1 /* $NetBSD: if_ethersubr.c,v 1.250 2017/12/09 10:51:30 maxv Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /* 33 * Copyright (c) 1982, 1989, 1993 34 * The Regents of the University of California. All rights reserved. 35 * 36 * Redistribution and use in source and binary forms, with or without 37 * modification, are permitted provided that the following conditions 38 * are met: 39 * 1. Redistributions of source code must retain the above copyright 40 * notice, this list of conditions and the following disclaimer. 41 * 2. Redistributions in binary form must reproduce the above copyright 42 * notice, this list of conditions and the following disclaimer in the 43 * documentation and/or other materials provided with the distribution. 44 * 3. Neither the name of the University nor the names of its contributors 45 * may be used to endorse or promote products derived from this software 46 * without specific prior written permission. 47 * 48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 58 * SUCH DAMAGE. 59 * 60 * @(#)if_ethersubr.c 8.2 (Berkeley) 4/4/96 61 */ 62 63 #include <sys/cdefs.h> 64 __KERNEL_RCSID(0, "$NetBSD: if_ethersubr.c,v 1.250 2017/12/09 10:51:30 maxv Exp $"); 65 66 #ifdef _KERNEL_OPT 67 #include "opt_inet.h" 68 #include "opt_atalk.h" 69 #include "opt_mbuftrace.h" 70 #include "opt_mpls.h" 71 #include "opt_gateway.h" 72 #include "opt_pppoe.h" 73 #include "opt_net_mpsafe.h" 74 #endif 75 76 #include "vlan.h" 77 #include "pppoe.h" 78 #include "bridge.h" 79 #include "arp.h" 80 #include "agr.h" 81 82 #include <sys/sysctl.h> 83 #include <sys/mbuf.h> 84 #include <sys/mutex.h> 85 #include <sys/ioctl.h> 86 #include <sys/errno.h> 87 #include <sys/device.h> 88 #include <sys/rnd.h> 89 #include <sys/rndsource.h> 90 #include <sys/cpu.h> 91 #include <sys/kmem.h> 92 93 #include <net/if.h> 94 #include <net/netisr.h> 95 #include <net/route.h> 96 #include <net/if_llc.h> 97 #include <net/if_dl.h> 98 #include <net/if_types.h> 99 #include <net/pktqueue.h> 100 101 #include <net/if_media.h> 102 #include <dev/mii/mii.h> 103 #include <dev/mii/miivar.h> 104 105 #if NARP == 0 106 /* 107 * XXX there should really be a way to issue this warning from within config(8) 108 */ 109 #error You have included NETATALK or a pseudo-device in your configuration that depends on the presence of ethernet interfaces, but have no such interfaces configured. Check if you really need pseudo-device bridge, pppoe, vlan or options NETATALK. 110 #endif 111 112 #include <net/bpf.h> 113 114 #include <net/if_ether.h> 115 #include <net/if_vlanvar.h> 116 117 #if NPPPOE > 0 118 #include <net/if_pppoe.h> 119 #endif 120 121 #if NAGR > 0 122 #include <net/agr/ieee8023_slowprotocols.h> /* XXX */ 123 #include <net/agr/ieee8023ad.h> 124 #include <net/agr/if_agrvar.h> 125 #endif 126 127 #if NBRIDGE > 0 128 #include <net/if_bridgevar.h> 129 #endif 130 131 #include <netinet/in.h> 132 #ifdef INET 133 #include <netinet/in_var.h> 134 #endif 135 #include <netinet/if_inarp.h> 136 137 #ifdef INET6 138 #ifndef INET 139 #include <netinet/in.h> 140 #endif 141 #include <netinet6/in6_var.h> 142 #include <netinet6/nd6.h> 143 #endif 144 145 146 #include "carp.h" 147 #if NCARP > 0 148 #include <netinet/ip_carp.h> 149 #endif 150 151 #ifdef NETATALK 152 #include <netatalk/at.h> 153 #include <netatalk/at_var.h> 154 #include <netatalk/at_extern.h> 155 156 #define llc_snap_org_code llc_un.type_snap.org_code 157 #define llc_snap_ether_type llc_un.type_snap.ether_type 158 159 extern u_char at_org_code[3]; 160 extern u_char aarp_org_code[3]; 161 #endif /* NETATALK */ 162 163 #ifdef MPLS 164 #include <netmpls/mpls.h> 165 #include <netmpls/mpls_var.h> 166 #endif 167 168 static struct timeval bigpktppslim_last; 169 static int bigpktppslim = 2; /* XXX */ 170 static int bigpktpps_count; 171 static kmutex_t bigpktpps_lock __cacheline_aligned; 172 173 174 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = 175 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 176 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] = 177 { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 }; 178 #define senderr(e) { error = (e); goto bad;} 179 180 static int ether_output(struct ifnet *, struct mbuf *, 181 const struct sockaddr *, const struct rtentry *); 182 183 /* 184 * Ethernet output routine. 185 * Encapsulate a packet of type family for the local net. 186 * Assumes that ifp is actually pointer to ethercom structure. 