1 /*- 2 * Copyright (c) 1982, 1989, 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_ethersubr.c 8.1 (Berkeley) 6/10/93 30 * $FreeBSD$ 31 */ 32 33 #include "opt_atalk.h" 34 #include "opt_inet.h" 35 #include "opt_inet6.h" 36 #include "opt_ipx.h" 37 #include "opt_mac.h" 38 #include "opt_netgraph.h" 39 #include "opt_carp.h" 40 41 #include <sys/param.h> 42 #include <sys/systm.h> 43 #include <sys/kernel.h> 44 #include <sys/mac.h> 45 #include <sys/malloc.h> 46 #include <sys/module.h> 47 #include <sys/mbuf.h> 48 #include <sys/random.h> 49 #include <sys/socket.h> 50 #include <sys/sockio.h> 51 #include <sys/sysctl.h> 52 53 #include <net/if.h> 54 #include <net/if_arp.h> 55 #include <net/netisr.h> 56 #include <net/route.h> 57 #include <net/if_llc.h> 58 #include <net/if_dl.h> 59 #include <net/if_types.h> 60 #include <net/bpf.h> 61 #include <net/ethernet.h> 62 #include <net/if_vlan_var.h> 63 64 #if defined(INET) || defined(INET6) 65 #include <netinet/in.h> 66 #include <netinet/in_var.h> 67 #include <netinet/if_ether.h> 68 #include <netinet/ip_fw.h> 69 #include <netinet/ip_dummynet.h> 70 #endif 71 #ifdef INET6 72 #include <netinet6/nd6.h> 73 #endif 74 75 #ifdef DEV_CARP 76 #include <netinet/ip_carp.h> 77 #endif 78 79 #ifdef IPX 80 #include <netipx/ipx.h> 81 #include <netipx/ipx_if.h> 82 #endif 83 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m); 84 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, 85 struct sockaddr *dst, short *tp, int *hlen); 86 87 #ifdef NETATALK 88 #include <netatalk/at.h> 89 #include <netatalk/at_var.h> 90 #include <netatalk/at_extern.h> 91 92 #define llc_snap_org_code llc_un.type_snap.org_code 93 #define llc_snap_ether_type llc_un.type_snap.ether_type 94 95 extern u_char at_org_code[3]; 96 extern u_char aarp_org_code[3]; 97 #endif /* NETATALK */ 98 99 /* netgraph node hooks for ng_ether(4) */ 100 void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp); 101 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m); 102 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp); 103 void (*ng_ether_attach_p)(struct ifnet *ifp); 104 void (*ng_ether_detach_p)(struct ifnet *ifp); 105 106 void (*vlan_input_p)(struct ifnet *, struct mbuf *); 107 108 /* bridge support */ 109 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *); 110 int (*bridge_output_p)(struct ifnet *, struct mbuf *, 111 struct sockaddr *, struct rtentry *); 112 void (*bridge_dn_p)(struct mbuf *, struct ifnet *); 113 114 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] = 115 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 116 117 static int ether_resolvemulti(struct ifnet *, struct sockaddr **, 118 struct sockaddr *); 119 120 /* XXX: should be in an arp support file, not here */ 121 MALLOC_DEFINE(M_ARPCOM, "arpcom", "802.* interface internals"); 122 123 #define senderr(e) do { error = (e); goto bad;} while (0) 124 125 #if defined(INET) || defined(INET6) 126 int 127 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 128 struct ip_fw **rule, int shared); 129 static int ether_ipfw; 130 #endif 131 132 /* 133 * Ethernet output routine. 134 * Encapsulate a packet of type family for the local net. 135 * Use trailer local net encapsulation if enough data in first 136 * packet leaves a multiple of 512 bytes of data in remainder. 137 * Assumes that ifp is actually pointer to arpcom structure. 138 */ 139 int 140 ether_output(struct ifnet *ifp, struct mbuf *m, 141 struct sockaddr *dst, struct rtentry *rt0) 142 { 143 short type; 144 int error, hdrcmplt = 0; 145 u_char esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN]; 146 struct ether_header *eh; 147 int loop_copy = 0; 148 int hlen; /* link layer header length */ 149 150 #ifdef MAC 151 error = mac_check_ifnet_transmit(ifp, m); 152 if (error) 153 senderr(error); 154 #endif 155 156 if (ifp->if_flags & IFF_MONITOR) 157 senderr(ENETDOWN); 158 if (!