if_ethersubr.c revision 1.242.6.8
1 /* $NetBSD: if_ethersubr.c,v 1.242.6.8 2021/10/25 18:16:07 martin 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 1.242.6.8 2021/10/25 18:16:07 martin Exp $"); 65 66 #ifdef _KERNEL_OPT 67 #include "opt_inet.h" 68 #include "opt_atalk.h" 69 #include "opt_mbuftrace.h" 70 #include "opt_mpls.h" 71 #include "opt_gateway.h" 72 #include "opt_pppoe.h" 73 #include "opt_net_mpsafe.h" 74 #endif 75 76 #include "vlan.h" 77 #include "pppoe.h" 78 #include "bridge.h" 79 #include "arp.h" 80 #include "agr.h" 81 82 #include <sys/sysctl.h> 83 #include <sys/mbuf.h> 84 #include <sys/mutex.h> 85 #include <sys/ioctl.h> 86 #include <sys/errno.h> 87 #include <sys/device.h> 88 #include <sys/rnd.h> 89 #include <sys/rndsource.h> 90 #include <sys/cpu.h> 91 #include <sys/kmem.h> 92 93 #include <net/if.h> 94 #include <net/netisr.h> 95 #include <net/route.h> 96 #include <net/if_llc.h> 97 #include <net/if_dl.h> 98 #include <net/if_types.h> 99 #include <net/pktqueue.h> 100 101 #include <net/if_media.h> 102 #include <dev/mii/mii.h> 103 #include <dev/mii/miivar.h> 104 105 #if NARP == 0 106 /* 107 * XXX there should really be a way to issue this warning from within config(8) 108 */ 109 #error You have included NETATALK or a pseudo-device in your configuration that depends on the presence of ethernet interfaces, but have no such interfaces configured. Check if you really need pseudo-device bridge, pppoe, vlan or options NETATALK. 110 #endif 111 112 #include <net/bpf.h> 113 114 #include <net/if_ether.h> 115 #include <net/if_vlanvar.h> 116 117 #if NPPPOE > 0 118 #include <net/if_pppoe.h> 119 #endif 120 121 #if NAGR > 0 122 #include <net/agr/ieee8023_slowprotocols.h> /* XXX */ 123 #include <net/agr/ieee8023ad.h> 124 #include <net/agr/if_agrvar.h> 125 #endif 126 127 #if NBRIDGE > 0 128 #include <net/if_bridgevar.h> 129 #endif 130 131 #include <netinet/in.h> 132 #ifdef INET 133 #include <netinet/in_var.h> 134 #endif 135 #include <netinet/if_inarp.h> 136 137 #ifdef INET6 138 #ifndef INET 139 #include <netinet/in.h> 140 #endif 141 #include <netinet6/in6_var.h> 142 #include <netinet6/nd6.h> 143 #endif 144 145 146 #include "carp.h" 147 #if NCARP > 0 148 #include <netinet/ip_carp.h> 149 #endif 150 151 #ifdef NETATALK 152 #include <netatalk/at.h> 153 #include <netatalk/at_var.h> 154 #include <netatalk/at_extern.h> 155 156 #define llc_snap_org_code llc_un.type_snap.org_code 157 #define llc_snap_ether_type llc_un.type_snap.ether_type 158 159 extern u_char at_org_code[3]; 160 extern u_char aarp_org_code[3]; 161 #endif /* NETATALK */ 162 163 #ifdef MPLS 164 #include <netmpls/mpls.h> 165 #include <netmpls/mpls_var.h> 166 #endif 167 168 static struct timeval bigpktppslim_last; 169 static int bigpktppslim = 2; /* XXX */ 170 static int bigpktpps_count; 171 static kmutex_t bigpktpps_lock __cacheline_aligned; 172 173 174 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = 175 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 176 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] = 177 { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 }; 178 #define senderr(e) { error = (e); goto bad;} 179 180 static int ether_output(struct ifnet *, struct mbuf *, 181 const struct sockaddr *, const struct rtentry *); 182 183 /* 184 * Ethernet output routine. 185 * Encapsulate a packet of type family for the local net. 186 * Assumes that ifp is actually pointer to ethercom structure. 187 */ 188 static int 189 ether_output(struct ifnet * const ifp0, struct mbuf * const m0, 190 const struct sockaddr * const dst, 191 const struct rtentry *rt) 192 { 193 uint16_t etype = 0; 194 int error = 0, hdrcmplt = 0; 195 uint8_t esrc[6], edst[6]; 196 struct mbuf *m = m0; 197 struct mbuf *mcopy = NULL; 198 struct ether_header *eh; 199 struct ifnet *ifp = ifp0; 200 #ifdef INET 201 struct arphdr *ah; 202 #endif /* INET */ 203 #ifdef NETATALK 204 struct at_ifaddr *aa; 205 #endif /* NETATALK */ 206 207 #ifdef MBUFTRACE 208 m_claimm(m, ifp->if_mowner); 209 #endif 210 211 #if NCARP > 0 212 if (ifp->if_type == IFT_CARP) { 213 struct ifaddr *ifa; 214 int s = pserialize_read_enter(); 215 216 /* loop back if this is going to the carp interface */ 217 if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP && 218 (ifa = ifa_ifwithaddr(dst)) != NULL) { 219 if (ifa->ifa_ifp == ifp0) { 220 pserialize_read_exit(s); 221 return looutput(ifp0, m, dst, rt); 222 } 223 } 224 pserialize_read_exit(s); 225 226 ifp = ifp->if_carpdev; 227 /* ac = (struct arpcom *)ifp; */ 228 229 if ((ifp0->if_flags & (IFF_UP | IFF_RUNNING)) != 230 (IFF_UP | IFF_RUNNING)) 231 senderr(ENETDOWN); 232 } 233 #endif /* NCARP > 0 */ 234 235 if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING)) 236 senderr(ENETDOWN); 237 238 switch (dst->sa_family) { 239 240 #ifdef INET 241 case AF_INET: 242 if (m->m_flags & M_BCAST) 243 (void)memcpy(edst, etherbroadcastaddr, sizeof(edst)); 244 else if (m->m_flags & M_MCAST) 245 ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst); 246 else if ((error = arpresolve(ifp, rt, m, dst, edst, 247 sizeof(edst))) != 0) { 248 return error == EWOULDBLOCK ? 