187 */ 188 static int 189 ether_output(struct ifnet * const ifp0, struct mbuf * const m0, 190 const struct sockaddr * const dst, 191 const struct rtentry *rt) 192 { 193 uint16_t etype = 0; 194 int error = 0, hdrcmplt = 0; 195 uint8_t esrc[6], edst[6]; 196 struct mbuf *m = m0; 197 struct mbuf *mcopy = NULL; 198 struct ether_header *eh; 199 struct ifnet *ifp = ifp0; 200 #ifdef INET 201 struct arphdr *ah; 202 #endif /* INET */ 203 #ifdef NETATALK 204 struct at_ifaddr *aa; 205 #endif /* NETATALK */ 206 207 #ifdef MBUFTRACE 208 m_claimm(m, ifp->if_mowner); 209 #endif 210 211 #if NCARP > 0 212 if (ifp->if_type == IFT_CARP) { 213 struct ifaddr *ifa; 214 int s = pserialize_read_enter(); 215 216 /* loop back if this is going to the carp interface */ 217 if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP && 218 (ifa = ifa_ifwithaddr(dst)) != NULL) { 219 if (ifa->ifa_ifp == ifp0) { 220 pserialize_read_exit(s); 221 return looutput(ifp0, m, dst, rt); 222 } 223 } 224 pserialize_read_exit(s); 225 226 ifp = ifp->if_carpdev; 227 /* ac = (struct arpcom *)ifp; */ 228 229 if ((ifp0->if_flags & (IFF_UP | IFF_RUNNING)) != 230 (IFF_UP | IFF_RUNNING)) 231 senderr(ENETDOWN); 232 } 233 #endif /* NCARP > 0 */ 234 235 if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING)) 236 senderr(ENETDOWN); 237 238 switch (dst->sa_family) { 239 240 #ifdef INET 241 case AF_INET: 242 if (m->m_flags & M_BCAST) 243 (void)memcpy(edst, etherbroadcastaddr, sizeof(edst)); 244 else if (m->m_flags & M_MCAST) 245 ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst); 246 else if ((error = arpresolve(ifp, rt, m, dst, edst, 247 sizeof(edst))) != 0) { 248 return error == EWOULDBLOCK ? 0 : error; 249 } 250 /* If broadcasting on a simplex interface, loopback a copy */ 251 if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX)) 252 mcopy = m_copy(m, 0, (int)M_COPYALL); 253 etype = htons(ETHERTYPE_IP); 254 break; 255 256 case AF_ARP: 257 ah = mtod(m, struct arphdr *); 258 if (m->m_flags & M_BCAST) 259 (void)memcpy(edst, etherbroadcastaddr, sizeof(edst)); 260 else { 261 void *tha = ar_tha(ah); 262 263 if (tha == NULL) { 264 /* fake with ARPHDR_IEEE1394 */ 265 m_freem(m); 266 return 0; 267 } 268 memcpy(edst, tha, sizeof(edst)); 269 } 270 271 ah->ar_hrd = htons(ARPHRD_ETHER); 272 273 switch (ntohs(ah->ar_op)) { 274 case ARPOP_REVREQUEST: 275 case ARPOP_REVREPLY: 276 etype = htons(ETHERTYPE_REVARP); 277 break; 278 279 case ARPOP_REQUEST: 280 case ARPOP_REPLY: 281 default: 282 etype = htons(ETHERTYPE_ARP); 283 } 284 285 break; 286 #endif 287 #ifdef INET6 288 case AF_INET6: 289 if (m->m_flags & M_BCAST) 290 (void)memcpy(edst, etherbroadcastaddr, sizeof(edst)); 291 else if (m->m_flags & M_MCAST) { 292 ETHER_MAP_IPV6_MULTICAST(&satocsin6(dst)->sin6_addr, 293 edst); 294 } else { 295 error = nd6_resolve(ifp, rt, m, dst, edst, 296 sizeof(edst)); 297 if (error != 0) 298 return error == EWOULDBLOCK ? 0 : error; 299 } 300 etype = htons(ETHERTYPE_IPV6); 301 break; 302 #endif 303 #ifdef NETATALK 304 case AF_APPLETALK: { 305 struct ifaddr *ifa; 306 int s; 307 308 KERNEL_LOCK(1, NULL); 309 if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) { 310 #ifdef NETATALKDEBUG 311 printf("aarpresolv failed\n"); 312 #endif /* NETATALKDEBUG */ 313 KERNEL_UNLOCK_ONE(NULL); 314 return (0); 315 } 316 /* 317 * ifaddr is the first thing in at_ifaddr 318 */ 319 s = pserialize_read_enter(); 320 ifa = at_ifawithnet((const struct sockaddr_at *)dst, ifp); 321 if (ifa == NULL) { 322 pserialize_read_exit(s); 323 KERNEL_UNLOCK_ONE(NULL); 324 goto bad; 325 } 326 aa = (struct at_ifaddr *)ifa; 327 328 /* 329 * In the phase 2 case, we need to prepend an mbuf for the 330 * llc header. Since we must preserve the value of m, 331 * which is passed to us by value, we m_copy() the first 332 * mbuf, and use it for our llc header. 333 */ 334 if (aa->aa_flags & AFA_PHASE2) { 335 struct llc llc; 336 337 M_PREPEND(m, sizeof(struct llc), M_DONTWAIT); 338 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 339 llc.llc_control = LLC_UI; 340 memcpy(llc.llc_snap_org_code, at_org_code, 341 sizeof(llc.llc_snap_org_code)); 342 llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK); 343 memcpy(mtod(m, void *), &llc, sizeof(struct llc)); 344 } else { 345 etype = htons(ETHERTYPE_ATALK); 346 } 347 pserialize_read_exit(s); 348 KERNEL_UNLOCK_ONE(NULL); 349 break; 350 } 351 #endif /* NETATALK */ 352 case pseudo_AF_HDRCMPLT: 353 hdrcmplt = 1; 354 memcpy(esrc, 355 ((const struct ether_header *)dst->sa_data)->ether_shost, 356 sizeof(esrc)); 357 /* FALLTHROUGH */ 358 359 case AF_UNSPEC: 360 memcpy(edst, 361 ((const struct ether_header *)dst->sa_data)->ether_dhost, 362 sizeof(edst)); 363 /* AF_UNSPEC doesn't swap the byte order of the ether_type. */ 364 etype = ((const struct ether_header *)dst->sa_data)->ether_type; 365 break; 366 367 default: 368 printf("%s: can't handle af%d\n", ifp->if_xname, 369 dst->sa_family); 370 senderr(EAFNOSUPPORT); 371 } 372 373 #ifdef MPLS 374 KERNEL_LOCK(1, NULL); 375 { 376 struct m_tag *mtag; 377 mtag = m_tag_find(m, PACKET_TAG_MPLS, NULL); 378 if (mtag != NULL) { 379 /* Having the tag itself indicates it's MPLS */ 380 etype = htons(ETHERTYPE_MPLS); 381 m_tag_delete(m, mtag); 382 } 383 } 384 KERNEL_UNLOCK_ONE(NULL); 385 #endif 386 387 if (mcopy) 388 (void)looutput(ifp, mcopy, dst, rt); 389 390 /* If no ether type is set, this must be a 802.2 formatted packet. 391 */ 392 if (etype == 0) 393 etype = htons(m->m_pkthdr.len); 394 /* 395 * Add local net header. If no space in first mbuf, 396 * allocate another. 