Home | History | Annotate | Line # | Download | only in net
if_ethersubr.c revision 1.222
      1 /*	$NetBSD: if_ethersubr.c,v 1.222 2016/04/28 00:16:56 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 1.222 2016/04/28 00:16:56 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 
     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 #ifndef NET_MPSAFE
    208 	KASSERT(KERNEL_LOCKED_P());
    209 #endif
    210 
    211 #ifdef MBUFTRACE
    212 	m_claimm(m, ifp->if_mowner);
    213 #endif
    214 
    215 #if NCARP > 0
    216 	if (ifp->if_type == IFT_CARP) {
    217 		struct ifaddr *ifa;
    218 
    219 		/* loop back if this is going to the carp interface */
    220 		if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP &&
    221 		    (ifa = ifa_ifwithaddr(dst)) != NULL &&
    222 		    ifa->ifa_ifp == ifp0)
    223 			return looutput(ifp0, m, dst, rt);
    224 
    225 		ifp = ifp->if_carpdev;
    226 		/* ac = (struct arpcom *)ifp; */
    227 
    228 		if ((ifp0->if_flags & (IFF_UP|IFF_RUNNING)) !=
    229 		    (IFF_UP|IFF_RUNNING))
    230 			senderr(ENETDOWN);
    231 	}
    232 #endif /* NCARP > 0 */
    233 
    234 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
    235 		senderr(ENETDOWN);
    236 
    237 	switch (dst->sa_family) {
    238 
    239 #ifdef INET
    240 	case AF_INET:
    241 		if (m->m_flags & M_BCAST)
    242 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
    243 		else if (m->m_flags & M_MCAST)
    244 			ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst);
    245 		else if ((error = arpresolve(ifp, rt, m, dst, edst,
    246 		    sizeof(edst))) != 0)
    247 			return error == EWOULDBLOCK ? 0 : error;
    248 		/* If broadcasting on a simplex interface, loopback a copy */
    249 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
    250 			mcopy = m_copy(m, 0, (int)M_COPYALL);
    251 		etype = htons(ETHERTYPE_IP);
    252 		break;
    253 
    254 	case AF_ARP:
    255 		ah = mtod(m, struct arphdr *);
    256 		if (m->m_flags & M_BCAST)
    257 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
    258 		else {
    259 			void *tha = ar_tha(ah);
    260 
    261 			if (tha == NULL) {
    262 				/* fake with ARPHDR_IEEE1394 */
    263 				return 0;
    264 			}
    265 			memcpy(edst, tha, sizeof(edst));
    266 		}
    267 
    268 		ah->ar_hrd = htons(ARPHRD_ETHER);
    269 
    270 		switch (ntohs(ah->ar_op)) {
    271 		case ARPOP_REVREQUEST:
    272 		case ARPOP_REVREPLY:
    273 			etype = htons(ETHERTYPE_REVARP);
    274 			break;
    275 
    276 		case ARPOP_REQUEST:
    277 		case ARPOP_REPLY:
    278 		default:
    279 			etype = htons(ETHERTYPE_ARP);
    280 		}
    281 
    282 		break;
    283 #endif
    284 #ifdef INET6
    285 	case AF_INET6:
    286 		if (!nd6_storelladdr(ifp, rt, m, dst, edst, sizeof(edst))){
    287 			/* something bad happened */
    288 			return (0);
    289 		}
    290 		etype = htons(ETHERTYPE_IPV6);
    291 		break;
    292 #endif
    293 #ifdef NETATALK
    294     case AF_APPLETALK:
    295 		if (aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) {
    296 #ifdef NETATALKDEBUG
    297 			printf("aarpresolv failed\n");
    298 #endif /* NETATALKDEBUG */
    299 			return (0);
    300 		}
    301 		/*
    302 		 * ifaddr is the first thing in at_ifaddr
    303 		 */
    304 		aa = (struct at_ifaddr *) at_ifawithnet(
    305 		    (const struct sockaddr_at *)dst, ifp);
    306 		if (aa == NULL)
    307 		    goto bad;
    308 
    309 		/*
    310 		 * In the phase 2 case, we need to prepend an mbuf for the
    311 		 * llc header.  Since we must preserve the value of m,
    312 		 * which is passed to us by value, we m_copy() the first
    313 		 * mbuf, and use it for our llc header.
    314 		 */
    315 		if (aa->aa_flags & AFA_PHASE2) {
    316 			struct llc llc;
    317 
    318 			M_PREPEND(m, sizeof(struct llc), M_DONTWAIT);
    319 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
    320 			llc.llc_control = LLC_UI;
    321 			memcpy(llc.llc_snap_org_code, at_org_code,
    322 			    sizeof(llc.llc_snap_org_code));
    323 			llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK);
    324 			memcpy(mtod(m, void *), &llc, sizeof(struct llc));
    325 		} else {
    326 			etype = htons(ETHERTYPE_ATALK);
    327 		}
    328 		break;
    329 #endif /* NETATALK */
    330 	case pseudo_AF_HDRCMPLT:
    331 		hdrcmplt = 1;
    332 		memcpy(esrc,
    333 		    ((const struct ether_header *)dst->sa_data)->ether_shost,
    334 		    sizeof(esrc));
    335 		/* FALLTHROUGH */
    336 
    337 	case AF_UNSPEC:
    338  		memcpy(edst,
    339 		    ((const struct ether_header *)dst->sa_data)->ether_dhost,
    340 		    sizeof(edst));
    341 		/* AF_UNSPEC doesn't swap the byte order of the ether_type. */
    342 		etype = ((const struct ether_header *)dst->sa_data)->ether_type;
    343 		break;
    344 
    345 	default:
    346 		printf("%s: can't handle af%d\n", ifp->if_xname,
    347 			dst->sa_family);
    348 		senderr(EAFNOSUPPORT);
    349 	}
    350 
    351 #ifdef MPLS
    352 	{
    353 		struct m_tag *mtag;
    354 		mtag = m_tag_find(m, PACKET_TAG_MPLS, NULL);
    355 		if (mtag != NULL) {
    356 			/* Having the tag itself indicates it's MPLS */
    357 			etype = htons(ETHERTYPE_MPLS);
    358 			m_tag_delete(m, mtag);
    359 		}
    360 	}
    361 #endif
    362 
    363 	if (mcopy)
    364 		(void)looutput(ifp, mcopy, dst, rt);
    365 
    366 	/* If no ether type is set, this must be a 802.2 formatted packet.
