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 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * @(#)if_ethersubr.c 8.1 (Berkeley) 6/10/93 34 * $FreeBSD: src/sys/net/if_ethersubr.c,v 1.70.2.33 2003/04/28 15:45:53 archie Exp $ 35 * $DragonFly: src/sys/net/if_ethersubr.c,v 1.40 2007/06/02 09:08:26 sephe Exp $ 36 */ 37 38 #include "opt_atalk.h" 39 #include "opt_inet.h" 40 #include "opt_inet6.h" 41 #include "opt_ipx.h" 42 #include "opt_netgraph.h" 43 44 #include <sys/param.h> 45 #include <sys/systm.h> 46 #include <sys/kernel.h> 47 #include <sys/malloc.h> 48 #include <sys/mbuf.h> 49 #include <sys/socket.h> 50 #include <sys/sockio.h> 51 #include <sys/sysctl.h> 52 53 #include <net/if.h> 54 #include <net/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/ifq_var.h> 60 #include <net/bpf.h> 61 #include <net/ethernet.h> 62 63 #if defined(INET) || defined(INET6) 64 #include <netinet/in.h> 65 #include <netinet/in_var.h> 66 #include <netinet/if_ether.h> 67 #include <net/ipfw/ip_fw.h> 68 #include <net/dummynet/ip_dummynet.h> 69 #endif 70 #ifdef INET6 71 #include <netinet6/nd6.h> 72 #endif 73 74 #ifdef IPX 75 #include <netproto/ipx/ipx.h> 76 #include <netproto/ipx/ipx_if.h> 77 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m); 78 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, struct sockaddr *dst, 79 short *tp, int *hlen); 80 #endif 81 82 #ifdef NS 83 #include <netns/ns.h> 84 #include <netns/ns_if.h> 85 ushort ns_nettype; 86 int ether_outputdebug = 0; 87 int ether_inputdebug = 0; 88 #endif 89 90 #ifdef NETATALK 91 #include <netproto/atalk/at.h> 92 #include <netproto/atalk/at_var.h> 93 #include <netproto/atalk/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 98 extern u_char at_org_code[3]; 99 extern u_char aarp_org_code[3]; 100 #endif /* NETATALK */ 101 102 /* netgraph node hooks for ng_ether(4) */ 103 void (*ng_ether_input_p)(struct ifnet *ifp, 104 struct mbuf **mp, struct ether_header *eh); 105 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, 106 struct mbuf *m, struct ether_header *eh); 107 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp); 108 void (*ng_ether_attach_p)(struct ifnet *ifp); 109 void (*ng_ether_detach_p)(struct ifnet *ifp); 110 111 int (*vlan_input_p)(struct ether_header *eh, struct mbuf *m); 112 int (*vlan_input_tag_p)(struct mbuf *m, uint16_t t); 113 114 static int ether_output(struct ifnet *, struct mbuf *, struct sockaddr *, 115 struct rtentry *); 116 117 /* 118 * if_bridge support 119 */ 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 static int ether_resolvemulti(struct ifnet *, struct sockaddr **, 126 struct sockaddr *); 127 128 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = { 129 0xff, 0xff, 0xff, 0xff, 0xff, 0xff 130 }; 131 132 #define gotoerr(e) do { error = (e); goto bad; } while (0) 133 #define IFP2AC(ifp) ((struct arpcom *)(ifp)) 134 135 static boolean_t ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 136 struct ip_fw **rule, struct ether_header *eh, 137 boolean_t shared); 138 139 static int ether_ipfw; 140 SYSCTL_DECL(_net_link); 141 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet"); 142 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW, 143 ðer_ipfw, 0, "Pass ether pkts through firewall"); 144 145 /* 146 * Ethernet output routine. 147 * Encapsulate a packet of type family for the local net. 148 * Use trailer local net encapsulation if enough data in first 149 * packet leaves a multiple of 512 bytes of data in remainder. 150 * Assumes that ifp is actually pointer to arpcom structure. 