if_ethersubr.c revision 183550
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: head/sys/net/if_ethersubr.c 183550 2008-10-02 15:37:58Z zec $ 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#include "opt_mbuf_profiling.h" 41 42#include <sys/param.h> 43#include <sys/systm.h> 44#include <sys/kernel.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#include <sys/vimage.h> 53 54#include <net/if.h> 55#include <net/if_arp.h> 56#include <net/netisr.h> 57#include <net/route.h> 58#include <net/if_llc.h> 59#include <net/if_dl.h> 60#include <net/if_types.h> 61#include <net/bpf.h> 62#include <net/ethernet.h> 63#include <net/if_bridgevar.h> 64#include <net/if_vlan_var.h> 65#include <net/pf_mtag.h> 66 67#if defined(INET) || defined(INET6) 68#include <netinet/in.h> 69#include <netinet/in_var.h> 70#include <netinet/if_ether.h> 71#include <netinet/ip_fw.h> 72#include <netinet/ip_dummynet.h> 73#endif 74#ifdef INET6 75#include <netinet6/nd6.h> 76#endif 77 78#ifdef DEV_CARP 79#include <netinet/ip_carp.h> 80#endif 81 82#ifdef IPX 83#include <netipx/ipx.h> 84#include <netipx/ipx_if.h> 85#endif 86int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m); 87int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, 88 struct sockaddr *dst, short *tp, int *hlen); 89 90#ifdef NETATALK 91#include <netatalk/at.h> 92#include <netatalk/at_var.h> 93#include <netatalk/at_extern.h> 94 95#define llc_snap_org_code llc_un.type_snap.org_code 96#define llc_snap_ether_type llc_un.type_snap.ether_type 97 98extern u_char at_org_code[3]; 99extern u_char aarp_org_code[3]; 100#endif /* NETATALK */ 101 102#include <security/mac/mac_framework.h> 103 104#ifdef CTASSERT 105CTASSERT(sizeof (struct ether_header) == ETHER_ADDR_LEN * 2 + 2); 106CTASSERT(sizeof (struct ether_addr) == ETHER_ADDR_LEN); 107#endif 108 109/* netgraph node hooks for ng_ether(4) */ 110void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp); 111void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m); 112int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp); 113void (*ng_ether_attach_p)(struct ifnet *ifp); 114void (*ng_ether_detach_p)(struct ifnet *ifp); 115 116void (*vlan_input_p)(struct ifnet *, struct mbuf *); 117 118/* if_bridge(4) support */ 119struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *); 120int (*bridge_output_p)(struct ifnet *, struct mbuf *, 121 struct sockaddr *, struct rtentry *); 122void (*bridge_dn_p)(struct mbuf *, struct ifnet *); 123 124/* if_lagg(4) support */ 125struct mbuf *(*lagg_input_p)(struct ifnet *, struct mbuf *); 126 127static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] = 128 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 129 130static int ether_resolvemulti(struct ifnet *, struct sockaddr **, 131 struct sockaddr *); 132 133/* XXX: should be in an arp support file, not here */ 134MALLOC_DEFINE(M_ARPCOM, "arpcom", "802.* interface internals"); 135 136#define ETHER_IS_BROADCAST(addr) \ 137 (bcmp(etherbroadcastaddr, (addr), ETHER_ADDR_LEN) == 0) 138 139#define senderr(e) do { error = (e); goto bad;} while (0) 140 141#if defined(INET) || defined(INET6) 142int 143ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 144 struct ip_fw **rule, int shared); 145static int ether_ipfw; 146#endif 147 148/* 149 * Ethernet output routine. 150 * Encapsulate a packet of type family for the local net. 151 * Use trailer local net encapsulation if enough data in first 152 * packet leaves a multiple of 512 bytes of data in remainder. 153 */ 154int 155ether_output(struct ifnet *ifp, struct mbuf *m, 156 struct sockaddr *dst, struct rtentry *rt0) 157{ 158 short type; 159 int error, hdrcmplt = 0; 160 u_char esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN]; 161 struct ether_header *eh; 162 struct pf_mtag *t; 163 int loop_copy = 1; 164 int hlen; /* link layer header length */ 165 166#ifdef MAC 167 error = mac_ifnet_check_transmit(ifp, m); 168 if (error) 169 senderr(error); 170#endif 171 172 M_PROFILE(m); 173 if (ifp->if_flags & IFF_MONITOR) 174 senderr(ENETDOWN); 175 if (!((ifp->if_flags & IFF_UP) && 176 (ifp->if_drv_flags & IFF_DRV_RUNNING))) 177 senderr(ENETDOWN); 178 179 hlen = ETHER_HDR_LEN; 180 switch (dst->sa_family) { 181#ifdef INET 182 case AF_INET: 183 error = arpresolve(ifp, rt0, m, dst, edst); 184 if (error) 185 return (error == EWOULDBLOCK ? 