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 1.50 2007/10/25 13:13:18 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 #include "opt_carp.h" 44 45 #include <sys/param.h> 46 #include <sys/systm.h> 47 #include <sys/kernel.h> 48 #include <sys/malloc.h> 49 #include <sys/mbuf.h> 50 #include <sys/socket.h> 51 #include <sys/sockio.h> 52 #include <sys/sysctl.h> 53 54 #include <net/if.h> 55 #include <net/netisr.h> 56 #include <net/route.h> 57 #include <net/if_llc.h> 58 #include <net/if_dl.h> 59 #include <net/if_types.h> 60 #include <net/ifq_var.h> 61 #include <net/bpf.h> 62 #include <net/ethernet.h> 63 64 #if defined(INET) || defined(INET6) 65 #include <netinet/in.h> 66 #include <netinet/in_var.h> 67 #include <netinet/if_ether.h> 68 #include <net/ipfw/ip_fw.h> 69 #include <net/dummynet/ip_dummynet.h> 70 #endif 71 #ifdef INET6 72 #include <netinet6/nd6.h> 73 #endif 74 75 #ifdef CARP 76 #include <netinet/ip_carp.h> 77 #endif 78 79 #ifdef IPX 80 #include <netproto/ipx/ipx.h> 81 #include <netproto/ipx/ipx_if.h> 82 int (*ef_inputp)(struct ifnet*, const struct ether_header *eh, struct mbuf *m); 83 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, struct sockaddr *dst, 84 short *tp, int *hlen); 85 #endif 86 87 #ifdef NS 88 #include <netns/ns.h> 89 #include <netns/ns_if.h> 90 ushort ns_nettype; 91 int ether_outputdebug = 0; 92 int ether_inputdebug = 0; 93 #endif 94 95 #ifdef NETATALK 96 #include <netproto/atalk/at.h> 97 #include <netproto/atalk/at_var.h> 98 #include <netproto/atalk/at_extern.h> 99 100 #define llc_snap_org_code llc_un.type_snap.org_code 101 #define llc_snap_ether_type llc_un.type_snap.ether_type 102 103 extern u_char at_org_code[3]; 104 extern u_char aarp_org_code[3]; 105 #endif /* NETATALK */ 106 107 /* netgraph node hooks for ng_ether(4) */ 108 void (*ng_ether_input_p)(struct ifnet *ifp, 109 struct mbuf **mp, const struct ether_header *eh); 110 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, 111 struct mbuf *m, const struct ether_header *eh); 112 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp); 113 void (*ng_ether_attach_p)(struct ifnet *ifp); 114 void (*ng_ether_detach_p)(struct ifnet *ifp); 115 116 int (*vlan_input_p)(const struct ether_header *eh, struct mbuf *m); 117 int (*vlan_input_tag_p)(struct mbuf *m, uint16_t t); 118 119 static int ether_output(struct ifnet *, struct mbuf *, struct sockaddr *, 120 struct rtentry *); 121 122 /* 123 * if_bridge support 124 */ 125 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *); 126 int (*bridge_output_p)(struct ifnet *, struct mbuf *, 127 struct sockaddr *, struct rtentry *); 128 void (*bridge_dn_p)(struct mbuf *, struct ifnet *); 129 130 static int ether_resolvemulti(struct ifnet *, struct sockaddr **, 131 struct sockaddr *); 132 133 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = { 134 0xff, 0xff, 0xff, 0xff, 0xff, 0xff 135 }; 136 137 #define gotoerr(e) do { error = (e); goto bad; } while (0) 138 #define IFP2AC(ifp) ((struct arpcom *)(ifp)) 139 140 static boolean_t ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, 141 struct ip_fw **rule, 142 const struct ether_header *eh, 143 boolean_t shared); 144 145 static int ether_ipfw; 146 SYSCTL_DECL(_net_link); 147 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet"); 148 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW, 149 ðer_ipfw, 0, "Pass ether pkts through firewall"); 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 * Assumes that ifp is actually pointer to arpcom structure. 