if_ethersubr.c revision 1.277
1 /* $NetBSD: if_ethersubr.c,v 1.277 2019/10/01 08:13:16 msaitoh 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 1.277 2019/10/01 08:13:16 msaitoh 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_copypacket(m, M_DONTWAIT); 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 KERNEL_UNLOCK_ONE(NULL); 311 return 0; 312 } 313 314 /* 315 * ifaddr is the first thing in at_ifaddr 316 */ 317 s = pserialize_read_enter(); 318 ifa = at_ifawithnet((const struct sockaddr_at *)dst, ifp); 319 if (ifa == NULL) { 320 pserialize_read_exit(s); 321 KERNEL_UNLOCK_ONE(NULL); 322 senderr(EADDRNOTAVAIL); 323 } 324 aa = (struct at_ifaddr *)ifa; 325 326 /* 327 * In the phase 2 case, we need to prepend an mbuf for the 328 * llc header. 329 */ 330 if (aa->aa_flags & AFA_PHASE2) { 331 struct llc llc; 332 333 M_PREPEND(m, sizeof(struct llc), M_DONTWAIT); 334 if (m == NULL) { 335 pserialize_read_exit(s); 336 KERNEL_UNLOCK_ONE(NULL); 337 senderr(ENOBUFS); 338 } 339 340 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 341 llc.llc_control = LLC_UI; 342 memcpy(llc.llc_snap_org_code, at_org_code, 343 sizeof(llc.llc_snap_org_code)); 344 llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK); 345 memcpy(mtod(m, void *), &llc, sizeof(struct llc)); 346 } else { 347 etype = htons(ETHERTYPE_ATALK); 348 } 349 pserialize_read_exit(s); 350 KERNEL_UNLOCK_ONE(NULL); 351 break; 352 } 353 #endif /* NETATALK */ 354 355 case pseudo_AF_HDRCMPLT: 356 hdrcmplt = 1; 357 memcpy(esrc, 358 ((const struct ether_header *)dst->sa_data)->ether_shost, 359 sizeof(esrc)); 360 /* FALLTHROUGH */ 361 362 case AF_UNSPEC: 363 memcpy(edst, 364 ((const struct ether_header *)dst->sa_data)->ether_dhost, 365 sizeof(edst)); 366 /* AF_UNSPEC doesn't swap the byte order of the ether_type. */ 367 etype = ((const struct ether_header *)dst->sa_data)->ether_type; 368 break; 369 370 default: 371 printf("%s: can't handle af%d\n", ifp->if_xname, 372 dst->sa_family); 373 senderr(EAFNOSUPPORT); 374 } 375 376 #ifdef MPLS 377 { 378 struct m_tag *mtag; 379 mtag = m_tag_find(m, PACKET_TAG_MPLS); 380 if (mtag != NULL) { 381 /* Having the tag itself indicates it's MPLS */ 382 etype = htons(ETHERTYPE_MPLS); 383 m_tag_delete(m, mtag); 384 } 385 } 386 #endif 387 388 if (mcopy) 389 (void)looutput(ifp, mcopy, dst, rt); 390 391 KASSERT((m->m_flags & M_PKTHDR) != 0); 392 393 /* 394 * If no ether type is set, this must be a 802.2 formatted packet. 395 */ 396 if (etype == 0) 397 etype = htons(m->m_pkthdr.len); 398 399 /* 400 * Add local net header. If no space in first mbuf, allocate another. 401 */ 402 M_PREPEND(m, sizeof(struct ether_header), M_DONTWAIT); 403 if (m == NULL) 404 senderr(ENOBUFS); 405 406 eh = mtod(m, struct ether_header *); 407 /* Note: etype is already in network byte order. */ 408 memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type)); 409 memcpy(eh->ether_dhost, edst, sizeof(edst)); 410 if (hdrcmplt) { 411 memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost)); 412 } else { 413 memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl), 414 sizeof(eh->ether_shost)); 415 } 416 417 #if NCARP > 0 418 if (ifp0 != ifp && ifp0->if_type == IFT_CARP) { 419 memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl), 420 sizeof(eh->ether_shost)); 421 } 422 #endif 423 424 if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0) 425 return error; 426 if (m == NULL) 427 return 0; 428 429 #if NBRIDGE > 0 430 /* 431 * Bridges require special output handling. 432 */ 433 if (ifp->if_bridge) 434 return bridge_output(ifp, m, NULL, NULL); 435 #endif 436 437 #if NCARP > 0 438 if (ifp != ifp0) 439 ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN; 440 #endif 441 442 #ifdef ALTQ 443 KERNEL_LOCK(1, NULL); 444 /* 445 * If ALTQ is enabled on the parent interface, do 446 * classification; the queueing discipline might not 447 * require classification, but might require the 448 * address family/header pointer in the pktattr. 