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 #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 #include <net/vnet.h> 67 68 #if defined(INET) || defined(INET6) 69 #include <netinet/in.h> 70 #include <netinet/in_var.h> 71 #include <netinet/if_ether.h> 72 #include <netinet/ip_fw.h> 73 #include <netinet/ip_dummynet.h> 74 #endif 75 #ifdef INET6 76 #include <netinet6/nd6.h> 77 #endif 78 79 #ifdef DEV_CARP 80 #include <netinet/ip_carp.h> 81 #endif 82 83 #ifdef IPX 84 #include <netipx/ipx.h> 85 #include <netipx/ipx_if.h> 86 #endif 87 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m); 88 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, 89 struct sockaddr *dst, short *tp, int *hlen); 90 91 #ifdef NETATALK 92 #include <netatalk/at.h> 93 #include <netatalk/at_var.h> 94 #include <netatalk/at_extern.h> 95 96 #define llc_snap_org_code llc_un.type_snap.org_code 97 #define llc_snap_ether_type llc_un.type_snap.ether_type 98 99 extern u_char at_org_code[3]; 100 extern u_char aarp_org_code[3]; 101 #endif /* NETATALK */ 102 103 #include <security/mac/mac_framework.h> 104 105 #ifdef CTASSERT 106 CTASSERT(sizeof (struct ether_header) == ETHER_ADDR_LEN * 2 + 2); 107 CTASSERT(sizeof (struct ether_addr) == ETHER_ADDR_LEN); 108 #endif 109 110 /* netgraph node hooks for ng_ether(4) */ 111 void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp); 112 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m); 113 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp); 114 void (*ng_ether_attach_p)(struct ifnet *ifp); 115 void (*ng_ether_detach_p)(struct ifnet *ifp); 116 117 void (*vlan_input_p)(struct ifnet *, struct mbuf *); 118 119 /* if_bridge(4) support */ 120 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *); 121 int (*bridge_output_p)(struct ifnet *, struct mbuf *, 122 struct sockaddr *, struct rtentry *); 123 void (*bridge_dn_p)(struct mbuf *, struct ifnet *); 124 125 /* if_lagg(4) support */ 126 struct mbuf *(*lagg_input_p)(struct ifnet *, struct mbuf *); 127 128 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] = 129 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 130 131 static int ether_resolvemulti(struct ifnet *, struct sockaddr **, 132 struct sockaddr *); 133 134 /* XXX: should be in an arp support file, not here */ 135 MALLOC_DEFINE(M_ARPCOM, "arpcom", "802.* interface internals"); 136 137 #define ETHER_IS_BROADCAST(addr) \ 138 (bcmp(etherbroadcastaddr, (addr), ETHER_ADDR_LEN) == 0) 139 140 #define senderr(e) do { error = (e); goto bad;} while (0) 141 142 #if defined(INET) || defined(INET6) 143 int 144 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 145 struct ip_fw **rule, int shared); 146 #ifdef VIMAGE_GLOBALS 147 static int ether_ipfw; 148 #endif 149 #endif 150 151 /* 152 * Ethernet output routine. 153 * Encapsulate a packet of type family for the local net. 154 * Use trailer local net encapsulation if enough data in first 155 * packet leaves a multiple of 512 bytes of data in remainder. 156 */ 157 int 158 ether_output(struct ifnet *ifp, struct mbuf *m, 159 struct sockaddr *dst, struct rtentry *rt0) 160 { 161 short type; 162 int error, hdrcmplt = 0; 163 u_char esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN]; 164 struct ether_header *eh; 165 struct pf_mtag *t; 166 int loop_copy = 1; 167 int hlen; /* link layer header length */ 168 169 #ifdef MAC 170 error = mac_ifnet_check_transmit(ifp, m); 171 if (error) 172 senderr(error); 173 #endif 174 175 M_PROFILE(m); 176 if (ifp->if_flags & IFF_MONITOR) 177 senderr(ENETDOWN); 178 if (!((ifp->if_flags & IFF_UP) && 179 (ifp->if_drv_flags & IFF_DRV_RUNNING))) 180 senderr(ENETDOWN); 181 182 hlen = ETHER_HDR_LEN; 183 switch (dst->sa_family) { 184 #ifdef INET 185 case AF_INET: 186 error = arpresolve(ifp, rt0, m, dst, edst); 187 if (error) 188 return (error == EWOULDBLOCK ? 