0 : error; 249 } 250 /* If broadcasting on a simplex interface, loopback a copy */ 251 if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX)) 252 mcopy = m_copy(m, 0, (int)M_COPYALL); 253 etype = htons(ETHERTYPE_IP); 254 break; 255 256 case AF_ARP: 257 ah = mtod(m, struct arphdr *); 258 if (m->m_flags & M_BCAST) 259 (void)memcpy(edst, etherbroadcastaddr, sizeof(edst)); 260 else { 261 void *tha = ar_tha(ah); 262 263 if (tha == NULL) { 264 /* fake with ARPHDR_IEEE1394 */ 265 m_freem(m); 266 return 0; 267 } 268 memcpy(edst, tha, sizeof(edst)); 269 } 270 271 ah->ar_hrd = htons(ARPHRD_ETHER); 272 273 switch (ntohs(ah->ar_op)) { 274 case ARPOP_REVREQUEST: 275 case ARPOP_REVREPLY: 276 etype = htons(ETHERTYPE_REVARP); 277 break; 278 279 case ARPOP_REQUEST: 280 case ARPOP_REPLY: 281 default: 282 etype = htons(ETHERTYPE_ARP); 283 } 284 285 break; 286 #endif 287 #ifdef INET6 288 case AF_INET6: 289 if (m->m_flags & M_BCAST) 290 (void)memcpy(edst, etherbroadcastaddr, sizeof(edst)); 291 else if (m->m_flags & M_MCAST) { 292 ETHER_MAP_IPV6_MULTICAST(&satocsin6(dst)->sin6_addr, 293 edst); 294 } else { 295 error = nd6_resolve(ifp, rt, m, dst, edst, 296 sizeof(edst)); 297 if (error != 0) 298 return error == EWOULDBLOCK ? 0 : error; 299 } 300 etype = htons(ETHERTYPE_IPV6); 301 break; 302 #endif 303 #ifdef NETATALK 304 case AF_APPLETALK: { 305 struct ifaddr *ifa; 306 int s; 307 308 KERNEL_LOCK(1, NULL); 309 if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) { 310 #ifdef NETATALKDEBUG 311 printf("aarpresolv failed\n"); 312 #endif /* NETATALKDEBUG */ 313 KERNEL_UNLOCK_ONE(NULL); 314 return (0); 315 } 316 /* 317 * ifaddr is the first thing in at_ifaddr 318 */ 319 s = pserialize_read_enter(); 320 ifa = at_ifawithnet((const struct sockaddr_at *)dst, ifp); 321 if (ifa == NULL) { 322 pserialize_read_exit(s); 323 KERNEL_UNLOCK_ONE(NULL); 324 goto bad; 325 } 326 aa = (struct at_ifaddr *)ifa; 327 328 /* 329 * In the phase 2 case, we need to prepend an mbuf for the 330 * llc header. Since we must preserve the value of m, 331 * which is passed to us by value, we m_copy() the first 332 * mbuf, and use it for our llc header. 333 */ 334 if (aa->aa_flags & AFA_PHASE2) { 335 struct llc llc; 336 337 M_PREPEND(m, sizeof(struct llc), M_DONTWAIT); 338 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 339 llc.llc_control = LLC_UI; 340 memcpy(llc.llc_snap_org_code, at_org_code, 341 sizeof(llc.llc_snap_org_code)); 342 llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK); 343 memcpy(mtod(m, void *), &llc, sizeof(struct llc)); 344 } else { 345 etype = htons(ETHERTYPE_ATALK); 346 } 347 pserialize_read_exit(s); 348 KERNEL_UNLOCK_ONE(NULL); 349 break; 350 } 351 #endif /* NETATALK */ 352 case pseudo_AF_HDRCMPLT: 353 hdrcmplt = 1; 354 memcpy(esrc, 355 ((const struct ether_header *)dst->sa_data)->ether_shost, 356 sizeof(esrc)); 357 /* FALLTHROUGH */ 358 359 case AF_UNSPEC: 360 memcpy(edst, 361 ((const struct ether_header *)dst->sa_data)->ether_dhost, 362 sizeof(edst)); 363 /* AF_UNSPEC doesn't swap the byte order of the ether_type. */ 364 etype = ((const struct ether_header *)dst->sa_data)->ether_type; 365 break; 366 367 default: 368 printf("%s: can't handle af%d\n", ifp->if_xname, 369 dst->sa_family); 370 senderr(EAFNOSUPPORT); 371 } 372 373 #ifdef MPLS 374 KERNEL_LOCK(1, NULL); 375 { 376 struct m_tag *mtag; 377 mtag = m_tag_find(m, PACKET_TAG_MPLS, NULL); 378 if (mtag != NULL) { 379 /* Having the tag itself indicates it's MPLS */ 380 etype = htons(ETHERTYPE_MPLS); 381 m_tag_delete(m, mtag); 382 } 383 } 384 KERNEL_UNLOCK_ONE(NULL); 385 #endif 386 387 if (mcopy) 388 (void)looutput(ifp, mcopy, dst, rt); 389 390 /* If no ether type is set, this must be a 802.2 formatted packet. 391 */ 392 if (etype == 0) 393 etype = htons(m->m_pkthdr.len); 394 /* 395 * Add local net header. If no space in first mbuf, 396 * allocate another. 397 */ 398 M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT); 399 if (m == 0) 400 senderr(ENOBUFS); 401 eh = mtod(m, struct ether_header *); 402 /* Note: etype is already in network byte order. */ 403 (void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type)); 404 memcpy(eh->ether_dhost, edst, sizeof(edst)); 405 if (hdrcmplt) 406 memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost)); 407 else 408 memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl), 409 sizeof(eh->ether_shost)); 410 411 #if NCARP > 0 412 if (ifp0 != ifp && ifp0->if_type == IFT_CARP) { 413 memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl), 414 sizeof(eh->ether_shost)); 415 } 416 #endif /* NCARP > 0 */ 417 418 if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0) 419 return (error); 420 if (m == NULL) 421 return (0); 422 423 #if NBRIDGE > 0 424 /* 425 * Bridges require special output handling. 