if_ethersubr.c revision 1.238
1 /* $NetBSD: if_ethersubr.c,v 1.238 2017/02/14 03:05:06 ozaki-r 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 1.238 2017/02/14 03:05:06 ozaki-r 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/malloc.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/rnd.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 #if NBRIDGE > 0 129 #include <net/if_bridgevar.h> 130 #endif 131 132 #include <netinet/in.h> 133 #ifdef INET 134 #include <netinet/in_var.h> 135 #endif 136 #include <netinet/if_inarp.h> 137 138 #ifdef INET6 139 #ifndef INET 140 #include <netinet/in.h> 141 #endif 142 #include <netinet6/in6_var.h> 143 #include <netinet6/nd6.h> 144 #endif 145 146 147 #include "carp.h" 148 #if NCARP > 0 149 #include <netinet/ip_carp.h> 150 #endif 151 152 #ifdef NETATALK 153 #include <netatalk/at.h> 154 #include <netatalk/at_var.h> 155 #include <netatalk/at_extern.h> 156 157 #define llc_snap_org_code llc_un.type_snap.org_code 158 #define llc_snap_ether_type llc_un.type_snap.ether_type 159 160 extern u_char at_org_code[3]; 161 extern u_char aarp_org_code[3]; 162 #endif /* NETATALK */ 163 164 #ifdef MPLS 165 #include <netmpls/mpls.h> 166 #include <netmpls/mpls_var.h> 167 #endif 168 169 static struct timeval bigpktppslim_last; 170 static int bigpktppslim = 2; /* XXX */ 171 static int bigpktpps_count; 172 static kmutex_t bigpktpps_lock __cacheline_aligned; 173 174 175 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = 176 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff }; 177 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] = 178 { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 }; 179 #define senderr(e) { error = (e); goto bad;} 180 181 static int ether_output(struct ifnet *, struct mbuf *, 182 const struct sockaddr *, const struct rtentry *); 183 184 /* 185 * Ethernet output routine. 186 * Encapsulate a packet of type family for the local net. 187 * Assumes that ifp is actually pointer to ethercom structure. 188 */ 189 static int 190 ether_output(struct ifnet * const ifp0, struct mbuf * const m0, 191 const struct sockaddr * const dst, 192 const struct rtentry *rt) 193 { 194 uint16_t etype = 0; 195 int error = 0, hdrcmplt = 0; 196 uint8_t esrc[6], edst[6]; 197 struct mbuf *m = m0; 198 struct mbuf *mcopy = NULL; 199 struct ether_header *eh; 200 struct ifnet *ifp = ifp0; 201 #ifdef INET 202 struct arphdr *ah; 203 #endif /* INET */ 204 #ifdef NETATALK 205 struct at_ifaddr *aa; 206 #endif /* NETATALK */ 207 208 /* 209 * some paths such as carp_output() call ethr_output() with "ifp" 210 * argument as other than ether ifnet. 211 */ 212 KASSERT(ifp->if_output != ether_output 213 || ifp->if_extflags & IFEF_OUTPUT_MPSAFE); 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 /* NCARP > 0 */ 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 #ifndef NET_MPSAFE 251 KERNEL_LOCK(1, NULL); 252 #endif 253 if (m->m_flags & M_BCAST) 254 (void)memcpy(edst, etherbroadcastaddr, sizeof(edst)); 255 else if (m->m_flags & M_MCAST) 256 ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst); 257 else if ((error = arpresolve(ifp, rt, m, dst, edst, 258 sizeof(edst))) != 0) { 259 #ifndef NET_MPSAFE 260 KERNEL_UNLOCK_ONE(NULL); 261 #endif 262 return error == EWOULDBLOCK ? 0 : error; 263 } 264 #ifndef NET_MPSAFE 265 KERNEL_UNLOCK_ONE(NULL); 266 #endif 267 /* If broadcasting on a simplex interface, loopback a copy */ 268 if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX)) 269 mcopy = m_copy(m, 0, (int)M_COPYALL); 270 etype = htons(ETHERTYPE_IP); 271 break; 272 273 case AF_ARP: 274 ah = mtod(m, struct arphdr *); 275 if (m->m_flags & M_BCAST) 276 (void)memcpy(edst, etherbroadcastaddr, sizeof(edst)); 277 else { 278 void *tha = ar_tha(ah); 279 280 if (tha == NULL) { 281 /* fake with ARPHDR_IEEE1394 */ 282 m_freem(m); 283 return 0; 284 } 285 memcpy(edst, tha, sizeof(edst)); 286 } 287 288 ah->ar_hrd = htons(ARPHRD_ETHER); 289 290 switch (ntohs(ah->ar_op)) { 291 case ARPOP_REVREQUEST: 292 case ARPOP_REVREPLY: 293 etype = htons(ETHERTYPE_REVARP); 294 break; 295 296 case ARPOP_REQUEST: 297 case ARPOP_REPLY: 298 default: 299 etype = htons(ETHERTYPE_ARP); 300 } 301 302 break; 303 #endif 304 #ifdef INET6 305 case AF_INET6: 306 if (m->m_flags & M_BCAST) 307 (void)memcpy(edst, etherbroadcastaddr, sizeof(edst)); 308 else if (m->m_flags & M_MCAST) { 309 ETHER_MAP_IPV6_MULTICAST(&satocsin6(dst)->sin6_addr, 310 edst); 311 } else { 312 error = nd6_resolve(ifp, rt, m, dst, edst, 313 sizeof(edst)); 314 if (error != 0) 315 return error == EWOULDBLOCK ? 