if_ethersubr.c revision 1.258
1 /* $NetBSD: if_ethersubr.c,v 1.258 2018/02/12 12:17:38 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 1.258 2018/02/12 12:17:38 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, const struct rtentry *rt) 191 { 192 uint8_t esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN]; 193 uint16_t etype = 0; 194 int error = 0, hdrcmplt = 0; 195 struct mbuf *m = m0; 196 struct mbuf *mcopy = NULL; 197 struct ether_header *eh; 198 struct ifnet *ifp = ifp0; 199 #ifdef INET 200 struct arphdr *ah; 201 #endif /* INET */ 202 #ifdef NETATALK 203 struct at_ifaddr *aa; 204 #endif /* NETATALK */ 205 206 #ifdef MBUFTRACE 207 m_claimm(m, ifp->if_mowner); 208 #endif 209 210 #if NCARP > 0 211 if (ifp->if_type == IFT_CARP) { 212 struct ifaddr *ifa; 213 int s = pserialize_read_enter(); 214 215 /* loop back if this is going to the carp interface */ 216 if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP && 217 (ifa = ifa_ifwithaddr(dst)) != NULL) { 218 if (ifa->ifa_ifp == ifp0) { 219 pserialize_read_exit(s); 220 return looutput(ifp0, m, dst, rt); 221 } 222 } 223 pserialize_read_exit(s); 224 225 ifp = ifp->if_carpdev; 226 /* ac = (struct arpcom *)ifp; */ 227 228 if ((ifp0->if_flags & (IFF_UP | IFF_RUNNING)) != 229 (IFF_UP | IFF_RUNNING)) 230 senderr(ENETDOWN); 231 } 232 #endif /* NCARP > 0 */ 233 234 if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING)) 235 senderr(ENETDOWN); 236 237 switch (dst->sa_family) { 238 239 #ifdef INET 240 case AF_INET: 241 if (m->m_flags & M_BCAST) 242 memcpy(edst, etherbroadcastaddr, sizeof(edst)); 243 else if (m->m_flags & M_MCAST) 244 ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst); 245 else if ((error = arpresolve(ifp, rt, m, dst, edst, 246 sizeof(edst))) != 0) { 247 return error == EWOULDBLOCK ? 0 : error; 248 } 249 /* If broadcasting on a simplex interface, loopback a copy */ 250 if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX)) 251 mcopy = m_copy(m, 0, M_COPYALL); 252 etype = htons(ETHERTYPE_IP); 253 break; 254 255 case AF_ARP: 256 ah = mtod(m, struct arphdr *); 257 if (m->m_flags & M_BCAST) 258 memcpy(edst, etherbroadcastaddr, sizeof(edst)); 259 else { 260 void *tha = ar_tha(ah); 261 262 if (tha == NULL) { 263 /* fake with ARPHRD_IEEE1394 */ 264 m_freem(m); 265 return 0; 266 } 267 memcpy(edst, tha, sizeof(edst)); 268 } 269 270 ah->ar_hrd = htons(ARPHRD_ETHER); 271 272 switch (ntohs(ah->ar_op)) { 273 case ARPOP_REVREQUEST: 274 case ARPOP_REVREPLY: 275 etype = htons(ETHERTYPE_REVARP); 276 break; 277 278 case ARPOP_REQUEST: 279 case ARPOP_REPLY: 280 default: 281 etype = htons(ETHERTYPE_ARP); 282 } 283 break; 284 #endif 285 286 #ifdef INET6 287 case AF_INET6: 288 if (m->m_flags & M_BCAST) 289 memcpy(edst, etherbroadcastaddr, sizeof(edst)); 290 else if (m->m_flags & M_MCAST) { 291 ETHER_MAP_IPV6_MULTICAST(&satocsin6(dst)->sin6_addr, 292 edst); 293 } else { 294 error = nd6_resolve(ifp, rt, m, dst, edst, 295 sizeof(edst)); 296 if (error != 0) 297 return error == EWOULDBLOCK ? 0 : error; 298 } 299 etype = htons(ETHERTYPE_IPV6); 300 break; 301 #endif 302 303 #ifdef NETATALK 304 case AF_APPLETALK: { 305 struct ifaddr *ifa; 306 int s; 307 308 KERNEL_LOCK(1, NULL); 309 310 if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) { 311 #ifdef NETATALKDEBUG 312 printf("aarpresolv failed\n"); 313 #endif 314 KERNEL_UNLOCK_ONE(NULL); 315 return (0); 316 } 317 318 /* 319 * ifaddr is the first thing in at_ifaddr 320 */ 321 s = pserialize_read_enter(); 322 ifa = at_ifawithnet((const struct sockaddr_at *)dst, ifp); 323 if (ifa == NULL) { 324 pserialize_read_exit(s); 325 KERNEL_UNLOCK_ONE(NULL); 326 /* XXX error? */ 327 goto bad; 328 } 329 aa = (struct at_ifaddr *)ifa; 330 331 /* 332 * In the phase 2 case, we need to prepend an mbuf for the 333 * llc header. Since we must preserve the value of m, 334 * which is passed to us by value, we m_copy() the first 335 * mbuf, and use it for our llc header. 336 */ 337 if (aa->aa_flags & AFA_PHASE2) { 338 struct llc llc; 339 340 M_PREPEND(m, sizeof(struct llc), M_DONTWAIT); 341 if (m == NULL) { 342 KERNEL_UNLOCK_ONE(NULL); 343 senderr(ENOBUFS); 344 } 345 346 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 347 llc.llc_control = LLC_UI; 348 memcpy(llc.llc_snap_org_code, at_org_code, 349 sizeof(llc.llc_snap_org_code)); 350 llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK); 351 memcpy(mtod(m, void *), &llc, sizeof(struct llc)); 352 } else { 353 etype = htons(ETHERTYPE_ATALK); 354 } 355 pserialize_read_exit(s); 356 KERNEL_UNLOCK_ONE(NULL); 357 break; 358 } 359 #endif /* NETATALK */ 360 361 case pseudo_AF_HDRCMPLT: 362 hdrcmplt = 1; 