dhost, th->iso88025_shost, 921 sizeof th->iso88025_dhost); 922 memcpy(th->iso88025_shost, IF_LLADDR(ifp), 923 sizeof th->iso88025_shost); 924 /* Set the source routing bit if neccesary */ 925 if (th->iso88025_dhost[0] & TR_RII) { 926 th->iso88025_dhost[0] &= ~TR_RII; 927 if (TR_RCF_RIFLEN(th->rcf) > 2) 928 th->iso88025_shost[0] |= TR_RII; 929 } 930 /* Copy the addresses, ac and fc into sa_data */ 931 memcpy(sa.sa_data, th->iso88025_dhost, 932 (sizeof th->iso88025_dhost) * 2); 933 sa.sa_data[(sizeof th->iso88025_dhost) * 2] = TR_AC; 934 sa.sa_data[(sizeof th->iso88025_dhost) * 2 + 1] = TR_LLC_FRAME; 935 break; 936 case IFT_ETHER: 937 case IFT_FDDI: 938 /* 939 * May not be correct for types not explictly 940 * listed, but it is our best guess. 941 */ 942 default: 943 eh = (struct ether_header *)sa.sa_data; 944 memcpy(eh->ether_dhost, ar_tha(ah), sizeof eh->ether_dhost); 945 eh->ether_type = htons(ETHERTYPE_ARP); 946 break; 947 } 948 sa.sa_family = AF_UNSPEC; 949 sa.sa_len = sizeof sa; 950 lwkt_serialize_enter(ifp->if_serializer); 951 (*ifp->if_output)(ifp, m, &sa, (struct rtentry *)0); 952 lwkt_serialize_exit(ifp->if_serializer); 953 return; 954 } 955 956 #ifdef SMP 957 958 static 959 int 960 arp_update_msghandler(struct lwkt_msg *lmsg) 961 { 962 struct netmsg_arp_update *msg = (struct netmsg_arp_update *)lmsg; 963 int nextcpu; 964 965 arp_update_oncpu(msg->m, msg->saddr, msg->create, FALSE); 966 967 nextcpu = mycpuid + 1; 968 if (nextcpu < ncpus) { 969 lwkt_forwardmsg(rtable_portfn(nextcpu), &msg->lmsg); 970 } else { 971 lwkt_replymsg(&msg->lmsg, 0); 972 } 973 return (0); 974 } 975 976 #endif 977 978 #endif 979 980 /* 981 * Free an arp entry. If the arp entry is actively referenced or represents 982 * a static entry we only clear it back to an unresolved state, otherwise 983 * we destroy the entry entirely. 984 * 985 * Note that static entries are created when route add ... -interface is used 986 * to create an interface route to a (direct) destination. 987 */ 988 static void 989 arptfree(struct llinfo_arp *la) 990 { 991 struct rtentry *rt = la->la_rt; 992 struct sockaddr_dl *sdl; 993 994 if (rt == NULL) 995 panic("arptfree"); 996 sdl = SDL(rt->rt_gateway); 997 if (sdl != NULL && 998 ((rt->rt_refcnt > 0 && sdl->sdl_family == AF_LINK) || 999 (rt->rt_flags & RTF_STATIC))) { 1000 sdl->sdl_alen = 0; 1001 la->la_preempt = la->la_asked = 0; 1002 rt->rt_flags &= ~RTF_REJECT; 1003 return; 1004 } 1005 rtrequest(RTM_DELETE, rt_key(rt), NULL, rt_mask(rt), 0, NULL); 1006 } 1007 1008 /* 1009 * Lookup or enter a new address in arptab. 1010 */ 1011 static struct llinfo_arp * 1012 arplookup(in_addr_t addr, boolean_t create, boolean_t proxy) 1013 { 1014 struct rtentry *rt; 1015 struct sockaddr_inarp sin = { sizeof sin, AF_INET }; 1016 const char *why = NULL; 1017 1018 sin.sin_addr.s_addr = addr; 1019 sin.sin_other = proxy ? SIN_PROXY : 0; 1020 if (create) 1021 rt = rtlookup((struct sockaddr *)&sin); 1022 else 1023 rt = rtpurelookup((struct sockaddr *)&sin); 1024 if (rt == NULL) 1025 return (NULL); 1026 rt->rt_refcnt--; 1027 1028 if (rt->rt_flags & RTF_GATEWAY) 1029 why = "host is not on local network"; 1030 else if (!(rt->rt_flags & RTF_LLINFO)) 1031 why = "could not allocate llinfo"; 1032 else if (rt->rt_gateway->sa_family != AF_LINK) 1033 why = "gateway route is not ours"; 1034 1035 if (why) { 1036 if (create) { 1037 log(LOG_DEBUG, "arplookup %s failed: %s\n", 1038 inet_ntoa(sin.sin_addr), why); 1039 } 1040 if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_WASCLONED)) { 1041 /* No references to this route. Purge it. */ 1042 rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway, 1043 rt_mask(rt), rt->rt_flags, NULL); 1044 } 1045 return (NULL); 1046 } 1047 return (rt->rt_llinfo); 1048 } 1049 1050 void 1051 arp_ifinit(struct ifnet *ifp, struct ifaddr *ifa) 1052 { 1053 if (IA_SIN(ifa)->sin_addr.s_addr != INADDR_ANY) 1054 arprequest(ifp, &IA_SIN(ifa)->sin_addr, &IA_SIN(ifa)->sin_addr, 1055 IF_LLADDR(ifp)); 1056 ifa->ifa_rtrequest = arp_rtrequest; 1057 ifa->ifa_flags |= RTF_CLONING; 1058 } 1059 1060 static void 1061 arp_init(void) 1062 { 1063 int cpu; 1064 1065 for (cpu = 0; cpu < ncpus2; cpu++) 1066 LIST_INIT(&llinfo_arp_list[cpu]); 1067 netisr_register(NETISR_ARP, cpu0_portfn, arpintr); 1068 } 1069 1070 SYSINIT(arp, SI_SUB_PROTO_DOMAIN, SI_ORDER_ANY, arp_init, 0); 1071