0 : error; 316 } 317 etype = htons(ETHERTYPE_IPV6); 318 break; 319 #endif 320 #ifdef NETATALK 321 case AF_APPLETALK: { 322 struct ifaddr *ifa; 323 int s; 324 325 KERNEL_LOCK(1, NULL); 326 if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) { 327 #ifdef NETATALKDEBUG 328 printf("aarpresolv failed\n"); 329 #endif /* NETATALKDEBUG */ 330 KERNEL_UNLOCK_ONE(NULL); 331 return (0); 332 } 333 /* 334 * ifaddr is the first thing in at_ifaddr 335 */ 336 s = pserialize_read_enter(); 337 ifa = at_ifawithnet((const struct sockaddr_at *)dst, ifp); 338 if (ifa == NULL) { 339 pserialize_read_exit(s); 340 KERNEL_UNLOCK_ONE(NULL); 341 goto bad; 342 } 343 aa = (struct at_ifaddr *)ifa; 344 345 /* 346 * In the phase 2 case, we need to prepend an mbuf for the 347 * llc header. Since we must preserve the value of m, 348 * which is passed to us by value, we m_copy() the first 349 * mbuf, and use it for our llc header. 350 */ 351 if (aa->aa_flags & AFA_PHASE2) { 352 struct llc llc; 353 354 M_PREPEND(m, sizeof(struct llc), M_DONTWAIT); 355 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; 356 llc.llc_control = LLC_UI; 357 memcpy(llc.llc_snap_org_code, at_org_code, 358 sizeof(llc.llc_snap_org_code)); 359 llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK); 360 memcpy(mtod(m, void *), &llc, sizeof(struct llc)); 361 } else { 362 etype = htons(ETHERTYPE_ATALK); 363 } 364 pserialize_read_exit(s); 365 KERNEL_UNLOCK_ONE(NULL); 366 break; 367 } 368 #endif /* NETATALK */ 369 case pseudo_AF_HDRCMPLT: 370 hdrcmplt = 1; 371 memcpy(esrc, 372 ((const struct ether_header *)dst->sa_data)->ether_shost, 373 sizeof(esrc)); 374 /* FALLTHROUGH */ 375 376 case AF_UNSPEC: 377 memcpy(edst, 378 ((const struct ether_header *)dst->sa_data)->ether_dhost, 379 sizeof(edst)); 380 /* AF_UNSPEC doesn't swap the byte order of the ether_type. */ 381 etype = ((const struct ether_header *)dst->sa_data)->ether_type; 382 break; 383 384 default: 385 printf("%s: can't handle af%d\n", ifp->if_xname, 386 dst->sa_family); 387 senderr(EAFNOSUPPORT); 388 } 389 390 #ifdef MPLS 391 KERNEL_LOCK(1, NULL); 392 { 393 struct m_tag *mtag; 394 mtag = m_tag_find(m, PACKET_TAG_MPLS, NULL); 395 if (mtag != NULL) { 396 /* Having the tag itself indicates it's MPLS */ 397 etype = htons(ETHERTYPE_MPLS); 398 m_tag_delete(m, mtag); 399 } 400 } 401 KERNEL_UNLOCK_ONE(NULL); 402 #endif 403 404 if (mcopy) 405 (void)looutput(ifp, mcopy, dst, rt); 406 407 /* If no ether type is set, this must be a 802.2 formatted packet. 408 */ 409 if (etype == 0) 410 etype = htons(m->m_pkthdr.len); 411 /* 412 * Add local net header. If no space in first mbuf, 413 * allocate another. 414 */ 415 M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT); 416 if (m == 0) 417 senderr(ENOBUFS); 418 eh = mtod(m, struct ether_header *); 419 /* Note: etype is already in network byte order. */ 420 (void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type)); 421 memcpy(eh->ether_dhost, edst, sizeof(edst)); 422 if (hdrcmplt) 423 memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost)); 424 else 425 memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl), 426 sizeof(eh->ether_shost)); 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 /* NCARP > 0 */ 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 ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN; 451 #endif /* NCARP > 0 */ 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 (m) 469 m_freem(m); 470 return (error); 471 } 472 473 #ifdef ALTQ 474 /* 475 * This routine is a slight hack to allow a packet to be classified 476 * if the Ethernet headers are present. It will go away when ALTQ's 477 * classification engine understands link headers. 