    367 	 */
    368 	if (etype == 0)
    369 		etype = htons(m->m_pkthdr.len);
    370 	/*
    371 	 * Add local net header.  If no space in first mbuf,
    372 	 * allocate another.
    373 	 */
    374 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
    375 	if (m == 0)
    376 		senderr(ENOBUFS);
    377 	eh = mtod(m, struct ether_header *);
    378 	/* Note: etype is already in network byte order. */
    379 	(void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
    380  	memcpy(eh->ether_dhost, edst, sizeof(edst));
    381 	if (hdrcmplt)
    382 		memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost));
    383 	else
    384 	 	memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl),
    385 		    sizeof(eh->ether_shost));
    386 
    387 #if NCARP > 0
    388 	if (ifp0 != ifp && ifp0->if_type == IFT_CARP) {
    389 	 	memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl),
    390 		    sizeof(eh->ether_shost));
    391 	}
    392 #endif /* NCARP > 0 */
    393 
    394 	if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0)
    395 		return (error);
    396 	if (m == NULL)
    397 		return (0);
    398 
    399 #if NBRIDGE > 0
    400 	/*
    401 	 * Bridges require special output handling.
    402 	 */
    403 	if (ifp->if_bridge)
    404 		return (bridge_output(ifp, m, NULL, NULL));
    405 #endif
    406 
    407 #if NCARP > 0
    408 	if (ifp != ifp0)
    409 		ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN;
    410 #endif /* NCARP > 0 */
    411 
    412 #ifdef ALTQ
    413 	/*
    414 	 * If ALTQ is enabled on the parent interface, do
    415 	 * classification; the queueing discipline might not
    416 	 * require classification, but might require the
    417 	 * address family/header pointer in the pktattr.
    418 	 */
    419 	if (ALTQ_IS_ENABLED(&ifp->if_snd))
    420 		altq_etherclassify(&ifp->if_snd, m);
    421 #endif
    422 	return ifq_enqueue(ifp, m);
    423 
    424 bad:
    425 	if (m)
    426 		m_freem(m);
    427 	return (error);
    428 }
    429 
    430 #ifdef ALTQ
    431 /*
    432  * This routine is a slight hack to allow a packet to be classified
    433  * if the Ethernet headers are present.  It will go away when ALTQ's
    434  * classification engine understands link headers.
    435  */
    436 void
    437 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m)
    438 {
    439 	struct ether_header *eh;
    440 	uint16_t ether_type;
    441 	int hlen, af, hdrsize;
    442 	void *hdr;
    443 
    444 	hlen = ETHER_HDR_LEN;
    445 	eh = mtod(m, struct ether_header *);
    446 
    447 	ether_type = htons(eh->ether_type);
    448 
    449 	if (ether_type < ETHERMTU) {
    450 		/* LLC/SNAP */
    451 		struct llc *llc = (struct llc *)(eh + 1);
    452 		hlen += 8;
    453 
    454 		if (m->m_len < hlen ||
    455 		    llc->llc_dsap != LLC_SNAP_LSAP ||
    456 		    llc->llc_ssap != LLC_SNAP_LSAP ||
    457 		    llc->llc_control != LLC_UI) {
    458 			/* Not SNAP. */
    459 			goto bad;
    460 		}
    461 
    462 		ether_type = htons(llc->llc_un.type_snap.ether_type);
    463 	}
    464 
    465 	switch (ether_type) {
    466 	case ETHERTYPE_IP:
    467 		af = AF_INET;
    468 		hdrsize = 20;		/* sizeof(struct ip) */
    469 		break;
    470 
    471 	case ETHERTYPE_IPV6:
    472 		af = AF_INET6;
    473 		hdrsize = 40;		/* sizeof(struct ip6_hdr) */
    474 		break;
    475 
    476 	default:
    477 		af = AF_UNSPEC;
    478 		hdrsize = 0;
    479 		break;
    480 	}
    481 
    482 	while (m->m_len <= hlen) {
    483 		hlen -= m->m_len;
    484 		m = m->m_next;
    485 	}
    486 	if (m->m_len < (hlen + hdrsize)) {
    487 		/*
    488 		 * protocol header not in a single mbuf.
    489 		 * We can't cope with this situation right
    490 		 * now (but it shouldn't ever happen, really, anyhow).
    491 		 */
    492 #ifdef DEBUG
    493 		printf("altq_etherclassify: headers span multiple mbufs: "
    494 		    "%d < %d\n", m->m_len, (hlen + hdrsize));
    495 #endif
    496 		goto bad;
    497 	}
    498 
    499 	m->m_data += hlen;
    500 	m->m_len -= hlen;
    501 
    502 	hdr = mtod(m, void *);
    503 
    504 	if (ALTQ_NEEDS_CLASSIFY(ifq))
    505 		m->m_pkthdr.pattr_class =
    506 		    (*ifq->altq_classify)(ifq->altq_clfier, m, af);
    507 	m->m_pkthdr.pattr_af = af;
    508 	m->m_pkthdr.pattr_hdr = hdr;
    509 
    510 	m->m_data -= hlen;
    511 	m->m_len += hlen;
    512 
    513 	return;
    514 
    515  bad:
    516 	m->m_pkthdr.pattr_class = NULL;
    517 	m->m_pkthdr.pattr_hdr = NULL;
    518 	m->m_pkthdr.pattr_af = AF_UNSPEC;
    519 }
    520 #endif /* ALTQ */
    521 
    522 /*
    523  * Process a received Ethernet packet;
    524  * the packet is in the mbuf chain m with
    525  * the ether header.