151 */ 152 static int 153 ether_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst, 154 struct rtentry *rt) 155 { 156 struct ether_header *eh, *deh; 157 u_char *edst; 158 int loop_copy = 0; 159 int hlen = ETHER_HDR_LEN; /* link layer header length */ 160 struct arpcom *ac = IFP2AC(ifp); 161 int error; 162 163 ASSERT_SERIALIZED(ifp->if_serializer); 164 165 if (ifp->if_flags & IFF_MONITOR) 166 gotoerr(ENETDOWN); 167 if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING)) 168 gotoerr(ENETDOWN); 169 170 M_PREPEND(m, sizeof(struct ether_header), MB_DONTWAIT); 171 if (m == NULL) 172 return (ENOBUFS); 173 eh = mtod(m, struct ether_header *); 174 edst = eh->ether_dhost; 175 176 /* 177 * Fill in the destination ethernet address and frame type. 178 */ 179 switch (dst->sa_family) { 180 #ifdef INET 181 case AF_INET: 182 if (!arpresolve(ifp, rt, m, dst, edst)) 183 return (0); /* if not yet resolved */ 184 eh->ether_type = htons(ETHERTYPE_IP); 185 break; 186 #endif 187 #ifdef INET6 188 case AF_INET6: 189 if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, edst)) 190 return (0); /* Something bad happenned. */ 191 eh->ether_type = htons(ETHERTYPE_IPV6); 192 break; 193 #endif 194 #ifdef IPX 195 case AF_IPX: 196 if (ef_outputp != NULL) { 197 error = ef_outputp(ifp, &m, dst, &eh->ether_type, 198 &hlen); 199 if (error) 200 goto bad; 201 } else { 202 eh->ether_type = htons(ETHERTYPE_IPX); 203 bcopy(&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host), 204 edst, ETHER_ADDR_LEN); 205 } 206 break; 207 #endif 208 #ifdef NETATALK 209 case AF_APPLETALK: { 210 struct at_ifaddr *aa; 211 212 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) { 213 error = 0; /* XXX */ 214 goto bad; 215 } 216 /* 217 * In the phase 2 case, need to prepend an mbuf for 218 * the llc header. Since we must preserve the value 219 * of m, which is passed to us by value, we m_copy() 220 * the first mbuf, and use it for our llc header. 221 */ 222 if (aa->aa_flags & AFA_PHASE2) { 223 struct llc llc; 224 225 M_PREPEND(m, sizeof(struct llc), MB_DONTWAIT); 226 eh = mtod(m, struct ether_header *); 227 edst = eh->ether_dhost; 228 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 229 llc.llc_control = LLC_UI; 230 bcopy(at_org_code, llc.llc_snap_org_code, 231 sizeof at_org_code); 232 llc.llc_snap_ether_type = htons(ETHERTYPE_AT); 233 bcopy(&llc, 234 mtod(m, caddr_t) + sizeof(struct ether_header), 235 sizeof(struct llc)); 236 eh->ether_type = htons(m->m_pkthdr.len); 237 hlen = sizeof(struct llc) + ETHER_HDR_LEN; 238 } else { 239 eh->ether_type = htons(ETHERTYPE_AT); 240 } 241 if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst)) 242 return (0); 243 break; 244 } 245 #endif 246 #ifdef NS 247 case AF_NS: 248 switch(ns_nettype) { 249 default: 250 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */ 251 eh->ether_type = 0x8137; 252 break; 253 case 0x0: /* Novell 802.3 */ 254 eh->ether_type = htons(m->m_pkthdr.len); 255 break; 256 case 0xe0e0: /* Novell 802.2 and Token-Ring */ 257 M_PREPEND(m, 3, MB_DONTWAIT); 258 eh = mtod(m, struct ether_header *); 259 edst = eh->ether_dhost; 260 eh->ether_type = htons(m->m_pkthdr.len); 261 cp = mtod(m, u_char *) + sizeof(struct ether_header); 262 *cp++ = 0xE0; 263 *cp++ = 0xE0; 264 *cp++ = 0x03; 265 break; 266 } 267 bcopy(&(((struct sockaddr_ns *)dst)->sns_addr.x_host), edst, 268 ETHER_ADDR_LEN); 269 /* 270 * XXX if ns_thishost is the same as the node's ethernet 271 * address then just the default code will catch this anyhow. 272 * So I'm not sure if this next clause should be here at all? 273 * [JRE] 274 */ 275 if (bcmp(edst, &ns_thishost, ETHER_ADDR_LEN) == 0) { 276 m->m_pkthdr.rcvif = ifp; 277 netisr_dispatch(NETISR_NS, m); 278 return (error); 279 } 280 if (bcmp(edst, &ns_broadhost, ETHER_ADDR_LEN) == 0) 281 m->m_flags |= M_BCAST; 282 break; 283 #endif 284 case pseudo_AF_HDRCMPLT: 285 case AF_UNSPEC: 286 loop_copy = -1; /* if this is for us, don't do it */ 287 deh = (struct ether_header *)dst->sa_data; 288 memcpy(edst, deh->ether_dhost, ETHER_ADDR_LEN); 289 eh->ether_type = deh->ether_type; 290 break; 291 292 default: 293 if_printf(ifp, "can't handle af%d\n", dst->sa_family); 294 gotoerr(EAFNOSUPPORT); 295 } 296 297 if (dst->sa_family == pseudo_AF_HDRCMPLT) /* unlikely */ 298 memcpy(eh->ether_shost, 299 ((struct ether_header *)dst->sa_data)->ether_shost, 300 ETHER_ADDR_LEN); 301 else 302 memcpy(eh->ether_shost, ac->ac_enaddr, ETHER_ADDR_LEN); 303 304 /* 305 * Bridges require special output handling. 