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/malloc.h> 45 #include <sys/module.h> 46 #include <sys/mbuf.h> 47 #include <sys/random.h> 48 #include <sys/socket.h> 49 #include <sys/sockio.h> 50 #include <sys/sysctl.h> 51 52 #include <net/if.h> 53 #include <net/if_arp.h> 54 #include <net/netisr.h> 55 #include <net/route.h> 56 #include <net/if_llc.h> 57 #include <net/if_dl.h> 58 #include <net/if_types.h> 59 #include <net/bpf.h> 60 #include <net/ethernet.h> 61 #include <net/if_bridgevar.h> 62 #include <net/if_vlan_var.h> 63 #include <net/pf_mtag.h> 64 65 #if defined(INET) || defined(INET6) 66 #include <netinet/in.h> 67 #include <netinet/in_var.h> 68 #include <netinet/if_ether.h> 69 #include <netinet/ip_fw.h> 70 #include <netinet/ip_dummynet.h> 71 #endif 72 #ifdef INET6 73 #include <netinet6/nd6.h> 74 #endif 75 76 #ifdef DEV_CARP 77 #include <netinet/ip_carp.h> 78 #endif 79 80 #ifdef IPX 81 #include <netipx/ipx.h> 82 #include <netipx/ipx_if.h> 83 #endif 84 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m); 85 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, 86 struct sockaddr *dst, short *tp, int *hlen); 87 88 #ifdef NETATALK 89 #include <netatalk/at.h> 90 #include <netatalk/at_var.h> 91 #include <netatalk/at_extern.h> 92 93 #define llc_snap_org_code llc_un.type_snap.org_code 94 #define llc_snap_ether_type llc_un.type_snap.ether_type 95 96 extern u_char at_org_code[3]; 97 extern u_char aarp_org_code[3]; 98 #endif /* NETATALK */ 99 100 #include <security/mac/mac_framework.h> 101 102 /* netgraph node hooks for ng_ether(4) */ 103 void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp); 104 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m); 105 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp); 106 void (*ng_ether_attach_p)(struct ifnet *ifp); 107 void (*ng_ether_detach_p)(struct ifnet *ifp); 108 109 void (*vlan_input_p)(struct ifnet *, struct mbuf *); 110 111 /* if_bridge(4) support */ 112 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *); 113 int (*bridge_output_p)(struct ifnet *, struct mbuf *, 114 struct sockaddr *, struct rtentry *); 115 void (*bridge_dn_p)(struct mbuf *, struct ifnet *); 116 117 /* if_lagg(4) support */ 118 struct mbuf *(*lagg_input_p)(struct ifnet *, struct mbuf *); 119 120 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] = 121 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 122 123 static int ether_resolvemulti(struct ifnet *, struct sockaddr **, 124 struct sockaddr *); 125 126 /* XXX: should be in an arp support file, not here */ 127 MALLOC_DEFINE(M_ARPCOM, "arpcom", "802.* interface internals"); 128 129 #define ETHER_IS_BROADCAST(addr) \ 130 (bcmp(etherbroadcastaddr, (addr), ETHER_ADDR_LEN) == 0) 131 132 #define senderr(e) do { error = (e); goto bad;} while (0) 133 134 #if defined(INET) || defined(INET6) 135 int 136 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 137 struct ip_fw **rule, int shared); 138 static int ether_ipfw; 139 #endif 140 141 /* 142 * Ethernet output routine. 143 * Encapsulate a packet of type family for the local net. 144 * Use trailer local net encapsulation if enough data in first 145 * packet leaves a multiple of 512 bytes of data in remainder. 