157 */ 158 static int 159 ether_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst, 160 struct rtentry *rt) 161 { 162 struct ether_header *eh, *deh; 163 u_char *edst; 164 int loop_copy = 0; 165 int hlen = ETHER_HDR_LEN; /* link layer header length */ 166 struct arpcom *ac = IFP2AC(ifp); 167 int error; 168 169 ASSERT_SERIALIZED(ifp->if_serializer); 170 171 if (ifp->if_flags & IFF_MONITOR) 172 gotoerr(ENETDOWN); 173 if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING)) 174 gotoerr(ENETDOWN); 175 176 M_PREPEND(m, sizeof(struct ether_header), MB_DONTWAIT); 177 if (m == NULL) 178 return (ENOBUFS); 179 eh = mtod(m, struct ether_header *); 180 edst = eh->ether_dhost; 181 182 /* 183 * Fill in the destination ethernet address and frame type. 184 */ 185 switch (dst->sa_family) { 186 #ifdef INET 187 case AF_INET: 188 if (!arpresolve(ifp, rt, m, dst, edst)) 189 return (0); /* if not yet resolved */ 190 eh->ether_type = htons(ETHERTYPE_IP); 191 break; 192 #endif 193 #ifdef INET6 194 case AF_INET6: 195 if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, edst)) 196 return (0); /* Something bad happenned. */ 197 eh->ether_type = htons(ETHERTYPE_IPV6); 198 break; 199 #endif 200 #ifdef IPX 201 case AF_IPX: 202 if (ef_outputp != NULL) { 203 error = ef_outputp(ifp, &m, dst, &eh->ether_type, 204 &hlen); 205 if (error) 206 goto bad; 207 } else { 208 eh->ether_type = htons(ETHERTYPE_IPX); 209 bcopy(&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host), 210 edst, ETHER_ADDR_LEN); 211 } 212 break; 213 #endif 214 #ifdef NETATALK 215 case AF_APPLETALK: { 216 struct at_ifaddr *aa; 217 218 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) { 219 error = 0; /* XXX */ 220 goto bad; 221 } 222 /* 223 * In the phase 2 case, need to prepend an mbuf for 224 * the llc header. Since we must preserve the value 225 * of m, which is passed to us by value, we m_copy() 226 * the first mbuf, and use it for our llc header. 227 */ 228 if (aa->aa_flags & AFA_PHASE2) { 229 struct llc llc; 230 231 M_PREPEND(m, sizeof(struct llc), MB_DONTWAIT); 232 eh = mtod(m, struct ether_header *); 233 edst = eh->ether_dhost; 234 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 235 llc.llc_control = LLC_UI; 236 bcopy(at_org_code, llc.llc_snap_org_code, 237 sizeof at_org_code); 238 llc.llc_snap_ether_type = htons(ETHERTYPE_AT); 239 bcopy(&llc, 240 mtod(m, caddr_t) + sizeof(struct ether_header), 241 sizeof(struct llc)); 242 eh->ether_type = htons(m->m_pkthdr.len); 243 hlen = sizeof(struct llc) + ETHER_HDR_LEN; 244 } else { 245 eh->ether_type = htons(ETHERTYPE_AT); 246 } 247 if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst)) 248 return (0); 249 break; 250 } 251 #endif 252 #ifdef NS 253 case AF_NS: 254 switch(ns_nettype) { 255 default: 256 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */ 257 eh->ether_type = 0x8137; 258 break; 259 case 0x0: /* Novell 802.3 */ 260 eh->ether_type = htons(m->m_pkthdr.len); 261 break; 262 case 0xe0e0: /* Novell 802.2 and Token-Ring */ 263 M_PREPEND(m, 3, MB_DONTWAIT); 264 eh = mtod(m, struct ether_header *); 265 edst = eh->ether_dhost; 266 eh->ether_type = htons(m->m_pkthdr.len); 267 cp = mtod(m, u_char *) + sizeof(struct ether_header); 268 *cp++ = 0xE0; 269 *cp++ = 0xE0; 270 *cp++ = 0x03; 271 break; 272 } 273 bcopy(&(((struct sockaddr_ns *)dst)->sns_addr.x_host), edst, 274 ETHER_ADDR_LEN); 275 /* 276 * XXX if ns_thishost is the same as the node's ethernet 277 * address then just the default code will catch this anyhow. 278 * So I'm not sure if this next clause should be here at all? 279 * [JRE] 280 */ 281 if (bcmp(edst, &ns_thishost, ETHER_ADDR_LEN) == 0) { 282 m->m_pkthdr.rcvif = ifp; 283 netisr_dispatch(NETISR_NS, m); 284 return (error); 285 } 286 if (bcmp(edst, &ns_broadhost, ETHER_ADDR_LEN) == 0) 287 m->m_flags |= M_BCAST; 288 break; 289 #endif 290 case pseudo_AF_HDRCMPLT: 291 case AF_UNSPEC: 292 loop_copy = -1; /* if this is for us, don't do it */ 293 deh = (struct ether_header *)dst->sa_data; 294 memcpy(edst, deh->ether_dhost, ETHER_ADDR_LEN); 295 eh->ether_type = deh->ether_type; 296 break; 297 298 default: 299 if_printf(ifp, "can't handle af%d\n", dst->sa_family); 300 gotoerr(EAFNOSUPPORT); 301 } 302 303 if (dst->sa_family == pseudo_AF_HDRCMPLT) /* unlikely */ 304 memcpy(eh->ether_shost, 305 ((struct ether_header *)dst->sa_data)->ether_shost, 306 ETHER_ADDR_LEN); 307 else 308 memcpy(eh->ether_shost, ac->ac_enaddr, ETHER_ADDR_LEN); 309 310 /* 311 * Bridges require special output handling. 