if_ethersubr.c revision 1.296
1 /* $NetBSD: if_ethersubr.c,v 1.296 2021/09/30 03:51:05 yamaguchi 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 1.296 2021/09/30 03:51:05 yamaguchi 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 #include "lagg.h" 82 83 #include <sys/sysctl.h> 84 #include <sys/mbuf.h> 85 #include <sys/mutex.h> 86 #include <sys/ioctl.h> 87 #include <sys/errno.h> 88 #include <sys/device.h> 89 #include <sys/entropy.h> 90 #include <sys/rndsource.h> 91 #include <sys/cpu.h> 92 #include <sys/kmem.h> 93 94 #include <net/if.h> 95 #include <net/netisr.h> 96 #include <net/route.h> 97 #include <net/if_llc.h> 98 #include <net/if_dl.h> 99 #include <net/if_types.h> 100 #include <net/pktqueue.h> 101 102 #include <net/if_media.h> 103 #include <dev/mii/mii.h> 104 #include <dev/mii/miivar.h> 105 106 #if NARP == 0 107 /* 108 * XXX there should really be a way to issue this warning from within config(8) 109 */ 110 #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. 111 #endif 112 113 #include <net/bpf.h> 114 115 #include <net/if_ether.h> 116 #include <net/if_vlanvar.h> 117 118 #if NPPPOE > 0 119 #include <net/if_pppoe.h> 120 #endif 121 122 #if NAGR > 0 123 #include <net/agr/ieee8023_slowprotocols.h> /* XXX */ 124 #include <net/agr/ieee8023ad.h> 125 #include <net/agr/if_agrvar.h> 126 #endif 127 128 #include <net/lagg/if_laggvar.h> 129 130 #if NBRIDGE > 0 131 #include <net/if_bridgevar.h> 132 #endif 133 134 #include <netinet/in.h> 135 #ifdef INET 136 #include <netinet/in_var.h> 137 #endif 138 #include <netinet/if_inarp.h> 139 140 #ifdef INET6 141 #ifndef INET 142 #include <netinet/in.h> 143 #endif 144 #include <netinet6/in6_var.h> 145 #include <netinet6/nd6.h> 146 #endif 147 148 #include "carp.h" 149 #if NCARP > 0 150 #include <netinet/ip_carp.h> 151 #endif 152 153 #ifdef NETATALK 154 #include <netatalk/at.h> 155 #include <netatalk/at_var.h> 156 #include <netatalk/at_extern.h> 157 158 #define llc_snap_org_code llc_un.type_snap.org_code 159 #define llc_snap_ether_type llc_un.type_snap.ether_type 160 161 extern u_char at_org_code[3]; 162 extern u_char aarp_org_code[3]; 163 #endif /* NETATALK */ 164 165 #ifdef MPLS 166 #include <netmpls/mpls.h> 167 #include <netmpls/mpls_var.h> 168 #endif 169 170 CTASSERT(sizeof(struct ether_addr) == 6); 171 CTASSERT(sizeof(struct ether_header) == 14); 172 173 #ifdef DIAGNOSTIC 174 static struct timeval bigpktppslim_last; 175 static int bigpktppslim = 2; /* XXX */ 176 static int bigpktpps_count; 177 static kmutex_t bigpktpps_lock __cacheline_aligned; 178 #endif 179 180 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = 181 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 182 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] = 183 { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 }; 184 #define senderr(e) { error = (e); goto bad;} 185 186 /* if_lagg(4) support */ 187 struct mbuf *(*lagg_input_ethernet_p)(struct ifnet *, struct mbuf *); 188 189 static int ether_output(struct ifnet *, struct mbuf *, 190 const struct sockaddr *, const struct rtentry *); 191 192 /* 193 * Ethernet output routine. 194 * Encapsulate a packet of type family for the local net. 195 * Assumes that ifp is actually pointer to ethercom structure. 196 */ 197 static int 198 ether_output(struct ifnet * const ifp0, struct mbuf * const m0, 199 const struct sockaddr * const dst, const struct rtentry *rt) 200 { 201 uint8_t esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN]; 202 uint16_t etype = 0; 203 int error = 0, hdrcmplt = 0; 204 struct mbuf *m = m0; 205 struct mbuf *mcopy = NULL; 206 struct ether_header *eh; 207 struct ifnet *ifp = ifp0; 208 #ifdef INET 209 struct arphdr *ah; 210 #endif 211 #ifdef NETATALK 212 struct at_ifaddr *aa; 213 #endif 214 215 #ifdef MBUFTRACE 216 m_claimm(m, ifp->if_mowner); 217 #endif 218 219 #if NCARP > 0 220 if (ifp->if_type == IFT_CARP) { 221 struct ifaddr *ifa; 222 int s = pserialize_read_enter(); 223 224 /* loop back if this is going to the carp interface */ 225 if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP && 226 (ifa = ifa_ifwithaddr(dst)) != NULL) { 227 if (ifa->ifa_ifp == ifp0) { 228 pserialize_read_exit(s); 229 return looutput(ifp0, m, dst, rt); 230 } 231 } 232 pserialize_read_exit(s); 233 234 ifp = ifp->if_carpdev; 235 /* ac = (struct arpcom *)ifp; */ 236 237 if ((ifp0->if_flags & (IFF_UP | IFF_RUNNING)) != 238 (IFF_UP | IFF_RUNNING)) 239 senderr(ENETDOWN); 240 } 241 #endif 242 243 if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING)) 244 senderr(ENETDOWN); 245 246 switch (dst->sa_family) { 247 248 #ifdef INET 249 case AF_INET: 250 if (m->m_flags & M_BCAST) { 251 memcpy(edst, etherbroadcastaddr, sizeof(edst)); 252 } else if (m->m_flags & M_MCAST) { 253 ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst); 254 } else { 255 error = arpresolve(ifp0, rt, m, dst, edst, sizeof(edst)); 256 if (error) 257 return (error == EWOULDBLOCK) ? 