((ifp->if_flags & IFF_UP) && 159 (ifp->if_drv_flags & IFF_DRV_RUNNING))) 160 senderr(ENETDOWN); 161 162 hlen = ETHER_HDR_LEN; 163 switch (dst->sa_family) { 164 #ifdef INET 165 case AF_INET: 166 error = arpresolve(ifp, rt0, m, dst, edst); 167 if (error) 168 return (error == EWOULDBLOCK ? 0 : error); 169 type = htons(ETHERTYPE_IP); 170 break; 171 case AF_ARP: 172 { 173 struct arphdr *ah; 174 ah = mtod(m, struct arphdr *); 175 ah->ar_hrd = htons(ARPHRD_ETHER); 176 177 loop_copy = -1; /* if this is for us, don't do it */ 178 179 switch(ntohs(ah->ar_op)) { 180 case ARPOP_REVREQUEST: 181 case ARPOP_REVREPLY: 182 type = htons(ETHERTYPE_REVARP); 183 break; 184 case ARPOP_REQUEST: 185 case ARPOP_REPLY: 186 default: 187 type = htons(ETHERTYPE_ARP); 188 break; 189 } 190 191 if (m->m_flags & M_BCAST) 192 bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN); 193 else 194 bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN); 195 196 } 197 break; 198 #endif 199 #ifdef INET6 200 case AF_INET6: 201 error = nd6_storelladdr(ifp, rt0, m, dst, (u_char *)edst); 202 if (error) 203 return error; 204 type = htons(ETHERTYPE_IPV6); 205 break; 206 #endif 207 #ifdef IPX 208 case AF_IPX: 209 if (ef_outputp) { 210 error = ef_outputp(ifp, &m, dst, &type, &hlen); 211 if (error) 212 goto bad; 213 } else 214 type = htons(ETHERTYPE_IPX); 215 bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host), 216 (caddr_t)edst, sizeof (edst)); 217 break; 218 #endif 219 #ifdef NETATALK 220 case AF_APPLETALK: 221 { 222 struct at_ifaddr *aa; 223 224 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) 225 senderr(EHOSTUNREACH); /* XXX */ 226 if (!aarpresolve(ifp, m, (struct sockaddr_at *)dst, edst)) 227 return (0); 228 /* 229 * In the phase 2 case, need to prepend an mbuf for the llc header. 230 */ 231 if ( aa->aa_flags & AFA_PHASE2 ) { 232 struct llc llc; 233 234 M_PREPEND(m, LLC_SNAPFRAMELEN, M_DONTWAIT); 235 if (m == NULL) 236 senderr(ENOBUFS); 237 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 238 llc.llc_control = LLC_UI; 239 bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code)); 240 llc.llc_snap_ether_type = htons( ETHERTYPE_AT ); 241 bcopy(&llc, mtod(m, caddr_t), LLC_SNAPFRAMELEN); 242 type = htons(m->m_pkthdr.len); 243 hlen = LLC_SNAPFRAMELEN + ETHER_HDR_LEN; 244 } else { 245 type = htons(ETHERTYPE_AT); 246 } 247 break; 248 } 249 #endif /* NETATALK */ 250 251 case pseudo_AF_HDRCMPLT: 252 hdrcmplt = 1; 253 eh = (struct ether_header *)dst->sa_data; 254 (void)memcpy(esrc, eh->ether_shost, sizeof (esrc)); 255 /* FALLTHROUGH */ 256 257 case AF_UNSPEC: 258 loop_copy = -1; /* if this is for us, don't do it */ 259 eh = (struct ether_header *)dst->sa_data; 260 (void)memcpy(edst, eh->ether_dhost, sizeof (edst)); 261 type = eh->ether_type; 262 break; 263 264 default: 265 if_printf(ifp, "can't handle af%d\n", dst->sa_family); 266 senderr(EAFNOSUPPORT); 267 } 268 269 /* 270 * Add local net header. If no space in first mbuf, 271 * allocate another. 272 */ 273 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 274 if (m == NULL) 275 senderr(ENOBUFS); 276 eh = mtod(m, struct ether_header *); 277 (void)memcpy(&eh->ether_type, &type, 278 sizeof(eh->ether_type)); 279 (void)memcpy(eh->ether_dhost, edst, sizeof (edst)); 280 if (hdrcmplt) 281 (void)memcpy(eh->ether_shost, esrc, 282 sizeof(eh->ether_shost)); 283 else 284 (void)memcpy(eh->ether_shost, IFP2ENADDR(ifp), 285 sizeof(eh->ether_shost)); 286 287 /* 288 * Bridges require special output handling. 289 */ 290 if (ifp->if_bridge) { 291 KASSERT(bridge_output_p != NULL, 292 ("ether_input: if_bridge not loaded!")); 293 return ((*bridge_output_p)(ifp, m, NULL, NULL)); 294 } 295 296 /* 297 * If a simplex interface, and the packet is being sent to our 298 * Ethernet address or a broadcast address, loopback a copy. 299 * XXX To make a simplex device behave exactly like a duplex 300 * device, we should copy in the case of sending to our own 301 * ethernet address (thus letting the original actually appear 302 * on the wire). However, we don't do that here for security 303 * reasons and compatibility with the original behavior. 