0 : error); 186 type = htons(ETHERTYPE_IP); 187 break; 188 case AF_ARP: 189 { 190 struct arphdr *ah; 191 ah = mtod(m, struct arphdr *); 192 ah->ar_hrd = htons(ARPHRD_ETHER); 193 194 loop_copy = 0; /* if this is for us, don't do it */ 195 196 switch(ntohs(ah->ar_op)) { 197 case ARPOP_REVREQUEST: 198 case ARPOP_REVREPLY: 199 type = htons(ETHERTYPE_REVARP); 200 break; 201 case ARPOP_REQUEST: 202 case ARPOP_REPLY: 203 default: 204 type = htons(ETHERTYPE_ARP); 205 break; 206 } 207 208 if (m->m_flags & M_BCAST) 209 bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN); 210 else 211 bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN); 212 213 } 214 break; 215#endif 216#ifdef INET6 217 case AF_INET6: 218 error = nd6_storelladdr(ifp, rt0, m, dst, (u_char *)edst); 219 if (error) 220 return error; 221 type = htons(ETHERTYPE_IPV6); 222 break; 223#endif 224#ifdef IPX 225 case AF_IPX: 226 if (ef_outputp) { 227 error = ef_outputp(ifp, &m, dst, &type, &hlen); 228 if (error) 229 goto bad; 230 } else 231 type = htons(ETHERTYPE_IPX); 232 bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host), 233 (caddr_t)edst, sizeof (edst)); 234 break; 235#endif 236#ifdef NETATALK 237 case AF_APPLETALK: 238 { 239 struct at_ifaddr *aa; 240 241 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) 242 senderr(EHOSTUNREACH); /* XXX */ 243 if (!aarpresolve(ifp, m, (struct sockaddr_at *)dst, edst)) 244 return (0); 245 /* 246 * In the phase 2 case, need to prepend an mbuf for the llc header. 247 */ 248 if ( aa->aa_flags & AFA_PHASE2 ) { 249 struct llc llc; 250 251 M_PREPEND(m, LLC_SNAPFRAMELEN, M_DONTWAIT); 252 if (m == NULL) 253 senderr(ENOBUFS); 254 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 255 llc.llc_control = LLC_UI; 256 bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code)); 257 llc.llc_snap_ether_type = htons( ETHERTYPE_AT ); 258 bcopy(&llc, mtod(m, caddr_t), LLC_SNAPFRAMELEN); 259 type = htons(m->m_pkthdr.len); 260 hlen = LLC_SNAPFRAMELEN + ETHER_HDR_LEN; 261 } else { 262 type = htons(ETHERTYPE_AT); 263 } 264 break; 265 } 266#endif /* NETATALK */ 267 268 case pseudo_AF_HDRCMPLT: 269 hdrcmplt = 1; 270 eh = (struct ether_header *)dst->sa_data; 271 (void)memcpy(esrc, eh->ether_shost, sizeof (esrc)); 272 /* FALLTHROUGH */ 273 274 case AF_UNSPEC: 275 loop_copy = 0; /* if this is for us, don't do it */ 276 eh = (struct ether_header *)dst->sa_data; 277 (void)memcpy(edst, eh->ether_dhost, sizeof (edst)); 278 type = eh->ether_type; 279 break; 280 281 default: 282 if_printf(ifp, "can't handle af%d\n", dst->sa_family); 283 senderr(EAFNOSUPPORT); 284 } 285 286 /* 287 * Add local net header. If no space in first mbuf, 288 * allocate another. 289 */ 290 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 291 if (m == NULL) 292 senderr(ENOBUFS); 293 eh = mtod(m, struct ether_header *); 294 (void)memcpy(&eh->ether_type, &type, 295 sizeof(eh->ether_type)); 296 (void)memcpy(eh->ether_dhost, edst, sizeof (edst)); 297 if (hdrcmplt) 298 (void)memcpy(eh->ether_shost, esrc, 299 sizeof(eh->ether_shost)); 300 else 301 (void)memcpy(eh->ether_shost, IF_LLADDR(ifp), 302 sizeof(eh->ether_shost)); 303 304 /* 305 * If a simplex interface, and the packet is being sent to our 306 * Ethernet address or a broadcast address, loopback a copy. 307 * XXX To make a simplex device behave exactly like a duplex 308 * device, we should copy in the case of sending to our own 309 * ethernet address (thus letting the original actually appear 310 * on the wire). However, we don't do that here for security 311 * reasons and compatibility with the original behavior. 312 */ 313 if ((ifp->if_flags & IFF_SIMPLEX) && loop_copy && 314 ((t = pf_find_mtag(m)) == NULL || !t->routed)) { 315 int csum_flags = 0; 316 317 if (m->m_pkthdr.csum_flags & CSUM_IP) 318 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID); 319 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) 320 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR); 321 322 if (m->m_flags & M_BCAST) { 323 struct mbuf *n; 324 325 /* 326 * Because if_simloop() modifies the packet, we need a 327 * writable copy through m_dup() instead of a readonly 328 * one as m_copy[m] would give us. The alternative would 329 * be to modify if_simloop() to handle the readonly mbuf, 330 * but performancewise it is mostly equivalent (trading 331 * extra data copying vs. extra locking). 