if_ethersubr.c revision 1.262
1 /* $NetBSD: if_ethersubr.c,v 1.262 2018/04/09 11:35:22 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 1.262 2018/04/09 11:35:22 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 #include "carp.h" 146 #if NCARP > 0 147 #include <netinet/ip_carp.h> 148 #endif 149 150 #ifdef NETATALK 151 #include <netatalk/at.h> 152 #include <netatalk/at_var.h> 153 #include <netatalk/at_extern.h> 154 155 #define llc_snap_org_code llc_un.type_snap.org_code 156 #define llc_snap_ether_type llc_un.type_snap.ether_type 157 158 extern u_char at_org_code[3]; 159 extern u_char aarp_org_code[3]; 160 #endif /* NETATALK */ 161 162 #ifdef MPLS 163 #include <netmpls/mpls.h> 164 #include <netmpls/mpls_var.h> 165 #endif 166 167 static struct timeval bigpktppslim_last; 168 static int bigpktppslim = 2; /* XXX */ 169 static int bigpktpps_count; 170 static kmutex_t bigpktpps_lock __cacheline_aligned; 171 172 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = 173 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 174 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] = 175 { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 }; 176 #define senderr(e) { error = (e); goto bad;} 177 178 static int ether_output(struct ifnet *, struct mbuf *, 179 const struct sockaddr *, const struct rtentry *); 180 181 /* 182 * Ethernet output routine. 183 * Encapsulate a packet of type family for the local net. 184 * Assumes that ifp is actually pointer to ethercom structure. 185 */ 186 static int 187 ether_output(struct ifnet * const ifp0, struct mbuf * const m0, 188 const struct sockaddr * const dst, const struct rtentry *rt) 189 { 190 uint8_t esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN]; 191 uint16_t etype = 0; 192 int error = 0, hdrcmplt = 0; 193 struct mbuf *m = m0; 194 struct mbuf *mcopy = NULL; 195 struct ether_header *eh; 196 struct ifnet *ifp = ifp0; 197 #ifdef INET 198 struct arphdr *ah; 199 #endif 200 #ifdef NETATALK 201 struct at_ifaddr *aa; 202 #endif 203 204 #ifdef MBUFTRACE 205 m_claimm(m, ifp->if_mowner); 206 #endif 207 208 #if NCARP > 0 209 if (ifp->if_type == IFT_CARP) { 210 struct ifaddr *ifa; 211 int s = pserialize_read_enter(); 212 213 /* loop back if this is going to the carp interface */ 214 if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP && 215 (ifa = ifa_ifwithaddr(dst)) != NULL) { 216 if (ifa->ifa_ifp == ifp0) { 217 pserialize_read_exit(s); 218 return looutput(ifp0, m, dst, rt); 219 } 220 } 221 pserialize_read_exit(s); 222 223 ifp = ifp->if_carpdev; 224 /* ac = (struct arpcom *)ifp; */ 225 226 if ((ifp0->if_flags & (IFF_UP | IFF_RUNNING)) != 227 (IFF_UP | IFF_RUNNING)) 228 senderr(ENETDOWN); 229 } 230 #endif 231 232 if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING)) 233 senderr(ENETDOWN); 234 235 switch (dst->sa_family) { 236 237 #ifdef INET 238 case AF_INET: 239 if (m->m_flags & M_BCAST) { 240 memcpy(edst, etherbroadcastaddr, sizeof(edst)); 241 } else if (m->m_flags & M_MCAST) { 242 ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst); 243 } else { 244 error = arpresolve(ifp, rt, m, dst, edst, sizeof(edst)); 245 if (error) 246 return (error == EWOULDBLOCK) ? 0 : error; 247 } 248 /* If broadcasting on a simplex interface, loopback a copy */ 249 if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX)) 250 mcopy = m_copy(m, 0, M_COPYALL); 251 etype = htons(ETHERTYPE_IP); 252 break; 253 254 case AF_ARP: 255 ah = mtod(m, struct arphdr *); 256 if (m->m_flags & M_BCAST) { 257 memcpy(edst, etherbroadcastaddr, sizeof(edst)); 258 } else { 259 void *tha = ar_tha(ah); 260 261 if (tha == NULL) { 262 /* fake with ARPHRD_IEEE1394 */ 263 m_freem(m); 264 return 0; 265 } 266 memcpy(edst, tha, sizeof(edst)); 267 } 268 269 ah->ar_hrd = htons(ARPHRD_ETHER); 270 271 switch (ntohs(ah->ar_op)) { 272 case ARPOP_REVREQUEST: 273 case ARPOP_REVREPLY: 274 etype = htons(ETHERTYPE_REVARP); 275 break; 276 277 case ARPOP_REQUEST: 278 case ARPOP_REPLY: 279 default: 280 etype = htons(ETHERTYPE_ARP); 281 } 282 break; 283 #endif 284 285 #ifdef INET6 286 case AF_INET6: 287 if (m->m_flags & M_BCAST) { 288 memcpy(edst, etherbroadcastaddr, sizeof(edst)); 289 } else if (m->m_flags & M_MCAST) { 290 ETHER_MAP_IPV6_MULTICAST(&satocsin6(dst)->sin6_addr, 291 edst); 292 } else { 293 error = nd6_resolve(ifp, rt, m, dst, edst, 294 sizeof(edst)); 295 if (error) 296 return (error == EWOULDBLOCK) ? 