if_ethersubr.c revision 1.242.6.10
1 /* $NetBSD: if_ethersubr.c,v 1.242.6.10 2022/10/10 16:09:12 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 1.242.6.10 2022/10/10 16:09:12 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 KASSERT(ehlen == sizeof(*eh)); 845 #if defined (LLC) || defined (NETATALK) 846 if (m->m_len < sizeof(*eh) + sizeof(struct llc)) { 847 goto dropanyway; 848 } 849 l = (struct llc *)(eh+1); 850 851 switch (l->llc_dsap) { 852 #ifdef NETATALK 853 case LLC_SNAP_LSAP: 854 switch (l->llc_control) { 855 case LLC_UI: 856 if (l->llc_ssap != LLC_SNAP_LSAP) { 857 goto dropanyway; 858 } 859 860 if (memcmp(&(l->llc_snap_org_code)[0], 861 at_org_code, sizeof(at_org_code)) == 0 && 862 ntohs(l->llc_snap_ether_type) == 863 ETHERTYPE_ATALK) { 864 inq = &atintrq2; 865 m_adj(m, sizeof(struct ether_header) 866 + sizeof(struct llc)); 867 isr = NETISR_ATALK; 868 break; 869 } 870 871 if (memcmp(&(l->llc_snap_org_code)[0], 872 aarp_org_code, 873 sizeof(aarp_org_code)) == 0 && 874 ntohs(l->llc_snap_ether_type) == 875 ETHERTYPE_AARP) { 876 m_adj(m, sizeof(struct ether_header) 877 + sizeof(struct llc)); 878 aarpinput(ifp, m); /* XXX */ 879 return; 880 } 881 882 default: 883 goto dropanyway; 884 } 885 break; 886 #endif 887 dropanyway: 888 default: 889 m_freem(m); 890 return; 891 } 892 #else /* LLC || NETATALK */ 893 m_freem(m); 894 return; 895 #endif /* LLC || NETATALK */ 896 } 897 898 if (__predict_true(pktq)) { 899 #ifdef NET_MPSAFE 900 const u_int h = curcpu()->ci_index; 901 #else 902 const uint32_t h = pktq_rps_hash(m); 903 #endif 904 if (__predict_false(!pktq_enqueue(pktq, m, h))) { 905 m_freem(m); 906 } 907 return; 908 } 909 910 if (__predict_false(!inq)) { 911 /* Should not happen. */ 912 m_freem(m); 913 return; 914 } 915 916 IFQ_LOCK(inq); 917 if (IF_QFULL(inq)) { 918 IF_DROP(inq); 919 IFQ_UNLOCK(inq); 920 m_freem(m); 921 } else { 922 IF_ENQUEUE(inq, m); 923 IFQ_UNLOCK(inq); 924 schednetisr(isr); 925 } 926 } 927 928 /* 929 * Convert Ethernet address to printable (loggable) representation. 930 */ 931 char * 932 ether_sprintf(const u_char *ap) 933 { 934 static char etherbuf[3 * ETHER_ADDR_LEN]; 935 return ether_snprintf(etherbuf, sizeof(etherbuf), ap); 936 } 937 938 char * 939 ether_snprintf(char *buf, size_t len, const u_char *ap) 940 { 941 char *cp = buf; 942 size_t i; 943 944 for (i = 0; i < len / 3; i++) { 945 *cp++ = hexdigits[*ap >> 4]; 946 *cp++ = hexdigits[*ap++ & 0xf]; 947 *cp++ = ':'; 948 } 949 *--cp = '\0'; 950 return buf; 951 } 952 953 /* 954 * Perform common duties while attaching to interface list 955 */ 956 void 957 ether_ifattach(struct ifnet *ifp, const uint8_t *lla) 958 { 959 struct ethercom *ec = (struct ethercom *)ifp; 960 961 ifp->if_type = IFT_ETHER; 962 ifp->if_hdrlen = ETHER_HDR_LEN; 963 ifp->if_dlt = DLT_EN10MB; 964 ifp->if_mtu = ETHERMTU; 965 ifp->if_output = ether_output; 966 ifp->_if_input = ether_input; 967 if (ifp->if_baudrate == 0) 968 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 969 970 if (lla != NULL) 971 if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla)); 972 973 LIST_INIT(&ec->ec_multiaddrs); 974 ec->ec_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NET); 975 ifp->if_broadcastaddr = etherbroadcastaddr; 976 bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header)); 977 #ifdef MBUFTRACE 978 strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname, 979 sizeof(ec->ec_tx_mowner.mo_name)); 980 strlcpy(ec->ec_tx_mowner.mo_descr, "tx", 