426 */ 427 if (ifp->if_bridge) 428 return (bridge_output(ifp, m, NULL, NULL)); 429 #endif 430 431 #if NCARP > 0 432 if (ifp != ifp0) 433 ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN; 434 #endif /* NCARP > 0 */ 435 436 #ifdef ALTQ 437 KERNEL_LOCK(1, NULL); 438 /* 439 * If ALTQ is enabled on the parent interface, do 440 * classification; the queueing discipline might not 441 * require classification, but might require the 442 * address family/header pointer in the pktattr. 443 */ 444 if (ALTQ_IS_ENABLED(&ifp->if_snd)) 445 altq_etherclassify(&ifp->if_snd, m); 446 KERNEL_UNLOCK_ONE(NULL); 447 #endif 448 return ifq_enqueue(ifp, m); 449 450 bad: 451 if (m) 452 m_freem(m); 453 return (error); 454 } 455 456 #ifdef ALTQ 457 /* 458 * This routine is a slight hack to allow a packet to be classified 459 * if the Ethernet headers are present. It will go away when ALTQ's 460 * classification engine understands link headers. 461 */ 462 void 463 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m) 464 { 465 struct ether_header *eh; 466 struct mbuf *mtop = m; 467 uint16_t ether_type; 468 int hlen, af, hdrsize; 469 void *hdr; 470 471 KASSERT((mtop->m_flags & M_PKTHDR) != 0); 472 473 hlen = ETHER_HDR_LEN; 474 eh = mtod(m, struct ether_header *); 475 476 ether_type = htons(eh->ether_type); 477 478 if (ether_type < ETHERMTU) { 479 /* LLC/SNAP */ 480 struct llc *llc = (struct llc *)(eh + 1); 481 hlen += 8; 482 483 if (m->m_len < hlen || 484 llc->llc_dsap != LLC_SNAP_LSAP || 485 llc->llc_ssap != LLC_SNAP_LSAP || 486 llc->llc_control != LLC_UI) { 487 /* Not SNAP. */ 488 goto bad; 489 } 490 491 ether_type = htons(llc->llc_un.type_snap.ether_type); 492 } 493 494 switch (ether_type) { 495 case ETHERTYPE_IP: 496 af = AF_INET; 497 hdrsize = 20; /* sizeof(struct ip) */ 498 break; 499 500 case ETHERTYPE_IPV6: 501 af = AF_INET6; 502 hdrsize = 40; /* sizeof(struct ip6_hdr) */ 503 break; 504 505 default: 506 af = AF_UNSPEC; 507 hdrsize = 0; 508 break; 509 } 510 511 while (m->m_len <= hlen) { 512 hlen -= m->m_len; 513 m = m->m_next; 514 if (m == NULL) 515 goto bad; 516 } 517 518 if (m->m_len < (hlen + hdrsize)) { 519 /* 520 * protocol header not in a single mbuf. 521 * We can't cope with this situation right 522 * now (but it shouldn't ever happen, really, anyhow). 523 */ 524 #ifdef DEBUG 525 printf("altq_etherclassify: headers span multiple mbufs: " 526 "%d < %d\n", m->m_len, (hlen + hdrsize)); 527 #endif 528 goto bad; 529 } 530 531 m->m_data += hlen; 532 m->m_len -= hlen; 533 534 hdr = mtod(m, void *); 535 536 if (ALTQ_NEEDS_CLASSIFY(ifq)) { 537 mtop->m_pkthdr.pattr_class = 538 (*ifq->altq_classify)(ifq->altq_clfier, m, af); 539 } 540 mtop->m_pkthdr.pattr_af = af; 541 mtop->m_pkthdr.pattr_hdr = hdr; 542 543 m->m_data -= hlen; 544 m->m_len += hlen; 545 546 return; 547 548 bad: 549 mtop->m_pkthdr.pattr_class = NULL; 550 mtop->m_pkthdr.pattr_hdr = NULL; 551 mtop->m_pkthdr.pattr_af = AF_UNSPEC; 552 } 553 #endif /* ALTQ */ 554 555 /* 556 * Process a received Ethernet packet; 557 * the packet is in the mbuf chain m with 558 * the ether header. 559 */ 560 void 561 ether_input(struct ifnet *ifp, struct mbuf *m) 562 { 563 struct ethercom *ec = (struct ethercom *) ifp; 564 pktqueue_t *pktq = NULL; 565 struct ifqueue *inq = NULL; 566 uint16_t etype; 567 struct ether_header *eh; 568 size_t ehlen; 569 static int earlypkts; 570 int isr = 0; 571 #if defined (LLC) || defined(NETATALK) 572 struct llc *l; 573 #endif 574 575 KASSERT(!cpu_intr_p()); 576 577 if ((ifp->if_flags & IFF_UP) == 0) { 578 m_freem(m); 579 return; 580 } 581 582 #ifdef MBUFTRACE 583 m_claimm(m, &ec->ec_rx_mowner); 584 #endif 585 eh = mtod(m, struct ether_header *); 586 etype = ntohs(eh->ether_type); 587 ehlen = sizeof(*eh); 588 589 if (__predict_false(earlypkts < 100 || !rnd_initial_entropy)) { 590 rnd_add_data(NULL, eh, ehlen, 0); 591 earlypkts++; 592 } 593 594 /* 595 * Determine if the packet is within its size limits. 596 */ 597 if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len > 598 ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) { 599 mutex_enter(&bigpktpps_lock); 600 if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count, 601 bigpktppslim)) { 602 printf("%s: discarding oversize frame (len=%d)\n", 603 ifp->if_xname, m->m_pkthdr.len); 604 } 605 mutex_exit(&bigpktpps_lock); 606 ifp->if_iqdrops++; 607 m_freem(m); 608 return; 609 } 610 611 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 612 /* 613 * If this is not a simplex interface, drop the packet 614 * if it came from us. 615 */ 616 if ((ifp->if_flags & IFF_SIMPLEX) == 0 && 617 memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost, 618 ETHER_ADDR_LEN) == 0) { 619 m_freem(m); 620 return; 621 } 622 623 if (memcmp(etherbroadcastaddr, 624 eh->ether_dhost, ETHER_ADDR_LEN) == 0) 625 m->m_flags |= M_BCAST; 626 else 627 m->m_flags |= M_MCAST; 628 ifp->if_imcasts++; 629 } 630 631 /* If the CRC is still on the packet, trim it off. */ 632 if (m->m_flags & M_HASFCS) { 633 m_adj(m, -ETHER_CRC_LEN); 634 m->m_flags &= ~M_HASFCS; 635 } 636 637 ifp->if_ibytes += m->m_pkthdr.len; 638 639 #if NCARP > 0 640 if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) { 641 /* 642 * clear M_PROMISC, in case the packets