0 : error; 297 } 298 etype = htons(ETHERTYPE_IPV6); 299 break; 300 #endif 301 302 #ifdef NETATALK 303 case AF_APPLETALK: { 304 struct ifaddr *ifa; 305 int s; 306 307 KERNEL_LOCK(1, NULL); 308 309 if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) { 310 #ifdef NETATALKDEBUG 311 printf("aarpresolve failed\n"); 312 #endif 313 KERNEL_UNLOCK_ONE(NULL); 314 return 0; 315 } 316 317 /* 318 * ifaddr is the first thing in at_ifaddr 319 */ 320 s = pserialize_read_enter(); 321 ifa = at_ifawithnet((const struct sockaddr_at *)dst, ifp); 322 if (ifa == NULL) { 323 pserialize_read_exit(s); 324 KERNEL_UNLOCK_ONE(NULL); 325 senderr(EADDRNOTAVAIL); 326 } 327 aa = (struct at_ifaddr *)ifa; 328 329 /* 330 * In the phase 2 case, we need to prepend an mbuf for the 331 * llc header. Since we must preserve the value of m, 332 * which is passed to us by value, we m_copy() the first 333 * mbuf, and use it for our llc header. 334 */ 335 if (aa->aa_flags & AFA_PHASE2) { 336 struct llc llc; 337 338 M_PREPEND(m, sizeof(struct llc), M_DONTWAIT); 339 if (m == NULL) { 340 pserialize_read_exit(s); 341 KERNEL_UNLOCK_ONE(NULL); 342 senderr(ENOBUFS); 343 } 344 345 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 346 llc.llc_control = LLC_UI; 347 memcpy(llc.llc_snap_org_code, at_org_code, 348 sizeof(llc.llc_snap_org_code)); 349 llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK); 350 memcpy(mtod(m, void *), &llc, sizeof(struct llc)); 351 } else { 352 etype = htons(ETHERTYPE_ATALK); 353 } 354 pserialize_read_exit(s); 355 KERNEL_UNLOCK_ONE(NULL); 356 break; 357 } 358 #endif /* NETATALK */ 359 360 case pseudo_AF_HDRCMPLT: 361 hdrcmplt = 1; 362 memcpy(esrc, 363 ((const struct ether_header *)dst->sa_data)->ether_shost, 364 sizeof(esrc)); 365 /* FALLTHROUGH */ 366 367 case AF_UNSPEC: 368 memcpy(edst, 369 ((const struct ether_header *)dst->sa_data)->ether_dhost, 370 sizeof(edst)); 371 /* AF_UNSPEC doesn't swap the byte order of the ether_type. */ 372 etype = ((const struct ether_header *)dst->sa_data)->ether_type; 373 break; 374 375 default: 376 printf("%s: can't handle af%d\n", ifp->if_xname, 377 dst->sa_family); 378 senderr(EAFNOSUPPORT); 379 } 380 381 #ifdef MPLS 382 { 383 struct m_tag *mtag; 384 mtag = m_tag_find(m, PACKET_TAG_MPLS, NULL); 385 if (mtag != NULL) { 386 /* Having the tag itself indicates it's MPLS */ 387 etype = htons(ETHERTYPE_MPLS); 388 m_tag_delete(m, mtag); 389 } 390 } 391 #endif 392 393 if (mcopy) 394 (void)looutput(ifp, mcopy, dst, rt); 395 396 KASSERT((m->m_flags & M_PKTHDR) != 0); 397 398 /* 399 * If no ether type is set, this must be a 802.2 formatted packet. 400 */ 401 if (etype == 0) 402 etype = htons(m->m_pkthdr.len); 403 404 /* 405 * Add local net header. If no space in first mbuf, allocate another. 406 */ 407 M_PREPEND(m, sizeof(struct ether_header), M_DONTWAIT); 408 if (m == NULL) 409 senderr(ENOBUFS); 410 411 eh = mtod(m, struct ether_header *); 412 /* Note: etype is already in network byte order. */ 413 memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type)); 414 memcpy(eh->ether_dhost, edst, sizeof(edst)); 415 if (hdrcmplt) { 416 memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost)); 417 } else { 418 memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl), 419 sizeof(eh->ether_shost)); 420 } 421 422 #if NCARP > 0 423 if (ifp0 != ifp && ifp0->if_type == IFT_CARP) { 424 memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl), 425 sizeof(eh->ether_shost)); 426 } 427 #endif 428 429 if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0) 430 return error; 431 if (m == NULL) 432 return 0; 433 434 #if NBRIDGE > 0 435 /* 436 * Bridges require special output handling. 437 */ 438 if (ifp->if_bridge) 439 return bridge_output(ifp, m, NULL, NULL); 440 #endif 441 442 #if NCARP > 0 443 if (ifp != ifp0) 444 ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN; 445 #endif 446 447 #ifdef ALTQ 448 KERNEL_LOCK(1, NULL); 449 /* 450 * If ALTQ is enabled on the parent interface, do 451 * classification; the queueing discipline might not 452 * require classification, but might require the 453 * address family/header pointer in the pktattr. 454 */ 455 if (ALTQ_IS_ENABLED(&ifp->if_snd)) 456 altq_etherclassify(&ifp->if_snd, m); 457 KERNEL_UNLOCK_ONE(NULL); 458 #endif 459 return ifq_enqueue(ifp, m); 460 461 bad: 462 if (m) 463 m_freem(m); 464 return error; 465 } 466 467 #ifdef ALTQ 468 /* 469 * This routine is a slight hack to allow a packet to be classified 470 * if the Ethernet headers are present. It will go away when ALTQ's 471 * classification engine understands link headers. 