332 * 333 * XXX This is a local workaround. A number of less 334 * often used kernel parts suffer from the same bug. 335 * See PR kern/105943 for a proposed general solution. 336 */ 337 if ((n = m_dup(m, M_DONTWAIT)) != NULL) { 338 n->m_pkthdr.csum_flags |= csum_flags; 339 if (csum_flags & CSUM_DATA_VALID) 340 n->m_pkthdr.csum_data = 0xffff; 341 (void)if_simloop(ifp, n, dst->sa_family, hlen); 342 } else 343 ifp->if_iqdrops++; 344 } else if (bcmp(eh->ether_dhost, eh->ether_shost, 345 ETHER_ADDR_LEN) == 0) { 346 m->m_pkthdr.csum_flags |= csum_flags; 347 if (csum_flags & CSUM_DATA_VALID) 348 m->m_pkthdr.csum_data = 0xffff; 349 (void) if_simloop(ifp, m, dst->sa_family, hlen); 350 return (0); /* XXX */ 351 } 352 } 353 354 /* 355 * Bridges require special output handling. 356 */ 357 if (ifp->if_bridge) { 358 BRIDGE_OUTPUT(ifp, m, error); 359 return (error); 360 } 361 362#ifdef DEV_CARP 363 if (ifp->if_carp && 364 (error = carp_output(ifp, m, dst, NULL))) 365 goto bad; 366#endif 367 368 /* Handle ng_ether(4) processing, if any */ 369 if (IFP2AC(ifp)->ac_netgraph != NULL) { 370 KASSERT(ng_ether_output_p != NULL, 371 ("ng_ether_output_p is NULL")); 372 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) { 373bad: if (m != NULL) 374 m_freem(m); 375 return (error); 376 } 377 if (m == NULL) 378 return (0); 379 } 380 381 /* Continue with link-layer output */ 382 return ether_output_frame(ifp, m); 383} 384 385/* 386 * Ethernet link layer output routine to send a raw frame to the device. 387 * 388 * This assumes that the 14 byte Ethernet header is present and contiguous 389 * in the first mbuf (if BRIDGE'ing). 390 */ 391int 392ether_output_frame(struct ifnet *ifp, struct mbuf *m) 393{ 394 int error; 395#if defined(INET) || defined(INET6) 396 INIT_VNET_NET(ifp->if_vnet); 397 struct ip_fw *rule = ip_dn_claim_rule(m); 398 399 if (IPFW_LOADED && V_ether_ipfw != 0) { 400 if (ether_ipfw_chk(&m, ifp, &rule, 0) == 0) { 401 if (m) { 402 m_freem(m); 403 return EACCES; /* pkt dropped */ 404 } else 405 return 0; /* consumed e.g. in a pipe */ 406 } 407 } 408#endif 409 410 /* 411 * Queue message on interface, update output statistics if 412 * successful, and start output if interface not yet active. 413 */ 414 IFQ_HANDOFF(ifp, m, error); 415 return (error); 416} 417 418#if defined(INET) || defined(INET6) 419/* 420 * ipfw processing for ethernet packets (in and out). 421 * The second parameter is NULL from ether_demux, and ifp from 422 * ether_output_frame. 423 */ 424int 425ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 426 struct ip_fw **rule, int shared) 427{ 428 INIT_VNET_IPFW(dst->if_vnet); 429 struct ether_header *eh; 430 struct ether_header save_eh; 431 struct mbuf *m; 432 int i; 433 struct ip_fw_args args; 434 435 if (*rule != NULL && V_fw_one_pass) 436 return 1; /* dummynet packet, already partially processed */ 437 438 /* 439 * I need some amt of data to be contiguous, and in case others need 440 * the packet (shared==1) also better be in the first mbuf. 441 */ 442 m = *m0; 443 i = min( m->m_pkthdr.len, max_protohdr); 444 if ( shared || m->m_len < i) { 445 m = m_pullup(m, i); 446 if (m == NULL) { 447 *m0 = m; 448 return 0; 449 } 450 } 451 eh = mtod(m, struct ether_header *); 452 save_eh = *eh; /* save copy for restore below */ 453 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */ 454 455 args.m = m; /* the packet we are looking at */ 456 args.oif = dst; /* destination, if any */ 457 args.rule = *rule; /* matching rule to restart */ 458 args.next_hop = NULL; /* we do not support forward yet */ 459 args.eh = &save_eh; /* MAC header for bridged/MAC packets */ 460 args.inp = NULL; /* used by ipfw uid/gid/jail rules */ 461 i = ip_fw_chk_ptr(&args); 462 m = args.m; 463 if (m != NULL) { 464 /* 465 * Restore Ethernet header, as needed, in case the 466 * mbuf chain was replaced by ipfw. 467 */ 468 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 469 if (m == NULL) { 470 *m0 = m; 471 return 0; 472 } 473 if (eh != mtod(m, struct ether_header *)) 474 bcopy(&save_eh, mtod(m, struct ether_header *), 475 ETHER_HDR_LEN); 476 } 477 *m0 = m; 478 *rule = args.rule; 479 480 if (i == IP_FW_DENY) /* drop */ 481 return 0; 482 483 KASSERT(m != NULL, ("ether_ipfw_chk: m is NULL")); 484 485 if (i == IP_FW_PASS) /* a PASS rule. */ 486 return 1; 487 488 if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) { 489 /* 490 * Pass the pkt to dummynet, which consumes it. 491 * If shared, make a copy and keep the original. 