Home | History | Annotate | Line # | Download | only in net
if_ethersubr.c revision 1.204.6.1
      1 /*	$NetBSD: if_ethersubr.c,v 1.204.6.1 2017/03/13 07:41:27 skrll 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 1.204.6.1 2017/03/13 07:41:27 skrll Exp $");
     65 
     66 #include "opt_inet.h"
     67 #include "opt_atalk.h"
     68 #include "opt_ipx.h"
     69 #include "opt_mbuftrace.h"
     70 #include "opt_mpls.h"
     71 #include "opt_gateway.h"
     72 #include "opt_pppoe.h"
     73 #include "vlan.h"
     74 #include "pppoe.h"
     75 #include "bridge.h"
     76 #include "arp.h"
     77 #include "agr.h"
     78 
     79 #include <sys/param.h>
     80 #include <sys/systm.h>
     81 #include <sys/sysctl.h>
     82 #include <sys/kernel.h>
     83 #include <sys/callout.h>
     84 #include <sys/malloc.h>
     85 #include <sys/mbuf.h>
     86 #include <sys/protosw.h>
     87 #include <sys/socket.h>
     88 #include <sys/ioctl.h>
     89 #include <sys/errno.h>
     90 #include <sys/syslog.h>
     91 #include <sys/kauth.h>
     92 #include <sys/cpu.h>
     93 #include <sys/intr.h>
     94 #include <sys/device.h>
     95 #include <sys/rnd.h>
     96 
     97 #include <net/if.h>
     98 #include <net/netisr.h>
     99 #include <net/route.h>
    100 #include <net/if_llc.h>
    101 #include <net/if_dl.h>
    102 #include <net/if_types.h>
    103 
    104 #include <net/if_media.h>
    105 #include <dev/mii/mii.h>
    106 #include <dev/mii/miivar.h>
    107 
    108 #if NARP == 0
    109 /*
    110  * XXX there should really be a way to issue this warning from within config(8)
    111  */
    112 #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.
    113 #endif
    114 
    115 #include <net/bpf.h>
    116 
    117 #include <net/if_ether.h>
    118 #include <net/if_vlanvar.h>
    119 
    120 #if NPPPOE > 0
    121 #include <net/if_pppoe.h>
    122 #endif
    123 
    124 #if NAGR > 0
    125 #include <net/agr/ieee8023_slowprotocols.h>	/* XXX */
    126 #include <net/agr/ieee8023ad.h>
    127 #include <net/agr/if_agrvar.h>
    128 #endif
    129 
    130 #if NBRIDGE > 0
    131 #include <net/if_bridgevar.h>
    132 #endif
    133 
    134 #include <netinet/in.h>
    135 #ifdef INET
    136 #include <netinet/in_var.h>
    137 #endif
    138 #include <netinet/if_inarp.h>
    139 
    140 #ifdef INET6
    141 #ifndef INET
    142 #include <netinet/in.h>
    143 #endif
    144 #include <netinet6/in6_var.h>
    145 #include <netinet6/nd6.h>
    146 #endif
    147 
    148 
    149 #include "carp.h"
    150 #if NCARP > 0
    151 #include <netinet/ip_carp.h>
    152 #endif
    153 
    154 #ifdef IPX
    155 #include <netipx/ipx.h>
    156 #include <netipx/ipx_if.h>
    157 #endif
    158 
    159 #ifdef NETATALK
    160 #include <netatalk/at.h>
    161 #include <netatalk/at_var.h>
    162 #include <netatalk/at_extern.h>
    163 
    164 #define llc_snap_org_code llc_un.type_snap.org_code
    165 #define llc_snap_ether_type llc_un.type_snap.ether_type
    166 
    167 extern u_char	at_org_code[3];
    168 extern u_char	aarp_org_code[3];
    169 #endif /* NETATALK */
    170 
    171 #ifdef MPLS
    172 #include <netmpls/mpls.h>
    173 #include <netmpls/mpls_var.h>
    174 #endif
    175 
    176 static struct timeval bigpktppslim_last;
    177 static int bigpktppslim = 2;	/* XXX */
    178 static int bigpktpps_count;
    179 static kmutex_t bigpktpps_lock __cacheline_aligned;
    180 
    181 
    182 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] =
    183     { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
    184 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] =
    185     { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 };
    186 #define senderr(e) { error = (e); goto bad;}
    187 
    188 static	int ether_output(struct ifnet *, struct mbuf *,
    189 	    const struct sockaddr *, struct rtentry *);
    190 
    191 /*
    192  * Ethernet output routine.
    193  * Encapsulate a packet of type family for the local net.
    194  * Assumes that ifp is actually pointer to ethercom structure.
    195  */
    196 static int
    197 ether_output(struct ifnet * const ifp0, struct mbuf * const m0,
    198 	const struct sockaddr * const dst,
    199 	struct rtentry *rt0)
    200 {
    201 	uint16_t etype = 0;
    202 	int error = 0, hdrcmplt = 0;
    203  	uint8_t esrc[6], edst[6];
    204 	struct mbuf *m = m0;
    205 	struct rtentry *rt;
    206 	struct mbuf *mcopy = NULL;
    207 	struct ether_header *eh;
    208 	struct ifnet *ifp = ifp0;
    209 	ALTQ_DECL(struct altq_pktattr pktattr;)
    210 #ifdef INET
    211 	struct arphdr *ah;
    212 #endif /* INET */
    213 #ifdef NETATALK
    214 	struct at_ifaddr *aa;
    215 #endif /* NETATALK */
    216 
    217 	KASSERT(KERNEL_LOCKED_P());
    218 
    219 #ifdef MBUFTRACE
    220 	m_claimm(m, ifp->if_mowner);
    221 #endif
    222 
    223 #if NCARP > 0
    224 	if (ifp->if_type == IFT_CARP) {
    225 		struct ifaddr *ifa;
    226 
    227 		/* loop back if this is going to the carp interface */
    228 		if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP &&
    229 		    (ifa = ifa_ifwithaddr(dst)) != NULL &&
    230 		    ifa->ifa_ifp == ifp0)
    231 			return looutput(ifp0, m, dst, rt0);
    232 
    233 		ifp = ifp->if_carpdev;
    234 		/* ac = (struct arpcom *)ifp; */
    235 
    236 		if ((ifp0->if_flags & (IFF_UP|IFF_RUNNING)) !=
    237 		    (IFF_UP|IFF_RUNNING))
    238 			senderr(ENETDOWN);
    239 	}
    240 #endif /* NCARP > 0 */
    241 
    242 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
    243 		senderr(ENETDOWN);
    244 	if ((rt = rt0) != NULL) {
    245 		if ((rt->rt_flags & RTF_UP) == 0) {
    246 			if ((rt0 = rt = rtalloc1(dst, 1)) != NULL) {
    247 				rt->rt_refcnt--;
    248 				if (rt->rt_ifp != ifp)