0 : error); 189 type = htons(ETHERTYPE_IP); 190 break; 191 case AF_ARP: 192 { 193 struct arphdr *ah; 194 ah = mtod(m, struct arphdr *); 195 ah->ar_hrd = htons(ARPHRD_ETHER); 196 197 loop_copy = 0; /* if this is for us, don't do it */ 198 199 switch(ntohs(ah->ar_op)) { 200 case ARPOP_REVREQUEST: 201 case ARPOP_REVREPLY: 202 type = htons(ETHERTYPE_REVARP); 203 break; 204 case ARPOP_REQUEST: 205 case ARPOP_REPLY: 206 default: 207 type = htons(ETHERTYPE_ARP); 208 break; 209 } 210 211 if (m->m_flags & M_BCAST) 212 bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN); 213 else 214 bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN); 215 216 } 217 break; 218 #endif 219 #ifdef INET6 220 case AF_INET6: 221 error = nd6_storelladdr(ifp, rt0, m, dst, (u_char *)edst); 222 if (error) 223 return error; 224 type = htons(ETHERTYPE_IPV6); 225 break; 226 #endif 227 #ifdef IPX 228 case AF_IPX: 229 if (ef_outputp) { 230 error = ef_outputp(ifp, &m, dst, &type, &hlen); 231 if (error) 232 goto bad; 233 } else 234 type = htons(ETHERTYPE_IPX); 235 bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host), 236 (caddr_t)edst, sizeof (edst)); 237 break; 238 #endif 239 #ifdef NETATALK 240 case AF_APPLETALK: 241 { 242 struct at_ifaddr *aa; 243 244 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) 245 senderr(EHOSTUNREACH); /* XXX */ 246 if (!aarpresolve(ifp, m, (struct sockaddr_at *)dst, edst)) 247 return (0); 248 /* 249 * In the phase 2 case, need to prepend an mbuf for the llc header. 250 */ 251 if ( aa->aa_flags & AFA_PHASE2 ) { 252 struct llc llc; 253 254 M_PREPEND(m, LLC_SNAPFRAMELEN, M_DONTWAIT); 255 if (m == NULL) 256 senderr(ENOBUFS); 257 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 258 llc.llc_control = LLC_UI; 259 bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code)); 260 llc.llc_snap_ether_type = htons( ETHERTYPE_AT ); 261 bcopy(&llc, mtod(m, caddr_t), LLC_SNAPFRAMELEN); 262 type = htons(m->m_pkthdr.len); 263 hlen = LLC_SNAPFRAMELEN + ETHER_HDR_LEN; 264 } else { 265 type = htons(ETHERTYPE_AT); 266 } 267 break; 268 } 269 #endif /* NETATALK */ 270 271 case pseudo_AF_HDRCMPLT: 272 hdrcmplt = 1; 273 eh = (struct ether_header *)dst->sa_data; 274 (void)memcpy(esrc, eh->ether_shost, sizeof (esrc)); 275 /* FALLTHROUGH */ 276 277 case AF_UNSPEC: 278 loop_copy = 0; /* if this is for us, don't do it */ 279 eh = (struct ether_header *)dst->sa_data; 280 (void)memcpy(edst, eh->ether_dhost, sizeof (edst)); 281 type = eh->ether_type; 282 break; 283 284 default: 285 if_printf(ifp, "can't handle af%d\n", dst->sa_family); 286 senderr(EAFNOSUPPORT); 287 } 288 289 /* 290 * Add local net header. If no space in first mbuf, 291 * allocate another. 292 */ 293 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 294 if (m == NULL) 295 senderr(ENOBUFS); 296 eh = mtod(m, struct ether_header *); 297 (void)memcpy(&eh->ether_type, &type, 298 sizeof(eh->ether_type)); 299 (void)memcpy(eh->ether_dhost, edst, sizeof (edst)); 300 if (hdrcmplt) 301 (void)memcpy(eh->ether_shost, esrc, 302 sizeof(eh->ether_shost)); 303 else 304 (void)memcpy(eh->ether_shost, IF_LLADDR(ifp), 305 sizeof(eh->ether_shost)); 306 307 /* 308 * If a simplex interface, and the packet is being sent to our 309 * Ethernet address or a broadcast address, loopback a copy. 310 * XXX To make a simplex device behave exactly like a duplex 311 * device, we should copy in the case of sending to our own 312 * ethernet address (thus letting the original actually appear 313 * on the wire). However, we don't do that here for security 314 * reasons and compatibility with the original behavior. 315 */ 316 if ((ifp->if_flags & IFF_SIMPLEX) && loop_copy && 317 ((t = pf_find_mtag(m)) == NULL || !t->routed)) { 318 int csum_flags = 0; 319 320 if (m->m_pkthdr.csum_flags & CSUM_IP) 321 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID); 322 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) 323 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR); 324 325 if (m->m_flags & M_BCAST) { 326 struct mbuf *n; 327 328 /* 329 * Because if_simloop() modifies the packet, we need a 330 * writable copy through m_dup() instead of a readonly 331 * one as m_copy[m] would give us. The alternative would 332 * be to modify if_simloop() to handle the readonly mbuf, 333 * but performancewise it is mostly equivalent (trading 334 * extra data copying vs. extra locking). 335 * 336 * XXX This is a local workaround. A number of less 337 * often used kernel parts suffer from the same bug. 338 * See PR kern/105943 for a proposed general solution. 