    526  */
    527 void
    528 ether_input(struct ifnet *ifp, struct mbuf *m)
    529 {
    530 	struct ethercom *ec = (struct ethercom *) ifp;
    531 	pktqueue_t *pktq = NULL;
    532 	struct ifqueue *inq = NULL;
    533 	uint16_t etype;
    534 	struct ether_header *eh;
    535 	size_t ehlen;
    536 	static int earlypkts;
    537 	int isr = 0;
    538 #if defined (LLC) || defined(NETATALK)
    539 	struct llc *l;
    540 #endif
    541 
    542 	KASSERT(!cpu_intr_p());
    543 
    544 	if ((ifp->if_flags & IFF_UP) == 0) {
    545 		m_freem(m);
    546 		return;
    547 	}
    548 
    549 #ifdef MBUFTRACE
    550 	m_claimm(m, &ec->ec_rx_mowner);
    551 #endif
    552 	eh = mtod(m, struct ether_header *);
    553 	etype = ntohs(eh->ether_type);
    554 	ehlen = sizeof(*eh);
    555 
    556 	if(__predict_false(earlypkts < 100 || !rnd_initial_entropy)) {
    557 		rnd_add_data(NULL, eh, ehlen, 0);
    558 		earlypkts++;
    559 	}
    560 
    561 	/*
    562 	 * Determine if the packet is within its size limits.
    563 	 */
    564 	if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len >
    565 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
    566 		mutex_enter(&bigpktpps_lock);
    567 		if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count,
    568 			    bigpktppslim)) {
    569 			printf("%s: discarding oversize frame (len=%d)\n",
    570 			    ifp->if_xname, m->m_pkthdr.len);
    571 		}
    572 		mutex_exit(&bigpktpps_lock);
    573 		m_freem(m);
    574 		return;
    575 	}
    576 
    577 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
    578 		/*
    579 		 * If this is not a simplex interface, drop the packet
    580 		 * if it came from us.
    581 		 */
    582 		if ((ifp->if_flags & IFF_SIMPLEX) == 0 &&
    583 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost,
    584 		    ETHER_ADDR_LEN) == 0) {
    585 			m_freem(m);
    586 			return;
    587 		}
    588 
    589 		if (memcmp(etherbroadcastaddr,
    590 		    eh->ether_dhost, ETHER_ADDR_LEN) == 0)
    591 			m->m_flags |= M_BCAST;
    592 		else
    593 			m->m_flags |= M_MCAST;
    594 		ifp->if_imcasts++;
    595 	}
    596 
    597 	/* If the CRC is still on the packet, trim it off. */
    598 	if (m->m_flags & M_HASFCS) {
    599 		m_adj(m, -ETHER_CRC_LEN);
    600 		m->m_flags &= ~M_HASFCS;
    601 	}
    602 
    603 	ifp->if_ibytes += m->m_pkthdr.len;
    604 
    605 #if NCARP > 0
    606 	if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) {
    607 		/*
    608 		 * clear M_PROMISC, in case the packets comes from a
    609 		 * vlan
    610 		 */
    611 		m->m_flags &= ~M_PROMISC;
    612 		if (carp_input(m, (uint8_t *)&eh->ether_shost,
    613 		    (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0)
    614 			return;
    615 	}
    616 #endif /* NCARP > 0 */
    617 	if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 &&
    618 	    (ifp->if_flags & IFF_PROMISC) != 0 &&
    619 	    memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost,
    620 		   ETHER_ADDR_LEN) != 0) {
    621 		m->m_flags |= M_PROMISC;
    622 	}
    623 
    624 	if ((m->m_flags & M_PROMISC) == 0) {
    625 		if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0)
    626 			return;
    627 		if (m == NULL)
    628 			return;
    629 
    630 		eh = mtod(m, struct ether_header *);
    631 		etype = ntohs(eh->ether_type);
    632 		ehlen = sizeof(*eh);
    633 	}
    634 
    635 #if NAGR > 0
    636 	if (ifp->if_agrprivate &&
    637 	    __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) {
    638 		m->m_flags &= ~M_PROMISC;
    639 		agr_input(ifp, m);
    640 		return;
    641 	}
    642 #endif /* NAGR > 0 */
    643 
    644 	/*
    645 	 * If VLANs are configured on the interface, check to
    646 	 * see if the device performed the decapsulation and
    647 	 * provided us with the tag.
    648 	 */
    649 	if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) {
    650 #if NVLAN > 0
    651 		/*
    652 		 * vlan_input() will either recursively call ether_input()
    653 		 * or drop the packet.
    654 		 */
    655 		vlan_input(ifp, m);
    656 #else
    657 		m_freem(m);
    658 #endif
    659 		return;
    660 	}
    661 
    662 	/*
    663 	 * Handle protocols that expect to have the Ethernet header
    664 	 * (and possibly FCS) intact.
    665 	 */
    666 	switch (etype) {
    667 	case ETHERTYPE_VLAN: {
    668 		struct ether_vlan_header *evl = (void *)eh;
    669 		/*
    670 		 * If there is a tag of 0, then the VLAN header was probably
    671 		 * just being used to store the priority.  Extract the ether
    672 		 * type, and if IP or IPV6, let them deal with it.
    673 		 */
    674 		if (m->m_len <= sizeof(*evl)
    675 		    && EVL_VLANOFTAG(evl->evl_tag) == 0) {
    676 			etype = ntohs(evl->evl_proto);
    677 			ehlen = sizeof(*evl);
    678 			if ((m->m_flags & M_PROMISC) == 0
    679 			    && (etype == ETHERTYPE_IP
    680 				|| etype == ETHERTYPE_IPV6))
    681 				break;
    682 		}
    683 #if NVLAN > 0
    684 		/*
    685 		 * vlan_input() will either recursively call ether_input()
    686 		 * or drop the packet.
