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_netgraph.h" 37 #include "opt_mbuf_profiling.h" 38 39 #include <sys/param.h> 40 #include <sys/systm.h> 41 #include <sys/kernel.h> 42 #include <sys/lock.h> 43 #include <sys/malloc.h> 44 #include <sys/module.h> 45 #include <sys/mbuf.h> 46 #include <sys/random.h> 47 #include <sys/socket.h> 48 #include <sys/sockio.h> 49 #include <sys/sysctl.h> 50 #include <sys/uuid.h> 51 52 #include <net/if.h> 53 #include <net/if_var.h> 54 #include <net/if_arp.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/bpf.h> 61 #include <net/ethernet.h> 62 #include <net/if_bridgevar.h> 63 #include <net/if_vlan_var.h> 64 #include <net/if_llatbl.h> 65 #include <net/pfil.h> 66 #include <net/vnet.h> 67 68 #include <netpfil/pf/pf_mtag.h> 69 70 #if defined(INET) || defined(INET6) 71 #include <netinet/in.h> 72 #include <netinet/in_var.h> 73 #include <netinet/if_ether.h> 74 #include <netinet/ip_carp.h> 75 #include <netinet/ip_var.h> 76 #endif 77 #ifdef INET6 78 #include <netinet6/nd6.h> 79 #endif 80 81 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m); 82 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, 83 const struct sockaddr *dst, short *tp, int *hlen); 84 85 #ifdef NETATALK 86 #include <netatalk/at.h> 87 #include <netatalk/at_var.h> 88 #include <netatalk/at_extern.h> 89 90 #define llc_snap_org_code llc_un.type_snap.org_code 91 #define llc_snap_ether_type llc_un.type_snap.ether_type 92 93 extern u_char at_org_code[3]; 94 extern u_char aarp_org_code[3]; 95 #endif /* NETATALK */ 96 97 #include <security/mac/mac_framework.h> 98 99 #ifdef CTASSERT 100 CTASSERT(sizeof (struct ether_header) == ETHER_ADDR_LEN * 2 + 2); 101 CTASSERT(sizeof (struct ether_addr) == ETHER_ADDR_LEN); 102 #endif 103 104 VNET_DEFINE(struct pfil_head, link_pfil_hook); /* Packet filter hooks */ 105 106 /* netgraph node hooks for ng_ether(4) */ 107 void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp); 108 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m); 109 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp); 110 void (*ng_ether_attach_p)(struct ifnet *ifp); 111 void (*ng_ether_detach_p)(struct ifnet *ifp); 112 113 void (*vlan_input_p)(struct ifnet *, struct mbuf *); 114 115 /* if_bridge(4) support */ 116 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *); 117 int (*bridge_output_p)(struct ifnet *, struct mbuf *, 118 struct sockaddr *, struct rtentry *); 119 void (*bridge_dn_p)(struct mbuf *, struct ifnet *); 120 121 /* if_lagg(4) support */ 122 struct mbuf *(*lagg_input_p)(struct ifnet *, struct mbuf *); 123 124 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] = 125 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 126 127 static int ether_resolvemulti(struct ifnet *, struct sockaddr **, 128 struct sockaddr *); 129 #ifdef VIMAGE 130 static void ether_reassign(struct ifnet *, struct vnet *, char *); 131 #endif 132 133 /* XXX: should be in an arp support file, not here */ 134 static MALLOC_DEFINE(M_ARPCOM, "arpcom", "802.* interface internals"); 135 136 #define ETHER_IS_BROADCAST(addr) \ 137 (bcmp(etherbroadcastaddr, (addr), ETHER_ADDR_LEN) == 0) 138 139 #define senderr(e) do { error = (e); goto bad;} while (0) 140 141 static void 142 update_mbuf_csumflags(struct mbuf *src, struct mbuf *dst) 143 { 144 int csum_flags = 0; 145 146 if (src->m_pkthdr.csum_flags & CSUM_IP) 147 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID); 148 if (src->m_pkthdr.csum_flags & CSUM_DELAY_DATA) 149 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR); 150 if (src->m_pkthdr.csum_flags & CSUM_SCTP) 151 csum_flags |= CSUM_SCTP_VALID; 152 dst->m_pkthdr.csum_flags |= csum_flags; 153 if (csum_flags & CSUM_DATA_VALID) 154 dst->m_pkthdr.csum_data = 0xffff; 155 } 156 157 /* 158 * Ethernet output routine. 159 * Encapsulate a packet of type family for the local net. 160 * Use trailer local net encapsulation if enough data in first 161 * packet leaves a multiple of 512 bytes of data in remainder. 