981 sizeof(ec->ec_tx_mowner.mo_descr)); 982 strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname, 983 sizeof(ec->ec_rx_mowner.mo_name)); 984 strlcpy(ec->ec_rx_mowner.mo_descr, "rx", 985 sizeof(ec->ec_rx_mowner.mo_descr)); 986 MOWNER_ATTACH(&ec->ec_tx_mowner); 987 MOWNER_ATTACH(&ec->ec_rx_mowner); 988 ifp->if_mowner = &ec->ec_tx_mowner; 989 #endif 990 } 991 992 void 993 ether_ifdetach(struct ifnet *ifp) 994 { 995 struct ethercom *ec = (void *) ifp; 996 struct ether_multi *enm; 997 998 /* 999 * Prevent further calls to ioctl (for example turning off 1000 * promiscuous mode from the bridge code), which eventually can 1001 * call if_init() which can cause panics because the interface 1002 * is in the process of being detached. Return device not configured 1003 * instead. 1004 */ 1005 ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio; 1006 1007 #if NBRIDGE > 0 1008 if (ifp->if_bridge) 1009 bridge_ifdetach(ifp); 1010 #endif 1011 bpf_detach(ifp); 1012 #if NVLAN > 0 1013 if (ec->ec_nvlans) 1014 vlan_ifdetach(ifp); 1015 #endif 1016 1017 ETHER_LOCK(ec); 1018 while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) { 1019 LIST_REMOVE(enm, enm_list); 1020 kmem_free(enm, sizeof(*enm)); 1021 ec->ec_multicnt--; 1022 } 1023 ETHER_UNLOCK(ec); 1024 1025 mutex_obj_free(ec->ec_lock); 1026 1027 ifp->if_mowner = NULL; 1028 MOWNER_DETACH(&ec->ec_rx_mowner); 1029 MOWNER_DETACH(&ec->ec_tx_mowner); 1030 } 1031 1032 #if 0 1033 /* 1034 * This is for reference. We have a table-driven version 1035 * of the little-endian crc32 generator, which is faster 1036 * than the double-loop. 1037 */ 1038 uint32_t 1039 ether_crc32_le(const uint8_t *buf, size_t len) 1040 { 1041 uint32_t c, crc, carry; 1042 size_t i, j; 1043 1044 crc = 0xffffffffU; /* initial value */ 1045 1046 for (i = 0; i < len; i++) { 1047 c = buf[i]; 1048 for (j = 0; j < 8; j++) { 1049 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1050 crc >>= 1; 1051 c >>= 1; 1052 if (carry) 1053 crc = (crc ^ ETHER_CRC_POLY_LE); 1054 } 1055 } 1056 1057 return (crc); 1058 } 1059 #else 1060 uint32_t 1061 ether_crc32_le(const uint8_t *buf, size_t len) 1062 { 1063 static const uint32_t crctab[] = { 1064 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1065 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1066 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1067 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1068 }; 1069 uint32_t crc; 1070 size_t i; 1071 1072 crc = 0xffffffffU; /* initial value */ 1073 1074 for (i = 0; i < len; i++) { 1075 crc ^= buf[i]; 1076 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1077 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1078 } 1079 1080 return (crc); 1081 } 1082 #endif 1083 1084 uint32_t 1085 ether_crc32_be(const uint8_t *buf, size_t len) 1086 { 1087 uint32_t c, crc, carry; 1088 size_t i, j; 1089 1090 crc = 0xffffffffU; /* initial value */ 1091 1092 for (i = 0; i < len; i++) { 1093 c = buf[i]; 1094 for (j = 0; j < 8; j++) { 1095 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1096 crc <<= 1; 1097 c >>= 1; 1098 if (carry) 1099 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1100 } 1101 } 1102 1103 return (crc); 1104 } 1105 1106 #ifdef INET 1107 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] = 1108 { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 }; 1109 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] = 1110 { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff }; 1111 #endif 1112 #ifdef INET6 1113 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] = 1114 { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 }; 1115 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] = 1116 { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff }; 1117 #endif 1118 1119 /* 1120 * ether_aton implementation, not using a static buffer. 