304 */ 305 if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1) && 306 m_tag_find(m, PACKET_TAG_PF_ROUTED, NULL) == NULL) { 307 int csum_flags = 0; 308 309 if (m->m_pkthdr.csum_flags & CSUM_IP) 310 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID); 311 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) 312 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR); 313 314 if ((m->m_flags & M_BCAST) || (loop_copy > 0)) { 315 struct mbuf *n; 316 317 if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) { 318 n->m_pkthdr.csum_flags |= csum_flags; 319 if (csum_flags & CSUM_DATA_VALID) 320 n->m_pkthdr.csum_data = 0xffff; 321 (void)if_simloop(ifp, n, dst->sa_family, hlen); 322 } else 323 ifp->if_iqdrops++; 324 } else if (bcmp(eh->ether_dhost, eh->ether_shost, 325 ETHER_ADDR_LEN) == 0) { 326 m->m_pkthdr.csum_flags |= csum_flags; 327 if (csum_flags & CSUM_DATA_VALID) 328 m->m_pkthdr.csum_data = 0xffff; 329 (void) if_simloop(ifp, m, dst->sa_family, hlen); 330 return (0); /* XXX */ 331 } 332 } 333 334 #ifdef DEV_CARP 335 if (ifp->if_carp && 336 (error = carp_output(ifp, m, dst, NULL))) 337 goto bad; 338 #endif 339 340 /* Handle ng_ether(4) processing, if any */ 341 if (IFP2AC(ifp)->ac_netgraph != NULL) { 342 KASSERT(ng_ether_output_p != NULL, 343 ("ng_ether_output_p is NULL")); 344 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) { 345 bad: if (m != NULL) 346 m_freem(m); 347 return (error); 348 } 349 if (m == NULL) 350 return (0); 351 } 352 353 /* Continue with link-layer output */ 354 return ether_output_frame(ifp, m); 355 } 356 357 /* 358 * Ethernet link layer output routine to send a raw frame to the device. 359 * 360 * This assumes that the 14 byte Ethernet header is present and contiguous 361 * in the first mbuf (if BRIDGE'ing). 362 */ 363 int 364 ether_output_frame(struct ifnet *ifp, struct mbuf *m) 365 { 366 #if defined(INET) || defined(INET6) 367 struct ip_fw *rule = ip_dn_claim_rule(m); 368 #else 369 void *rule = NULL; 370 #endif 371 int error; 372 373 #if defined(INET) || defined(INET6) 374 if (IPFW_LOADED && ether_ipfw != 0) { 375 if (ether_ipfw_chk(&m, ifp, &rule, 0) == 0) { 376 if (m) { 377 m_freem(m); 378 return EACCES; /* pkt dropped */ 379 } else 380 return 0; /* consumed e.g. in a pipe */ 381 } 382 } 383 #endif 384 385 /* 386 * Queue message on interface, update output statistics if 387 * successful, and start output if interface not yet active. 388 */ 389 IFQ_HANDOFF(ifp, m, error); 390 return (error); 391 } 392 393 #if defined(INET) || defined(INET6) 394 /* 395 * ipfw processing for ethernet packets (in and out). 396 * The second parameter is NULL from ether_demux, and ifp from 397 * ether_output_frame. 398 */ 399 int 400 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 401 struct ip_fw **rule, int shared) 402 { 403 struct ether_header *eh; 404 struct ether_header save_eh; 405 struct mbuf *m; 406 int i; 407 struct ip_fw_args args; 408 409 if (*rule != NULL && fw_one_pass) 410 return 1; /* dummynet packet, already partially processed */ 411 412 /* 413 * I need some amt of data to be contiguous, and in case others need 414 * the packet (shared==1) also better be in the first mbuf. 415 */ 416 m = *m0; 417 i = min( m->m_pkthdr.len, max_protohdr); 418 if ( shared || m->m_len < i) { 419 m = m_pullup(m, i); 420 if (m == NULL) { 421 *m0 = m; 422 return 0; 423 } 424 } 425 eh = mtod(m, struct ether_header *); 426 save_eh = *eh; /* save copy for restore below */ 427 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */ 428 429 args.m = m; /* the packet we are looking at */ 430 args.oif = dst; /* destination, if any */ 431 args.rule = *rule; /* matching rule to restart */ 432 args.next_hop = NULL; /* we do not support forward yet */ 433 args.eh = &save_eh; /* MAC header for bridged/MAC packets */ 434 i = ip_fw_chk_ptr(&args); 435 m = args.m; 436 if (m != NULL) { 437 /* 438 * Restore Ethernet header, as needed, in