312 */ 313 if (ifp->if_bridge) { 314 KASSERT(bridge_output_p != NULL, 315 ("%s: if_bridge not loaded!", __func__)); 316 return ((*bridge_output_p)(ifp, m, NULL, NULL)); 317 } 318 319 /* 320 * If a simplex interface, and the packet is being sent to our 321 * Ethernet address or a broadcast address, loopback a copy. 322 * XXX To make a simplex device behave exactly like a duplex 323 * device, we should copy in the case of sending to our own 324 * ethernet address (thus letting the original actually appear 325 * on the wire). However, we don't do that here for security 326 * reasons and compatibility with the original behavior. 327 */ 328 if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) { 329 int csum_flags = 0; 330 331 if (m->m_pkthdr.csum_flags & CSUM_IP) 332 csum_flags |= (CSUM_IP_CHECKED | CSUM_IP_VALID); 333 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) 334 csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); 335 if ((m->m_flags & M_BCAST) || (loop_copy > 0)) { 336 struct mbuf *n; 337 338 if ((n = m_copypacket(m, MB_DONTWAIT)) != NULL) { 339 n->m_pkthdr.csum_flags |= csum_flags; 340 if (csum_flags & CSUM_DATA_VALID) 341 n->m_pkthdr.csum_data = 0xffff; 342 if_simloop(ifp, n, dst->sa_family, hlen); 343 } else 344 ifp->if_iqdrops++; 345 } else if (bcmp(eh->ether_dhost, eh->ether_shost, 346 ETHER_ADDR_LEN) == 0) { 347 m->m_pkthdr.csum_flags |= csum_flags; 348 if (csum_flags & CSUM_DATA_VALID) 349 m->m_pkthdr.csum_data = 0xffff; 350 if_simloop(ifp, m, dst->sa_family, hlen); 351 return (0); /* XXX */ 352 } 353 } 354 355 #ifdef CARP 356 if (ifp->if_carp && (error = carp_output(ifp, m, dst, NULL))) 357 goto bad; 358 #endif 359 360 361 /* Handle ng_ether(4) processing, if any */ 362 if (ng_ether_output_p != NULL) { 363 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) 364 goto bad; 365 if (m == NULL) 366 return (0); 367 } 368 369 /* Continue with link-layer output */ 370 return ether_output_frame(ifp, m); 371 372 bad: 373 m_freem(m); 374 return (error); 375 } 376 377 /* 378 * Ethernet link layer output routine to send a raw frame to the device. 379 * 380 * This assumes that the 14 byte Ethernet header is present and contiguous 381 * in the first mbuf. 382 */ 383 int 384 ether_output_frame(struct ifnet *ifp, struct mbuf *m) 385 { 386 struct ip_fw *rule = NULL; 387 int error = 0; 388 struct altq_pktattr pktattr; 389 struct m_tag *mtag; 390 391 ASSERT_SERIALIZED(ifp->if_serializer); 392 393 /* Extract info from dummynet tag */ 394 mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL); 395 if (mtag != NULL) { 396 rule = ((struct dn_pkt *)m_tag_data(mtag))->rule; 397 398 m_tag_delete(m, mtag); 399 mtag = NULL; 400 } 401 402 if (ifq_is_enabled(&ifp->if_snd)) 403 altq_etherclassify(&ifp->if_snd, m, &pktattr); 404 crit_enter(); 405 if (IPFW_LOADED && ether_ipfw != 0) { 406 struct ether_header save_eh, *eh; 407 408 eh = mtod(m, struct ether_header *); 409 save_eh = *eh; 410 m_adj(m, ETHER_HDR_LEN); 411 if (!ether_ipfw_chk(&m, ifp, &rule, eh, FALSE)) { 412 crit_exit(); 413 if (m != NULL) { 414 m_freem(m); 415 return ENOBUFS; /* pkt dropped */ 416 } else 417 return 0; /* consumed e.g. in a pipe */ 418 } 419 eh = mtod(m, struct ether_header *); 420 /* packet was ok, restore the ethernet header */ 421 if ((void *)(eh + 1) == (void *)m->m_data) { 422 m->m_data -= ETHER_HDR_LEN ; 423 m->m_len += ETHER_HDR_LEN ; 424 m->m_pkthdr.len += ETHER_HDR_LEN ; 425 } else { 426 M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT); 427 if (m == NULL) /* nope... */ { 428 crit_exit(); 429 return ENOBUFS; 430 } 431 bcopy(&save_eh, mtod(m, struct ether_header *), 432 ETHER_HDR_LEN); 433 } 434 } 435 crit_exit(); 436 437 /* 438 * Queue message on interface, update output statistics if 439 * successful, and start output if interface not yet active. 