397 */ 398 M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT); 399 if (m == 0) 400 senderr(ENOBUFS); 401 eh = mtod(m, struct ether_header *); 402 /* Note: etype is already in network byte order. */ 403 (void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type)); 404 memcpy(eh->ether_dhost, edst, sizeof(edst)); 405 if (hdrcmplt) 406 memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost)); 407 else 408 memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl), 409 sizeof(eh->ether_shost)); 410 411 #if NCARP > 0 412 if (ifp0 != ifp && ifp0->if_type == IFT_CARP) { 413 memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl), 414 sizeof(eh->ether_shost)); 415 } 416 #endif /* NCARP > 0 */ 417 418 if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0) 419 return (error); 420 if (m == NULL) 421 return (0); 422 423 #if NBRIDGE > 0 424 /* 425 * Bridges require special output handling. 426 */ 427 if (ifp->if_bridge) 428 return (bridge_output(ifp, m, NULL, NULL)); 429 #endif 430 431 #if NCARP > 0 432 if (ifp != ifp0) 433 ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN; 434 #endif /* NCARP > 0 */ 435 436 #ifdef ALTQ 437 KERNEL_LOCK(1, NULL); 438 /* 439 * If ALTQ is enabled on the parent interface, do 440 * classification; the queueing discipline might not 441 * require classification, but might require the 442 * address family/header pointer in the pktattr. 443 */ 444 if (ALTQ_IS_ENABLED(&ifp->if_snd)) 445 altq_etherclassify(&ifp->if_snd, m); 446 KERNEL_UNLOCK_ONE(NULL); 447 #endif 448 return ifq_enqueue(ifp, m); 449 450 bad: 451 if (m) 452 m_freem(m); 453 return (error); 454 } 455 456 #ifdef ALTQ 457 /* 458 * This routine is a slight hack to allow a packet to be classified 459 * if the Ethernet headers are present. It will go away when ALTQ's 460 * classification engine understands link headers. 461 */ 462 void 463 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m) 464 { 465 struct ether_header *eh; 466 uint16_t ether_type; 467 int hlen, af, hdrsize; 468 void *hdr; 469 470 hlen = ETHER_HDR_LEN; 471 eh = mtod(m, struct ether_header *); 472 473 ether_type = htons(eh->ether_type); 474 475 if (ether_type < ETHERMTU) { 476 /* LLC/SNAP */ 477 struct llc *llc = (struct llc *)(eh + 1); 478 hlen += 8; 479 480 if (m->m_len < hlen || 481 llc->llc_dsap != LLC_SNAP_LSAP || 482 llc->llc_ssap != LLC_SNAP_LSAP || 483 llc->llc_control != LLC_UI) { 484 /* Not SNAP. */ 485 goto bad; 486 } 487 488 ether_type = htons(llc->llc_un.type_snap.ether_type); 489 } 490 491 switch (ether_type) { 492 case ETHERTYPE_IP: 493 af = AF_INET; 494 hdrsize = 20; /* sizeof(struct ip) */ 495 break; 496 497 case ETHERTYPE_IPV6: 498 af = AF_INET6; 499 hdrsize = 40; /* sizeof(struct ip6_hdr) */ 500 break; 501 502 default: 503 af = AF_UNSPEC; 504 hdrsize = 0; 505 break; 506 } 507 508 while (m->m_len <= hlen) { 509 hlen -= m->m_len; 510 m = m->m_next; 511 } 512 if (m->m_len < (hlen + hdrsize)) { 513 /* 514 * protocol header not in a single mbuf. 515 * We can't cope with this situation right 516 * now (but it shouldn't ever happen, really, anyhow). 517 */ 518 #ifdef DEBUG 519 printf("altq_etherclassify: headers span multiple mbufs: " 520 "%d < %d\n", m->m_len, (hlen + hdrsize)); 521 #endif 522 goto bad; 523 } 524 525 m->m_data += hlen; 526 m->m_len -= hlen; 527 528 hdr = mtod(m, void *); 529 530 if (ALTQ_NEEDS_CLASSIFY(ifq)) 531 m->m_pkthdr.pattr_class = 532 (*ifq->altq_classify)(ifq->altq_clfier, m, af); 533 m->m_pkthdr.pattr_af = af; 534 m->m_pkthdr.pattr_hdr = hdr; 535 536 m->m_data -= hlen; 537 m->m_len += hlen; 538 539 return; 540 541 bad: 542 m->m_pkthdr.pattr_class = NULL; 543 m->m_pkthdr.pattr_hdr = NULL; 544 m->m_pkthdr.pattr_af = AF_UNSPEC; 545 } 546 #endif /* ALTQ */ 547 548 /* 549 * Process a received Ethernet packet; 550 * the packet is in the mbuf chain m with 551 * the ether header. 552 */ 553 void 554 ether_input(struct ifnet *ifp, struct mbuf *m) 555 { 556 struct ethercom *ec = (struct ethercom *) ifp; 557 pktqueue_t *pktq = NULL; 558 struct ifqueue *inq = NULL; 559 uint16_t etype; 560 struct ether_header *eh; 561 size_t ehlen; 562 static int earlypkts; 563 int isr = 0; 564 #if defined (LLC) || defined(NETATALK) 565 struct llc *l; 566 #endif 567 568 KASSERT(!cpu_intr_p()); 569 570 if ((ifp->if_flags & IFF_UP) == 0) { 571 m_freem(m); 572 return; 573 } 574 575 #ifdef MBUFTRACE 576 m_claimm(m, &ec->ec_rx_mowner); 577 #endif 578 eh = mtod(m, struct ether_header *); 579 etype = ntohs(eh->ether_type); 580 ehlen = sizeof(*eh); 581 582 if (__predict_false(earlypkts < 100 || !rnd_initial_entropy)) { 583 rnd_add_data(NULL, eh, ehlen, 0); 584 earlypkts++; 585 } 586 587 /* 588 * Determine if the packet is within its size limits. 