comes from a 643 * vlan 644 */ 645 m->m_flags &= ~M_PROMISC; 646 if (carp_input(m, (uint8_t *)&eh->ether_shost, 647 (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0) 648 return; 649 } 650 #endif /* NCARP > 0 */ 651 652 if ((m->m_flags & (M_BCAST | M_MCAST | M_PROMISC)) == 0 && 653 (ifp->if_flags & IFF_PROMISC) != 0 && 654 memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost, 655 ETHER_ADDR_LEN) != 0) { 656 m->m_flags |= M_PROMISC; 657 } 658 659 if ((m->m_flags & M_PROMISC) == 0) { 660 if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0) 661 return; 662 if (m == NULL) 663 return; 664 665 eh = mtod(m, struct ether_header *); 666 etype = ntohs(eh->ether_type); 667 ehlen = sizeof(*eh); 668 } 669 670 #if NAGR > 0 671 if (ifp->if_agrprivate && 672 __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) { 673 m->m_flags &= ~M_PROMISC; 674 agr_input(ifp, m); 675 return; 676 } 677 #endif /* NAGR > 0 */ 678 679 /* 680 * If VLANs are configured on the interface, check to 681 * see if the device performed the decapsulation and 682 * provided us with the tag. 683 */ 684 if (ec->ec_nvlans && vlan_has_tag(m)) { 685 #if NVLAN > 0 686 /* 687 * vlan_input() will either recursively call ether_input() 688 * or drop the packet. 689 */ 690 vlan_input(ifp, m); 691 #else 692 m_freem(m); 693 #endif 694 return; 695 } 696 697 /* 698 * Handle protocols that expect to have the Ethernet header 699 * (and possibly FCS) intact. 700 */ 701 switch (etype) { 702 case ETHERTYPE_VLAN: { 703 struct ether_vlan_header *evl = (void *)eh; 704 /* 705 * If there is a tag of 0, then the VLAN header was probably 706 * just being used to store the priority. Extract the ether 707 * type, and if IP or IPV6, let them deal with it. 708 */ 709 if (m->m_len <= sizeof(*evl) 710 && EVL_VLANOFTAG((ntohs(evl->evl_tag)) == 0) { 711 etype = ntohs(evl->evl_proto); 712 ehlen = sizeof(*evl); 713 if ((m->m_flags & M_PROMISC) == 0 714 && (etype == ETHERTYPE_IP 715 || etype == ETHERTYPE_IPV6)) 716 break; 717 } 718 #if NVLAN > 0 719 /* 720 * vlan_input() will either recursively call ether_input() 721 * or drop the packet. 722 */ 723 if (((struct ethercom *)ifp)->ec_nvlans != 0) 724 vlan_input(ifp, m); 725 else 726 #endif /* NVLAN > 0 */ 727 m_freem(m); 728 return; 729 } 730 #if NPPPOE > 0 731 case ETHERTYPE_PPPOEDISC: 732 pppoedisc_input(ifp, m); 733 return; 734 case ETHERTYPE_PPPOE: 735 pppoe_input(ifp, m); 736 return; 737 #endif /* NPPPOE > 0 */ 738 case ETHERTYPE_SLOWPROTOCOLS: { 739 uint8_t subtype; 740 741 #if defined(DIAGNOSTIC) 742 if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) { 743 panic("ether_input: too short slow protocol packet"); 744 } 745 #endif 746 m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype); 747 switch (subtype) { 748 #if NAGR > 0 749 case SLOWPROTOCOLS_SUBTYPE_LACP: 750 if (ifp->if_agrprivate) { 751 ieee8023ad_lacp_input(ifp, m); 752 return; 753 } 754 break; 755 756 case SLOWPROTOCOLS_SUBTYPE_MARKER: 757 if (ifp->if_agrprivate) { 758 ieee8023ad_marker_input(ifp, m); 759 return; 760 } 761 break; 762 #endif /* NAGR > 0 */ 763 default: 764 if (subtype == 0 || subtype > 10) { 765 /* illegal value */ 766 m_freem(m); 767 return; 768 } 769 /* unknown subtype */ 770 break; 771 } 772 /* FALLTHROUGH */ 773 } 774 default: 775 if (m->m_flags & M_PROMISC) { 776 m_freem(m); 777 return; 778 } 779 } 780 781 /* If the CRC is still on the packet, trim it off. */ 782 if (m->m_flags & M_HASFCS) { 783 m_adj(m, -ETHER_CRC_LEN); 784 m->m_flags &= ~M_HASFCS; 785 } 786 787 if (etype > ETHERMTU + sizeof (struct ether_header)) { 788 /* Strip off the Ethernet header. */ 789 m_adj(m, ehlen); 790 791 switch (etype) { 792 #ifdef INET 793 case ETHERTYPE_IP: 794 #ifdef GATEWAY 795 if (ipflow_fastforward(m)) 796 return; 797 #endif 798 pktq = ip_pktq; 799 break; 800 801 case ETHERTYPE_ARP: 802 isr = NETISR_ARP; 803 inq = &arpintrq; 804 break; 805 806 case ETHERTYPE_REVARP: 807 revarpinput(m); /* XXX queue? */ 808 return; 809 #endif 810 #ifdef INET6 811 case ETHERTYPE_IPV6: 812 if (__predict_false(!in6_present)) { 813 m_freem(m); 814 return; 815 } 816 #ifdef GATEWAY 817 if (ip6flow_fastforward(&m)) 818 return; 819 #endif 820 pktq = ip6_pktq; 821 break; 822 #endif 823 #ifdef NETATALK 824 case ETHERTYPE_ATALK: 825 isr = NETISR_ATALK; 826 inq = &atintrq1; 827 break; 828 case ETHERTYPE_AARP: 829 /* probably this should be done with a NETISR as well */ 830 aarpinput(ifp, m); /* XXX */ 831 return; 832 #endif /* NETATALK */ 833 #ifdef MPLS 834 case ETHERTYPE_MPLS: 835 isr = NETISR_MPLS; 836 inq = &mplsintrq; 837 break; 838 #endif 839 default: 840 m_freem(m); 841 return; 842 } 843 } else { 844 #if defined (LLC) || defined (NETATALK) 845 if (m->m_len < ehlen + sizeof(struct llc)) { 846 goto dropanyway; 847 } 848 l = (struct llc *)(eh+1); 849 switch (l->llc_dsap) { 850 #ifdef NETATALK 851 case LLC_SNAP_LSAP: 852 switch (l->llc_control) { 853 case LLC_UI: 854 if (l->llc_ssap != LLC_SNAP_LSAP) { 855 goto dropanyway; 856 } 857 858 if (memcmp(&(l->llc_snap_org_code)[0], 859 