146 */ 147 int 148 ether_output(struct ifnet *ifp, struct mbuf *m, 149 struct sockaddr *dst, struct rtentry *rt0) 150 { 151 short type; 152 int error, hdrcmplt = 0; 153 u_char esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN]; 154 struct ether_header *eh; 155 struct pf_mtag *t; 156 int loop_copy = 1; 157 int hlen; /* link layer header length */ 158 159 #ifdef MAC 160 error = mac_ifnet_check_transmit(ifp, m); 161 if (error) 162 senderr(error); 163 #endif 164 165 if (ifp->if_flags & IFF_MONITOR) 166 senderr(ENETDOWN); 167 if (!((ifp->if_flags & IFF_UP) && 168 (ifp->if_drv_flags & IFF_DRV_RUNNING))) 169 senderr(ENETDOWN); 170 171 hlen = ETHER_HDR_LEN; 172 switch (dst->sa_family) { 173 #ifdef INET 174 case AF_INET: 175 error = arpresolve(ifp, rt0, m, dst, edst); 176 if (error) 177 return (error == EWOULDBLOCK ? 0 : error); 178 type = htons(ETHERTYPE_IP); 179 break; 180 case AF_ARP: 181 { 182 struct arphdr *ah; 183 ah = mtod(m, struct arphdr *); 184 ah->ar_hrd = htons(ARPHRD_ETHER); 185 186 loop_copy = 0; /* if this is for us, don't do it */ 187 188 switch(ntohs(ah->ar_op)) { 189 case ARPOP_REVREQUEST: 190 case ARPOP_REVREPLY: 191 type = htons(ETHERTYPE_REVARP); 192 break; 193 case ARPOP_REQUEST: 194 case ARPOP_REPLY: 195 default: 196 type = htons(ETHERTYPE_ARP); 197 break; 198 } 199 200 if (m->m_flags & M_BCAST) 201 bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN); 202 else 203 bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN); 204 205 } 206 break; 207 #endif 208 #ifdef INET6 209 case AF_INET6: 210 error = nd6_storelladdr(ifp, rt0, m, dst, (u_char *)edst); 211 if (error) 212 return error; 213 type = htons(ETHERTYPE_IPV6); 214 break; 215 #endif 216 #ifdef IPX 217 case AF_IPX: 218 if (ef_outputp) { 219 error = ef_outputp(ifp, &m, dst, &type, &hlen); 220 if (error) 221 goto bad; 222 } else 223 type = htons(ETHERTYPE_IPX); 224 bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host), 225 (caddr_t)edst, sizeof (edst)); 226 break; 227 #endif 228 #ifdef NETATALK 229 case AF_APPLETALK: 230 { 231 struct at_ifaddr *aa; 232 233 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) 234 senderr(EHOSTUNREACH); /* XXX */ 235 if (!aarpresolve(ifp, m, (struct sockaddr_at *)dst, edst)) 236 return (0); 237 /* 238 * In the phase 2 case, need to prepend an mbuf for the llc header. 239 */ 240 if ( aa->aa_flags & AFA_PHASE2 ) { 241 struct llc llc; 242 243 M_PREPEND(m, LLC_SNAPFRAMELEN, M_DONTWAIT); 244 if (m == NULL) 245 senderr(ENOBUFS); 246 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 247 llc.llc_control = LLC_UI; 248 bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code)); 249 llc.llc_snap_ether_type = htons( ETHERTYPE_AT ); 250 bcopy(&llc, mtod(m, caddr_t), LLC_SNAPFRAMELEN); 251 type = htons(m->m_pkthdr.len); 252 hlen = LLC_SNAPFRAMELEN + ETHER_HDR_LEN; 253 } else { 254 type = htons(ETHERTYPE_AT); 255 } 256 break; 257 } 258 #endif /* NETATALK */ 259 260 case pseudo_AF_HDRCMPLT: 261 hdrcmplt = 1; 262 eh = (struct ether_header *)dst->sa_data; 263 (void)memcpy(esrc, eh->ether_shost, sizeof (esrc)); 264 /* FALLTHROUGH */ 265 266 case AF_UNSPEC: 267 loop_copy = 0; /* if this is for us, don't do it */ 268 eh = (struct ether_header *)dst->sa_data; 269 (void)memcpy(edst, eh->ether_dhost, sizeof (edst)); 270 type = eh->ether_type; 271 break; 272 273 default: 274 if_printf(ifp, "can't handle af%d\n", dst->sa_family); 275 senderr(EAFNOSUPPORT); 276 } 277 278 /* 279 * Add local net header. If no space in first mbuf, 280 * allocate another. 