case the 439 * mbuf chain was replaced by ipfw. 440 */ 441 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 442 if (m == NULL) { 443 *m0 = m; 444 return 0; 445 } 446 if (eh != mtod(m, struct ether_header *)) 447 bcopy(&save_eh, mtod(m, struct ether_header *), 448 ETHER_HDR_LEN); 449 } 450 *m0 = m; 451 *rule = args.rule; 452 453 if (i == IP_FW_DENY) /* drop */ 454 return 0; 455 456 KASSERT(m != NULL, ("ether_ipfw_chk: m is NULL")); 457 458 if (i == IP_FW_PASS) /* a PASS rule. */ 459 return 1; 460 461 if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) { 462 /* 463 * Pass the pkt to dummynet, which consumes it. 464 * If shared, make a copy and keep the original. 465 */ 466 if (shared) { 467 m = m_copypacket(m, M_DONTWAIT); 468 if (m == NULL) 469 return 0; 470 } else { 471 /* 472 * Pass the original to dummynet and 473 * nothing back to the caller 474 */ 475 *m0 = NULL ; 476 } 477 ip_dn_io_ptr(m, dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args); 478 return 0; 479 } 480 /* 481 * XXX at some point add support for divert/forward actions. 482 * If none of the above matches, we have to drop the pkt. 483 */ 484 return 0; 485 } 486 #endif 487 488 /* 489 * Process a received Ethernet packet; the packet is in the 490 * mbuf chain m with the ethernet header at the front. 491 */ 492 static void 493 ether_input(struct ifnet *ifp, struct mbuf *m) 494 { 495 struct ether_header *eh; 496 u_short etype; 497 498 /* 499 * Do consistency checks to verify assumptions 500 * made by code past this point. 501 */ 502 if ((m->m_flags & M_PKTHDR) == 0) { 503 if_printf(ifp, "discard frame w/o packet header\n"); 504 ifp->if_ierrors++; 505 m_freem(m); 506 return; 507 } 508 if (m->m_len < ETHER_HDR_LEN) { 509 /* XXX maybe should pullup? */ 510 if_printf(ifp, "discard frame w/o leading ethernet " 511 "header (len %u pkt len %u)\n", 512 m->m_len, m->m_pkthdr.len); 513 ifp->if_ierrors++; 514 m_freem(m); 515 return; 516 } 517 eh = mtod(m, struct ether_header *); 518 etype = ntohs(eh->ether_type); 519 if (m->m_pkthdr.len > 520 ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) { 521 if_printf(ifp, "discard oversize frame " 522 "(ether type %x flags %x len %u > max %lu)\n", 523 etype, m->m_flags, m->m_pkthdr.len, 524 ETHER_MAX_FRAME(ifp, etype, 525 m->m_flags & M_HASFCS)); 526 ifp->if_ierrors++; 527 m_freem(m); 528 return; 529 } 530 if (m->m_pkthdr.rcvif == NULL) { 531 if_printf(ifp, "discard frame w/o interface pointer\n"); 532 ifp->if_ierrors++; 533 m_freem(m); 534 return; 535 } 536 #ifdef DIAGNOSTIC 537 if (m->m_pkthdr.rcvif != ifp) { 538 if_printf(ifp, "Warning, frame marked as received on %s\n", 539 m->m_pkthdr.rcvif->if_xname); 540 } 541 #endif 542 543 #ifdef MAC 544 /* 545 * Tag the mbuf with an appropriate MAC label before any other 546 * consumers can get to it. 547 */ 548 mac_create_mbuf_from_ifnet(ifp, m); 549 #endif 550 551 /* 552 * Give bpf a chance at the packet. 553 */ 554 BPF_MTAP(ifp, m); 555 556 if (ifp->if_flags & IFF_MONITOR) { 557 /* 558 * Interface marked for monitoring; discard packet. 559 */ 560 m_freem(m); 561 return; 562 } 563 564 /* If the CRC is still on the packet, trim it off. */ 565 if (m->m_flags & M_HASFCS) { 566 m_adj(m, -ETHER_CRC_LEN); 567 m->m_flags &= ~M_HASFCS; 568 } 569 570 ifp->if_ibytes += m->m_pkthdr.len; 571 572 /* Handle ng_ether(4) processing, if any */ 573 if (IFP2AC(ifp)->ac_netgraph != NULL) { 574 KASSERT(ng_ether_input_p != NULL, 575 ("ng_ether_input_p is NULL")); 576 (*ng_ether_input_p)(ifp, &m); 577 if (m == NULL) 578 return; 579 } 580 581 /* 582 * Tap the packet off here for a bridge. bridge_input() 583 * will return NULL if it has consumed the packet, otherwise 584 * it gets processed as normal. Note that bridge_input() 585 * will always return the original packet if we need to 586 * process it locally. 