449 */ 450 if (ALTQ_IS_ENABLED(&ifp->if_snd)) 451 altq_etherclassify(&ifp->if_snd, m); 452 KERNEL_UNLOCK_ONE(NULL); 453 #endif 454 return ifq_enqueue(ifp, m); 455 456 bad: 457 if (m) 458 m_freem(m); 459 return error; 460 } 461 462 #ifdef ALTQ 463 /* 464 * This routine is a slight hack to allow a packet to be classified 465 * if the Ethernet headers are present. It will go away when ALTQ's 466 * classification engine understands link headers. 467 * 468 * XXX: We may need to do m_pullups here. First to ensure struct ether_header 469 * is indeed contiguous, then to read the LLC and so on. 470 */ 471 void 472 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m) 473 { 474 struct ether_header *eh; 475 struct mbuf *mtop = m; 476 uint16_t ether_type; 477 int hlen, af, hdrsize; 478 void *hdr; 479 480 KASSERT((mtop->m_flags & M_PKTHDR) != 0); 481 482 hlen = ETHER_HDR_LEN; 483 eh = mtod(m, struct ether_header *); 484 485 ether_type = htons(eh->ether_type); 486 487 if (ether_type < ETHERMTU) { 488 /* LLC/SNAP */ 489 struct llc *llc = (struct llc *)(eh + 1); 490 hlen += 8; 491 492 if (m->m_len < hlen || 493 llc->llc_dsap != LLC_SNAP_LSAP || 494 llc->llc_ssap != LLC_SNAP_LSAP || 495 llc->llc_control != LLC_UI) { 496 /* Not SNAP. */ 497 goto bad; 498 } 499 500 ether_type = htons(llc->llc_un.type_snap.ether_type); 501 } 502 503 switch (ether_type) { 504 case ETHERTYPE_IP: 505 af = AF_INET; 506 hdrsize = 20; /* sizeof(struct ip) */ 507 break; 508 509 case ETHERTYPE_IPV6: 510 af = AF_INET6; 511 hdrsize = 40; /* sizeof(struct ip6_hdr) */ 512 break; 513 514 default: 515 af = AF_UNSPEC; 516 hdrsize = 0; 517 break; 518 } 519 520 while (m->m_len <= hlen) { 521 hlen -= m->m_len; 522 m = m->m_next; 523 if (m == NULL) 524 goto bad; 525 } 526 527 if (m->m_len < (hlen + hdrsize)) { 528 /* 529 * protocol header not in a single mbuf. 530 * We can't cope with this situation right 531 * now (but it shouldn't ever happen, really, anyhow). 532 */ 533 #ifdef DEBUG 534 printf("altq_etherclassify: headers span multiple mbufs: " 535 "%d < %d\n", m->m_len, (hlen + hdrsize)); 536 #endif 537 goto bad; 538 } 539 540 m->m_data += hlen; 541 m->m_len -= hlen; 542 543 hdr = mtod(m, void *); 544 545 if (ALTQ_NEEDS_CLASSIFY(ifq)) { 546 mtop->m_pkthdr.pattr_class = 547 (*ifq->altq_classify)(ifq->altq_clfier, m, af); 548 } 549 mtop->m_pkthdr.pattr_af = af; 550 mtop->m_pkthdr.pattr_hdr = hdr; 551 552 m->m_data -= hlen; 553 m->m_len += hlen; 554 555 return; 556 557 bad: 558 mtop->m_pkthdr.pattr_class = NULL; 559 mtop->m_pkthdr.pattr_hdr = NULL; 560 mtop->m_pkthdr.pattr_af = AF_UNSPEC; 561 } 562 #endif /* ALTQ */ 563 564 /* 565 * Process a received Ethernet packet; 566 * the packet is in the mbuf chain m with 567 * the ether header. 568 */ 569 void 570 ether_input(struct ifnet *ifp, struct mbuf *m) 571 { 572 struct ethercom *ec = (struct ethercom *) ifp; 573 pktqueue_t *pktq = NULL; 574 struct ifqueue *inq = NULL; 575 uint16_t etype; 576 struct ether_header *eh; 577 size_t ehlen; 578 static int earlypkts; 579 int isr = 0; 580 #if defined (LLC) || defined(NETATALK) 581 struct llc *l; 582 #endif 583 584 KASSERT(!cpu_intr_p()); 585 KASSERT((m->m_flags & M_PKTHDR) != 0); 586 587 if ((ifp->if_flags & IFF_UP) == 0) { 588 m_freem(m); 589 return; 590 } 591 if (m->m_len < sizeof(*eh)) { 592 m = m_pullup(m, sizeof(*eh)); 593 if (m == NULL) 594 return; 595 } 596 597 #ifdef MBUFTRACE 598 m_claimm(m, &ec->ec_rx_mowner); 599 #endif 600 eh = mtod(m, struct ether_header *); 601 etype = ntohs(eh->ether_type); 602 ehlen = sizeof(*eh); 603 604 if (__predict_false(earlypkts < 100 || !rnd_initial_entropy)) { 605 rnd_add_data(NULL, eh, ehlen, 0); 606 earlypkts++; 607 } 608 609 /* 610 * Determine if the packet is within its size limits. For MPLS the 611 * header length is variable, so we skip the check. 