Home | History | Annotate | Line # | Download | only in net
if_ethersubr.c revision 1.197
      1 /*	$NetBSD: if_ethersubr.c,v 1.197 2014/05/13 19:36:16 bouyer 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 1.197 2014/05/13 19:36:16 bouyer 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/kernel.h>
     82 #include <sys/callout.h>
     83 #include <sys/malloc.h>
     84 #include <sys/mbuf.h>
     85 #include <sys/protosw.h>
     86 #include <sys/socket.h>
     87 #include <sys/ioctl.h>
     88 #include <sys/errno.h>
     89 #include <sys/syslog.h>
     90 #include <sys/kauth.h>
     91 #include <sys/cpu.h>
     92 #include <sys/intr.h>
     93 #include <sys/device.h>
     94 
     95 #include <net/if.h>
     96 #include <net/netisr.h>
     97 #include <net/route.h>
     98 #include <net/if_llc.h>
     99 #include <net/if_dl.h>
    100 #include <net/if_types.h>
    101 
    102 #include <net/if_media.h>
    103 #include <dev/mii/mii.h>
    104 #include <dev/mii/miivar.h>
    105 
    106 #if NARP == 0
    107 /*
    108  * XXX there should really be a way to issue this warning from within config(8)
    109  */
    110 #error You have included NETATALK or a pseudo-device in your configuration that depends on the presence of ethernet interfaces, but have no such interfaces configured. Check if you really need pseudo-device bridge, pppoe, vlan or options NETATALK.
    111 #endif
    112 
    113 #include <net/bpf.h>
    114 
    115 #include <net/if_ether.h>
    116 #include <net/if_vlanvar.h>
    117 
    118 #if NPPPOE > 0
    119 #include <net/if_pppoe.h>
    120 #endif
    121 
    122 #if NAGR > 0
    123 #include <net/agr/ieee8023_slowprotocols.h>	/* XXX */
    124 #include <net/agr/ieee8023ad.h>
    125 #include <net/agr/if_agrvar.h>
    126 #endif
    127 
    128 #if NBRIDGE > 0
    129 #include <net/if_bridgevar.h>
    130 #endif
    131 
    132 #include <netinet/in.h>
    133 #ifdef INET
    134 #include <netinet/in_var.h>
    135 #endif
    136 #include <netinet/if_inarp.h>
    137 
    138 #ifdef INET6
    139 #ifndef INET
    140 #include <netinet/in.h>
    141 #endif
    142 #include <netinet6/in6_var.h>
    143 #include <netinet6/nd6.h>
    144 #endif
    145 
    146 
    147 #include "carp.h"
    148 #if NCARP > 0
    149 #include <netinet/ip_carp.h>
    150 #endif
    151 
    152 #ifdef IPX
    153 #include <netipx/ipx.h>
    154 #include <netipx/ipx_if.h>
    155 #endif
    156 
    157 #ifdef NETATALK
    158 #include <netatalk/at.h>
    159 #include <netatalk/at_var.h>
    160 #include <netatalk/at_extern.h>
    161 
    162 #define llc_snap_org_code llc_un.type_snap.org_code
    163 #define llc_snap_ether_type llc_un.type_snap.ether_type
    164 
    165 extern u_char	at_org_code[3];
    166 extern u_char	aarp_org_code[3];
    167 #endif /* NETATALK */
    168 
    169 #ifdef MPLS
    170 #include <netmpls/mpls.h>
    171 #include <netmpls/mpls_var.h>
    172 #endif
    173 
    174 static struct timeval bigpktppslim_last;
    175 static int bigpktppslim = 2;	/* XXX */
    176 static int bigpktpps_count;
    177 
    178 
    179 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] =
    180     { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
    181 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] =
    182     { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 };
    183 #define senderr(e) { error = (e); goto bad;}
    184 
    185 static	int ether_output(struct ifnet *, struct mbuf *,
    186 	    const struct sockaddr *, struct rtentry *);
    187 
    188 /*
    189  * Ethernet output routine.
    190  * Encapsulate a packet of type family for the local net.
    191  * Assumes that ifp is actually pointer to ethercom structure.