478 */ 479 void 480 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m) 481 { 482 struct ether_header *eh; 483 uint16_t ether_type; 484 int hlen, af, hdrsize; 485 void *hdr; 486 487 hlen = ETHER_HDR_LEN; 488 eh = mtod(m, struct ether_header *); 489 490 ether_type = htons(eh->ether_type); 491 492 if (ether_type < ETHERMTU) { 493 /* LLC/SNAP */ 494 struct llc *llc = (struct llc *)(eh + 1); 495 hlen += 8; 496 497 if (m->m_len < hlen || 498 llc->llc_dsap != LLC_SNAP_LSAP || 499 llc->llc_ssap != LLC_SNAP_LSAP || 500 llc->llc_control != LLC_UI) { 501 /* Not SNAP. */ 502 goto bad; 503 } 504 505 ether_type = htons(llc->llc_un.type_snap.ether_type); 506 } 507 508 switch (ether_type) { 509 case ETHERTYPE_IP: 510 af = AF_INET; 511 hdrsize = 20; /* sizeof(struct ip) */ 512 break; 513 514 case ETHERTYPE_IPV6: 515 af = AF_INET6; 516 hdrsize = 40; /* sizeof(struct ip6_hdr) */ 517 break; 518 519 default: 520 af = AF_UNSPEC; 521 hdrsize = 0; 522 break; 523 } 524 525 while (m->m_len <= hlen) { 526 hlen -= m->m_len; 527 m = m->m_next; 528 } 529 if (m->m_len < (hlen + hdrsize)) { 530 /* 531 * protocol header not in a single mbuf. 532 * We can't cope with this situation right 533 * now (but it shouldn't ever happen, really, anyhow). 534 */ 535 #ifdef DEBUG 536 printf("altq_etherclassify: headers span multiple mbufs: " 537 "%d < %d\n", m->m_len, (hlen + hdrsize)); 538 #endif 539 goto bad; 540 } 541 542 m->m_data += hlen; 543 m->m_len -= hlen; 544 545 hdr = mtod(m, void *); 546 547 if (ALTQ_NEEDS_CLASSIFY(ifq)) 548 m->m_pkthdr.pattr_class = 549 (*ifq->altq_classify)(ifq->altq_clfier, m, af); 550 m->m_pkthdr.pattr_af = af; 551 m->m_pkthdr.pattr_hdr = hdr; 552 553 m->m_data -= hlen; 554 m->m_len += hlen; 555 556 return; 557 558 bad: 559 m->m_pkthdr.pattr_class = NULL; 560 m->m_pkthdr.pattr_hdr = NULL; 561 m->m_pkthdr.pattr_af = AF_UNSPEC; 562 } 563 #endif /* ALTQ */ 564 565 /* 566 * Process a received Ethernet packet; 567 * the packet is in the mbuf chain m with 568 * the ether header. 569 */ 570 void 571 ether_input(struct ifnet *ifp, struct mbuf *m) 572 { 573 struct ethercom *ec = (struct ethercom *) ifp; 574 pktqueue_t *pktq = NULL; 575 struct ifqueue *inq = NULL; 576 uint16_t etype; 577 struct ether_header *eh; 578 size_t ehlen; 579 static int earlypkts; 580 int isr = 0; 581 #if defined (LLC) || defined(NETATALK) 582 struct llc *l; 583 #endif 584 585 KASSERT(!cpu_intr_p()); 586 587 if ((ifp->if_flags & IFF_UP) == 0) { 588 m_freem(m); 589 return; 590 } 591 592 #ifdef MBUFTRACE 593 m_claimm(m, &ec->ec_rx_mowner); 594 #endif 595 eh = mtod(m, struct ether_header *); 596 etype = ntohs(eh->ether_type); 597 ehlen = sizeof(*eh); 598 599 if(__predict_false(earlypkts < 100 || !rnd_initial_entropy)) { 600 rnd_add_data(NULL, eh, ehlen, 0); 601 earlypkts++; 602 } 603 604 /* 605 * Determine if the packet is within its size limits. 606 */ 607 if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len > 608 ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) { 609 mutex_enter(&bigpktpps_lock); 610 if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count, 611 bigpktppslim)) { 612 printf("%s: discarding oversize frame (len=%d)\n", 613 ifp->if_xname, m->m_pkthdr.len); 614 } 615 mutex_exit(&bigpktpps_lock); 616 m_freem(m); 617 return; 618 } 619 620 if (ETHER_IS_MULTICAST(eh->ether_dhost)) { 621 /* 622 * If this is not a simplex interface, drop the packet 623 * if it came from us. 624 */ 625 if ((ifp->if_flags & IFF_SIMPLEX) == 0 && 626 memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost, 627 ETHER_ADDR_LEN) == 0) { 628 m_freem(m); 629 return; 630 } 631 632 if (memcmp(etherbroadcastaddr, 633 eh->ether_dhost, ETHER_ADDR_LEN) == 0) 634 m->m_flags |= M_BCAST; 635 else 636 m->m_flags |= M_MCAST; 637 ifp->if_imcasts++; 638 } 639 640 /* If the CRC is still on the packet, trim it off. */ 641 if (m->m_flags & M_HASFCS) { 642 m_adj(m, -ETHER_CRC_LEN); 643 m->m_flags &= ~M_HASFCS; 644 } 645 646 ifp->if_ibytes += m->m_pkthdr.len; 647 648 #if NCARP > 0 649 if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) { 650 /* 651 * clear M_PROMISC, in case the packets comes from a 652 * vlan 653 */ 654 m->m_flags &= ~M_PROMISC; 655 if (carp_input(m, (uint8_t *)&eh->ether_shost, 656 (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0) 657 return; 658 } 659 #endif /* NCARP > 0 */ 660 if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 && 661 (ifp->if_flags & IFF_PROMISC) != 0 && 662 memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost, 663 ETHER_ADDR_LEN) != 0) { 664 m->m_flags |= M_PROMISC; 665 } 666 667 if ((m->m_flags & M_PROMISC) == 0) { 668 if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0) 669 return; 670 if (m == NULL) 671 return; 672 673 eh = mtod(m, struct ether_header *); 674 etype = ntohs(eh->ether_type); 675 ehlen = sizeof(*eh); 676 } 677 678 #if NAGR > 0 679 if (ifp->if_agrprivate && 680 __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) { 681 m->m_flags &= ~M_PROMISC; 682 agr_input(ifp, m); 683 return; 684 } 685 #endif /* NAGR > 0 */ 686 687 /* 688 * If VLANs are configured on the interface, check to 689 * see if the device performed the decapsulation and 690 * provided us with the tag. 691 */ 692 if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) { 693 #if NVLAN > 0 694 /* 695 * vlan_input() will either recursively call ether_input() 696 * or drop the packet. 697 */ 698 vlan_input(ifp, m); 699 #else 700 m_freem(m); 701 #endif 702 return; 703 } 704 705 /* 706 * Handle protocols that expect to have the Ethernet header 707 * (and possibly FCS) intact. 708 */ 709 switch (etype) { 710 case ETHERTYPE_VLAN: { 711 struct ether_vlan_header *evl = (void *)eh; 712 /* 713 * If there is a tag of 0, then the VLAN header was probably 714 * just being used to store the priority. Extract the ether 715 * type, and if IP or IPV6, let them deal with it. 716 */ 717 if (m->m_len <= sizeof(*evl) 718 && EVL_VLANOFTAG(evl->evl_tag) == 0) { 719 etype = ntohs(evl->evl_proto); 720 ehlen = sizeof(*evl); 721 if ((m->m_flags & M_PROMISC) == 0 722 && (etype == ETHERTYPE_IP 723 || etype == ETHERTYPE_IPV6)) 724 break; 725 } 726 #if NVLAN > 0 727 /* 728 * vlan_input() will either recursively call ether_input() 729 * or drop the packet. 730 */ 731 if (((struct ethercom *)ifp)->ec_nvlans != 0) 732 vlan_input(ifp, m); 733 else 734 #endif /* NVLAN > 0 */ 735 m_freem(m); 736 return; 737 } 738 #if NPPPOE > 0 739 case ETHERTYPE_PPPOEDISC: 740 pppoedisc_input(ifp, m); 741 return; 742 case ETHERTYPE_PPPOE: 743 pppoe_input(ifp, m); 744 return; 745 #endif /* NPPPOE > 0 */ 746 case ETHERTYPE_SLOWPROTOCOLS: { 747 uint8_t subtype; 748 749 #if defined(DIAGNOSTIC) 750 if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) { 751 panic("ether_input: too short slow protocol packet"); 752 } 753 #endif 754 m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype); 755 switch (subtype) { 756 #if NAGR > 0 757 case SLOWPROTOCOLS_SUBTYPE_LACP: 758 if (ifp->if_agrprivate) { 759 ieee8023ad_lacp_input(ifp, m); 760 return; 761 } 762 break; 763 764 case SLOWPROTOCOLS_SUBTYPE_MARKER: 765 if (ifp->if_agrprivate) { 766 ieee8023ad_marker_input(ifp, m); 767 return; 768 } 769 break; 770 #endif /* NAGR > 0 */ 771 default: 772 if (subtype == 0 || subtype > 10) { 773 /* illegal value */ 774 m_freem(m); 775 return; 776 } 777 /* unknown subtype */ 778 break; 779 } 780 /* FALLTHROUGH */ 781 } 782 default: 783 if (m->m_flags & M_PROMISC) { 784 m_freem(m); 785 return; 786 } 787 } 788 789 /* If the CRC is still on the packet, trim it off. */ 790 if (m->m_flags & M_HASFCS) { 791 m_adj(m, -ETHER_CRC_LEN); 792 m->m_flags &= ~M_HASFCS; 793 } 794 795 if (etype > ETHERMTU + sizeof (struct ether_header)) { 796 /* Strip off the Ethernet header. */ 797 m_adj(m, ehlen); 798 799 switch (etype) { 800 #ifdef INET 801 case ETHERTYPE_IP: 802 #ifdef GATEWAY 803 if (ipflow_fastforward(m)) 804 return; 805 #endif 806 pktq = ip_pktq; 807 break; 808 809 case ETHERTYPE_ARP: 810 isr = NETISR_ARP; 811 inq = &arpintrq; 812 break; 813 814 case ETHERTYPE_REVARP: 815 revarpinput(m); /* XXX queue? */ 816 return; 817 #endif 818 #ifdef INET6 819 case ETHERTYPE_IPV6: 820 if (__predict_false(!in6_present)) { 821 m_freem(m); 822 return; 823 } 824 #ifdef GATEWAY 825 if (ip6flow_fastforward(&m)) 826 return; 827 #endif 828 pktq = ip6_pktq; 829 break; 830 #endif 831 #ifdef NETATALK 832 case ETHERTYPE_ATALK: 833 isr = NETISR_ATALK; 834 inq = &atintrq1; 835 break; 836 case ETHERTYPE_AARP: 837 /* probably this should be done with a NETISR as well */ 838 aarpinput(ifp, m); /* XXX */ 839 return; 840 #endif /* NETATALK */ 841 #ifdef MPLS 842 case ETHERTYPE_MPLS: 843 isr = NETISR_MPLS; 844 inq = &mplsintrq; 845 break; 846 #endif 847 default: 848 m_freem(m); 849 return; 850 } 851 } else { 852 #if defined (LLC) || defined (NETATALK) 853 l = (struct llc *)(eh+1); 