162 */ 163 int 164 ether_output(struct ifnet *ifp, struct mbuf *m, 165 const struct sockaddr *dst, struct route *ro) 166 { 167 short type; 168 int error = 0, hdrcmplt = 0; 169 u_char esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN]; 170 struct llentry *lle = NULL; 171 struct rtentry *rt0 = NULL; 172 struct ether_header *eh; 173 struct pf_mtag *t; 174 int loop_copy = 1; 175 int hlen; /* link layer header length */ 176 177 if (ro != NULL) { 178 if (!(m->m_flags & (M_BCAST | M_MCAST))) 179 lle = ro->ro_lle; 180 rt0 = ro->ro_rt; 181 } 182 #ifdef MAC 183 error = mac_ifnet_check_transmit(ifp, m); 184 if (error) 185 senderr(error); 186 #endif 187 188 M_PROFILE(m); 189 if (ifp->if_flags & IFF_MONITOR) 190 senderr(ENETDOWN); 191 if (!((ifp->if_flags & IFF_UP) && 192 (ifp->if_drv_flags & IFF_DRV_RUNNING))) 193 senderr(ENETDOWN); 194 195 hlen = ETHER_HDR_LEN; 196 switch (dst->sa_family) { 197 #ifdef INET 198 case AF_INET: 199 if (lle != NULL && (lle->la_flags & LLE_VALID)) 200 memcpy(edst, &lle->ll_addr.mac16, sizeof(edst)); 201 else 202 error = arpresolve(ifp, rt0, m, dst, edst, &lle); 203 if (error) 204 return (error == EWOULDBLOCK ? 0 : error); 205 type = htons(ETHERTYPE_IP); 206 break; 207 case AF_ARP: 208 { 209 struct arphdr *ah; 210 ah = mtod(m, struct arphdr *); 211 ah->ar_hrd = htons(ARPHRD_ETHER); 212 213 loop_copy = 0; /* if this is for us, don't do it */ 214 215 switch(ntohs(ah->ar_op)) { 216 case ARPOP_REVREQUEST: 217 case ARPOP_REVREPLY: 218 type = htons(ETHERTYPE_REVARP); 219 break; 220 case ARPOP_REQUEST: 221 case ARPOP_REPLY: 222 default: 223 type = htons(ETHERTYPE_ARP); 224 break; 225 } 226 227 if (m->m_flags & M_BCAST) 228 bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN); 229 else 230 bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN); 231 232 } 233 break; 234 #endif 235 #ifdef INET6 236 case AF_INET6: 237 if (lle != NULL && (lle->la_flags & LLE_VALID)) 238 memcpy(edst, &lle->ll_addr.mac16, sizeof(edst)); 239 else 240 error = nd6_storelladdr(ifp, m, dst, (u_char *)edst, &lle); 241 if (error) 242 return error; 243 type = htons(ETHERTYPE_IPV6); 244 break; 245 #endif 246 #ifdef NETATALK 247 case AF_APPLETALK: 248 { 249 struct at_ifaddr *aa; 250 251 if ((aa = at_ifawithnet((const struct sockaddr_at *)dst)) == NULL) 252 senderr(EHOSTUNREACH); /* XXX */ 253 if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) { 254 ifa_free(&aa->aa_ifa); 255 return (0); 256 } 257 /* 258 * In the phase 2 case, need to prepend an mbuf for the llc header. 259 */ 260 if ( aa->aa_flags & AFA_PHASE2 ) { 261 struct llc llc; 262 263 ifa_free(&aa->aa_ifa); 264 M_PREPEND(m, LLC_SNAPFRAMELEN, M_NOWAIT); 265 if (m == NULL) 266 senderr(ENOBUFS); 267 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 268 llc.llc_control = LLC_UI; 269 bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code)); 270 llc.llc_snap_ether_type = htons( ETHERTYPE_AT ); 271 bcopy(&llc, mtod(m, caddr_t), LLC_SNAPFRAMELEN); 272 type = htons(m->m_pkthdr.len); 273 hlen = LLC_SNAPFRAMELEN + ETHER_HDR_LEN; 274 } else { 275 ifa_free(&aa->aa_ifa); 276 type = htons(ETHERTYPE_AT); 277 } 278 break; 279 } 280 #endif /* NETATALK */ 281 282 case pseudo_AF_HDRCMPLT: 283 { 284 const struct ether_header *eh; 285 286 hdrcmplt = 1; 287 eh = (const struct ether_header *)dst->sa_data; 288 (void)memcpy(esrc, eh->ether_shost, sizeof (esrc)); 289 /* FALLTHROUGH */ 290 291 case AF_UNSPEC: 292 loop_copy = 0; /* if this is for us, don't do it */ 293 eh = (const struct ether_header *)dst->sa_data; 294 (void)memcpy(edst, eh->ether_dhost, sizeof (edst)); 295 type = eh->ether_type; 296 break; 297 } 298 default: 299 if_printf(ifp, "can't handle af%d\n", dst->sa_family); 300 senderr(EAFNOSUPPORT); 301 } 302 303 if (lle != NULL && (lle->la_flags & LLE_IFADDR)) { 304 update_mbuf_csumflags(m, m); 305 return (if_simloop(ifp, m, dst->sa_family, 0)); 306 } 307 308 /* 309 * Add local net header. If no space in first mbuf, 310 * allocate another. 