281 */ 282 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 283 if (m == NULL) 284 senderr(ENOBUFS); 285 eh = mtod(m, struct ether_header *); 286 (void)memcpy(&eh->ether_type, &type, 287 sizeof(eh->ether_type)); 288 (void)memcpy(eh->ether_dhost, edst, sizeof (edst)); 289 if (hdrcmplt) 290 (void)memcpy(eh->ether_shost, esrc, 291 sizeof(eh->ether_shost)); 292 else 293 (void)memcpy(eh->ether_shost, IF_LLADDR(ifp), 294 sizeof(eh->ether_shost)); 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 && 306 ((t = pf_find_mtag(m)) == NULL || !t->routed)) { 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) { 315 struct mbuf *n; 316 317 /* 318 * Because if_simloop() modifies the packet, we need a 319 * writable copy through m_dup() instead of a readonly 320 * one as m_copy[m] would give us. The alternative would 321 * be to modify if_simloop() to handle the readonly mbuf, 322 * but performancewise it is mostly equivalent (trading 323 * extra data copying vs. extra locking). 324 * 325 * XXX This is a local workaround. A number of less 326 * often used kernel parts suffer from the same bug. 327 * See PR kern/105943 for a proposed general solution. 328 */ 329 if ((n = m_dup(m, M_DONTWAIT)) != NULL) { 330 n->m_pkthdr.csum_flags |= csum_flags; 331 if (csum_flags & CSUM_DATA_VALID) 332 n->m_pkthdr.csum_data = 0xffff; 333 (void)if_simloop(ifp, n, dst->sa_family, hlen); 334 } else 335 ifp->if_iqdrops++; 336 } else if (bcmp(eh->ether_dhost, eh->ether_shost, 337 ETHER_ADDR_LEN) == 0) { 338 m->m_pkthdr.csum_flags |= csum_flags; 339 if (csum_flags & CSUM_DATA_VALID) 340 m->m_pkthdr.csum_data = 0xffff; 341 (void) if_simloop(ifp, m, dst->sa_family, hlen); 342 return (0); /* XXX */ 343 } 344 } 345 346 /* 347 * Bridges require special output handling. 348 */ 349 if (ifp->if_bridge) { 350 BRIDGE_OUTPUT(ifp, m, error); 351 return (error); 352 } 353 354 #ifdef DEV_CARP 355 if (ifp->if_carp && 356 (error = carp_output(ifp, m, dst, NULL))) 357 goto bad; 358 #endif 359 360 /* Handle ng_ether(4) processing, if any */ 361 if (IFP2AC(ifp)->ac_netgraph != NULL) { 362 KASSERT(ng_ether_output_p != NULL, 363 ("ng_ether_output_p is NULL")); 364 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) { 365 bad: if (m != NULL) 366 m_freem(m); 367 return (error); 368 } 369 if (m == NULL) 370 return (0); 371 } 372 373 /* Continue with link-layer output */ 374 return ether_output_frame(ifp, m); 375 } 376 377 /* 378 * Ethernet link layer output routine to send a raw frame to the device. 379 * 380 * This assumes that the 14 byte Ethernet header is present and contiguous 381 * in the first mbuf (if BRIDGE'ing). 382 */ 383 int 384 ether_output_frame(struct ifnet *ifp, struct mbuf *m) 385 { 386 int error; 387 #if defined(INET) || defined(INET6) 388 struct ip_fw *rule = ip_dn_claim_rule(m); 389 390 if (IPFW_LOADED && ether_ipfw != 0) { 391 if (ether_ipfw_chk(&m, ifp, &rule, 0) == 0) { 392 if (m) { 393 m_freem(m); 394 return EACCES; /* pkt dropped */ 395 } else 396 return 0; /* consumed e.g. in a pipe */ 397 } 398 } 399 #endif 400 401 /* 402 * Queue message on interface, update output statistics if 403 * successful, and start output if interface not yet active. 404 */ 405 IFQ_HANDOFF(ifp, m, error); 406 return (error); 407 } 408 409 #if defined(INET) || defined(INET6) 410 /* 411 * ipfw processing for ethernet packets (in and out). 