1121 */ 1122 int 1123 ether_aton_r(u_char *dest, size_t len, const char *str) 1124 { 1125 const u_char *cp = (const void *)str; 1126 u_char *ep; 1127 1128 #define atox(c) (((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10)) 1129 1130 if (len < ETHER_ADDR_LEN) 1131 return ENOSPC; 1132 1133 ep = dest + ETHER_ADDR_LEN; 1134 1135 while (*cp) { 1136 if (!isxdigit(*cp)) 1137 return EINVAL; 1138 1139 *dest = atox(*cp); 1140 cp++; 1141 if (isxdigit(*cp)) { 1142 *dest = (*dest << 4) | atox(*cp); 1143 cp++; 1144 } 1145 dest++; 1146 1147 if (dest == ep) 1148 return (*cp == '\0') ? 0 : ENAMETOOLONG; 1149 1150 switch (*cp) { 1151 case ':': 1152 case '-': 1153 case '.': 1154 cp++; 1155 break; 1156 } 1157 } 1158 return ENOBUFS; 1159 } 1160 1161 /* 1162 * Convert a sockaddr into an Ethernet address or range of Ethernet 1163 * addresses. 1164 */ 1165 int 1166 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN], 1167 uint8_t addrhi[ETHER_ADDR_LEN]) 1168 { 1169 #ifdef INET 1170 const struct sockaddr_in *sin; 1171 #endif /* INET */ 1172 #ifdef INET6 1173 const struct sockaddr_in6 *sin6; 1174 #endif /* INET6 */ 1175 1176 switch (sa->sa_family) { 1177 1178 case AF_UNSPEC: 1179 memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN); 1180 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1181 break; 1182 1183 #ifdef INET 1184 case AF_INET: 1185 sin = satocsin(sa); 1186 if (sin->sin_addr.s_addr == INADDR_ANY) { 1187 /* 1188 * An IP address of INADDR_ANY means listen to 1189 * or stop listening to all of the Ethernet 1190 * multicast addresses used for IP. 1191 * (This is for the sake of IP multicast routers.) 1192 */ 1193 memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN); 1194 memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN); 1195 } else { 1196 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo); 1197 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1198 } 1199 break; 1200 #endif 1201 #ifdef INET6 1202 case AF_INET6: 1203 sin6 = satocsin6(sa); 1204 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1205 /* 1206 * An IP6 address of 0 means listen to or stop 1207 * listening to all of the Ethernet multicast 1208 * address used for IP6. 1209 * (This is used for multicast routers.) 1210 */ 1211 memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN); 1212 memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN); 1213 } else { 1214 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo); 1215 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1216 } 1217 break; 1218 #endif 1219 1220 default: 1221 return EAFNOSUPPORT; 1222 } 1223 return 0; 1224 } 1225 1226 /* 1227 * Add an Ethernet multicast address or range of addresses to the list for a 1228 * given interface. 1229 */ 1230 int 1231 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec) 1232 { 1233 struct ether_multi *enm, *_enm; 1234 u_char addrlo[ETHER_ADDR_LEN]; 1235 u_char addrhi[ETHER_ADDR_LEN]; 1236 int error = 0; 1237 1238 /* Allocate out of lock */ 1239 enm = kmem_alloc(sizeof(*enm), KM_SLEEP); 1240 1241 ETHER_LOCK(ec); 1242 error = ether_multiaddr(sa, addrlo, addrhi); 1243 if (error != 0) 1244 goto out; 1245 1246 /* 1247 * Verify that we have valid Ethernet multicast addresses. 