440 */ 441 error = ifq_handoff(ifp, m, &pktattr); 442 return (error); 443 } 444 445 /* 446 * ipfw processing for ethernet packets (in and out). 447 * The second parameter is NULL from ether_demux(), and ifp from 448 * ether_output_frame(). 449 */ 450 static boolean_t 451 ether_ipfw_chk( 452 struct mbuf **m0, 453 struct ifnet *dst, 454 struct ip_fw **rule, 455 const struct ether_header *eh, 456 boolean_t shared) 457 { 458 struct ether_header save_eh = *eh; /* might be a ptr in m */ 459 struct ip_fw_args args; 460 struct m_tag *mtag; 461 int i; 462 463 if (*rule != NULL && fw_one_pass) 464 return TRUE; /* dummynet packet, already partially processed */ 465 466 /* 467 * I need some amount of data to be contiguous, and in case others 468 * need the packet (shared==TRUE), it also better be in the first mbuf. 469 */ 470 i = min((*m0)->m_pkthdr.len, max_protohdr); 471 if (shared || (*m0)->m_len < i) { 472 *m0 = m_pullup(*m0, i); 473 if (*m0 == NULL) 474 return FALSE; 475 } 476 477 args.m = *m0; /* the packet we are looking at */ 478 args.oif = dst; /* destination, if any */ 479 if ((mtag = m_tag_find(*m0, PACKET_TAG_IPFW_DIVERT, NULL)) != NULL) 480 m_tag_delete(*m0, mtag); 481 args.rule = *rule; /* matching rule to restart */ 482 args.next_hop = NULL; /* we do not support forward yet */ 483 args.eh = &save_eh; /* MAC header for bridged/MAC packets */ 484 i = ip_fw_chk_ptr(&args); 485 *m0 = args.m; 486 *rule = args.rule; 487 488 if ((i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL) /* drop */ 489 return FALSE; 490 491 if (i == 0) /* a PASS rule. */ 492 return TRUE; 493 494 if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG)) { 495 /* 496 * Pass the pkt to dummynet, which consumes it. 497 * If shared, make a copy and keep the original. 498 */ 499 struct mbuf *m ; 500 501 if (shared) { 502 m = m_copypacket(*m0, MB_DONTWAIT); 503 if (m == NULL) 504 return FALSE; 505 } else { 506 m = *m0 ; /* pass the original to dummynet */ 507 *m0 = NULL ; /* and nothing back to the caller */ 508 } 509 /* 510 * Prepend the header, optimize for the common case of 511 * eh pointing into the mbuf. 512 */ 513 if ((const void *)(eh + 1) == (void *)m->m_data) { 514 m->m_data -= ETHER_HDR_LEN ; 515 m->m_len += ETHER_HDR_LEN ; 516 m->m_pkthdr.len += ETHER_HDR_LEN ; 517 } else { 518 M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT); 519 if (m == NULL) 520 return FALSE; 521 bcopy(&save_eh, mtod(m, struct ether_header *), 522 ETHER_HDR_LEN); 523 } 524 ip_dn_io_ptr(m, (i & 0xffff), 525 dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args); 526 return FALSE; 527 } 528 /* 529 * XXX at some point add support for divert/forward actions. 530 * If none of the above matches, we have to drop the pkt. 531 */ 532 return FALSE; 533 } 534 535 /* 536 * Process a received Ethernet packet. 537 * 538 * The ethernet header is assumed to be in the mbuf so the caller 539 * MUST MAKE SURE that there are at least sizeof(struct ether_header) 540 * bytes in the first mbuf. 541 * 542 * This allows us to concentrate in one place a bunch of code which 543 * is replicated in all device drivers. Also, many functions called 544 * from ether_input() try to put the eh back into the mbuf, so we 545 * can later propagate the 'contiguous packet' interface to them. 546 * 547 * NOTA BENE: for all drivers "eh" is a pointer into the first mbuf or 548 * cluster, right before m_data. So be very careful when working on m, 549 * as you could destroy *eh !! 550 * 551 * First we perform any link layer operations, then continue to the 552 * upper layers with ether_demux(). 