    687 		 */
    688 		if (((struct ethercom *)ifp)->ec_nvlans != 0)
    689 			vlan_input(ifp, m);
    690 		else
    691 #endif /* NVLAN > 0 */
    692 			m_freem(m);
    693 		return;
    694 	}
    695 #if NPPPOE > 0
    696 	case ETHERTYPE_PPPOEDISC:
    697 		pppoedisc_input(ifp, m);
    698 		return;
    699 	case ETHERTYPE_PPPOE:
    700 		pppoe_input(ifp, m);
    701 		return;
    702 #endif /* NPPPOE > 0 */
    703 	case ETHERTYPE_SLOWPROTOCOLS: {
    704 		uint8_t subtype;
    705 
    706 #if defined(DIAGNOSTIC)
    707 		if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) {
    708 			panic("ether_input: too short slow protocol packet");
    709 		}
    710 #endif
    711 		m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype);
    712 		switch (subtype) {
    713 #if NAGR > 0
    714 		case SLOWPROTOCOLS_SUBTYPE_LACP:
    715 			if (ifp->if_agrprivate) {
    716 				ieee8023ad_lacp_input(ifp, m);
    717 				return;
    718 			}
    719 			break;
    720 
    721 		case SLOWPROTOCOLS_SUBTYPE_MARKER:
    722 			if (ifp->if_agrprivate) {
    723 				ieee8023ad_marker_input(ifp, m);
    724 				return;
    725 			}
    726 			break;
    727 #endif /* NAGR > 0 */
    728 		default:
    729 			if (subtype == 0 || subtype > 10) {
    730 				/* illegal value */
    731 				m_freem(m);
    732 				return;
    733 			}
    734 			/* unknown subtype */
    735 			break;
    736 		}
    737 		/* FALLTHROUGH */
    738 	}
    739 	default:
    740 		if (m->m_flags & M_PROMISC) {
    741 			m_freem(m);
    742 			return;
    743 		}
    744 	}
    745 
    746 	/* If the CRC is still on the packet, trim it off. */
    747 	if (m->m_flags & M_HASFCS) {
    748 		m_adj(m, -ETHER_CRC_LEN);
    749 		m->m_flags &= ~M_HASFCS;
    750 	}
    751 
    752 	if (etype > ETHERMTU + sizeof (struct ether_header)) {
    753 		/* Strip off the Ethernet header. */
    754 		m_adj(m, ehlen);
    755 
    756 		switch (etype) {
    757 #ifdef INET
    758 		case ETHERTYPE_IP:
    759 #ifdef GATEWAY
    760 			if (ipflow_fastforward(m))
    761 				return;
    762 #endif
    763 			pktq = ip_pktq;
    764 			break;
    765 
    766 		case ETHERTYPE_ARP:
    767 			isr = NETISR_ARP;
    768 			inq = &arpintrq;
    769 			break;
    770 
    771 		case ETHERTYPE_REVARP:
    772 			revarpinput(m);	/* XXX queue? */
    773 			return;
    774 #endif
    775 #ifdef INET6
    776 		case ETHERTYPE_IPV6:
    777 			if (__predict_false(!in6_present)) {
    778 				m_freem(m);
    779 				return;
    780 			}
    781 #ifdef GATEWAY
    782 			if (ip6flow_fastforward(&m))
    783 				return;
    784 #endif
    785 			pktq = ip6_pktq;
    786 			break;
    787 #endif
    788 #ifdef NETATALK
    789 		case ETHERTYPE_ATALK:
    790 			isr = NETISR_ATALK;
    791 			inq = &atintrq1;
    792 			break;
    793 		case ETHERTYPE_AARP:
    794 			/* probably this should be done with a NETISR as well */
    795 			aarpinput(ifp, m); /* XXX */
    796 			return;
    797 #endif /* NETATALK */
    798 #ifdef MPLS
    799 		case ETHERTYPE_MPLS:
    800 			isr = NETISR_MPLS;
    801 			inq = &mplsintrq;
    802 			break;
    803 #endif
    804 		default:
    805 			m_freem(m);
    806 			return;
    807 		}
    808 	} else {
    809 #if defined (LLC) || defined (NETATALK)
    810 		l = (struct llc *)(eh+1);
    811 		switch (l->llc_dsap) {
    812 #ifdef NETATALK
    813 		case LLC_SNAP_LSAP:
    814 			switch (l->llc_control) {
    815 			case LLC_UI:
    816 				if (l->llc_ssap != LLC_SNAP_LSAP) {
    817 					goto dropanyway;
    818 				}
    819 
    820 				if (memcmp(&(l->llc_snap_org_code)[0],
    821 				    at_org_code, sizeof(at_org_code)) == 0 &&
    822 				    ntohs(l->llc_snap_ether_type) ==
    823 				    ETHERTYPE_ATALK) {
    824 					inq = &atintrq2;
    825 					m_adj(m, sizeof(struct ether_header)
    826 					    + sizeof(struct llc));
    827 					isr = NETISR_ATALK;
    828 					break;
    829 				}
    830 
    831 				if (memcmp(&(l->llc_snap_org_code)[0],
    832 				    aarp_org_code,
    833 				    sizeof(aarp_org_code)) == 0 &&
    834 				    ntohs(l->llc_snap_ether_type) ==
    835 				    ETHERTYPE_AARP) {
    836 					m_adj( m, sizeof(struct ether_header)
    837 					    + sizeof(struct llc));
    838 					aarpinput(ifp, m); /* XXX */
    839 				    return;
    840 				}
    841 
    842 			default:
    843 				goto dropanyway;
    844 			}
    845 			break;
    846 		dropanyway:
    847 #endif
    848 		default:
    849 			m_freem(m);
    850 			return;
    851 		}
    852 #else /* ISO || LLC || NETATALK*/
    853 		m_freem(m);
    854 		return;
    855 #endif /* ISO || LLC || NETATALK*/
    856 	}
    857 
    858 	if (__predict_true(pktq)) {
    859 		const uint32_t h = pktq_rps_hash(m);
    860 		if (__predict_false(!pktq_enqueue(pktq, m, h))) {
    861 			m_freem(m);
    862 		}
    863 		return;
    864 	}
    865 
    866 	if (__predict_false(!inq)) {
    867 		/* Should not happen. */
    868 		m_freem(m);
    869 		return;
    870 	}
    871 	if (IF_QFULL(inq)) {
    872 		IF_DROP(inq);
    873 		m_freem(m);
    874 	} else {
    875 		IF_ENQUEUE(inq, m);
    876 		schednetisr(isr);
    877 	}
    878 }
    879 
    880 /*
    881  * Convert Ethernet address to printable (loggable) representation.