1248 */ 1249 if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) { 1250 error = EINVAL; 1251 goto out; 1252 } 1253 1254 /* 1255 * See if the address range is already in the list. 1256 */ 1257 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, _enm); 1258 if (_enm != NULL) { 1259 /* 1260 * Found it; just increment the reference count. 1261 */ 1262 ++_enm->enm_refcount; 1263 error = 0; 1264 goto out; 1265 } 1266 1267 /* 1268 * Link a new multicast record into the interface's multicast list. 1269 */ 1270 memcpy(enm->enm_addrlo, addrlo, ETHER_ADDR_LEN); 1271 memcpy(enm->enm_addrhi, addrhi, ETHER_ADDR_LEN); 1272 enm->enm_refcount = 1; 1273 LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list); 1274 ec->ec_multicnt++; 1275 1276 /* 1277 * Return ENETRESET to inform the driver that the list has changed 1278 * and its reception filter should be adjusted accordingly. 1279 */ 1280 error = ENETRESET; 1281 enm = NULL; 1282 1283 out: 1284 ETHER_UNLOCK(ec); 1285 if (enm != NULL) 1286 kmem_free(enm, sizeof(*enm)); 1287 return error; 1288 } 1289 1290 /* 1291 * Delete a multicast address record. 1292 */ 1293 int 1294 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec) 1295 { 1296 struct ether_multi *enm; 1297 u_char addrlo[ETHER_ADDR_LEN]; 1298 u_char addrhi[ETHER_ADDR_LEN]; 1299 int error; 1300 1301 ETHER_LOCK(ec); 1302 error = ether_multiaddr(sa, addrlo, addrhi); 1303 if (error != 0) 1304 goto error; 1305 1306 /* 1307 * Look up the address in our list. 1308 */ 1309 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm); 1310 if (enm == NULL) { 1311 error = ENXIO; 1312 goto error; 1313 } 1314 if (--enm->enm_refcount != 0) { 1315 /* 1316 * Still some claims to this record. 1317 */ 1318 error = 0; 1319 goto error; 1320 } 1321 1322 /* 1323 * No remaining claims to this record; unlink and free it. 1324 */ 1325 LIST_REMOVE(enm, enm_list); 1326 ec->ec_multicnt--; 1327 ETHER_UNLOCK(ec); 1328 kmem_free(enm, sizeof(*enm)); 1329 1330 /* 1331 * Return ENETRESET to inform the driver that the list has changed 1332 * and its reception filter should be adjusted accordingly. 1333 */ 1334 return ENETRESET; 1335 1336 error: 1337 ETHER_UNLOCK(ec); 1338 return error; 1339 } 1340 1341 void 1342 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb) 1343 { 1344 ec->ec_ifflags_cb = cb; 1345 } 1346 1347 /* 1348 * Common ioctls for Ethernet interfaces. Note, we must be 1349 * called at splnet(). 1350 */ 1351 int 1352 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data) 1353 { 1354 struct ethercom *ec = (void *) ifp; 1355 struct eccapreq *eccr; 1356 struct ifreq *ifr = (struct ifreq *)data; 1357 struct if_laddrreq *iflr = data; 1358 const struct sockaddr_dl *sdl; 1359 static const uint8_t zero[ETHER_ADDR_LEN]; 1360 int error; 1361 1362 switch (cmd) { 1363 case SIOCINITIFADDR: 1364 { 1365 struct ifaddr *ifa = (struct ifaddr *)data; 1366 if (ifa->ifa_addr->sa_family != AF_LINK 1367 && (ifp->if_flags & (IFF_UP | IFF_RUNNING)) != 1368 (IFF_UP | IFF_RUNNING)) { 1369 ifp->if_flags |= IFF_UP; 1370 if ((error = (*ifp->if_init)(ifp)) != 0) 1371 return error; 1372 } 1373 #ifdef INET 1374 if (ifa->ifa_addr->sa_family == AF_INET) 1375 arp_ifinit(ifp, ifa); 1376 #endif 1377 return 0; 1378 } 1379 1380 case SIOCSIFMTU: 1381 { 1382 int maxmtu; 1383 1384 if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU) 1385 