589 */ 590 if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len > 591 ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) { 592 mutex_enter(&bigpktpps_lock); 593 if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count, 594 bigpktppslim)) { 595 printf("%s: discarding oversize frame (len=%d)\n", 596 ifp->if_xname, m->m_pkthdr.len); 597 } 598 mutex_exit(&bigpktpps_lock); 599 m_freem(m); 600 return; 601 } 602 603 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 604 /* 605 * If this is not a simplex interface, drop the packet 606 * if it came from us. 607 */ 608 if ((ifp->if_flags & IFF_SIMPLEX) == 0 && 609 memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost, 610 ETHER_ADDR_LEN) == 0) { 611 m_freem(m); 612 return; 613 } 614 615 if (memcmp(etherbroadcastaddr, 616 eh->ether_dhost, ETHER_ADDR_LEN) == 0) 617 m->m_flags |= M_BCAST; 618 else 619 m->m_flags |= M_MCAST; 620 ifp->if_imcasts++; 621 } 622 623 /* If the CRC is still on the packet, trim it off. */ 624 if (m->m_flags & M_HASFCS) { 625 m_adj(m, -ETHER_CRC_LEN); 626 m->m_flags &= ~M_HASFCS; 627 } 628 629 ifp->if_ibytes += m->m_pkthdr.len; 630 631 #if NCARP > 0 632 if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) { 633 /* 634 * clear M_PROMISC, in case the packets comes from a 635 * vlan 636 */ 637 m->m_flags &= ~M_PROMISC; 638 if (carp_input(m, (uint8_t *)&eh->ether_shost, 639 (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0) 640 return; 641 } 642 #endif /* NCARP > 0 */ 643 644 if ((m->m_flags & (M_BCAST | M_MCAST | M_PROMISC)) == 0 && 645 (ifp->if_flags & IFF_PROMISC) != 0 && 646 memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost, 647 ETHER_ADDR_LEN) != 0) { 648 m->m_flags |= M_PROMISC; 649 } 650 651 if ((m->m_flags & M_PROMISC) == 0) { 652 if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0) 653 return; 654 if (m == NULL) 655 return; 656 657 eh = mtod(m, struct ether_header *); 658 etype = ntohs(eh->ether_type); 659 ehlen = sizeof(*eh); 660 } 661 662 #if NAGR > 0 663 if (ifp->if_agrprivate && 664 __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) { 665 m->m_flags &= ~M_PROMISC; 666 agr_input(ifp, m); 667 return; 668 } 669 #endif /* NAGR > 0 */ 670 671 /* 672 * If VLANs are configured on the interface, check to 673 * see if the device performed the decapsulation and 674 * provided us with the tag. 675 */ 676 if (ec->ec_nvlans && vlan_has_tag(m)) { 677 #if NVLAN > 0 678 /* 679 * vlan_input() will either recursively call ether_input() 680 * or drop the packet. 681 */ 682 vlan_input(ifp, m); 683 #else 684 m_freem(m); 685 #endif 686 return; 687 } 688 689 /* 690 * Handle protocols that expect to have the Ethernet header 691 * (and possibly FCS) intact. 692 */ 693 switch (etype) { 694 case ETHERTYPE_VLAN: { 695 struct ether_vlan_header *evl = (void *)eh; 696 /* 697 * If there is a tag of 0, then the VLAN header was probably 698 * just being used to store the priority. Extract the ether 699 * type, and if IP or IPV6, let them deal with it. 700 */ 701 if (m->m_len <= sizeof(*evl) 702 && EVL_VLANOFTAG(evl->evl_tag) == 0) { 703 etype = ntohs(evl->evl_proto); 704 ehlen = sizeof(*evl); 705 if ((m->m_flags & M_PROMISC) == 0 706 && (etype == ETHERTYPE_IP 707 || etype == ETHERTYPE_IPV6)) 708 break; 709 } 710 #if NVLAN > 0 711 /* 712 * vlan_input() will either recursively call ether_input() 713 * or drop the packet. 714 */ 715 if (((struct ethercom *)ifp)->ec_nvlans != 0) 716 vlan_input(ifp, m); 717 else 718 #endif /* NVLAN > 0 */ 719 m_freem(m); 720 return; 721 } 722 #if NPPPOE > 0 723 case ETHERTYPE_PPPOEDISC: 724 pppoedisc_input(ifp, m); 725 return; 726 case ETHERTYPE_PPPOE: 727 pppoe_input(ifp, m); 728 return; 729 #endif /* NPPPOE > 0 */ 730 case ETHERTYPE_SLOWPROTOCOLS: { 731 uint8_t subtype; 732 733 #if defined(DIAGNOSTIC) 734 if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) { 735 panic("ether_input: too short slow protocol packet"); 736 } 737 #endif 738 m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype); 739 switch (subtype) { 740 #if NAGR > 0 741 case SLOWPROTOCOLS_SUBTYPE_LACP: 742 if (ifp->if_agrprivate) { 743 ieee8023ad_lacp_input(ifp, m); 744 return; 745 } 746 break; 747 748 case SLOWPROTOCOLS_SUBTYPE_MARKER: 749 if (ifp->if_agrprivate) { 750 ieee8023ad_marker_input(ifp, m); 751 return; 752 } 753 break; 754 #endif /* NAGR > 0 */ 755 default: 756 if (subtype == 0 || subtype > 10) { 757 /* illegal value */ 758 m_freem(m); 759 return; 760 } 761 /* unknown subtype */ 762 break; 763 } 764 /* FALLTHROUGH */ 765 } 766 default: 767 if (m->m_flags & M_PROMISC) { 768 m_freem(m); 769 return; 770 } 771 } 772 773 /* If the CRC is still on the packet, trim it off. */ 774 if (m->m_flags & M_HASFCS) { 775 m_adj(m, -ETHER_CRC_LEN); 776 m->m_flags &= ~M_HASFCS; 777 } 778 779 if (etype > ETHERMTU + sizeof (struct ether_header)) { 780 /* Strip off the Ethernet header. */ 781 m_adj(m, ehlen); 782 783 switch (etype) { 784 #ifdef INET 785 case ETHERTYPE_IP: 786 #ifdef GATEWAY 787 if (ipflow_fastforward(m)) 788 return; 789 #endif 790 pktq = ip_pktq; 791 break; 792 793 case ETHERTYPE_ARP: 794 isr = NETISR_ARP; 795 inq = &arpintrq; 796 break; 797 798 case ETHERTYPE_REVARP: 799 revarpinput(m); /* XXX queue? */ 800 return; 801 #endif 802 #ifdef INET6 803 case ETHERTYPE_IPV6: 804 if (__predict_false(!in6_present)) { 805 m_freem(m); 806 return; 807 } 808 #ifdef GATEWAY 809 if (ip6flow_fastforward(&m)) 810 return; 811 #endif 812 pktq = ip6_pktq; 813 break; 814 #endif 815 #ifdef NETATALK 816 case ETHERTYPE_ATALK: 817 isr = NETISR_ATALK; 818 inq = &atintrq1; 819 break; 820 case ETHERTYPE_AARP: 821 /* probably this should be done with a NETISR as well */ 822 aarpinput(ifp, m); /* XXX */ 823 return; 824 #endif /* NETATALK */ 825 #ifdef MPLS 826 case ETHERTYPE_MPLS: 827 isr = NETISR_MPLS; 828 inq = &mplsintrq; 