0 : error; 258 } 259 /* If broadcasting on a simplex interface, loopback a copy */ 260 if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX)) 261 mcopy = m_copypacket(m, M_DONTWAIT); 262 etype = htons(ETHERTYPE_IP); 263 break; 264 265 case AF_ARP: 266 ah = mtod(m, struct arphdr *); 267 if (m->m_flags & M_BCAST) { 268 memcpy(edst, etherbroadcastaddr, sizeof(edst)); 269 } else { 270 void *tha = ar_tha(ah); 271 272 if (tha == NULL) { 273 /* fake with ARPHRD_IEEE1394 */ 274 m_freem(m); 275 return 0; 276 } 277 memcpy(edst, tha, sizeof(edst)); 278 } 279 280 ah->ar_hrd = htons(ARPHRD_ETHER); 281 282 switch (ntohs(ah->ar_op)) { 283 case ARPOP_REVREQUEST: 284 case ARPOP_REVREPLY: 285 etype = htons(ETHERTYPE_REVARP); 286 break; 287 288 case ARPOP_REQUEST: 289 case ARPOP_REPLY: 290 default: 291 etype = htons(ETHERTYPE_ARP); 292 } 293 break; 294 #endif 295 296 #ifdef INET6 297 case AF_INET6: 298 if (m->m_flags & M_BCAST) { 299 memcpy(edst, etherbroadcastaddr, sizeof(edst)); 300 } else if (m->m_flags & M_MCAST) { 301 ETHER_MAP_IPV6_MULTICAST(&satocsin6(dst)->sin6_addr, 302 edst); 303 } else { 304 error = nd6_resolve(ifp0, rt, m, dst, edst, 305 sizeof(edst)); 306 if (error) 307 return (error == EWOULDBLOCK) ? 0 : error; 308 } 309 etype = htons(ETHERTYPE_IPV6); 310 break; 311 #endif 312 313 #ifdef NETATALK 314 case AF_APPLETALK: { 315 struct ifaddr *ifa; 316 int s; 317 318 KERNEL_LOCK(1, NULL); 319 320 if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) { 321 KERNEL_UNLOCK_ONE(NULL); 322 return 0; 323 } 324 325 /* 326 * ifaddr is the first thing in at_ifaddr 327 */ 328 s = pserialize_read_enter(); 329 ifa = at_ifawithnet((const struct sockaddr_at *)dst, ifp); 330 if (ifa == NULL) { 331 pserialize_read_exit(s); 332 KERNEL_UNLOCK_ONE(NULL); 333 senderr(EADDRNOTAVAIL); 334 } 335 aa = (struct at_ifaddr *)ifa; 336 337 /* 338 * In the phase 2 case, we need to prepend an mbuf for the 339 * llc header. 340 */ 341 if (aa->aa_flags & AFA_PHASE2) { 342 struct llc llc; 343 344 M_PREPEND(m, sizeof(struct llc), M_DONTWAIT); 345 if (m == NULL) { 346 pserialize_read_exit(s); 347 KERNEL_UNLOCK_ONE(NULL); 348 senderr(ENOBUFS); 349 } 350 351 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 352 llc.llc_control = LLC_UI; 353 memcpy(llc.llc_snap_org_code, at_org_code, 354 sizeof(llc.llc_snap_org_code)); 355 llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK); 356 memcpy(mtod(m, void *), &llc, sizeof(struct llc)); 357 } else { 358 etype = htons(ETHERTYPE_ATALK); 359 } 360 pserialize_read_exit(s); 361 KERNEL_UNLOCK_ONE(NULL); 362 break; 363 } 364 #endif /* NETATALK */ 365 366 case pseudo_AF_HDRCMPLT: 367 hdrcmplt = 1; 368 memcpy(esrc, 369 ((const struct ether_header *)dst->sa_data)->ether_shost, 370 sizeof(esrc)); 371 /* FALLTHROUGH */ 372 373 case AF_UNSPEC: 374 memcpy(edst, 375 ((const struct ether_header *)dst->sa_data)->ether_dhost, 376 sizeof(edst)); 377 /* AF_UNSPEC doesn't swap the byte order of the ether_type. */ 378 etype = ((const struct ether_header *)dst->sa_data)->ether_type; 379 break; 380 381 default: 382 printf("%s: can't handle af%d\n", ifp->if_xname, 383 dst->sa_family); 384 senderr(EAFNOSUPPORT); 385 } 386 387 #ifdef MPLS 388 { 389 struct m_tag *mtag; 390 mtag = m_tag_find(m, PACKET_TAG_MPLS); 391 if (mtag != NULL) { 392 /* Having the tag itself indicates it's MPLS */ 393 etype = htons(ETHERTYPE_MPLS); 394 m_tag_delete(m, mtag); 395 } 396 } 397 #endif 398 399 if (mcopy) 400 (void)looutput(ifp, mcopy, dst, rt); 401 402 KASSERT((m->m_flags & M_PKTHDR) != 0); 403 404 /* 405 * If no ether type is set, this must be a 802.2 formatted packet. 406 */ 407 if (etype == 0) 408 etype = htons(m->m_pkthdr.len); 409 410 /* 411 * Add local net header. If no space in first mbuf, allocate another. 412 */ 413 M_PREPEND(m, sizeof(struct ether_header), M_DONTWAIT); 414 if (m == NULL) 415 senderr(ENOBUFS); 416 417 eh = mtod(m, struct ether_header *); 418 /* Note: etype is already in network byte order. */ 419 memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type)); 420 memcpy(eh->ether_dhost, edst, sizeof(edst)); 421 if (hdrcmplt) { 422 memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost)); 423 } else { 424 memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl), 425 sizeof(eh->ether_shost)); 426 } 427 428 #if NCARP > 0 429 if (ifp0 != ifp && ifp0->if_type == IFT_CARP) { 430 memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl), 431 sizeof(eh->ether_shost)); 432 } 433 #endif 434 435 if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0) 436 return error; 437 if (m == NULL) 438 return 0; 439 440 #if NBRIDGE > 0 441 /* 442 * Bridges require special output handling. 443 */ 444 if (ifp->if_bridge) 445 return bridge_output(ifp, m, NULL, NULL); 446 #endif 447 448 #if NCARP > 0 449 if (ifp != ifp0) 450 if_statadd(ifp0, if_obytes, m->m_pkthdr.len + ETHER_HDR_LEN); 451 #endif 452 453 #ifdef ALTQ 454 KERNEL_LOCK(1, NULL); 455 /* 456 * If ALTQ is enabled on the parent interface, do 457 * classification; the queueing discipline might not 458 * require classification, but might require the 459 * address family/header pointer in the pktattr. 460 */ 461 if (ALTQ_IS_ENABLED(&ifp->if_snd)) 462 altq_etherclassify(&ifp->if_snd, m); 463 KERNEL_UNLOCK_ONE(NULL); 464 #endif 465 return ifq_enqueue(ifp, m); 466 467 bad: 468 if_statinc(ifp, if_oerrors); 469 if (m) 470 m_freem(m); 471 return error; 472 } 473 474 #ifdef ALTQ 475 /* 476 * This routine is a slight hack to allow a packet to be classified 477 * if the Ethernet headers are present. It will go away when ALTQ's 478 * classification engine understands link headers. 