    882  */
    883 char *
    884 ether_sprintf(const u_char *ap)
    885 {
    886 	static char etherbuf[3 * ETHER_ADDR_LEN];
    887 	return ether_snprintf(etherbuf, sizeof(etherbuf), ap);
    888 }
    889 
    890 char *
    891 ether_snprintf(char *buf, size_t len, const u_char *ap)
    892 {
    893 	char *cp = buf;
    894 	size_t i;
    895 
    896 	for (i = 0; i < len / 3; i++) {
    897 		*cp++ = hexdigits[*ap >> 4];
    898 		*cp++ = hexdigits[*ap++ & 0xf];
    899 		*cp++ = ':';
    900 	}
    901 	*--cp = '\0';
    902 	return buf;
    903 }
    904 
    905 /*
    906  * Perform common duties while attaching to interface list
    907  */
    908 void
    909 ether_ifattach(struct ifnet *ifp, const uint8_t *lla)
    910 {
    911 	struct ethercom *ec = (struct ethercom *)ifp;
    912 
    913 	ifp->if_type = IFT_ETHER;
    914 	ifp->if_hdrlen = ETHER_HDR_LEN;
    915 	ifp->if_dlt = DLT_EN10MB;
    916 	ifp->if_mtu = ETHERMTU;
    917 	ifp->if_output = ether_output;
    918 	ifp->_if_input = ether_input;
    919 	if (ifp->if_baudrate == 0)
    920 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
    921 
    922 	if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla));
    923 
    924 	LIST_INIT(&ec->ec_multiaddrs);
    925 	ifp->if_broadcastaddr = etherbroadcastaddr;
    926 	bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header));
    927 #ifdef MBUFTRACE
    928 	strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname,
    929 	    sizeof(ec->ec_tx_mowner.mo_name));
    930 	strlcpy(ec->ec_tx_mowner.mo_descr, "tx",
    931 	    sizeof(ec->ec_tx_mowner.mo_descr));
    932 	strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname,
    933 	    sizeof(ec->ec_rx_mowner.mo_name));
    934 	strlcpy(ec->ec_rx_mowner.mo_descr, "rx",
    935 	    sizeof(ec->ec_rx_mowner.mo_descr));
    936 	MOWNER_ATTACH(&ec->ec_tx_mowner);
    937 	MOWNER_ATTACH(&ec->ec_rx_mowner);
    938 	ifp->if_mowner = &ec->ec_tx_mowner;
    939 #endif
    940 }
    941 
    942 void
    943 ether_ifdetach(struct ifnet *ifp)
    944 {
    945 	struct ethercom *ec = (void *) ifp;
    946 	struct ether_multi *enm;
    947 	int s;
    948 
    949 	/*
    950 	 * Prevent further calls to ioctl (for example turning off
    951 	 * promiscuous mode from the bridge code), which eventually can
    952 	 * call if_init() which can cause panics because the interface
    953 	 * is in the process of being detached. Return device not configured
    954 	 * instead.
    955 	 */
    956 	ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio;
    957 
    958 #if NBRIDGE > 0
    959 	if (ifp->if_bridge)
    960 		bridge_ifdetach(ifp);
    961 #endif
    962 
    963 	bpf_detach(ifp);
    964 
    965 #if NVLAN > 0
    966 	if (ec->ec_nvlans)
    967 		vlan_ifdetach(ifp);
    968 #endif
    969 
    970 	s = splnet();
    971 	while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) {
    972 		LIST_REMOVE(enm, enm_list);
    973 		free(enm, M_IFMADDR);
    974 		ec->ec_multicnt--;
    975 	}
    976 	splx(s);
    977 
    978 	ifp->if_mowner = NULL;
    979 	MOWNER_DETACH(&ec->ec_rx_mowner);
    980 	MOWNER_DETACH(&ec->ec_tx_mowner);
    981 }
    982 
    983 #if 0
    984 /*
    985  * This is for reference.  We have a table-driven version
    986  * of the little-endian crc32 generator, which is faster
    987  * than the double-loop.
    988  */
    989 uint32_t
    990 ether_crc32_le(const uint8_t *buf, size_t len)
    991 {
    992 	uint32_t c, crc, carry;
    993 	size_t i, j;
    994 
    995 	crc = 0xffffffffU;	/* initial value */
    996 
    997 	for (i = 0; i < len; i++) {
    998 		c = buf[i];
    999 		for (j = 0; j < 8; j++) {
   1000 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
   1001 			crc >>= 1;
   1002 			c >>= 1;
   1003 			if (carry)
   1004 				crc = (crc ^ ETHER_CRC_POLY_LE);
   1005 		}
   1006 	}
   1007 
   1008 	return (crc);
   1009 }
   1010 #else
   1011 uint32_t
   1012 ether_crc32_le(const uint8_t *buf, size_t len)
   1013 {
   1014 	static const uint32_t crctab[] = {
   1015 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
   1016 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
   1017 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
   1018 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
   1019 	};
   1020 	uint32_t crc;
   1021 	size_t i;
   1022 
   1023 	crc = 0xffffffffU;	/* initial value */
   1024 
   1025 	for (i = 0; i < len; i++) {
   1026 		crc ^= buf[i];
   1027 		crc = (crc >> 4) ^ crctab[crc & 0xf];
   1028 		crc = (crc >> 4) ^ crctab[crc & 0xf];
   1029 	}
   1030 
   1031 	return (crc);
   1032 }
   1033 #endif
   1034 
   1035 uint32_t
   1036 ether_crc32_be(const uint8_t *buf, size_t len)
   1037 {
   1038 	uint32_t c, crc, carry;
   1039 	size_t i, j;
   1040 
   1041 	crc = 0xffffffffU;	/* initial value */
   1042 
   1043 	for (i = 0; i < len; i++) {
   1044 		c = buf[i];
   1045 		for (j = 0; j < 8; j++) {
   1046 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
   1047 			crc <<= 1;
   1048 			c >>= 1;
   1049 			if (carry)
   1050 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
   1051 		}
   1052 	}
   1053 
   1054 	return (crc);
   1055 }
   1056 
   1057 #ifdef INET
   1058 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] =
   1059     { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 };
   1060 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] =
   1061     { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff };
   1062 #endif
   1063 #ifdef INET6
   1064 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] =
   1065     { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 };
   1066 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] =
   1067     { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff };
   1068 #endif
   1069 
   1070 /*
   1071  * ether_aton implementation, not using a static buffer.