306 */ 307 if (ifp->if_bridge) { 308 KASSERT(bridge_output_p != NULL, 309 ("%s: if_bridge not loaded!", __func__)); 310 return ((*bridge_output_p)(ifp, m, NULL, NULL)); 311 } 312 313 /* 314 * If a simplex interface, and the packet is being sent to our 315 * Ethernet address or a broadcast address, loopback a copy. 316 * XXX To make a simplex device behave exactly like a duplex 317 * device, we should copy in the case of sending to our own 318 * ethernet address (thus letting the original actually appear 319 * on the wire). However, we don't do that here for security 320 * reasons and compatibility with the original behavior. 321 */ 322 if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) { 323 int csum_flags = 0; 324 325 if (m->m_pkthdr.csum_flags & CSUM_IP) 326 csum_flags |= (CSUM_IP_CHECKED | CSUM_IP_VALID); 327 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) 328 csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); 329 if ((m->m_flags & M_BCAST) || (loop_copy > 0)) { 330 struct mbuf *n; 331 332 if ((n = m_copypacket(m, MB_DONTWAIT)) != NULL) { 333 n->m_pkthdr.csum_flags |= csum_flags; 334 if (csum_flags & CSUM_DATA_VALID) 335 n->m_pkthdr.csum_data = 0xffff; 336 if_simloop(ifp, n, dst->sa_family, hlen); 337 } else 338 ifp->if_iqdrops++; 339 } else if (bcmp(eh->ether_dhost, eh->ether_shost, 340 ETHER_ADDR_LEN) == 0) { 341 m->m_pkthdr.csum_flags |= csum_flags; 342 if (csum_flags & CSUM_DATA_VALID) 343 m->m_pkthdr.csum_data = 0xffff; 344 if_simloop(ifp, m, dst->sa_family, hlen); 345 return (0); /* XXX */ 346 } 347 } 348 349 /* Handle ng_ether(4) processing, if any */ 350 if (ng_ether_output_p != NULL) { 351 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) 352 goto bad; 353 if (m == NULL) 354 return (0); 355 } 356 357 /* Continue with link-layer output */ 358 return ether_output_frame(ifp, m); 359 360 bad: 361 m_freem(m); 362 return (error); 363 } 364 365 /* 366 * Ethernet link layer output routine to send a raw frame to the device. 367 * 368 * This assumes that the 14 byte Ethernet header is present and contiguous 369 * in the first mbuf. 370 */ 371 int 372 ether_output_frame(struct ifnet *ifp, struct mbuf *m) 373 { 374 struct ip_fw *rule = NULL; 375 int error = 0; 376 struct altq_pktattr pktattr; 377 378 ASSERT_SERIALIZED(ifp->if_serializer); 379 380 /* Extract info from dummynet tag, ignore others */ 381 while (m->m_type == MT_TAG) { 382 if (m->m_flags == PACKET_TAG_DUMMYNET) { 383 rule = ((struct dn_pkt *)m)->rule; 384 break; 385 } 386 m = m->m_next; 387 } 388 389 if (ifq_is_enabled(&ifp->if_snd)) 390 altq_etherclassify(&ifp->if_snd, m, &pktattr); 391 crit_enter(); 392 if (IPFW_LOADED && ether_ipfw != 0) { 393 struct ether_header save_eh, *eh; 394 395 eh = mtod(m, struct ether_header *); 396 save_eh = *eh; 397 m_adj(m, ETHER_HDR_LEN); 398 if (!ether_ipfw_chk(&m, ifp, &rule, eh, FALSE)) { 399 crit_exit(); 400 if (m != NULL) { 401 m_freem(m); 402 return ENOBUFS; /* pkt dropped */ 403 } else 404 return 0; /* consumed e.g. in a pipe */ 405 } 406 eh = mtod(m, struct ether_header *); 407 /* packet was ok, restore the ethernet header */ 408 if ((void *)(eh + 1) == (void *)m->m_data) { 409 m->m_data -= ETHER_HDR_LEN ; 410 m->m_len += ETHER_HDR_LEN ; 411 m->m_pkthdr.len += ETHER_HDR_LEN ; 412 } else { 413 M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT); 414 if (m == NULL) /* nope... */ { 415 crit_exit(); 416 return ENOBUFS; 417 } 418 bcopy(&save_eh, mtod(m, struct ether_header *), 419 ETHER_HDR_LEN); 420 } 421 } 422 crit_exit(); 423 424 /* 425 * Queue message on interface, update output statistics if 426 * successful, and start output if interface not yet active. 