363 memcpy(esrc, 364 ((const struct ether_header *)dst->sa_data)->ether_shost, 365 sizeof(esrc)); 366 /* FALLTHROUGH */ 367 368 case AF_UNSPEC: 369 memcpy(edst, 370 ((const struct ether_header *)dst->sa_data)->ether_dhost, 371 sizeof(edst)); 372 /* AF_UNSPEC doesn't swap the byte order of the ether_type. */ 373 etype = ((const struct ether_header *)dst->sa_data)->ether_type; 374 break; 375 376 default: 377 printf("%s: can't handle af%d\n", ifp->if_xname, 378 dst->sa_family); 379 senderr(EAFNOSUPPORT); 380 } 381 382 #ifdef MPLS 383 KERNEL_LOCK(1, NULL); 384 { 385 struct m_tag *mtag; 386 mtag = m_tag_find(m, PACKET_TAG_MPLS, NULL); 387 if (mtag != NULL) { 388 /* Having the tag itself indicates it's MPLS */ 389 etype = htons(ETHERTYPE_MPLS); 390 m_tag_delete(m, mtag); 391 } 392 } 393 KERNEL_UNLOCK_ONE(NULL); 394 #endif 395 396 if (mcopy) 397 (void)looutput(ifp, mcopy, dst, rt); 398 399 KASSERT((m->m_flags & M_PKTHDR) != 0); 400 401 /* 402 * If no ether type is set, this must be a 802.2 formatted packet. 403 */ 404 if (etype == 0) 405 etype = htons(m->m_pkthdr.len); 406 407 /* 408 * Add local net header. If no space in first mbuf, allocate another. 409 */ 410 M_PREPEND(m, sizeof(struct ether_header), M_DONTWAIT); 411 if (m == NULL) 412 senderr(ENOBUFS); 413 414 eh = mtod(m, struct ether_header *); 415 /* Note: etype is already in network byte order. */ 416 memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type)); 417 memcpy(eh->ether_dhost, edst, sizeof(edst)); 418 if (hdrcmplt) 419 memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost)); 420 else 421 memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl), 422 sizeof(eh->ether_shost)); 423 424 #if NCARP > 0 425 if (ifp0 != ifp && ifp0->if_type == IFT_CARP) { 426 memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl), 427 sizeof(eh->ether_shost)); 428 } 429 #endif 430 431 if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0) 432 return (error); 433 if (m == NULL) 434 return (0); 435 436 #if NBRIDGE > 0 437 /* 438 * Bridges require special output handling. 439 */ 440 if (ifp->if_bridge) 441 return bridge_output(ifp, m, NULL, NULL); 442 #endif 443 444 #if NCARP > 0 445 if (ifp != ifp0) 446 ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN; 447 #endif 448 449 #ifdef ALTQ 450 KERNEL_LOCK(1, NULL); 451 /* 452 * If ALTQ is enabled on the parent interface, do 453 * classification; the queueing discipline might not 454 * require classification, but might require the 455 * address family/header pointer in the pktattr. 456 */ 457 if (ALTQ_IS_ENABLED(&ifp->if_snd)) 458 altq_etherclassify(&ifp->if_snd, m); 459 KERNEL_UNLOCK_ONE(NULL); 460 #endif 461 return ifq_enqueue(ifp, m); 462 463 bad: 464 if (m) 465 m_freem(m); 466 return (error); 467 } 468 469 #ifdef ALTQ 470 /* 471 * This routine is a slight hack to allow a packet to be classified 472 * if the Ethernet headers are present. It will go away when ALTQ's 473 * classification engine understands link headers. 474 */ 475 void 476 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m) 477 { 478 struct ether_header *eh; 479 struct mbuf *mtop = m; 480 uint16_t ether_type; 481 int hlen, af, hdrsize; 482 void *hdr; 483 484 KASSERT((mtop->m_flags & M_PKTHDR) != 0); 485 486 hlen = ETHER_HDR_LEN; 487 eh = mtod(m, struct ether_header *); 488 489 ether_type = htons(eh->ether_type); 490 491 if (ether_type < ETHERMTU) { 492 /* LLC/SNAP */ 493 struct llc *llc = (struct llc *)(eh + 1); 494 hlen += 8; 495 496 if (m->m_len < hlen || 497 llc->llc_dsap != LLC_SNAP_LSAP || 498 llc->llc_ssap != LLC_SNAP_LSAP || 499 llc->llc_control != LLC_UI) { 500 /* Not SNAP. */ 501 goto bad; 502 } 503 504 ether_type = htons(llc->llc_un.type_snap.ether_type); 505 } 506 507 switch (ether_type) { 508 case ETHERTYPE_IP: 509 af = AF_INET; 510 hdrsize = 20; /* sizeof(struct ip) */ 511 break; 512 513 case ETHERTYPE_IPV6: 514 af = AF_INET6; 515 hdrsize = 40; /* sizeof(struct ip6_hdr) */ 516 break; 517 518 default: 519 af = AF_UNSPEC; 520 hdrsize = 0; 521 break; 522 } 523 524 while (m->m_len <= hlen) { 525 hlen -= m->m_len; 526 m = m->m_next; 527 if (m == NULL) 528 goto bad; 529 } 530 531 if (m->m_len < (hlen + hdrsize)) { 532 /* 533 * protocol header not in a single mbuf. 534 * We can't cope with this situation right 535 * now (but it shouldn't ever happen, really, anyhow). 