587 */ 588 if (ifp->if_bridge) { 589 KASSERT(bridge_input_p != NULL, 590 ("ether_input: if_bridge not loaded!")); 591 592 /* Mark the packet as broadcast or multicast. This is also set 593 * further down the code in ether_demux() but since the bridge 594 * input routine rarely returns a mbuf for further processing, 595 * it is an acceptable duplication. 596 */ 597 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 598 if (bcmp(etherbroadcastaddr, eh->ether_dhost, 599 sizeof(etherbroadcastaddr)) == 0) 600 m->m_flags |= M_BCAST; 601 else 602 m->m_flags |= M_MCAST; 603 } 604 605 m = (*bridge_input_p)(ifp, m); 606 if (m == NULL) 607 return; 608 /* 609 * Bridge has determined that the packet is for us. 610 * Update our interface pointer -- we may have had 611 * to "bridge" the packet locally. 612 */ 613 ifp = m->m_pkthdr.rcvif; 614 } 615 616 /* First chunk of an mbuf contains good entropy */ 617 if (harvest.ethernet) 618 random_harvest(m, 16, 3, 0, RANDOM_NET); 619 ether_demux(ifp, m); 620 } 621 622 /* 623 * Upper layer processing for a received Ethernet packet. 624 */ 625 void 626 ether_demux(struct ifnet *ifp, struct mbuf *m) 627 { 628 struct ether_header *eh; 629 int isr; 630 u_short ether_type; 631 #if defined(NETATALK) 632 struct llc *l; 633 #endif 634 #if defined(INET) || defined(INET6) 635 struct ip_fw *rule = ip_dn_claim_rule(m); 636 #endif 637 638 KASSERT(ifp != NULL, ("ether_demux: NULL interface pointer")); 639 640 eh = mtod(m, struct ether_header *); 641 ether_type = ntohs(eh->ether_type); 642 643 #if defined(INET) || defined(INET6) 644 if (rule) /* packet was already bridged */ 645 goto post_stats; 646 #endif 647 648 if (!(ifp->if_bridge) && 649 !((ether_type == ETHERTYPE_VLAN || m->m_flags & M_VLANTAG) && 650 ifp->if_nvlans > 0)) { 651 #ifdef DEV_CARP 652 /* 653 * XXX: Okay, we need to call carp_forus() and - if it is for 654 * us jump over code that does the normal check 655 * "IFP2ENADDR(ifp) == ether_dhost". The check sequence is a bit 656 * different from OpenBSD, so we jump over as few code as 657 * possible, to catch _all_ sanity checks. This needs 658 * evaluation, to see if the carp ether_dhost values break any 659 * of these checks! 660 */ 661 if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost)) 662 goto pre_stats; 663 #endif 664 /* 665 * Discard packet if upper layers shouldn't see it because it 666 * was unicast to a different Ethernet address. If the driver 667 * is working properly, then this situation can only happen 668 * when the interface is in promiscuous mode. 669 * 670 * If VLANs are active, and this packet has a VLAN tag, do 671 * not drop it here but pass it on to the VLAN layer, to 672 * give them a chance to consider it as well (e. g. in case 673 * bridging is only active on a VLAN). They will drop it if 674 * it's undesired. 675 */ 676 if ((ifp->if_flags & IFF_PROMISC) != 0 677 && !ETHER_IS_MULTICAST(eh->ether_dhost) 678 && bcmp(eh->ether_dhost, 679 IFP2ENADDR(ifp), ETHER_ADDR_LEN) != 0 680 && (ifp->if_flags & IFF_PPROMISC) == 0) { 681 m_freem(m); 682 return; 683 } 684 } 685 686 #ifdef DEV_CARP 687 pre_stats: 688 #endif 689 /* Discard packet if interface is not up */ 690 if ((ifp->if_flags & IFF_UP) == 0) { 691 m_freem(m); 692 return; 693 } 694 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 695 if (bcmp(etherbroadcastaddr, eh->ether_dhost, 696 sizeof(etherbroadcastaddr)) == 0) 697 m->m_flags |= M_BCAST; 698 else 699 m->m_flags |= M_MCAST; 700 } 701 if (m->m_flags & (M_BCAST|M_MCAST)) 702 ifp->if_imcasts++; 703 704 #if defined(INET) || defined(INET6) 705 post_stats: 706 if (IPFW_LOADED && ether_ipfw != 0) { 707 if (ether_ipfw_chk(&m, NULL, &rule, 0) == 0) { 708 if (m) 709 m_freem(m); 710 return; 711 } 712 } 713 #endif 714 715 /* 716 * Check to see if the device performed the VLAN decapsulation and 717 * provided us with the tag. 718 */ 719 if (m->m_flags & M_VLANTAG) { 720 /* 721 * If no VLANs are configured, drop. 722 */ 723 if (ifp->if_nvlans == 0) { 724 ifp->if_noproto++; 725 m_freem(m); 726 return; 727 } 728 /* 729 * vlan_input() will either recursively call ether_input() 730 * or drop the packet. 