at_org_code, sizeof(at_org_code)) == 0 && 860 ntohs(l->llc_snap_ether_type) == 861 ETHERTYPE_ATALK) { 862 inq = &atintrq2; 863 m_adj(m, sizeof(struct ether_header) 864 + sizeof(struct llc)); 865 isr = NETISR_ATALK; 866 break; 867 } 868 869 if (memcmp(&(l->llc_snap_org_code)[0], 870 aarp_org_code, 871 sizeof(aarp_org_code)) == 0 && 872 ntohs(l->llc_snap_ether_type) == 873 ETHERTYPE_AARP) { 874 m_adj( m, sizeof(struct ether_header) 875 + sizeof(struct llc)); 876 aarpinput(ifp, m); /* XXX */ 877 return; 878 } 879 880 default: 881 goto dropanyway; 882 } 883 break; 884 #endif 885 dropanyway: 886 default: 887 m_freem(m); 888 return; 889 } 890 #else /* LLC || NETATALK */ 891 m_freem(m); 892 return; 893 #endif /* LLC || NETATALK */ 894 } 895 896 if (__predict_true(pktq)) { 897 #ifdef NET_MPSAFE 898 const u_int h = curcpu()->ci_index; 899 #else 900 const uint32_t h = pktq_rps_hash(m); 901 #endif 902 if (__predict_false(!pktq_enqueue(pktq, m, h))) { 903 m_freem(m); 904 } 905 return; 906 } 907 908 if (__predict_false(!inq)) { 909 /* Should not happen. */ 910 m_freem(m); 911 return; 912 } 913 914 IFQ_LOCK(inq); 915 if (IF_QFULL(inq)) { 916 IF_DROP(inq); 917 IFQ_UNLOCK(inq); 918 m_freem(m); 919 } else { 920 IF_ENQUEUE(inq, m); 921 IFQ_UNLOCK(inq); 922 schednetisr(isr); 923 } 924 } 925 926 /* 927 * Convert Ethernet address to printable (loggable) representation. 928 */ 929 char * 930 ether_sprintf(const u_char *ap) 931 { 932 static char etherbuf[3 * ETHER_ADDR_LEN]; 933 return ether_snprintf(etherbuf, sizeof(etherbuf), ap); 934 } 935 936 char * 937 ether_snprintf(char *buf, size_t len, const u_char *ap) 938 { 939 char *cp = buf; 940 size_t i; 941 942 for (i = 0; i < len / 3; i++) { 943 *cp++ = hexdigits[*ap >> 4]; 944 *cp++ = hexdigits[*ap++ & 0xf]; 945 *cp++ = ':'; 946 } 947 *--cp = '\0'; 948 return buf; 949 } 950 951 /* 952 * Perform common duties while attaching to interface list 953 */ 954 void 955 ether_ifattach(struct ifnet *ifp, const uint8_t *lla) 956 { 957 struct ethercom *ec = (struct ethercom *)ifp; 958 959 ifp->if_type = IFT_ETHER; 960 ifp->if_hdrlen = ETHER_HDR_LEN; 961 ifp->if_dlt = DLT_EN10MB; 962 ifp->if_mtu = ETHERMTU; 963 ifp->if_output = ether_output; 964 ifp->_if_input = ether_input; 965 if (ifp->if_baudrate == 0) 966 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 967 968 if (lla != NULL) 969 if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla)); 970 971 LIST_INIT(&ec->ec_multiaddrs); 972 ec->ec_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NET); 973 ifp->if_broadcastaddr = etherbroadcastaddr; 974 bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header)); 975 #ifdef MBUFTRACE 976 strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname, 977 sizeof(ec->ec_tx_mowner.mo_name)); 978 strlcpy(ec->ec_tx_mowner.mo_descr, "tx", 979 sizeof(ec->ec_tx_mowner.mo_descr)); 980 strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname, 981 sizeof(ec->ec_rx_mowner.mo_name)); 982 strlcpy(ec->ec_rx_mowner.mo_descr, "rx", 983 sizeof(ec->ec_rx_mowner.mo_descr)); 984 MOWNER_ATTACH(&ec->ec_tx_mowner); 985 MOWNER_ATTACH(&ec->ec_rx_mowner); 986 ifp->if_mowner = &ec->ec_tx_mowner; 987 #endif 988 } 989 990 void 991 ether_ifdetach(struct ifnet *ifp) 992 { 993 struct ethercom *ec = (void *) ifp; 994 struct ether_multi *enm; 995 996 /* 997 * Prevent further calls to ioctl (for example turning off 998 * promiscuous mode from the bridge code), which eventually can 999 * call if_init() which can cause panics because the interface 1000 * is in the process of being detached. Return device not configured 1001 * instead. 1002 */ 1003 ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio; 1004 1005 #if NBRIDGE > 0 1006 if (ifp->if_bridge) 1007 bridge_ifdetach(ifp); 1008 #endif 1009 bpf_detach(ifp); 1010 #if NVLAN > 0 1011 if (ec->ec_nvlans) 1012 vlan_ifdetach(ifp); 1013 #endif 1014 1015 ETHER_LOCK(ec); 1016 while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) { 1017 LIST_REMOVE(enm, enm_list); 1018 kmem_free(enm, sizeof(*enm)); 1019 ec->ec_multicnt--; 1020 } 1021 ETHER_UNLOCK(ec); 1022 1023 mutex_obj_free(ec->ec_lock); 1024 1025 ifp->if_mowner = NULL; 1026 MOWNER_DETACH(&ec->ec_rx_mowner); 1027 MOWNER_DETACH(&ec->ec_tx_mowner); 1028 } 1029 1030 #if 0 1031 /* 1032 * This is for reference. We have a table-driven version 1033 * of the little-endian crc32 generator, which is faster 1034 * than the double-loop. 1035 */ 1036 uint32_t 1037 ether_crc32_le(const uint8_t *buf, size_t len) 1038 { 1039 uint32_t c, crc, carry; 1040 size_t i, j; 1041 1042 crc = 0xffffffffU; /* initial value */ 1043 1044 for (i = 0; i < len; i++) { 1045 c = buf[i]; 1046 for (j = 0; j < 8; j++) { 1047 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1048 crc >>= 1; 1049 c >>= 1; 1050 if (carry) 1051 crc = (crc ^ ETHER_CRC_POLY_LE); 1052 } 1053 } 1054 1055 return (crc); 1056 } 1057 #else 1058 uint32_t 1059 ether_crc32_le(const uint8_t *buf, size_t len) 1060 { 1061 static const uint32_t crctab[] = { 1062 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1063 