612 */ 613 if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len > 614 ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) { 615 mutex_enter(&bigpktpps_lock); 616 if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count, 617 bigpktppslim)) { 618 printf("%s: discarding oversize frame (len=%d)\n", 619 ifp->if_xname, m->m_pkthdr.len); 620 } 621 mutex_exit(&bigpktpps_lock); 622 ifp->if_iqdrops++; 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 if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) { 763 m_freem(m); 764 return; 765 } 766 767 m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype); 768 switch (subtype) { 769 #if NAGR > 0 770 case SLOWPROTOCOLS_SUBTYPE_LACP: 771 if (ifp->if_agrprivate) { 772 ieee8023ad_lacp_input(ifp, m); 773 return; 774 } 775 break; 776 777 case SLOWPROTOCOLS_SUBTYPE_MARKER: 778 if (ifp->if_agrprivate) { 779 ieee8023ad_marker_input(ifp, m); 780 return; 781 } 782 break; 783 #endif 784 785 default: 786 if (subtype == 0 || subtype > 10) { 787 /* illegal value */ 788 m_freem(m); 789 return; 790 } 791 /* unknown subtype */ 792 break; 793 } 794 } 795 /* FALLTHROUGH */ 796 default: 797 if (m->m_flags & M_PROMISC) { 798 m_freem(m); 799 return; 800 } 801 } 802 803 /* If the CRC is still on the packet, trim it off. */ 804 if (m->m_flags & M_HASFCS) { 805 m_adj(m, -ETHER_CRC_LEN); 806 m->m_flags &= ~M_HASFCS; 807 } 808 809 if (etype > ETHERMTU + sizeof(struct ether_header)) { 810 /* Strip off the Ethernet header. */ 811 m_adj(m, ehlen); 812 813 switch (etype) { 814 #ifdef INET 815 case ETHERTYPE_IP: 816 #ifdef GATEWAY 817 if (ipflow_fastforward(m)) 818 return; 819 #endif 820 pktq = ip_pktq; 821 break; 822 823 case ETHERTYPE_ARP: 824 isr = NETISR_ARP; 825 inq = &arpintrq; 826 break; 827 828 case ETHERTYPE_REVARP: 829 revarpinput(m); /* XXX queue? */ 830 return; 831 #endif 832 833 #ifdef INET6 834 case ETHERTYPE_IPV6: 835 if (__predict_false(!in6_present)) { 836 m_freem(m); 837 return; 838 } 839 #ifdef GATEWAY 840 if (ip6flow_fastforward(&m)) 841 return; 842 #endif 843 pktq = ip6_pktq; 844 break; 845 #endif 846 847 #ifdef NETATALK 848 case ETHERTYPE_ATALK: 849 isr = NETISR_ATALK; 850 inq = &atintrq1; 851 break; 852 853 case ETHERTYPE_AARP: 854 aarpinput(ifp, m); /* XXX queue? */ 855 return; 856 #endif 857 858 #ifdef MPLS 859 case ETHERTYPE_MPLS: 860 isr = NETISR_MPLS; 861 inq = &mplsintrq; 862 break; 863 #endif 864 865 default: 866 m_freem(m); 867 return; 868 } 869 } else { 870 KASSERT(ehlen == sizeof(*eh)); 871 #if defined (LLC) || defined (NETATALK) 872 if (m->m_len < sizeof(*eh) + sizeof(struct llc)) { 873 goto dropanyway; 874 } 875 l = (struct llc *)(eh+1); 876 877 switch (l->llc_dsap) { 878 #ifdef NETATALK 879 case LLC_SNAP_LSAP: 880 switch (l->llc_control) { 881 case LLC_UI: 882 if (l->llc_ssap != LLC_SNAP_LSAP) { 883 goto dropanyway; 884 } 885 886 if (memcmp(&(l->llc_snap_org_code)[0], 887 at_org_code, sizeof(at_org_code)) == 0 && 888 ntohs(l->llc_snap_ether_type) == 889 ETHERTYPE_ATALK) { 890 inq = &atintrq2; 891 m_adj(m, sizeof(struct ether_header) 892 + sizeof(struct llc)); 893 isr = NETISR_ATALK; 894 break; 895 } 896 897 if (memcmp(&(l->llc_snap_org_code)[0], 898 aarp_org_code, 899 sizeof(aarp_org_code)) == 0 && 900 ntohs(l->llc_snap_ether_type) == 901 ETHERTYPE_AARP) { 902 m_adj(m, sizeof(struct ether_header) 903 + sizeof(struct llc)); 904 aarpinput(ifp, m); /* XXX queue? */ 905 return; 906 } 907 908 default: 909 goto dropanyway; 910 } 911 break; 912 #endif 913 dropanyway: 914 default: 915 m_freem(m); 916 return; 917 } 918 #else /* LLC || NETATALK */ 919 m_freem(m); 920 return; 921 #endif /* LLC || NETATALK */ 922 } 923 924 if (__predict_true(pktq)) { 925 #ifdef NET_MPSAFE 926 const u_int h = curcpu()->ci_index; 927 #else 928 const uint32_t h = pktq_rps_hash(m); 929 #endif 930 if (__predict_false(!pktq_enqueue(pktq, m, h))) { 931 m_freem(m); 932 } 933 return; 934 } 935 936 if (__predict_false(!inq)) { 937 /* Should not happen. */ 938 m_freem(m); 939 return; 940 } 941 942 IFQ_LOCK(inq); 943 if (IF_QFULL(inq)) { 944 IF_DROP(inq); 945 IFQ_UNLOCK(inq); 946 m_freem(m); 947 } else { 948 IF_ENQUEUE(inq, m); 949 IFQ_UNLOCK(inq); 950 schednetisr(isr); 951 } 952 } 953 954 /* 955 * Convert Ethernet address to printable (loggable) representation. 