    249 					return (*rt->rt_ifp->if_output)
    250 							(ifp, m0, dst, rt);
    251 			} else
    252 				senderr(EHOSTUNREACH);
    253 		}
    254 		if ((rt->rt_flags & RTF_GATEWAY) && dst->sa_family != AF_NS) {
    255 			if (rt->rt_gwroute == NULL)
    256 				goto lookup;
    257 			if (((rt = rt->rt_gwroute)->rt_flags & RTF_UP) == 0) {
    258 				rtfree(rt); rt = rt0;
    259 			lookup: rt->rt_gwroute = rtalloc1(rt->rt_gateway, 1);
    260 				if ((rt = rt->rt_gwroute) == NULL)
    261 					senderr(EHOSTUNREACH);
    262 				/* the "G" test below also prevents rt == rt0 */
    263 				if ((rt->rt_flags & RTF_GATEWAY) ||
    264 				    (rt->rt_ifp != ifp)) {
    265 					rt->rt_refcnt--;
    266 					rt0->rt_gwroute = NULL;
    267 					senderr(EHOSTUNREACH);
    268 				}
    269 			}
    270 		}
    271 		if (rt->rt_flags & RTF_REJECT)
    272 			if (rt->rt_rmx.rmx_expire == 0 ||
    273 			    (u_long) time_second < rt->rt_rmx.rmx_expire)
    274 				senderr(rt == rt0 ? EHOSTDOWN : EHOSTUNREACH);
    275 	}
    276 
    277 	switch (dst->sa_family) {
    278 
    279 #ifdef INET
    280 	case AF_INET:
    281 		if (m->m_flags & M_BCAST)
    282 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
    283 		else if (m->m_flags & M_MCAST)
    284 			ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst);
    285 		else if (!arpresolve(ifp, rt, m, dst, edst))
    286 			return (0);	/* if not yet resolved */
    287 		/* If broadcasting on a simplex interface, loopback a copy */
    288 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
    289 			mcopy = m_copy(m, 0, (int)M_COPYALL);
    290 		etype = htons(ETHERTYPE_IP);
    291 		break;
    292 
    293 	case AF_ARP:
    294 		ah = mtod(m, struct arphdr *);
    295 		if (m->m_flags & M_BCAST)
    296 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
    297 		else {
    298 			void *tha = ar_tha(ah);
    299 
    300 			if (tha == NULL) {
    301 				/* fake with ARPHDR_IEEE1394 */
    302 				m_freem(m);
    303 				return 0;
    304 			}
    305 			memcpy(edst, tha, sizeof(edst));
    306 		}
    307 
    308 		ah->ar_hrd = htons(ARPHRD_ETHER);
    309 
    310 		switch (ntohs(ah->ar_op)) {
    311 		case ARPOP_REVREQUEST:
    312 		case ARPOP_REVREPLY:
    313 			etype = htons(ETHERTYPE_REVARP);
    314 			break;
    315 
    316 		case ARPOP_REQUEST:
    317 		case ARPOP_REPLY:
    318 		default:
    319 			etype = htons(ETHERTYPE_ARP);
    320 		}
    321 
    322 		break;
    323 #endif
    324 #ifdef INET6
    325 	case AF_INET6:
    326 		if (!nd6_storelladdr(ifp, rt, m, dst, edst, sizeof(edst))){
    327 			/* something bad happened */
    328 			return (0);
    329 		}
    330 		etype = htons(ETHERTYPE_IPV6);
    331 		break;
    332 #endif
    333 #ifdef NETATALK
    334     case AF_APPLETALK:
    335 		if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) {
    336 #ifdef NETATALKDEBUG
    337 			printf("aarpresolv failed\n");
    338 #endif /* NETATALKDEBUG */
    339 			return (0);
    340 		}
    341 		/*
    342 		 * ifaddr is the first thing in at_ifaddr
    343 		 */
    344 		aa = (struct at_ifaddr *) at_ifawithnet(
    345 		    (const struct sockaddr_at *)dst, ifp);
    346 		if (aa == NULL)
    347 		    goto bad;
    348 
    349 		/*
    350 		 * In the phase 2 case, we need to prepend an mbuf for the
    351 		 * llc header.  Since we must preserve the value of m,
    352 		 * which is passed to us by value, we m_copy() the first
    353 		 * mbuf, and use it for our llc header.
    354 		 */
    355 		if (aa->aa_flags & AFA_PHASE2) {
    356 			struct llc llc;
    357 
    358 			M_PREPEND(m, sizeof(struct llc), M_DONTWAIT);
    359 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
    360 			llc.llc_control = LLC_UI;
    361 			memcpy(llc.llc_snap_org_code, at_org_code,
    362 			    sizeof(llc.llc_snap_org_code));
    363 			llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK);
    364 			memcpy(mtod(m, void *), &llc, sizeof(struct llc));
    365 		} else {
    366 			etype = htons(ETHERTYPE_ATALK);
    367 		}
    368 		break;
    369 #endif /* NETATALK */
    370 #ifdef IPX
    371 	case AF_IPX:
    372 		etype = htons(ETHERTYPE_IPX);
    373  		memcpy(edst,
    374 		    &(((const struct sockaddr_ipx *)dst)->sipx_addr.x_host),
    375 		    sizeof(edst));
    376 		/* If broadcasting on a simplex interface, loopback a copy */
    377 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
    378 			mcopy = m_copy(m, 0, (int)M_COPYALL);
    379 		break;
    380 #endif
    381 	case pseudo_AF_HDRCMPLT:
    382 		hdrcmplt = 1;
    383 		memcpy(esrc,
    384 		    ((const struct ether_header *)dst->sa_data)->ether_shost,
    385 		    sizeof(esrc));
    386 		/* FALLTHROUGH */
    387 
    388 	case AF_UNSPEC:
    389  		memcpy(edst,
    390 		    ((const struct ether_header *)dst->sa_data)->ether_dhost,
    391 		    sizeof(edst));
    392 		/* AF_UNSPEC doesn't swap the byte order of the ether_type. */
    393 		etype = ((const struct ether_header *)dst->sa_data)->ether_type;
    394 		break;
    395 
    396 	default:
    397 		printf("%s: can't handle af%d\n", ifp->if_xname,
    398 			dst->sa_family);
    399 		senderr(EAFNOSUPPORT);
    400 	}
    401 
    402 #ifdef MPLS
    403 	if (rt0 != NULL && rt_gettag(rt0) != NULL &&
    404 	    rt_gettag(rt0)->sa_family == AF_MPLS &&
    405 	    (m->m_flags & (M_MCAST | M_BCAST)) == 0) {
    406 		union mpls_shim msh;
    407 		msh.s_addr = MPLS_GETSADDR(rt0);
    408 		if (msh.shim.label != MPLS_LABEL_IMPLNULL)
    409 			etype = htons(ETHERTYPE_MPLS);
    410 	}
    411 #endif
    412 
    413 	if (mcopy)
    414 		(void)looutput(ifp, mcopy, dst, rt);
    415 
    416 	/* If no ether type is set, this must be a 802.2 formatted packet.