829 break; 830 #endif 831 default: 832 m_freem(m); 833 return; 834 } 835 } else { 836 #if defined (LLC) || defined (NETATALK) 837 if (m->m_len < ehlen + sizeof(struct llc)) { 838 goto dropanyway; 839 } 840 l = (struct llc *)(eh+1); 841 switch (l->llc_dsap) { 842 #ifdef NETATALK 843 case LLC_SNAP_LSAP: 844 switch (l->llc_control) { 845 case LLC_UI: 846 if (l->llc_ssap != LLC_SNAP_LSAP) { 847 goto dropanyway; 848 } 849 850 if (memcmp(&(l->llc_snap_org_code)[0], 851 at_org_code, sizeof(at_org_code)) == 0 && 852 ntohs(l->llc_snap_ether_type) == 853 ETHERTYPE_ATALK) { 854 inq = &atintrq2; 855 m_adj(m, sizeof(struct ether_header) 856 + sizeof(struct llc)); 857 isr = NETISR_ATALK; 858 break; 859 } 860 861 if (memcmp(&(l->llc_snap_org_code)[0], 862 aarp_org_code, 863 sizeof(aarp_org_code)) == 0 && 864 ntohs(l->llc_snap_ether_type) == 865 ETHERTYPE_AARP) { 866 m_adj( m, sizeof(struct ether_header) 867 + sizeof(struct llc)); 868 aarpinput(ifp, m); /* XXX */ 869 return; 870 } 871 872 default: 873 goto dropanyway; 874 } 875 break; 876 #endif 877 dropanyway: 878 default: 879 m_freem(m); 880 return; 881 } 882 #else /* LLC || NETATALK */ 883 m_freem(m); 884 return; 885 #endif /* LLC || NETATALK */ 886 } 887 888 if (__predict_true(pktq)) { 889 #ifdef NET_MPSAFE 890 const u_int h = curcpu()->ci_index; 891 #else 892 const uint32_t h = pktq_rps_hash(m); 893 #endif 894 if (__predict_false(!pktq_enqueue(pktq, m, h))) { 895 m_freem(m); 896 } 897 return; 898 } 899 900 if (__predict_false(!inq)) { 901 /* Should not happen. */ 902 m_freem(m); 903 return; 904 } 905 906 IFQ_LOCK(inq); 907 if (IF_QFULL(inq)) { 908 IF_DROP(inq); 909 IFQ_UNLOCK(inq); 910 m_freem(m); 911 } else { 912 IF_ENQUEUE(inq, m); 913 IFQ_UNLOCK(inq); 914 schednetisr(isr); 915 } 916 } 917 918 /* 919 * Convert Ethernet address to printable (loggable) representation. 920 */ 921 char * 922 ether_sprintf(const u_char *ap) 923 { 924 static char etherbuf[3 * ETHER_ADDR_LEN]; 925 return ether_snprintf(etherbuf, sizeof(etherbuf), ap); 926 } 927 928 char * 929 ether_snprintf(char *buf, size_t len, const u_char *ap) 930 { 931 char *cp = buf; 932 size_t i; 933 934 for (i = 0; i < len / 3; i++) { 935 *cp++ = hexdigits[*ap >> 4]; 936 *cp++ = hexdigits[*ap++ & 0xf]; 937 *cp++ = ':'; 938 } 939 *--cp = '\0'; 940 return buf; 941 } 942 943 /* 944 * Perform common duties while attaching to interface list 945 */ 946 void 947 ether_ifattach(struct ifnet *ifp, const uint8_t *lla) 948 { 949 struct ethercom *ec = (struct ethercom *)ifp; 950 951 ifp->if_type = IFT_ETHER; 952 ifp->if_hdrlen = ETHER_HDR_LEN; 953 ifp->if_dlt = DLT_EN10MB; 954 ifp->if_mtu = ETHERMTU; 955 ifp->if_output = ether_output; 956 ifp->_if_input = ether_input; 957 if (ifp->if_baudrate == 0) 958 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 959 960 if (lla != NULL) 961 if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla)); 962 963 LIST_INIT(&ec->ec_multiaddrs); 964 ec->ec_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NET); 965 ifp->if_broadcastaddr = etherbroadcastaddr; 966 bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header)); 967 #ifdef MBUFTRACE 968 strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname, 969 sizeof(ec->ec_tx_mowner.mo_name)); 970 strlcpy(ec->ec_tx_mowner.mo_descr, "tx", 971 sizeof(ec->ec_tx_mowner.mo_descr)); 972 strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname, 973 sizeof(ec->ec_rx_mowner.mo_name)); 974 strlcpy(ec->ec_rx_mowner.mo_descr, "rx", 975 sizeof(ec->ec_rx_mowner.mo_descr)); 976 MOWNER_ATTACH(&ec->ec_tx_mowner); 977 MOWNER_ATTACH(&ec->ec_rx_mowner); 978 ifp->if_mowner = &ec->ec_tx_mowner; 979 #endif 980 } 981 982 void 983 ether_ifdetach(struct ifnet *ifp) 984 { 985 struct ethercom *ec = (void *) ifp; 986 struct ether_multi *enm; 987 988 /* 989 * Prevent further calls to ioctl (for example turning off 990 * promiscuous mode from the bridge code), which eventually can 991 * call if_init() which can cause panics because the interface 992 * is in the process of being detached. Return device not configured 993 * instead. 994 */ 995 ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio; 996 997 #if NBRIDGE > 0 998 if (ifp->if_bridge) 999 bridge_ifdetach(ifp); 1000 #endif 1001 1002 bpf_detach(ifp); 1003 1004 #if NVLAN > 0 1005 if (ec->ec_nvlans) 1006 vlan_ifdetach(ifp); 1007 #endif 1008 1009 ETHER_LOCK(ec); 1010 while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) { 1011 LIST_REMOVE(enm, enm_list); 1012 kmem_free(enm, sizeof(*enm)); 1013 ec->ec_multicnt--; 1014 } 1015 ETHER_UNLOCK(ec); 1016 1017 mutex_destroy(ec->ec_lock); 1018 1019 ifp->if_mowner = NULL; 1020 MOWNER_DETACH(&ec->ec_rx_mowner); 1021 MOWNER_DETACH(&ec->ec_tx_mowner); 1022 } 1023 1024 #if 0 1025 /* 1026 * This is for reference. We have a table-driven version 1027 * of the little-endian crc32 generator, which is faster 1028 * than the double-loop. 1029 */ 1030 uint32_t 1031 ether_crc32_le(const uint8_t *buf, size_t len) 1032 { 1033 uint32_t c, crc, carry; 1034 size_t i, j; 1035 1036 crc = 0xffffffffU; /* initial value */ 1037 1038 for (i = 0; i < len; i++) { 1039 c = buf[i]; 1040 for (j = 