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1064 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1065 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1066 }; 1067 uint32_t crc; 1068 size_t i; 1069 1070 crc = 0xffffffffU; /* initial value */ 1071 1072 for (i = 0; i < len; i++) { 1073 crc ^= buf[i]; 1074 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1075 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1076 } 1077 1078 return (crc); 1079 } 1080 #endif 1081 1082 uint32_t 1083 ether_crc32_be(const uint8_t *buf, size_t len) 1084 { 1085 uint32_t c, crc, carry; 1086 size_t i, j; 1087 1088 crc = 0xffffffffU; /* initial value */ 1089 1090 for (i = 0; i < len; i++) { 1091 c = buf[i]; 1092 for (j = 0; j < 8; j++) { 1093 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1094 crc <<= 1; 1095 c >>= 1; 1096 if (carry) 1097 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1098 } 1099 } 1100 1101 return (crc); 1102 } 1103 1104 #ifdef INET 1105 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] = 1106 { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 }; 1107 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] = 1108 { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff }; 1109 #endif 1110 #ifdef INET6 1111 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] = 1112 { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 }; 1113 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] = 1114 { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff }; 1115 #endif 1116 1117 /* 1118 * ether_aton implementation, not using a static buffer. 1119 */ 1120 int 1121 ether_aton_r(u_char *dest, size_t len, const char *str) 1122 { 1123 const u_char *cp = (const void *)str; 1124 u_char *ep; 1125 1126 #define atox(c) (((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10)) 1127 1128 if (len < ETHER_ADDR_LEN) 1129 return ENOSPC; 1130 1131 ep = dest + ETHER_ADDR_LEN; 1132 1133 while (*cp) { 1134 if (!isxdigit(*cp)) 1135 return EINVAL; 1136 1137 *dest = atox(*cp); 1138 cp++; 1139 if (isxdigit(*cp)) { 1140 *dest = (*dest << 4) | atox(*cp); 1141 cp++; 1142 } 1143 dest++; 1144 1145 if (dest == ep) 1146 return (*cp == '\0') ? 0 : ENAMETOOLONG; 1147 1148 switch (*cp) { 1149 case ':': 1150 case '-': 1151 case '.': 1152 cp++; 1153 break; 1154 } 1155 } 1156 return ENOBUFS; 1157 } 1158 1159 /* 1160 * Convert a sockaddr into an Ethernet address or range of Ethernet 1161 * addresses. 1162 */ 1163 int 1164 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN], 1165 uint8_t addrhi[ETHER_ADDR_LEN]) 1166 { 1167 #ifdef INET 1168 const struct sockaddr_in *sin; 1169 #endif /* INET */ 1170 #ifdef INET6 1171 const struct sockaddr_in6 *sin6; 1172 #endif /* INET6 */ 1173 1174 switch (sa->sa_family) { 1175 1176 case AF_UNSPEC: 1177 memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN); 1178 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1179 break; 1180 1181 #ifdef INET 1182 case AF_INET: 1183 sin = satocsin(sa); 1184 if (sin->sin_addr.s_addr == INADDR_ANY) { 1185 /* 1186 * An IP address of INADDR_ANY means listen to 1187 * or stop listening to all of the Ethernet 1188 * multicast addresses used for IP. 1189 * (This is for the sake of IP multicast routers.) 1190 */ 1191 memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN); 1192 memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN); 1193 } else { 1194 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo); 1195 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1196 } 1197 break; 1198 #endif 1199 #ifdef INET6 1200 case AF_INET6: 1201 sin6 = satocsin6(sa); 1202 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1203 /* 1204 * An IP6 address of 0 means listen to or stop 1205 * listening to all of the Ethernet multicast 1206 * address used for IP6. 1207 * (This is used for multicast routers.) 1208 */ 1209 memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN); 1210 memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN); 1211 } else { 1212 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo); 1213 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1214 } 1215 break; 1216 #endif 1217 1218 default: 1219 return EAFNOSUPPORT; 1220 } 1221 return 0; 1222 } 1223 1224 /* 1225 * Add an Ethernet multicast address or range of addresses to the list for a 1226 * given interface. 1227 */ 1228 int 1229 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec) 1230 { 1231 struct ether_multi *enm, *_enm; 1232 u_char addrlo[ETHER_ADDR_LEN]; 1233 u_char addrhi[ETHER_ADDR_LEN]; 1234 int error = 0; 1235 1236 /* Allocate out of lock */ 1237 enm = kmem_alloc(sizeof(*enm), KM_SLEEP); 1238 1239 ETHER_LOCK(ec); 1240 error = ether_multiaddr(sa, addrlo, addrhi); 1241 if (error != 0) 1242 goto out; 1243 1244 /* 1245 * Verify that we have valid Ethernet multicast addresses. 1246 */ 1247 if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) { 1248 error = EINVAL; 1249 goto out; 1250 } 1251 1252 /* 1253 * See if the address range is already in the list. 1254 */ 1255 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, _enm); 1256 if (_enm != NULL) { 1257 /* 1258 * Found it; just increment the reference count. 