536 */ 537 #ifdef DEBUG 538 printf("altq_etherclassify: headers span multiple mbufs: " 539 "%d < %d\n", m->m_len, (hlen + hdrsize)); 540 #endif 541 goto bad; 542 } 543 544 m->m_data += hlen; 545 m->m_len -= hlen; 546 547 hdr = mtod(m, void *); 548 549 if (ALTQ_NEEDS_CLASSIFY(ifq)) { 550 mtop->m_pkthdr.pattr_class = 551 (*ifq->altq_classify)(ifq->altq_clfier, m, af); 552 } 553 mtop->m_pkthdr.pattr_af = af; 554 mtop->m_pkthdr.pattr_hdr = hdr; 555 556 m->m_data -= hlen; 557 m->m_len += hlen; 558 559 return; 560 561 bad: 562 mtop->m_pkthdr.pattr_class = NULL; 563 mtop->m_pkthdr.pattr_hdr = NULL; 564 mtop->m_pkthdr.pattr_af = AF_UNSPEC; 565 } 566 #endif /* ALTQ */ 567 568 /* 569 * Process a received Ethernet packet; 570 * the packet is in the mbuf chain m with 571 * the ether header. 572 */ 573 void 574 ether_input(struct ifnet *ifp, struct mbuf *m) 575 { 576 struct ethercom *ec = (struct ethercom *) ifp; 577 pktqueue_t *pktq = NULL; 578 struct ifqueue *inq = NULL; 579 uint16_t etype; 580 struct ether_header *eh; 581 size_t ehlen; 582 static int earlypkts; 583 int isr = 0; 584 #if defined (LLC) || defined(NETATALK) 585 struct llc *l; 586 #endif 587 588 KASSERT(!cpu_intr_p()); 589 KASSERT((m->m_flags & M_PKTHDR) != 0); 590 591 if ((ifp->if_flags & IFF_UP) == 0) { 592 m_freem(m); 593 return; 594 } 595 596 #ifdef MBUFTRACE 597 m_claimm(m, &ec->ec_rx_mowner); 598 #endif 599 eh = mtod(m, struct ether_header *); 600 etype = ntohs(eh->ether_type); 601 ehlen = sizeof(*eh); 602 603 if (__predict_false(earlypkts < 100 || !rnd_initial_entropy)) { 604 rnd_add_data(NULL, eh, ehlen, 0); 605 earlypkts++; 606 } 607 608 /* 609 * Determine if the packet is within its size limits. For MPLS the 610 * header length is variable, so we skip the check. 611 */ 612 if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len > 613 ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) { 614 mutex_enter(&bigpktpps_lock); 615 if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count, 616 bigpktppslim)) { 617 printf("%s: discarding oversize frame (len=%d)\n", 618 ifp->if_xname, m->m_pkthdr.len); 619 } 620 mutex_exit(&bigpktpps_lock); 621 m_freem(m); 622 return; 623 } 624 625 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 626 /* 627 * If this is not a simplex interface, drop the packet 628 * if it came from us. 629 */ 630 if ((ifp->if_flags & IFF_SIMPLEX) == 0 && 631 memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost, 632 ETHER_ADDR_LEN) == 0) { 633 m_freem(m); 634 return; 635 } 636 637 if (memcmp(etherbroadcastaddr, 638 eh->ether_dhost, ETHER_ADDR_LEN) == 0) 639 m->m_flags |= M_BCAST; 640 else 641 m->m_flags |= M_MCAST; 642 ifp->if_imcasts++; 643 } 644 645 /* If the CRC is still on the packet, trim it off. */ 646 if (m->m_flags & M_HASFCS) { 647 m_adj(m, -ETHER_CRC_LEN); 648 m->m_flags &= ~M_HASFCS; 649 } 650 651 ifp->if_ibytes += m->m_pkthdr.len; 652 653 #if NCARP > 0 654 if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) { 655 /* 656 * clear M_PROMISC, in case the packets comes from a 657 * vlan 658 */ 659 m->m_flags &= ~M_PROMISC; 660 if (carp_input(m, (uint8_t *)&eh->ether_shost, 661 (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0) 662 return; 663 } 664 #endif 665 666 if ((m->m_flags & (M_BCAST | M_MCAST | M_PROMISC)) == 0 && 667 (ifp->if_flags & IFF_PROMISC) != 0 && 668 memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost, 669 ETHER_ADDR_LEN) != 0) { 670 m->m_flags |= M_PROMISC; 671 } 672 673 if ((m->m_flags & M_PROMISC) == 0) { 674 if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0) 675 return; 676 if (m == NULL) 677 return; 678 679 eh = mtod(m, struct ether_header *); 680 etype = ntohs(eh->ether_type); 681 } 682 683 #if NAGR > 0 684 if (ifp->if_agrprivate && 685 __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) { 686 m->m_flags &= ~M_PROMISC; 687 agr_input(ifp, m); 688 return; 689 } 690 #endif 691 692 /* 693 * If VLANs are configured on the interface, check to 694 * see if the device performed the decapsulation and 695 * provided us with the tag. 696 */ 697 if (ec->ec_nvlans && vlan_has_tag(m)) { 698 #if NVLAN > 0 699 /* 700 * vlan_input() will either recursively call ether_input() 701 * or drop the packet. 702 */ 703 vlan_input(ifp, m); 704 #else 705 m_freem(m); 706 #endif 707 return; 708 } 709 710 /* 711 * Handle protocols that expect to have the Ethernet header 712 * (and possibly FCS) intact. 713 */ 714 switch (etype) { 715 case ETHERTYPE_VLAN: { 716 struct ether_vlan_header *evl = (void *)eh; 717 718 /* 719 * If there is a tag of 0, then the VLAN header was probably 720 * just being used to store the priority. Extract the ether 721 * type, and if IP or IPV6, let them deal with it. 722 */ 723 if (m->m_len >= sizeof(*evl) && 724 EVL_VLANOFTAG(evl->evl_tag) == 0) { 725 etype = ntohs(evl->evl_proto); 726 ehlen = sizeof(*evl); 727 if ((m->m_flags & M_PROMISC) == 0 && 728 (etype == ETHERTYPE_IP || 729 etype == ETHERTYPE_IPV6)) 730 break; 731 } 732 733 #if NVLAN > 0 734 /* 735 * vlan_input() will either recursively call ether_input() 736 * or drop the packet. 