553 */ 554 void 555 ether_input(struct ifnet *ifp, struct mbuf *m) 556 { 557 struct ether_header *eh; 558 559 ASSERT_SERIALIZED(ifp->if_serializer); 560 561 if (m->m_len < sizeof(struct ether_header)) { 562 /* XXX error in the caller. */ 563 m_freem(m); 564 return; 565 } 566 m->m_pkthdr.rcvif = ifp; 567 568 BPF_MTAP(ifp, m); 569 570 ifp->if_ibytes += m->m_pkthdr.len; 571 572 if (ifp->if_flags & IFF_MONITOR) { 573 /* 574 * Interface marked for monitoring; discard packet. 575 */ 576 m_freem(m); 577 return; 578 } 579 580 /* 581 * Tap the packet off here for a bridge. bridge_input() 582 * will return NULL if it has consumed the packet, otherwise 583 * it gets processed as normal. Note that bridge_input() 584 * will always return the original packet if we need to 585 * process it locally. 586 */ 587 if (ifp->if_bridge) { 588 KASSERT(bridge_input_p != NULL, 589 ("%s: if_bridge not loaded!", __func__)); 590 591 if(m->m_flags & M_PROTO1) { 592 m->m_flags &= ~M_PROTO1; 593 } else { 594 /* clear M_PROMISC, in case the packets comes from a vlan */ 595 /* m->m_flags &= ~M_PROMISC; */ 596 lwkt_serialize_exit(ifp->if_serializer); 597 m = (*bridge_input_p)(ifp, m); 598 lwkt_serialize_enter(ifp->if_serializer); 599 if (m == NULL) 600 return; 601 602 KASSERT(ifp == m->m_pkthdr.rcvif, 603 ("bridge_input_p changed rcvif\n")); 604 } 605 } 606 607 eh = mtod(m, struct ether_header *); 608 609 /* XXX old crufty stuff, needs to be removed */ 610 m_adj(m, sizeof(struct ether_header)); 611 /* XXX */ 612 /* m->m_pkthdr.len = m->m_len; */ 613 614 /* Handle ng_ether(4) processing, if any */ 615 if (ng_ether_input_p != NULL) { 616 lwkt_serialize_exit(ifp->if_serializer); 617 (*ng_ether_input_p)(ifp, &m, eh); 618 lwkt_serialize_enter(ifp->if_serializer); 619 if (m == NULL) 620 return; 621 } 622 623 /* Continue with upper layer processing */ 624 ether_demux(ifp, eh, m); 625 } 626 627 /* 628 * Upper layer processing for a received Ethernet packet. 629 */ 630 void 631 ether_demux(struct ifnet *ifp, struct ether_header *eh0, struct mbuf *m) 632 { 633 struct ether_header eh; 634 int isr; 635 u_short ether_type; 636 struct ip_fw *rule = NULL; 637 struct m_tag *mtag; 638 #ifdef NETATALK 639 struct llc *l; 640 #endif 641 642 eh = *eh0; 643 644 /* Extract info from dummynet tag */ 645 mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL); 646 if (mtag != NULL) { 647 rule = ((struct dn_pkt *)m_tag_data(mtag))->rule; 648 KKASSERT(ifp == NULL); 649 ifp = m->m_pkthdr.rcvif; 650 651 m_tag_delete(m, mtag); 652 mtag = NULL; 653 } 654 if (rule) /* packet is passing the second time */ 655 goto post_stats; 656 657 #ifdef CARP 658 /* 659 * XXX: Okay, we need to call carp_forus() and - if it is for 660 * us jump over code that does the normal check 661 * "ac_enaddr == ether_dhost". The check sequence is a bit 662 * different from OpenBSD, so we jump over as few code as 663 * possible, to catch _all_ sanity checks. This needs 664 * evaluation, to see if the carp ether_dhost values break any 665 * of these checks! 666 */ 667 if (ifp->if_carp && carp_forus(ifp->if_carp, eh.ether_dhost)) 668 goto pre_stats; 669 #endif 670 671 /* 672 * Discard packet if upper layers shouldn't see it because 673 * it was unicast to a different Ethernet address. If the 674 * driver is working properly, then this situation can only 675 * happen when the interface is in promiscuous mode. 676 */ 677 if (((ifp->if_flags & (IFF_PROMISC | IFF_PPROMISC)) == IFF_PROMISC) && 678 (eh.ether_dhost[0] & 1) == 0 && 679 bcmp(eh.ether_dhost, IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN)) { 680 m_freem(m); 681 return; 682 } 683 684 #ifdef CARP 685 pre_stats: 686 #endif 687 688 /* Discard packet if interface is not up */ 689 if (!