854 switch (l->llc_dsap) { 855 #ifdef NETATALK 856 case LLC_SNAP_LSAP: 857 switch (l->llc_control) { 858 case LLC_UI: 859 if (l->llc_ssap != LLC_SNAP_LSAP) { 860 goto dropanyway; 861 } 862 863 if (memcmp(&(l->llc_snap_org_code)[0], 864 at_org_code, sizeof(at_org_code)) == 0 && 865 ntohs(l->llc_snap_ether_type) == 866 ETHERTYPE_ATALK) { 867 inq = &atintrq2; 868 m_adj(m, sizeof(struct ether_header) 869 + sizeof(struct llc)); 870 isr = NETISR_ATALK; 871 break; 872 } 873 874 if (memcmp(&(l->llc_snap_org_code)[0], 875 aarp_org_code, 876 sizeof(aarp_org_code)) == 0 && 877 ntohs(l->llc_snap_ether_type) == 878 ETHERTYPE_AARP) { 879 m_adj( m, sizeof(struct ether_header) 880 + sizeof(struct llc)); 881 aarpinput(ifp, m); /* XXX */ 882 return; 883 } 884 885 default: 886 goto dropanyway; 887 } 888 break; 889 dropanyway: 890 #endif 891 default: 892 m_freem(m); 893 return; 894 } 895 #else /* ISO || LLC || NETATALK*/ 896 m_freem(m); 897 return; 898 #endif /* ISO || LLC || NETATALK*/ 899 } 900 901 if (__predict_true(pktq)) { 902 #ifdef NET_MPSAFE 903 const u_int h = curcpu()->ci_index; 904 #else 905 const uint32_t h = pktq_rps_hash(m); 906 #endif 907 if (__predict_false(!pktq_enqueue(pktq, m, h))) { 908 m_freem(m); 909 } 910 return; 911 } 912 913 if (__predict_false(!inq)) { 914 /* Should not happen. */ 915 m_freem(m); 916 return; 917 } 918 919 IFQ_LOCK(inq); 920 if (IF_QFULL(inq)) { 921 IF_DROP(inq); 922 IFQ_UNLOCK(inq); 923 m_freem(m); 924 } else { 925 IF_ENQUEUE(inq, m); 926 IFQ_UNLOCK(inq); 927 schednetisr(isr); 928 } 929 } 930 931 /* 932 * Convert Ethernet address to printable (loggable) representation. 933 */ 934 char * 935 ether_sprintf(const u_char *ap) 936 { 937 static char etherbuf[3 * ETHER_ADDR_LEN]; 938 return ether_snprintf(etherbuf, sizeof(etherbuf), ap); 939 } 940 941 char * 942 ether_snprintf(char *buf, size_t len, const u_char *ap) 943 { 944 char *cp = buf; 945 size_t i; 946 947 for (i = 0; i < len / 3; i++) { 948 *cp++ = hexdigits[*ap >> 4]; 949 *cp++ = hexdigits[*ap++ & 0xf]; 950 *cp++ = ':'; 951 } 952 *--cp = '\0'; 953 return buf; 954 } 955 956 /* 957 * Perform common duties while attaching to interface list 958 */ 959 void 960 ether_ifattach(struct ifnet *ifp, const uint8_t *lla) 961 { 962 struct ethercom *ec = (struct ethercom *)ifp; 963 964 ifp->if_extflags |= IFEF_OUTPUT_MPSAFE; 965 ifp->if_type = IFT_ETHER; 966 ifp->if_hdrlen = ETHER_HDR_LEN; 967 ifp->if_dlt = DLT_EN10MB; 968 ifp->if_mtu = ETHERMTU; 969 ifp->if_output = ether_output; 970 ifp->_if_input = ether_input; 971 if (ifp->if_baudrate == 0) 972 ifp->if_baudrate = IF_Mbps(10); /* just a default */ 973 974 if (lla != NULL) 975 if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla)); 976 977 LIST_INIT(&ec->ec_multiaddrs); 978 ec->ec_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NET); 979 ifp->if_broadcastaddr = etherbroadcastaddr; 980 bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header)); 981 #ifdef MBUFTRACE 982 strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname, 983 sizeof(ec->ec_tx_mowner.mo_name)); 984 strlcpy(ec->ec_tx_mowner.mo_descr, "tx", 985 sizeof(ec->ec_tx_mowner.mo_descr)); 986 strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname, 987 sizeof(ec->ec_rx_mowner.mo_name)); 988 strlcpy(ec->ec_rx_mowner.mo_descr, "rx", 989 sizeof(ec->ec_rx_mowner.mo_descr)); 990 MOWNER_ATTACH(&ec->ec_tx_mowner); 991 MOWNER_ATTACH(&ec->ec_rx_mowner); 992 ifp->if_mowner = &ec->ec_tx_mowner; 993 #endif 994 } 995 996 void 997 ether_ifdetach(struct ifnet *ifp) 998 { 999 struct ethercom *ec = (void *) ifp; 1000 struct ether_multi *enm; 1001 1002 /* 1003 * Prevent further calls to ioctl (for example turning off 1004 * promiscuous mode from the bridge code), which eventually can 1005 * call if_init() which can cause panics because the interface 1006 * is in the process of being detached. Return device not configured 1007 * instead. 