   1072  */
   1073 int
   1074 ether_aton_r(u_char *dest, size_t len, const char *str)
   1075 {
   1076         const u_char *cp = (const void *)str;
   1077 	u_char *ep;
   1078 
   1079 #define atox(c)	(((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10))
   1080 
   1081 	if (len < ETHER_ADDR_LEN)
   1082 		return ENOSPC;
   1083 
   1084 	ep = dest + ETHER_ADDR_LEN;
   1085 
   1086 	while (*cp) {
   1087                 if (!isxdigit(*cp))
   1088                         return EINVAL;
   1089 		*dest = atox(*cp);
   1090 		cp++;
   1091                 if (isxdigit(*cp)) {
   1092                         *dest = (*dest << 4) | atox(*cp);
   1093 			dest++;
   1094 			cp++;
   1095                 } else
   1096 			dest++;
   1097 		if (dest == ep)
   1098 			return *cp == '\0' ? 0 : ENAMETOOLONG;
   1099 		switch (*cp) {
   1100 		case ':':
   1101 		case '-':
   1102 		case '.':
   1103 			cp++;
   1104 			break;
   1105 		}
   1106         }
   1107 	return ENOBUFS;
   1108 }
   1109 
   1110 /*
   1111  * Convert a sockaddr into an Ethernet address or range of Ethernet
   1112  * addresses.
   1113  */
   1114 int
   1115 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN],
   1116     uint8_t addrhi[ETHER_ADDR_LEN])
   1117 {
   1118 #ifdef INET
   1119 	const struct sockaddr_in *sin;
   1120 #endif /* INET */
   1121 #ifdef INET6
   1122 	const struct sockaddr_in6 *sin6;
   1123 #endif /* INET6 */
   1124 
   1125 	switch (sa->sa_family) {
   1126 
   1127 	case AF_UNSPEC:
   1128 		memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN);
   1129 		memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
   1130 		break;
   1131 
   1132 #ifdef INET
   1133 	case AF_INET:
   1134 		sin = satocsin(sa);
   1135 		if (sin->sin_addr.s_addr == INADDR_ANY) {
   1136 			/*
   1137 			 * An IP address of INADDR_ANY means listen to
   1138 			 * or stop listening to all of the Ethernet
   1139 			 * multicast addresses used for IP.
   1140 			 * (This is for the sake of IP multicast routers.)
   1141 			 */
   1142 			memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN);
   1143 			memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN);
   1144 		}
   1145 		else {
   1146 			ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo);
   1147 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
   1148 		}
   1149 		break;
   1150 #endif
   1151 #ifdef INET6
   1152 	case AF_INET6:
   1153 		sin6 = satocsin6(sa);
   1154 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
   1155 			/*
   1156 			 * An IP6 address of 0 means listen to or stop
   1157 			 * listening to all of the Ethernet multicast
   1158 			 * address used for IP6.
   1159 			 * (This is used for multicast routers.)
   1160 			 */
   1161 			memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN);
   1162 			memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN);
   1163 		} else {
   1164 			ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo);
   1165 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
   1166 		}
   1167 		break;
   1168 #endif
   1169 
   1170 	default:
   1171 		return EAFNOSUPPORT;
   1172 	}
   1173 	return 0;
   1174 }
   1175 
   1176 /*
   1177  * Add an Ethernet multicast address or range of addresses to the list for a
   1178  * given interface.
   1179  */
   1180 int
   1181 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec)
   1182 {
   1183 	struct ether_multi *enm;
   1184 	u_char addrlo[ETHER_ADDR_LEN];
   1185 	u_char addrhi[ETHER_ADDR_LEN];
   1186 	int s = splnet(), error;
   1187 
   1188 	error = ether_multiaddr(sa, addrlo, addrhi);
   1189 	if (error != 0) {
   1190 		splx(s);
   1191 		return error;
   1192 	}
   1193 
   1194 	/*
   1195 	 * Verify that we have valid Ethernet multicast addresses.
   1196 	 */
   1197 	if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) {
   1198 		splx(s);
   1199 		return EINVAL;
   1200 	}
   1201 	/*
   1202 	 * See if the address range is already in the list.
   1203 	 */
   1204 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
   1205 	if (enm != NULL) {
   1206 		/*
   1207 		 * Found it; just increment the reference count.
   1208 		 */
   1209 		++enm->enm_refcount;
   1210 		splx(s);
   1211 		return 0;
   1212 	}
   1213 	/*
   1214 	 * New address or range; malloc a new multicast record
   1215 	 * and link it into the interface's multicast list.
   1216 	 */
   1217 	enm = (struct ether_multi *)malloc(sizeof(*enm), M_IFMADDR, M_NOWAIT);
   1218 	if (enm == NULL) {
   1219 		splx(s);
   1220 		return ENOBUFS;
   1221 	}
   1222 	memcpy(enm->enm_addrlo, addrlo, 6);
   1223 	memcpy(enm->enm_addrhi, addrhi, 6);
   1224 	enm->enm_refcount = 1;
   1225 	LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list);
   1226 	ec->ec_multicnt++;
   1227 	splx(s);
   1228 	/*
   1229 	 * Return ENETRESET to inform the driver that the list has changed
   1230 	 * and its reception filter should be adjusted accordingly.