412 * The second parameter is NULL from ether_demux, and ifp from 413 * ether_output_frame. 414 */ 415 int 416 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 417 struct ip_fw **rule, int shared) 418 { 419 struct ether_header *eh; 420 struct ether_header save_eh; 421 struct mbuf *m; 422 int i; 423 struct ip_fw_args args; 424 425 if (*rule != NULL && fw_one_pass) 426 return 1; /* dummynet packet, already partially processed */ 427 428 /* 429 * I need some amt of data to be contiguous, and in case others need 430 * the packet (shared==1) also better be in the first mbuf. 431 */ 432 m = *m0; 433 i = min( m->m_pkthdr.len, max_protohdr); 434 if ( shared || m->m_len < i) { 435 m = m_pullup(m, i); 436 if (m == NULL) { 437 *m0 = m; 438 return 0; 439 } 440 } 441 eh = mtod(m, struct ether_header *); 442 save_eh = *eh; /* save copy for restore below */ 443 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */ 444 445 args.m = m; /* the packet we are looking at */ 446 args.oif = dst; /* destination, if any */ 447 args.rule = *rule; /* matching rule to restart */ 448 args.next_hop = NULL; /* we do not support forward yet */ 449 args.eh = &save_eh; /* MAC header for bridged/MAC packets */ 450 args.inp = NULL; /* used by ipfw uid/gid/jail rules */ 451 i = ip_fw_chk_ptr(&args); 452 m = args.m; 453 if (m != NULL) { 454 /* 455 * Restore Ethernet header, as needed, in case the 456 * mbuf chain was replaced by ipfw. 457 */ 458 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 459 if (m == NULL) { 460 *m0 = m; 461 return 0; 462 } 463 if (eh != mtod(m, struct ether_header *)) 464 bcopy(&save_eh, mtod(m, struct ether_header *), 465 ETHER_HDR_LEN); 466 } 467 *m0 = m; 468 *rule = args.rule; 469 470 if (i == IP_FW_DENY) /* drop */ 471 return 0; 472 473 KASSERT(m != NULL, ("ether_ipfw_chk: m is NULL")); 474 475 if (i == IP_FW_PASS) /* a PASS rule. */ 476 return 1; 477 478 if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) { 479 /* 480 * Pass the pkt to dummynet, which consumes it. 481 * If shared, make a copy and keep the original. 482 */ 483 if (shared) { 484 m = m_copypacket(m, M_DONTWAIT); 485 if (m == NULL) 486 return 0; 487 } else { 488 /* 489 * Pass the original to dummynet and 490 * nothing back to the caller 491 */ 492 *m0 = NULL ; 493 } 494 ip_dn_io_ptr(&m, dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args); 495 return 0; 496 } 497 /* 498 * XXX at some point add support for divert/forward actions. 499 * If none of the above matches, we have to drop the pkt. 500 */ 501 return 0; 502 } 503 #endif 504 505 /* 506 * Process a received Ethernet packet; the packet is in the 507 * mbuf chain m with the ethernet header at the front. 508 */ 509 static void 510 ether_input(struct ifnet *ifp, struct mbuf *m) 511 { 512 struct ether_header *eh; 513 u_short etype; 514 515 if ((ifp->if_flags & IFF_UP) == 0) { 516 m_freem(m); 517 return; 518 } 519 #ifdef DIAGNOSTIC 520 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 521 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n"); 522 m_freem(m); 523 return; 524 } 525 #endif 526 /* 527 * Do consistency checks to verify assumptions 528 * made by code past this point. 