956 */ 957 char * 958 ether_sprintf(const u_char *ap) 959 { 960 static char etherbuf[3 * ETHER_ADDR_LEN]; 961 return ether_snprintf(etherbuf, sizeof(etherbuf), ap); 962 } 963 964 char * 965 ether_snprintf(char *buf, size_t len, const u_char *ap) 966 { 967 char *cp = buf; 968 size_t i; 969 970 for (i = 0; i < len / 3; i++) { 971 *cp++ = hexdigits[*ap >> 4]; 972 *cp++ = hexdigits[*ap++ & 0xf]; 973 *cp++ = ':'; 974 } 975 *--cp = '\0'; 976 return buf; 977 } 978 979 /* 980 * Perform common duties while attaching to interface list 981 */ 982 void 983 ether_ifattach(struct ifnet *ifp, const uint8_t *lla) 984 { 985 struct ethercom *ec = (struct ethercom *)ifp; 986 987 ifp->if_type = IFT_ETHER; 988 ifp->if_hdrlen = ETHER_HDR_LEN; 989 ifp->if_dlt = DLT_EN10MB; 990 ifp->if_mtu = ETHERMTU; 991 ifp->if_output = ether_output; 992 ifp->_if_input = ether_input; 993 if (ifp->if_baudrate == 0) 994 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 995 996 if (lla != NULL) 997 if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla)); 998 999 LIST_INIT(&ec->ec_multiaddrs); 1000 SIMPLEQ_INIT(&ec->ec_vids); 1001 ec->ec_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NET); 1002 ec->ec_flags = 0; 1003 ifp->if_broadcastaddr = etherbroadcastaddr; 1004 bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header)); 1005 #ifdef MBUFTRACE 1006 strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname, 1007 sizeof(ec->ec_tx_mowner.mo_name)); 1008 strlcpy(ec->ec_tx_mowner.mo_descr, "tx", 1009 sizeof(ec->ec_tx_mowner.mo_descr)); 1010 strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname, 1011 sizeof(ec->ec_rx_mowner.mo_name)); 1012 strlcpy(ec->ec_rx_mowner.mo_descr, "rx", 1013 sizeof(ec->ec_rx_mowner.mo_descr)); 1014 MOWNER_ATTACH(&ec->ec_tx_mowner); 1015 MOWNER_ATTACH(&ec->ec_rx_mowner); 1016 ifp->if_mowner = &ec->ec_tx_mowner; 1017 #endif 1018 } 1019 1020 void 1021 ether_ifdetach(struct ifnet *ifp) 1022 { 1023 struct ethercom *ec = (void *) ifp; 1024 struct ether_multi *enm; 1025 1026 IFNET_ASSERT_UNLOCKED(ifp); 1027 /* 1028 * Prevent further calls to ioctl (for example turning off 1029 * promiscuous mode from the bridge code), which eventually can 1030 * call if_init() which can cause panics because the interface 1031 * is in the process of being detached. Return device not configured 1032 * instead. 1033 */ 1034 ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio; 1035 1036 #if NBRIDGE > 0 1037 if (ifp->if_bridge) 1038 bridge_ifdetach(ifp); 1039 #endif 1040 bpf_detach(ifp); 1041 #if NVLAN > 0 1042 if (ec->ec_nvlans) 1043 vlan_ifdetach(ifp); 1044 #endif 1045 1046 ETHER_LOCK(ec); 1047 KASSERT(ec->ec_nvlans == 0); 1048 while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) { 1049 LIST_REMOVE(enm, enm_list); 1050 kmem_free(enm, sizeof(*enm)); 1051 ec->ec_multicnt--; 1052 } 1053 ETHER_UNLOCK(ec); 1054 1055 mutex_obj_free(ec->ec_lock); 1056 ec->ec_lock = NULL; 1057 1058 ifp->if_mowner = NULL; 1059 MOWNER_DETACH(&ec->ec_rx_mowner); 1060 MOWNER_DETACH(&ec->ec_tx_mowner); 1061 } 1062 1063 #if 0 1064 /* 1065 * This is for reference. We have a table-driven version 1066 * of the little-endian crc32 generator, which is faster 1067 * than the double-loop. 1068 */ 1069 uint32_t 1070 ether_crc32_le(const uint8_t *buf, size_t len) 1071 { 1072 uint32_t c, crc, carry; 1073 size_t i, j; 1074 1075 crc = 0xffffffffU; /* initial value */ 1076 1077 for (i = 0; i < len; i++) { 1078 c = buf[i]; 1079 for (j = 0; j < 8; j++) { 1080 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1081 crc >>= 1; 1082 c >>= 1; 1083 if (carry) 1084 crc = (crc ^ ETHER_CRC_POLY_LE); 1085 } 1086 } 