339 */ 340 if ((n = m_dup(m, M_DONTWAIT)) != NULL) { 341 n->m_pkthdr.csum_flags |= csum_flags; 342 if (csum_flags & CSUM_DATA_VALID) 343 n->m_pkthdr.csum_data = 0xffff; 344 (void)if_simloop(ifp, n, dst->sa_family, hlen); 345 } else 346 ifp->if_iqdrops++; 347 } else if (bcmp(eh->ether_dhost, eh->ether_shost, 348 ETHER_ADDR_LEN) == 0) { 349 m->m_pkthdr.csum_flags |= csum_flags; 350 if (csum_flags & CSUM_DATA_VALID) 351 m->m_pkthdr.csum_data = 0xffff; 352 (void) if_simloop(ifp, m, dst->sa_family, hlen); 353 return (0); /* XXX */ 354 } 355 } 356 357 /* 358 * Bridges require special output handling. 359 */ 360 if (ifp->if_bridge) { 361 BRIDGE_OUTPUT(ifp, m, error); 362 return (error); 363 } 364 365 #ifdef DEV_CARP 366 if (ifp->if_carp && 367 (error = carp_output(ifp, m, dst, NULL))) 368 goto bad; 369 #endif 370 371 /* Handle ng_ether(4) processing, if any */ 372 if (IFP2AC(ifp)->ac_netgraph != NULL) { 373 KASSERT(ng_ether_output_p != NULL, 374 ("ng_ether_output_p is NULL")); 375 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) { 376 bad: if (m != NULL) 377 m_freem(m); 378 return (error); 379 } 380 if (m == NULL) 381 return (0); 382 } 383 384 /* Continue with link-layer output */ 385 return ether_output_frame(ifp, m); 386 } 387 388 /* 389 * Ethernet link layer output routine to send a raw frame to the device. 390 * 391 * This assumes that the 14 byte Ethernet header is present and contiguous 392 * in the first mbuf (if BRIDGE'ing). 393 */ 394 int 395 ether_output_frame(struct ifnet *ifp, struct mbuf *m) 396 { 397 #if defined(INET) || defined(INET6) 398 INIT_VNET_NET(ifp->if_vnet); 399 struct ip_fw *rule = ip_dn_claim_rule(m); 400 401 if (IPFW_LOADED && V_ether_ipfw != 0) { 402 if (ether_ipfw_chk(&m, ifp, &rule, 0) == 0) { 403 if (m) { 404 m_freem(m); 405 return EACCES; /* pkt dropped */ 406 } else 407 return 0; /* consumed e.g. in a pipe */ 408 } 409 } 410 #endif 411 412 /* 413 * Queue message on interface, update output statistics if 414 * successful, and start output if interface not yet active. 415 */ 416 return ((ifp->if_transmit)(ifp, m)); 417 } 418 419 #if defined(INET) || defined(INET6) 420 /* 421 * ipfw processing for ethernet packets (in and out). 422 * The second parameter is NULL from ether_demux, and ifp from 423 * ether_output_frame. 424 */ 425 int 426 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 427 struct ip_fw **rule, int shared) 428 { 429 INIT_VNET_IPFW(dst->if_vnet); 430 struct ether_header *eh; 431 struct ether_header save_eh; 432 struct mbuf *m; 433 int i; 434 struct ip_fw_args args; 435 436 if (*rule != NULL && V_fw_one_pass) 437 return 1; /* dummynet packet, already partially processed */ 438 439 /* 440 * I need some amt of data to be contiguous, and in case others need 441 * the packet (shared==1) also better be in the first mbuf. 442 */ 443 m = *m0; 444 i = min( m->m_pkthdr.len, max_protohdr); 445 if ( shared || m->m_len < i) { 446 m = m_pullup(m, i); 447 if (m == NULL) { 448 *m0 = m; 449 return 0; 450 } 451 } 452 eh = mtod(m, struct ether_header *); 453 save_eh = *eh; /* save copy for restore below */ 454 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */ 455 456 args.m = m; /* the packet we are looking at */ 457 args.oif = dst; /* destination, if any */ 458 args.rule = *rule; /* matching rule to restart */ 459 args.next_hop = NULL; /* we do not support forward yet */ 460 args.eh = &save_eh; /* MAC header for bridged/MAC packets */ 461 args.inp = NULL; /* used by ipfw uid/gid/jail rules */ 462 i = ip_fw_chk_ptr(&args); 463 m = args.m; 464 if (m != NULL) { 465 /* 466 * Restore Ethernet header, as needed, in case the 467 * mbuf chain was replaced by ipfw. 468 */ 469 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 470 if (m == NULL) { 471 *m0 = m; 472 return 0; 473 } 474 if (eh != mtod(m, struct ether_header *)) 475 bcopy(&save_eh, mtod(m, struct ether_header *), 476 ETHER_HDR_LEN); 477 } 478 *m0 = m; 479 *rule = args.rule; 480 481 if (i == IP_FW_DENY) /* drop */ 482 return 0; 483 484 KASSERT(m != NULL, ("ether_ipfw_chk: m is NULL")); 485 486 if (i == IP_FW_PASS) /* a PASS rule. */ 487 return 1; 488 489 if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) { 490 /* 491 * Pass the pkt to dummynet, which consumes it. 492 * If shared, make a copy and keep the original. 