1008 */ 1009 ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio; 1010 1011 #if NBRIDGE > 0 1012 if (ifp->if_bridge) 1013 bridge_ifdetach(ifp); 1014 #endif 1015 1016 bpf_detach(ifp); 1017 1018 #if NVLAN > 0 1019 if (ec->ec_nvlans) 1020 vlan_ifdetach(ifp); 1021 #endif 1022 1023 mutex_enter(ec->ec_lock); 1024 while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) { 1025 LIST_REMOVE(enm, enm_list); 1026 kmem_free(enm, sizeof(*enm)); 1027 ec->ec_multicnt--; 1028 } 1029 mutex_exit(ec->ec_lock); 1030 1031 mutex_destroy(ec->ec_lock); 1032 1033 ifp->if_mowner = NULL; 1034 MOWNER_DETACH(&ec->ec_rx_mowner); 1035 MOWNER_DETACH(&ec->ec_tx_mowner); 1036 } 1037 1038 #if 0 1039 /* 1040 * This is for reference. We have a table-driven version 1041 * of the little-endian crc32 generator, which is faster 1042 * than the double-loop. 1043 */ 1044 uint32_t 1045 ether_crc32_le(const uint8_t *buf, size_t len) 1046 { 1047 uint32_t c, crc, carry; 1048 size_t i, j; 1049 1050 crc = 0xffffffffU; /* initial value */ 1051 1052 for (i = 0; i < len; i++) { 1053 c = buf[i]; 1054 for (j = 0; j < 8; j++) { 1055 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); 1056 crc >>= 1; 1057 c >>= 1; 1058 if (carry) 1059 crc = (crc ^ ETHER_CRC_POLY_LE); 1060 } 1061 } 1062 1063 return (crc); 1064 } 1065 #else 1066 uint32_t 1067 ether_crc32_le(const uint8_t *buf, size_t len) 1068 { 1069 static const uint32_t crctab[] = { 1070 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, 1071 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, 1072 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, 1073 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c 1074 }; 1075 uint32_t crc; 1076 size_t i; 1077 1078 crc = 0xffffffffU; /* initial value */ 1079 1080 for (i = 0; i < len; i++) { 1081 crc ^= buf[i]; 1082 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1083 crc = (crc >> 4) ^ crctab[crc & 0xf]; 1084 } 1085 1086 return (crc); 1087 } 1088 #endif 1089 1090 uint32_t 1091 ether_crc32_be(const uint8_t *buf, size_t len) 1092 { 1093 uint32_t c, crc, carry; 1094 size_t i, j; 1095 1096 crc = 0xffffffffU; /* initial value */ 1097 1098 for (i = 0; i < len; i++) { 1099 c = buf[i]; 1100 for (j = 0; j < 8; j++) { 1101 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); 1102 crc <<= 1; 1103 c >>= 1; 1104 if (carry) 1105 crc = (crc ^ ETHER_CRC_POLY_BE) | carry; 1106 } 1107 } 1108 1109 return (crc); 1110 } 1111 1112 #ifdef INET 1113 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] = 1114 { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 }; 1115 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] = 1116 { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff }; 1117 #endif 1118 #ifdef INET6 1119 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] = 1120 { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 }; 1121 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] = 1122 { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff }; 1123 #endif 1124 1125 /* 1126 * ether_aton implementation, not using a static buffer. 1127 */ 1128 int 1129 ether_aton_r(u_char *dest, size_t len, const char *str) 1130 { 1131 const u_char *cp = (const void *)str; 1132 u_char *ep; 1133 1134 #define atox(c) (((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10)) 1135 1136 if (len < ETHER_ADDR_LEN) 1137 return ENOSPC; 1138 1139 ep = dest + ETHER_ADDR_LEN; 1140 1141 while (*cp) { 1142 if (!isxdigit(*cp)) 1143 return EINVAL; 1144 *dest = atox(*cp); 1145 cp++; 1146 if (isxdigit(*cp)) { 1147 *dest = (*dest << 4) | atox(*cp); 1148 dest++; 1149 cp++; 1150 } else 1151 dest++; 1152 if (dest == ep) 1153 return *cp == '\0' ? 0 : ENAMETOOLONG; 1154 switch (*cp) { 1155 case ':': 1156 case '-': 1157 case '.': 1158 cp++; 1159 break; 1160 } 1161 } 1162 return ENOBUFS; 1163 } 1164 1165 /* 1166 * Convert a sockaddr into an Ethernet address or range of Ethernet 1167 * addresses. 1168 */ 1169 int 1170 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN], 1171 uint8_t addrhi[ETHER_ADDR_LEN]) 1172 { 1173 #ifdef INET 1174 const struct sockaddr_in *sin; 1175 #endif /* INET */ 1176 #ifdef INET6 1177 const struct sockaddr_in6 *sin6; 1178 #endif /* INET6 */ 1179 1180 switch (sa->sa_family) { 1181 1182 case AF_UNSPEC: 1183 memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN); 1184 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1185 break; 1186 1187 #ifdef INET 1188 case AF_INET: 1189 sin = satocsin(sa); 1190 if (sin->sin_addr.s_addr == INADDR_ANY) { 1191 /* 1192 * An IP address of INADDR_ANY means listen to 1193 * or stop listening to all of the Ethernet 1194 * multicast addresses used for IP. 1195 * (This is for the sake of IP multicast routers.) 