    417 	 */
    418 	if (etype == 0)
    419 		etype = htons(m->m_pkthdr.len);
    420 	/*
    421 	 * Add local net header.  If no space in first mbuf,
    422 	 * allocate another.
    423 	 */
    424 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
    425 	if (m == 0)
    426 		senderr(ENOBUFS);
    427 	eh = mtod(m, struct ether_header *);
    428 	/* Note: etype is already in network byte order. */
    429 	(void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
    430  	memcpy(eh->ether_dhost, edst, sizeof(edst));
    431 	if (hdrcmplt)
    432 		memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost));
    433 	else
    434 	 	memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl),
    435 		    sizeof(eh->ether_shost));
    436 
    437 #if NCARP > 0
    438 	if (ifp0 != ifp && ifp0->if_type == IFT_CARP) {
    439 	 	memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl),
    440 		    sizeof(eh->ether_shost));
    441 	}
    442 #endif /* NCARP > 0 */
    443 
    444 	if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0)
    445 		return (error);
    446 	if (m == NULL)
    447 		return (0);
    448 
    449 #if NBRIDGE > 0
    450 	/*
    451 	 * Bridges require special output handling.
    452 	 */
    453 	if (ifp->if_bridge)
    454 		return (bridge_output(ifp, m, NULL, NULL));
    455 #endif
    456 
    457 #if NCARP > 0
    458 	if (ifp != ifp0)
    459 		ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN;
    460 #endif /* NCARP > 0 */
    461 
    462 #ifdef ALTQ
    463 	/*
    464 	 * If ALTQ is enabled on the parent interface, do
    465 	 * classification; the queueing discipline might not
    466 	 * require classification, but might require the
    467 	 * address family/header pointer in the pktattr.
    468 	 */
    469 	if (ALTQ_IS_ENABLED(&ifp->if_snd))
    470 		altq_etherclassify(&ifp->if_snd, m, &pktattr);
    471 #endif
    472 	return ifq_enqueue(ifp, m ALTQ_COMMA ALTQ_DECL(&pktattr));
    473 
    474 bad:
    475 	if (m)
    476 		m_freem(m);
    477 	return (error);
    478 }
    479 
    480 #ifdef ALTQ
    481 /*
    482  * This routine is a slight hack to allow a packet to be classified
    483  * if the Ethernet headers are present.  It will go away when ALTQ's
    484  * classification engine understands link headers.
    485  */
    486 void
    487 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
    488     struct altq_pktattr *pktattr)
    489 {
    490 	struct ether_header *eh;
    491 	uint16_t ether_type;
    492 	int hlen, af, hdrsize;
    493 	void *hdr;
    494 
    495 	hlen = ETHER_HDR_LEN;
    496 	eh = mtod(m, struct ether_header *);
    497 
    498 	ether_type = htons(eh->ether_type);
    499 
    500 	if (ether_type < ETHERMTU) {
    501 		/* LLC/SNAP */
    502 		struct llc *llc = (struct llc *)(eh + 1);
    503 		hlen += 8;
    504 
    505 		if (m->m_len < hlen ||
    506 		    llc->llc_dsap != LLC_SNAP_LSAP ||
    507 		    llc->llc_ssap != LLC_SNAP_LSAP ||
    508 		    llc->llc_control != LLC_UI) {
    509 			/* Not SNAP. */
    510 			goto bad;
    511 		}
    512 
    513 		ether_type = htons(llc->llc_un.type_snap.ether_type);
    514 	}
    515 
    516 	switch (ether_type) {
    517 	case ETHERTYPE_IP:
    518 		af = AF_INET;
    519 		hdrsize = 20;		/* sizeof(struct ip) */
    520 		break;
    521 
    522 	case ETHERTYPE_IPV6:
    523 		af = AF_INET6;
    524 		hdrsize = 40;		/* sizeof(struct ip6_hdr) */
    525 		break;
    526 
    527 	default:
    528 		af = AF_UNSPEC;
    529 		hdrsize = 0;
    530 		break;
    531 	}
    532 
    533 	while (m->m_len <= hlen) {
    534 		hlen -= m->m_len;
    535 		m = m->m_next;
    536 	}
    537 	if (m->m_len < (hlen + hdrsize)) {
    538 		/*
    539 		 * protocol header not in a single mbuf.
    540 		 * We can't cope with this situation right
    541 		 * now (but it shouldn't ever happen, really, anyhow).
    542 		 */
    543 #ifdef DEBUG
    544 		printf("altq_etherclassify: headers span multiple mbufs: "
    545 		    "%d < %d\n", m->m_len, (hlen + hdrsize));
    546 #endif
    547 		goto bad;
    548 	}
    549 
    550 	m->m_data += hlen;
    551 	m->m_len -= hlen;
    552 
    553 	hdr = mtod(m, void *);
    554 
    555 	if (ALTQ_NEEDS_CLASSIFY(ifq))
    556 		pktattr->pattr_class =
    557 		    (*ifq->altq_classify)(ifq->altq_clfier, m, af);
    558 	pktattr->pattr_af = af;
    559 	pktattr->pattr_hdr = hdr;
    560 
    561 	m->m_data -= hlen;
    562 	m->m_len += hlen;
    563 
    564 	return;
    565 
    566  bad:
    567 	pktattr->pattr_class = NULL;
    568 	pktattr->pattr_hdr = NULL;
    569 	pktattr->pattr_af = AF_UNSPEC;
    570 }
    571 #endif /* ALTQ */
    572 
    573 /*
    574  * Process a received Ethernet packet;
    575  * the packet is in the mbuf chain m with
    576  * the ether header.