1087 1088 return (crc); 1089 } 1090 #else 1091 uint32_t 1092 ether_crc32_le(const uint8_t *buf, size_t len) 1093 { 1094 static const uint32_t crctab[] = { 1095 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1096 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1097 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1098 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1099 }; 1100 uint32_t crc; 1101 size_t i; 1102 1103 crc = 0xffffffffU; /* initial value */ 1104 1105 for (i = 0; i < len; i++) { 1106 crc ^= buf[i]; 1107 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1108 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1109 } 1110 1111 return (crc); 1112 } 1113 #endif 1114 1115 uint32_t 1116 ether_crc32_be(const uint8_t *buf, size_t len) 1117 { 1118 uint32_t c, crc, carry; 1119 size_t i, j; 1120 1121 crc = 0xffffffffU; /* initial value */ 1122 1123 for (i = 0; i < len; i++) { 1124 c = buf[i]; 1125 for (j = 0; j < 8; j++) { 1126 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1127 crc <<= 1; 1128 c >>= 1; 1129 if (carry) 1130 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1131 } 1132 } 1133 1134 return (crc); 1135 } 1136 1137 #ifdef INET 1138 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] = 1139 { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 }; 1140 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] = 1141 { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff }; 1142 #endif 1143 #ifdef INET6 1144 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] = 1145 { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 }; 1146 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] = 1147 { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff }; 1148 #endif 1149 1150 /* 1151 * ether_aton implementation, not using a static buffer. 1152 */ 1153 int 1154 ether_aton_r(u_char *dest, size_t len, const char *str) 1155 { 1156 const u_char *cp = (const void *)str; 1157 u_char *ep; 1158 1159 #define atox(c) (((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10)) 1160 1161 if (len < ETHER_ADDR_LEN) 1162 return ENOSPC; 1163 1164 ep = dest + ETHER_ADDR_LEN; 1165 1166 while (*cp) { 1167 if (!isxdigit(*cp)) 1168 return EINVAL; 1169 1170 *dest = atox(*cp); 1171 cp++; 1172 if (isxdigit(*cp)) { 1173 *dest = (*dest << 4) | atox(*cp); 1174 cp++; 1175 } 1176 dest++; 1177 1178 if (dest == ep) 1179 return (*cp == '\0') ? 0 : ENAMETOOLONG; 1180 1181 switch (*cp) { 1182 case ':': 1183 case '-': 1184 case '.': 1185 cp++; 1186 break; 1187 } 1188 } 1189 return ENOBUFS; 1190 } 1191 1192 /* 1193 * Convert a sockaddr into an Ethernet address or range of Ethernet 1194 * addresses. 1195 */ 1196 int 1197 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN], 1198 uint8_t addrhi[ETHER_ADDR_LEN]) 1199 { 1200 #ifdef INET 1201 const struct sockaddr_in *sin; 1202 #endif 1203 #ifdef INET6 1204 const struct sockaddr_in6 *sin6; 1205 #endif 1206 1207 switch (sa->sa_family) { 1208 1209 case AF_UNSPEC: 1210 memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN); 1211 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1212 break; 1213 1214 #ifdef INET 1215 case AF_INET: 1216 sin = satocsin(sa); 1217 if (sin->sin_addr.s_addr == INADDR_ANY) { 1218 /* 1219 * An IP address of INADDR_ANY means listen to 1220 * or stop listening to all of the Ethernet 1221 * multicast addresses used for IP. 1222 * (This is for the sake of IP multicast routers.) 1223 */ 1224 memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN); 1225 memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN); 1226 } else { 1227 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo); 1228 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1229 } 1230 break; 1231 #endif 1232 #ifdef INET6 1233 case AF_INET6: 1234 sin6 = satocsin6(sa); 1235 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1236 /* 1237 * An IP6 address of 0 means listen to or stop 1238 * listening to all of the Ethernet multicast 1239 * address used for IP6. 