427 */ 428 error = ifq_handoff(ifp, m, &pktattr); 429 return (error); 430 } 431 432 /* 433 * ipfw processing for ethernet packets (in and out). 434 * The second parameter is NULL from ether_demux(), and ifp from 435 * ether_output_frame(). 436 */ 437 static boolean_t 438 ether_ipfw_chk( 439 struct mbuf **m0, 440 struct ifnet *dst, 441 struct ip_fw **rule, 442 struct ether_header *eh, 443 boolean_t shared) 444 { 445 struct ether_header save_eh = *eh; /* might be a ptr in m */ 446 struct ip_fw_args args; 447 struct m_tag *mtag; 448 int i; 449 450 if (*rule != NULL && fw_one_pass) 451 return TRUE; /* dummynet packet, already partially processed */ 452 453 /* 454 * I need some amount of data to be contiguous, and in case others 455 * need the packet (shared==TRUE), it also better be in the first mbuf. 456 */ 457 i = min((*m0)->m_pkthdr.len, max_protohdr); 458 if (shared || (*m0)->m_len < i) { 459 *m0 = m_pullup(*m0, i); 460 if (*m0 == NULL) 461 return FALSE; 462 } 463 464 args.m = *m0; /* the packet we are looking at */ 465 args.oif = dst; /* destination, if any */ 466 if ((mtag = m_tag_find(*m0, PACKET_TAG_IPFW_DIVERT, NULL)) != NULL) 467 m_tag_delete(*m0, mtag); 468 args.rule = *rule; /* matching rule to restart */ 469 args.next_hop = NULL; /* we do not support forward yet */ 470 args.eh = &save_eh; /* MAC header for bridged/MAC packets */ 471 i = ip_fw_chk_ptr(&args); 472 *m0 = args.m; 473 *rule = args.rule; 474 475 if ((i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL) /* drop */ 476 return FALSE; 477 478 if (i == 0) /* a PASS rule. */ 479 return TRUE; 480 481 if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG)) { 482 /* 483 * Pass the pkt to dummynet, which consumes it. 484 * If shared, make a copy and keep the original. 485 */ 486 struct mbuf *m ; 487 488 if (shared) { 489 m = m_copypacket(*m0, MB_DONTWAIT); 490 if (m == NULL) 491 return FALSE; 492 } else { 493 m = *m0 ; /* pass the original to dummynet */ 494 *m0 = NULL ; /* and nothing back to the caller */ 495 } 496 /* 497 * Prepend the header, optimize for the common case of 498 * eh pointing into the mbuf. 499 */ 500 if ((void *)(eh + 1) == (void *)m->m_data) { 501 m->m_data -= ETHER_HDR_LEN ; 502 m->m_len += ETHER_HDR_LEN ; 503 m->m_pkthdr.len += ETHER_HDR_LEN ; 504 } else { 505 M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT); 506 if (m == NULL) 507 return FALSE; 508 bcopy(&save_eh, mtod(m, struct ether_header *), 509 ETHER_HDR_LEN); 510 } 511 ip_dn_io_ptr(m, (i & 0xffff), 512 dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args); 513 return FALSE; 514 } 515 /* 516 * XXX at some point add support for divert/forward actions. 517 * If none of the above matches, we have to drop the pkt. 518 */ 519 return FALSE; 520 } 521 522 /* 523 * XXX merge this function with ether_input. 524 */ 525 static void 526 ether_input_internal(struct ifnet *ifp, struct mbuf *m) 527 { 528 ether_input(ifp, NULL, m); 529 } 530 531 /* 532 * Process a received Ethernet packet. We have two different interfaces: 533 * one (conventional) assumes the packet in the mbuf, with the ethernet 534 * header provided separately in *eh. The second one (new) has everything 535 * in the mbuf, and we can tell it because eh == NULL. 536 * The caller MUST MAKE SURE that there are at least 537 * sizeof(struct ether_header) bytes in the first mbuf. 538 * 539 * This allows us to concentrate in one place a bunch of code which 540 * is replicated in all device drivers. Also, many functions called 541 * from ether_input() try to put the eh back into the mbuf, so we 542 * can later propagate the 'contiguous packet' interface to them, 543 * and handle the old interface just here. 544 * 545 * NOTA BENE: for many drivers "eh" is a pointer into the first mbuf or 546 * cluster, right before m_data. So be very careful when working on m, 547 * as you could destroy *eh !! 