   1231 	 */
   1232 	return ENETRESET;
   1233 }
   1234 
   1235 /*
   1236  * Delete a multicast address record.
   1237  */
   1238 int
   1239 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec)
   1240 {
   1241 	struct ether_multi *enm;
   1242 	u_char addrlo[ETHER_ADDR_LEN];
   1243 	u_char addrhi[ETHER_ADDR_LEN];
   1244 	int s = splnet(), error;
   1245 
   1246 	error = ether_multiaddr(sa, addrlo, addrhi);
   1247 	if (error != 0) {
   1248 		splx(s);
   1249 		return (error);
   1250 	}
   1251 
   1252 	/*
   1253 	 * Look ur the address in our list.
   1254 	 */
   1255 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
   1256 	if (enm == NULL) {
   1257 		splx(s);
   1258 		return (ENXIO);
   1259 	}
   1260 	if (--enm->enm_refcount != 0) {
   1261 		/*
   1262 		 * Still some claims to this record.
   1263 		 */
   1264 		splx(s);
   1265 		return (0);
   1266 	}
   1267 	/*
   1268 	 * No remaining claims to this record; unlink and free it.
   1269 	 */
   1270 	LIST_REMOVE(enm, enm_list);
   1271 	free(enm, M_IFMADDR);
   1272 	ec->ec_multicnt--;
   1273 	splx(s);
   1274 	/*
   1275 	 * Return ENETRESET to inform the driver that the list has changed
   1276 	 * and its reception filter should be adjusted accordingly.
   1277 	 */
   1278 	return (ENETRESET);
   1279 }
   1280 
   1281 void
   1282 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb)
   1283 {
   1284 	ec->ec_ifflags_cb = cb;
   1285 }
   1286 
   1287 /*
   1288  * Common ioctls for Ethernet interfaces.  Note, we must be
   1289  * called at splnet().
   1290  */
   1291 int
   1292 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data)
   1293 {
   1294 	struct ethercom *ec = (void *) ifp;
   1295 	struct eccapreq *eccr;
   1296 	struct ifreq *ifr = (struct ifreq *)data;
   1297 	struct if_laddrreq *iflr = data;
   1298 	const struct sockaddr_dl *sdl;
   1299 	static const uint8_t zero[ETHER_ADDR_LEN];
   1300 	int error;
   1301 
   1302 	switch (cmd) {
   1303 	case SIOCINITIFADDR:
   1304 	    {
   1305 		struct ifaddr *ifa = (struct ifaddr *)data;
   1306 		if (ifa->ifa_addr->sa_family != AF_LINK
   1307 		    && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) !=
   1308 		       (IFF_UP|IFF_RUNNING)) {
   1309 			ifp->if_flags |= IFF_UP;
   1310 			if ((error = (*ifp->if_init)(ifp)) != 0)
   1311 				return error;
   1312 		}
   1313 #ifdef INET
   1314 		if (ifa->ifa_addr->sa_family == AF_INET)
   1315 			arp_ifinit(ifp, ifa);
   1316 #endif /* INET */
   1317 		return 0;
   1318 	    }
   1319 
   1320 	case SIOCSIFMTU:
   1321 	    {
   1322 		int maxmtu;
   1323 
   1324 		if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU)
   1325 			maxmtu = ETHERMTU_JUMBO;
   1326 		else
   1327 			maxmtu = ETHERMTU;
   1328 
   1329 		if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu)
   1330 			return EINVAL;
   1331 		else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET)
   1332 			return error;
   1333 		else if (ifp->if_flags & IFF_UP) {
   1334 			/* Make sure the device notices the MTU change. */
   1335 			return (*ifp->if_init)(ifp);
   1336 		} else
   1337 			return 0;
   1338 	    }
   1339 
   1340 	case SIOCSIFFLAGS:
   1341 		if ((error = ifioctl_common(ifp, cmd, data)) != 0)
   1342 			return error;
   1343 		switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) {
   1344 		case IFF_RUNNING:
   1345 			/*
   1346 			 * If interface is marked down and it is running,
   1347 			 * then stop and disable it.
   1348 			 */
   1349 			(*ifp->if_stop)(ifp, 1);
   1350 			break;
   1351 		case IFF_UP:
   1352 			/*
   1353 			 * If interface is marked up and it is stopped, then
   1354 			 * start it.
   1355 			 */
   1356 			return (*ifp->if_init)(ifp);
   1357 		case IFF_UP|IFF_RUNNING:
   1358 			error = 0;
   1359 			if (ec->ec_ifflags_cb == NULL ||
   1360 			    (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) {
   1361 				/*
   1362 				 * Reset the interface to pick up
   1363 				 * changes in any other flags that
   1364 				 * affect the hardware state.
   1365 				 */
   1366 				return (*ifp->if_init)(ifp);
   1367 			} else
   1368 				return error;
   1369 		case 0:
   1370 			break;
   1371 		}
   1372 		return 0;
   1373 	case SIOCGETHERCAP:
   1374 		eccr = (struct eccapreq *)data;
   1375 		eccr->eccr_capabilities = ec->ec_capabilities;
   1376 		eccr->eccr_capenable = ec->ec_capenable;
   1377 		return 0;
   1378 	case SIOCADDMULTI:
   1379 		return ether_addmulti(ifreq_getaddr(cmd, ifr), ec);
   1380 	case SIOCDELMULTI:
   1381 		return ether_delmulti(ifreq_getaddr(cmd, ifr), ec);
   1382 	case SIOCSIFMEDIA:
   1383 	case SIOCGIFMEDIA:
   1384 		if (ec->ec_mii == NULL)
   1385 			return ENOTTY;
   1386 		return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd);
   1387 	case SIOCALIFADDR:
   1388 		sdl = satocsdl(sstocsa(&iflr->addr));
   1389 		if (sdl->sdl_family != AF_LINK)
   1390 			;
   1391 		else if (ETHER_IS_MULTICAST(CLLADDR(sdl)))
   1392 			return EINVAL;
   1393 		else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0)
   1394 			return EINVAL;
   1395 		/*FALLTHROUGH*/
   1396 	default:
   1397 		return ifioctl_common(ifp, cmd, data);
   1398 	}
   1399 	return 0;
   1400 }
   1401 
   1402 /*
   1403  * Enable/disable passing VLAN packets if the parent interface supports it.