0; j < 8; j++) { 1041 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1042 crc >>= 1; 1043 c >>= 1; 1044 if (carry) 1045 crc = (crc ^ ETHER_CRC_POLY_LE); 1046 } 1047 } 1048 1049 return (crc); 1050 } 1051 #else 1052 uint32_t 1053 ether_crc32_le(const uint8_t *buf, size_t len) 1054 { 1055 static const uint32_t crctab[] = { 1056 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1057 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1058 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1059 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1060 }; 1061 uint32_t crc; 1062 size_t i; 1063 1064 crc = 0xffffffffU; /* initial value */ 1065 1066 for (i = 0; i < len; i++) { 1067 crc ^= buf[i]; 1068 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1069 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1070 } 1071 1072 return (crc); 1073 } 1074 #endif 1075 1076 uint32_t 1077 ether_crc32_be(const uint8_t *buf, size_t len) 1078 { 1079 uint32_t c, crc, carry; 1080 size_t i, j; 1081 1082 crc = 0xffffffffU; /* initial value */ 1083 1084 for (i = 0; i < len; i++) { 1085 c = buf[i]; 1086 for (j = 0; j < 8; j++) { 1087 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1088 crc <<= 1; 1089 c >>= 1; 1090 if (carry) 1091 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1092 } 1093 } 1094 1095 return (crc); 1096 } 1097 1098 #ifdef INET 1099 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] = 1100 { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 }; 1101 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] = 1102 { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff }; 1103 #endif 1104 #ifdef INET6 1105 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] = 1106 { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 }; 1107 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] = 1108 { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff }; 1109 #endif 1110 1111 /* 1112 * ether_aton implementation, not using a static buffer. 1113 */ 1114 int 1115 ether_aton_r(u_char *dest, size_t len, const char *str) 1116 { 1117 const u_char *cp = (const void *)str; 1118 u_char *ep; 1119 1120 #define atox(c) (((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10)) 1121 1122 if (len < ETHER_ADDR_LEN) 1123 return ENOSPC; 1124 1125 ep = dest + ETHER_ADDR_LEN; 1126 1127 while (*cp) { 1128 if (!isxdigit(*cp)) 1129 return EINVAL; 1130 *dest = atox(*cp); 1131 cp++; 1132 if (isxdigit(*cp)) { 1133 *dest = (*dest << 4) | atox(*cp); 1134 dest++; 1135 cp++; 1136 } else { 1137 dest++; 1138 } 1139 if (dest == ep) 1140 return *cp == '\0' ? 0 : ENAMETOOLONG; 1141 switch (*cp) { 1142 case ':': 1143 case '-': 1144 case '.': 1145 cp++; 1146 break; 1147 } 1148 } 1149 return ENOBUFS; 1150 } 1151 1152 /* 1153 * Convert a sockaddr into an Ethernet address or range of Ethernet 1154 * addresses. 1155 */ 1156 int 1157 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN], 1158 uint8_t addrhi[ETHER_ADDR_LEN]) 1159 { 1160 #ifdef INET 1161 const struct sockaddr_in *sin; 1162 #endif /* INET */ 1163 #ifdef INET6 1164 const struct sockaddr_in6 *sin6; 1165 #endif /* INET6 */ 1166 1167 switch (sa->sa_family) { 1168 1169 case AF_UNSPEC: 1170 memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN); 1171 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1172 break; 1173 1174 #ifdef INET 1175 case AF_INET: 1176 sin = satocsin(sa); 1177 if (sin->sin_addr.s_addr == INADDR_ANY) { 1178 /* 1179 * An IP address of INADDR_ANY means listen to 1180 * or stop listening to all of the Ethernet 1181 * multicast addresses used for IP. 1182 * (This is for the sake of IP multicast routers.) 1183 */ 1184 memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN); 1185 memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN); 1186 } 1187 else { 1188 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo); 1189 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1190 } 1191 break; 1192 #endif 1193 #ifdef INET6 1194 case AF_INET6: 1195 sin6 = satocsin6(sa); 1196 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1197 /* 1198 * An IP6 address of 0 means listen to or stop 1199 * listening to all of the Ethernet multicast 1200 * address used for IP6. 1201 * (This is used for multicast routers.) 1202 */ 1203 memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN); 1204 memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN); 1205 } else { 1206 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo); 1207 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1208 } 1209 break; 1210 #endif 1211 1212 default: 1213 return EAFNOSUPPORT; 1214 } 1215 return 0; 1216 } 1217 1218 /* 1219 * Add an Ethernet multicast address or range of addresses to the list for a 1220 * given interface. 1221 */ 1222 int 1223 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec) 1224 { 1225 struct ether_multi *enm, *_enm; 1226 u_char addrlo[ETHER_ADDR_LEN]; 1227 u_char addrhi[ETHER_ADDR_LEN]; 1228 int error = 0; 1229 1230 /* Allocate out of lock */ 1231 enm = kmem_alloc(sizeof(*enm), KM_SLEEP); 1232 if (enm == NULL) 1233 return ENOBUFS; 1234 1235 ETHER_LOCK(ec); 1236 error = ether_multiaddr(sa, addrlo, addrhi); 1237 if (error != 0) 1238 goto out; 1239 1240 /* 1241 * Verify that we have valid Ethernet multicast addresses. 