311 */ 312 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT); 313 if (m == NULL) 314 senderr(ENOBUFS); 315 eh = mtod(m, struct ether_header *); 316 (void)memcpy(&eh->ether_type, &type, 317 sizeof(eh->ether_type)); 318 (void)memcpy(eh->ether_dhost, edst, sizeof (edst)); 319 if (hdrcmplt) 320 (void)memcpy(eh->ether_shost, esrc, 321 sizeof(eh->ether_shost)); 322 else 323 (void)memcpy(eh->ether_shost, IF_LLADDR(ifp), 324 sizeof(eh->ether_shost)); 325 326 /* 327 * If a simplex interface, and the packet is being sent to our 328 * Ethernet address or a broadcast address, loopback a copy. 329 * XXX To make a simplex device behave exactly like a duplex 330 * device, we should copy in the case of sending to our own 331 * ethernet address (thus letting the original actually appear 332 * on the wire). However, we don't do that here for security 333 * reasons and compatibility with the original behavior. 334 */ 335 if ((ifp->if_flags & IFF_SIMPLEX) && loop_copy && 336 ((t = pf_find_mtag(m)) == NULL || !t->routed)) { 337 if (m->m_flags & M_BCAST) { 338 struct mbuf *n; 339 340 /* 341 * Because if_simloop() modifies the packet, we need a 342 * writable copy through m_dup() instead of a readonly 343 * one as m_copy[m] would give us. The alternative would 344 * be to modify if_simloop() to handle the readonly mbuf, 345 * but performancewise it is mostly equivalent (trading 346 * extra data copying vs. extra locking). 347 * 348 * XXX This is a local workaround. A number of less 349 * often used kernel parts suffer from the same bug. 350 * See PR kern/105943 for a proposed general solution. 351 */ 352 if ((n = m_dup(m, M_NOWAIT)) != NULL) { 353 update_mbuf_csumflags(m, n); 354 (void)if_simloop(ifp, n, dst->sa_family, hlen); 355 } else 356 ifp->if_iqdrops++; 357 } else if (bcmp(eh->ether_dhost, eh->ether_shost, 358 ETHER_ADDR_LEN) == 0) { 359 update_mbuf_csumflags(m, m); 360 (void) if_simloop(ifp, m, dst->sa_family, hlen); 361 return (0); /* XXX */ 362 } 363 } 364 365 /* 366 * Bridges require special output handling. 367 */ 368 if (ifp->if_bridge) { 369 BRIDGE_OUTPUT(ifp, m, error); 370 return (error); 371 } 372 373 #if defined(INET) || defined(INET6) 374 if (ifp->if_carp && 375 (error = (*carp_output_p)(ifp, m, dst))) 376 goto bad; 377 #endif 378 379 /* Handle ng_ether(4) processing, if any */ 380 if (IFP2AC(ifp)->ac_netgraph != NULL) { 381 KASSERT(ng_ether_output_p != NULL, 382 ("ng_ether_output_p is NULL")); 383 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) { 384 bad: if (m != NULL) 385 m_freem(m); 386 return (error); 387 } 388 if (m == NULL) 389 return (0); 390 } 391 392 /* Continue with link-layer output */ 393 return ether_output_frame(ifp, m); 394 } 395 396 /* 397 * Ethernet link layer output routine to send a raw frame to the device. 398 * 399 * This assumes that the 14 byte Ethernet header is present and contiguous 400 * in the first mbuf (if BRIDGE'ing). 401 */ 402 int 403 ether_output_frame(struct ifnet *ifp, struct mbuf *m) 404 { 405 int i; 406 407 if (PFIL_HOOKED(&V_link_pfil_hook)) { 408 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_OUT, NULL); 409 410 if (i != 0) 411 return (EACCES); 412 413 if (m == NULL) 414 return (0); 415 } 416 417 /* 418 * Queue message on interface, update output statistics if 419 * successful, and start output if interface not yet active. 420 */ 421 return ((ifp->if_transmit)(ifp, m)); 422 } 423 424 #if defined(INET) || defined(INET6) 425 #endif 426 427 /* 428 * Process a received Ethernet packet; the packet is in the 429 * mbuf chain m with the ethernet header at the front. 430 */ 431 static void 432 ether_input_internal(struct ifnet *ifp, struct mbuf *m) 433 { 434 struct ether_header *eh; 435 u_short etype; 436 437 if ((ifp->if_flags & IFF_UP) == 0) { 438 m_freem(m); 439 return; 440 } 441 #ifdef DIAGNOSTIC 442 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 443 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n"); 444 m_freem(m); 445 return; 446 } 447 #endif 448 /* 449 * Do consistency checks to verify assumptions 450 * made by code past this point. 