1240 * (This is used for multicast routers.) 1241 */ 1242 memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN); 1243 memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN); 1244 } else { 1245 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo); 1246 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1247 } 1248 break; 1249 #endif 1250 1251 default: 1252 return EAFNOSUPPORT; 1253 } 1254 return 0; 1255 } 1256 1257 /* 1258 * Add an Ethernet multicast address or range of addresses to the list for a 1259 * given interface. 1260 */ 1261 int 1262 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec) 1263 { 1264 struct ether_multi *enm, *_enm; 1265 u_char addrlo[ETHER_ADDR_LEN]; 1266 u_char addrhi[ETHER_ADDR_LEN]; 1267 int error = 0; 1268 1269 /* Allocate out of lock */ 1270 enm = kmem_alloc(sizeof(*enm), KM_SLEEP); 1271 1272 ETHER_LOCK(ec); 1273 error = ether_multiaddr(sa, addrlo, addrhi); 1274 if (error != 0) 1275 goto out; 1276 1277 /* 1278 * Verify that we have valid Ethernet multicast addresses. 1279 */ 1280 if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) { 1281 error = EINVAL; 1282 goto out; 1283 } 1284 1285 /* 1286 * See if the address range is already in the list. 1287 */ 1288 _enm = ether_lookup_multi(addrlo, addrhi, ec); 1289 if (_enm != NULL) { 1290 /* 1291 * Found it; just increment the reference count. 1292 */ 1293 ++_enm->enm_refcount; 1294 error = 0; 1295 goto out; 1296 } 1297 1298 /* 1299 * Link a new multicast record into the interface's multicast list. 1300 */ 1301 memcpy(enm->enm_addrlo, addrlo, ETHER_ADDR_LEN); 1302 memcpy(enm->enm_addrhi, addrhi, ETHER_ADDR_LEN); 1303 enm->enm_refcount = 1; 1304 LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list); 1305 ec->ec_multicnt++; 1306 1307 /* 1308 * Return ENETRESET to inform the driver that the list has changed 1309 * and its reception filter should be adjusted accordingly. 1310 */ 1311 error = ENETRESET; 1312 enm = NULL; 1313 1314 out: 1315 ETHER_UNLOCK(ec); 1316 if (enm != NULL) 1317 kmem_free(enm, sizeof(*enm)); 1318 return error; 1319 } 1320 1321 /* 1322 * Delete a multicast address record. 1323 */ 1324 int 1325 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec) 1326 { 1327 struct ether_multi *enm; 1328 u_char addrlo[ETHER_ADDR_LEN]; 1329 u_char addrhi[ETHER_ADDR_LEN]; 1330 int error; 1331 1332 ETHER_LOCK(ec); 1333 error = ether_multiaddr(sa, addrlo, addrhi); 1334 if (error != 0) 1335 goto error; 1336 1337 /* 1338 * Look up the address in our list. 1339 */ 1340 enm = ether_lookup_multi(addrlo, addrhi, ec); 1341 if (enm == NULL) { 1342 error = ENXIO; 1343 goto error; 1344 } 1345 if (--enm->enm_refcount != 0) { 1346 /* 1347 * Still some claims to this record. 1348 */ 1349 error = 0; 1350 goto error; 1351 } 1352 1353 /* 1354 * No remaining claims to this record; unlink and free it. 1355 */ 1356 LIST_REMOVE(enm, enm_list); 1357 ec->ec_multicnt--; 1358 ETHER_UNLOCK(ec); 1359 kmem_free(enm, sizeof(*enm)); 1360 1361 /* 1362 * Return ENETRESET to inform the driver that the list has changed 1363 * and its reception filter should be adjusted accordingly. 1364 */ 1365 return ENETRESET; 1366 1367 error: 1368 ETHER_UNLOCK(ec); 1369 return error; 1370 } 1371 1372 void 1373 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb) 1374 { 1375 ec->ec_ifflags_cb = cb; 1376 } 1377 1378 void 1379 ether_set_vlan_cb(struct ethercom *ec, ether_vlancb_t cb) 1380 { 1381 1382 ec->ec_vlan_cb = cb; 1383 } 1384 1385 static int 1386 ether_ioctl_reinit(struct ethercom *ec) 1387 { 1388 struct ifnet *ifp = &ec->ec_if; 1389 int error; 1390 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 1424 return 0; 1425 } 1426 1427 /* 1428 * Common ioctls for Ethernet interfaces. Note, we must be 1429 * called at splnet(). 