1242 */ 1243 if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) { 1244 error = EINVAL; 1245 goto out; 1246 } 1247 /* 1248 * See if the address range is already in the list. 1249 */ 1250 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, _enm); 1251 if (_enm != NULL) { 1252 /* 1253 * Found it; just increment the reference count. 1254 */ 1255 ++_enm->enm_refcount; 1256 error = 0; 1257 goto out; 1258 } 1259 /* 1260 * Link a new multicast record into the interface's multicast list. 1261 */ 1262 memcpy(enm->enm_addrlo, addrlo, 6); 1263 memcpy(enm->enm_addrhi, addrhi, 6); 1264 enm->enm_refcount = 1; 1265 LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list); 1266 ec->ec_multicnt++; 1267 /* 1268 * Return ENETRESET to inform the driver that the list has changed 1269 * and its reception filter should be adjusted accordingly. 1270 */ 1271 error = ENETRESET; 1272 enm = NULL; 1273 out: 1274 ETHER_UNLOCK(ec); 1275 if (enm != NULL) 1276 kmem_free(enm, sizeof(*enm)); 1277 return error; 1278 } 1279 1280 /* 1281 * Delete a multicast address record. 1282 */ 1283 int 1284 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec) 1285 { 1286 struct ether_multi *enm; 1287 u_char addrlo[ETHER_ADDR_LEN]; 1288 u_char addrhi[ETHER_ADDR_LEN]; 1289 int error; 1290 1291 ETHER_LOCK(ec); 1292 error = ether_multiaddr(sa, addrlo, addrhi); 1293 if (error != 0) 1294 goto error; 1295 1296 /* 1297 * Look ur the address in our list. 1298 */ 1299 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm); 1300 if (enm == NULL) { 1301 error = ENXIO; 1302 goto error; 1303 } 1304 if (--enm->enm_refcount != 0) { 1305 /* 1306 * Still some claims to this record. 1307 */ 1308 error = 0; 1309 goto error; 1310 } 1311 /* 1312 * No remaining claims to this record; unlink and free it. 1313 */ 1314 LIST_REMOVE(enm, enm_list); 1315 ec->ec_multicnt--; 1316 ETHER_UNLOCK(ec); 1317 1318 kmem_free(enm, sizeof(*enm)); 1319 /* 1320 * Return ENETRESET to inform the driver that the list has changed 1321 * and its reception filter should be adjusted accordingly. 1322 */ 1323 return ENETRESET; 1324 error: 1325 ETHER_UNLOCK(ec); 1326 return error; 1327 } 1328 1329 void 1330 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb) 1331 { 1332 ec->ec_ifflags_cb = cb; 1333 } 1334 1335 /* 1336 * Common ioctls for Ethernet interfaces. Note, we must be 1337 * called at splnet(). 1338 */ 1339 int 1340 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data) 1341 { 1342 struct ethercom *ec = (void *) ifp; 1343 struct eccapreq *eccr; 1344 struct ifreq *ifr = (struct ifreq *)data; 1345 struct if_laddrreq *iflr = data; 1346 const struct sockaddr_dl *sdl; 1347 static const uint8_t zero[ETHER_ADDR_LEN]; 1348 int error; 1349 1350 switch (cmd) { 1351 case SIOCINITIFADDR: 1352 { 1353 struct ifaddr *ifa = (struct ifaddr *)data; 1354 if (ifa->ifa_addr->sa_family != AF_LINK 1355 && (ifp->if_flags & (IFF_UP | IFF_RUNNING)) != 1356 (IFF_UP | IFF_RUNNING)) { 1357 ifp->if_flags |= IFF_UP; 1358 if ((error = (*ifp->if_init)(ifp)) != 0) 1359 return error; 1360 } 1361 #ifdef INET 1362 if (ifa->ifa_addr->sa_family == AF_INET) 1363 arp_ifinit(ifp, ifa); 1364 #endif /* INET */ 1365 return 0; 1366 } 1367 1368 case SIOCSIFMTU: 1369 { 1370 int maxmtu; 1371 1372 if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU) 1373 maxmtu = ETHERMTU_JUMBO; 1374 else 1375 maxmtu = ETHERMTU; 1376 1377 if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu) 1378 return EINVAL; 1379 else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET) 1380 return error; 1381 else if (ifp->if_flags & IFF_UP) { 1382 /* Make sure the device notices the MTU change. */ 1383 return (*ifp->if_init)(ifp); 1384 } else 1385 return 0; 1386 } 1387 1388 case SIOCSIFFLAGS: 1389 if ((error = ifioctl_common(ifp, cmd, data)) != 0) 1390 return error; 1391 switch (ifp->if_flags & (IFF_UP | IFF_RUNNING)) { 1392 case IFF_RUNNING: 1393 /* 1394 * If interface is marked down and it is running, 1395 * then stop and disable it. 1396 */ 1397 (*ifp->if_stop)(ifp, 1); 1398 break; 1399 case IFF_UP: 1400 /* 1401 * If interface is marked up and it is stopped, then 1402 * start it. 1403 */ 1404 return (*ifp->if_init)(ifp); 1405 case IFF_UP | IFF_RUNNING: 1406 error = 0; 1407 if (ec->ec_ifflags_cb != NULL) { 1408 error = (*ec->ec_ifflags_cb)(ec); 1409 if (error == ENETRESET) { 1410 /* 1411 * Reset the interface to pick up 1412 * changes in any other flags that 1413 * affect the hardware state. 