737 */ 738 if (ec->ec_nvlans != 0) 739 vlan_input(ifp, m); 740 else 741 #endif 742 m_freem(m); 743 744 return; 745 } 746 747 #if NPPPOE > 0 748 case ETHERTYPE_PPPOEDISC: 749 pppoedisc_input(ifp, m); 750 return; 751 752 case ETHERTYPE_PPPOE: 753 pppoe_input(ifp, m); 754 return; 755 #endif 756 757 case ETHERTYPE_SLOWPROTOCOLS: { 758 uint8_t subtype; 759 760 KASSERTMSG((m->m_pkthdr.len >= sizeof(*eh) + sizeof(subtype)), 761 "too short slow protocol packet"); 762 763 m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype); 764 switch (subtype) { 765 #if NAGR > 0 766 case SLOWPROTOCOLS_SUBTYPE_LACP: 767 if (ifp->if_agrprivate) { 768 ieee8023ad_lacp_input(ifp, m); 769 return; 770 } 771 break; 772 773 case SLOWPROTOCOLS_SUBTYPE_MARKER: 774 if (ifp->if_agrprivate) { 775 ieee8023ad_marker_input(ifp, m); 776 return; 777 } 778 break; 779 #endif /* NAGR > 0 */ 780 781 default: 782 if (subtype == 0 || subtype > 10) { 783 /* illegal value */ 784 m_freem(m); 785 return; 786 } 787 /* unknown subtype */ 788 break; 789 } 790 /* FALLTHROUGH */ 791 } 792 793 default: 794 if (m->m_flags & M_PROMISC) { 795 m_freem(m); 796 return; 797 } 798 } 799 800 /* If the CRC is still on the packet, trim it off. */ 801 if (m->m_flags & M_HASFCS) { 802 m_adj(m, -ETHER_CRC_LEN); 803 m->m_flags &= ~M_HASFCS; 804 } 805 806 if (etype > ETHERMTU + sizeof(struct ether_header)) { 807 /* Strip off the Ethernet header. */ 808 m_adj(m, ehlen); 809 810 switch (etype) { 811 #ifdef INET 812 case ETHERTYPE_IP: 813 #ifdef GATEWAY 814 if (ipflow_fastforward(m)) 815 return; 816 #endif 817 pktq = ip_pktq; 818 break; 819 820 case ETHERTYPE_ARP: 821 isr = NETISR_ARP; 822 inq = &arpintrq; 823 break; 824 825 case ETHERTYPE_REVARP: 826 revarpinput(m); /* XXX queue? */ 827 return; 828 #endif 829 830 #ifdef INET6 831 case ETHERTYPE_IPV6: 832 if (__predict_false(!in6_present)) { 833 m_freem(m); 834 return; 835 } 836 #ifdef GATEWAY 837 if (ip6flow_fastforward(&m)) 838 return; 839 #endif 840 pktq = ip6_pktq; 841 break; 842 #endif 843 844 #ifdef NETATALK 845 case ETHERTYPE_ATALK: 846 isr = NETISR_ATALK; 847 inq = &atintrq1; 848 break; 849 850 case ETHERTYPE_AARP: 851 aarpinput(ifp, m); /* XXX queue? */ 852 return; 853 #endif 854 855 #ifdef MPLS 856 case ETHERTYPE_MPLS: 857 isr = NETISR_MPLS; 858 inq = &mplsintrq; 859 break; 860 #endif 861 862 default: 863 m_freem(m); 864 return; 865 } 866 } else { 867 KASSERT(ehlen == sizeof(*eh)); 868 #if defined (LLC) || defined (NETATALK) 869 if (m->m_len < sizeof(*eh) + sizeof(struct llc)) { 870 goto dropanyway; 871 } 872 l = (struct llc *)(eh+1); 873 874 switch (l->llc_dsap) { 875 #ifdef NETATALK 876 case LLC_SNAP_LSAP: 877 switch (l->llc_control) { 878 case LLC_UI: 879 if (l->llc_ssap != LLC_SNAP_LSAP) { 880 goto dropanyway; 881 } 882 883 if (memcmp(&(l->llc_snap_org_code)[0], 884 at_org_code, sizeof(at_org_code)) == 0 && 885 ntohs(l->llc_snap_ether_type) == 886 ETHERTYPE_ATALK) { 887 inq = &atintrq2; 888 m_adj(m, sizeof(struct ether_header) 889 + sizeof(struct llc)); 890 isr = NETISR_ATALK; 891 break; 892 } 893 894 if (memcmp(&(l->llc_snap_org_code)[0], 895 aarp_org_code, 896 sizeof(aarp_org_code)) == 0 && 897 ntohs(l->llc_snap_ether_type) == 898 ETHERTYPE_AARP) { 899 m_adj(m, sizeof(struct ether_header) 900 + sizeof(struct llc)); 901 aarpinput(ifp, m); /* XXX */ 902 return; 903 } 904 905 default: 906 goto dropanyway; 907 } 908 break; 909 #endif 910 dropanyway: 911 default: 912 m_freem(m); 913 return; 914 } 915 #else /* LLC || NETATALK */ 916 m_freem(m); 917 return; 918 #endif /* LLC || NETATALK */ 919 } 920 921 if (__predict_true(pktq)) { 922 #ifdef NET_MPSAFE 923 const u_int h = curcpu()->ci_index; 924 #else 925 const uint32_t h = pktq_rps_hash(m); 926 #endif 927 if (__predict_false(!pktq_enqueue(pktq, m, h))) { 928 m_freem(m); 929 } 930 return; 931 } 932 933 if (__predict_false(!inq)) { 934 /* Should not happen. */ 935 m_freem(m); 936 return; 937 } 938 939 IFQ_LOCK(inq); 940 if (IF_QFULL(inq)) { 941 IF_DROP(inq); 942 IFQ_UNLOCK(inq); 943 m_freem(m); 944 } else { 945 IF_ENQUEUE(inq, m); 946 IFQ_UNLOCK(inq); 947 schednetisr(isr); 948 } 949 } 950 951 /* 952 * Convert Ethernet address to printable (loggable) representation. 