479 * 480 * XXX: We may need to do m_pullups here. First to ensure struct ether_header 481 * is indeed contiguous, then to read the LLC and so on. 482 */ 483 void 484 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m) 485 { 486 struct ether_header *eh; 487 struct mbuf *mtop = m; 488 uint16_t ether_type; 489 int hlen, af, hdrsize; 490 void *hdr; 491 492 KASSERT((mtop->m_flags & M_PKTHDR) != 0); 493 494 hlen = ETHER_HDR_LEN; 495 eh = mtod(m, struct ether_header *); 496 497 ether_type = htons(eh->ether_type); 498 499 if (ether_type < ETHERMTU) { 500 /* LLC/SNAP */ 501 struct llc *llc = (struct llc *)(eh + 1); 502 hlen += 8; 503 504 if (m->m_len < hlen || 505 llc->llc_dsap != LLC_SNAP_LSAP || 506 llc->llc_ssap != LLC_SNAP_LSAP || 507 llc->llc_control != LLC_UI) { 508 /* Not SNAP. */ 509 goto bad; 510 } 511 512 ether_type = htons(llc->llc_un.type_snap.ether_type); 513 } 514 515 switch (ether_type) { 516 case ETHERTYPE_IP: 517 af = AF_INET; 518 hdrsize = 20; /* sizeof(struct ip) */ 519 break; 520 521 case ETHERTYPE_IPV6: 522 af = AF_INET6; 523 hdrsize = 40; /* sizeof(struct ip6_hdr) */ 524 break; 525 526 default: 527 af = AF_UNSPEC; 528 hdrsize = 0; 529 break; 530 } 531 532 while (m->m_len <= hlen) { 533 hlen -= m->m_len; 534 m = m->m_next; 535 if (m == NULL) 536 goto bad; 537 } 538 539 if (m->m_len < (hlen + hdrsize)) { 540 /* 541 * protocol header not in a single mbuf. 542 * We can't cope with this situation right 543 * now (but it shouldn't ever happen, really, anyhow). 544 */ 545 #ifdef DEBUG 546 printf("altq_etherclassify: headers span multiple mbufs: " 547 "%d < %d\n", m->m_len, (hlen + hdrsize)); 548 #endif 549 goto bad; 550 } 551 552 m->m_data += hlen; 553 m->m_len -= hlen; 554 555 hdr = mtod(m, void *); 556 557 if (ALTQ_NEEDS_CLASSIFY(ifq)) { 558 mtop->m_pkthdr.pattr_class = 559 (*ifq->altq_classify)(ifq->altq_clfier, m, af); 560 } 561 mtop->m_pkthdr.pattr_af = af; 562 mtop->m_pkthdr.pattr_hdr = hdr; 563 564 m->m_data -= hlen; 565 m->m_len += hlen; 566 567 return; 568 569 bad: 570 mtop->m_pkthdr.pattr_class = NULL; 571 mtop->m_pkthdr.pattr_hdr = NULL; 572 mtop->m_pkthdr.pattr_af = AF_UNSPEC; 573 } 574 #endif /* ALTQ */ 575 576 #if defined (LLC) || defined (NETATALK) 577 static void 578 ether_input_llc(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh) 579 { 580 struct ifqueue *inq = NULL; 581 int isr = 0; 582 struct llc *l; 583 584 if (m->m_len < sizeof(*eh) + sizeof(struct llc)) 585 goto drop; 586 587 l = (struct llc *)(eh+1); 588 switch (l->llc_dsap) { 589 #ifdef NETATALK 590 case LLC_SNAP_LSAP: 591 switch (l->llc_control) { 592 case LLC_UI: 593 if (l->llc_ssap != LLC_SNAP_LSAP) 594 goto drop; 595 596 if (memcmp(&(l->llc_snap_org_code)[0], 597 at_org_code, sizeof(at_org_code)) == 0 && 598 ntohs(l->llc_snap_ether_type) == 599 ETHERTYPE_ATALK) { 600 inq = &atintrq2; 601 m_adj(m, sizeof(struct ether_header) 602 + sizeof(struct llc)); 603 isr = NETISR_ATALK; 604 break; 605 } 606 607 if (memcmp(&(l->llc_snap_org_code)[0], 608 aarp_org_code, 609 sizeof(aarp_org_code)) == 0 && 610 ntohs(l->llc_snap_ether_type) == 611 ETHERTYPE_AARP) { 612 m_adj(m, sizeof(struct ether_header) 613 + sizeof(struct llc)); 614 aarpinput(ifp, m); /* XXX queue? */ 615 return; 616 } 617 618 default: 619 goto drop; 620 } 621 break; 622 #endif 623 default: 624 goto drop; 625 } 626 627 KASSERT(inq != NULL); 628 IFQ_ENQUEUE_ISR(inq, m, isr); 629 return; 630 631 drop: 632 m_freem(m); 633 if_statinc(ifp, if_ierrors); /* XXX should have a dedicated counter? */ 634 return; 635 } 636 #endif /* defined (LLC) || defined (NETATALK) */ 637 638 /* 639 * Process a received Ethernet packet; 640 * the packet is in the mbuf chain m with 641 * the ether header. 642 */ 643 void 644 ether_input(struct ifnet *ifp, struct mbuf *m) 645 { 646 struct ethercom *ec = (struct ethercom *) ifp; 647 pktqueue_t *pktq = NULL; 648 struct ifqueue *inq = NULL; 649 uint16_t etype; 650 struct ether_header *eh; 651 size_t ehlen; 652 static int earlypkts; 653 int isr = 0; 654 #if NAGR > 0 655 void *agrprivate; 656 #endif 657 658 KASSERT(!cpu_intr_p()); 659 KASSERT((m->m_flags & M_PKTHDR) != 0); 660 661 if ((ifp->if_flags & IFF_UP) == 0) 662 goto drop; 663 664 #ifdef MBUFTRACE 665 m_claimm(m, &ec->ec_rx_mowner); 666 #endif 667 668 if (__predict_false(m->m_len < sizeof(*eh))) { 669 if ((m = m_pullup(m, sizeof(*eh))) == NULL) 670 goto dropped; 671 } 672 673 eh = mtod(m, struct ether_header *); 674 etype = ntohs(eh->ether_type); 675 ehlen = sizeof(*eh); 676 677 if (__predict_false(earlypkts < 100 || 678 entropy_epoch() == (unsigned)-1)) { 679 rnd_add_data(NULL, eh, ehlen, 0); 680 earlypkts++; 681 } 682 683 /* 684 * Determine if the packet is within its size limits. For MPLS the 685 * header length is variable, so we skip the check. 