472 * 473 * XXX: We may need to do m_pullups here. First to ensure struct ether_header 474 * is indeed contiguous, then to read the LLC and so on. 475 */ 476 void 477 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m) 478 { 479 struct ether_header *eh; 480 struct mbuf *mtop = m; 481 uint16_t ether_type; 482 int hlen, af, hdrsize; 483 void *hdr; 484 485 KASSERT((mtop->m_flags & M_PKTHDR) != 0); 486 487 hlen = ETHER_HDR_LEN; 488 eh = mtod(m, struct ether_header *); 489 490 ether_type = htons(eh->ether_type); 491 492 if (ether_type < ETHERMTU) { 493 /* LLC/SNAP */ 494 struct llc *llc = (struct llc *)(eh + 1); 495 hlen += 8; 496 497 if (m->m_len < hlen || 498 llc->llc_dsap != LLC_SNAP_LSAP || 499 llc->llc_ssap != LLC_SNAP_LSAP || 500 llc->llc_control != LLC_UI) { 501 /* Not SNAP. */ 502 goto bad; 503 } 504 505 ether_type = htons(llc->llc_un.type_snap.ether_type); 506 } 507 508 switch (ether_type) { 509 case ETHERTYPE_IP: 510 af = AF_INET; 511 hdrsize = 20; /* sizeof(struct ip) */ 512 break; 513 514 case ETHERTYPE_IPV6: 515 af = AF_INET6; 516 hdrsize = 40; /* sizeof(struct ip6_hdr) */ 517 break; 518 519 default: 520 af = AF_UNSPEC; 521 hdrsize = 0; 522 break; 523 } 524 525 while (m->m_len <= hlen) { 526 hlen -= m->m_len; 527 m = m->m_next; 528 if (m == NULL) 529 goto bad; 530 } 531 532 if (m->m_len < (hlen + hdrsize)) { 533 /* 534 * protocol header not in a single mbuf. 535 * We can't cope with this situation right 536 * now (but it shouldn't ever happen, really, anyhow). 537 */ 538 #ifdef DEBUG 539 printf("altq_etherclassify: headers span multiple mbufs: " 540 "%d < %d\n", m->m_len, (hlen + hdrsize)); 541 #endif 542 goto bad; 543 } 544 545 m->m_data += hlen; 546 m->m_len -= hlen; 547 548 hdr = mtod(m, void *); 549 550 if (ALTQ_NEEDS_CLASSIFY(ifq)) { 551 mtop->m_pkthdr.pattr_class = 552 (*ifq->altq_classify)(ifq->altq_clfier, m, af); 553 } 554 mtop->m_pkthdr.pattr_af = af; 555 mtop->m_pkthdr.pattr_hdr = hdr; 556 557 m->m_data -= hlen; 558 m->m_len += hlen; 559 560 return; 561 562 bad: 563 mtop->m_pkthdr.pattr_class = NULL; 564 mtop->m_pkthdr.pattr_hdr = NULL; 565 mtop->m_pkthdr.pattr_af = AF_UNSPEC; 566 } 567 #endif /* ALTQ */ 568 569 /* 570 * Process a received Ethernet packet; 571 * the packet is in the mbuf chain m with 572 * the ether header. 573 */ 574 void 575 ether_input(struct ifnet *ifp, struct mbuf *m) 576 { 577 struct ethercom *ec = (struct ethercom *) ifp; 578 pktqueue_t *pktq = NULL; 579 struct ifqueue *inq = NULL; 580 uint16_t etype; 581 struct ether_header *eh; 582 size_t ehlen; 583 static int earlypkts; 584 int isr = 0; 585 #if defined (LLC) || defined(NETATALK) 586 struct llc *l; 587 #endif 588 589 KASSERT(!cpu_intr_p()); 590 KASSERT((m->m_flags & M_PKTHDR) != 0); 591 KASSERT(m->m_len >= sizeof(*eh)); 592 593 if ((ifp->if_flags & IFF_UP) == 0) { 594 m_freem(m); 595 return; 596 } 597 598 #ifdef MBUFTRACE 599 m_claimm(m, &ec->ec_rx_mowner); 600 #endif 601 eh = mtod(m, struct ether_header *); 602 etype = ntohs(eh->ether_type); 603 ehlen = sizeof(*eh); 604 605 if (__predict_false(earlypkts < 100 || !rnd_initial_entropy)) { 606 rnd_add_data(NULL, eh, ehlen, 0); 607 earlypkts++; 608 } 609 610 /* 611 * Determine if the packet is within its size limits. For MPLS the 612 * header length is variable, so we skip the check. 613 */ 614 if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len > 615 ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) { 616 mutex_enter(&bigpktpps_lock); 617 if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count, 618 bigpktppslim)) { 619 printf("%s: discarding oversize frame (len=%d)\n", 620 ifp->if_xname, m->m_pkthdr.len); 621 } 622 mutex_exit(&bigpktpps_lock); 623 m_freem(m); 624 return; 625 } 626 627 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 628 /* 629 * If this is not a simplex interface, drop the packet 630 * if it came from us. 631 */ 632 if ((ifp->if_flags & IFF_SIMPLEX) == 0 && 633 memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost, 634 ETHER_ADDR_LEN) == 0) { 635 m_freem(m); 636 return; 637 } 638 639 if (memcmp(etherbroadcastaddr, 640 eh->ether_dhost, ETHER_ADDR_LEN) == 0) 641 m->m_flags |= M_BCAST; 642 else 643 m->m_flags |= M_MCAST; 644 ifp->if_imcasts++; 645 } 646 647 /* If the CRC is still on the packet, trim it off. */ 648 if (m->m_flags & M_HASFCS) { 649 m_adj(m, -ETHER_CRC_LEN); 650 m->m_flags &= ~M_HASFCS; 651 } 652 653 ifp->if_ibytes += m->m_pkthdr.len; 654 655 #if NCARP > 0 656 if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) { 657 /* 658 * Clear M_PROMISC, in case the packet comes from a 659 * vlan. 660 */ 661 m->m_flags &= ~M_PROMISC; 662 if (carp_input(m, (uint8_t *)&eh->ether_shost, 663 (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0) 664 return; 665 } 666 #endif 667 668 if ((m->m_flags & (M_BCAST | M_MCAST | M_PROMISC)) == 0 && 669 (ifp->if_flags & IFF_PROMISC) != 0 && 670 memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost, 671 ETHER_ADDR_LEN) != 0) { 672 m->m_flags |= M_PROMISC; 673 } 674 675 if ((m->m_flags & M_PROMISC) == 0) { 676 if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0) 677 return; 678 if (m == NULL) 679 return; 680 681 eh = mtod(m, struct ether_header *); 682 etype = ntohs(eh->ether_type); 683 } 684 685 #if NAGR > 0 686 if (ifp->if_agrprivate && 687 __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) { 688 m->m_flags &= ~M_PROMISC; 689 agr_input(ifp, m); 690 return; 691 } 692 #endif 693 694 /* 695 * If VLANs are configured on the interface, check to 696 * see if the device performed the decapsulation and 697 * provided us with the tag. 698 */ 699 if (ec->ec_nvlans && vlan_has_tag(m)) { 700 #if NVLAN > 0 701 /* 702 * vlan_input() will either recursively call ether_input() 703 * or drop the packet. 