   1404  * Return:
   1405  * 	 0: Ok
   1406  *	-1: Parent interface does not support vlans
   1407  *	>0: Error
   1408  */
   1409 int
   1410 ether_enable_vlan_mtu(struct ifnet *ifp)
   1411 {
   1412 	int error;
   1413 	struct ethercom *ec = (void *)ifp;
   1414 
   1415 	/* Already have VLAN's do nothing. */
   1416 	if (ec->ec_nvlans != 0)
   1417 		return 0;
   1418 
   1419 	/* Parent does not support VLAN's */
   1420 	if ((ec->ec_capabilities & ETHERCAP_VLAN_MTU) == 0)
   1421 		return -1;
   1422 
   1423 	/*
   1424 	 * Parent supports the VLAN_MTU capability,
   1425 	 * i.e. can Tx/Rx larger than ETHER_MAX_LEN frames;
   1426 	 * enable it.
   1427 	 */
   1428 	ec->ec_capenable |= ETHERCAP_VLAN_MTU;
   1429 
   1430 	/* Interface is down, defer for later */
   1431 	if ((ifp->if_flags & IFF_UP) == 0)
   1432 		return 0;
   1433 
   1434 	if ((error = if_flags_set(ifp, ifp->if_flags)) == 0)
   1435 		return 0;
   1436 
   1437 	ec->ec_capenable &= ~ETHERCAP_VLAN_MTU;
   1438 	return error;
   1439 }
   1440 
   1441 int
   1442 ether_disable_vlan_mtu(struct ifnet *ifp)
   1443 {
   1444 	int error;
   1445 	struct ethercom *ec = (void *)ifp;
   1446 
   1447 	/* We still have VLAN's, defer for later */
   1448 	if (ec->ec_nvlans != 0)
   1449 		return 0;
   1450 
   1451 	/* Parent does not support VLAB's, nothing to do. */
   1452 	if ((ec->ec_capenable & ETHERCAP_VLAN_MTU) == 0)
   1453 		return -1;
   1454 
   1455 	/*
   1456 	 * Disable Tx/Rx of VLAN-sized frames.
   1457 	 */
   1458 	ec->ec_capenable &= ~ETHERCAP_VLAN_MTU;
   1459 
   1460 	/* Interface is down, defer for later */
   1461 	if ((ifp->if_flags & IFF_UP) == 0)
   1462 		return 0;
   1463 
   1464 	if ((error = if_flags_set(ifp, ifp->if_flags)) == 0)
   1465 		return 0;
   1466 
   1467 	ec->ec_capenable |= ETHERCAP_VLAN_MTU;
   1468 	return error;
   1469 }
   1470 
   1471 static int
   1472 ether_multicast_sysctl(SYSCTLFN_ARGS)
   1473 {
   1474 	struct ether_multi *enm;
   1475 	struct ether_multi_sysctl addr;
   1476 	struct ifnet *ifp;
   1477 	struct ethercom *ec;
   1478 	int error;
   1479 	size_t written;
   1480 
   1481 	if (namelen != 1)
   1482 		return EINVAL;
   1483 
   1484 	ifp = if_byindex(name[0]);
   1485 	if (ifp == NULL)
   1486 		return ENODEV;
   1487 	if (ifp->if_type != IFT_ETHER) {
   1488 		*oldlenp = 0;
   1489 		return 0;
   1490 	}
   1491 	ec = (struct ethercom *)ifp;
   1492 
   1493 	if (oldp == NULL) {
   1494 		*oldlenp = ec->ec_multicnt * sizeof(addr);
   1495 		return 0;
   1496 	}
   1497 
   1498 	memset(&addr, 0, sizeof(addr));
   1499 	error = 0;
   1500 	written = 0;
   1501 
   1502 	LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) {
   1503 		if (written + sizeof(addr) > *oldlenp)
   1504 			break;
   1505 		addr.enm_refcount = enm->enm_refcount;
   1506 		memcpy(addr.enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN);
   1507 		memcpy(addr.enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN);
   1508 		error = sysctl_copyout(l, &addr, oldp, sizeof(addr));
   1509 		if (error)
   1510 			break;
   1511 		written += sizeof(addr);
   1512 		oldp = (char *)oldp + sizeof(addr);
   1513 	}
   1514 
   1515 	*oldlenp = written;
   1516 	return error;
   1517 }
   1518 
   1519 SYSCTL_SETUP(sysctl_net_ether_setup, "sysctl net.ether subtree setup")
   1520 {
   1521 	const struct sysctlnode *rnode = NULL;
   1522 
   1523 	sysctl_createv(clog, 0, NULL, &rnode,
   1524 		       CTLFLAG_PERMANENT,
   1525 		       CTLTYPE_NODE, "ether",
   1526 		       SYSCTL_DESCR("Ethernet-specific information"),
   1527 		       NULL, 0, NULL, 0,
   1528 		       CTL_NET, CTL_CREATE, CTL_EOL);
   1529 
   1530 	sysctl_createv(clog, 0, &rnode, NULL,
   1531 		       CTLFLAG_PERMANENT,
   1532 		       CTLTYPE_NODE, "multicast",
   1533 		       SYSCTL_DESCR("multicast addresses"),
   1534 		       ether_multicast_sysctl, 0, NULL, 0,
   1535 		       CTL_CREATE, CTL_EOL);
   1536 }
   1537 
   1538 void
   1539 etherinit(void)
   1540 {
   1541 	mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET);
   1542 }
   1543