953 */ 954 char * 955 ether_sprintf(const u_char *ap) 956 { 957 static char etherbuf[3 * ETHER_ADDR_LEN]; 958 return ether_snprintf(etherbuf, sizeof(etherbuf), ap); 959 } 960 961 char * 962 ether_snprintf(char *buf, size_t len, const u_char *ap) 963 { 964 char *cp = buf; 965 size_t i; 966 967 for (i = 0; i < len / 3; i++) { 968 *cp++ = hexdigits[*ap >> 4]; 969 *cp++ = hexdigits[*ap++ & 0xf]; 970 *cp++ = ':'; 971 } 972 *--cp = '\0'; 973 return buf; 974 } 975 976 /* 977 * Perform common duties while attaching to interface list 978 */ 979 void 980 ether_ifattach(struct ifnet *ifp, const uint8_t *lla) 981 { 982 struct ethercom *ec = (struct ethercom *)ifp; 983 984 ifp->if_type = IFT_ETHER; 985 ifp->if_hdrlen = ETHER_HDR_LEN; 986 ifp->if_dlt = DLT_EN10MB; 987 ifp->if_mtu = ETHERMTU; 988 ifp->if_output = ether_output; 989 ifp->_if_input = ether_input; 990 if (ifp->if_baudrate == 0) 991 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 992 993 if (lla != NULL) 994 if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla)); 995 996 LIST_INIT(&ec->ec_multiaddrs); 997 ec->ec_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NET); 998 ifp->if_broadcastaddr = etherbroadcastaddr; 999 bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header)); 1000 #ifdef MBUFTRACE 1001 strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname, 1002 sizeof(ec->ec_tx_mowner.mo_name)); 1003 strlcpy(ec->ec_tx_mowner.mo_descr, "tx", 1004 sizeof(ec->ec_tx_mowner.mo_descr)); 1005 strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname, 1006 sizeof(ec->ec_rx_mowner.mo_name)); 1007 strlcpy(ec->ec_rx_mowner.mo_descr, "rx", 1008 sizeof(ec->ec_rx_mowner.mo_descr)); 1009 MOWNER_ATTACH(&ec->ec_tx_mowner); 1010 MOWNER_ATTACH(&ec->ec_rx_mowner); 1011 ifp->if_mowner = &ec->ec_tx_mowner; 1012 #endif 1013 } 1014 1015 void 1016 ether_ifdetach(struct ifnet *ifp) 1017 { 1018 struct ethercom *ec = (void *) ifp; 1019 struct ether_multi *enm; 1020 1021 /* 1022 * Prevent further calls to ioctl (for example turning off 1023 * promiscuous mode from the bridge code), which eventually can 1024 * call if_init() which can cause panics because the interface 1025 * is in the process of being detached. Return device not configured 1026 * instead. 1027 */ 1028 ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio; 1029 1030 #if NBRIDGE > 0 1031 if (ifp->if_bridge) 1032 bridge_ifdetach(ifp); 1033 #endif 1034 bpf_detach(ifp); 1035 #if NVLAN > 0 1036 if (ec->ec_nvlans) 1037 vlan_ifdetach(ifp); 1038 #endif 1039 1040 ETHER_LOCK(ec); 1041 while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) { 1042 LIST_REMOVE(enm, enm_list); 1043 kmem_free(enm, sizeof(*enm)); 1044 ec->ec_multicnt--; 1045 } 1046 ETHER_UNLOCK(ec); 1047 1048 mutex_obj_free(ec->ec_lock); 1049 ec->ec_lock = NULL; 1050 1051 ifp->if_mowner = NULL; 1052 MOWNER_DETACH(&ec->ec_rx_mowner); 1053 MOWNER_DETACH(&ec->ec_tx_mowner); 1054 } 1055 1056 #if 0 1057 /* 1058 * This is for reference. We have a table-driven version 1059 * of the little-endian crc32 generator, which is faster 1060 * than the double-loop. 1061 */ 1062 uint32_t 1063 ether_crc32_le(const uint8_t *buf, size_t len) 1064 { 1065 uint32_t c, crc, carry; 1066 size_t i, j; 1067 1068 crc = 0xffffffffU; /* initial value */ 1069 1070 for (i = 0; i < len; i++) { 1071 c = buf[i]; 1072 for (j = 0; j < 8; j++) { 1073 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1074 crc >>= 1; 1075 c >>= 1; 1076 if (carry) 1077 crc = (crc ^ ETHER_CRC_POLY_LE); 1078 } 1079 } 1080 1081 return (crc); 1082 } 1083 #else 1084 uint32_t 1085 ether_crc32_le(const uint8_t *buf, size_t len) 1086 { 1087 static const uint32_t crctab[] = { 1088 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1089 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1090 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1091 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1092 }; 1093 uint32_t crc; 1094 size_t i; 1095 1096 crc = 0xffffffffU; /* initial value */ 1097 1098 for (i = 0; i < len; i++) { 1099 crc ^= buf[i]; 1100 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1101 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1102 } 1103 1104 return (crc); 1105 } 1106 #endif 1107 1108 uint32_t 1109 ether_crc32_be(const uint8_t *buf, size_t len) 1110 { 1111 uint32_t c, crc, carry; 1112 size_t i, j; 1113 1114 crc = 0xffffffffU; /* initial value */ 1115 1116 for (i = 0; i < len; i++) { 1117 c = buf[i]; 1118 for (j = 0; j < 8; j++) { 1119 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1120 crc <<= 1; 1121 c >>= 1; 1122 if (carry) 1123 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1124 } 1125 } 1126 1127 return (crc); 1128 } 1129 1130 #ifdef INET 1131 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] = 1132 { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 }; 1133 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] = 1134 { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff }; 1135 #endif 1136 #ifdef INET6 1137 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] = 1138 { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 }; 1139 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] = 1140 { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff }; 1141 #endif 1142 1143 /* 1144 * ether_aton implementation, not using a static buffer. 