maxmtu = ETHERMTU_JUMBO; 1386 else 1387 maxmtu = ETHERMTU; 1388 1389 if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu) 1390 return EINVAL; 1391 else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET) 1392 return error; 1393 else if (ifp->if_flags & IFF_UP) { 1394 /* Make sure the device notices the MTU change. */ 1395 return (*ifp->if_init)(ifp); 1396 } else 1397 return 0; 1398 } 1399 1400 case SIOCSIFFLAGS: 1401 if ((error = ifioctl_common(ifp, cmd, data)) != 0) 1402 return error; 1403 switch (ifp->if_flags & (IFF_UP | IFF_RUNNING)) { 1404 case IFF_RUNNING: 1405 /* 1406 * If interface is marked down and it is running, 1407 * then stop and disable it. 1408 */ 1409 (*ifp->if_stop)(ifp, 1); 1410 break; 1411 case IFF_UP: 1412 /* 1413 * If interface is marked up and it is stopped, then 1414 * start it. 1415 */ 1416 return (*ifp->if_init)(ifp); 1417 case IFF_UP | IFF_RUNNING: 1418 error = 0; 1419 if (ec->ec_ifflags_cb != NULL) { 1420 error = (*ec->ec_ifflags_cb)(ec); 1421 if (error == ENETRESET) { 1422 /* 1423 * Reset the interface to pick up 1424 * changes in any other flags that 1425 * affect the hardware state. 1426 */ 1427 return (*ifp->if_init)(ifp); 1428 } 1429 } else 1430 error = (*ifp->if_init)(ifp); 1431 return error; 1432 case 0: 1433 break; 1434 } 1435 return 0; 1436 case SIOCGETHERCAP: 1437 eccr = (struct eccapreq *)data; 1438 eccr->eccr_capabilities = ec->ec_capabilities; 1439 eccr->eccr_capenable = ec->ec_capenable; 1440 return 0; 1441 case SIOCADDMULTI: 1442 return ether_addmulti(ifreq_getaddr(cmd, ifr), ec); 1443 case SIOCDELMULTI: 1444 return ether_delmulti(ifreq_getaddr(cmd, ifr), ec); 1445 case SIOCSIFMEDIA: 1446 case SIOCGIFMEDIA: 1447 if (ec->ec_mii == NULL) 1448 return ENOTTY; 1449 return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd); 1450 case SIOCALIFADDR: 1451 sdl = satocsdl(sstocsa(&iflr->addr)); 1452 if (sdl->sdl_family != AF_LINK) 1453 ; 1454 else if (ETHER_IS_MULTICAST(CLLADDR(sdl))) 1455 return EINVAL; 1456 else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0) 1457 return EINVAL; 1458 /*FALLTHROUGH*/ 1459 default: 1460 return ifioctl_common(ifp, cmd, data); 1461 } 1462 return 0; 1463 } 1464 1465 /* 1466 * Enable/disable passing VLAN packets if the parent interface supports it. 1467 * Return: 1468 * 0: Ok 1469 * -1: Parent interface does not support vlans 1470 * >0: Error 1471 */ 1472 int 1473 ether_enable_vlan_mtu(struct ifnet *ifp) 1474 { 1475 int error; 1476 struct ethercom *ec = (void *)ifp; 1477 1478 /* Parent does not support VLAN's */ 1479 if ((ec->ec_capabilities & ETHERCAP_VLAN_MTU) == 0) 1480 return -1; 1481 1482 /* 1483 * Parent supports the VLAN_MTU capability, 1484 * i.e. can Tx/Rx larger than ETHER_MAX_LEN frames; 1485 * enable it. 1486 */ 1487 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1488 1489 /* Interface is down, defer for later */ 1490 if ((ifp->if_flags & IFF_UP) == 0) 1491 return 0; 1492 1493 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1494 return 0; 1495 1496 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1497 return error; 1498 } 1499 1500 int 1501 ether_disable_vlan_mtu(struct ifnet *ifp) 1502 { 1503 int error; 1504 struct ethercom *ec = (void *)ifp; 1505 1506 /* We still have VLAN's, defer for later */ 1507 if (ec->ec_nvlans != 0) 1508 return 0; 1509 1510 /* Parent does not support VLAB's, nothing to do. */ 1511 if ((ec->ec_capenable & ETHERCAP_VLAN_MTU) == 0) 1512 return -1; 1513 1514 /* 1515 * Disable Tx/Rx of VLAN-sized frames. 