451 */ 452 if ((m->m_flags & M_PKTHDR) == 0) { 453 if_printf(ifp, "discard frame w/o packet header\n"); 454 ifp->if_ierrors++; 455 m_freem(m); 456 return; 457 } 458 if (m->m_len < ETHER_HDR_LEN) { 459 /* XXX maybe should pullup? */ 460 if_printf(ifp, "discard frame w/o leading ethernet " 461 "header (len %u pkt len %u)\n", 462 m->m_len, m->m_pkthdr.len); 463 ifp->if_ierrors++; 464 m_freem(m); 465 return; 466 } 467 eh = mtod(m, struct ether_header *); 468 etype = ntohs(eh->ether_type); 469 if (m->m_pkthdr.rcvif == NULL) { 470 if_printf(ifp, "discard frame w/o interface pointer\n"); 471 ifp->if_ierrors++; 472 m_freem(m); 473 return; 474 } 475 #ifdef DIAGNOSTIC 476 if (m->m_pkthdr.rcvif != ifp) { 477 if_printf(ifp, "Warning, frame marked as received on %s\n", 478 m->m_pkthdr.rcvif->if_xname); 479 } 480 #endif 481 482 CURVNET_SET_QUIET(ifp->if_vnet); 483 484 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 485 if (ETHER_IS_BROADCAST(eh->ether_dhost)) 486 m->m_flags |= M_BCAST; 487 else 488 m->m_flags |= M_MCAST; 489 ifp->if_imcasts++; 490 } 491 492 #ifdef MAC 493 /* 494 * Tag the mbuf with an appropriate MAC label before any other 495 * consumers can get to it. 496 */ 497 mac_ifnet_create_mbuf(ifp, m); 498 #endif 499 500 /* 501 * Give bpf a chance at the packet. 502 */ 503 ETHER_BPF_MTAP(ifp, m); 504 505 /* 506 * If the CRC is still on the packet, trim it off. We do this once 507 * and once only in case we are re-entered. Nothing else on the 508 * Ethernet receive path expects to see the FCS. 509 */ 510 if (m->m_flags & M_HASFCS) { 511 m_adj(m, -ETHER_CRC_LEN); 512 m->m_flags &= ~M_HASFCS; 513 } 514 515 if (!(ifp->if_capenable & IFCAP_HWSTATS)) 516 ifp->if_ibytes += m->m_pkthdr.len; 517 518 /* Allow monitor mode to claim this frame, after stats are updated. */ 519 if (ifp->if_flags & IFF_MONITOR) { 520 m_freem(m); 521 CURVNET_RESTORE(); 522 return; 523 } 524 525 /* Handle input from a lagg(4) port */ 526 if (ifp->if_type == IFT_IEEE8023ADLAG) { 527 KASSERT(lagg_input_p != NULL, 528 ("%s: if_lagg not loaded!", __func__)); 529 m = (*lagg_input_p)(ifp, m); 530 if (m != NULL) 531 ifp = m->m_pkthdr.rcvif; 532 else { 533 CURVNET_RESTORE(); 534 return; 535 } 536 } 537 538 /* 539 * If the hardware did not process an 802.1Q tag, do this now, 540 * to allow 802.1P priority frames to be passed to the main input 541 * path correctly. 542 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels. 543 */ 544 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) { 545 struct ether_vlan_header *evl; 546 547 if (m->m_len < sizeof(*evl) && 548 (m = m_pullup(m, sizeof(*evl))) == NULL) { 549 #ifdef DIAGNOSTIC 550 if_printf(ifp, "cannot pullup VLAN header\n"); 551 #endif 552 ifp->if_ierrors++; 553 m_freem(m); 554 CURVNET_RESTORE(); 555 return; 556 } 557 558 evl = mtod(m, struct ether_vlan_header *); 559 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag); 560 m->m_flags |= M_VLANTAG; 561 562 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN, 563 ETHER_HDR_LEN - ETHER_TYPE_LEN); 564 m_adj(m, ETHER_VLAN_ENCAP_LEN); 565 eh = mtod(m, struct ether_header *); 566 } 567 568 M_SETFIB(m, ifp->if_fib); 569 570 /* Allow ng_ether(4) to claim this frame. */ 571 if (IFP2AC(ifp)->ac_netgraph != NULL) { 572 KASSERT(ng_ether_input_p != NULL, 573 ("%s: ng_ether_input_p is NULL", __func__)); 574 m->m_flags &= ~M_PROMISC; 575 (*ng_ether_input_p)(ifp, &m); 576 if (m == NULL) { 577 CURVNET_RESTORE(); 578 return; 579 } 580 eh = mtod(m, struct ether_header *); 581 } 582 583 /* 584 * Allow if_bridge(4) to claim this frame. 585 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it 586 * and the frame should be delivered locally. 587 */ 588 if (ifp->if_bridge != NULL) { 589 m->m_flags &= ~M_PROMISC; 590 BRIDGE_INPUT(ifp, m); 591 if (m == NULL) { 592 CURVNET_RESTORE(); 593 return; 594 } 595 eh = mtod(m, struct ether_header *); 596 } 597 598 #if defined(INET) || defined(INET6) 599 /* 600 * Clear M_PROMISC on frame so that carp(4) will see it when the 601 * mbuf flows up to Layer 3. 