1145 */ 1146 int 1147 ether_aton_r(u_char *dest, size_t len, const char *str) 1148 { 1149 const u_char *cp = (const void *)str; 1150 u_char *ep; 1151 1152 #define atox(c) (((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10)) 1153 1154 if (len < ETHER_ADDR_LEN) 1155 return ENOSPC; 1156 1157 ep = dest + ETHER_ADDR_LEN; 1158 1159 while (*cp) { 1160 if (!isxdigit(*cp)) 1161 return EINVAL; 1162 1163 *dest = atox(*cp); 1164 cp++; 1165 if (isxdigit(*cp)) { 1166 *dest = (*dest << 4) | atox(*cp); 1167 cp++; 1168 } 1169 dest++; 1170 1171 if (dest == ep) 1172 return (*cp == '\0') ? 0 : ENAMETOOLONG; 1173 1174 switch (*cp) { 1175 case ':': 1176 case '-': 1177 case '.': 1178 cp++; 1179 break; 1180 } 1181 } 1182 return ENOBUFS; 1183 } 1184 1185 /* 1186 * Convert a sockaddr into an Ethernet address or range of Ethernet 1187 * addresses. 1188 */ 1189 int 1190 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN], 1191 uint8_t addrhi[ETHER_ADDR_LEN]) 1192 { 1193 #ifdef INET 1194 const struct sockaddr_in *sin; 1195 #endif /* INET */ 1196 #ifdef INET6 1197 const struct sockaddr_in6 *sin6; 1198 #endif /* INET6 */ 1199 1200 switch (sa->sa_family) { 1201 1202 case AF_UNSPEC: 1203 memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN); 1204 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1205 break; 1206 1207 #ifdef INET 1208 case AF_INET: 1209 sin = satocsin(sa); 1210 if (sin->sin_addr.s_addr == INADDR_ANY) { 1211 /* 1212 * An IP address of INADDR_ANY means listen to 1213 * or stop listening to all of the Ethernet 1214 * multicast addresses used for IP. 1215 * (This is for the sake of IP multicast routers.) 1216 */ 1217 memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN); 1218 memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN); 1219 } else { 1220 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo); 1221 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1222 } 1223 break; 1224 #endif 1225 #ifdef INET6 1226 case AF_INET6: 1227 sin6 = satocsin6(sa); 1228 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1229 /* 1230 * An IP6 address of 0 means listen to or stop 1231 * listening to all of the Ethernet multicast 1232 * address used for IP6. 1233 * (This is used for multicast routers.) 1234 */ 1235 memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN); 1236 memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN); 1237 } else { 1238 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo); 1239 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1240 } 1241 break; 1242 #endif 1243 1244 default: 1245 return EAFNOSUPPORT; 1246 } 1247 return 0; 1248 } 1249 1250 /* 1251 * Add an Ethernet multicast address or range of addresses to the list for a 1252 * given interface. 1253 */ 1254 int 1255 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec) 1256 { 1257 struct ether_multi *enm, *_enm; 1258 u_char addrlo[ETHER_ADDR_LEN]; 1259 u_char addrhi[ETHER_ADDR_LEN]; 1260 int error = 0; 1261 1262 /* Allocate out of lock */ 1263 enm = kmem_alloc(sizeof(*enm), KM_SLEEP); 1264 1265 ETHER_LOCK(ec); 1266 error = ether_multiaddr(sa, addrlo, addrhi); 1267 if (error != 0) 1268 goto out; 1269 1270 /* 1271 * Verify that we have valid Ethernet multicast addresses. 1272 */ 1273 if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) { 1274 error = EINVAL; 1275 goto out; 1276 } 1277 1278 /* 1279 * See if the address range is already in the list. 1280 */ 1281 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, _enm); 1282 if (_enm != NULL) { 1283 /* 1284 * Found it; just increment the reference count. 1285 */ 1286 ++_enm->enm_refcount; 1287 error = 0; 1288 goto out; 1289 } 1290 1291 /* 1292 * Link a new multicast record into the interface's multicast list. 1293 */ 1294 memcpy(enm->enm_addrlo, addrlo, ETHER_ADDR_LEN); 1295 memcpy(enm->enm_addrhi, addrhi, ETHER_ADDR_LEN); 1296 enm->enm_refcount = 1; 1297 LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list); 1298 ec->ec_multicnt++; 1299 1300 /* 1301 * Return ENETRESET to inform the driver that the list has changed 1302 * and its reception filter should be adjusted accordingly. 1303 */ 1304 error = ENETRESET; 1305 enm = NULL; 1306 1307 out: 1308 ETHER_UNLOCK(ec); 1309 if (enm != NULL) 1310 kmem_free(enm, sizeof(*enm)); 1311 return error; 1312 } 1313 1314 /* 1315 * Delete a multicast address record. 