529 */ 530 if ((m->m_flags & M_PKTHDR) == 0) { 531 if_printf(ifp, "discard frame w/o packet header\n"); 532 ifp->if_ierrors++; 533 m_freem(m); 534 return; 535 } 536 if (m->m_len < ETHER_HDR_LEN) { 537 /* XXX maybe should pullup? */ 538 if_printf(ifp, "discard frame w/o leading ethernet " 539 "header (len %u pkt len %u)\n", 540 m->m_len, m->m_pkthdr.len); 541 ifp->if_ierrors++; 542 m_freem(m); 543 return; 544 } 545 eh = mtod(m, struct ether_header *); 546 etype = ntohs(eh->ether_type); 547 if (m->m_pkthdr.rcvif == NULL) { 548 if_printf(ifp, "discard frame w/o interface pointer\n"); 549 ifp->if_ierrors++; 550 m_freem(m); 551 return; 552 } 553 #ifdef DIAGNOSTIC 554 if (m->m_pkthdr.rcvif != ifp) { 555 if_printf(ifp, "Warning, frame marked as received on %s\n", 556 m->m_pkthdr.rcvif->if_xname); 557 } 558 #endif 559 560 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 561 if (ETHER_IS_BROADCAST(eh->ether_dhost)) 562 m->m_flags |= M_BCAST; 563 else 564 m->m_flags |= M_MCAST; 565 ifp->if_imcasts++; 566 } 567 568 #ifdef MAC 569 /* 570 * Tag the mbuf with an appropriate MAC label before any other 571 * consumers can get to it. 572 */ 573 mac_ifnet_create_mbuf(ifp, m); 574 #endif 575 576 /* 577 * Give bpf a chance at the packet. 578 */ 579 ETHER_BPF_MTAP(ifp, m); 580 581 /* 582 * If the CRC is still on the packet, trim it off. We do this once 583 * and once only in case we are re-entered. Nothing else on the 584 * Ethernet receive path expects to see the FCS. 585 */ 586 if (m->m_flags & M_HASFCS) { 587 m_adj(m, -ETHER_CRC_LEN); 588 m->m_flags &= ~M_HASFCS; 589 } 590 591 ifp->if_ibytes += m->m_pkthdr.len; 592 593 /* Allow monitor mode to claim this frame, after stats are updated. */ 594 if (ifp->if_flags & IFF_MONITOR) { 595 m_freem(m); 596 return; 597 } 598 599 /* Handle input from a lagg(4) port */ 600 if (ifp->if_type == IFT_IEEE8023ADLAG) { 601 KASSERT(lagg_input_p != NULL, 602 ("%s: if_lagg not loaded!", __func__)); 603 m = (*lagg_input_p)(ifp, m); 604 if (m != NULL) 605 ifp = m->m_pkthdr.rcvif; 606 else 607 return; 608 } 609 610 /* 611 * If the hardware did not process an 802.1Q tag, do this now, 612 * to allow 802.1P priority frames to be passed to the main input 613 * path correctly. 614 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels. 615 */ 616 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) { 617 struct ether_vlan_header *evl; 618 619 if (m->m_len < sizeof(*evl) && 620 (m = m_pullup(m, sizeof(*evl))) == NULL) { 621 #ifdef DIAGNOSTIC 622 if_printf(ifp, "cannot pullup VLAN header\n"); 623 #endif 624 ifp->if_ierrors++; 625 m_freem(m); 626 return; 627 } 628 629 evl = mtod(m, struct ether_vlan_header *); 630 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag); 631 m->m_flags |= M_VLANTAG; 632 633 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN, 634 ETHER_HDR_LEN - ETHER_TYPE_LEN); 635 m_adj(m, ETHER_VLAN_ENCAP_LEN); 636 } 637 638 /* Allow ng_ether(4) to claim this frame. */ 639 if (IFP2AC(ifp)->ac_netgraph != NULL) { 640 KASSERT(ng_ether_input_p != NULL, 641 ("%s: ng_ether_input_p is NULL", __func__)); 642 m->m_flags &= ~M_PROMISC; 643 (*ng_ether_input_p)(ifp, &m); 644 if (m == NULL) 645 return; 646 } 647 648 /* 649 * Allow if_bridge(4) to claim this frame. 650 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it 651 * and the frame should be delivered locally. 652 */ 653 if (ifp->if_bridge != NULL) { 654 m->m_flags &= ~M_PROMISC; 655 BRIDGE_INPUT(ifp, m); 656 if (m == NULL) 657 return; 658 } 659 660 #ifdef DEV_CARP 661 /* 662 * Clear M_PROMISC on frame so that carp(4) will see it when the 663 * mbuf flows up to Layer 3. 