548 * 549 * First we perform any link layer operations, then continue 550 * to the upper layers with ether_demux(). 551 */ 552 void 553 ether_input(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m) 554 { 555 ASSERT_SERIALIZED(ifp->if_serializer); 556 557 /* XXX old crufty stuff, needs to be removed */ 558 if (eh != NULL) { 559 kprintf("ether_input got mbuf without embedded ethernet header"); 560 m_free(m); 561 return; 562 } 563 564 if (m->m_len < sizeof(struct ether_header)) { 565 /* XXX error in the caller. */ 566 m_freem(m); 567 return; 568 } 569 m->m_pkthdr.rcvif = ifp; 570 eh = mtod(m, struct ether_header *); 571 572 BPF_MTAP(ifp, m); 573 574 ifp->if_ibytes += m->m_pkthdr.len; 575 576 if (ifp->if_flags & IFF_MONITOR) { 577 /* 578 * Interface marked for monitoring; discard packet. 579 */ 580 m_freem(m); 581 return; 582 } 583 584 /* 585 * Tap the packet off here for a bridge. bridge_input() 586 * will return NULL if it has consumed the packet, otherwise 587 * it gets processed as normal. Note that bridge_input() 588 * will always return the original packet if we need to 589 * process it locally. 590 */ 591 if (ifp->if_bridge) { 592 KASSERT(bridge_input_p != NULL, 593 ("%s: if_bridge not loaded!", __func__)); 594 595 if(m->m_flags & M_PROTO1) { 596 m->m_flags &= ~M_PROTO1; 597 } else { 598 /* clear M_PROMISC, in case the packets comes from a vlan */ 599 /* m->m_flags &= ~M_PROMISC; */ 600 lwkt_serialize_exit(ifp->if_serializer); 601 m = (*bridge_input_p)(ifp, m); 602 lwkt_serialize_enter(ifp->if_serializer); 603 if (m == NULL) 604 return; 605 606 /* 607 * Bridge has determined that the packet is for us. 608 * Update our interface pointer -- we may have had 609 * to "bridge" the packet locally. 610 */ 611 ifp = m->m_pkthdr.rcvif; 612 } 613 } 614 615 /* XXX old crufty stuff, needs to be removed */ 616 m_adj(m, sizeof(struct ether_header)); 617 /* XXX */ 618 /* m->m_pkthdr.len = m->m_len; */ 619 620 /* Handle ng_ether(4) processing, if any */ 621 if (ng_ether_input_p != NULL) { 622 lwkt_serialize_exit(ifp->if_serializer); 623 (*ng_ether_input_p)(ifp, &m, eh); 624 lwkt_serialize_enter(ifp->if_serializer); 625 if (m == NULL) 626 return; 627 } 628 629 /* Continue with upper layer processing */ 630 ether_demux(ifp, eh, m); 631 } 632 633 /* 634 * Upper layer processing for a received Ethernet packet. 635 */ 636 void 637 ether_demux(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m) 638 { 639 int isr; 640 u_short ether_type; 641 struct ip_fw *rule = NULL; 642 #ifdef NETATALK 643 struct llc *l; 644 #endif 645 646 /* Extract info from dummynet tag, ignore others */ 647 while (m->m_type == MT_TAG) { 648 if (m->m_flags == PACKET_TAG_DUMMYNET) { 649 rule = ((struct dn_pkt *)m)->rule; 650 ifp = m->m_next->m_pkthdr.rcvif; 651 break; 652 } 653 m = m->m_next; 654 } 655 if (rule) /* packet is passing the second time */ 656 goto post_stats; 657 658 /* 659 * Discard packet if upper layers shouldn't see it because 660 * it was unicast to a different Ethernet address. If the 661 * driver is working properly, then this situation can only 662 * happen when the interface is in promiscuous mode. 663 */ 664 if (((ifp->if_flags & (IFF_PROMISC | IFF_PPROMISC)) == IFF_PROMISC) && 665 (eh->ether_dhost[0] & 1) == 0 && 666 bcmp(eh->ether_dhost, IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN)) { 667 m_freem(m); 668 return; 669 } 670 /* Discard packet if interface is not up */ 671 if (!