1196 */ 1197 memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN); 1198 memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN); 1199 } 1200 else { 1201 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo); 1202 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1203 } 1204 break; 1205 #endif 1206 #ifdef INET6 1207 case AF_INET6: 1208 sin6 = satocsin6(sa); 1209 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { 1210 /* 1211 * An IP6 address of 0 means listen to or stop 1212 * listening to all of the Ethernet multicast 1213 * address used for IP6. 1214 * (This is used for multicast routers.) 1215 */ 1216 memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN); 1217 memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN); 1218 } else { 1219 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo); 1220 memcpy(addrhi, addrlo, ETHER_ADDR_LEN); 1221 } 1222 break; 1223 #endif 1224 1225 default: 1226 return EAFNOSUPPORT; 1227 } 1228 return 0; 1229 } 1230 1231 /* 1232 * Add an Ethernet multicast address or range of addresses to the list for a 1233 * given interface. 1234 */ 1235 int 1236 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec) 1237 { 1238 struct ether_multi *enm, *_enm; 1239 u_char addrlo[ETHER_ADDR_LEN]; 1240 u_char addrhi[ETHER_ADDR_LEN]; 1241 int error = 0; 1242 1243 /* Allocate out of lock */ 1244 /* XXX still can be called in softint */ 1245 enm = kmem_intr_alloc(sizeof(*enm), KM_SLEEP); 1246 if (enm == NULL) 1247 return ENOBUFS; 1248 1249 mutex_enter(ec->ec_lock); 1250 error = ether_multiaddr(sa, addrlo, addrhi); 1251 if (error != 0) 1252 goto out; 1253 1254 /* 1255 * Verify that we have valid Ethernet multicast addresses. 1256 */ 1257 if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) { 1258 error = EINVAL; 1259 goto out; 1260 } 1261 /* 1262 * See if the address range is already in the list. 1263 */ 1264 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, _enm); 1265 if (_enm != NULL) { 1266 /* 1267 * Found it; just increment the reference count. 1268 */ 1269 ++_enm->enm_refcount; 1270 error = 0; 1271 goto out; 1272 } 1273 /* 1274 * New address or range; malloc a new multicast record 1275 * and link it into the interface's multicast list. 1276 */ 1277 memcpy(enm->enm_addrlo, addrlo, 6); 1278 memcpy(enm->enm_addrhi, addrhi, 6); 1279 enm->enm_refcount = 1; 1280 LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list); 1281 ec->ec_multicnt++; 1282 /* 1283 * Return ENETRESET to inform the driver that the list has changed 1284 * and its reception filter should be adjusted accordingly. 1285 */ 1286 error = ENETRESET; 1287 enm = NULL; 1288 out: 1289 mutex_exit(ec->ec_lock); 1290 if (enm != NULL) 1291 kmem_free(enm, sizeof(*enm)); 1292 return error; 1293 } 1294 1295 /* 1296 * Delete a multicast address record. 1297 */ 1298 int 1299 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec) 1300 { 1301 struct ether_multi *enm; 1302 u_char addrlo[ETHER_ADDR_LEN]; 1303 u_char addrhi[ETHER_ADDR_LEN]; 1304 int error; 1305 1306 mutex_enter(ec->ec_lock); 1307 error = ether_multiaddr(sa, addrlo, addrhi); 1308 if (error != 0) 1309 goto error; 1310 1311 /* 1312 * Look ur the address in our list. 1313 */ 1314 ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm); 1315 if (enm == NULL) { 1316 error = ENXIO; 1317 goto error; 1318 } 1319 if (--enm->enm_refcount != 0) { 1320 /* 1321 * Still some claims to this record. 1322 */ 1323 error = 0; 1324 goto error; 1325 } 1326 /* 1327 * No remaining claims to this record; unlink and free it. 1328 */ 1329 LIST_REMOVE(enm, enm_list); 1330 ec->ec_multicnt--; 1331 mutex_exit(ec->ec_lock); 1332 1333 kmem_free(enm, sizeof(*enm)); 1334 /* 1335 * Return ENETRESET to inform the driver that the list has changed 1336 * and its reception filter should be adjusted accordingly. 1337 */ 1338 return ENETRESET; 1339 error: 1340 mutex_exit(ec->ec_lock); 1341 return error; 1342 } 1343 1344 void 1345 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb) 1346 { 1347 ec->ec_ifflags_cb = cb; 1348 } 1349 1350 /* 1351 * Common ioctls for Ethernet interfaces. Note, we must be 1352 * called at splnet(). 