664 * FreeBSD's implementation of carp(4) uses the inprotosw 665 * to dispatch IPPROTO_CARP. carp(4) also allocates its own 666 * Ethernet addresses of the form 00:00:5e:00:01:xx, which 667 * is outside the scope of the M_PROMISC test below. 668 * TODO: Maintain a hash table of ethernet addresses other than 669 * ether_dhost which may be active on this ifp. 670 */ 671 if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost)) { 672 m->m_flags &= ~M_PROMISC; 673 } else 674 #endif 675 { 676 /* 677 * If the frame received was not for our MAC address, set the 678 * M_PROMISC flag on the mbuf chain. The frame may need to 679 * be seen by the rest of the Ethernet input path in case of 680 * re-entry (e.g. bridge, vlan, netgraph) but should not be 681 * seen by upper protocol layers. 682 */ 683 if (!ETHER_IS_MULTICAST(eh->ether_dhost) && 684 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0) 685 m->m_flags |= M_PROMISC; 686 } 687 688 /* First chunk of an mbuf contains good entropy */ 689 if (harvest.ethernet) 690 random_harvest(m, 16, 3, 0, RANDOM_NET); 691 692 ether_demux(ifp, m); 693 } 694 695 /* 696 * Upper layer processing for a received Ethernet packet. 697 */ 698 void 699 ether_demux(struct ifnet *ifp, struct mbuf *m) 700 { 701 struct ether_header *eh; 702 int isr; 703 u_short ether_type; 704 #if defined(NETATALK) 705 struct llc *l; 706 #endif 707 708 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__)); 709 710 #if defined(INET) || defined(INET6) 711 /* 712 * Allow dummynet and/or ipfw to claim the frame. 713 * Do not do this for PROMISC frames in case we are re-entered. 714 */ 715 if (IPFW_LOADED && ether_ipfw != 0 && !(m->m_flags & M_PROMISC)) { 716 struct ip_fw *rule = ip_dn_claim_rule(m); 717 718 if (ether_ipfw_chk(&m, NULL, &rule, 0) == 0) { 719 if (m) 720 m_freem(m); /* dropped; free mbuf chain */ 721 return; /* consumed */ 722 } 723 } 724 #endif 725 eh = mtod(m, struct ether_header *); 726 ether_type = ntohs(eh->ether_type); 727 728 /* 729 * If this frame has a VLAN tag other than 0, call vlan_input() 730 * if its module is loaded. Otherwise, drop. 731 */ 732 if ((m->m_flags & M_VLANTAG) && 733 EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) { 734 if (ifp->if_vlantrunk == NULL) { 735 ifp->if_noproto++; 736 m_freem(m); 737 return; 738 } 739 KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!", 740 __func__)); 741 /* Clear before possibly re-entering ether_input(). */ 742 m->m_flags &= ~M_PROMISC; 743 (*vlan_input_p)(ifp, m); 744 return; 745 } 746 747 /* 748 * Pass promiscuously received frames to the upper layer if the user 749 * requested this by setting IFF_PPROMISC. Otherwise, drop them. 750 */ 751 if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) { 752 m_freem(m); 753 return; 754 } 755 756 /* 757 * Reset layer specific mbuf flags to avoid confusing upper layers. 758 * Strip off Ethernet header. 759 */ 760 m->m_flags &= ~M_VLANTAG; 761 m->m_flags &= ~(M_PROTOFLAGS); 762 m_adj(m, ETHER_HDR_LEN); 763 764 /* 765 * Dispatch frame to upper layer. 