493 */ 494 if (shared) { 495 m = m_copypacket(m, M_DONTWAIT); 496 if (m == NULL) 497 return 0; 498 } else { 499 /* 500 * Pass the original to dummynet and 501 * nothing back to the caller 502 */ 503 *m0 = NULL ; 504 } 505 ip_dn_io_ptr(&m, dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args); 506 return 0; 507 } 508 /* 509 * XXX at some point add support for divert/forward actions. 510 * If none of the above matches, we have to drop the pkt. 511 */ 512 return 0; 513 } 514 #endif 515 516 /* 517 * Process a received Ethernet packet; the packet is in the 518 * mbuf chain m with the ethernet header at the front. 519 */ 520 static void 521 ether_input(struct ifnet *ifp, struct mbuf *m) 522 { 523 struct ether_header *eh; 524 u_short etype; 525 526 if ((ifp->if_flags & IFF_UP) == 0) { 527 m_freem(m); 528 return; 529 } 530 #ifdef DIAGNOSTIC 531 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 532 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n"); 533 m_freem(m); 534 return; 535 } 536 #endif 537 /* 538 * Do consistency checks to verify assumptions 539 * made by code past this point. 540 */ 541 if ((m->m_flags & M_PKTHDR) == 0) { 542 if_printf(ifp, "discard frame w/o packet header\n"); 543 ifp->if_ierrors++; 544 m_freem(m); 545 return; 546 } 547 if (m->m_len < ETHER_HDR_LEN) { 548 /* XXX maybe should pullup? */ 549 if_printf(ifp, "discard frame w/o leading ethernet " 550 "header (len %u pkt len %u)\n", 551 m->m_len, m->m_pkthdr.len); 552 ifp->if_ierrors++; 553 m_freem(m); 554 return; 555 } 556 eh = mtod(m, struct ether_header *); 557 etype = ntohs(eh->ether_type); 558 if (m->m_pkthdr.rcvif == NULL) { 559 if_printf(ifp, "discard frame w/o interface pointer\n"); 560 ifp->if_ierrors++; 561 m_freem(m); 562 return; 563 } 564 #ifdef DIAGNOSTIC 565 if (m->m_pkthdr.rcvif != ifp) { 566 if_printf(ifp, "Warning, frame marked as received on %s\n", 567 m->m_pkthdr.rcvif->if_xname); 568 } 569 #endif 570 571 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 572 if (ETHER_IS_BROADCAST(eh->ether_dhost)) 573 m->m_flags |= M_BCAST; 574 else 575 m->m_flags |= M_MCAST; 576 ifp->if_imcasts++; 577 } 578 579 #ifdef MAC 580 /* 581 * Tag the mbuf with an appropriate MAC label before any other 582 * consumers can get to it. 583 */ 584 mac_ifnet_create_mbuf(ifp, m); 585 #endif 586 587 /* 588 * Give bpf a chance at the packet. 589 */ 590 ETHER_BPF_MTAP(ifp, m); 591 592 /* 593 * If the CRC is still on the packet, trim it off. We do this once 594 * and once only in case we are re-entered. Nothing else on the 595 * Ethernet receive path expects to see the FCS. 596 */ 597 if (m->m_flags & M_HASFCS) { 598 m_adj(m, -ETHER_CRC_LEN); 599 m->m_flags &= ~M_HASFCS; 600 } 601 602 ifp->if_ibytes += m->m_pkthdr.len; 603 604 /* Allow monitor mode to claim this frame, after stats are updated. */ 605 if (ifp->if_flags & IFF_MONITOR) { 606 m_freem(m); 607 return; 608 } 609 610 /* Handle input from a lagg(4) port */ 611 if (ifp->if_type == IFT_IEEE8023ADLAG) { 612 KASSERT(lagg_input_p != NULL, 613 ("%s: if_lagg not loaded!", __func__)); 614 m = (*lagg_input_p)(ifp, m); 615 if (m != NULL) 616 ifp = m->m_pkthdr.rcvif; 617 else 618 return; 619 } 620 621 /* 622 * If the hardware did not process an 802.1Q tag, do this now, 623 * to allow 802.1P priority frames to be passed to the main input 624 * path correctly. 625 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels. 626 */ 627 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) { 628 struct ether_vlan_header *evl; 629 630 if (m->m_len < sizeof(*evl) && 631 (m = m_pullup(m, sizeof(*evl))) == NULL) { 632 #ifdef DIAGNOSTIC 633 if_printf(ifp, "cannot pullup VLAN header\n"); 634 #endif 635 ifp->if_ierrors++; 636 m_freem(m); 637 return; 638 } 639 640 evl = mtod(m, struct ether_vlan_header *); 641 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag); 642 m->m_flags |= M_VLANTAG; 643 644 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN, 645 ETHER_HDR_LEN - ETHER_TYPE_LEN); 646 m_adj(m, ETHER_VLAN_ENCAP_LEN); 647 } 648 649 /* Allow ng_ether(4) to claim this frame. */ 650 if (IFP2AC(ifp)->ac_netgraph != NULL) { 651 KASSERT(ng_ether_input_p != NULL, 652 ("%s: ng_ether_input_p is NULL", __func__)); 653 m->m_flags &= ~M_PROMISC; 654 (*ng_ether_input_p)(ifp, &m); 655 if (m == NULL) 656 return; 657 } 658 659 /* 660 * Allow if_bridge(4) to claim this frame. 