731 */ 732 KASSERT(vlan_input_p != NULL,("ether_input: VLAN not loaded!")); 733 (*vlan_input_p)(ifp, m); 734 return; 735 } 736 737 /* 738 * Handle protocols that expect to have the Ethernet header 739 * (and possibly FCS) intact. 740 */ 741 switch (ether_type) { 742 case ETHERTYPE_VLAN: 743 if (ifp->if_nvlans != 0) { 744 KASSERT(vlan_input_p,("ether_input: VLAN not loaded!")); 745 (*vlan_input_p)(ifp, m); 746 } else { 747 ifp->if_noproto++; 748 m_freem(m); 749 } 750 return; 751 } 752 753 /* Strip off Ethernet header. */ 754 m_adj(m, ETHER_HDR_LEN); 755 756 /* If the CRC is still on the packet, trim it off. */ 757 if (m->m_flags & M_HASFCS) { 758 m_adj(m, -ETHER_CRC_LEN); 759 m->m_flags &= ~M_HASFCS; 760 } 761 762 switch (ether_type) { 763 #ifdef INET 764 case ETHERTYPE_IP: 765 if (ip_fastforward(m)) 766 return; 767 isr = NETISR_IP; 768 break; 769 770 case ETHERTYPE_ARP: 771 if (ifp->if_flags & IFF_NOARP) { 772 /* Discard packet if ARP is disabled on interface */ 773 m_freem(m); 774 return; 775 } 776 isr = NETISR_ARP; 777 break; 778 #endif 779 #ifdef IPX 780 case ETHERTYPE_IPX: 781 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 782 return; 783 isr = NETISR_IPX; 784 break; 785 #endif 786 #ifdef INET6 787 case ETHERTYPE_IPV6: 788 isr = NETISR_IPV6; 789 break; 790 #endif 791 #ifdef NETATALK 792 case ETHERTYPE_AT: 793 isr = NETISR_ATALK1; 794 break; 795 case ETHERTYPE_AARP: 796 isr = NETISR_AARP; 797 break; 798 #endif /* NETATALK */ 799 default: 800 #ifdef IPX 801 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 802 return; 803 #endif /* IPX */ 804 #if defined(NETATALK) 805 if (ether_type > ETHERMTU) 806 goto discard; 807 l = mtod(m, struct llc *); 808 if (l->llc_dsap == LLC_SNAP_LSAP && 809 l->llc_ssap == LLC_SNAP_LSAP && 810 l->llc_control == LLC_UI) { 811 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code, 812 sizeof(at_org_code)) == 0 && 813 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) { 814 m_adj(m, LLC_SNAPFRAMELEN); 815 isr = NETISR_ATALK2; 816 break; 817 } 818 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code, 819 sizeof(aarp_org_code)) == 0 && 820 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) { 821 m_adj(m, LLC_SNAPFRAMELEN); 822 isr = NETISR_AARP; 823 break; 824 } 825 } 826 #endif /* NETATALK */ 827 goto discard; 828 } 829 netisr_dispatch(isr, m); 830 return; 831 832 discard: 833 /* 834 * Packet is to be discarded. If netgraph is present, 835 * hand the packet to it for last chance processing; 836 * otherwise dispose of it. 837 */ 838 if (IFP2AC(ifp)->ac_netgraph != NULL) { 839 KASSERT(ng_ether_input_orphan_p != NULL, 840 ("ng_ether_input_orphan_p is NULL")); 841 /* 842 * Put back the ethernet header so netgraph has a 843 * consistent view of inbound packets. 844 */ 845 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 846 (*ng_ether_input_orphan_p)(ifp, m); 847 return; 848 } 849 m_freem(m); 850 } 851 852 /* 853 * Convert Ethernet address to printable (loggable) representation. 854 * This routine is for compatibility; it's better to just use 855 * 856 * printf("%6D", <pointer to address>, ":"); 857 * 858 * since there's no static buffer involved. 859 */ 860 char * 861 ether_sprintf(const u_char *ap) 862 { 863 static char etherbuf[18]; 864 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":"); 865 return (etherbuf); 866 } 867 868 /* 869 * Perform common duties while attaching to interface list 870 */ 871 void 872 ether_ifattach(struct ifnet *ifp, const u_int8_t *llc) 873 { 874 int i; 875 struct ifaddr *ifa; 876 struct sockaddr_dl *sdl; 877 878 ifp->if_addrlen = ETHER_ADDR_LEN; 879 ifp->if_hdrlen = ETHER_HDR_LEN; 880 if_attach(ifp); 881 ifp->if_mtu = ETHERMTU; 882 ifp->if_output = ether_output; 883 ifp->if_input = ether_input; 884 ifp->if_resolvemulti = ether_resolvemulti; 885 if (ifp->if_baudrate == 0) 886 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 887 ifp->if_broadcastaddr = etherbroadcastaddr; 888 