602 * FreeBSD's implementation of carp(4) uses the inprotosw 603 * to dispatch IPPROTO_CARP. carp(4) also allocates its own 604 * Ethernet addresses of the form 00:00:5e:00:01:xx, which 605 * is outside the scope of the M_PROMISC test below. 606 * TODO: Maintain a hash table of ethernet addresses other than 607 * ether_dhost which may be active on this ifp. 608 */ 609 if (ifp->if_carp && (*carp_forus_p)(ifp, eh->ether_dhost)) { 610 m->m_flags &= ~M_PROMISC; 611 } else 612 #endif 613 { 614 /* 615 * If the frame received was not for our MAC address, set the 616 * M_PROMISC flag on the mbuf chain. The frame may need to 617 * be seen by the rest of the Ethernet input path in case of 618 * re-entry (e.g. bridge, vlan, netgraph) but should not be 619 * seen by upper protocol layers. 620 */ 621 if (!ETHER_IS_MULTICAST(eh->ether_dhost) && 622 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0) 623 m->m_flags |= M_PROMISC; 624 } 625 626 if (harvest.ethernet) 627 random_harvest(&(m->m_data), 12, 2, RANDOM_NET_ETHER); 628 629 ether_demux(ifp, m); 630 CURVNET_RESTORE(); 631 } 632 633 /* 634 * Ethernet input dispatch; by default, direct dispatch here regardless of 635 * global configuration. 636 */ 637 static void 638 ether_nh_input(struct mbuf *m) 639 { 640 641 ether_input_internal(m->m_pkthdr.rcvif, m); 642 } 643 644 static struct netisr_handler ether_nh = { 645 .nh_name = "ether", 646 .nh_handler = ether_nh_input, 647 .nh_proto = NETISR_ETHER, 648 .nh_policy = NETISR_POLICY_SOURCE, 649 .nh_dispatch = NETISR_DISPATCH_DIRECT, 650 }; 651 652 static void 653 ether_init(__unused void *arg) 654 { 655 656 netisr_register(ðer_nh); 657 } 658 SYSINIT(ether, SI_SUB_INIT_IF, SI_ORDER_ANY, ether_init, NULL); 659 660 static void 661 vnet_ether_init(__unused void *arg) 662 { 663 int i; 664 665 /* Initialize packet filter hooks. */ 666 V_link_pfil_hook.ph_type = PFIL_TYPE_AF; 667 V_link_pfil_hook.ph_af = AF_LINK; 668 if ((i = pfil_head_register(&V_link_pfil_hook)) != 0) 669 printf("%s: WARNING: unable to register pfil link hook, " 670 "error %d\n", __func__, i); 671 } 672 VNET_SYSINIT(vnet_ether_init, SI_SUB_PROTO_IF, SI_ORDER_ANY, 673 vnet_ether_init, NULL); 674 675 static void 676 vnet_ether_destroy(__unused void *arg) 677 { 678 int i; 679 680 if ((i = pfil_head_unregister(&V_link_pfil_hook)) != 0) 681 printf("%s: WARNING: unable to unregister pfil link hook, " 682 "error %d\n", __func__, i); 683 } 684 VNET_SYSUNINIT(vnet_ether_uninit, SI_SUB_PROTO_IF, SI_ORDER_ANY, 685 vnet_ether_destroy, NULL); 686 687 688 689 static void 690 ether_input(struct ifnet *ifp, struct mbuf *m) 691 { 692 693 struct mbuf *mn; 694 695 /* 696 * The drivers are allowed to pass in a chain of packets linked with 697 * m_nextpkt. We split them up into separate packets here and pass 698 * them up. This allows the drivers to amortize the receive lock. 699 */ 700 while (m) { 701 mn = m->m_nextpkt; 702 m->m_nextpkt = NULL; 703 704 /* 705 * We will rely on rcvif being set properly in the deferred context, 706 * so assert it is correct here. 707 */ 708 KASSERT(m->m_pkthdr.rcvif == ifp, ("%s: ifnet mismatch", __func__)); 709 netisr_dispatch(NETISR_ETHER, m); 710 m = mn; 711 } 712 } 713 714 /* 715 * Upper layer processing for a received Ethernet packet. 716 */ 717 void 718 ether_demux(struct ifnet *ifp, struct mbuf *m) 719 { 720 struct ether_header *eh; 721 int i, isr; 722 u_short ether_type; 723 #if defined(NETATALK) 724 struct llc *l; 725 #endif 726 727 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__)); 728 729 /* Do not grab PROMISC frames in case we are re-entered. */ 730 if (PFIL_HOOKED(&V_link_pfil_hook) && !