661 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it 662 * and the frame should be delivered locally. 663 */ 664 if (ifp->if_bridge != NULL) { 665 m->m_flags &= ~M_PROMISC; 666 BRIDGE_INPUT(ifp, m); 667 if (m == NULL) 668 return; 669 } 670 671 #ifdef DEV_CARP 672 /* 673 * Clear M_PROMISC on frame so that carp(4) will see it when the 674 * mbuf flows up to Layer 3. 675 * FreeBSD's implementation of carp(4) uses the inprotosw 676 * to dispatch IPPROTO_CARP. carp(4) also allocates its own 677 * Ethernet addresses of the form 00:00:5e:00:01:xx, which 678 * is outside the scope of the M_PROMISC test below. 679 * TODO: Maintain a hash table of ethernet addresses other than 680 * ether_dhost which may be active on this ifp. 681 */ 682 if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost)) { 683 m->m_flags &= ~M_PROMISC; 684 } else 685 #endif 686 { 687 /* 688 * If the frame received was not for our MAC address, set the 689 * M_PROMISC flag on the mbuf chain. The frame may need to 690 * be seen by the rest of the Ethernet input path in case of 691 * re-entry (e.g. bridge, vlan, netgraph) but should not be 692 * seen by upper protocol layers. 693 */ 694 if (!ETHER_IS_MULTICAST(eh->ether_dhost) && 695 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0) 696 m->m_flags |= M_PROMISC; 697 } 698 699 /* First chunk of an mbuf contains good entropy */ 700 if (harvest.ethernet) 701 random_harvest(m, 16, 3, 0, RANDOM_NET); 702 703 ether_demux(ifp, m); 704 } 705 706 /* 707 * Upper layer processing for a received Ethernet packet. 708 */ 709 void 710 ether_demux(struct ifnet *ifp, struct mbuf *m) 711 { 712 struct ether_header *eh; 713 int isr; 714 u_short ether_type; 715 #if defined(NETATALK) 716 struct llc *l; 717 #endif 718 719 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__)); 720 721 #if defined(INET) || defined(INET6) 722 INIT_VNET_NET(ifp->if_vnet); 723 /* 724 * Allow dummynet and/or ipfw to claim the frame. 725 * Do not do this for PROMISC frames in case we are re-entered. 726 */ 727 if (IPFW_LOADED && V_ether_ipfw != 0 && !(m->m_flags & M_PROMISC)) { 728 struct ip_fw *rule = ip_dn_claim_rule(m); 729 730 if (ether_ipfw_chk(&m, NULL, &rule, 0) == 0) { 731 if (m) 732 m_freem(m); /* dropped; free mbuf chain */ 733 return; /* consumed */ 734 } 735 } 736 #endif 737 eh = mtod(m, struct ether_header *); 738 ether_type = ntohs(eh->ether_type); 739 740 /* 741 * If this frame has a VLAN tag other than 0, call vlan_input() 742 * if its module is loaded. Otherwise, drop. 743 */ 744 if ((m->m_flags & M_VLANTAG) && 745 EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) { 746 if (ifp->if_vlantrunk == NULL) { 747 ifp->if_noproto++; 748 m_freem(m); 749 return; 750 } 751 KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!", 752 __func__)); 753 /* Clear before possibly re-entering ether_input(). */ 754 m->m_flags &= ~M_PROMISC; 755 (*vlan_input_p)(ifp, m); 756 return; 757 } 758 759 /* 760 * Pass promiscuously received frames to the upper layer if the user 761 * requested this by setting IFF_PPROMISC. Otherwise, drop them. 762 */ 763 if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) { 764 m_freem(m); 765 return; 766 } 767 768 /* 769 * Reset layer specific mbuf flags to avoid confusing upper layers. 770 * Strip off Ethernet header. 771 */ 772 m->m_flags &= ~M_VLANTAG; 773 m->m_flags &= ~(M_PROTOFLAGS); 774 m_adj(m, ETHER_HDR_LEN); 775 776 /* 777 * Dispatch frame to upper layer. 