889 ifa = ifaddr_byindex(ifp->if_index); 890 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__)); 891 sdl = (struct sockaddr_dl *)ifa->ifa_addr; 892 sdl->sdl_type = IFT_ETHER; 893 sdl->sdl_alen = ifp->if_addrlen; 894 bcopy(llc, LLADDR(sdl), ifp->if_addrlen); 895 /* 896 * XXX: This doesn't belong here; we do it until 897 * XXX: all drivers are cleaned up 898 */ 899 if (llc != IFP2ENADDR(ifp)) 900 bcopy(llc, IFP2ENADDR(ifp), ifp->if_addrlen); 901 902 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN); 903 if (ng_ether_attach_p != NULL) 904 (*ng_ether_attach_p)(ifp); 905 906 /* Announce Ethernet MAC address if non-zero. */ 907 for (i = 0; i < ifp->if_addrlen; i++) 908 if (llc[i] != 0) 909 break; 910 if (i != ifp->if_addrlen) 911 if_printf(ifp, "Ethernet address: %6D\n", llc, ":"); 912 if (debug_mpsafenet && (ifp->if_flags & IFF_NEEDSGIANT) != 0) 913 if_printf(ifp, "if_start running deferred for Giant\n"); 914 } 915 916 /* 917 * Perform common duties while detaching an Ethernet interface 918 */ 919 void 920 ether_ifdetach(struct ifnet *ifp) 921 { 922 if (IFP2AC(ifp)->ac_netgraph != NULL) { 923 KASSERT(ng_ether_detach_p != NULL, 924 ("ng_ether_detach_p is NULL")); 925 (*ng_ether_detach_p)(ifp); 926 } 927 bpfdetach(ifp); 928 if_detach(ifp); 929 } 930 931 SYSCTL_DECL(_net_link); 932 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet"); 933 #if defined(INET) || defined(INET6) 934 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW, 935 ðer_ipfw,0,"Pass ether pkts through firewall"); 936 #endif 937 938 #if 0 939 /* 940 * This is for reference. We have a table-driven version 941 * of the little-endian crc32 generator, which is faster 942 * than the double-loop. 943 */ 944 uint32_t 945 ether_crc32_le(const uint8_t *buf, size_t len) 946 { 947 size_t i; 948 uint32_t crc; 949 int bit; 950 uint8_t data; 951 952 crc = 0xffffffff; /* initial value */ 953 954 for (i = 0; i < len; i++) { 955 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) 956 carry = (crc ^ data) & 1; 957 crc >>= 1; 958 if (carry) 959 crc = (crc ^ ETHER_CRC_POLY_LE); 960 } 961 962 return (crc); 963 } 964 #else 965 uint32_t 966 ether_crc32_le(const uint8_t *buf, size_t len) 967 { 968 static const uint32_t crctab[] = { 969 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 970 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 971 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 972 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 973 }; 974 size_t i; 975 uint32_t crc; 976 977 crc = 0xffffffff; /* initial value */ 978 979 for (i = 0; i < len; i++) { 980 crc ^= buf[i]; 981 crc = (crc >> 4) ^ crctab[crc & 0xf]; 982 crc = (crc >> 4) ^ crctab[crc & 0xf]; 983 } 984 985 return (crc); 986 } 987 #endif 988 989 uint32_t 990 ether_crc32_be(const uint8_t *buf, size_t len) 991 { 992 size_t i; 993 uint32_t crc, carry; 994 int bit; 995 uint8_t data; 996 997 crc = 0xffffffff; /* initial value */ 998 999 for (i = 0; i < len; i++) { 1000 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) { 1001 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01); 1002 crc <<= 1; 1003 if (carry) 1004 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1005 } 1006 } 1007 1008 return (crc); 1009 } 1010 1011 int 1012 ether_ioctl(struct ifnet *ifp, int command, caddr_t data) 1013 { 1014 struct ifaddr *ifa = (struct ifaddr *) data; 1015 struct ifreq *ifr = (struct ifreq *) data; 1016 int error = 0; 1017 1018 switch (command) { 1019 case SIOCSIFADDR: 1020 ifp->if_flags |= IFF_UP; 1021 1022 switch (ifa->ifa_addr->sa_family) { 1023 #ifdef INET 1024 case AF_INET: 1025 ifp->if_init(ifp->if_softc); /* before arpwhohas */ 1026 arp_ifinit(ifp, ifa); 1027 break; 1028 #endif 1029 #ifdef IPX 1030 /* 1031 * XXX - This code is probably wrong 1032 */ 1033 case AF_IPX: 1034 { 1035 struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr); 1036 1037 if (ipx_nullhost(*ina)) 1038 ina->x_host = 1039 *(union ipx_host *) 1040 IFP2ENADDR(ifp); 1041 