766 */ 767 switch (ether_type) { 768 #ifdef INET 769 case ETHERTYPE_IP: 770 if ((m = ip_fastforward(m)) == NULL) 771 return; 772 isr = NETISR_IP; 773 break; 774 775 case ETHERTYPE_ARP: 776 if (ifp->if_flags & IFF_NOARP) { 777 /* Discard packet if ARP is disabled on interface */ 778 m_freem(m); 779 return; 780 } 781 isr = NETISR_ARP; 782 break; 783 #endif 784 #ifdef IPX 785 case ETHERTYPE_IPX: 786 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 787 return; 788 isr = NETISR_IPX; 789 break; 790 #endif 791 #ifdef INET6 792 case ETHERTYPE_IPV6: 793 isr = NETISR_IPV6; 794 break; 795 #endif 796 #ifdef NETATALK 797 case ETHERTYPE_AT: 798 isr = NETISR_ATALK1; 799 break; 800 case ETHERTYPE_AARP: 801 isr = NETISR_AARP; 802 break; 803 #endif /* NETATALK */ 804 default: 805 #ifdef IPX 806 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 807 return; 808 #endif /* IPX */ 809 #if defined(NETATALK) 810 if (ether_type > ETHERMTU) 811 goto discard; 812 l = mtod(m, struct llc *); 813 if (l->llc_dsap == LLC_SNAP_LSAP && 814 l->llc_ssap == LLC_SNAP_LSAP && 815 l->llc_control == LLC_UI) { 816 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code, 817 sizeof(at_org_code)) == 0 && 818 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) { 819 m_adj(m, LLC_SNAPFRAMELEN); 820 isr = NETISR_ATALK2; 821 break; 822 } 823 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code, 824 sizeof(aarp_org_code)) == 0 && 825 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) { 826 m_adj(m, LLC_SNAPFRAMELEN); 827 isr = NETISR_AARP; 828 break; 829 } 830 } 831 #endif /* NETATALK */ 832 goto discard; 833 } 834 netisr_dispatch(isr, m); 835 return; 836 837 discard: 838 /* 839 * Packet is to be discarded. If netgraph is present, 840 * hand the packet to it for last chance processing; 841 * otherwise dispose of it. 842 */ 843 if (IFP2AC(ifp)->ac_netgraph != NULL) { 844 KASSERT(ng_ether_input_orphan_p != NULL, 845 ("ng_ether_input_orphan_p is NULL")); 846 /* 847 * Put back the ethernet header so netgraph has a 848 * consistent view of inbound packets. 849 */ 850 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 851 (*ng_ether_input_orphan_p)(ifp, m); 852 return; 853 } 854 m_freem(m); 855 } 856 857 /* 858 * Convert Ethernet address to printable (loggable) representation. 859 * This routine is for compatibility; it's better to just use 860 * 861 * printf("%6D", <pointer to address>, ":"); 862 * 863 * since there's no static buffer involved. 864 */ 865 char * 866 ether_sprintf(const u_char *ap) 867 { 868 static char etherbuf[18]; 869 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":"); 870 return (etherbuf); 871 } 872 873 /* 874 * Perform common duties while attaching to interface list 875 */ 876 void 877 ether_ifattach(struct ifnet *ifp, const u_int8_t *lla) 878 { 879 int i; 880 struct ifaddr *ifa; 881 struct sockaddr_dl *sdl; 882 883 ifp->if_addrlen = ETHER_ADDR_LEN; 884 ifp->if_hdrlen = ETHER_HDR_LEN; 885 if_attach(ifp); 886 ifp->if_mtu = ETHERMTU; 887 ifp->if_output = ether_output; 888 ifp->if_input = ether_input; 889 ifp->if_resolvemulti = ether_resolvemulti; 890 if (ifp->if_baudrate == 0) 891 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 892 ifp->if_broadcastaddr = etherbroadcastaddr; 893 894 ifa = ifp->if_addr; 895 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__)); 896 sdl = (struct sockaddr_dl *)ifa->ifa_addr; 897 sdl->sdl_type = IFT_ETHER; 898 sdl->sdl_alen = ifp->if_addrlen; 899 bcopy(lla, LLADDR(sdl), ifp->if_addrlen); 900 901 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN); 902 if (ng_ether_attach_p != NULL) 903 (*ng_ether_attach_p)(ifp); 904 905 /* Announce Ethernet MAC address if non-zero. */ 906 for (i = 0; i < ifp->if_addrlen; i++) 907 if (lla[i] != 0) 908 break; 909 if (i != ifp->if_addrlen) 910 if_printf(ifp, "Ethernet address: %6D\n", lla, ":"); 911 if (ifp->if_flags & IFF_NEEDSGIANT) 912 if_printf(ifp, "if_start running deferred for Giant\n"); 913 } 914 915 /* 916 * Perform common duties while detaching an Ethernet interface 917 */ 918 void 919 ether_ifdetach(struct ifnet *ifp) 920 { 921 if (IFP2AC(ifp)->ac_netgraph != NULL) { 922 KASSERT(ng_ether_detach_p != NULL, 923 ("ng_ether_detach_p is NULL")); 924 (*ng_ether_detach_p)(ifp); 925 } 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, u_long 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 IF_LLADDR(ifp); 1041 else { 1042 bcopy((caddr_t) ina->x_host.c_host, 1043 (caddr_t) IF_LLADDR(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(IF_LLADDR(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 void 1213 ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen) 1214 { 1215 struct ether_vlan_header vlan; 1216 struct mbuf mv, mb; 1217 1218 KASSERT((m->m_flags & M_VLANTAG) != 0, 1219 ("%s: vlan information not present", __func__)); 1220 KASSERT(m->m_len >= sizeof(struct ether_header), 1221 ("%s: mbuf not large enough for header", __func__)); 1222 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header)); 1223 vlan.evl_proto = vlan.evl_encap_proto; 1224 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN); 1225 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag); 1226 m->m_len -= sizeof(struct ether_header); 1227 m->m_data += sizeof(struct ether_header); 1228 /* 1229 * If a data link has been supplied by the caller, then we will need to 1230 * re-create a stack allocated mbuf chain with the following structure: 1231 * 1232 * (1) mbuf #1 will contain the supplied data link 1233 * (2) mbuf #2 will contain the vlan header 1234 * (3) mbuf #3 will contain the original mbuf's packet data 1235 * 1236 * Otherwise, submit the packet and vlan header via bpf_mtap2(). 1237 */ 1238 if (data != NULL) { 1239 mv.m_next = m; 1240 mv.m_data = (caddr_t)&vlan; 1241 mv.m_len = sizeof(vlan); 1242 mb.m_next = &mv; 1243 mb.m_data = data; 1244 mb.m_len = dlen; 1245 bpf_mtap(bp, &mb); 1246 } else 1247 bpf_mtap2(bp, &vlan, sizeof(vlan), m); 1248 m->m_len += sizeof(struct ether_header); 1249 m->m_data -= sizeof(struct ether_header); 1250 } 1251 1252 struct mbuf * 1253 ether_vlanencap(struct mbuf *m, uint16_t tag) 1254 { 1255 struct ether_vlan_header *evl; 1256 1257 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_DONTWAIT); 1258 if (m == NULL) 1259 return (NULL); 1260 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */ 1261 1262 if (m->m_len < sizeof(*evl)) { 1263 m = m_pullup(m, sizeof(*evl)); 1264 if (m == NULL) 1265 return (NULL); 1266 } 1267 1268 /* 1269 * Transform the Ethernet header into an Ethernet header 1270 * with 802.1Q encapsulation. 1271 */ 1272 evl = mtod(m, struct ether_vlan_header *); 1273 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN, 1274 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN); 1275 evl->evl_encap_proto = htons(ETHERTYPE_VLAN); 1276 evl->evl_tag = htons(tag); 1277 return (m); 1278 } 1279 1280 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY); 1281 MODULE_VERSION(ether, 1); 1282