704 */ 705 vlan_input(ifp, m); 706 #else 707 m_freem(m); 708 #endif 709 return; 710 } 711 712 /* 713 * Handle protocols that expect to have the Ethernet header 714 * (and possibly FCS) intact. 715 */ 716 switch (etype) { 717 case ETHERTYPE_VLAN: { 718 struct ether_vlan_header *evl = (void *)eh; 719 720 /* 721 * If there is a tag of 0, then the VLAN header was probably 722 * just being used to store the priority. Extract the ether 723 * type, and if IP or IPV6, let them deal with it. 724 */ 725 if (m->m_len >= sizeof(*evl) && 726 EVL_VLANOFTAG(evl->evl_tag) == 0) { 727 etype = ntohs(evl->evl_proto); 728 ehlen = sizeof(*evl); 729 if ((m->m_flags & M_PROMISC) == 0 && 730 (etype == ETHERTYPE_IP || 731 etype == ETHERTYPE_IPV6)) 732 break; 733 } 734 735 #if NVLAN > 0 736 /* 737 * vlan_input() will either recursively call ether_input() 738 * or drop the packet. 739 */ 740 if (ec->ec_nvlans != 0) 741 vlan_input(ifp, m); 742 else 743 #endif 744 m_freem(m); 745 746 return; 747 } 748 749 #if NPPPOE > 0 750 case ETHERTYPE_PPPOEDISC: 751 pppoedisc_input(ifp, m); 752 return; 753 754 case ETHERTYPE_PPPOE: 755 pppoe_input(ifp, m); 756 return; 757 #endif 758 759 case ETHERTYPE_SLOWPROTOCOLS: { 760 uint8_t subtype; 761 762 KASSERTMSG((m->m_pkthdr.len >= sizeof(*eh) + sizeof(subtype)), 763 "too short slow protocol packet"); 764 765 m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype); 766 switch (subtype) { 767 #if NAGR > 0 768 case SLOWPROTOCOLS_SUBTYPE_LACP: 769 if (ifp->if_agrprivate) { 770 ieee8023ad_lacp_input(ifp, m); 771 return; 772 } 773 break; 774 775 case SLOWPROTOCOLS_SUBTYPE_MARKER: 776 if (ifp->if_agrprivate) { 777 ieee8023ad_marker_input(ifp, m); 778 return; 779 } 780 break; 781 #endif 782 783 default: 784 if (subtype == 0 || subtype > 10) { 785 /* illegal value */ 786 m_freem(m); 787 return; 788 } 789 /* unknown subtype */ 790 break; 791 } 792 /* FALLTHROUGH */ 793 } 794 795 default: 796 if (m->m_flags & M_PROMISC) { 797 m_freem(m); 798 return; 799 } 800 } 801 802 /* If the CRC is still on the packet, trim it off. */ 803 if (m->m_flags & M_HASFCS) { 804 m_adj(m, -ETHER_CRC_LEN); 805 m->m_flags &= ~M_HASFCS; 806 } 807 808 if (etype > ETHERMTU + sizeof(struct ether_header)) { 809 /* Strip off the Ethernet header. */ 810 m_adj(m, ehlen); 811 812 switch (etype) { 813 #ifdef INET 814 case ETHERTYPE_IP: 815 #ifdef GATEWAY 816 if (ipflow_fastforward(m)) 817 return; 818 #endif 819 pktq = ip_pktq; 820 break; 821 822 case ETHERTYPE_ARP: 823 isr = NETISR_ARP; 824 inq = &arpintrq; 825 break; 826 827 case ETHERTYPE_REVARP: 828 revarpinput(m); /* XXX queue? */ 829 return; 830 #endif 831 832 #ifdef INET6 833 case ETHERTYPE_IPV6: 834 if (__predict_false(!in6_present)) { 835 m_freem(m); 836 return; 837 } 838 #ifdef GATEWAY 839 if (ip6flow_fastforward(&m)) 840 return; 841 #endif 842 pktq = ip6_pktq; 843 break; 844 #endif 845 846 #ifdef NETATALK 847 case ETHERTYPE_ATALK: 848 isr = NETISR_ATALK; 849 inq = &atintrq1; 850 break; 851 852 case ETHERTYPE_AARP: 853 aarpinput(ifp, m); /* XXX queue? */ 854 return; 855 #endif 856 857 #ifdef MPLS 858 case ETHERTYPE_MPLS: 859 isr = NETISR_MPLS; 860 inq = &mplsintrq; 861 break; 862 #endif 863 864 default: 865 m_freem(m); 866 return; 867 } 868 } else { 869 KASSERT(ehlen == sizeof(*eh)); 870 #if defined (LLC) || defined (NETATALK) 871 if (m->m_len < sizeof(*eh) + sizeof(struct llc)) { 872 goto dropanyway; 873 } 874 l = (struct llc *)(eh+1); 875 876 switch (l->llc_dsap) { 877 #ifdef NETATALK 878 case LLC_SNAP_LSAP: 879 switch (l->llc_control) { 880 case LLC_UI: 881 if (l->llc_ssap != LLC_SNAP_LSAP) { 882 goto dropanyway; 883 } 884 885 if (memcmp(&(l->llc_snap_org_code)[0], 886 at_org_code, sizeof(at_org_code)) == 0 && 887 ntohs(l->llc_snap_ether_type) == 888 ETHERTYPE_ATALK) { 889 inq = &atintrq2; 890 m_adj(m, sizeof(struct ether_header) 891 + sizeof(struct llc)); 892 isr = NETISR_ATALK; 893 break; 894 } 895 896 if (memcmp(&(l->llc_snap_org_code)[0], 897 aarp_org_code, 898 sizeof(aarp_org_code)) == 0 && 899 ntohs(l->llc_snap_ether_type) == 900 ETHERTYPE_AARP) { 901 m_adj(m, sizeof(struct ether_header) 902 + sizeof(struct llc)); 903 aarpinput(ifp, m); /* XXX queue? */ 904 return; 905 } 906 907 default: 908 goto dropanyway; 909 } 910 break; 911 #endif 912 dropanyway: 913 default: 914 m_freem(m); 915 return; 916 } 917 #else /* LLC || NETATALK */ 918 m_freem(m); 919 return; 920 #endif /* LLC || NETATALK */ 921 } 922 923 if (__predict_true(pktq)) { 924 #ifdef NET_MPSAFE 925 const u_int h = curcpu()->ci_index; 926 #else 927 const uint32_t h = pktq_rps_hash(m); 928 #endif 929 if (__predict_false(!pktq_enqueue(pktq, m, h))) { 930 m_freem(m); 931 } 932 return; 933 } 934 935 if (__predict_false(!inq)) { 936 /* Should not happen. */ 937 m_freem(m); 938 return; 939 } 940 941 IFQ_LOCK(inq); 942 if (IF_QFULL(inq)) { 943 IF_DROP(inq); 944 IFQ_UNLOCK(inq); 945 m_freem(m); 946 } else { 947 IF_ENQUEUE(inq, m); 948 IFQ_UNLOCK(inq); 949 schednetisr(isr); 950 } 951 } 952 953 /* 954 * Convert Ethernet address to printable (loggable) representation. 