(ifp->if_flags & IFF_UP)) { 690 m_freem(m); 691 return; 692 } 693 if (eh.ether_dhost[0] & 1) { 694 if (bcmp(ifp->if_broadcastaddr, eh.ether_dhost, 695 ifp->if_addrlen) == 0) 696 m->m_flags |= M_BCAST; 697 else 698 m->m_flags |= M_MCAST; 699 ifp->if_imcasts++; 700 } 701 702 post_stats: 703 if (IPFW_LOADED && ether_ipfw != 0) { 704 if (!ether_ipfw_chk(&m, NULL, &rule, &eh, FALSE)) { 705 m_freem(m); 706 return; 707 } 708 } 709 710 ether_type = ntohs(eh.ether_type); 711 712 switch (ether_type) { 713 #ifdef INET 714 case ETHERTYPE_IP: 715 if (ipflow_fastforward(m, ifp->if_serializer)) 716 return; 717 isr = NETISR_IP; 718 break; 719 720 case ETHERTYPE_ARP: 721 if (ifp->if_flags & IFF_NOARP) { 722 /* Discard packet if ARP is disabled on interface */ 723 m_freem(m); 724 return; 725 } 726 isr = NETISR_ARP; 727 break; 728 #endif 729 730 #ifdef INET6 731 case ETHERTYPE_IPV6: 732 isr = NETISR_IPV6; 733 break; 734 #endif 735 736 #ifdef IPX 737 case ETHERTYPE_IPX: 738 if (ef_inputp && ef_inputp(ifp, &eh, m) == 0) 739 return; 740 isr = NETISR_IPX; 741 break; 742 #endif 743 744 #ifdef NS 745 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */ 746 isr = NETISR_NS; 747 break; 748 749 #endif 750 751 #ifdef NETATALK 752 case ETHERTYPE_AT: 753 isr = NETISR_ATALK1; 754 break; 755 case ETHERTYPE_AARP: 756 isr = NETISR_AARP; 757 break; 758 #endif 759 760 case ETHERTYPE_VLAN: 761 if (vlan_input_p != NULL) 762 (*vlan_input_p)(&eh, m); 763 else { 764 m->m_pkthdr.rcvif->if_noproto++; 765 m_freem(m); 766 } 767 return; 768 769 default: 770 #ifdef IPX 771 if (ef_inputp && ef_inputp(ifp, &eh, m) == 0) 772 return; 773 #endif 774 #ifdef NS 775 checksum = mtod(m, ushort *); 776 /* Novell 802.3 */ 777 if ((ether_type <= ETHERMTU) && 778 ((*checksum == 0xffff) || (*checksum == 0xE0E0))) { 779 if (*checksum == 0xE0E0) { 780 m->m_pkthdr.len -= 3; 781 m->m_len -= 3; 782 m->m_data += 3; 783 } 784 isr = NETISR_NS; 785 break; 786 } 787 #endif 788 #ifdef NETATALK 789 if (ether_type > ETHERMTU) 790 goto dropanyway; 791 l = mtod(m, struct llc *); 792 if (l->llc_dsap == LLC_SNAP_LSAP && 793 l->llc_ssap == LLC_SNAP_LSAP && 794 l->llc_control == LLC_UI) { 795 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code, 796 sizeof at_org_code) == 0 && 797 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) { 798 m_adj(m, sizeof(struct llc)); 799 isr = NETISR_ATALK2; 800 break; 801 } 802 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code, 803 sizeof aarp_org_code) == 0 && 804 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) { 805 m_adj(m, sizeof(struct llc)); 806 isr = NETISR_AARP; 807 break; 808 } 809 } 810 dropanyway: 811 #endif 812 if (ng_ether_input_orphan_p != NULL) 813 (*ng_ether_input_orphan_p)(ifp, m, &eh); 814 else 815 m_freem(m); 816 return; 817 } 818 netisr_dispatch(isr, m); 819 } 820 821 /* 822 * Perform common duties while attaching to interface list 823 */ 824 825 void 826 ether_ifattach(struct ifnet *ifp, uint8_t *lla, lwkt_serialize_t serializer) 827 { 828 ether_ifattach_bpf(ifp, lla, DLT_EN10MB, sizeof(struct ether_header), 829 serializer); 830 } 831 832 void 833 ether_ifattach_bpf(struct ifnet *ifp, uint8_t *lla, u_int dlt, u_int hdrlen, 834 lwkt_serialize_t serializer) 835 { 836 struct sockaddr_dl *sdl; 837 838 ifp->if_type = IFT_ETHER; 839 ifp->if_addrlen = ETHER_ADDR_LEN; 840 ifp->if_hdrlen = ETHER_HDR_LEN; 841 if_attach(ifp, serializer); 842 ifp->if_mtu = ETHERMTU; 843 if (ifp->if_baudrate == 0) 844 ifp->if_baudrate = 10000000; 845 ifp->if_output = ether_output; 846 ifp->if_input = ether_input; 847 ifp->if_resolvemulti = ether_resolvemulti; 848 ifp->if_broadcastaddr = etherbroadcastaddr; 849 sdl = IF_LLSOCKADDR(ifp); 850 sdl->sdl_type = IFT_ETHER; 851 sdl->sdl_alen = ifp->if_addrlen; 852 bcopy(lla, LLADDR(sdl), ifp->if_addrlen); 853 /* 854 * XXX Keep the current drivers happy. 