1259 */ 1260 ++_enm->enm_refcount; 1261 error = 0; 1262 goto out; 1263 } 1264 1265 /* 1266 * Link a new multicast record into the interface's multicast list. 1267 */ 1268 memcpy(enm->enm_addrlo, addrlo, ETHER_ADDR_LEN); 1269 memcpy(enm->enm_addrhi, addrhi, ETHER_ADDR_LEN); 1270 enm->enm_refcount = 1; 1271 LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list); 1272 ec->ec_multicnt++; 1273 1274 /* 1275 * Return ENETRESET to inform the driver that the list has changed 1276 * and its reception filter should be adjusted accordingly. 1277 */ 1278 error = ENETRESET; 1279 enm = NULL; 1280 1281 out: 1282 ETHER_UNLOCK(ec); 1283 if (enm != NULL) 1284 kmem_free(enm, sizeof(*enm)); 1285 return error; 1286 } 1287 1288 /* 1289 * Delete a multicast address record. 1290 */ 1291 int 1292 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec) 1293 { 1294 struct ether_multi *enm; 1295 u_char addrlo[ETHER_ADDR_LEN]; 1296 u_char addrhi[ETHER_ADDR_LEN]; 1297 int error; 1298 1299 ETHER_LOCK(ec); 1300 error = ether_multiaddr(sa, addrlo, addrhi); 1301 if (error != 0) 1302 goto error; 1303 1304 /* 1305 * Look up the address in our list. 1306 */ 1307 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm); 1308 if (enm == NULL) { 1309 error = ENXIO; 1310 goto error; 1311 } 1312 if (--enm->enm_refcount != 0) { 1313 /* 1314 * Still some claims to this record. 1315 */ 1316 error = 0; 1317 goto error; 1318 } 1319 1320 /* 1321 * No remaining claims to this record; unlink and free it. 1322 */ 1323 LIST_REMOVE(enm, enm_list); 1324 ec->ec_multicnt--; 1325 ETHER_UNLOCK(ec); 1326 kmem_free(enm, sizeof(*enm)); 1327 1328 /* 1329 * Return ENETRESET to inform the driver that the list has changed 1330 * and its reception filter should be adjusted accordingly. 1331 */ 1332 return ENETRESET; 1333 1334 error: 1335 ETHER_UNLOCK(ec); 1336 return error; 1337 } 1338 1339 void 1340 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb) 1341 { 1342 ec->ec_ifflags_cb = cb; 1343 } 1344 1345 /* 1346 * Common ioctls for Ethernet interfaces. Note, we must be 1347 * called at splnet(). 1348 */ 1349 int 1350 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data) 1351 { 1352 struct ethercom *ec = (void *) ifp; 1353 struct eccapreq *eccr; 1354 struct ifreq *ifr = (struct ifreq *)data; 1355 struct if_laddrreq *iflr = data; 1356 const struct sockaddr_dl *sdl; 1357 static const uint8_t zero[ETHER_ADDR_LEN]; 1358 int error; 1359 1360 switch (cmd) { 1361 case SIOCINITIFADDR: 1362 { 1363 struct ifaddr *ifa = (struct ifaddr *)data; 1364 if (ifa->ifa_addr->sa_family != AF_LINK 1365 && (ifp->if_flags & (IFF_UP | IFF_RUNNING)) != 1366 (IFF_UP | IFF_RUNNING)) { 1367 ifp->if_flags |= IFF_UP; 1368 if ((error = (*ifp->if_init)(ifp)) != 0) 1369 return error; 1370 } 1371 #ifdef INET 1372 if (ifa->ifa_addr->sa_family == AF_INET) 1373 arp_ifinit(ifp, ifa); 1374 #endif 1375 return 0; 1376 } 1377 1378 case SIOCSIFMTU: 1379 { 1380 int maxmtu; 1381 1382 if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU) 1383 maxmtu = ETHERMTU_JUMBO; 1384 else 1385 maxmtu = ETHERMTU; 1386 1387 if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu) 1388 return EINVAL; 1389 else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET) 1390 return error; 1391 else if (ifp->if_flags & IFF_UP) { 1392 /* Make sure the device notices the MTU change. */ 1393 return (*ifp->if_init)(ifp); 1394 } else 1395 return 0; 1396 } 1397 1398 case SIOCSIFFLAGS: 1399 if ((error = ifioctl_common(ifp, cmd, data)) != 0) 1400 return error; 1401 switch (ifp->if_flags & (IFF_UP | IFF_RUNNING)) { 1402 case IFF_RUNNING: 1403 /* 1404 * If interface is marked down and it is running, 1405 * then stop and disable it. 1406 */ 1407 (*ifp->if_stop)(ifp, 1); 1408 break; 1409 case IFF_UP: 1410 /* 1411 * If interface is marked up and it is stopped, then 1412 * start it. 1413 */ 1414 return (*ifp->if_init)(ifp); 1415 case IFF_UP | IFF_RUNNING: 1416 error = 0; 1417 if (ec->ec_ifflags_cb != NULL) { 1418 error = (*ec->ec_ifflags_cb)(ec); 1419 if (error == ENETRESET) { 1420 /* 1421 * Reset the interface to pick up 1422 * changes in any other flags that 1423 * affect the hardware state. 1424 */ 1425 return (*ifp->if_init)(ifp); 1426 } 1427 } else 1428 error = (*ifp->if_init)(ifp); 1429 return error; 1430 case 0: 1431 break; 1432 } 1433 return 0; 1434 case SIOCGETHERCAP: 1435 eccr = (struct eccapreq *)data; 1436 eccr->eccr_capabilities = ec->ec_capabilities; 1437 eccr->eccr_capenable = ec->ec_capenable; 1438 return 0; 1439 case SIOCADDMULTI: 1440 return ether_addmulti(ifreq_getaddr(cmd, ifr), ec); 1441 case SIOCDELMULTI: 1442 return ether_delmulti(ifreq_getaddr(cmd, ifr), ec); 1443 case SIOCSIFMEDIA: 1444 case SIOCGIFMEDIA: 1445 if (ec->ec_mii == NULL) 1446 return ENOTTY; 1447 return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd); 1448 case SIOCALIFADDR: 1449 sdl = satocsdl(sstocsa(&iflr->addr)); 1450 if (sdl->sdl_family != AF_LINK) 