    577  */
    578 void
    579 ether_input(struct ifnet *ifp, struct mbuf *m)
    580 {
    581 	struct ethercom *ec = (struct ethercom *) ifp;
    582 	pktqueue_t *pktq = NULL;
    583 	struct ifqueue *inq = NULL;
    584 	uint16_t etype;
    585 	struct ether_header *eh;
    586 	size_t ehlen;
    587 	static int earlypkts;
    588 	int isr = 0;
    589 #if defined (LLC) || defined(NETATALK)
    590 	struct llc *l;
    591 #endif
    592 
    593 	if ((ifp->if_flags & IFF_UP) == 0) {
    594 		m_freem(m);
    595 		return;
    596 	}
    597 
    598 #ifdef MBUFTRACE
    599 	m_claimm(m, &ec->ec_rx_mowner);
    600 #endif
    601 	eh = mtod(m, struct ether_header *);
    602 	etype = ntohs(eh->ether_type);
    603 	ehlen = sizeof(*eh);
    604 
    605 	if(__predict_false(earlypkts < 100 || !rnd_initial_entropy)) {
    606 		rnd_add_data(NULL, eh, ehlen, 0);
    607 		earlypkts++;
    608 	}
    609 
    610 	/*
    611 	 * Determine if the packet is within its size limits.
    612 	 */
    613 	if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len >
    614 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
    615 		mutex_enter(&bigpktpps_lock);
    616 		if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count,
    617 			    bigpktppslim)) {
    618 			printf("%s: discarding oversize frame (len=%d)\n",
    619 			    ifp->if_xname, m->m_pkthdr.len);
    620 		}
    621 		mutex_exit(&bigpktpps_lock);
    622 		m_freem(m);
    623 		return;
    624 	}
    625 
    626 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
    627 		/*
    628 		 * If this is not a simplex interface, drop the packet
    629 		 * if it came from us.
    630 		 */
    631 		if ((ifp->if_flags & IFF_SIMPLEX) == 0 &&
    632 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost,
    633 		    ETHER_ADDR_LEN) == 0) {
    634 			m_freem(m);
    635 			return;
    636 		}
    637 
    638 		if (memcmp(etherbroadcastaddr,
    639 		    eh->ether_dhost, ETHER_ADDR_LEN) == 0)
    640 			m->m_flags |= M_BCAST;
    641 		else
    642 			m->m_flags |= M_MCAST;
    643 		ifp->if_imcasts++;
    644 	}
    645 
    646 	/* If the CRC is still on the packet, trim it off. */
    647 	if (m->m_flags & M_HASFCS) {
    648 		m_adj(m, -ETHER_CRC_LEN);
    649 		m->m_flags &= ~M_HASFCS;
    650 	}
    651 
    652 	ifp->if_ibytes += m->m_pkthdr.len;
    653 
    654 #if NCARP > 0
    655 	if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) {
    656 		/*
    657 		 * clear M_PROMISC, in case the packets comes from a
    658 		 * vlan
    659 		 */
    660 		m->m_flags &= ~M_PROMISC;
    661 		if (carp_input(m, (uint8_t *)&eh->ether_shost,
    662 		    (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0)
    663 			return;
    664 	}
    665 #endif /* NCARP > 0 */
    666 	if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 &&
    667 	    (ifp->if_flags & IFF_PROMISC) != 0 &&
    668 	    memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost,
    669 		   ETHER_ADDR_LEN) != 0) {
    670 		m->m_flags |= M_PROMISC;
    671 	}
    672 
    673 	if ((m->m_flags & M_PROMISC) == 0) {
    674 		if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0)
    675 			return;
    676 		if (m == NULL)
    677 			return;
    678 
    679 		eh = mtod(m, struct ether_header *);
    680 		etype = ntohs(eh->ether_type);
    681 		ehlen = sizeof(*eh);
    682 	}
    683 
    684 #if NAGR > 0
    685 	if (ifp->if_agrprivate &&
    686 	    __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) {
    687 		m->m_flags &= ~M_PROMISC;
    688 		agr_input(ifp, m);
    689 		return;
    690 	}
    691 #endif /* NAGR > 0 */
    692 
    693 	/*
    694 	 * If VLANs are configured on the interface, check to
    695 	 * see if the device performed the decapsulation and
    696 	 * provided us with the tag.
    697 	 */
    698 	if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) {
    699 #if NVLAN > 0
    700 		/*
    701 		 * vlan_input() will either recursively call ether_input()
    702 		 * or drop the packet.
    703 		 */
    704 		vlan_input(ifp, m);
    705 #else
    706 		m_freem(m);
    707 #endif
    708 		return;
    709 	}
    710 
    711 	/*
    712 	 * Handle protocols that expect to have the Ethernet header
    713 	 * (and possibly FCS) intact.
    714 	 */
    715 	switch (etype) {
    716 	case ETHERTYPE_VLAN: {
    717 		struct ether_vlan_header *evl = (void *)eh;
    718 		/*
    719 		 * If there is a tag of 0, then the VLAN header was probably
    720 		 * just being used to store the priority.  Extract the ether
    721 		 * type, and if IP or IPV6, let them deal with it.
    722 		 */
    723 		if (m->m_len <= sizeof(*evl)
    724 		    && EVL_VLANOFTAG(evl->evl_tag) == 0) {
    725 			etype = ntohs(evl->evl_proto);
    726 			ehlen = sizeof(*evl);
    727 			if ((m->m_flags & M_PROMISC) == 0
    728 			    && (etype == ETHERTYPE_IP
    729 				|| etype == ETHERTYPE_IPV6))
    730 				break;
    731 		}
    732 #if NVLAN > 0
    733 		/*
    734 		 * vlan_input() will either recursively call ether_input()
    735 		 * or drop the packet.