778 */ 779 switch (ether_type) { 780 #ifdef INET 781 case ETHERTYPE_IP: 782 if ((m = ip_fastforward(m)) == NULL) 783 return; 784 isr = NETISR_IP; 785 break; 786 787 case ETHERTYPE_ARP: 788 if (ifp->if_flags & IFF_NOARP) { 789 /* Discard packet if ARP is disabled on interface */ 790 m_freem(m); 791 return; 792 } 793 isr = NETISR_ARP; 794 break; 795 #endif 796 #ifdef IPX 797 case ETHERTYPE_IPX: 798 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 799 return; 800 isr = NETISR_IPX; 801 break; 802 #endif 803 #ifdef INET6 804 case ETHERTYPE_IPV6: 805 isr = NETISR_IPV6; 806 break; 807 #endif 808 #ifdef NETATALK 809 case ETHERTYPE_AT: 810 isr = NETISR_ATALK1; 811 break; 812 case ETHERTYPE_AARP: 813 isr = NETISR_AARP; 814 break; 815 #endif /* NETATALK */ 816 default: 817 #ifdef IPX 818 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 819 return; 820 #endif /* IPX */ 821 #if defined(NETATALK) 822 if (ether_type > ETHERMTU) 823 goto discard; 824 l = mtod(m, struct llc *); 825 if (l->llc_dsap == LLC_SNAP_LSAP && 826 l->llc_ssap == LLC_SNAP_LSAP && 827 l->llc_control == LLC_UI) { 828 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code, 829 sizeof(at_org_code)) == 0 && 830 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) { 831 m_adj(m, LLC_SNAPFRAMELEN); 832 isr = NETISR_ATALK2; 833 break; 834 } 835 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code, 836 sizeof(aarp_org_code)) == 0 && 837 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) { 838 m_adj(m, LLC_SNAPFRAMELEN); 839 isr = NETISR_AARP; 840 break; 841 } 842 } 843 #endif /* NETATALK */ 844 goto discard; 845 } 846 netisr_dispatch(isr, m); 847 return; 848 849 discard: 850 /* 851 * Packet is to be discarded. If netgraph is present, 852 * hand the packet to it for last chance processing; 853 * otherwise dispose of it. 854 */ 855 if (IFP2AC(ifp)->ac_netgraph != NULL) { 856 KASSERT(ng_ether_input_orphan_p != NULL, 857 ("ng_ether_input_orphan_p is NULL")); 858 /* 859 * Put back the ethernet header so netgraph has a 860 * consistent view of inbound packets. 861 */ 862 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT); 863 (*ng_ether_input_orphan_p)(ifp, m); 864 return; 865 } 866 m_freem(m); 867 } 868 869 /* 870 * Convert Ethernet address to printable (loggable) representation. 871 * This routine is for compatibility; it's better to just use 872 * 873 * printf("%6D", <pointer to address>, ":"); 874 * 875 * since there's no static buffer involved. 876 */ 877 char * 878 ether_sprintf(const u_char *ap) 879 { 880 static char etherbuf[18]; 881 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":"); 882 return (etherbuf); 883 } 884 885 /* 886 * Perform common duties while attaching to interface list 887 */ 888 void 889 ether_ifattach(struct ifnet *ifp, const u_int8_t *lla) 890 { 891 int i; 892 struct ifaddr *ifa; 893 struct sockaddr_dl *sdl; 894 895 ifp->if_addrlen = ETHER_ADDR_LEN; 896 ifp->if_hdrlen = ETHER_HDR_LEN; 897 if_attach(ifp); 898 ifp->if_mtu = ETHERMTU; 899 ifp->if_output = ether_output; 900 ifp->if_input = ether_input; 901 ifp->if_resolvemulti = ether_resolvemulti; 902 if (ifp->if_baudrate == 0) 903 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 904 ifp->if_broadcastaddr = etherbroadcastaddr; 905 906 ifa = ifp->if_addr; 907 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__)); 908 sdl = (struct sockaddr_dl *)ifa->ifa_addr; 909 sdl->sdl_type = IFT_ETHER; 910 sdl->sdl_alen = ifp->if_addrlen; 911 bcopy(lla, LLADDR(sdl), ifp->if_addrlen); 912 913 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN); 914 if (ng_ether_attach_p != NULL) 915 (*ng_ether_attach_p)(ifp); 916 917 /* Announce Ethernet MAC address if non-zero. */ 918 for (i = 0; i < ifp->if_addrlen; i++) 919 if (lla[i] != 0) 920 break; 921 if (i != ifp->if_addrlen) 922 if_printf(ifp, "Ethernet address: %6D\n", lla, ":"); 923 } 924 925 /* 926 * Perform common duties while detaching an Ethernet interface 927 */ 928 void 929 ether_ifdetach(struct ifnet *ifp) 930 { 931 if (IFP2AC(ifp)->ac_netgraph != NULL) { 932 KASSERT(ng_ether_detach_p != NULL, 933 ("ng_ether_detach_p is NULL")); 934 (*ng_ether_detach_p)(ifp); 935 } 936 937 bpfdetach(ifp); 938 if_detach(ifp); 939 } 940 941 SYSCTL_DECL(_net_link); 942 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet"); 943 #if defined(INET) || defined(INET6) 944 SYSCTL_V_INT(V_NET, vnet_net, _net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW, 945 ether_ipfw, 0, "Pass ether pkts through firewall"); 946 #endif 947 948 #if 0 949 /* 950 * This is for reference. We have a table-driven version 951 * of the little-endian crc32 generator, which is faster 952 * than the double-loop. 953 */ 954 uint32_t 955 ether_crc32_le(const uint8_t *buf, size_t len) 956 { 957 