492 */ 493 if (shared) { 494 m = m_copypacket(m, M_DONTWAIT); 495 if (m == NULL) 496 return 0; 497 } else { 498 /* 499 * Pass the original to dummynet and 500 * nothing back to the caller 501 */ 502 *m0 = NULL ; 503 } 504 ip_dn_io_ptr(&m, dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args); 505 return 0; 506 } 507 /* 508 * XXX at some point add support for divert/forward actions. 509 * If none of the above matches, we have to drop the pkt. 510 */ 511 return 0; 512} 513#endif 514 515/* 516 * Process a received Ethernet packet; the packet is in the 517 * mbuf chain m with the ethernet header at the front. 518 */ 519static void 520ether_input(struct ifnet *ifp, struct mbuf *m) 521{ 522 struct ether_header *eh; 523 u_short etype; 524 525 if ((ifp->if_flags & IFF_UP) == 0) { 526 m_freem(m); 527 return; 528 } 529#ifdef DIAGNOSTIC 530 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 531 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n"); 532 m_freem(m); 533 return; 534 } 535#endif 536 /* 537 * Do consistency checks to verify assumptions 538 * made by code past this point. 539 */ 540 if ((m->m_flags & M_PKTHDR) == 0) { 541 if_printf(ifp, "discard frame w/o packet header\n"); 542 ifp->if_ierrors++; 543 m_freem(m); 544 return; 545 } 546 if (m->m_len < ETHER_HDR_LEN) { 547 /* XXX maybe should pullup? */ 548 if_printf(ifp, "discard frame w/o leading ethernet " 549 "header (len %u pkt len %u)\n", 550 m->m_len, m->m_pkthdr.len); 551 ifp->if_ierrors++; 552 m_freem(m); 553 return; 554 } 555 eh = mtod(m, struct ether_header *); 556 etype = ntohs(eh->ether_type); 557 if (m->m_pkthdr.rcvif == NULL) { 558 if_printf(ifp, "discard frame w/o interface pointer\n"); 559 ifp->if_ierrors++; 560 m_freem(m); 561 return; 562 } 563#ifdef DIAGNOSTIC 564 if (m->m_pkthdr.rcvif != ifp) { 565 if_printf(ifp, "Warning, frame marked as received on %s\n", 566 m->m_pkthdr.rcvif->if_xname); 567 } 568#endif 569 570 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 571 if (ETHER_IS_BROADCAST(eh->ether_dhost)) 572 m->m_flags |= M_BCAST; 573 else 574 m->m_flags |= M_MCAST; 575 ifp->if_imcasts++; 576 } 577 578#ifdef MAC 579 /* 580 * Tag the mbuf with an appropriate MAC label before any other 581 * consumers can get to it. 582 */ 583 mac_ifnet_create_mbuf(ifp, m); 584#endif 585 586 /* 587 * Give bpf a chance at the packet. 588 */ 589 ETHER_BPF_MTAP(ifp, m); 590 591 /* 592 * If the CRC is still on the packet, trim it off. We do this once 593 * and once only in case we are re-entered. Nothing else on the 594 * Ethernet receive path expects to see the FCS. 595 */ 596 if (m->m_flags & M_HASFCS) { 597 m_adj(m, -ETHER_CRC_LEN); 598 m->m_flags &= ~M_HASFCS; 599 } 600 601 ifp->if_ibytes += m->m_pkthdr.len; 602 603 /* Allow monitor mode to claim this frame, after stats are updated. */ 604 if (ifp->if_flags & IFF_MONITOR) { 605 m_freem(m); 606 return; 607 } 608 609 /* Handle input from a lagg(4) port */ 610 if (ifp->if_type == IFT_IEEE8023ADLAG) { 611 KASSERT(lagg_input_p != NULL, 612 ("%s: if_lagg not loaded!", __func__)); 613 m = (*lagg_input_p)(ifp, m); 614 if (m != NULL) 615 ifp = m->m_pkthdr.rcvif; 616 else 617 return; 618 } 619 620 /* 621 * If the hardware did not process an 802.1Q tag, do this now, 622 * to allow 802.1P priority frames to be passed to the main input 623 * path correctly. 624 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels. 625 */ 626 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) { 627 struct ether_vlan_header *evl; 628 629 if (m->m_len < sizeof(*evl) && 630 (m = m_pullup(m, sizeof(*evl))) == NULL) { 631#ifdef DIAGNOSTIC 632 if_printf(ifp, "cannot pullup VLAN header\n"); 633#endif 634 ifp->if_ierrors++; 635 m_freem(m); 636 return; 637 } 638 639 evl = mtod(m, struct ether_vlan_header *); 640 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag); 641 m->m_flags |= M_VLANTAG; 642 643 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN, 644 ETHER_HDR_LEN - ETHER_TYPE_LEN); 645 m_adj(m, ETHER_VLAN_ENCAP_LEN); 646 } 647 648 /* Allow ng_ether(4) to claim this frame. */ 649 if (IFP2AC(ifp)->ac_netgraph != NULL) { 650 KASSERT(ng_ether_input_p != NULL, 651 ("%s: ng_ether_input_p is NULL", __func__)); 652 m->m_flags &= ~M_PROMISC; 653 (*ng_ether_input_p)(ifp, &m); 654 if (m == NULL) 655 return; 656 } 657 658 /* 659 * Allow if_bridge(4) to claim this frame. 