(m->m_flags & M_PROMISC)) { 731 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_IN, NULL); 732 733 if (i != 0 || m == NULL) 734 return; 735 } 736 737 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_clrprotoflags(m); 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 INET6 797 case ETHERTYPE_IPV6: 798 isr = NETISR_IPV6; 799 break; 800 #endif 801 #ifdef NETATALK 802 case ETHERTYPE_AT: 803 isr = NETISR_ATALK1; 804 break; 805 case ETHERTYPE_AARP: 806 isr = NETISR_AARP; 807 break; 808 #endif /* NETATALK */ 809 default: 810 #if defined(NETATALK) 811 if (ether_type > ETHERMTU) 812 goto discard; 813 l = mtod(m, struct llc *); 814 if (l->llc_dsap == LLC_SNAP_LSAP && 815 l->llc_ssap == LLC_SNAP_LSAP && 816 l->llc_control == LLC_UI) { 817 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code, 818 sizeof(at_org_code)) == 0 && 819 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) { 820 m_adj(m, LLC_SNAPFRAMELEN); 821 isr = NETISR_ATALK2; 822 break; 823 } 824 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code, 825 sizeof(aarp_org_code)) == 0 && 826 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) { 827 m_adj(m, LLC_SNAPFRAMELEN); 828 isr = NETISR_AARP; 829 break; 830 } 831 } 832 #endif /* NETATALK */ 833 goto discard; 834 } 835 netisr_dispatch(isr, m); 836 return; 837 838 discard: 839 /* 840 * Packet is to be discarded. If netgraph is present, 841 * hand the packet to it for last chance processing; 842 * otherwise dispose of it. 843 */ 844 if (IFP2AC(ifp)->ac_netgraph != NULL) { 845 KASSERT(ng_ether_input_orphan_p != NULL, 846 ("ng_ether_input_orphan_p is NULL")); 847 /* 848 * Put back the ethernet header so netgraph has a 849 * consistent view of inbound packets. 850 */ 851 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT); 852 (*ng_ether_input_orphan_p)(ifp, m); 853 return; 854 } 855 m_freem(m); 856 } 857 858 /* 859 * Convert Ethernet address to printable (loggable) representation. 860 * This routine is for compatibility; it's better to just use 861 * 862 * printf("%6D", <pointer to address>, ":"); 863 * 864 * since there's no static buffer involved. 865 */ 866 char * 867 ether_sprintf(const u_char *ap) 868 { 869 static char etherbuf[18]; 870 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":"); 871 return (etherbuf); 872 } 873 874 /* 875 * Perform common duties while attaching to interface list 876 */ 877 void 878 ether_ifattach(struct ifnet *ifp, const u_int8_t *lla) 879 { 880 int i; 881 struct ifaddr *ifa; 882 struct sockaddr_dl *sdl; 883 884 ifp->if_addrlen = ETHER_ADDR_LEN; 885 ifp->if_hdrlen = ETHER_HDR_LEN; 886 if_attach(ifp); 887 ifp->if_mtu = ETHERMTU; 888 ifp->if_output = ether_output; 889 ifp->if_input = ether_input; 890 ifp->if_resolvemulti = ether_resolvemulti; 891 #ifdef VIMAGE 892 ifp->if_reassign = ether_reassign; 893 #endif 894 if (ifp->if_baudrate == 0) 895 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 896 ifp->if_broadcastaddr = etherbroadcastaddr; 897 898 ifa = ifp->if_addr; 899 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__)); 900 sdl = (struct sockaddr_dl *)ifa->ifa_addr; 901 sdl->sdl_type = IFT_ETHER; 902 sdl->sdl_alen = ifp->if_addrlen; 903 bcopy(lla, LLADDR(sdl), ifp->if_addrlen); 904 905 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN); 906 if (ng_ether_attach_p != NULL) 907 (*ng_ether_attach_p)(ifp); 908 909 /* Announce Ethernet MAC address if non-zero. */ 910 for (i = 0; i < ifp->if_addrlen; i++) 911 if (lla[i] != 0) 912 break; 913 if (i != ifp->if_addrlen) 914 if_printf(ifp, "Ethernet address: %6D\n", lla, ":"); 915 916 uuid_ether_add(LLADDR(sdl)); 917 } 918 919 /* 920 * Perform common duties while detaching an Ethernet interface 921 */ 922 void 923 ether_ifdetach(struct ifnet *ifp) 924 { 925 struct sockaddr_dl *sdl; 926 927 sdl = (struct sockaddr_dl *)(ifp->if_addr->ifa_addr); 928 uuid_ether_del(LLADDR(sdl)); 929 930 if (IFP2AC(ifp)->ac_netgraph != NULL) { 931 KASSERT(ng_ether_detach_p != NULL, 932 ("ng_ether_detach_p is NULL")); 933 (*ng_ether_detach_p)(ifp); 934 } 935 936 bpfdetach(ifp); 937 if_detach(ifp); 938 } 939 940 #ifdef VIMAGE 941 void 942 ether_reassign(struct ifnet *ifp, struct vnet *new_vnet, char *unused __unused) 943 { 944 945 if (IFP2AC(ifp)->ac_netgraph != NULL) { 946 KASSERT(ng_ether_detach_p != NULL, 947 ("ng_ether_detach_p is NULL")); 948 (*ng_ether_detach_p)(ifp); 949 } 950 951 if (ng_ether_attach_p != NULL) { 952 CURVNET_SET_QUIET(new_vnet); 