1430 */ 1431 int 1432 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data) 1433 { 1434 struct ethercom *ec = (void *)ifp; 1435 struct eccapreq *eccr; 1436 struct ifreq *ifr = (struct ifreq *)data; 1437 struct if_laddrreq *iflr = data; 1438 const struct sockaddr_dl *sdl; 1439 static const uint8_t zero[ETHER_ADDR_LEN]; 1440 int error; 1441 1442 switch (cmd) { 1443 case SIOCINITIFADDR: 1444 { 1445 struct ifaddr *ifa = (struct ifaddr *)data; 1446 if (ifa->ifa_addr->sa_family != AF_LINK 1447 && (ifp->if_flags & (IFF_UP | IFF_RUNNING)) != 1448 (IFF_UP | IFF_RUNNING)) { 1449 ifp->if_flags |= IFF_UP; 1450 if ((error = (*ifp->if_init)(ifp)) != 0) 1451 return error; 1452 } 1453 #ifdef INET 1454 if (ifa->ifa_addr->sa_family == AF_INET) 1455 arp_ifinit(ifp, ifa); 1456 #endif 1457 return 0; 1458 } 1459 1460 case SIOCSIFMTU: 1461 { 1462 int maxmtu; 1463 1464 if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU) 1465 maxmtu = ETHERMTU_JUMBO; 1466 else 1467 maxmtu = ETHERMTU; 1468 1469 if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu) 1470 return EINVAL; 1471 else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET) 1472 return error; 1473 else if (ifp->if_flags & IFF_UP) { 1474 /* Make sure the device notices the MTU change. */ 1475 return (*ifp->if_init)(ifp); 1476 } else 1477 return 0; 1478 } 1479 1480 case SIOCSIFFLAGS: 1481 if ((error = ifioctl_common(ifp, cmd, data)) != 0) 1482 return error; 1483 return ether_ioctl_reinit(ec); 1484 case SIOCGIFFLAGS: 1485 error = ifioctl_common(ifp, cmd, data); 1486 if (error == 0) { 1487 /* Set IFF_ALLMULTI for backcompat */ 1488 ifr->ifr_flags |= (ec->ec_flags & ETHER_F_ALLMULTI) ? 1489 IFF_ALLMULTI : 0; 1490 } 1491 return error; 1492 case SIOCGETHERCAP: 1493 eccr = (struct eccapreq *)data; 1494 eccr->eccr_capabilities = ec->ec_capabilities; 1495 eccr->eccr_capenable = ec->ec_capenable; 1496 return 0; 1497 case SIOCSETHERCAP: 1498 eccr = (struct eccapreq *)data; 1499 if ((eccr->eccr_capenable & ~ec->ec_capabilities) != 0) 1500 return EINVAL; 1501 if (eccr->eccr_capenable == ec->ec_capenable) 1502 return 0; 1503 #if 0 /* notyet */ 1504 ec->ec_capenable = (ec->ec_capenable & ETHERCAP_CANTCHANGE) 1505 | (eccr->eccr_capenable & ~ETHERCAP_CANTCHANGE); 1506 #else 1507 ec->ec_capenable = eccr->eccr_capenable; 1508 #endif 1509 return ether_ioctl_reinit(ec); 1510 case SIOCADDMULTI: 1511 return ether_addmulti(ifreq_getaddr(cmd, ifr), ec); 1512 case SIOCDELMULTI: 1513 return ether_delmulti(ifreq_getaddr(cmd, ifr), ec); 1514 case SIOCSIFMEDIA: 1515 case SIOCGIFMEDIA: 1516 if (ec->ec_mii != NULL) 1517 return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, 1518 cmd); 1519 else if (ec->ec_ifmedia != NULL) 1520 return ifmedia_ioctl(ifp, ifr, ec->ec_ifmedia, cmd); 1521 else 1522 return ENOTTY; 1523 break; 1524 case SIOCALIFADDR: 1525 sdl = satocsdl(sstocsa(&iflr->addr)); 1526 if (sdl->sdl_family != AF_LINK) 1527 ; 1528 else if (ETHER_IS_MULTICAST(CLLADDR(sdl))) 1529 return EINVAL; 1530 else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0) 1531 return EINVAL; 1532 /*FALLTHROUGH*/ 1533 default: 1534 return ifioctl_common(ifp, cmd, data); 1535 } 1536 return 0; 1537 } 1538 1539 /* 1540 * Enable/disable passing VLAN packets if the parent interface supports it. 1541 * Return: 1542 * 0: Ok 1543 * -1: Parent interface does not support vlans 1544 * >0: Error 1545 */ 1546 int 1547 ether_enable_vlan_mtu(struct ifnet *ifp) 1548 { 1549 int error; 1550 struct ethercom *ec = (void *)ifp; 1551 1552 /* Parent does not support VLAN's */ 1553 if ((ec->ec_capabilities & ETHERCAP_VLAN_MTU) == 0) 1554 return -1; 1555 1556 /* 1557 * Parent supports the VLAN_MTU capability, 1558 * i.e. can Tx/Rx larger than ETHER_MAX_LEN frames; 1559 * enable it. 1560 */ 1561 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1562 1563 /* Interface is down, defer for later */ 1564 if ((ifp->if_flags & IFF_UP) == 0) 1565 return 0; 1566 1567 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1568 return 0; 1569 1570 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1571 return error; 1572 } 1573 1574 int 1575 ether_disable_vlan_mtu(struct ifnet *ifp) 1576 { 1577 int error; 1578 struct ethercom *ec = (void *)ifp; 1579 1580 /* We still have VLAN's, defer for later */ 1581 if (ec->ec_nvlans != 0) 1582 return 0; 1583 1584 /* Parent does not support VLAB's, nothing to do. */ 1585 if ((ec->ec_capenable & ETHERCAP_VLAN_MTU) == 0) 1586 return -1; 1587 1588 /* 1589 * Disable Tx/Rx of VLAN-sized frames. 