855 * XXX Remove once all drivers have been cleaned up 856 */ 857 if (lla != IFP2AC(ifp)->ac_enaddr) 858 bcopy(lla, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen); 859 bpfattach(ifp, dlt, hdrlen); 860 if (ng_ether_attach_p != NULL) 861 (*ng_ether_attach_p)(ifp); 862 863 if_printf(ifp, "MAC address: %6D\n", lla, ":"); 864 } 865 866 /* 867 * Perform common duties while detaching an Ethernet interface 868 */ 869 void 870 ether_ifdetach(struct ifnet *ifp) 871 { 872 if_down(ifp); 873 874 if (ng_ether_detach_p != NULL) 875 (*ng_ether_detach_p)(ifp); 876 bpfdetach(ifp); 877 if_detach(ifp); 878 } 879 880 int 881 ether_ioctl(struct ifnet *ifp, int command, caddr_t data) 882 { 883 struct ifaddr *ifa = (struct ifaddr *) data; 884 struct ifreq *ifr = (struct ifreq *) data; 885 int error = 0; 886 887 #define IF_INIT(ifp) \ 888 do { \ 889 if (((ifp)->if_flags & IFF_UP) == 0) { \ 890 (ifp)->if_flags |= IFF_UP; \ 891 (ifp)->if_init((ifp)->if_softc); \ 892 } \ 893 } while (0) 894 895 ASSERT_SERIALIZED(ifp->if_serializer); 896 897 switch (command) { 898 case SIOCSIFADDR: 899 switch (ifa->ifa_addr->sa_family) { 900 #ifdef INET 901 case AF_INET: 902 IF_INIT(ifp); /* before arpwhohas */ 903 arp_ifinit(ifp, ifa); 904 break; 905 #endif 906 #ifdef IPX 907 /* 908 * XXX - This code is probably wrong 909 */ 910 case AF_IPX: 911 { 912 struct ipx_addr *ina = &IA_SIPX(ifa)->sipx_addr; 913 struct arpcom *ac = IFP2AC(ifp); 914 915 if (ipx_nullhost(*ina)) 916 ina->x_host = *(union ipx_host *) ac->ac_enaddr; 917 else 918 bcopy(ina->x_host.c_host, ac->ac_enaddr, 919 sizeof ac->ac_enaddr); 920 921 IF_INIT(ifp); /* Set new address. */ 922 break; 923 } 924 #endif 925 #ifdef NS 926 /* 927 * XXX - This code is probably wrong 928 */ 929 case AF_NS: 930 { 931 struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr); 932 struct arpcom *ac = IFP2AC(ifp); 933 934 if (ns_nullhost(*ina)) 935 ina->x_host = *(union ns_host *)(ac->ac_enaddr); 936 else 937 bcopy(ina->x_host.c_host, ac->ac_enaddr, 938 sizeof ac->ac_enaddr); 939 940 /* 941 * Set new address 942 */ 943 IF_INIT(ifp); 944 break; 945 } 946 #endif 947 default: 948 IF_INIT(ifp); 949 break; 950 } 951 break; 952 953 case SIOCGIFADDR: 954 bcopy(IFP2AC(ifp)->ac_enaddr, 955 ((struct sockaddr *)ifr->ifr_data)->sa_data, 956 ETHER_ADDR_LEN); 957 break; 958 959 case SIOCSIFMTU: 960 /* 961 * Set the interface MTU. 962 */ 963 if (ifr->ifr_mtu > ETHERMTU) { 964 error = EINVAL; 965 } else { 966 ifp->if_mtu = ifr->ifr_mtu; 967 } 968 break; 969 default: 970 error = EINVAL; 971 break; 972 } 973 return (error); 974 975 #undef IF_INIT 976 } 977 978 int 979 ether_resolvemulti( 980 struct ifnet *ifp, 981 struct sockaddr **llsa, 982 struct sockaddr *sa) 983 { 984 struct sockaddr_dl *sdl; 985 struct sockaddr_in *sin; 986 #ifdef INET6 987 struct sockaddr_in6 *sin6; 988 #endif 989 u_char *e_addr; 990 991 switch(sa->sa_family) { 992 case AF_LINK: 993 /* 994 * No mapping needed. Just check that it's a valid MC address. 995 */ 996 sdl = (struct sockaddr_dl *)sa; 997 e_addr = LLADDR(sdl); 998 if ((e_addr[0] & 1) != 1) 999 return EADDRNOTAVAIL; 1000 *llsa = 0; 1001 return 0; 1002 1003 #ifdef INET 1004 case AF_INET: 1005 sin = (struct sockaddr_in *)sa; 1006 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) 1007 return EADDRNOTAVAIL; 1008 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, 1009 M_WAITOK | M_ZERO); 1010 sdl->sdl_len = sizeof *sdl; 1011 sdl->sdl_family = AF_LINK; 1012 sdl->sdl_index = ifp->if_index; 1013 sdl->sdl_type = IFT_ETHER; 1014 sdl->sdl_alen = ETHER_ADDR_LEN; 1015 e_addr = LLADDR(sdl); 1016 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr); 1017 *llsa = (struct sockaddr *)sdl; 1018 return 0; 1019 #endif 1020 #ifdef INET6 1021 case AF_INET6: 1022 sin6 = (struct sockaddr_in6 *)sa; 1023 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1024 /* 1025 * An IP6 address of 0 means listen to all 1026 * of the Ethernet multicast address used for IP6. 1027 * (This is used for multicast routers.) 