953 (*ng_ether_attach_p)(ifp); 954 CURVNET_RESTORE(); 955 } 956 } 957 #endif 958 959 SYSCTL_DECL(_net_link); 960 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet"); 961 962 #if 0 963 /* 964 * This is for reference. We have a table-driven version 965 * of the little-endian crc32 generator, which is faster 966 * than the double-loop. 967 */ 968 uint32_t 969 ether_crc32_le(const uint8_t *buf, size_t len) 970 { 971 size_t i; 972 uint32_t crc; 973 int bit; 974 uint8_t data; 975 976 crc = 0xffffffff; /* initial value */ 977 978 for (i = 0; i < len; i++) { 979 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) { 980 carry = (crc ^ data) & 1; 981 crc >>= 1; 982 if (carry) 983 crc = (crc ^ ETHER_CRC_POLY_LE); 984 } 985 } 986 987 return (crc); 988 } 989 #else 990 uint32_t 991 ether_crc32_le(const uint8_t *buf, size_t len) 992 { 993 static const uint32_t crctab[] = { 994 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 995 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 996 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 997 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 998 }; 999 size_t i; 1000 uint32_t crc; 1001 1002 crc = 0xffffffff; /* initial value */ 1003 1004 for (i = 0; i < len; i++) { 1005 crc ^= buf[i]; 1006 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1007 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1008 } 1009 1010 return (crc); 1011 } 1012 #endif 1013 1014 uint32_t 1015 ether_crc32_be(const uint8_t *buf, size_t len) 1016 { 1017 size_t i; 1018 uint32_t crc, carry; 1019 int bit; 1020 uint8_t data; 1021 1022 crc = 0xffffffff; /* initial value */ 1023 1024 for (i = 0; i < len; i++) { 1025 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) { 1026 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01); 1027 crc <<= 1; 1028 if (carry) 1029 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1030 } 1031 } 1032 1033 return (crc); 1034 } 1035 1036 int 1037 ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data) 1038 { 1039 struct ifaddr *ifa = (struct ifaddr *) data; 1040 struct ifreq *ifr = (struct ifreq *) data; 1041 int error = 0; 1042 1043 switch (command) { 1044 case SIOCSIFADDR: 1045 ifp->if_flags |= IFF_UP; 1046 1047 switch (ifa->ifa_addr->sa_family) { 1048 #ifdef INET 1049 case AF_INET: 1050 ifp->if_init(ifp->if_softc); /* before arpwhohas */ 1051 arp_ifinit(ifp, ifa); 1052 break; 1053 #endif 1054 default: 1055 ifp->if_init(ifp->if_softc); 1056 break; 1057 } 1058 break; 1059 1060 case SIOCGIFADDR: 1061 { 1062 struct sockaddr *sa; 1063 1064 sa = (struct sockaddr *) & ifr->ifr_data; 1065 bcopy(IF_LLADDR(ifp), 1066 (caddr_t) sa->sa_data, ETHER_ADDR_LEN); 1067 } 1068 break; 1069 1070 case SIOCSIFMTU: 1071 /* 1072 * Set the interface MTU. 1073 */ 1074 if (ifr->ifr_mtu > ETHERMTU) { 1075 error = EINVAL; 1076 } else { 1077 ifp->if_mtu = ifr->ifr_mtu; 1078 } 1079 break; 1080 default: 1081 error = EINVAL; /* XXX netbsd has ENOTTY??? */ 1082 break; 1083 } 1084 return (error); 1085 } 1086 1087 static int 1088 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa, 1089 struct sockaddr *sa) 1090 { 1091 struct sockaddr_dl *sdl; 1092 #ifdef INET 1093 struct sockaddr_in *sin; 1094 #endif 1095 #ifdef INET6 1096 struct sockaddr_in6 *sin6; 1097 #endif 1098 u_char *e_addr; 1099 1100 switch(sa->sa_family) { 1101 case AF_LINK: 1102 /* 1103 * No mapping needed. Just check that it's a valid MC address. 1104 */ 1105 sdl = (struct sockaddr_dl *)sa; 1106 e_addr = LLADDR(sdl); 1107 if (!ETHER_IS_MULTICAST(e_addr)) 1108 return EADDRNOTAVAIL; 1109 *llsa = 0; 1110 return 0; 1111 1112 #ifdef INET 1113 case AF_INET: 1114 sin = (struct sockaddr_in *)sa; 1115 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) 1116 return EADDRNOTAVAIL; 1117 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER); 1118 sdl->sdl_alen = ETHER_ADDR_LEN; 1119 e_addr = LLADDR(sdl); 1120 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr); 1121 *llsa = (struct sockaddr *)sdl; 1122 return 0; 1123 #endif 1124 #ifdef INET6 1125 case AF_INET6: 1126 sin6 = (struct sockaddr_in6 *)sa; 1127 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1128 /* 1129 * An IP6 address of 0 means listen to all 1130 * of the Ethernet multicast address used for IP6. 