    736 		 */
    737 		if (((struct ethercom *)ifp)->ec_nvlans != 0)
    738 			vlan_input(ifp, m);
    739 		else
    740 #endif /* NVLAN > 0 */
    741 			m_freem(m);
    742 		return;
    743 	}
    744 #if NPPPOE > 0
    745 	case ETHERTYPE_PPPOEDISC:
    746 	case ETHERTYPE_PPPOE:
    747 		if (m->m_flags & M_PROMISC) {
    748 			m_freem(m);
    749 			return;
    750 		}
    751 #ifndef PPPOE_SERVER
    752 		if (m->m_flags & (M_MCAST | M_BCAST)) {
    753 			m_freem(m);
    754 			return;
    755 		}
    756 #endif
    757 
    758 		if (etype == ETHERTYPE_PPPOEDISC)
    759 			inq = &ppoediscinq;
    760 		else
    761 			inq = &ppoeinq;
    762 		if (IF_QFULL(inq)) {
    763 			IF_DROP(inq);
    764 			m_freem(m);
    765 		} else {
    766 			IF_ENQUEUE(inq, m);
    767 			softint_schedule(pppoe_softintr);
    768 		}
    769 		return;
    770 #endif /* NPPPOE > 0 */
    771 	case ETHERTYPE_SLOWPROTOCOLS: {
    772 		uint8_t subtype;
    773 
    774 #if defined(DIAGNOSTIC)
    775 		if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) {
    776 			panic("ether_input: too short slow protocol packet");
    777 		}
    778 #endif
    779 		m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype);
    780 		switch (subtype) {
    781 #if NAGR > 0
    782 		case SLOWPROTOCOLS_SUBTYPE_LACP:
    783 			if (ifp->if_agrprivate) {
    784 				ieee8023ad_lacp_input(ifp, m);
    785 				return;
    786 			}
    787 			break;
    788 
    789 		case SLOWPROTOCOLS_SUBTYPE_MARKER:
    790 			if (ifp->if_agrprivate) {
    791 				ieee8023ad_marker_input(ifp, m);
    792 				return;
    793 			}
    794 			break;
    795 #endif /* NAGR > 0 */
    796 		default:
    797 			if (subtype == 0 || subtype > 10) {
    798 				/* illegal value */
    799 				m_freem(m);
    800 				return;
    801 			}
    802 			/* unknown subtype */
    803 			break;
    804 		}
    805 		/* FALLTHROUGH */
    806 	}
    807 	default:
    808 		if (m->m_flags & M_PROMISC) {
    809 			m_freem(m);
    810 			return;
    811 		}
    812 	}
    813 
    814 	/* If the CRC is still on the packet, trim it off. */
    815 	if (m->m_flags & M_HASFCS) {
    816 		m_adj(m, -ETHER_CRC_LEN);
    817 		m->m_flags &= ~M_HASFCS;
    818 	}
    819 
    820 	if (etype > ETHERMTU + sizeof (struct ether_header)) {
    821 		/* Strip off the Ethernet header. */
    822 		m_adj(m, ehlen);
    823 
    824 		switch (etype) {
    825 #ifdef INET
    826 		case ETHERTYPE_IP:
    827 #ifdef GATEWAY
    828 			if (ipflow_fastforward(m))
    829 				return;
    830 #endif
    831 			pktq = ip_pktq;
    832 			break;
    833 
    834 		case ETHERTYPE_ARP:
    835 			isr = NETISR_ARP;
    836 			inq = &arpintrq;
    837 			break;
    838 
    839 		case ETHERTYPE_REVARP:
    840 			revarpinput(m);	/* XXX queue? */
    841 			return;
    842 #endif
    843 #ifdef INET6
    844 		case ETHERTYPE_IPV6:
    845 			if (__predict_false(!in6_present)) {
    846 				m_freem(m);
    847 				return;
    848 			}
    849 #ifdef GATEWAY
    850 			if (ip6flow_fastforward(&m))
    851 				return;
    852 #endif
    853 			pktq = ip6_pktq;
    854 			break;
    855 #endif
    856 #ifdef IPX
    857 		case ETHERTYPE_IPX:
    858 			isr = NETISR_IPX;
    859 			inq = &ipxintrq;
    860 			break;
    861 #endif
    862 #ifdef NETATALK
    863 		case ETHERTYPE_ATALK:
    864 			isr = NETISR_ATALK;
    865 			inq = &atintrq1;
    866 			break;
    867 		case ETHERTYPE_AARP:
    868 			/* probably this should be done with a NETISR as well */
    869 			aarpinput(ifp, m); /* XXX */
    870 			return;
    871 #endif /* NETATALK */
    872 #ifdef MPLS
    873 		case ETHERTYPE_MPLS:
    874 			isr = NETISR_MPLS;
    875 			inq = &mplsintrq;
    876 			break;
    877 #endif
    878 		default:
    879 			m_freem(m);
    880 			return;
    881 		}
    882 	} else {
    883 #if defined (LLC) || defined (NETATALK)
    884 		l = (struct llc *)(eh+1);
    885 		switch (l->llc_dsap) {
    886 #ifdef NETATALK
    887 		case LLC_SNAP_LSAP:
    888 			switch (l->llc_control) {
    889 			case LLC_UI:
    890 				if (l->llc_ssap != LLC_SNAP_LSAP) {
    891 					goto dropanyway;
    892 				}
    893 
    894 				if (memcmp(&(l->llc_snap_org_code)[0],
    895 				    at_org_code, sizeof(at_org_code)) == 0 &&
    896 				    ntohs(l->llc_snap_ether_type) ==
    897 				    ETHERTYPE_ATALK) {
    898 					inq = &atintrq2;
    899 					m_adj(m, sizeof(struct ether_header)
    900 					    + sizeof(struct llc));
    901 					isr = NETISR_ATALK;
    902 					break;
    903 				}
    904 
    905 				if (memcmp(&(l->llc_snap_org_code)[0],
    906 				    aarp_org_code,
    907 				    sizeof(aarp_org_code)) == 0 &&
    908 				    ntohs(l->llc_snap_ether_type) ==
    909 				    ETHERTYPE_AARP) {
    910 					m_adj( m, sizeof(struct ether_header)
    911 					    + sizeof(struct llc));
    912 					aarpinput(ifp, m); /* XXX */
    913 				    return;
    914 				}
    915 
    916 			default:
    917 				goto dropanyway;
    918 			}
    919 			break;
    920 		dropanyway:
    921 #endif
    922 		default:
    923 			m_freem(m);
    924 			return;
    925 		}
    926 #else /* ISO || LLC || NETATALK*/
    927 		m_freem(m);
    928 		return;
    929 #endif /* ISO || LLC || NETATALK*/
    930 	}
    931 
    932 	if (__predict_true(pktq)) {
    933 		const uint32_t h = pktq_rps_hash(m);
    934 		if (__predict_false(!pktq_enqueue(pktq, m, h))) {
    935 			m_freem(m);
    936 		}
    937 		return;
    938 	}
    939 
    940 	if (__predict_false(!inq)) {
    941 		/* Should not happen. */
    942 		m_freem(m);
    943 		return;
    944 	}
    945 	if (IF_QFULL(inq)) {
    946 		IF_DROP(inq);
    947 		m_freem(m);
    948 	} else {
    949 		IF_ENQUEUE(inq, m);
    950 		schednetisr(isr);
    951 	}
    952 }
    953 
    954 /*
    955  * Convert Ethernet address to printable (loggable) representation.