955 */ 956 char * 957 ether_sprintf(const u_char *ap) 958 { 959 static char etherbuf[3 * ETHER_ADDR_LEN]; 960 return ether_snprintf(etherbuf, sizeof(etherbuf), ap); 961 } 962 963 char * 964 ether_snprintf(char *buf, size_t len, const u_char *ap) 965 { 966 char *cp = buf; 967 size_t i; 968 969 for (i = 0; i < len / 3; i++) { 970 *cp++ = hexdigits[*ap >> 4]; 971 *cp++ = hexdigits[*ap++ & 0xf]; 972 *cp++ = ':'; 973 } 974 *--cp = '\0'; 975 return buf; 976 } 977 978 /* 979 * Perform common duties while attaching to interface list 980 */ 981 void 982 ether_ifattach(struct ifnet *ifp, const uint8_t *lla) 983 { 984 struct ethercom *ec = (struct ethercom *)ifp; 985 986 ifp->if_type = IFT_ETHER; 987 ifp->if_hdrlen = ETHER_HDR_LEN; 988 ifp->if_dlt = DLT_EN10MB; 989 ifp->if_mtu = ETHERMTU; 990 ifp->if_output = ether_output; 991 ifp->_if_input = ether_input; 992 if (ifp->if_baudrate == 0) 993 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 994 995 if (lla != NULL) 996 if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla)); 997 998 LIST_INIT(&ec->ec_multiaddrs); 999 ec->ec_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NET); 1000 ifp->if_broadcastaddr = etherbroadcastaddr; 1001 bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header)); 1002 #ifdef MBUFTRACE 1003 strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname, 1004 sizeof(ec->ec_tx_mowner.mo_name)); 1005 strlcpy(ec->ec_tx_mowner.mo_descr, "tx", 1006 sizeof(ec->ec_tx_mowner.mo_descr)); 1007 strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname, 1008 sizeof(ec->ec_rx_mowner.mo_name)); 1009 strlcpy(ec->ec_rx_mowner.mo_descr, "rx", 1010 sizeof(ec->ec_rx_mowner.mo_descr)); 1011 MOWNER_ATTACH(&ec->ec_tx_mowner); 1012 MOWNER_ATTACH(&ec->ec_rx_mowner); 1013 ifp->if_mowner = &ec->ec_tx_mowner; 1014 #endif 1015 } 1016 1017 void 1018 ether_ifdetach(struct ifnet *ifp) 1019 { 1020 struct ethercom *ec = (void *) ifp; 1021 struct ether_multi *enm; 1022 1023 /* 1024 * Prevent further calls to ioctl (for example turning off 1025 * promiscuous mode from the bridge code), which eventually can 1026 * call if_init() which can cause panics because the interface 1027 * is in the process of being detached. Return device not configured 1028 * instead. 1029 */ 1030 ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio; 1031 1032 #if NBRIDGE > 0 1033 if (ifp->if_bridge) 1034 bridge_ifdetach(ifp); 1035 #endif 1036 bpf_detach(ifp); 1037 #if NVLAN > 0 1038 if (ec->ec_nvlans) 1039 vlan_ifdetach(ifp); 1040 #endif 1041 1042 ETHER_LOCK(ec); 1043 while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) { 1044 LIST_REMOVE(enm, enm_list); 1045 kmem_free(enm, sizeof(*enm)); 1046 ec->ec_multicnt--; 1047 } 1048 ETHER_UNLOCK(ec); 1049 1050 mutex_obj_free(ec->ec_lock); 1051 ec->ec_lock = NULL; 1052 1053 ifp->if_mowner = NULL; 1054 MOWNER_DETACH(&ec->ec_rx_mowner); 1055 MOWNER_DETACH(&ec->ec_tx_mowner); 1056 } 1057 1058 #if 0 1059 /* 1060 * This is for reference. We have a table-driven version 1061 * of the little-endian crc32 generator, which is faster 1062 * than the double-loop. 1063 */ 1064 uint32_t 1065 ether_crc32_le(const uint8_t *buf, size_t len) 1066 { 1067 uint32_t c, crc, carry; 1068 size_t i, j; 1069 1070 crc = 0xffffffffU; /* initial value */ 1071 1072 for (i = 0; i < len; i++) { 1073 c = buf[i]; 1074 for (j = 0; j < 8; j++) { 1075 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1076 crc >>= 1; 1077 c >>= 1; 1078 if (carry) 1079 crc = (crc ^ ETHER_CRC_POLY_LE); 1080 } 1081 } 1082 1083 return (crc); 1084 } 1085 #else 1086 uint32_t 1087 ether_crc32_le(const uint8_t *buf, size_t len) 1088 { 1089 static const uint32_t crctab[] = { 1090 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1091 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1092 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1093 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1094 }; 1095 uint32_t crc; 1096 size_t i; 1097 1098 crc = 0xffffffffU; /* initial value */ 1099 1100 for (i = 0; i < len; i++) { 1101 crc ^= buf[i]; 1102 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1103 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1104 } 1105 1106 return (crc); 1107 } 1108 #endif 1109 1110 uint32_t 1111 ether_crc32_be(const uint8_t *buf, size_t len) 1112 { 1113 uint32_t c, crc, carry; 1114 size_t i, j; 1115 1116 crc = 0xffffffffU; /* initial value */ 1117 1118 for (i = 0; i < len; i++) { 1119 c = buf[i]; 1120 for (j = 0; j < 8; j++) { 1121 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1122 crc <<= 1; 1123 c >>= 1; 1124 if (carry) 1125 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1126 } 1127 } 1128 1129 return (crc); 1130 } 1131 1132 #ifdef INET 1133 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] = 1134 { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 }; 1135 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] = 1136 { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff }; 1137 #endif 1138 #ifdef INET6 1139 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] = 1140 { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 }; 1141 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] = 1142 { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff }; 1143 #endif 1144 1145 /* 1146 * ether_aton implementation, not using a static buffer. 1147 */ 1148 int 1149 ether_aton_r(u_char *dest, size_t len, const char *str) 1150 { 1151 const u_char *cp = (const void *)str; 1152 u_char *ep; 1153 1154 #define atox(c) (((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10)) 1155 1156 if (len < ETHER_ADDR_LEN) 1157 return ENOSPC; 1158 1159 ep = dest + ETHER_ADDR_LEN; 1160 1161 while (*cp) { 1162 if (!isxdigit(*cp)) 1163 return EINVAL; 1164 1165 *dest = atox(*cp); 1166 cp++; 1167 if (isxdigit(*cp)) { 1168 *dest = (*dest << 4) | atox(*cp); 1169 cp++; 1170 } 1171 dest++; 1172 1173 if (dest == ep) 1174 return (*cp == '\0') ? 0 : ENAMETOOLONG; 1175 1176 switch (*cp) { 1177 case ':': 1178 case '-': 1179 case '.': 1180 cp++; 1181 break; 1182 } 1183 } 1184 return ENOBUFS; 1185 } 1186 1187 /* 1188 * Convert a sockaddr into an Ethernet address or range of Ethernet 1189 * addresses. 1190 */ 1191 int 1192 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN], 1193 uint8_t addrhi[ETHER_ADDR_LEN]) 1194 { 1195 #ifdef INET 1196 const struct sockaddr_in *sin; 1197 #endif 1198 #ifdef INET6 1199 const struct sockaddr_in6 *sin6; 1200 #endif 1201 1202 switch (sa->sa_family) { 1203 1204 case AF_UNSPEC: 1205 memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN); 1206 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1207 break; 1208 1209 #ifdef INET 1210 case AF_INET: 1211 sin = satocsin(sa); 1212 if (sin->sin_addr.s_addr == INADDR_ANY) { 1213 /* 1214 * An IP address of INADDR_ANY means listen to 1215 * or stop listening to all of the Ethernet 1216 * multicast addresses used for IP. 1217 * (This is for the sake of IP multicast routers.) 1218 */ 1219 memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN); 1220 memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN); 1221 } else { 1222 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo); 1223 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1224 } 1225 break; 1226 #endif 1227 #ifdef INET6 1228 case AF_INET6: 1229 sin6 = satocsin6(sa); 1230 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1231 /* 1232 * An IP6 address of 0 means listen to or stop 1233 * listening to all of the Ethernet multicast 1234 * address used for IP6. 1235 * (This is used for multicast routers.) 1236 */ 1237 memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN); 1238 memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN); 1239 } else { 1240 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo); 1241 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1242 } 1243 break; 1244 #endif 1245 1246 default: 1247 return EAFNOSUPPORT; 1248 } 1249 return 0; 1250 } 1251 1252 /* 1253 * Add an Ethernet multicast address or range of addresses to the list for a 1254 * given interface. 1255 */ 1256 int 1257 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec) 1258 { 1259 struct ether_multi *enm, *_enm; 1260 u_char addrlo[ETHER_ADDR_LEN]; 1261 u_char addrhi[ETHER_ADDR_LEN]; 1262 int error = 0; 1263 1264 /* Allocate out of lock */ 1265 enm = kmem_alloc(sizeof(*enm), KM_SLEEP); 1266 1267 ETHER_LOCK(ec); 1268 error = ether_multiaddr(sa, addrlo, addrhi); 1269 if (error != 0) 1270 goto out; 1271 1272 /* 1273 * Verify that we have valid Ethernet multicast addresses. 1274 */ 1275 if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) { 1276 error = EINVAL; 1277 goto out; 1278 } 1279 1280 /* 1281 * See if the address range is already in the list. 1282 */ 1283 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, _enm); 1284 if (_enm != NULL) { 1285 /* 1286 * Found it; just increment the reference count. 1287 */ 1288 ++_enm->enm_refcount; 1289 error = 0; 1290 goto out; 1291 } 1292 1293 /* 1294 * Link a new multicast record into the interface's multicast list. 1295 */ 1296 memcpy(enm->enm_addrlo, addrlo, ETHER_ADDR_LEN); 1297 memcpy(enm->enm_addrhi, addrhi, ETHER_ADDR_LEN); 1298 enm->enm_refcount = 1; 1299 LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list); 1300 ec->ec_multicnt++; 1301 1302 /* 1303 * Return ENETRESET to inform the driver that the list has changed 1304 * and its reception filter should be adjusted accordingly. 