(ifp->if_flags & IFF_UP)) { 672 m_freem(m); 673 return; 674 } 675 if (eh->ether_dhost[0] & 1) { 676 if (bcmp(ifp->if_broadcastaddr, eh->ether_dhost, 677 ifp->if_addrlen) == 0) 678 m->m_flags |= M_BCAST; 679 else 680 m->m_flags |= M_MCAST; 681 ifp->if_imcasts++; 682 } 683 684 post_stats: 685 if (IPFW_LOADED && ether_ipfw != 0) { 686 if (!ether_ipfw_chk(&m, NULL, &rule, eh, FALSE)) { 687 m_freem(m); 688 return; 689 } 690 eh = mtod(m, struct ether_header *); 691 } 692 693 ether_type = ntohs(eh->ether_type); 694 695 switch (ether_type) { 696 #ifdef INET 697 case ETHERTYPE_IP: 698 if (ipflow_fastforward(m, ifp->if_serializer)) 699 return; 700 isr = NETISR_IP; 701 break; 702 703 case ETHERTYPE_ARP: 704 if (ifp->if_flags & IFF_NOARP) { 705 /* Discard packet if ARP is disabled on interface */ 706 m_freem(m); 707 return; 708 } 709 isr = NETISR_ARP; 710 break; 711 #endif 712 713 #ifdef INET6 714 case ETHERTYPE_IPV6: 715 isr = NETISR_IPV6; 716 break; 717 #endif 718 719 #ifdef IPX 720 case ETHERTYPE_IPX: 721 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 722 return; 723 isr = NETISR_IPX; 724 break; 725 #endif 726 727 #ifdef NS 728 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */ 729 isr = NETISR_NS; 730 break; 731 732 #endif 733 734 #ifdef NETATALK 735 case ETHERTYPE_AT: 736 isr = NETISR_ATALK1; 737 break; 738 case ETHERTYPE_AARP: 739 isr = NETISR_AARP; 740 break; 741 #endif 742 743 case ETHERTYPE_VLAN: 744 if (vlan_input_p != NULL) 745 (*vlan_input_p)(eh, m); 746 else { 747 m->m_pkthdr.rcvif->if_noproto++; 748 m_freem(m); 749 } 750 return; 751 752 default: 753 #ifdef IPX 754 if (ef_inputp && ef_inputp(ifp, eh, m) == 0) 755 return; 756 #endif 757 #ifdef NS 758 checksum = mtod(m, ushort *); 759 /* Novell 802.3 */ 760 if ((ether_type <= ETHERMTU) && 761 ((*checksum == 0xffff) || (*checksum == 0xE0E0))) { 762 if (*checksum == 0xE0E0) { 763 m->m_pkthdr.len -= 3; 764 m->m_len -= 3; 765 m->m_data += 3; 766 } 767 isr = NETISR_NS; 768 break; 769 } 770 #endif 771 #ifdef NETATALK 772 if (ether_type > ETHERMTU) 773 goto dropanyway; 774 l = mtod(m, struct llc *); 775 if (l->llc_dsap == LLC_SNAP_LSAP && 776 l->llc_ssap == LLC_SNAP_LSAP && 777 l->llc_control == LLC_UI) { 778 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code, 779 sizeof at_org_code) == 0 && 780 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) { 781 m_adj(m, sizeof(struct llc)); 782 isr = NETISR_ATALK2; 783 break; 784 } 785 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code, 786 sizeof aarp_org_code) == 0 && 787 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) { 788 m_adj(m, sizeof(struct llc)); 789 isr = NETISR_AARP; 790 break; 791 } 792 } 793 dropanyway: 794 #endif 795 if (ng_ether_input_orphan_p != NULL) 796 (*ng_ether_input_orphan_p)(ifp, m, eh); 797 else 798 m_freem(m); 799 return; 800 } 801 netisr_dispatch(isr, m); 802 } 803 804 /* 805 * Perform common duties while attaching to interface list 806 */ 807 808 void 809 ether_ifattach(struct ifnet *ifp, uint8_t *lla, lwkt_serialize_t serializer) 810 { 811 ether_ifattach_bpf(ifp, lla, DLT_EN10MB, sizeof(struct ether_header), 812 serializer); 813 } 814 815 void 816 ether_ifattach_bpf(struct ifnet *ifp, uint8_t *lla, u_int dlt, u_int hdrlen, 817 lwkt_serialize_t serializer) 818 { 819 struct sockaddr_dl *sdl; 820 821 ifp->if_type = IFT_ETHER; 822 ifp->if_addrlen = ETHER_ADDR_LEN; 823 ifp->if_hdrlen = ETHER_HDR_LEN; 824 if_attach(ifp, serializer); 825 ifp->if_mtu = ETHERMTU; 826 if (ifp->if_baudrate == 0) 827 ifp->if_baudrate = 10000000; 828 ifp->if_output = ether_output; 829 ifp->if_input = ether_input_internal; 830 ifp->if_resolvemulti = ether_resolvemulti; 831 ifp->if_broadcastaddr = etherbroadcastaddr; 832 sdl = IF_LLSOCKADDR(ifp); 833 sdl->sdl_type = IFT_ETHER; 834 sdl->sdl_alen = ifp->if_addrlen; 835 bcopy(lla, LLADDR(sdl), ifp->if_addrlen); 836 /* 837 * XXX Keep the current drivers happy. 