size_t i; 958 uint32_t crc; 959 int bit; 960 uint8_t data; 961 962 crc = 0xffffffff; /* initial value */ 963 964 for (i = 0; i < len; i++) { 965 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) { 966 carry = (crc ^ data) & 1; 967 crc >>= 1; 968 if (carry) 969 crc = (crc ^ ETHER_CRC_POLY_LE); 970 } 971 } 972 973 return (crc); 974 } 975 #else 976 uint32_t 977 ether_crc32_le(const uint8_t *buf, size_t len) 978 { 979 static const uint32_t crctab[] = { 980 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 981 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 982 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 983 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 984 }; 985 size_t i; 986 uint32_t crc; 987 988 crc = 0xffffffff; /* initial value */ 989 990 for (i = 0; i < len; i++) { 991 crc ^= buf[i]; 992 crc = (crc >> 4) ^ crctab[crc & 0xf]; 993 crc = (crc >> 4) ^ crctab[crc & 0xf]; 994 } 995 996 return (crc); 997 } 998 #endif 999 1000 uint32_t 1001 ether_crc32_be(const uint8_t *buf, size_t len) 1002 { 1003 size_t i; 1004 uint32_t crc, carry; 1005 int bit; 1006 uint8_t data; 1007 1008 crc = 0xffffffff; /* initial value */ 1009 1010 for (i = 0; i < len; i++) { 1011 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) { 1012 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01); 1013 crc <<= 1; 1014 if (carry) 1015 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1016 } 1017 } 1018 1019 return (crc); 1020 } 1021 1022 int 1023 ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data) 1024 { 1025 struct ifaddr *ifa = (struct ifaddr *) data; 1026 struct ifreq *ifr = (struct ifreq *) data; 1027 int error = 0; 1028 1029 switch (command) { 1030 case SIOCSIFADDR: 1031 ifp->if_flags |= IFF_UP; 1032 1033 switch (ifa->ifa_addr->sa_family) { 1034 #ifdef INET 1035 case AF_INET: 1036 ifp->if_init(ifp->if_softc); /* before arpwhohas */ 1037 arp_ifinit(ifp, ifa); 1038 break; 1039 #endif 1040 #ifdef IPX 1041 /* 1042 * XXX - This code is probably wrong 1043 */ 1044 case AF_IPX: 1045 { 1046 struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr); 1047 1048 if (ipx_nullhost(*ina)) 1049 ina->x_host = 1050 *(union ipx_host *) 1051 IF_LLADDR(ifp); 1052 else { 1053 bcopy((caddr_t) ina->x_host.c_host, 1054 (caddr_t) IF_LLADDR(ifp), 1055 ETHER_ADDR_LEN); 1056 } 1057 1058 /* 1059 * Set new address 1060 */ 1061 ifp->if_init(ifp->if_softc); 1062 break; 1063 } 1064 #endif 1065 default: 1066 ifp->if_init(ifp->if_softc); 1067 break; 1068 } 1069 break; 1070 1071 case SIOCGIFADDR: 1072 { 1073 struct sockaddr *sa; 1074 1075 sa = (struct sockaddr *) & ifr->ifr_data; 1076 bcopy(IF_LLADDR(ifp), 1077 (caddr_t) sa->sa_data, ETHER_ADDR_LEN); 1078 } 1079 break; 1080 1081 case SIOCSIFMTU: 1082 /* 1083 * Set the interface MTU. 1084 */ 1085 if (ifr->ifr_mtu > ETHERMTU) { 1086 error = EINVAL; 1087 } else { 1088 ifp->if_mtu = ifr->ifr_mtu; 1089 } 1090 break; 1091 default: 1092 error = EINVAL; /* XXX netbsd has ENOTTY??? */ 1093 break; 1094 } 1095 return (error); 1096 } 1097 1098 static int 1099 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa, 1100 struct sockaddr *sa) 1101 { 1102 struct sockaddr_dl *sdl; 1103 #ifdef INET 1104 struct sockaddr_in *sin; 1105 #endif 1106 #ifdef INET6 1107 struct sockaddr_in6 *sin6; 1108 #endif 1109 u_char *e_addr; 1110 1111 switch(sa->sa_family) { 1112 case AF_LINK: 1113 /* 1114 * No mapping needed. Just check that it's a valid MC address. 1115 */ 1116 sdl = (struct sockaddr_dl *)sa; 1117 e_addr = LLADDR(sdl); 1118 if (!ETHER_IS_MULTICAST(e_addr)) 1119 return EADDRNOTAVAIL; 1120 *llsa = 0; 1121 return 0; 1122 1123 #ifdef INET 1124 case AF_INET: 1125 sin = (struct sockaddr_in *)sa; 1126 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) 1127 return EADDRNOTAVAIL; 1128 sdl = malloc(sizeof *sdl, M_IFMADDR, 1129 M_NOWAIT|M_ZERO); 1130 if (sdl == NULL) 1131 return ENOMEM; 1132 sdl->sdl_len = sizeof *sdl; 1133 sdl->sdl_family = AF_LINK; 1134 sdl->sdl_index = ifp->if_index; 1135 sdl->sdl_type = IFT_ETHER; 1136 sdl->sdl_alen = ETHER_ADDR_LEN; 1137 e_addr = LLADDR(sdl); 1138 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr); 1139 *llsa = (struct sockaddr *)sdl; 1140 return 0; 1141 #endif 1142 #ifdef INET6 1143 case AF_INET6: 1144 sin6 = (struct sockaddr_in6 *)sa; 1145 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1146 /* 1147 * An IP6 address of 0 means listen to all 1148 * of the Ethernet multicast address used for IP6. 