660 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it 661 * and the frame should be delivered locally. 662 */ 663 if (ifp->if_bridge != NULL) { 664 m->m_flags &= ~M_PROMISC; 665 BRIDGE_INPUT(ifp, m); 666 if (m == NULL) 667 return; 668 } 669 670#ifdef DEV_CARP 671 /* 672 * Clear M_PROMISC on frame so that carp(4) will see it when the 673 * mbuf flows up to Layer 3. 674 * FreeBSD's implementation of carp(4) uses the inprotosw 675 * to dispatch IPPROTO_CARP. carp(4) also allocates its own 676 * Ethernet addresses of the form 00:00:5e:00:01:xx, which 677 * is outside the scope of the M_PROMISC test below. 678 * TODO: Maintain a hash table of ethernet addresses other than 679 * ether_dhost which may be active on this ifp. 680 */ 681 if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost)) { 682 m->m_flags &= ~M_PROMISC; 683 } else 684#endif 685 { 686 /* 687 * If the frame received was not for our MAC address, set the 688 * M_PROMISC flag on the mbuf chain. The frame may need to 689 * be seen by the rest of the Ethernet input path in case of 690 * re-entry (e.g. bridge, vlan, netgraph) but should not be 691 * seen by upper protocol layers. 692 */ 693 if (!ETHER_IS_MULTICAST(eh->ether_dhost) && 694 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0) 695 m->m_flags |= M_PROMISC; 696 } 697 698 /* First chunk of an mbuf contains good entropy */ 699 if (harvest.ethernet) 700 random_harvest(m, 16, 3, 0, RANDOM_NET); 701 702 ether_demux(ifp, m); 703} 704 705/* 706 * Upper layer processing for a received Ethernet packet. 707 */ 708void 709ether_demux(struct ifnet *ifp, struct mbuf *m) 710{ 711 struct ether_header *eh; 712 int isr; 713 u_short ether_type; 714#if defined(NETATALK) 715 struct llc *l; 716#endif 717 718 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__)); 719 720#if defined(INET) || defined(INET6) 721 INIT_VNET_NET(ifp->if_vnet); 722 /* 723 * Allow dummynet and/or ipfw to claim the frame. 724 * Do not do this for PROMISC frames in case we are re-entered. 725 */ 726 if (IPFW_LOADED && V_ether_ipfw != 0 && !(m->m_flags & M_PROMISC)) { 727 struct ip_fw *rule = ip_dn_claim_rule(m); 728 729 if (ether_ipfw_chk(&m, NULL, &rule, 0) == 0) { 730 if (m) 731 m_freem(m); /* dropped; free mbuf chain */ 732 return; /* consumed */ 733 } 734 } 735#endif 736 eh = mtod(m, struct ether_header *); 737 ether_type = ntohs(eh->ether_type); 738 739 /* 740 * If this frame has a VLAN tag other than 0, call vlan_input() 741 * if its module is loaded. Otherwise, drop. 742 */ 743 if ((m->m_flags & M_VLANTAG) && 744 EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) { 745 if (ifp->if_vlantrunk == NULL) { 746 ifp->if_noproto++; 747 m_freem(m); 748 return; 749 } 750 KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!", 751 __func__)); 752 /* Clear before possibly re-entering ether_input(). */ 753 m->m_flags &= ~M_PROMISC; 754 (*vlan_input_p)(ifp, m); 755 return; 756 } 757 758 /* 759 * Pass promiscuously received frames to the upper layer if the user 760 * requested this by setting IFF_PPROMISC. Otherwise, drop them. 761 */ 762 if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) { 763 m_freem(m); 764 return; 765 } 766 767 /* 768 * Reset layer specific mbuf flags to avoid confusing upper layers. 769 * Strip off Ethernet header. 770 */ 771 m->m_flags &= ~M_VLANTAG; 772 m->m_flags &= ~(M_PROTOFLAGS); 773 m_adj(m, ETHER_HDR_LEN); 774 775 /* 776 * Dispatch frame to upper layer. 