686 */ 687 if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len > 688 ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) { 689 #ifdef DIAGNOSTIC 690 mutex_enter(&bigpktpps_lock); 691 if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count, 692 bigpktppslim)) { 693 printf("%s: discarding oversize frame (len=%d)\n", 694 ifp->if_xname, m->m_pkthdr.len); 695 } 696 mutex_exit(&bigpktpps_lock); 697 #endif 698 goto drop; 699 } 700 701 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 702 /* 703 * If this is not a simplex interface, drop the packet 704 * if it came from us. 705 */ 706 if ((ifp->if_flags & IFF_SIMPLEX) == 0 && 707 memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost, 708 ETHER_ADDR_LEN) == 0) { 709 goto drop; 710 } 711 712 if (memcmp(etherbroadcastaddr, 713 eh->ether_dhost, ETHER_ADDR_LEN) == 0) 714 m->m_flags |= M_BCAST; 715 else 716 m->m_flags |= M_MCAST; 717 if_statinc(ifp, if_imcasts); 718 } 719 720 /* If the CRC is still on the packet, trim it off. */ 721 if (m->m_flags & M_HASFCS) { 722 m_adj(m, -ETHER_CRC_LEN); 723 m->m_flags &= ~M_HASFCS; 724 } 725 726 if_statadd(ifp, if_ibytes, m->m_pkthdr.len); 727 728 #if NCARP > 0 729 if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) { 730 /* 731 * Clear M_PROMISC, in case the packet comes from a 732 * vlan. 733 */ 734 m->m_flags &= ~M_PROMISC; 735 if (carp_input(m, (uint8_t *)&eh->ether_shost, 736 (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0) 737 return; 738 } 739 #endif 740 741 if ((m->m_flags & (M_BCAST | M_MCAST | M_PROMISC)) == 0 && 742 (ifp->if_flags & IFF_PROMISC) != 0 && 743 memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost, 744 ETHER_ADDR_LEN) != 0) { 745 m->m_flags |= M_PROMISC; 746 } 747 748 if ((m->m_flags & M_PROMISC) == 0) { 749 if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0) 750 return; 751 if (m == NULL) 752 return; 753 754 eh = mtod(m, struct ether_header *); 755 etype = ntohs(eh->ether_type); 756 } 757 758 #if NAGR > 0 759 if (ifp->if_type != IFT_IEEE8023ADLAG) { 760 agrprivate = ifp->if_lagg; 761 } else { 762 agrprivate = NULL; 763 } 764 if (agrprivate != NULL && 765 __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) { 766 m->m_flags &= ~M_PROMISC; 767 agr_input(ifp, m); 768 return; 769 } 770 #endif 771 772 /* Handle input from a lagg(4) port */ 773 if (ifp->if_type == IFT_IEEE8023ADLAG) { 774 KASSERT(lagg_input_ethernet_p != NULL); 775 m = (*lagg_input_ethernet_p)(ifp, m); 776 if (m == NULL) 777 return; 778 } 779 780 /* 781 * If VLANs are configured on the interface, check to 782 * see if the device performed the decapsulation and 783 * provided us with the tag. 784 */ 785 if (ec->ec_nvlans && vlan_has_tag(m)) { 786 #if NVLAN > 0 787 /* 788 * vlan_input() will either recursively call ether_input() 789 * or drop the packet. 790 */ 791 vlan_input(ifp, m); 792 return; 793 #else 794 goto drop; 795 #endif 796 } 797 798 /* 799 * Handle protocols that expect to have the Ethernet header 800 * (and possibly FCS) intact. 801 */ 802 switch (etype) { 803 case ETHERTYPE_VLAN: { 804 struct ether_vlan_header *evl = (void *)eh; 805 806 /* 807 * If there is a tag of 0, then the VLAN header was probably 808 * just being used to store the priority. Extract the ether 809 * type, and if IP or IPV6, let them deal with it. 810 */ 811 if (m->m_len >= sizeof(*evl) && 812 EVL_VLANOFTAG(evl->evl_tag) == 0) { 813 etype = ntohs(evl->evl_proto); 814 ehlen = sizeof(*evl); 815 if ((m->m_flags & M_PROMISC) == 0 && 816 (etype == ETHERTYPE_IP || 817 etype == ETHERTYPE_IPV6)) 818 break; 819 } 820 821 #if NVLAN > 0 822 /* 823 * vlan_input() will either recursively call ether_input() 824 * or drop the packet. 825 */ 826 if (ec->ec_nvlans != 0) { 827 vlan_input(ifp, m); 828 return; 829 } else 830 #endif 831 goto drop; 832 } 833 834 #if NPPPOE > 0 835 case ETHERTYPE_PPPOEDISC: 836 pppoedisc_input(ifp, m); 837 return; 838 839 case ETHERTYPE_PPPOE: 840 pppoe_input(ifp, m); 841 return; 842 #endif 843 844 case ETHERTYPE_SLOWPROTOCOLS: { 845 uint8_t subtype; 846 847 if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) 848 goto drop; 849 850 m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype); 851 switch (subtype) { 852 #if NAGR > 0 853 case SLOWPROTOCOLS_SUBTYPE_LACP: 854 if (agrprivate != NULL) { 855 ieee8023ad_lacp_input(ifp, m); 856 return; 857 } 858 break; 859 860 case SLOWPROTOCOLS_SUBTYPE_MARKER: 861 if (agrprivate != NULL) { 862 ieee8023ad_marker_input(ifp, m); 863 return; 864 } 865 break; 866 #endif 867 868 default: 869 if (subtype == 0 || subtype > 10) { 870 /* illegal value */ 871 goto drop; 872 } 873 /* unknown subtype */ 874 break; 875 } 876 } 877 /* FALLTHROUGH */ 878 default: 879 if (m->m_flags & M_PROMISC) 880 goto drop; 881 } 882 883 /* If the CRC is still on the packet, trim it off. */ 884 if (m->m_flags & M_HASFCS) { 885 m_adj(m, -ETHER_CRC_LEN); 886 m->m_flags &= ~M_HASFCS; 887 } 888 889 /* etype represents the size of the payload in this case */ 890 if (etype <= ETHERMTU + sizeof(struct ether_header)) { 891 KASSERT(ehlen == sizeof(*eh)); 892 #if defined (LLC) || defined (NETATALK) 893 ether_input_llc(ifp, m, eh); 894 return; 895 #else 896 goto drop; 897 #endif 898 } 899 900 /* Strip off the Ethernet header. */ 901 m_adj(m, ehlen); 902 903 switch (etype) { 904 #ifdef INET 905 case ETHERTYPE_IP: 906 #ifdef GATEWAY 907 if (ipflow_fastforward(m)) 908 return; 909 #endif 910 pktq = ip_pktq; 911 break; 912 913 case ETHERTYPE_ARP: 914 isr = NETISR_ARP; 915 inq = &arpintrq; 916 break; 917 918 case ETHERTYPE_REVARP: 919 revarpinput(m); /* XXX queue? */ 920 return; 921 #endif 922 923 #ifdef INET6 924 case ETHERTYPE_IPV6: 925 if (__predict_false(!in6_present)) 926 goto