1028 */ 1029 ifp->if_flags |= IFF_ALLMULTI; 1030 *llsa = 0; 1031 return 0; 1032 } 1033 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) 1034 return EADDRNOTAVAIL; 1035 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, 1036 M_WAITOK | M_ZERO); 1037 sdl->sdl_len = sizeof *sdl; 1038 sdl->sdl_family = AF_LINK; 1039 sdl->sdl_index = ifp->if_index; 1040 sdl->sdl_type = IFT_ETHER; 1041 sdl->sdl_alen = ETHER_ADDR_LEN; 1042 e_addr = LLADDR(sdl); 1043 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr); 1044 *llsa = (struct sockaddr *)sdl; 1045 return 0; 1046 #endif 1047 1048 default: 1049 /* 1050 * Well, the text isn't quite right, but it's the name 1051 * that counts... 1052 */ 1053 return EAFNOSUPPORT; 1054 } 1055 } 1056 1057 #if 0 1058 /* 1059 * This is for reference. We have a table-driven version 1060 * of the little-endian crc32 generator, which is faster 1061 * than the double-loop. 1062 */ 1063 uint32_t 1064 ether_crc32_le(const uint8_t *buf, size_t len) 1065 { 1066 uint32_t c, crc, carry; 1067 size_t i, j; 1068 1069 crc = 0xffffffffU; /* initial value */ 1070 1071 for (i = 0; i < len; i++) { 1072 c = buf[i]; 1073 for (j = 0; j < 8; j++) { 1074 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1075 crc >>= 1; 1076 c >>= 1; 1077 if (carry) 1078 crc = (crc ^ ETHER_CRC_POLY_LE); 1079 } 1080 } 1081 1082 return (crc); 1083 } 1084 #else 1085 uint32_t 1086 ether_crc32_le(const uint8_t *buf, size_t len) 1087 { 1088 static const uint32_t crctab[] = { 1089 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1090 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1091 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1092 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1093 }; 1094 uint32_t crc; 1095 size_t i; 1096 1097 crc = 0xffffffffU; /* initial value */ 1098 1099 for (i = 0; i < len; i++) { 1100 crc ^= buf[i]; 1101 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1102 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1103 } 1104 1105 return (crc); 1106 } 1107 #endif 1108 1109 uint32_t 1110 ether_crc32_be(const uint8_t *buf, size_t len) 1111 { 1112 uint32_t c, crc, carry; 1113 size_t i, j; 1114 1115 crc = 0xffffffffU; /* initial value */ 1116 1117 for (i = 0; i < len; i++) { 1118 c = buf[i]; 1119 for (j = 0; j < 8; j++) { 1120 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1121 crc <<= 1; 1122 c >>= 1; 1123 if (carry) 1124 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1125 } 1126 } 1127 1128 return (crc); 1129 } 1130 1131 /* 1132 * find the size of ethernet header, and call classifier 1133 */ 1134 void 1135 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m, 1136 struct altq_pktattr *pktattr) 1137 { 1138 struct ether_header *eh; 1139 uint16_t ether_type; 1140 int hlen, af, hdrsize; 1141 caddr_t hdr; 1142 1143 hlen = sizeof(struct ether_header); 1144 eh = mtod(m, struct ether_header *); 1145 1146 ether_type = ntohs(eh->ether_type); 1147 if (ether_type < ETHERMTU) { 1148 /* ick! LLC/SNAP */ 1149 struct llc *llc = (struct llc *)(eh + 1); 1150 hlen += 8; 1151 1152 if (m->m_len < hlen || 1153 llc->llc_dsap != LLC_SNAP_LSAP || 1154 llc->llc_ssap != LLC_SNAP_LSAP || 1155 llc->llc_control != LLC_UI) 1156 goto bad; /* not snap! */ 1157 1158 ether_type = ntohs(llc->llc_un.type_snap.ether_type); 1159 } 1160 1161 if (ether_type == ETHERTYPE_IP) { 1162 af = AF_INET; 1163 hdrsize = 20; /* sizeof(struct ip) */ 1164 #ifdef INET6 1165 } else if (ether_type == ETHERTYPE_IPV6) { 1166 af = AF_INET6; 1167 hdrsize = 40; /* sizeof(struct ip6_hdr) */ 1168 #endif 1169 } else 1170 goto bad; 1171 1172 while (m->m_len <= hlen) { 1173 hlen -= m->m_len; 1174 m = m->m_next; 1175 } 1176 hdr = m->m_data + hlen; 1177 if (m->m_len < hlen + hdrsize) { 1178 /* 1179 * ip header is not in a single mbuf. this should not 1180 * happen in the current code. 1181 * (todo: use m_pulldown in the future) 1182 */ 1183 goto bad; 1184 } 1185 m->m_data += hlen; 1186 m->m_len -= hlen; 1187 ifq_classify(ifq, m, af, pktattr); 1188 m->m_data -= hlen; 1189 m->m_len += hlen; 1190 1191 return; 1192 1193 bad: 1194 pktattr->pattr_class = NULL; 1195 pktattr->pattr_hdr = NULL; 1196 pktattr->pattr_af = AF_UNSPEC; 1197 } 1198