1131 * (This is used for multicast routers.) 1132 */ 1133 ifp->if_flags |= IFF_ALLMULTI; 1134 *llsa = 0; 1135 return 0; 1136 } 1137 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) 1138 return EADDRNOTAVAIL; 1139 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER); 1140 sdl->sdl_alen = ETHER_ADDR_LEN; 1141 e_addr = LLADDR(sdl); 1142 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr); 1143 *llsa = (struct sockaddr *)sdl; 1144 return 0; 1145 #endif 1146 1147 default: 1148 /* 1149 * Well, the text isn't quite right, but it's the name 1150 * that counts... 1151 */ 1152 return EAFNOSUPPORT; 1153 } 1154 } 1155 1156 static void* 1157 ether_alloc(u_char type, struct ifnet *ifp) 1158 { 1159 struct arpcom *ac; 1160 1161 ac = malloc(sizeof(struct arpcom), M_ARPCOM, M_WAITOK | M_ZERO); 1162 ac->ac_ifp = ifp; 1163 1164 return (ac); 1165 } 1166 1167 static void 1168 ether_free(void *com, u_char type) 1169 { 1170 1171 free(com, M_ARPCOM); 1172 } 1173 1174 static int 1175 ether_modevent(module_t mod, int type, void *data) 1176 { 1177 1178 switch (type) { 1179 case MOD_LOAD: 1180 if_register_com_alloc(IFT_ETHER, ether_alloc, ether_free); 1181 break; 1182 case MOD_UNLOAD: 1183 if_deregister_com_alloc(IFT_ETHER); 1184 break; 1185 default: 1186 return EOPNOTSUPP; 1187 } 1188 1189 return (0); 1190 } 1191 1192 static moduledata_t ether_mod = { 1193 "ether", 1194 ether_modevent, 1195 0 1196 }; 1197 1198 void 1199 ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen) 1200 { 1201 struct ether_vlan_header vlan; 1202 struct mbuf mv, mb; 1203 1204 KASSERT((m->m_flags & M_VLANTAG) != 0, 1205 ("%s: vlan information not present", __func__)); 1206 KASSERT(m->m_len >= sizeof(struct ether_header), 1207 ("%s: mbuf not large enough for header", __func__)); 1208 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header)); 1209 vlan.evl_proto = vlan.evl_encap_proto; 1210 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN); 1211 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag); 1212 m->m_len -= sizeof(struct ether_header); 1213 m->m_data += sizeof(struct ether_header); 1214 /* 1215 * If a data link has been supplied by the caller, then we will need to 1216 * re-create a stack allocated mbuf chain with the following structure: 1217 * 1218 * (1) mbuf #1 will contain the supplied data link 1219 * (2) mbuf #2 will contain the vlan header 1220 * (3) mbuf #3 will contain the original mbuf's packet data 1221 * 1222 * Otherwise, submit the packet and vlan header via bpf_mtap2(). 1223 */ 1224 if (data != NULL) { 1225 mv.m_next = m; 1226 mv.m_data = (caddr_t)&vlan; 1227 mv.m_len = sizeof(vlan); 1228 mb.m_next = &mv; 1229 mb.m_data = data; 1230 mb.m_len = dlen; 1231 bpf_mtap(bp, &mb); 1232 } else 1233 bpf_mtap2(bp, &vlan, sizeof(vlan), m); 1234 m->m_len += sizeof(struct ether_header); 1235 m->m_data -= sizeof(struct ether_header); 1236 } 1237 1238 struct mbuf * 1239 ether_vlanencap(struct mbuf *m, uint16_t tag) 1240 { 1241 struct ether_vlan_header *evl; 1242 1243 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_NOWAIT); 1244 if (m == NULL) 1245 return (NULL); 1246 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */ 1247 1248 if (m->m_len < sizeof(*evl)) { 1249 m = m_pullup(m, sizeof(*evl)); 1250 if (m == NULL) 1251 return (NULL); 1252 } 1253 1254 /* 1255 * Transform the Ethernet header into an Ethernet header 1256 * with 802.1Q encapsulation. 1257 */ 1258 evl = mtod(m, struct ether_vlan_header *); 1259 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN, 1260 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN); 1261 evl->evl_encap_proto = htons(ETHERTYPE_VLAN); 1262 evl->evl_tag = htons(tag); 1263 return (m); 1264 } 1265 1266 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY); 1267 MODULE_VERSION(ether, 1); 1268