1516 */ 1517 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1518 1519 /* Interface is down, defer for later */ 1520 if ((ifp->if_flags & IFF_UP) == 0) 1521 return 0; 1522 1523 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1524 return 0; 1525 1526 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1527 return error; 1528 } 1529 1530 static int 1531 ether_multicast_sysctl(SYSCTLFN_ARGS) 1532 { 1533 struct ether_multi *enm; 1534 struct ifnet *ifp; 1535 struct ethercom *ec; 1536 int error = 0; 1537 size_t written; 1538 struct psref psref; 1539 int bound; 1540 unsigned int multicnt; 1541 struct ether_multi_sysctl *addrs; 1542 int i; 1543 1544 if (namelen != 1) 1545 return EINVAL; 1546 1547 bound = curlwp_bind(); 1548 ifp = if_get_byindex(name[0], &psref); 1549 if (ifp == NULL) { 1550 error = ENODEV; 1551 goto out; 1552 } 1553 if (ifp->if_type != IFT_ETHER) { 1554 if_put(ifp, &psref); 1555 *oldlenp = 0; 1556 goto out; 1557 } 1558 ec = (struct ethercom *)ifp; 1559 1560 if (oldp == NULL) { 1561 if_put(ifp, &psref); 1562 *oldlenp = ec->ec_multicnt * sizeof(*addrs); 1563 goto out; 1564 } 1565 1566 /* 1567 * ec->ec_lock is a spin mutex so we cannot call sysctl_copyout, which 1568 * is sleepable, while holding it. Copy data to a local buffer first 1569 * with the lock taken and then call sysctl_copyout without holding it. 1570 */ 1571 retry: 1572 multicnt = ec->ec_multicnt; 1573 1574 if (multicnt == 0) { 1575 if_put(ifp, &psref); 1576 *oldlenp = 0; 1577 goto out; 1578 } 1579 1580 addrs = kmem_zalloc(sizeof(*addrs) * multicnt, KM_SLEEP); 1581 1582 ETHER_LOCK(ec); 1583 if (multicnt != ec->ec_multicnt) { 1584 /* The number of multicast addresses has changed */ 1585 ETHER_UNLOCK(ec); 1586 kmem_free(addrs, sizeof(*addrs) * multicnt); 1587 goto retry; 1588 } 1589 1590 i = 0; 1591 LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) { 1592 struct ether_multi_sysctl *addr = &addrs[i]; 1593 addr->enm_refcount = enm->enm_refcount; 1594 memcpy(addr->enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN); 1595 memcpy(addr->enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN); 1596 i++; 1597 } 1598 ETHER_UNLOCK(ec); 1599 1600 error = 0; 1601 written = 0; 1602 for (i = 0; i < multicnt; i++) { 1603 struct ether_multi_sysctl *addr = &addrs[i]; 1604 1605 if (written + sizeof(*addr) > *oldlenp) 1606 break; 1607 error = sysctl_copyout(l, addr, oldp, sizeof(*addr)); 1608 if (error) 1609 break; 1610 written += sizeof(*addr); 1611 oldp = (char *)oldp + sizeof(*addr); 1612 } 1613 kmem_free(addrs, sizeof(*addrs) * multicnt); 1614 1615 if_put(ifp, &psref); 1616 1617 *oldlenp = written; 1618 out: 1619 curlwp_bindx(bound); 1620 return error; 1621 } 1622 1623 static void 1624 ether_sysctl_setup(struct sysctllog **clog) 1625 { 1626 const struct sysctlnode *rnode = NULL; 1627 1628 sysctl_createv(clog, 0, NULL, &rnode, 1629 CTLFLAG_PERMANENT, 1630 CTLTYPE_NODE, "ether", 1631 SYSCTL_DESCR("Ethernet-specific information"), 1632 NULL, 0, NULL, 0, 1633 CTL_NET, CTL_CREATE, CTL_EOL); 1634 1635 sysctl_createv(clog, 0, &rnode, NULL, 1636 CTLFLAG_PERMANENT, 1637 CTLTYPE_NODE, "multicast", 1638 SYSCTL_DESCR("multicast addresses"), 1639 ether_multicast_sysctl, 0, NULL, 0, 1640 CTL_CREATE, CTL_EOL); 1641 } 1642 1643 void 1644 etherinit(void) 1645 { 1646 1647 mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET); 1648 ether_sysctl_setup(NULL); 1649 } 1650
Indexes created Tue Sep 30 06:09:59 GMT 2025