838 * XXX Remove once all drivers have been cleaned up 839 */ 840 if (lla != IFP2AC(ifp)->ac_enaddr) 841 bcopy(lla, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen); 842 bpfattach(ifp, dlt, hdrlen); 843 if (ng_ether_attach_p != NULL) 844 (*ng_ether_attach_p)(ifp); 845 846 if_printf(ifp, "MAC address: %6D\n", lla, ":"); 847 } 848 849 /* 850 * Perform common duties while detaching an Ethernet interface 851 */ 852 void 853 ether_ifdetach(struct ifnet *ifp) 854 { 855 if_down(ifp); 856 857 if (ng_ether_detach_p != NULL) 858 (*ng_ether_detach_p)(ifp); 859 bpfdetach(ifp); 860 if_detach(ifp); 861 } 862 863 int 864 ether_ioctl(struct ifnet *ifp, int command, caddr_t data) 865 { 866 struct ifaddr *ifa = (struct ifaddr *) data; 867 struct ifreq *ifr = (struct ifreq *) data; 868 int error = 0; 869 870 ASSERT_SERIALIZED(ifp->if_serializer); 871 872 switch (command) { 873 case SIOCSIFADDR: 874 ifp->if_flags |= IFF_UP; 875 876 switch (ifa->ifa_addr->sa_family) { 877 #ifdef INET 878 case AF_INET: 879 ifp->if_init(ifp->if_softc); /* before arpwhohas */ 880 arp_ifinit(ifp, ifa); 881 break; 882 #endif 883 #ifdef IPX 884 /* 885 * XXX - This code is probably wrong 886 */ 887 case AF_IPX: 888 { 889 struct ipx_addr *ina = &IA_SIPX(ifa)->sipx_addr; 890 struct arpcom *ac = IFP2AC(ifp); 891 892 if (ipx_nullhost(*ina)) 893 ina->x_host = *(union ipx_host *) ac->ac_enaddr; 894 else 895 bcopy(ina->x_host.c_host, ac->ac_enaddr, 896 sizeof ac->ac_enaddr); 897 898 ifp->if_init(ifp->if_softc); /* Set new address. */ 899 break; 900 } 901 #endif 902 #ifdef NS 903 /* 904 * XXX - This code is probably wrong 905 */ 906 case AF_NS: 907 { 908 struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr); 909 struct arpcom *ac = IFP2AC(ifp); 910 911 if (ns_nullhost(*ina)) 912 ina->x_host = *(union ns_host *)(ac->ac_enaddr); 913 else 914 bcopy(ina->x_host.c_host, ac->ac_enaddr, 915 sizeof ac->ac_enaddr); 916 917 /* 918 * Set new address 919 */ 920 ifp->if_init(ifp->if_softc); 921 break; 922 } 923 #endif 924 default: 925 ifp->if_init(ifp->if_softc); 926 break; 927 } 928 break; 929 930 case SIOCGIFADDR: 931 bcopy(IFP2AC(ifp)->ac_enaddr, 932 ((struct sockaddr *)ifr->ifr_data)->sa_data, 933 ETHER_ADDR_LEN); 934 break; 935 936 case SIOCSIFMTU: 937 /* 938 * Set the interface MTU. 939 */ 940 if (ifr->ifr_mtu > ETHERMTU) { 941 error = EINVAL; 942 } else { 943 ifp->if_mtu = ifr->ifr_mtu; 944 } 945 break; 946 default: 947 error = EINVAL; 948 break; 949 } 950 return (error); 951 } 952 953 int 954 ether_resolvemulti( 955 struct ifnet *ifp, 956 struct sockaddr **llsa, 957 struct sockaddr *sa) 958 { 959 struct sockaddr_dl *sdl; 960 struct sockaddr_in *sin; 961 #ifdef INET6 962 struct sockaddr_in6 *sin6; 963 #endif 964 u_char *e_addr; 965 966 switch(sa->sa_family) { 967 case AF_LINK: 968 /* 969 * No mapping needed. Just check that it's a valid MC address. 970 */ 971 sdl = (struct sockaddr_dl *)sa; 972 e_addr = LLADDR(sdl); 973 if ((e_addr[0] & 1) != 1) 974 return EADDRNOTAVAIL; 975 *llsa = 0; 976 return 0; 977 978 #ifdef INET 979 case AF_INET: 980 sin = (struct sockaddr_in *)sa; 981 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) 982 return EADDRNOTAVAIL; 983 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, 984 M_WAITOK | M_ZERO); 985 sdl->sdl_len = sizeof *sdl; 986 sdl->sdl_family = AF_LINK; 987 sdl->sdl_index = ifp->if_index; 988 sdl->sdl_type = IFT_ETHER; 989 sdl->sdl_alen = ETHER_ADDR_LEN; 990 e_addr = LLADDR(sdl); 991 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr); 992 *llsa = (struct sockaddr *)sdl; 993 return 0; 994 #endif 995 #ifdef INET6 996 case AF_INET6: 997 sin6 = (struct sockaddr_in6 *)sa; 998 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 999 /* 1000 * An IP6 address of 0 means listen to all 1001 * of the Ethernet multicast address used for IP6. 1002 * (This is used for multicast routers.) 