777 */ 778 switch (ether_type) { 779#ifdef INET 780 case ETHERTYPE_IP: 781 if ((m = ip_fastforward(m)) == NULL) 782 return; 783 isr = NETISR_IP; 784 break; 785 786 case ETHERTYPE_ARP: 787 if (ifp->if_flags & IFF_NOARP) { 788 /* Discard packet if ARP is disabled on interface */ 789 m_freem(m); 790 return; 791 } 792 isr = NETISR_ARP; 793 break; 794#endif 795#ifdef IPX 796 case ETHERTYPE_IPX: 797 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 798 return; 799 isr = NETISR_IPX; 800 break; 801#endif 802#ifdef INET6 803 case ETHERTYPE_IPV6: 804 isr = NETISR_IPV6; 805 break; 806#endif 807#ifdef NETATALK 808 case ETHERTYPE_AT: 809 isr = NETISR_ATALK1; 810 break; 811 case ETHERTYPE_AARP: 812 isr = NETISR_AARP; 813 break; 814#endif /* NETATALK */ 815 default: 816#ifdef IPX 817 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 818 return; 819#endif /* IPX */ 820#if defined(NETATALK) 821 if (ether_type > ETHERMTU) 822 goto discard; 823 l = mtod(m, struct llc *); 824 if (l->llc_dsap == LLC_SNAP_LSAP && 825 l->llc_ssap == LLC_SNAP_LSAP && 826 l->llc_control == LLC_UI) { 827 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code, 828 sizeof(at_org_code)) == 0 && 829 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) { 830 m_adj(m, LLC_SNAPFRAMELEN); 831 isr = NETISR_ATALK2; 832 break; 833 } 834 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code, 835 sizeof(aarp_org_code)) == 0 && 836 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) { 837 m_adj(m, LLC_SNAPFRAMELEN); 838 isr = NETISR_AARP; 839 break; 840 } 841 } 842#endif /* NETATALK */ 843 goto discard; 844 } 845 netisr_dispatch(isr, m); 846 return; 847 848discard: 849 /* 850 * Packet is to be discarded. If netgraph is present, 851 * hand the packet to it for last chance processing; 852 * otherwise dispose of it. 853 */ 854 if (IFP2AC(ifp)->ac_netgraph != NULL) { 855 KASSERT(ng_ether_input_orphan_p != NULL, 856 ("ng_ether_input_orphan_p is NULL")); 857 /* 858 * Put back the ethernet header so netgraph has a 859 * consistent view of inbound packets. 860 */ 861 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 862 (*ng_ether_input_orphan_p)(ifp, m); 863 return; 864 } 865 m_freem(m); 866} 867 868/* 869 * Convert Ethernet address to printable (loggable) representation. 870 * This routine is for compatibility; it's better to just use 871 * 872 * printf("%6D", <pointer to address>, ":"); 873 * 874 * since there's no static buffer involved. 875 */ 876char * 877ether_sprintf(const u_char *ap) 878{ 879 static char etherbuf[18]; 880 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":"); 881 return (etherbuf); 882} 883 884/* 885 * Perform common duties while attaching to interface list 886 */ 887void 888ether_ifattach(struct ifnet *ifp, const u_int8_t *lla) 889{ 890 int i; 891 struct ifaddr *ifa; 892 struct sockaddr_dl *sdl; 893 894 ifp->if_addrlen = ETHER_ADDR_LEN; 895 ifp->if_hdrlen = ETHER_HDR_LEN; 896 if_attach(ifp); 897 ifp->if_mtu = ETHERMTU; 898 ifp->if_output = ether_output; 899 ifp->if_input = ether_input; 900 ifp->if_resolvemulti = ether_resolvemulti; 901 if (ifp->if_baudrate == 0) 902 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 903 ifp->if_broadcastaddr = etherbroadcastaddr; 904 905 ifa = ifp->if_addr; 906 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__)); 907 sdl = (struct sockaddr_dl *)ifa->ifa_addr; 908 sdl->sdl_type = IFT_ETHER; 909 sdl->sdl_alen = ifp->if_addrlen; 910 bcopy(lla, LLADDR(sdl), ifp->if_addrlen); 911 912 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN); 913 if (ng_ether_attach_p != NULL) 914 (*ng_ether_attach_p)(ifp); 915 916 /* Announce Ethernet MAC address if non-zero. */ 917 for (i = 0; i < ifp->if_addrlen; i++) 918 if (lla[i] != 0) 919 break; 920 if (i != ifp->if_addrlen) 921 if_printf(ifp, "Ethernet address: %6D\n", lla, ":"); 922} 923 924/* 925 * Perform common duties while detaching an Ethernet interface 926 */ 927void 928ether_ifdetach(struct ifnet *ifp) 929{ 930 if (IFP2AC(ifp)->ac_netgraph != NULL) { 931 KASSERT(ng_ether_detach_p != NULL, 932 ("ng_ether_detach_p is NULL")); 933 (*ng_ether_detach_p)(ifp); 934 } 935 936 bpfdetach(ifp); 937 if_detach(ifp); 938} 939 940SYSCTL_DECL(_net_link); 941SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet"); 942#if defined(INET) || defined(INET6) 943SYSCTL_V_INT(V_NET, vnet_net, _net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW, 944 ether_ipfw, 0, "Pass ether pkts through firewall"); 945#endif 946 947#if 0 948/* 949 * This is for reference. We have a table-driven version 950 * of the little-endian crc32 generator, which