drop; 927 #ifdef GATEWAY 928 if (ip6flow_fastforward(&m)) 929 return; 930 #endif 931 pktq = ip6_pktq; 932 break; 933 #endif 934 935 #ifdef NETATALK 936 case ETHERTYPE_ATALK: 937 isr = NETISR_ATALK; 938 inq = &atintrq1; 939 break; 940 941 case ETHERTYPE_AARP: 942 aarpinput(ifp, m); /* XXX queue? */ 943 return; 944 #endif 945 946 #ifdef MPLS 947 case ETHERTYPE_MPLS: 948 isr = NETISR_MPLS; 949 inq = &mplsintrq; 950 break; 951 #endif 952 953 default: 954 goto drop; 955 } 956 957 if (__predict_true(pktq)) { 958 #ifdef NET_MPSAFE 959 const u_int h = curcpu()->ci_index; 960 #else 961 const uint32_t h = pktq_rps_hash(m); 962 #endif 963 if (__predict_false(!pktq_enqueue(pktq, m, h))) { 964 m_freem(m); 965 } 966 return; 967 } 968 969 if (__predict_false(!inq)) { 970 /* Should not happen. */ 971 goto drop; 972 } 973 974 IFQ_ENQUEUE_ISR(inq, m, isr); 975 return; 976 977 drop: 978 m_freem(m); 979 dropped: 980 if_statinc(ifp, if_ierrors); /* XXX should have a dedicated counter? */ 981 } 982 983 /* 984 * Convert Ethernet address to printable (loggable) representation. 985 */ 986 char * 987 ether_sprintf(const u_char *ap) 988 { 989 static char etherbuf[3 * ETHER_ADDR_LEN]; 990 return ether_snprintf(etherbuf, sizeof(etherbuf), ap); 991 } 992 993 char * 994 ether_snprintf(char *buf, size_t len, const u_char *ap) 995 { 996 char *cp = buf; 997 size_t i; 998 999 for (i = 0; i < len / 3; i++) { 1000 *cp++ = hexdigits[*ap >> 4]; 1001 *cp++ = hexdigits[*ap++ & 0xf]; 1002 *cp++ = ':'; 1003 } 1004 *--cp = '\0'; 1005 return buf; 1006 } 1007 1008 /* 1009 * Perform common duties while attaching to interface list 1010 */ 1011 void 1012 ether_ifattach(struct ifnet *ifp, const uint8_t *lla) 1013 { 1014 struct ethercom *ec = (struct ethercom *)ifp; 1015 1016 ifp->if_type = IFT_ETHER; 1017 ifp->if_hdrlen = ETHER_HDR_LEN; 1018 ifp->if_dlt = DLT_EN10MB; 1019 ifp->if_mtu = ETHERMTU; 1020 ifp->if_output = ether_output; 1021 ifp->_if_input = ether_input; 1022 if (ifp->if_baudrate == 0) 1023 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 1024 1025 if (lla != NULL) 1026 if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla)); 1027 1028 LIST_INIT(&ec->ec_multiaddrs); 1029 SIMPLEQ_INIT(&ec->ec_vids); 1030 ec->ec_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NET); 1031 ec->ec_flags = 0; 1032 ifp->if_broadcastaddr = etherbroadcastaddr; 1033 bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header)); 1034 #ifdef MBUFTRACE 1035 mowner_init_owner(&ec->ec_tx_mowner, ifp->if_xname, "tx"); 1036 mowner_init_owner(&ec->ec_rx_mowner, ifp->if_xname, "rx"); 1037 MOWNER_ATTACH(&ec->ec_tx_mowner); 1038 MOWNER_ATTACH(&ec->ec_rx_mowner); 1039 ifp->if_mowner = &ec->ec_tx_mowner; 1040 #endif 1041 } 1042 1043 void 1044 ether_ifdetach(struct ifnet *ifp) 1045 { 1046 struct ethercom *ec = (void *) ifp; 1047 struct ether_multi *enm; 1048 1049 IFNET_ASSERT_UNLOCKED(ifp); 1050 /* 1051 * Prevent further calls to ioctl (for example turning off 1052 * promiscuous mode from the bridge code), which eventually can 1053 * call if_init() which can cause panics because the interface 1054 * is in the process of being detached. Return device not configured 1055 * instead. 1056 */ 1057 ifp->if_ioctl = __FPTRCAST(int (*)(struct ifnet *, u_long, void *), 1058 enxio); 1059 1060 #if NBRIDGE > 0 1061 if (ifp->if_bridge) 1062 bridge_ifdetach(ifp); 1063 #endif 1064 bpf_detach(ifp); 1065 #if NVLAN > 0 1066 if (ec->ec_nvlans) 1067 vlan_ifdetach(ifp); 1068 #endif 1069 1070 #if NLAGG > 0 1071 if (ifp->if_lagg) 1072 lagg_ifdetach(ifp); 1073 #endif 1074 1075 ETHER_LOCK(ec); 1076 KASSERT(ec->ec_nvlans == 0); 1077 while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) { 1078 LIST_REMOVE(enm, enm_list); 1079 kmem_free(enm, sizeof(*enm)); 1080 ec->ec_multicnt--; 1081 } 1082 ETHER_UNLOCK(ec); 1083 1084 mutex_obj_free(ec->ec_lock); 1085 ec->ec_lock = NULL; 1086 1087 ifp->if_mowner = NULL; 1088 MOWNER_DETACH(&ec->ec_rx_mowner); 1089 MOWNER_DETACH(&ec->ec_tx_mowner); 1090 } 1091 1092 #if 0 1093 /* 1094 * This is for reference. We have a table-driven version 1095 * of the little-endian crc32 generator, which is faster 1096 * than the double-loop. 1097 */ 1098 uint32_t 1099 ether_crc32_le(const uint8_t *buf, size_t len) 1100 { 1101 uint32_t c, crc, carry; 1102 size_t i, j; 1103 1104 crc = 0xffffffffU; /* initial value */ 1105 1106 for (i = 0; i < len; i++) { 1107 c = buf[i]; 1108 for (j = 0; j < 8; j++) { 1109 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1110 crc >>= 1; 1111 c >>= 1; 1112 if (carry) 1113 crc = (crc ^ ETHER_CRC_POLY_LE); 1114 } 1115 } 1116 1117 return (crc); 1118 } 1119 #else 1120 uint32_t 1121 ether_crc32_le(const uint8_t *buf, size_t len) 1122 { 1123 static const uint32_t crctab[] = { 1124 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1125 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1126 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1127 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1128 }; 1129 uint32_t crc; 1130 size_t i; 1131 1132 crc = 0xffffffffU; /* initial value */ 1133 1134 for (i = 0; i < len; i++) { 1135 crc ^= buf[i]; 1136 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1137 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1138 } 1139 1140 return (crc); 1141 } 1142 #endif 1143 1144 uint32_t 1145 ether_crc32_be(const uint8_t *buf, size_t len) 1146 { 1147 uint32_t c, crc, carry; 1148 size_t i, j; 1149 1150 crc = 0xffffffffU; /* initial value */ 1151 1152 for (i = 0; i < len; i++) { 1153 c = buf[i]; 1154 for (j = 0; j < 8; j++) { 1155 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1156 crc <<= 1; 1157 c >>= 1; 1158 if (carry) 1159 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1160 } 1161 } 1162 1163 return (crc); 1164 } 1165 1166 #ifdef INET 1167 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] = 1168 { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 }; 1169 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] = 1170 { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff }; 1171 #endif 1172 #ifdef INET6 1173 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] = 1174 { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 }; 1175 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] = 1176 { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff }; 1177 #endif 1178 1179 /* 1180 * ether_aton implementation, not using a static buffer. 1181 */ 1182 int 1183 ether_aton_r(u_char *dest, size_t len, const char *str) 1184 { 1185 const u_char *cp = (const void *)str; 1186 u_char *ep; 1187 1188 #define atox(c) (((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10)) 1189 1190 if (len < ETHER_ADDR_LEN) 1191 return ENOSPC; 1192 1193 ep = dest + ETHER_ADDR_LEN; 1194 1195 while (*cp) { 1196 if (!isxdigit(*cp)) 1197 return EINVAL; 1198 1199 *dest = atox(*cp); 1200 cp++; 1201 if (isxdigit(*cp)) { 1202 *dest = (*dest << 4) | atox(*cp); 1203 cp++; 1204 } 1205 dest++; 1206 1207 if (dest == ep) 1208 return (*cp == '\0') ? 0 : ENAMETOOLONG; 1209 1210 switch (*cp) { 1211 case ':': 1212 case '-': 1213 case '.': 1214 cp++; 1215 break; 1216 } 1217 } 1218 return ENOBUFS; 1219 } 1220 1221 /* 1222 * Convert a sockaddr into an Ethernet address or range of Ethernet 1223 * addresses. 1224 */ 1225 int 1226 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN], 1227 uint8_t addrhi[ETHER_ADDR_LEN]) 1228 { 1229 #ifdef INET 1230 const struct sockaddr_in *sin; 1231 #endif 1232 #ifdef INET6 1233 const struct sockaddr_in6 *sin6; 1234 #endif 1235 1236 switch (sa->sa_family) { 1237 1238 case AF_UNSPEC: 1239 memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN); 1240 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1241 break; 1242 1243 #ifdef INET 1244 case AF_INET: 1245 sin = satocsin(sa); 1246 if (sin->sin_addr.s_addr == INADDR_ANY) { 1247 /* 1248 * An IP address of INADDR_ANY means listen to 1249 * or stop listening to all of the Ethernet 1250 * multicast addresses used for IP. 1251 * (This is for the sake of IP multicast routers.) 1252 */ 1253 memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN); 1254 memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN); 1255 } else { 1256 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo); 1257 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1258 } 1259 break; 1260 #endif 1261 #ifdef INET6 1262 case AF_INET6: 1263 sin6 = satocsin6(sa); 1264 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1265 /* 1266 * An IP6 address of 0 means listen to or stop 1267 * listening to all of the Ethernet multicast 1268 * address used for IP6. 1269 * (This is used for multicast routers.) 1270 */ 1271 memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN); 1272 memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN); 1273 } else { 1274 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo); 1275 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1276 } 1277 break; 1278 #endif 1279 1280 default: 1281 return EAFNOSUPPORT; 1282 } 1283 return 0; 1284 } 1285 1286 /* 1287 * Add an Ethernet multicast address or range of addresses to the list for a 1288 * given interface. 1289 */ 1290 int 1291 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec) 1292 { 1293 struct ether_multi *enm, *_enm; 1294 u_char addrlo[ETHER_ADDR_LEN]; 1295 u_char addrhi[ETHER_ADDR_LEN]; 1296 int error = 0; 1297 1298 /* Allocate out of lock */ 1299 enm = kmem_alloc(sizeof(*enm), KM_SLEEP); 1300 1301 ETHER_LOCK(ec); 1302 error = ether_multiaddr(sa, addrlo, addrhi); 1303 if (error != 0) 1304 goto out; 1305 1306 /* 1307 * Verify that we have valid Ethernet multicast addresses. 1308 */ 1309 if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) { 1310 error = EINVAL; 1311 goto out; 1312 } 1313 1314 /* 1315 * See if the address range is already in the list. 1316 */ 1317 _enm = ether_lookup_multi(addrlo, addrhi, ec); 1318 if (_enm != NULL) { 1319 /* 1320 * Found it; just increment the reference count. 1321 */ 1322 ++_enm->enm_refcount; 1323 error = 0; 1324 goto out; 1325 } 1326 1327 /* 1328 * Link a new multicast record into the interface's multicast list. 1329 */ 1330 memcpy(enm->enm_addrlo, addrlo, ETHER_ADDR_LEN); 1331 memcpy(enm->enm_addrhi, addrhi, ETHER_ADDR_LEN); 1332 enm->enm_refcount = 1; 1333 LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list); 1334 ec->ec_multicnt++; 1335 1336 /* 1337 * Return ENETRESET to inform the driver that the list has changed 1338 * and its reception filter should be adjusted accordingly. 