1590 */ 1591 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1592 1593 /* Interface is down, defer for later */ 1594 if ((ifp->if_flags & IFF_UP) == 0) 1595 return 0; 1596 1597 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1598 return 0; 1599 1600 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1601 return error; 1602 } 1603 1604 static int 1605 ether_multicast_sysctl(SYSCTLFN_ARGS) 1606 { 1607 struct ether_multi *enm; 1608 struct ifnet *ifp; 1609 struct ethercom *ec; 1610 int error = 0; 1611 size_t written; 1612 struct psref psref; 1613 int bound; 1614 unsigned int multicnt; 1615 struct ether_multi_sysctl *addrs; 1616 int i; 1617 1618 if (namelen != 1) 1619 return EINVAL; 1620 1621 bound = curlwp_bind(); 1622 ifp = if_get_byindex(name[0], &psref); 1623 if (ifp == NULL) { 1624 error = ENODEV; 1625 goto out; 1626 } 1627 if (ifp->if_type != IFT_ETHER) { 1628 if_put(ifp, &psref); 1629 *oldlenp = 0; 1630 goto out; 1631 } 1632 ec = (struct ethercom *)ifp; 1633 1634 if (oldp == NULL) { 1635 if_put(ifp, &psref); 1636 *oldlenp = ec->ec_multicnt * sizeof(*addrs); 1637 goto out; 1638 } 1639 1640 /* 1641 * ec->ec_lock is a spin mutex so we cannot call sysctl_copyout, which 1642 * is sleepable, while holding it. Copy data to a local buffer first 1643 * with the lock taken and then call sysctl_copyout without holding it. 1644 */ 1645 retry: 1646 multicnt = ec->ec_multicnt; 1647 1648 if (multicnt == 0) { 1649 if_put(ifp, &psref); 1650 *oldlenp = 0; 1651 goto out; 1652 } 1653 1654 addrs = kmem_zalloc(sizeof(*addrs) * multicnt, KM_SLEEP); 1655 1656 ETHER_LOCK(ec); 1657 if (multicnt != ec->ec_multicnt) { 1658 /* The number of multicast addresses has changed */ 1659 ETHER_UNLOCK(ec); 1660 kmem_free(addrs, sizeof(*addrs) * multicnt); 1661 goto retry; 1662 } 1663 1664 i = 0; 1665 LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) { 1666 struct ether_multi_sysctl *addr = &addrs[i]; 1667 addr->enm_refcount = enm->enm_refcount; 1668 memcpy(addr->enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN); 1669 memcpy(addr->enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN); 1670 i++; 1671 } 1672 ETHER_UNLOCK(ec); 1673 1674 error = 0; 1675 written = 0; 1676 for (i = 0; i < multicnt; i++) { 1677 struct ether_multi_sysctl *addr = &addrs[i]; 1678 1679 if (written + sizeof(*addr) > *oldlenp) 1680 break; 1681 error = sysctl_copyout(l, addr, oldp, sizeof(*addr)); 1682 if (error) 1683 break; 1684 written += sizeof(*addr); 1685 oldp = (char *)oldp + sizeof(*addr); 1686 } 1687 kmem_free(addrs, sizeof(*addrs) * multicnt); 1688 1689 if_put(ifp, &psref); 1690 1691 *oldlenp = written; 1692 out: 1693 curlwp_bindx(bound); 1694 return error; 1695 } 1696 1697 static void 1698 ether_sysctl_setup(struct sysctllog **clog) 1699 { 1700 const struct sysctlnode *rnode = NULL; 1701 1702 sysctl_createv(clog, 0, NULL, &rnode, 1703 CTLFLAG_PERMANENT, 1704 CTLTYPE_NODE, "ether", 1705 SYSCTL_DESCR("Ethernet-specific information"), 1706 NULL, 0, NULL, 0, 1707 CTL_NET, CTL_CREATE, CTL_EOL); 1708 1709 sysctl_createv(clog, 0, &rnode, NULL, 1710 CTLFLAG_PERMANENT, 1711 CTLTYPE_NODE, "multicast", 1712 SYSCTL_DESCR("multicast addresses"), 1713 ether_multicast_sysctl, 0, NULL, 0, 1714 CTL_CREATE, CTL_EOL); 1715 } 1716 1717 void 1718 etherinit(void) 1719 { 1720 1721 mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET); 1722 ether_sysctl_setup(NULL); 1723 } 1724
Indexes created Fri Sep 26 08:10:20 GMT 2025