is faster 951 * than the double-loop. 952 */ 953uint32_t 954ether_crc32_le(const uint8_t *buf, size_t len) 955{ 956 size_t i; 957 uint32_t crc; 958 int bit; 959 uint8_t data; 960 961 crc = 0xffffffff; /* initial value */ 962 963 for (i = 0; i < len; i++) { 964 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) { 965 carry = (crc ^ data) & 1; 966 crc >>= 1; 967 if (carry) 968 crc = (crc ^ ETHER_CRC_POLY_LE); 969 } 970 } 971 972 return (crc); 973} 974#else 975uint32_t 976ether_crc32_le(const uint8_t *buf, size_t len) 977{ 978 static const uint32_t crctab[] = { 979 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 980 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 981 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 982 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 983 }; 984 size_t i; 985 uint32_t crc; 986 987 crc = 0xffffffff; /* initial value */ 988 989 for (i = 0; i < len; i++) { 990 crc ^= buf[i]; 991 crc = (crc >> 4) ^ crctab[crc & 0xf]; 992 crc = (crc >> 4) ^ crctab[crc & 0xf]; 993 } 994 995 return (crc); 996} 997#endif 998 999uint32_t 1000ether_crc32_be(const uint8_t *buf, size_t len) 1001{ 1002 size_t i; 1003 uint32_t crc, carry; 1004 int bit; 1005 uint8_t data; 1006 1007 crc = 0xffffffff; /* initial value */ 1008 1009 for (i = 0; i < len; i++) { 1010 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) { 1011 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01); 1012 crc <<= 1; 1013 if (carry) 1014 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1015 } 1016 } 1017 1018 return (crc); 1019} 1020 1021int 1022ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data) 1023{ 1024 struct ifaddr *ifa = (struct ifaddr *) data; 1025 struct ifreq *ifr = (struct ifreq *) data; 1026 int error = 0; 1027 1028 switch (command) { 1029 case SIOCSIFADDR: 1030 ifp->if_flags |= IFF_UP; 1031 1032 switch (ifa->ifa_addr->sa_family) { 1033#ifdef INET 1034 case AF_INET: 1035 ifp->if_init(ifp->if_softc); /* before arpwhohas */ 1036 arp_ifinit(ifp, ifa); 1037 break; 1038#endif 1039#ifdef IPX 1040 /* 1041 * XXX - This code is probably wrong 1042 */ 1043 case AF_IPX: 1044 { 1045 struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr); 1046 1047 if (ipx_nullhost(*ina)) 1048 ina->x_host = 1049 *(union ipx_host *) 1050 IF_LLADDR(ifp); 1051 else { 1052 bcopy((caddr_t) ina->x_host.c_host, 1053 (caddr_t) IF_LLADDR(ifp), 1054 ETHER_ADDR_LEN); 1055 } 1056 1057 /* 1058 * Set new address 1059 */ 1060 ifp->if_init(ifp->if_softc); 1061 break; 1062 } 1063#endif 1064 default: 1065 ifp->if_init(ifp->if_softc); 1066 break; 1067 } 1068 break; 1069 1070 case SIOCGIFADDR: 1071 { 1072 struct sockaddr *sa; 1073 1074 sa = (struct sockaddr *) & ifr->ifr_data; 1075 bcopy(IF_LLADDR(ifp), 1076 (caddr_t) sa->sa_data, ETHER_ADDR_LEN); 1077 } 1078 break; 1079 1080 case SIOCSIFMTU: 1081 /* 1082 * Set the interface MTU. 1083 */ 1084 if (ifr->ifr_mtu > ETHERMTU) { 1085 error = EINVAL; 1086 } else { 1087 ifp->if_mtu = ifr->ifr_mtu; 1088 } 1089 break; 1090 default: 1091 error = EINVAL; /* XXX netbsd has ENOTTY??? */ 1092 break; 1093 } 1094 return (error); 1095} 1096 1097static int 1098ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa, 1099 struct sockaddr *sa) 1100{ 1101 struct sockaddr_dl *sdl; 1102#ifdef INET 1103 struct sockaddr_in *sin; 1104#endif 1105#ifdef INET6 1106 struct sockaddr_in6 *sin6; 1107#endif 1108 u_char *e_addr; 1109 1110 switch(sa->sa_family) { 1111 case AF_LINK: 1112 /* 1113 * No mapping needed. Just check that it's a valid MC address. 1114 */ 1115 sdl = (struct sockaddr_dl *)sa; 1116 e_addr = LLADDR(sdl); 1117 if (!ETHER_IS_MULTICAST(e_addr)) 1118 return EADDRNOTAVAIL; 1119 *llsa = 0; 1120 return 0; 1121 1122#ifdef INET 1123 case AF_INET: 1124 sin = (struct sockaddr_in *)sa; 1125 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) 1126 return EADDRNOTAVAIL; 1127 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, 1128 M_NOWAIT|M_ZERO); 1129 if (sdl == NULL) 1130 return ENOMEM; 1131 sdl->sdl_len = sizeof *sdl; 1132 sdl->sdl_family = AF_LINK; 1133 sdl->sdl_index = ifp->if_index; 1134 sdl->sdl_type = IFT_ETHER; 1135 sdl->sdl_alen = ETHER_ADDR_LEN; 1136 e_addr = LLADDR(sdl); 1137 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr); 1138 *llsa = (struct sockaddr *)sdl; 1139 return 0; 1140#endif 1141#ifdef INET6 1142 case AF_INET6: 1143 sin6 = (struct sockaddr_in6 *)sa; 1144 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1145 /* 1146 * An IP6 address of 0 means listen to all 1147 * of the Ethernet multicast address used for IP6. 