1339 */ 1340 error = ENETRESET; 1341 enm = NULL; 1342 1343 out: 1344 ETHER_UNLOCK(ec); 1345 if (enm != NULL) 1346 kmem_free(enm, sizeof(*enm)); 1347 return error; 1348 } 1349 1350 /* 1351 * Delete a multicast address record. 1352 */ 1353 int 1354 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec) 1355 { 1356 struct ether_multi *enm; 1357 u_char addrlo[ETHER_ADDR_LEN]; 1358 u_char addrhi[ETHER_ADDR_LEN]; 1359 int error; 1360 1361 ETHER_LOCK(ec); 1362 error = ether_multiaddr(sa, addrlo, addrhi); 1363 if (error != 0) 1364 goto error; 1365 1366 /* 1367 * Look up the address in our list. 1368 */ 1369 enm = ether_lookup_multi(addrlo, addrhi, ec); 1370 if (enm == NULL) { 1371 error = ENXIO; 1372 goto error; 1373 } 1374 if (--enm->enm_refcount != 0) { 1375 /* 1376 * Still some claims to this record. 1377 */ 1378 error = 0; 1379 goto error; 1380 } 1381 1382 /* 1383 * No remaining claims to this record; unlink and free it. 1384 */ 1385 LIST_REMOVE(enm, enm_list); 1386 ec->ec_multicnt--; 1387 ETHER_UNLOCK(ec); 1388 kmem_free(enm, sizeof(*enm)); 1389 1390 /* 1391 * Return ENETRESET to inform the driver that the list has changed 1392 * and its reception filter should be adjusted accordingly. 1393 */ 1394 return ENETRESET; 1395 1396 error: 1397 ETHER_UNLOCK(ec); 1398 return error; 1399 } 1400 1401 void 1402 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb) 1403 { 1404 ec->ec_ifflags_cb = cb; 1405 } 1406 1407 void 1408 ether_set_vlan_cb(struct ethercom *ec, ether_vlancb_t cb) 1409 { 1410 1411 ec->ec_vlan_cb = cb; 1412 } 1413 1414 static int 1415 ether_ioctl_reinit(struct ethercom *ec) 1416 { 1417 struct ifnet *ifp = &ec->ec_if; 1418 int error; 1419 1420 switch (ifp->if_flags & (IFF_UP | IFF_RUNNING)) { 1421 case IFF_RUNNING: 1422 /* 1423 * If interface is marked down and it is running, 1424 * then stop and disable it. 1425 */ 1426 (*ifp->if_stop)(ifp, 1); 1427 break; 1428 case IFF_UP: 1429 /* 1430 * If interface is marked up and it is stopped, then 1431 * start it. 1432 */ 1433 return (*ifp->if_init)(ifp); 1434 case IFF_UP | IFF_RUNNING: 1435 error = 0; 1436 if (ec->ec_ifflags_cb != NULL) { 1437 error = (*ec->ec_ifflags_cb)(ec); 1438 if (error == ENETRESET) { 1439 /* 1440 * Reset the interface to pick up 1441 * changes in any other flags that 1442 * affect the hardware state. 1443 */ 1444 return (*ifp->if_init)(ifp); 1445 } 1446 } else 1447 error = (*ifp->if_init)(ifp); 1448 return error; 1449 case 0: 1450 break; 1451 } 1452 1453 return 0; 1454 } 1455 1456 /* 1457 * Common ioctls for Ethernet interfaces. Note, we must be 1458 * called at splnet(). 1459 */ 1460 int 1461 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data) 1462 { 1463 struct ethercom *ec = (void *)ifp; 1464 struct eccapreq *eccr; 1465 struct ifreq *ifr = (struct ifreq *)data; 1466 struct if_laddrreq *iflr = data; 1467 const struct sockaddr_dl *sdl; 1468 static const uint8_t zero[ETHER_ADDR_LEN]; 1469 int error; 1470 1471 switch (cmd) { 1472 case SIOCINITIFADDR: 1473 { 1474 struct ifaddr *ifa = (struct ifaddr *)data; 1475 if (ifa->ifa_addr->sa_family != AF_LINK 1476 && (ifp->if_flags & (IFF_UP | IFF_RUNNING)) != 1477 (IFF_UP | IFF_RUNNING)) { 1478 ifp->if_flags |= IFF_UP; 1479 if ((error = (*ifp->if_init)(ifp)) != 0) 1480 return error; 1481 } 1482 #ifdef INET 1483 if (ifa->ifa_addr->sa_family == AF_INET) 1484 arp_ifinit(ifp, ifa); 1485 #endif 1486 return 0; 1487 } 1488 1489 case SIOCSIFMTU: 1490 { 1491 int maxmtu; 1492 1493 if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU) 1494 maxmtu = ETHERMTU_JUMBO; 1495 else 1496 maxmtu = ETHERMTU; 1497 1498 if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu) 1499 return EINVAL; 1500 else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET) 1501 return error; 1502 else if (ifp->if_flags & IFF_UP) { 1503 /* Make sure the device notices the MTU change. */ 1504 return (*ifp->if_init)(ifp); 1505 } else 1506 return 0; 1507 } 1508 1509 case SIOCSIFFLAGS: 1510 if ((error = ifioctl_common(ifp, cmd, data)) != 0) 1511 return error; 1512 return ether_ioctl_reinit(ec); 1513 case SIOCGIFFLAGS: 1514 error = ifioctl_common(ifp, cmd, data); 1515 if (error == 0) { 1516 /* Set IFF_ALLMULTI for backcompat */ 1517 ifr->ifr_flags |= (ec->ec_flags & ETHER_F_ALLMULTI) ? 1518 IFF_ALLMULTI : 0; 1519 } 1520 return error; 1521 case SIOCGETHERCAP: 1522 eccr = (struct eccapreq *)data; 1523 eccr->eccr_capabilities = ec->ec_capabilities; 1524 eccr->eccr_capenable = ec->ec_capenable; 1525 return 0; 1526 case SIOCSETHERCAP: 1527 eccr = (struct eccapreq *)data; 1528 if ((eccr->eccr_capenable & ~ec->ec_capabilities) != 0) 1529 return EINVAL; 1530 if (eccr->eccr_capenable == ec->ec_capenable) 1531 return 0; 1532 #if 0 /* notyet */ 1533 ec->ec_capenable = (ec->ec_capenable & ETHERCAP_CANTCHANGE) 1534 | (eccr->eccr_capenable & ~ETHERCAP_CANTCHANGE); 1535 #else 1536 ec->ec_capenable = eccr->eccr_capenable; 1537 #endif 1538 return ether_ioctl_reinit(ec); 1539 case SIOCADDMULTI: 1540 return ether_addmulti(ifreq_getaddr(cmd, ifr), ec); 1541 case SIOCDELMULTI: 1542 return ether_delmulti(ifreq_getaddr(cmd, ifr), ec); 1543 case SIOCSIFMEDIA: 1544 case SIOCGIFMEDIA: 1545 if (ec->ec_mii != NULL) 1546 return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, 1547 cmd); 1548 else if (ec->ec_ifmedia != NULL) 1549 return ifmedia_ioctl(ifp, ifr, ec->ec_ifmedia, cmd); 1550 else 1551 return ENOTTY; 1552 break; 1553 case SIOCALIFADDR: 1554 sdl = satocsdl(sstocsa(&iflr->addr)); 1555 if (sdl->sdl_family != AF_LINK) 1556 ; 1557 else if (ETHER_IS_MULTICAST(CLLADDR(sdl))) 1558 return EINVAL; 1559 else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0) 1560 return EINVAL; 1561 /*FALLTHROUGH*/ 1562 default: 1563 return ifioctl_common(ifp, cmd, data); 1564 } 1565 return 0; 1566 } 1567 1568 /* 1569 * Enable/disable passing VLAN packets if the parent interface supports it. 1570 * Return: 1571 * 0: Ok 1572 * -1: Parent interface does not support vlans 1573 * >0: Error 1574 */ 1575 int 1576 ether_enable_vlan_mtu(struct ifnet *ifp) 1577 { 1578 int error; 1579 struct ethercom *ec = (void *)ifp; 1580 1581 /* Parent does not support VLAN's */ 1582 if ((ec->ec_capabilities & ETHERCAP_VLAN_MTU) == 0) 1583 return -1; 1584 1585 /* 1586 * Parent supports the VLAN_MTU capability, 1587 * i.e. can Tx/Rx larger than ETHER_MAX_LEN frames; 1588 * enable it. 1589 */ 1590 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1591 1592 /* Interface is down, defer for later */ 1593 if ((ifp->if_flags & IFF_UP) == 0) 1594 return 0; 1595 1596 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1597 return 0; 1598 1599 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1600 return error; 1601 } 1602 1603 int 1604 ether_disable_vlan_mtu(struct ifnet *ifp) 1605 { 1606 int error; 1607 struct ethercom *ec = (void *)ifp; 1608 1609 /* We still have VLAN's, defer for later */ 1610 if (ec->ec_nvlans != 0) 1611 return 0; 1612 1613 /* Parent does not support VLAB's, nothing to do. */ 1614 if ((ec->ec_capenable & ETHERCAP_VLAN_MTU) == 0) 1615 return -1; 1616 1617 /* 1618 * Disable Tx/Rx of VLAN-sized frames. 1619 */ 1620 ec->ec_capenable &= ~ETHERCAP_VLAN_MTU; 1621 1622 /* Interface is down, defer for later */ 1623 if ((ifp->if_flags & IFF_UP) == 0) 1624 return 0; 1625 1626 if ((error = if_flags_set(ifp, ifp->if_flags)) == 0) 1627 return 0; 1628 1629 ec->ec_capenable |= ETHERCAP_VLAN_MTU; 1630 return error; 1631 } 1632 1633 static int 1634 ether_multicast_sysctl(SYSCTLFN_ARGS) 1635 { 1636 struct ether_multi *enm; 1637 struct ifnet *ifp; 1638 struct ethercom *ec; 1639 int error = 0; 1640 size_t written; 1641 struct psref psref; 1642 int bound; 1643 unsigned int multicnt; 1644 struct ether_multi_sysctl *addrs; 1645 int i; 1646 1647 if (namelen != 1) 1648 return EINVAL; 1649 1650 bound = curlwp_bind(); 1651 ifp = if_get_byindex(name[0], &psref); 1652 if (ifp == NULL) { 1653 error = ENODEV; 1654 goto out; 1655 } 1656 if (ifp->if_type != IFT_ETHER) { 1657 if_put(ifp, &psref); 1658 *oldlenp = 0; 1659 goto out; 1660 } 1661 ec = (struct ethercom *)ifp; 1662 1663 if (oldp == NULL) { 1664 if_put(ifp, &psref); 1665 *oldlenp = ec->ec_multicnt * sizeof(*addrs); 1666 goto out; 1667 } 1668 1669 /* 1670 * ec->ec_lock is a spin mutex so we cannot call sysctl_copyout, which 1671 * is sleepable, while holding it. Copy data to a local buffer first 1672 * with the lock taken and then call sysctl_copyout without holding it. 1673 */ 1674 retry: 1675 multicnt = ec->ec_multicnt; 1676 1677 if (multicnt == 0) { 1678 if_put(ifp, &psref); 1679 *oldlenp = 0; 1680 goto out; 1681 } 1682 1683 addrs = kmem_zalloc(sizeof(*addrs) * multicnt, KM_SLEEP); 1684 1685 ETHER_LOCK(ec); 1686 if (multicnt != ec->ec_multicnt) { 1687 /* The number of multicast addresses has changed */ 1688 ETHER_UNLOCK(ec); 1689 kmem_free(addrs, sizeof(*addrs) * multicnt); 1690 goto retry; 1691 } 1692 1693 i = 0; 1694 LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) { 1695 struct ether_multi_sysctl *addr = &addrs[i]; 1696 addr->enm_refcount = enm->enm_refcount; 1697 memcpy(addr->enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN); 1698 memcpy(addr->enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN); 1699 i++; 1700 } 1701 ETHER_UNLOCK(ec); 1702 1703 error = 0; 1704 written = 0; 1705 for (i = 0; i < multicnt; i++) { 1706 struct ether_multi_sysctl *addr = &addrs[i]; 1707 1708 if (written + sizeof(*addr) > *oldlenp) 1709 break; 1710 error = sysctl_copyout(l, addr, oldp, sizeof(*addr)); 1711 if (error) 1712 break; 1713 written += sizeof(*addr); 1714 oldp = (char *)oldp + sizeof(*addr); 1715 } 1716 kmem_free(addrs, sizeof(*addrs) * multicnt); 1717 1718 if_put(ifp, &psref); 1719 1720 *oldlenp = written; 1721 out: 1722 curlwp_bindx(bound); 1723 return error; 1724 } 1725 1726 static void 1727 ether_sysctl_setup(struct sysctllog **clog) 1728 { 1729 const struct sysctlnode *rnode = NULL; 1730 1731 sysctl_createv(clog, 0, NULL, &rnode, 1732 CTLFLAG_PERMANENT, 1733 CTLTYPE_NODE, "ether", 1734 SYSCTL_DESCR("Ethernet-specific information"), 1735 NULL, 0, NULL, 0, 1736 CTL_NET, CTL_CREATE, CTL_EOL); 1737 1738 sysctl_createv(clog, 0, &rnode, NULL, 1739 CTLFLAG_PERMANENT, 1740 CTLTYPE_NODE, "multicast", 1741 SYSCTL_DESCR("multicast addresses"), 1742 ether_multicast_sysctl, 0, NULL, 0, 1743 CTL_CREATE, CTL_EOL); 1744 } 1745 1746 void 1747 etherinit(void) 1748 { 1749 1750 #ifdef DIAGNOSTIC 1751 mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET); 1752 #endif 1753 ether_sysctl_setup(NULL); 1754 } 1755
Indexes created Tue Sep 30 10:09:55 GMT 2025