    956  */
    957 char *
    958 ether_sprintf(const u_char *ap)
    959 {
    960 	static char etherbuf[3 * ETHER_ADDR_LEN];
    961 	return ether_snprintf(etherbuf, sizeof(etherbuf), ap);
    962 }
    963 
    964 char *
    965 ether_snprintf(char *buf, size_t len, const u_char *ap)
    966 {
    967 	char *cp = buf;
    968 	size_t i;
    969 
    970 	for (i = 0; i < len / 3; i++) {
    971 		*cp++ = hexdigits[*ap >> 4];
    972 		*cp++ = hexdigits[*ap++ & 0xf];
    973 		*cp++ = ':';
    974 	}
    975 	*--cp = '\0';
    976 	return buf;
    977 }
    978 
    979 /*
    980  * Perform common duties while attaching to interface list
    981  */
    982 void
    983 ether_ifattach(struct ifnet *ifp, const uint8_t *lla)
    984 {
    985 	struct ethercom *ec = (struct ethercom *)ifp;
    986 
    987 	ifp->if_type = IFT_ETHER;
    988 	ifp->if_hdrlen = ETHER_HDR_LEN;
    989 	ifp->if_dlt = DLT_EN10MB;
    990 	ifp->if_mtu = ETHERMTU;
    991 	ifp->if_output = ether_output;
    992 	ifp->if_input = ether_input;
    993 	if (ifp->if_baudrate == 0)
    994 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
    995 
    996 	if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla));
    997 
    998 	LIST_INIT(&ec->ec_multiaddrs);
    999 	ifp->if_broadcastaddr = etherbroadcastaddr;
   1000 	bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header));
   1001 #ifdef MBUFTRACE
   1002 	strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname,
   1003 	    sizeof(ec->ec_tx_mowner.mo_name));
   1004 	strlcpy(ec->ec_tx_mowner.mo_descr, "tx",
   1005 	    sizeof(ec->ec_tx_mowner.mo_descr));
   1006 	strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname,
   1007 	    sizeof(ec->ec_rx_mowner.mo_name));
   1008 	strlcpy(ec->ec_rx_mowner.mo_descr, "rx",
   1009 	    sizeof(ec->ec_rx_mowner.mo_descr));
   1010 	MOWNER_ATTACH(&ec->ec_tx_mowner);
   1011 	MOWNER_ATTACH(&ec->ec_rx_mowner);
   1012 	ifp->if_mowner = &ec->ec_tx_mowner;
   1013 #endif
   1014 }
   1015 
   1016 void
   1017 ether_ifdetach(struct ifnet *ifp)
   1018 {
   1019 	struct ethercom *ec = (void *) ifp;
   1020 	struct ether_multi *enm;
   1021 	int s;
   1022 
   1023 	/*
   1024 	 * Prevent further calls to ioctl (for example turning off
   1025 	 * promiscuous mode from the bridge code), which eventually can
   1026 	 * call if_init() which can cause panics because the interface
   1027 	 * is in the process of being detached. Return device not configured
   1028 	 * instead.
   1029 	 */
   1030 	ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio;
   1031 
   1032 #if NBRIDGE > 0
   1033 	if (ifp->if_bridge)
   1034 		bridge_ifdetach(ifp);
   1035 #endif
   1036 
   1037 	bpf_detach(ifp);
   1038 
   1039 #if NVLAN > 0
   1040 	if (ec->ec_nvlans)
   1041 		vlan_ifdetach(ifp);
   1042 #endif
   1043 
   1044 	s = splnet();
   1045 	while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) {
   1046 		LIST_REMOVE(enm, enm_list);
   1047 		free(enm, M_IFMADDR);
   1048 		ec->ec_multicnt--;
   1049 	}
   1050 	splx(s);
   1051 
   1052 #if 0	/* done in if_detach() */
   1053 	if_free_sadl(ifp);
   1054 #endif
   1055 
   1056 	ifp->if_mowner = NULL;
   1057 	MOWNER_DETACH(&ec->ec_rx_mowner);
   1058 	MOWNER_DETACH(&ec->ec_tx_mowner);
   1059 }
   1060 
   1061 #if 0
   1062 /*
   1063  * This is for reference.  We have a table-driven version
   1064  * of the little-endian crc32 generator, which is faster
   1065  * than the double-loop.
   1066  */
   1067 uint32_t
   1068 ether_crc32_le(const uint8_t *buf, size_t len)
   1069 {
   1070 	uint32_t c, crc, carry;
   1071 	size_t i, j;
   1072 
   1073 	crc = 0xffffffffU;	/* initial value */
   1074 
   1075 	for (i = 0; i < len; i++) {
   1076 		c = buf[i];
   1077 		for (j = 0; j < 8; j++) {
   1078 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
   1079 			crc >>= 1;
   1080 			c >>= 1;
   1081 			if (carry)
   1082 				crc = (crc ^ ETHER_CRC_POLY_LE);
   1083 		}
   1084 	}
   1085 
   1086 	return (crc);
   1087 }
   1088 #else
   1089 uint32_t
   1090 ether_crc32_le(const uint8_t *buf, size_t len)
   1091 {
   1092 	static const uint32_t crctab[] = {
   1093 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
   1094 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
   1095 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
   1096 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
   1097 	};
   1098 	uint32_t crc;
   1099 	size_t i;
   1100 
   1101 	crc = 0xffffffffU;	/* initial value */
   1102 
   1103 	for (i = 0; i < len; i++) {
   1104 		crc ^= buf[i];
   1105 		crc = (crc >> 4) ^ crctab[crc & 0xf];
   1106 		crc = (crc >> 4) ^ crctab[crc & 0xf];
   1107 	}
   1108 
   1109 	return (crc);
   1110 }
   1111 #endif
   1112 
   1113 uint32_t
   1114 ether_crc32_be(const uint8_t *buf, size_t len)
   1115 {
   1116 	uint32_t c, crc, carry;
   1117 	size_t i, j;
   1118 
   1119 	crc = 0xffffffffU;	/* initial value */
   1120 
   1121 	for (i = 0; i < len; i++) {
   1122 		c = buf[i];
   1123 		for (j = 0; j < 8; j++) {
   1124 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
   1125 			crc <<= 1;
   1126 			c >>= 1;
   1127 			if (carry)
   1128 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
   1129 		}
   1130 	}
   1131 
   1132 	return (crc);
   1133 }
   1134 
   1135 #ifdef INET
   1136 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] =
   1137     { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 };
   1138 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] =
   1139     { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff };
   1140 #endif
   1141 #ifdef INET6
   1142 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] =
   1143     { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 };
   1144 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] =
   1145     { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff };
   1146 #endif
   1147 
   1148 /*
   1149  * ether_aton implementation, not using a static buffer.
   1150  */
   1151 int
   1152 ether_aton_r(u_char *dest, size_t len, const char *str)
   1153 {
   1154         const u_char *cp = (const void *)str;
   1155 	u_char *ep;
   1156 
   1157 #define atox(c)	(((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10))
   1158 
   1159 	if (len < ETHER_ADDR_LEN)
   1160 		return ENOSPC;
   1161 
   1162 	ep = dest + ETHER_ADDR_LEN;
   1163 
   1164 	while (*cp) {
   1165                 if (!isxdigit(*cp))
   1166                         return EINVAL;
   1167 		*dest = atox(*cp);
   1168 		cp++;
   1169                 if (isxdigit(*cp)) {
   1170                         *dest = (*dest << 4) | atox(*cp);
   1171 			dest++;
   1172 			cp++;
   1173                 } else
   1174 			dest++;
   1175 		if (dest == ep)
   1176 			return *cp == '\0' ? 0 : ENAMETOOLONG;
   1177 		switch (*cp) {
   1178 		case ':':
   1179 		case '-':
   1180 		case '.':
   1181 			cp++;
   1182 			break;
   1183 		}
   1184         }
   1185 	return ENOBUFS;
   1186 }
   1187 
   1188 /*
   1189  * Convert a sockaddr into an Ethernet address or range of Ethernet
   1190  * addresses.