1316 */ 1317 int 1318 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec) 1319 { 1320 struct ether_multi *enm; 1321 u_char addrlo[ETHER_ADDR_LEN]; 1322 u_char addrhi[ETHER_ADDR_LEN]; 1323 int error; 1324 1325 ETHER_LOCK(ec); 1326 error = ether_multiaddr(sa, addrlo, addrhi); 1327 if (error != 0) 1328 goto error; 1329 1330 /* 1331 * Look up the address in our list. 1332 */ 1333 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm); 1334 if (enm == NULL) { 1335 error = ENXIO; 1336 goto error; 1337 } 1338 if (--enm->enm_refcount != 0) { 1339 /* 1340 * Still some claims to this record. 1341 */ 1342 error = 0; 1343 goto error; 1344 } 1345 1346 /* 1347 * No remaining claims to this record; unlink and free it. 1348 */ 1349 LIST_REMOVE(enm, enm_list); 1350 ec->ec_multicnt--; 1351 ETHER_UNLOCK(ec); 1352 kmem_free(enm, sizeof(*enm)); 1353 1354 /* 1355 * Return ENETRESET to inform the driver that the list has changed 1356 * and its reception filter should be adjusted accordingly. 1357 */ 1358 return ENETRESET; 1359 1360 error: 1361 ETHER_UNLOCK(ec); 1362 return error; 1363 } 1364 1365 void 1366 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb) 1367 { 1368 ec->ec_ifflags_cb = cb; 1369 } 1370 1371 /* 1372 * Common ioctls for Ethernet interfaces. Note, we must be 1373 * called at splnet(). 1374 */ 1375 int 1376 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data) 1377 { 1378 struct ethercom *ec = (void *) ifp; 1379 struct eccapreq *eccr; 1380 struct ifreq *ifr = (struct ifreq *)data; 1381 struct if_laddrreq *iflr = data; 1382 const struct sockaddr_dl *sdl; 1383 static const uint8_t zero[ETHER_ADDR_LEN]; 1384 int error; 1385 1386 switch (cmd) { 1387 case SIOCINITIFADDR: 1388 { 1389 struct ifaddr *ifa = (struct ifaddr *)data; 1390 if (ifa->ifa_addr->sa_family != AF_LINK 1391 && (ifp->if_flags & (IFF_UP | IFF_RUNNING)) != 1392 (IFF_UP | IFF_RUNNING)) { 1393 ifp->if_flags |= IFF_UP; 1394 if ((error = (*ifp->if_init)(ifp)) != 0) 1395 return error; 1396 } 1397 #ifdef INET 1398 if (ifa->ifa_addr->sa_family == AF_INET) 1399 arp_ifinit(ifp, ifa); 1400 #endif 1401 return 0; 1402 } 1403 1404 case SIOCSIFMTU: 1405 { 1406 int maxmtu; 1407 1408 if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU) 1409 maxmtu = ETHERMTU_JUMBO; 1410 else 1411 maxmtu = ETHERMTU; 1412 1413 if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu) 1414 return EINVAL; 1415 else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET) 1416 return error; 1417 else if (ifp->if_flags & IFF_UP) { 1418 /* Make sure the device notices the MTU change. */ 1419 return (*ifp->if_init)(ifp); 1420 } else 1421 return 0; 1422 } 1423 1424 case SIOCSIFFLAGS: 1425 if ((error = ifioctl_common(ifp, cmd, data)) != 0) 1426 return error; 1427 switch (ifp->if_flags & (IFF_UP | IFF_RUNNING)) { 1428 case IFF_RUNNING: 1429 /* 1430 * If interface is marked down and it is running, 1431 * then stop and disable it. 1432 */ 1433 (*ifp->if_stop)(ifp, 1); 1434 break; 1435 case IFF_UP: 1436 /* 1437 * If interface is marked up and it is stopped, then 1438 * start it. 1439 */ 1440 return (*ifp->if_init)(ifp); 1441 case IFF_UP | IFF_RUNNING: 1442 error = 0; 1443 if (ec->ec_ifflags_cb != NULL) { 1444 error = (*ec->ec_ifflags_cb)(ec); 1445 if (error == ENETRESET) { 1446 /* 1447 * Reset the interface to pick up 1448 * changes in any other flags that 1449 * affect the hardware state. 1450 */ 1451 return (*ifp->if_init)(ifp); 1452 } 1453 } else 1454 error = (*ifp->if_init)(ifp); 1455 return error; 1456 case 0: 1457 break; 1458 } 1459 return 0; 1460 case SIOCGETHERCAP: 1461 eccr = (struct eccapreq *)data; 1462 eccr->eccr_capabilities = ec->ec_capabilities; 1463 eccr->eccr_capenable = ec->ec_capenable; 1464 return 0; 1465 case SIOCADDMULTI: 1466 return ether_addmulti(ifreq_getaddr(cmd, ifr), ec); 1467 case SIOCDELMULTI: 1468 return ether_delmulti(ifreq_getaddr(cmd, ifr), ec); 1469 case SIOCSIFMEDIA: 1470 case SIOCGIFMEDIA: 1471 if (ec->ec_mii == NULL) 1472 return ENOTTY; 1473 return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd); 1474 case SIOCALIFADDR: 1475 sdl = satocsdl(sstocsa(&iflr->addr)); 1476 if (sdl->sdl_family != AF_LINK) 1477 ; 1478 else if (ETHER_IS_MULTICAST(CLLADDR(sdl))) 1479 return EINVAL; 1480 else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0) 1481 return EINVAL; 1482 /*FALLTHROUGH*/ 1483 default: 1484 return ifioctl_common(ifp, cmd, data); 1485 } 1486 return 0; 1487 } 1488 1489 /* 1490 * Enable/disable passing VLAN packets if the parent interface supports it. 1491 * Return: 1492 * 0: Ok 1493 * -1: Parent interface does not support vlans 1494 * >0: Error 1495 */ 1496 int 1497 ether_enable_vlan_mtu(struct ifnet *ifp) 1498 { 1499 int error; 1500 struct ethercom *ec = (void *)ifp; 1501 1502 /* Parent does not support VLAN's */ 1503 if ((ec->ec_capabilities & ETHERCAP_VLAN_MTU) == 0) 1504 return -1; 1505 1506 /* 1507 * Parent supports the VLAN_MTU capability, 1508 * i.e. can Tx/Rx larger than ETHER_MAX_LEN frames; 1509 * enable it. 1510 */ 1511 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1512 1513 /* Interface is down, defer for later */ 1514 if ((ifp->if_flags & IFF_UP) == 0) 1515 return 0; 1516 1517 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1518 return 0; 1519 1520 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1521 return error; 1522 } 1523 1524 int 1525 ether_disable_vlan_mtu(struct ifnet *ifp) 1526 { 1527 int error; 1528 struct ethercom *ec = (void *)ifp; 1529 1530 /* We still have VLAN's, defer for later */ 1531 if (ec->ec_nvlans != 0) 1532 return 0; 1533 1534 /* Parent does not support VLAB's, nothing to do. */ 1535 if ((ec->ec_capenable & ETHERCAP_VLAN_MTU) == 0) 1536 return -1; 1537 1538 /* 1539 * Disable Tx/Rx of VLAN-sized frames. 1540 */ 1541 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1542 1543 /* Interface is down, defer for later */ 1544 if ((ifp->if_flags & IFF_UP) == 0) 1545 return 0; 1546 1547 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1548 return 0; 1549 1550 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1551 return error; 1552 } 1553 1554 static int 1555 ether_multicast_sysctl(SYSCTLFN_ARGS) 1556 { 1557 struct ether_multi *enm; 1558 struct ifnet *ifp; 1559 struct ethercom *ec; 1560 int error = 0; 1561 size_t written; 1562 struct psref psref; 1563 int bound; 1564 unsigned int multicnt; 1565 struct ether_multi_sysctl *addrs; 1566 int i; 1567 1568 if (namelen != 1) 1569 return EINVAL; 1570 1571 bound = curlwp_bind(); 1572 ifp = if_get_byindex(name[0], &psref); 1573 if (ifp == NULL) { 1574 error = ENODEV; 1575 goto out; 1576 } 1577 if (ifp->if_type != IFT_ETHER) { 1578 if_put(ifp, &psref); 1579 *oldlenp = 0; 1580 goto out; 1581 } 1582 ec = (struct ethercom *)ifp; 1583 1584 if (oldp == NULL) { 1585 if_put(ifp, &psref); 1586 *oldlenp = ec->ec_multicnt * sizeof(*addrs); 1587 goto out; 1588 } 1589 1590 /* 1591 * ec->ec_lock is a spin mutex so we cannot call sysctl_copyout, which 1592 * is sleepable, while holding it. Copy data to a local buffer first 1593 * with the lock taken and then call sysctl_copyout without holding it. 1594 */ 1595 retry: 1596 multicnt = ec->ec_multicnt; 1597 1598 if (multicnt == 0) { 1599 if_put(ifp, &psref); 1600 *oldlenp = 0; 1601 goto out; 1602 } 1603 1604 addrs = kmem_zalloc(sizeof(*addrs) * multicnt, KM_SLEEP); 1605 1606 ETHER_LOCK(ec); 1607 if (multicnt != ec->ec_multicnt) { 1608 /* The number of multicast addresses has changed */ 1609 ETHER_UNLOCK(ec); 1610 kmem_free(addrs, sizeof(*addrs) * multicnt); 1611 goto retry; 1612 } 1613 1614 i = 0; 1615 LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) { 1616 struct ether_multi_sysctl *addr = &addrs[i]; 1617 addr->enm_refcount = enm->enm_refcount; 1618 memcpy(addr->enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN); 1619 memcpy(addr->enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN); 1620 i++; 1621 } 1622 ETHER_UNLOCK(ec); 1623 1624 error = 0; 1625 written = 0; 1626 for (i = 0; i < multicnt; i++) { 1627 struct ether_multi_sysctl *addr = &addrs[i]; 1628 1629 if (written + sizeof(*addr) > *oldlenp) 1630 break; 1631 error = sysctl_copyout(l, addr, oldp, sizeof(*addr)); 1632 if (error) 1633 break; 1634 written += sizeof(*addr); 1635 oldp = (char *)oldp + sizeof(*addr); 1636 } 1637 kmem_free(addrs, sizeof(*addrs) * multicnt); 1638 1639 if_put(ifp, &psref); 1640 1641 *oldlenp = written; 1642 out: 1643 curlwp_bindx(bound); 1644 return error; 1645 } 1646 1647 static void 1648 ether_sysctl_setup(struct sysctllog **clog) 1649 { 1650 const struct sysctlnode *rnode = NULL; 1651 1652 sysctl_createv(clog, 0, NULL, &rnode, 1653 CTLFLAG_PERMANENT, 1654 CTLTYPE_NODE, "ether", 1655 SYSCTL_DESCR("Ethernet-specific information"), 1656 NULL, 0, NULL, 0, 1657 CTL_NET, CTL_CREATE, CTL_EOL); 1658 1659 sysctl_createv(clog, 0, &rnode, NULL, 1660 CTLFLAG_PERMANENT, 1661 CTLTYPE_NODE, "multicast", 1662 SYSCTL_DESCR("multicast addresses"), 1663 ether_multicast_sysctl, 0, NULL, 0, 1664 CTL_CREATE, CTL_EOL); 1665 } 1666 1667 void 1668 etherinit(void) 1669 { 1670 1671 mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET); 1672 ether_sysctl_setup(NULL); 1673 } 1674
Indexes created Tue Sep 23 07:09:52 GMT 2025