1149 * (This is used for multicast routers.) 1150 */ 1151 ifp->if_flags |= IFF_ALLMULTI; 1152 *llsa = 0; 1153 return 0; 1154 } 1155 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) 1156 return EADDRNOTAVAIL; 1157 sdl = malloc(sizeof *sdl, M_IFMADDR, 1158 M_NOWAIT|M_ZERO); 1159 if (sdl == NULL) 1160 return (ENOMEM); 1161 sdl->sdl_len = sizeof *sdl; 1162 sdl->sdl_family = AF_LINK; 1163 sdl->sdl_index = ifp->if_index; 1164 sdl->sdl_type = IFT_ETHER; 1165 sdl->sdl_alen = ETHER_ADDR_LEN; 1166 e_addr = LLADDR(sdl); 1167 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr); 1168 *llsa = (struct sockaddr *)sdl; 1169 return 0; 1170 #endif 1171 1172 default: 1173 /* 1174 * Well, the text isn't quite right, but it's the name 1175 * that counts... 1176 */ 1177 return EAFNOSUPPORT; 1178 } 1179 } 1180 1181 static void* 1182 ether_alloc(u_char type, struct ifnet *ifp) 1183 { 1184 struct arpcom *ac; 1185 1186 ac = malloc(sizeof(struct arpcom), M_ARPCOM, M_WAITOK | M_ZERO); 1187 ac->ac_ifp = ifp; 1188 1189 return (ac); 1190 } 1191 1192 static void 1193 ether_free(void *com, u_char type) 1194 { 1195 1196 free(com, M_ARPCOM); 1197 } 1198 1199 static int 1200 ether_modevent(module_t mod, int type, void *data) 1201 { 1202 1203 switch (type) { 1204 case MOD_LOAD: 1205 if_register_com_alloc(IFT_ETHER, ether_alloc, ether_free); 1206 break; 1207 case MOD_UNLOAD: 1208 if_deregister_com_alloc(IFT_ETHER); 1209 break; 1210 default: 1211 return EOPNOTSUPP; 1212 } 1213 1214 return (0); 1215 } 1216 1217 static moduledata_t ether_mod = { 1218 "ether", 1219 ether_modevent, 1220 0 1221 }; 1222 1223 void 1224 ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen) 1225 { 1226 struct ether_vlan_header vlan; 1227 struct mbuf mv, mb; 1228 1229 KASSERT((m->m_flags & M_VLANTAG) != 0, 1230 ("%s: vlan information not present", __func__)); 1231 KASSERT(m->m_len >= sizeof(struct ether_header), 1232 ("%s: mbuf not large enough for header", __func__)); 1233 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header)); 1234 vlan.evl_proto = vlan.evl_encap_proto; 1235 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN); 1236 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag); 1237 m->m_len -= sizeof(struct ether_header); 1238 m->m_data += sizeof(struct ether_header); 1239 /* 1240 * If a data link has been supplied by the caller, then we will need to 1241 * re-create a stack allocated mbuf chain with the following structure: 1242 * 1243 * (1) mbuf #1 will contain the supplied data link 1244 * (2) mbuf #2 will contain the vlan header 1245 * (3) mbuf #3 will contain the original mbuf's packet data 1246 * 1247 * Otherwise, submit the packet and vlan header via bpf_mtap2(). 1248 */ 1249 if (data != NULL) { 1250 mv.m_next = m; 1251 mv.m_data = (caddr_t)&vlan; 1252 mv.m_len = sizeof(vlan); 1253 mb.m_next = &mv; 1254 mb.m_data = data; 1255 mb.m_len = dlen; 1256 bpf_mtap(bp, &mb); 1257 } else 1258 bpf_mtap2(bp, &vlan, sizeof(vlan), m); 1259 m->m_len += sizeof(struct ether_header); 1260 m->m_data -= sizeof(struct ether_header); 1261 } 1262 1263 struct mbuf * 1264 ether_vlanencap(struct mbuf *m, uint16_t tag) 1265 { 1266 struct ether_vlan_header *evl; 1267 1268 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_DONTWAIT); 1269 if (m == NULL) 1270 return (NULL); 1271 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */ 1272 1273 if (m->m_len < sizeof(*evl)) { 1274 m = m_pullup(m, sizeof(*evl)); 1275 if (m == NULL) 1276 return (NULL); 1277 } 1278 1279 /* 1280 * Transform the Ethernet header into an Ethernet header 1281 * with 802.1Q encapsulation. 1282 */ 1283 evl = mtod(m, struct ether_vlan_header *); 1284 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN, 1285 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN); 1286 evl->evl_encap_proto = htons(ETHERTYPE_VLAN); 1287 evl->evl_tag = htons(tag); 1288 return (m); 1289 } 1290 1291 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY); 1292 MODULE_VERSION(ether, 1); 1293