   1191  */
   1192 int
   1193 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN],
   1194     uint8_t addrhi[ETHER_ADDR_LEN])
   1195 {
   1196 #ifdef INET
   1197 	const struct sockaddr_in *sin;
   1198 #endif /* INET */
   1199 #ifdef INET6
   1200 	const struct sockaddr_in6 *sin6;
   1201 #endif /* INET6 */
   1202 
   1203 	switch (sa->sa_family) {
   1204 
   1205 	case AF_UNSPEC:
   1206 		memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN);
   1207 		memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
   1208 		break;
   1209 
   1210 #ifdef INET
   1211 	case AF_INET:
   1212 		sin = satocsin(sa);
   1213 		if (sin->sin_addr.s_addr == INADDR_ANY) {
   1214 			/*
   1215 			 * An IP address of INADDR_ANY means listen to
   1216 			 * or stop listening to all of the Ethernet
   1217 			 * multicast addresses used for IP.
   1218 			 * (This is for the sake of IP multicast routers.)
   1219 			 */
   1220 			memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN);
   1221 			memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN);
   1222 		}
   1223 		else {
   1224 			ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo);
   1225 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
   1226 		}
   1227 		break;
   1228 #endif
   1229 #ifdef INET6
   1230 	case AF_INET6:
   1231 		sin6 = satocsin6(sa);
   1232 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
   1233 			/*
   1234 			 * An IP6 address of 0 means listen to or stop
   1235 			 * listening to all of the Ethernet multicast
   1236 			 * address used for IP6.
   1237 			 * (This is used for multicast routers.)
   1238 			 */
   1239 			memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN);
   1240 			memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN);
   1241 		} else {
   1242 			ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo);
   1243 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
   1244 		}
   1245 		break;
   1246 #endif
   1247 
   1248 	default:
   1249 		return EAFNOSUPPORT;
   1250 	}
   1251 	return 0;
   1252 }
   1253 
   1254 /*
   1255  * Add an Ethernet multicast address or range of addresses to the list for a
   1256  * given interface.
   1257  */
   1258 int
   1259 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec)
   1260 {
   1261 	struct ether_multi *enm;
   1262 	u_char addrlo[ETHER_ADDR_LEN];
   1263 	u_char addrhi[ETHER_ADDR_LEN];
   1264 	int s = splnet(), error;
   1265 
   1266 	error = ether_multiaddr(sa, addrlo, addrhi);
   1267 	if (error != 0) {
   1268 		splx(s);
   1269 		return error;
   1270 	}
   1271 
   1272 	/*
   1273 	 * Verify that we have valid Ethernet multicast addresses.
   1274 	 */
   1275 	if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) {
   1276 		splx(s);
   1277 		return EINVAL;
   1278 	}
   1279 	/*
   1280 	 * See if the address range is already in the list.
   1281 	 */
   1282 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
   1283 	if (enm != NULL) {
   1284 		/*
   1285 		 * Found it; just increment the reference count.
   1286 		 */
   1287 		++enm->enm_refcount;
   1288 		splx(s);
   1289 		return 0;
   1290 	}
   1291 	/*
   1292 	 * New address or range; malloc a new multicast record
   1293 	 * and link it into the interface's multicast list.
   1294 	 */
   1295 	enm = (struct ether_multi *)malloc(sizeof(*enm), M_IFMADDR, M_NOWAIT);
   1296 	if (enm == NULL) {
   1297 		splx(s);
   1298 		return ENOBUFS;
   1299 	}
   1300 	memcpy(enm->enm_addrlo, addrlo, 6);
   1301 	memcpy(enm->enm_addrhi, addrhi, 6);
   1302 	enm->enm_refcount = 1;
   1303 	LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list);
   1304 	ec->ec_multicnt++;
   1305 	splx(s);
   1306 	/*
   1307 	 * Return ENETRESET to inform the driver that the list has changed
   1308 	 * and its reception filter should be adjusted accordingly.
   1309 	 */
   1310 	return ENETRESET;
   1311 }
   1312 
   1313 /*
   1314  * Delete a multicast address record.
   1315  */
   1316 int
   1317 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec)
   1318 {
   1319 	struct ether_multi *enm;
   1320 	u_char addrlo[ETHER_ADDR_LEN];
   1321 	u_char addrhi[ETHER_ADDR_LEN];
   1322 	int s = splnet(), error;
   1323 
   1324 	error = ether_multiaddr(sa, addrlo, addrhi);
   1325 	if (error != 0) {
   1326 		splx(s);
   1327 		return (error);
   1328 	}
   1329 
   1330 	/*
   1331 	 * Look ur the address in our list.
   1332 	 */
   1333 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
   1334 	if (enm == NULL) {
   1335 		splx(s);
   1336 		return (ENXIO);
   1337 	}
   1338 	if (--enm->enm_refcount != 0) {
   1339 		/*
   1340 		 * Still some claims to this record.
   1341 		 */
   1342 		splx(s);
   1343 		return (0);
   1344 	}
   1345 	/*
   1346 	 * No remaining claims to this record; unlink and free it.
   1347 	 */
   1348 	LIST_REMOVE(enm, enm_list);
   1349 	free(enm, M_IFMADDR);
   1350 	ec->ec_multicnt--;
   1351 	splx(s);
   1352 	/*
   1353 	 * Return ENETRESET to inform the driver that the list has changed
   1354 	 * and its reception filter should be adjusted accordingly.
   1355 	 */
   1356 	return (ENETRESET);
   1357 }
   1358 
   1359 void
   1360 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb)
   1361 {
   1362 	ec->ec_ifflags_cb = cb;
   1363 }
   1364 
   1365 /*
   1366  * Common ioctls for Ethernet interfaces.  Note, we must be
   1367  * called at splnet().