1414 */ 1415 return (*ifp->if_init)(ifp); 1416 } 1417 } else 1418 error = (*ifp->if_init)(ifp); 1419 return error; 1420 case 0: 1421 break; 1422 } 1423 return 0; 1424 case SIOCGETHERCAP: 1425 eccr = (struct eccapreq *)data; 1426 eccr->eccr_capabilities = ec->ec_capabilities; 1427 eccr->eccr_capenable = ec->ec_capenable; 1428 return 0; 1429 case SIOCADDMULTI: 1430 return ether_addmulti(ifreq_getaddr(cmd, ifr), ec); 1431 case SIOCDELMULTI: 1432 return ether_delmulti(ifreq_getaddr(cmd, ifr), ec); 1433 case SIOCSIFMEDIA: 1434 case SIOCGIFMEDIA: 1435 if (ec->ec_mii == NULL) 1436 return ENOTTY; 1437 return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd); 1438 case SIOCALIFADDR: 1439 sdl = satocsdl(sstocsa(&iflr->addr)); 1440 if (sdl->sdl_family != AF_LINK) 1441 ; 1442 else if (ETHER_IS_MULTICAST(CLLADDR(sdl))) 1443 return EINVAL; 1444 else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0) 1445 return EINVAL; 1446 /*FALLTHROUGH*/ 1447 default: 1448 return ifioctl_common(ifp, cmd, data); 1449 } 1450 return 0; 1451 } 1452 1453 /* 1454 * Enable/disable passing VLAN packets if the parent interface supports it. 1455 * Return: 1456 * 0: Ok 1457 * -1: Parent interface does not support vlans 1458 * >0: Error 1459 */ 1460 int 1461 ether_enable_vlan_mtu(struct ifnet *ifp) 1462 { 1463 int error; 1464 struct ethercom *ec = (void *)ifp; 1465 1466 /* Parent does not support VLAN's */ 1467 if ((ec->ec_capabilities & ETHERCAP_VLAN_MTU) == 0) 1468 return -1; 1469 1470 /* 1471 * Parent supports the VLAN_MTU capability, 1472 * i.e. can Tx/Rx larger than ETHER_MAX_LEN frames; 1473 * enable it. 1474 */ 1475 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1476 1477 /* Interface is down, defer for later */ 1478 if ((ifp->if_flags & IFF_UP) == 0) 1479 return 0; 1480 1481 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1482 return 0; 1483 1484 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1485 return error; 1486 } 1487 1488 int 1489 ether_disable_vlan_mtu(struct ifnet *ifp) 1490 { 1491 int error; 1492 struct ethercom *ec = (void *)ifp; 1493 1494 /* We still have VLAN's, defer for later */ 1495 if (ec->ec_nvlans != 0) 1496 return 0; 1497 1498 /* Parent does not support VLAB's, nothing to do. */ 1499 if ((ec->ec_capenable & ETHERCAP_VLAN_MTU) == 0) 1500 return -1; 1501 1502 /* 1503 * Disable Tx/Rx of VLAN-sized frames. 1504 */ 1505 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1506 1507 /* Interface is down, defer for later */ 1508 if ((ifp->if_flags & IFF_UP) == 0) 1509 return 0; 1510 1511 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1512 return 0; 1513 1514 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1515 return error; 1516 } 1517 1518 static int 1519 ether_multicast_sysctl(SYSCTLFN_ARGS) 1520 { 1521 struct ether_multi *enm; 1522 struct ifnet *ifp; 1523 struct ethercom *ec; 1524 int error = 0; 1525 size_t written; 1526 struct psref psref; 1527 int bound; 1528 unsigned int multicnt; 1529 struct ether_multi_sysctl *addrs; 1530 int i; 1531 1532 if (namelen != 1) 1533 return EINVAL; 1534 1535 bound = curlwp_bind(); 1536 ifp = if_get_byindex(name[0], &psref); 1537 if (ifp == NULL) { 1538 error = ENODEV; 1539 goto out; 1540 } 1541 if (ifp->if_type != IFT_ETHER) { 1542 if_put(ifp, &psref); 1543 *oldlenp = 0; 1544 goto out; 1545 } 1546 ec = (struct ethercom *)ifp; 1547 1548 if (oldp == NULL) { 1549 if_put(ifp, &psref); 1550 *oldlenp = ec->ec_multicnt * sizeof(*addrs); 1551 goto out; 1552 } 1553 1554 /* 1555 * ec->ec_lock is a spin mutex so we cannot call sysctl_copyout, which 1556 * is sleepable, with holding it. Copy data to a local buffer first 1557 * with holding it and then call sysctl_copyout without holding it. 1558 */ 1559 retry: 1560 multicnt = ec->ec_multicnt; 1561 addrs = kmem_alloc(sizeof(*addrs) * multicnt, KM_SLEEP); 1562 1563 ETHER_LOCK(ec); 1564 if (multicnt < ec->ec_multicnt) { 1565 /* The number of multicast addresses have increased */ 1566 ETHER_UNLOCK(ec); 1567 kmem_free(addrs, sizeof(*addrs) * multicnt); 1568 goto retry; 1569 } 1570 1571 i = 0; 1572 LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) { 1573 struct ether_multi_sysctl *addr = &addrs[i]; 1574 addr->enm_refcount = enm->enm_refcount; 1575 memcpy(addr->enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN); 1576 memcpy(addr->enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN); 1577 i++; 1578 } 1579 ETHER_UNLOCK(ec); 1580 1581 error = 0; 1582 written = 0; 1583 for (i = 0; i < multicnt; i++) { 1584 struct ether_multi_sysctl *addr = &addrs[i]; 1585 1586 if (written + sizeof(*addr) > *oldlenp) 1587 break; 1588 error = sysctl_copyout(l, addr, oldp, sizeof(*addr)); 1589 if (error) 1590 break; 1591 written += sizeof(*addr); 1592 oldp = (char *)oldp + sizeof(*addr); 1593 } 1594 kmem_free(addrs, sizeof(*addrs) * multicnt); 1595 1596 if_put(ifp, &psref); 1597 1598 *oldlenp = written; 1599 out: 1600 curlwp_bindx(bound); 1601 return error; 1602 } 1603 1604 static void 1605 ether_sysctl_setup(struct sysctllog **clog) 1606 { 1607 const struct sysctlnode *rnode = NULL; 1608 1609 sysctl_createv(clog, 0, NULL, &rnode, 1610 CTLFLAG_PERMANENT, 1611 CTLTYPE_NODE, "ether", 1612 SYSCTL_DESCR("Ethernet-specific information"), 1613 NULL, 0, NULL, 0, 1614 CTL_NET, CTL_CREATE, CTL_EOL); 1615 1616 sysctl_createv(clog, 0, &rnode, NULL, 1617 CTLFLAG_PERMANENT, 1618 CTLTYPE_NODE, "multicast", 1619 SYSCTL_DESCR("multicast addresses"), 1620 ether_multicast_sysctl, 0, NULL, 0, 1621 CTL_CREATE, CTL_EOL); 1622 } 1623 1624 void 1625 etherinit(void) 1626 { 1627 1628 mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET); 1629 ether_sysctl_setup(NULL); 1630 } 1631