1451 ; 1452 else if (ETHER_IS_MULTICAST(CLLADDR(sdl))) 1453 return EINVAL; 1454 else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0) 1455 return EINVAL; 1456 /*FALLTHROUGH*/ 1457 default: 1458 return ifioctl_common(ifp, cmd, data); 1459 } 1460 return 0; 1461 } 1462 1463 /* 1464 * Enable/disable passing VLAN packets if the parent interface supports it. 1465 * Return: 1466 * 0: Ok 1467 * -1: Parent interface does not support vlans 1468 * >0: Error 1469 */ 1470 int 1471 ether_enable_vlan_mtu(struct ifnet *ifp) 1472 { 1473 int error; 1474 struct ethercom *ec = (void *)ifp; 1475 1476 /* Parent does not support VLAN's */ 1477 if ((ec->ec_capabilities & ETHERCAP_VLAN_MTU) == 0) 1478 return -1; 1479 1480 /* 1481 * Parent supports the VLAN_MTU capability, 1482 * i.e. can Tx/Rx larger than ETHER_MAX_LEN frames; 1483 * enable it. 1484 */ 1485 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1486 1487 /* Interface is down, defer for later */ 1488 if ((ifp->if_flags & IFF_UP) == 0) 1489 return 0; 1490 1491 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1492 return 0; 1493 1494 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1495 return error; 1496 } 1497 1498 int 1499 ether_disable_vlan_mtu(struct ifnet *ifp) 1500 { 1501 int error; 1502 struct ethercom *ec = (void *)ifp; 1503 1504 /* We still have VLAN's, defer for later */ 1505 if (ec->ec_nvlans != 0) 1506 return 0; 1507 1508 /* Parent does not support VLAB's, nothing to do. */ 1509 if ((ec->ec_capenable & ETHERCAP_VLAN_MTU) == 0) 1510 return -1; 1511 1512 /* 1513 * Disable Tx/Rx of VLAN-sized frames. 1514 */ 1515 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1516 1517 /* Interface is down, defer for later */ 1518 if ((ifp->if_flags & IFF_UP) == 0) 1519 return 0; 1520 1521 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1522 return 0; 1523 1524 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1525 return error; 1526 } 1527 1528 static int 1529 ether_multicast_sysctl(SYSCTLFN_ARGS) 1530 { 1531 struct ether_multi *enm; 1532 struct ifnet *ifp; 1533 struct ethercom *ec; 1534 int error = 0; 1535 size_t written; 1536 struct psref psref; 1537 int bound; 1538 unsigned int multicnt; 1539 struct ether_multi_sysctl *addrs; 1540 int i; 1541 1542 if (namelen != 1) 1543 return EINVAL; 1544 1545 bound = curlwp_bind(); 1546 ifp = if_get_byindex(name[0], &psref); 1547 if (ifp == NULL) { 1548 error = ENODEV; 1549 goto out; 1550 } 1551 if (ifp->if_type != IFT_ETHER) { 1552 if_put(ifp, &psref); 1553 *oldlenp = 0; 1554 goto out; 1555 } 1556 ec = (struct ethercom *)ifp; 1557 1558 if (oldp == NULL) { 1559 if_put(ifp, &psref); 1560 *oldlenp = ec->ec_multicnt * sizeof(*addrs); 1561 goto out; 1562 } 1563 1564 /* 1565 * ec->ec_lock is a spin mutex so we cannot call sysctl_copyout, which 1566 * is sleepable, while holding it. Copy data to a local buffer first 1567 * with the lock taken and then call sysctl_copyout without holding it. 1568 */ 1569 retry: 1570 multicnt = ec->ec_multicnt; 1571 1572 if (multicnt == 0) { 1573 if_put(ifp, &psref); 1574 *oldlenp = 0; 1575 goto out; 1576 } 1577 1578 addrs = kmem_zalloc(sizeof(*addrs) * multicnt, KM_SLEEP); 1579 1580 ETHER_LOCK(ec); 1581 if (multicnt != ec->ec_multicnt) { 1582 /* The number of multicast addresses has changed */ 1583 ETHER_UNLOCK(ec); 1584 kmem_free(addrs, sizeof(*addrs) * multicnt); 1585 goto retry; 1586 } 1587 1588 i = 0; 1589 LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) { 1590 struct ether_multi_sysctl *addr = &addrs[i]; 1591 addr->enm_refcount = enm->enm_refcount; 1592 memcpy(addr->enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN); 1593 memcpy(addr->enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN); 1594 i++; 1595 } 1596 ETHER_UNLOCK(ec); 1597 1598 error = 0; 1599 written = 0; 1600 for (i = 0; i < multicnt; i++) { 1601 struct ether_multi_sysctl *addr = &addrs[i]; 1602 1603 if (written + sizeof(*addr) > *oldlenp) 1604 break; 1605 error = sysctl_copyout(l, addr, oldp, sizeof(*addr)); 1606 if (error) 1607 break; 1608 written += sizeof(*addr); 1609 oldp = (char *)oldp + sizeof(*addr); 1610 } 1611 kmem_free(addrs, sizeof(*addrs) * multicnt); 1612 1613 if_put(ifp, &psref); 1614 1615 *oldlenp = written; 1616 out: 1617 curlwp_bindx(bound); 1618 return error; 1619 } 1620 1621 static void 1622 ether_sysctl_setup(struct sysctllog **clog) 1623 { 1624 const struct sysctlnode *rnode = NULL; 1625 1626 sysctl_createv(clog, 0, NULL, &rnode, 1627 CTLFLAG_PERMANENT, 1628 CTLTYPE_NODE, "ether", 1629 SYSCTL_DESCR("Ethernet-specific information"), 1630 NULL, 0, NULL, 0, 1631 CTL_NET, CTL_CREATE, CTL_EOL); 1632 1633 sysctl_createv(clog, 0, &rnode, NULL, 1634 CTLFLAG_PERMANENT, 1635 CTLTYPE_NODE, "multicast", 1636 SYSCTL_DESCR("multicast addresses"), 1637 ether_multicast_sysctl, 0, NULL, 0, 1638 CTL_CREATE, CTL_EOL); 1639 } 1640 1641 void 1642 etherinit(void) 1643 { 1644 1645 mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET); 1646 ether_sysctl_setup(NULL); 1647 } 1648
Indexes created Tue Sep 30 06:09:59 GMT 2025