1305 */ 1306 error = ENETRESET; 1307 enm = NULL; 1308 1309 out: 1310 ETHER_UNLOCK(ec); 1311 if (enm != NULL) 1312 kmem_free(enm, sizeof(*enm)); 1313 return error; 1314 } 1315 1316 /* 1317 * Delete a multicast address record. 1318 */ 1319 int 1320 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec) 1321 { 1322 struct ether_multi *enm; 1323 u_char addrlo[ETHER_ADDR_LEN]; 1324 u_char addrhi[ETHER_ADDR_LEN]; 1325 int error; 1326 1327 ETHER_LOCK(ec); 1328 error = ether_multiaddr(sa, addrlo, addrhi); 1329 if (error != 0) 1330 goto error; 1331 1332 /* 1333 * Look up the address in our list. 1334 */ 1335 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm); 1336 if (enm == NULL) { 1337 error = ENXIO; 1338 goto error; 1339 } 1340 if (--enm->enm_refcount != 0) { 1341 /* 1342 * Still some claims to this record. 1343 */ 1344 error = 0; 1345 goto error; 1346 } 1347 1348 /* 1349 * No remaining claims to this record; unlink and free it. 1350 */ 1351 LIST_REMOVE(enm, enm_list); 1352 ec->ec_multicnt--; 1353 ETHER_UNLOCK(ec); 1354 kmem_free(enm, sizeof(*enm)); 1355 1356 /* 1357 * Return ENETRESET to inform the driver that the list has changed 1358 * and its reception filter should be adjusted accordingly. 1359 */ 1360 return ENETRESET; 1361 1362 error: 1363 ETHER_UNLOCK(ec); 1364 return error; 1365 } 1366 1367 void 1368 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb) 1369 { 1370 ec->ec_ifflags_cb = cb; 1371 } 1372 1373 /* 1374 * Common ioctls for Ethernet interfaces. Note, we must be 1375 * called at splnet(). 1376 */ 1377 int 1378 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data) 1379 { 1380 struct ethercom *ec = (void *) ifp; 1381 struct eccapreq *eccr; 1382 struct ifreq *ifr = (struct ifreq *)data; 1383 struct if_laddrreq *iflr = data; 1384 const struct sockaddr_dl *sdl; 1385 static const uint8_t zero[ETHER_ADDR_LEN]; 1386 int error; 1387 1388 switch (cmd) { 1389 case SIOCINITIFADDR: 1390 { 1391 struct ifaddr *ifa = (struct ifaddr *)data; 1392 if (ifa->ifa_addr->sa_family != AF_LINK 1393 && (ifp->if_flags & (IFF_UP | IFF_RUNNING)) != 1394 (IFF_UP | IFF_RUNNING)) { 1395 ifp->if_flags |= IFF_UP; 1396 if ((error = (*ifp->if_init)(ifp)) != 0) 1397 return error; 1398 } 1399 #ifdef INET 1400 if (ifa->ifa_addr->sa_family == AF_INET) 1401 arp_ifinit(ifp, ifa); 1402 #endif 1403 return 0; 1404 } 1405 1406 case SIOCSIFMTU: 1407 { 1408 int maxmtu; 1409 1410 if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU) 1411 maxmtu = ETHERMTU_JUMBO; 1412 else 1413 maxmtu = ETHERMTU; 1414 1415 if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu) 1416 return EINVAL; 1417 else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET) 1418 return error; 1419 else if (ifp->if_flags & IFF_UP) { 1420 /* Make sure the device notices the MTU change. */ 1421 return (*ifp->if_init)(ifp); 1422 } else 1423 return 0; 1424 } 1425 1426 case SIOCSIFFLAGS: 1427 if ((error = ifioctl_common(ifp, cmd, data)) != 0) 1428 return error; 1429 switch (ifp->if_flags & (IFF_UP | IFF_RUNNING)) { 1430 case IFF_RUNNING: 1431 /* 1432 * If interface is marked down and it is running, 1433 * then stop and disable it. 1434 */ 1435 (*ifp->if_stop)(ifp, 1); 1436 break; 1437 case IFF_UP: 1438 /* 1439 * If interface is marked up and it is stopped, then 1440 * start it. 1441 */ 1442 return (*ifp->if_init)(ifp); 1443 case IFF_UP | IFF_RUNNING: 1444 error = 0; 1445 if (ec->ec_ifflags_cb != NULL) { 1446 error = (*ec->ec_ifflags_cb)(ec); 1447 if (error == ENETRESET) { 1448 /* 1449 * Reset the interface to pick up 1450 * changes in any other flags that 1451 * affect the hardware state. 1452 */ 1453 return (*ifp->if_init)(ifp); 1454 } 1455 } else 1456 error = (*ifp->if_init)(ifp); 1457 return error; 1458 case 0: 1459 break; 1460 } 1461 return 0; 1462 case SIOCGETHERCAP: 1463 eccr = (struct eccapreq *)data; 1464 eccr->eccr_capabilities = ec->ec_capabilities; 1465 eccr->eccr_capenable = ec->ec_capenable; 1466 return 0; 1467 case SIOCADDMULTI: 1468 return ether_addmulti(ifreq_getaddr(cmd, ifr), ec); 1469 case SIOCDELMULTI: 1470 return ether_delmulti(ifreq_getaddr(cmd, ifr), ec); 1471 case SIOCSIFMEDIA: 1472 case SIOCGIFMEDIA: 1473 if (ec->ec_mii == NULL) 1474 return ENOTTY; 1475 return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd); 1476 case SIOCALIFADDR: 1477 sdl = satocsdl(sstocsa(&iflr->addr)); 1478 if (sdl->sdl_family != AF_LINK) 1479 ; 1480 else if (ETHER_IS_MULTICAST(CLLADDR(sdl))) 1481 return EINVAL; 1482 else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0) 1483 return EINVAL; 1484 /*FALLTHROUGH*/ 1485 default: 1486 return