1003 */ 1004 ifp->if_flags |= IFF_ALLMULTI; 1005 *llsa = 0; 1006 return 0; 1007 } 1008 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) 1009 return EADDRNOTAVAIL; 1010 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, 1011 M_WAITOK | M_ZERO); 1012 sdl->sdl_len = sizeof *sdl; 1013 sdl->sdl_family = AF_LINK; 1014 sdl->sdl_index = ifp->if_index; 1015 sdl->sdl_type = IFT_ETHER; 1016 sdl->sdl_alen = ETHER_ADDR_LEN; 1017 e_addr = LLADDR(sdl); 1018 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr); 1019 *llsa = (struct sockaddr *)sdl; 1020 return 0; 1021 #endif 1022 1023 default: 1024 /* 1025 * Well, the text isn't quite right, but it's the name 1026 * that counts... 1027 */ 1028 return EAFNOSUPPORT; 1029 } 1030 } 1031 1032 #if 0 1033 /* 1034 * This is for reference. We have a table-driven version 1035 * of the little-endian crc32 generator, which is faster 1036 * than the double-loop. 1037 */ 1038 uint32_t 1039 ether_crc32_le(const uint8_t *buf, size_t len) 1040 { 1041 uint32_t c, crc, carry; 1042 size_t i, j; 1043 1044 crc = 0xffffffffU; /* initial value */ 1045 1046 for (i = 0; i < len; i++) { 1047 c = buf[i]; 1048 for (j = 0; j < 8; j++) { 1049 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1050 crc >>= 1; 1051 c >>= 1; 1052 if (carry) 1053 crc = (crc ^ ETHER_CRC_POLY_LE); 1054 } 1055 } 1056 1057 return (crc); 1058 } 1059 #else 1060 uint32_t 1061 ether_crc32_le(const uint8_t *buf, size_t len) 1062 { 1063 static const uint32_t crctab[] = { 1064 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1065 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1066 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1067 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1068 }; 1069 uint32_t crc; 1070 size_t i; 1071 1072 crc = 0xffffffffU; /* initial value */ 1073 1074 for (i = 0; i < len; i++) { 1075 crc ^= buf[i]; 1076 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1077 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1078 } 1079 1080 return (crc); 1081 } 1082 #endif 1083 1084 uint32_t 1085 ether_crc32_be(const uint8_t *buf, size_t len) 1086 { 1087 uint32_t c, crc, carry; 1088 size_t i, j; 1089 1090 crc = 0xffffffffU; /* initial value */ 1091 1092 for (i = 0; i < len; i++) { 1093 c = buf[i]; 1094 for (j = 0; j < 8; j++) { 1095 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1096 crc <<= 1; 1097 c >>= 1; 1098 if (carry) 1099 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1100 } 1101 } 1102 1103 return (crc); 1104 } 1105 1106 /* 1107 * find the size of ethernet header, and call classifier 1108 */ 1109 void 1110 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m, 1111 struct altq_pktattr *pktattr) 1112 { 1113 struct ether_header *eh; 1114 uint16_t ether_type; 1115 int hlen, af, hdrsize; 1116 caddr_t hdr; 1117 1118 hlen = sizeof(struct ether_header); 1119 eh = mtod(m, struct ether_header *); 1120 1121 ether_type = ntohs(eh->ether_type); 1122 if (ether_type < ETHERMTU) { 1123 /* ick! LLC/SNAP */ 1124 struct llc *llc = (struct llc *)(eh + 1); 1125 hlen += 8; 1126 1127 if (m->m_len < hlen || 1128 llc->llc_dsap != LLC_SNAP_LSAP || 1129 llc->llc_ssap != LLC_SNAP_LSAP || 1130 llc->llc_control != LLC_UI) 1131 goto bad; /* not snap! */ 1132 1133 ether_type = ntohs(llc->llc_un.type_snap.ether_type); 1134 } 1135 1136 if (ether_type == ETHERTYPE_IP) { 1137 af = AF_INET; 1138 hdrsize = 20; /* sizeof(struct ip) */ 1139 #ifdef INET6 1140 } else if (ether_type == ETHERTYPE_IPV6) { 1141 af = AF_INET6; 1142 hdrsize = 40; /* sizeof(struct ip6_hdr) */ 1143 #endif 1144 } else 1145 goto bad; 1146 1147 while (m->m_len <= hlen) { 1148 hlen -= m->m_len; 1149 m = m->m_next; 1150 } 1151 hdr = m->m_data + hlen; 1152 if (m->m_len < hlen + hdrsize) { 1153 /* 1154 * ip header is not in a single mbuf. this should not 1155 * happen in the current code. 1156 * (todo: use m_pulldown in the future) 1157 */ 1158 goto bad; 1159 } 1160 m->m_data += hlen; 1161 m->m_len -= hlen; 1162 ifq_classify(ifq, m, af, pktattr); 1163 m->m_data -= hlen; 1164 m->m_len += hlen; 1165 1166 return; 1167 1168 bad: 1169 pktattr->pattr_class = NULL; 1170 pktattr->pattr_hdr = NULL; 1171 pktattr->pattr_af = AF_UNSPEC; 1172 } 1173