   1368  */
   1369 int
   1370 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data)
   1371 {
   1372 	struct ethercom *ec = (void *) ifp;
   1373 	struct eccapreq *eccr;
   1374 	struct ifreq *ifr = (struct ifreq *)data;
   1375 	struct if_laddrreq *iflr = data;
   1376 	const struct sockaddr_dl *sdl;
   1377 	static const uint8_t zero[ETHER_ADDR_LEN];
   1378 	int error;
   1379 
   1380 	switch (cmd) {
   1381 	case SIOCINITIFADDR:
   1382 	    {
   1383 		struct ifaddr *ifa = (struct ifaddr *)data;
   1384 		if (ifa->ifa_addr->sa_family != AF_LINK
   1385 		    && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) !=
   1386 		       (IFF_UP|IFF_RUNNING)) {
   1387 			ifp->if_flags |= IFF_UP;
   1388 			if ((error = (*ifp->if_init)(ifp)) != 0)
   1389 				return error;
   1390 		}
   1391 #ifdef INET
   1392 		if (ifa->ifa_addr->sa_family == AF_INET)
   1393 			arp_ifinit(ifp, ifa);
   1394 #endif /* INET */
   1395 		return 0;
   1396 	    }
   1397 
   1398 	case SIOCSIFMTU:
   1399 	    {
   1400 		int maxmtu;
   1401 
   1402 		if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU)
   1403 			maxmtu = ETHERMTU_JUMBO;
   1404 		else
   1405 			maxmtu = ETHERMTU;
   1406 
   1407 		if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu)
   1408 			return EINVAL;
   1409 		else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET)
   1410 			return error;
   1411 		else if (ifp->if_flags & IFF_UP) {
   1412 			/* Make sure the device notices the MTU change. */
   1413 			return (*ifp->if_init)(ifp);
   1414 		} else
   1415 			return 0;
   1416 	    }
   1417 
   1418 	case SIOCSIFFLAGS:
   1419 		if ((error = ifioctl_common(ifp, cmd, data)) != 0)
   1420 			return error;
   1421 		switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) {
   1422 		case IFF_RUNNING:
   1423 			/*
   1424 			 * If interface is marked down and it is running,
   1425 			 * then stop and disable it.
   1426 			 */
   1427 			(*ifp->if_stop)(ifp, 1);
   1428 			break;
   1429 		case IFF_UP:
   1430 			/*
   1431 			 * If interface is marked up and it is stopped, then
   1432 			 * start it.
   1433 			 */
   1434 			return (*ifp->if_init)(ifp);
   1435 		case IFF_UP|IFF_RUNNING:
   1436 			error = 0;
   1437 			if (ec->ec_ifflags_cb == NULL ||
   1438 			    (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) {
   1439 				/*
   1440 				 * Reset the interface to pick up
   1441 				 * changes in any other flags that
   1442 				 * affect the hardware state.
   1443 				 */
   1444 				return (*ifp->if_init)(ifp);
   1445 			} else
   1446 				return error;
   1447 		case 0:
   1448 			break;
   1449 		}
   1450 		return 0;
   1451 	case SIOCGETHERCAP:
   1452 		eccr = (struct eccapreq *)data;
   1453 		eccr->eccr_capabilities = ec->ec_capabilities;
   1454 		eccr->eccr_capenable = ec->ec_capenable;
   1455 		return 0;
   1456 	case SIOCADDMULTI:
   1457 		return ether_addmulti(ifreq_getaddr(cmd, ifr), ec);
   1458 	case SIOCDELMULTI:
   1459 		return ether_delmulti(ifreq_getaddr(cmd, ifr), ec);
   1460 	case SIOCSIFMEDIA:
   1461 	case SIOCGIFMEDIA:
   1462 		if (ec->ec_mii == NULL)
   1463 			return ENOTTY;
   1464 		return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd);
   1465 	case SIOCALIFADDR:
   1466 		sdl = satocsdl(sstocsa(&iflr->addr));
   1467 		if (sdl->sdl_family != AF_LINK)
   1468 			;
   1469 		else if (ETHER_IS_MULTICAST(CLLADDR(sdl)))
   1470 			return EINVAL;
   1471 		else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0)
   1472 			return EINVAL;
   1473 		/*FALLTHROUGH*/
   1474 	default:
   1475 		return ifioctl_common(ifp, cmd, data);
   1476 	}
   1477 	return 0;
   1478 }
   1479 
   1480 static int
   1481 ether_multicast_sysctl(SYSCTLFN_ARGS)
   1482 {
   1483 	struct ether_multi *enm;
   1484 	struct ether_multi_sysctl addr;
   1485 	struct ifnet *ifp;
   1486 	struct ethercom *ec;
   1487 	int error;
   1488 	size_t written;
   1489 
   1490 	if (namelen != 1)
   1491 		return EINVAL;
   1492 
   1493 	ifp = if_byindex(name[0]);
   1494 	if (ifp == NULL)
   1495 		return ENODEV;
   1496 	if (ifp->if_type != IFT_ETHER) {
   1497 		*oldlenp = 0;
   1498 		return 0;
   1499 	}
   1500 	ec = (struct ethercom *)ifp;
   1501 
   1502 	if (oldp == NULL) {
   1503 		*oldlenp = ec->ec_multicnt * sizeof(addr);
   1504 		return 0;
   1505 	}
   1506 
   1507 	memset(&addr, 0, sizeof(addr));
   1508 	error = 0;
   1509 	written = 0;
   1510 
   1511 	LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) {
   1512 		if (written + sizeof(addr) > *oldlenp)
   1513 			break;
   1514 		addr.enm_refcount = enm->enm_refcount;
   1515 		memcpy(addr.enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN);
   1516 		memcpy(addr.enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN);
   1517 		error = sysctl_copyout(l, &addr, oldp, sizeof(addr));
   1518 		if (error)
   1519 			break;
   1520 		written += sizeof(addr);
   1521 		oldp = (char *)oldp + sizeof(addr);
   1522 	}
   1523 
   1524 	*oldlenp = written;
   1525 	return error;
   1526 }
   1527 
   1528 SYSCTL_SETUP(sysctl_net_ether_setup, "sysctl net.ether subtree setup")
   1529 {
   1530 	const struct sysctlnode *rnode = NULL;
   1531 
   1532 	sysctl_createv(clog, 0, NULL, &rnode,
   1533 		       CTLFLAG_PERMANENT,
   1534 		       CTLTYPE_NODE, "ether",
   1535 		       SYSCTL_DESCR("Ethernet-specific information"),
   1536 		       NULL, 0, NULL, 0,
   1537 		       CTL_NET, CTL_CREATE, CTL_EOL);
   1538 
   1539 	sysctl_createv(clog, 0, &rnode, NULL,
   1540 		       CTLFLAG_PERMANENT,
   1541 		       CTLTYPE_NODE, "multicast",
   1542 		       SYSCTL_DESCR("multicast addresses"),
   1543 		       ether_multicast_sysctl, 0, NULL, 0,
   1544 		       CTL_CREATE, CTL_EOL);
   1545 }
   1546 
   1547 void
   1548 etherinit(void)
   1549 {
   1550 	mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET);
   1551 }
   1552