ifioctl_common(ifp, cmd, data); 1487 } 1488 return 0; 1489 } 1490 1491 /* 1492 * Enable/disable passing VLAN packets if the parent interface supports it. 1493 * Return: 1494 * 0: Ok 1495 * -1: Parent interface does not support vlans 1496 * >0: Error 1497 */ 1498 int 1499 ether_enable_vlan_mtu(struct ifnet *ifp) 1500 { 1501 int error; 1502 struct ethercom *ec = (void *)ifp; 1503 1504 /* Parent does not support VLAN's */ 1505 if ((ec->ec_capabilities & ETHERCAP_VLAN_MTU) == 0) 1506 return -1; 1507 1508 /* 1509 * Parent supports the VLAN_MTU capability, 1510 * i.e. can Tx/Rx larger than ETHER_MAX_LEN frames; 1511 * enable it. 1512 */ 1513 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1514 1515 /* Interface is down, defer for later */ 1516 if ((ifp->if_flags & IFF_UP) == 0) 1517 return 0; 1518 1519 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1520 return 0; 1521 1522 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1523 return error; 1524 } 1525 1526 int 1527 ether_disable_vlan_mtu(struct ifnet *ifp) 1528 { 1529 int error; 1530 struct ethercom *ec = (void *)ifp; 1531 1532 /* We still have VLAN's, defer for later */ 1533 if (ec->ec_nvlans != 0) 1534 return 0; 1535 1536 /* Parent does not support VLAB's, nothing to do. */ 1537 if ((ec->ec_capenable & ETHERCAP_VLAN_MTU) == 0) 1538 return -1; 1539 1540 /* 1541 * Disable Tx/Rx of VLAN-sized frames. 1542 */ 1543 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1544 1545 /* Interface is down, defer for later */ 1546 if ((ifp->if_flags & IFF_UP) == 0) 1547 return 0; 1548 1549 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1550 return 0; 1551 1552 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1553 return error; 1554 } 1555 1556 static int 1557 ether_multicast_sysctl(SYSCTLFN_ARGS) 1558 { 1559 struct ether_multi *enm; 1560 struct ifnet *ifp; 1561 struct ethercom *ec; 1562 int error = 0; 1563 size_t written; 1564 struct psref psref; 1565 int bound; 1566 unsigned int multicnt; 1567 struct ether_multi_sysctl *addrs; 1568 int i; 1569 1570 if (namelen != 1) 1571 return EINVAL; 1572 1573 bound = curlwp_bind(); 1574 ifp = if_get_byindex(name[0], &psref); 1575 if (ifp == NULL) { 1576 error = ENODEV; 1577 goto out; 1578 } 1579 if (ifp->if_type != IFT_ETHER) { 1580 if_put(ifp, &psref); 1581 *oldlenp = 0; 1582 goto out; 1583 } 1584 ec = (struct ethercom *)ifp; 1585 1586 if (oldp == NULL) { 1587 if_put(ifp, &psref); 1588 *oldlenp = ec->ec_multicnt * sizeof(*addrs); 1589 goto out; 1590 } 1591 1592 /* 1593 * ec->ec_lock is a spin mutex so we cannot call sysctl_copyout, which 1594 * is sleepable, while holding it. Copy data to a local buffer first 1595 * with the lock taken and then call sysctl_copyout without holding it. 1596 */ 1597 retry: 1598 multicnt = ec->ec_multicnt; 1599 1600 if (multicnt == 0) { 1601 if_put(ifp, &psref); 1602 *oldlenp = 0; 1603 goto out; 1604 } 1605 1606 addrs = kmem_zalloc(sizeof(*addrs) * multicnt, KM_SLEEP); 1607 1608 ETHER_LOCK(ec); 1609 if (multicnt != ec->ec_multicnt) { 1610 /* The number of multicast addresses has changed */ 1611 ETHER_UNLOCK(ec); 1612 kmem_free(addrs, sizeof(*addrs) * multicnt); 1613 goto retry; 1614 } 1615 1616 i = 0; 1617 LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) { 1618 struct ether_multi_sysctl *addr = &addrs[i]; 1619 addr->enm_refcount = enm->enm_refcount; 1620 memcpy(addr->enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN); 1621 memcpy(addr->enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN); 1622 i++; 1623 } 1624 ETHER_UNLOCK(ec); 1625 1626 error = 0; 1627 written = 0; 1628 for (i = 0; i < multicnt; i++) { 1629 struct ether_multi_sysctl *addr = &addrs[i]; 1630 1631 if (written + sizeof(*addr) > *oldlenp) 1632 break; 1633 error = sysctl_copyout(l, addr, oldp, sizeof(*addr)); 1634 if (error) 1635 break; 1636 written += sizeof(*addr); 1637 oldp = (char *)oldp + sizeof(*addr); 1638 } 1639 kmem_free(addrs, sizeof(*addrs) * multicnt); 1640 1641 if_put(ifp, &psref); 1642 1643 *oldlenp = written; 1644 out: 1645 curlwp_bindx(bound); 1646 return error; 1647 } 1648 1649 static void 1650 ether_sysctl_setup(struct sysctllog **clog) 1651 { 1652 const struct sysctlnode *rnode = NULL; 1653 1654 sysctl_createv(clog, 0, NULL, &rnode, 1655 CTLFLAG_PERMANENT, 1656 CTLTYPE_NODE, "ether", 1657 SYSCTL_DESCR("Ethernet-specific information"), 1658 NULL, 0, NULL, 0, 1659 CTL_NET, CTL_CREATE, CTL_EOL); 1660 1661 sysctl_createv(clog, 0, &rnode, NULL, 1662 CTLFLAG_PERMANENT, 1663 CTLTYPE_NODE, "multicast", 1664 SYSCTL_DESCR("multicast addresses"), 1665 ether_multicast_sysctl, 0, NULL, 0, 1666 CTL_CREATE, CTL_EOL); 1667 } 1668 1669 void 1670 etherinit(void) 1671 { 1672 1673 mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET); 1674 ether_sysctl_setup(NULL); 1675 } 1676
Indexes created Sun Sep 21 19:09:51 GMT 2025