1148 * (This is used for multicast routers.) 1149 */ 1150 ifp->if_flags |= IFF_ALLMULTI; 1151 *llsa = 0; 1152 return 0; 1153 } 1154 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) 1155 return EADDRNOTAVAIL; 1156 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, 1157 M_NOWAIT|M_ZERO); 1158 if (sdl == NULL) 1159 return (ENOMEM); 1160 sdl->sdl_len = sizeof *sdl; 1161 sdl->sdl_family = AF_LINK; 1162 sdl->sdl_index = ifp->if_index; 1163 sdl->sdl_type = IFT_ETHER; 1164 sdl->sdl_alen = ETHER_ADDR_LEN; 1165 e_addr = LLADDR(sdl); 1166 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr); 1167 *llsa = (struct sockaddr *)sdl; 1168 return 0; 1169#endif 1170 1171 default: 1172 /* 1173 * Well, the text isn't quite right, but it's the name 1174 * that counts... 1175 */ 1176 return EAFNOSUPPORT; 1177 } 1178} 1179 1180static void* 1181ether_alloc(u_char type, struct ifnet *ifp) 1182{ 1183 struct arpcom *ac; 1184 1185 ac = malloc(sizeof(struct arpcom), M_ARPCOM, M_WAITOK | M_ZERO); 1186 ac->ac_ifp = ifp; 1187 1188 return (ac); 1189} 1190 1191static void 1192ether_free(void *com, u_char type) 1193{ 1194 1195 free(com, M_ARPCOM); 1196} 1197 1198static int 1199ether_modevent(module_t mod, int type, void *data) 1200{ 1201 1202 switch (type) { 1203 case MOD_LOAD: 1204 if_register_com_alloc(IFT_ETHER, ether_alloc, ether_free); 1205 break; 1206 case MOD_UNLOAD: 1207 if_deregister_com_alloc(IFT_ETHER); 1208 break; 1209 default: 1210 return EOPNOTSUPP; 1211 } 1212 1213 return (0); 1214} 1215 1216static moduledata_t ether_mod = { 1217 "ether", 1218 ether_modevent, 1219 0 1220}; 1221 1222void 1223ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen) 1224{ 1225 struct ether_vlan_header vlan; 1226 struct mbuf mv, mb; 1227 1228 KASSERT((m->m_flags & M_VLANTAG) != 0, 1229 ("%s: vlan information not present", __func__)); 1230 KASSERT(m->m_len >= sizeof(struct ether_header), 1231 ("%s: mbuf not large enough for header", __func__)); 1232 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header)); 1233 vlan.evl_proto = vlan.evl_encap_proto; 1234 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN); 1235 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag); 1236 m->m_len -= sizeof(struct ether_header); 1237 m->m_data += sizeof(struct ether_header); 1238 /* 1239 * If a data link has been supplied by the caller, then we will need to 1240 * re-create a stack allocated mbuf chain with the following structure: 1241 * 1242 * (1) mbuf #1 will contain the supplied data link 1243 * (2) mbuf #2 will contain the vlan header 1244 * (3) mbuf #3 will contain the original mbuf's packet data 1245 * 1246 * Otherwise, submit the packet and vlan header via bpf_mtap2(). 1247 */ 1248 if (data != NULL) { 1249 mv.m_next = m; 1250 mv.m_data = (caddr_t)&vlan; 1251 mv.m_len = sizeof(vlan); 1252 mb.m_next = &mv; 1253 mb.m_data = data; 1254 mb.m_len = dlen; 1255 bpf_mtap(bp, &mb); 1256 } else 1257 bpf_mtap2(bp, &vlan, sizeof(vlan), m); 1258 m->m_len += sizeof(struct ether_header); 1259 m->m_data -= sizeof(struct ether_header); 1260} 1261 1262struct mbuf * 1263ether_vlanencap(struct mbuf *m, uint16_t tag) 1264{ 1265 struct ether_vlan_header *evl; 1266 1267 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_DONTWAIT); 1268 if (m == NULL) 1269 return (NULL); 1270 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */ 1271 1272 if (m->m_len < sizeof(*evl)) { 1273 m = m_pullup(m, sizeof(*evl)); 1274 if (m == NULL) 1275 return (NULL); 1276 } 1277 1278 /* 1279 * Transform the Ethernet header into an Ethernet header 1280 * with 802.1Q encapsulation. 1281 */ 1282 evl = mtod(m, struct ether_vlan_header *); 1283 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN, 1284 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN); 1285 evl->evl_encap_proto = htons(ETHERTYPE_VLAN); 1286 evl->evl_tag = htons(tag); 1287 return (m); 1288} 1289 1290DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY); 1291MODULE_VERSION(ether, 1); 1292