1353 */ 1354 int 1355 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data) 1356 { 1357 struct ethercom *ec = (void *) ifp; 1358 struct eccapreq *eccr; 1359 struct ifreq *ifr = (struct ifreq *)data; 1360 struct if_laddrreq *iflr = data; 1361 const struct sockaddr_dl *sdl; 1362 static const uint8_t zero[ETHER_ADDR_LEN]; 1363 int error; 1364 1365 switch (cmd) { 1366 case SIOCINITIFADDR: 1367 { 1368 struct ifaddr *ifa = (struct ifaddr *)data; 1369 if (ifa->ifa_addr->sa_family != AF_LINK 1370 && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) != 1371 (IFF_UP|IFF_RUNNING)) { 1372 ifp->if_flags |= IFF_UP; 1373 if ((error = (*ifp->if_init)(ifp)) != 0) 1374 return error; 1375 } 1376 #ifdef INET 1377 if (ifa->ifa_addr->sa_family == AF_INET) 1378 arp_ifinit(ifp, ifa); 1379 #endif /* INET */ 1380 return 0; 1381 } 1382 1383 case SIOCSIFMTU: 1384 { 1385 int maxmtu; 1386 1387 if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU) 1388 maxmtu = ETHERMTU_JUMBO; 1389 else 1390 maxmtu = ETHERMTU; 1391 1392 if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu) 1393 return EINVAL; 1394 else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET) 1395 return error; 1396 else if (ifp->if_flags & IFF_UP) { 1397 /* Make sure the device notices the MTU change. */ 1398 return (*ifp->if_init)(ifp); 1399 } else 1400 return 0; 1401 } 1402 1403 case SIOCSIFFLAGS: 1404 if ((error = ifioctl_common(ifp, cmd, data)) != 0) 1405 return error; 1406 switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) { 1407 case IFF_RUNNING: 1408 /* 1409 * If interface is marked down and it is running, 1410 * then stop and disable it. 1411 */ 1412 (*ifp->if_stop)(ifp, 1); 1413 break; 1414 case IFF_UP: 1415 /* 1416 * If interface is marked up and it is stopped, then 1417 * start it. 1418 */ 1419 return (*ifp->if_init)(ifp); 1420 case IFF_UP|IFF_RUNNING: 1421 error = 0; 1422 if (ec->ec_ifflags_cb == NULL || 1423 (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) { 1424 /* 1425 * Reset the interface to pick up 1426 * changes in any other flags that 1427 * affect the hardware state. 1428 */ 1429 return (*ifp->if_init)(ifp); 1430 } else 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, with holding it. Copy data to a local buffer first 1569 * with holding it and then call sysctl_copyout without holding it. 1570 */ 1571 retry: 1572 multicnt = ec->ec_multicnt; 1573 addrs = kmem_alloc(sizeof(*addrs) * multicnt, KM_SLEEP); 1574 1575 mutex_enter(ec->ec_lock); 1576 if (multicnt < ec->ec_multicnt) { 1577 /* The number of multicast addresses have increased */ 1578 mutex_exit(ec->ec_lock); 1579 kmem_free(addrs, sizeof(*addrs) * multicnt); 1580 goto retry; 1581 } 1582 1583 i = 0; 1584 LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) { 1585 struct ether_multi_sysctl *addr = &addrs[i]; 1586 addr->enm_refcount = enm->enm_refcount; 1587 memcpy(addr->enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN); 1588 memcpy(addr->enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN); 1589 i++; 1590 } 1591 mutex_exit(ec->ec_lock); 1592 1593 error = 0; 1594 written = 0; 1595 for (i = 0; i < multicnt; i++) { 1596 struct ether_multi_sysctl *addr = &addrs[i]; 1597 1598 if (written + sizeof(*addr) > *oldlenp) 1599 break; 1600 error = sysctl_copyout(l, addr, oldp, sizeof(*addr)); 1601 if (error) 1602 break; 1603 written += sizeof(*addr); 1604 oldp = (char *)oldp + sizeof(*addr); 1605 } 1606 kmem_free(addrs, sizeof(*addrs) * multicnt); 1607 1608 if_put(ifp, &psref); 1609 1610 *oldlenp = written; 1611 out: 1612 curlwp_bindx(bound); 1613 return error; 1614 } 1615 1616 static void 1617 ether_sysctl_setup(struct sysctllog **clog) 1618 { 1619 const struct sysctlnode *rnode = NULL; 1620 1621 sysctl_createv(clog, 0, NULL, &rnode, 1622 CTLFLAG_PERMANENT, 1623 CTLTYPE_NODE, "ether", 1624 SYSCTL_DESCR("Ethernet-specific information"), 1625 NULL, 0, NULL, 0, 1626 CTL_NET, CTL_CREATE, CTL_EOL); 1627 1628 sysctl_createv(clog, 0, &rnode, NULL, 1629 CTLFLAG_PERMANENT, 1630 CTLTYPE_NODE, "multicast", 1631 SYSCTL_DESCR("multicast addresses"), 1632 ether_multicast_sysctl, 0, NULL, 0, 1633 CTL_CREATE, CTL_EOL); 1634 } 1635 1636 void 1637 etherinit(void) 1638 { 1639 1640 mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET); 1641 ether_sysctl_setup(NULL); 1642 } 1643
Indexes created Sun Sep 21 20:09:37 GMT 2025