else { 1042 bcopy((caddr_t) ina->x_host.c_host, 1043 (caddr_t) IFP2ENADDR(ifp), 1044 ETHER_ADDR_LEN); 1045 } 1046 1047 /* 1048 * Set new address 1049 */ 1050 ifp->if_init(ifp->if_softc); 1051 break; 1052 } 1053 #endif 1054 default: 1055 ifp->if_init(ifp->if_softc); 1056 break; 1057 } 1058 break; 1059 1060 case SIOCGIFADDR: 1061 { 1062 struct sockaddr *sa; 1063 1064 sa = (struct sockaddr *) & ifr->ifr_data; 1065 bcopy(IFP2ENADDR(ifp), 1066 (caddr_t) sa->sa_data, ETHER_ADDR_LEN); 1067 } 1068 break; 1069 1070 case SIOCSIFMTU: 1071 /* 1072 * Set the interface MTU. 1073 */ 1074 if (ifr->ifr_mtu > ETHERMTU) { 1075 error = EINVAL; 1076 } else { 1077 ifp->if_mtu = ifr->ifr_mtu; 1078 } 1079 break; 1080 default: 1081 error = EINVAL; /* XXX netbsd has ENOTTY??? */ 1082 break; 1083 } 1084 return (error); 1085 } 1086 1087 static int 1088 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa, 1089 struct sockaddr *sa) 1090 { 1091 struct sockaddr_dl *sdl; 1092 #ifdef INET 1093 struct sockaddr_in *sin; 1094 #endif 1095 #ifdef INET6 1096 struct sockaddr_in6 *sin6; 1097 #endif 1098 u_char *e_addr; 1099 1100 switch(sa->sa_family) { 1101 case AF_LINK: 1102 /* 1103 * No mapping needed. Just check that it's a valid MC address. 1104 */ 1105 sdl = (struct sockaddr_dl *)sa; 1106 e_addr = LLADDR(sdl); 1107 if (!ETHER_IS_MULTICAST(e_addr)) 1108 return EADDRNOTAVAIL; 1109 *llsa = 0; 1110 return 0; 1111 1112 #ifdef INET 1113 case AF_INET: 1114 sin = (struct sockaddr_in *)sa; 1115 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) 1116 return EADDRNOTAVAIL; 1117 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, 1118 M_NOWAIT|M_ZERO); 1119 if (sdl == NULL) 1120 return ENOMEM; 1121 sdl->sdl_len = sizeof *sdl; 1122 sdl->sdl_family = AF_LINK; 1123 sdl->sdl_index = ifp->if_index; 1124 sdl->sdl_type = IFT_ETHER; 1125 sdl->sdl_alen = ETHER_ADDR_LEN; 1126 e_addr = LLADDR(sdl); 1127 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr); 1128 *llsa = (struct sockaddr *)sdl; 1129 return 0; 1130 #endif 1131 #ifdef INET6 1132 case AF_INET6: 1133 sin6 = (struct sockaddr_in6 *)sa; 1134 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1135 /* 1136 * An IP6 address of 0 means listen to all 1137 * of the Ethernet multicast address used for IP6. 1138 * (This is used for multicast routers.) 1139 */ 1140 ifp->if_flags |= IFF_ALLMULTI; 1141 *llsa = 0; 1142 return 0; 1143 } 1144 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) 1145 return EADDRNOTAVAIL; 1146 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, 1147 M_NOWAIT|M_ZERO); 1148 if (sdl == NULL) 1149 return (ENOMEM); 1150 sdl->sdl_len = sizeof *sdl; 1151 sdl->sdl_family = AF_LINK; 1152 sdl->sdl_index = ifp->if_index; 1153 sdl->sdl_type = IFT_ETHER; 1154 sdl->sdl_alen = ETHER_ADDR_LEN; 1155 e_addr = LLADDR(sdl); 1156 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr); 1157 *llsa = (struct sockaddr *)sdl; 1158 return 0; 1159 #endif 1160 1161 default: 1162 /* 1163 * Well, the text isn't quite right, but it's the name 1164 * that counts... 1165 */ 1166 return EAFNOSUPPORT; 1167 } 1168 } 1169 1170 static void* 1171 ether_alloc(u_char type, struct ifnet *ifp) 1172 { 1173 struct arpcom *ac; 1174 1175 ac = malloc(sizeof(struct arpcom), M_ARPCOM, M_WAITOK | M_ZERO); 1176 ac->ac_ifp = ifp; 1177 1178 return (ac); 1179 } 1180 1181 static void 1182 ether_free(void *com, u_char type) 1183 { 1184 1185 free(com, M_ARPCOM); 1186 } 1187 1188 static int 1189 ether_modevent(module_t mod, int type, void *data) 1190 { 1191 1192 switch (type) { 1193 case MOD_LOAD: 1194 if_register_com_alloc(IFT_ETHER, ether_alloc, ether_free); 1195 break; 1196 case MOD_UNLOAD: 1197 if_deregister_com_alloc(IFT_ETHER); 1198 break; 1199 default: 1200 return EOPNOTSUPP; 1201 } 1202 1203 return (0); 1204 } 1205 1206 static moduledata_t ether_mod = { 1207 "ether", 1208 ether_modevent, 1209 0 1210 }; 1211 1212 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY); 1213 MODULE_VERSION(ether, 1); 1214