    192  */
    193 static int
    194 ether_output(struct ifnet * const ifp0, struct mbuf * const m0,
    195 	const struct sockaddr * const dst,
    196 	struct rtentry *rt0)
    197 {
    198 	uint16_t etype = 0;
    199 	int error = 0, hdrcmplt = 0;
    200  	uint8_t esrc[6], edst[6];
    201 	struct mbuf *m = m0;
    202 	struct rtentry *rt;
    203 	struct mbuf *mcopy = NULL;
    204 	struct ether_header *eh;
    205 	struct ifnet *ifp = ifp0;
    206 	ALTQ_DECL(struct altq_pktattr pktattr;)
    207 #ifdef INET
    208 	struct arphdr *ah;
    209 #endif /* INET */
    210 #ifdef NETATALK
    211 	struct at_ifaddr *aa;
    212 #endif /* NETATALK */
    213 
    214 	KASSERT(KERNEL_LOCKED_P());
    215 
    216 #ifdef MBUFTRACE
    217 	m_claimm(m, ifp->if_mowner);
    218 #endif
    219 
    220 #if NCARP > 0
    221 	if (ifp->if_type == IFT_CARP) {
    222 		struct ifaddr *ifa;
    223 
    224 		/* loop back if this is going to the carp interface */
    225 		if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP &&
    226 		    (ifa = ifa_ifwithaddr(dst)) != NULL &&
    227 		    ifa->ifa_ifp == ifp0)
    228 			return looutput(ifp0, m, dst, rt0);
    229 
    230 		ifp = ifp->if_carpdev;
    231 		/* ac = (struct arpcom *)ifp; */
    232 
    233 		if ((ifp0->if_flags & (IFF_UP|IFF_RUNNING)) !=
    234 		    (IFF_UP|IFF_RUNNING))
    235 			senderr(ENETDOWN);
    236 	}
    237 #endif /* NCARP > 0 */
    238 
    239 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
    240 		senderr(ENETDOWN);
    241 	if ((rt = rt0) != NULL) {
    242 		if ((rt->rt_flags & RTF_UP) == 0) {
    243 			if ((rt0 = rt = rtalloc1(dst, 1)) != NULL) {
    244 				rt->rt_refcnt--;
    245 				if (rt->rt_ifp != ifp)
    246 					return (*rt->rt_ifp->if_output)
    247 							(ifp, m0, dst, rt);
    248 			} else
    249 				senderr(EHOSTUNREACH);
    250 		}
    251 		if ((rt->rt_flags & RTF_GATEWAY) && dst->sa_family != AF_NS) {
    252 			if (rt->rt_gwroute == NULL)
    253 				goto lookup;
    254 			if (((rt = rt->rt_gwroute)->rt_flags & RTF_UP) == 0) {
    255 				rtfree(rt); rt = rt0;
    256 			lookup: rt->rt_gwroute = rtalloc1(rt->rt_gateway, 1);
    257 				if ((rt = rt->rt_gwroute) == NULL)
    258 					senderr(EHOSTUNREACH);
    259 				/* the "G" test below also prevents rt == rt0 */
    260 				if ((rt->rt_flags & RTF_GATEWAY) ||
    261 				    (rt->rt_ifp != ifp)) {
    262 					rt->rt_refcnt--;
    263 					rt0->rt_gwroute = NULL;
    264 					senderr(EHOSTUNREACH);
    265 				}
    266 			}
    267 		}
    268 		if (rt->rt_flags & RTF_REJECT)
    269 			if (rt->rt_rmx.rmx_expire == 0 ||
    270 			    (u_long) time_second < rt->rt_rmx.rmx_expire)
    271 				senderr(rt == rt0 ? EHOSTDOWN : EHOSTUNREACH);
    272 	}
    273 
    274 	switch (dst->sa_family) {
    275 
    276 #ifdef INET
    277 	case AF_INET:
    278 		if (m->m_flags & M_BCAST)
    279 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
    280 		else if (m->m_flags & M_MCAST)
    281 			ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst);
    282 		else if (!arpresolve(ifp, rt, m, dst, edst))
    283 			return (0);	/* if not yet resolved */
    284 		/* If broadcasting on a simplex interface, loopback a copy */
    285 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
    286 			mcopy = m_copy(m, 0, (int)M_COPYALL);
    287 		etype = htons(ETHERTYPE_IP);
    288 		break;
    289 
    290 	case AF_ARP:
    291 		ah = mtod(m, struct arphdr *);
    292 		if (m->m_flags & M_BCAST)
    293 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
    294 		else {
    295 			void *tha = ar_tha(ah);
    296 
    297 			if (tha == NULL) {
    298 				/* fake with ARPHDR_IEEE1394 */
    299 				return 0;
    300 			}
    301 			memcpy(edst, tha, sizeof(edst));
    302 		}
    303 
    304 		ah->ar_hrd = htons(ARPHRD_ETHER);
    305 
    306 		switch (ntohs(ah->ar_op)) {
    307 		case ARPOP_REVREQUEST:
    308 		case ARPOP_REVREPLY:
    309 			etype = htons(ETHERTYPE_REVARP);
    310 			break;
    311 
    312 		case ARPOP_REQUEST:
    313 		case ARPOP_REPLY:
    314 		default:
    315 			etype = htons(ETHERTYPE_ARP);
    316 		}
    317 
    318 		break;
    319 #endif
    320 #ifdef INET6
    321 	case AF_INET6:
    322 		if (!nd6_storelladdr(ifp, rt, m, dst, edst, sizeof(edst))){
    323 			/* something bad happened */
    324 			return (0);
    325 		}
    326 		etype = htons(ETHERTYPE_IPV6);
    327 		break;
    328 #endif
    329 #ifdef NETATALK
    330     case AF_APPLETALK:
    331 		if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) {
    332 #ifdef NETATALKDEBUG
    333 			printf("aarpresolv failed\n");
    334 #endif /* NETATALKDEBUG */
    335 			return (0);
    336 		}
    337 		/*
    338 		 * ifaddr is the first thing in at_ifaddr
    339 		 */
    340 		aa = (struct at_ifaddr *) at_ifawithnet(
    341 		    (const struct sockaddr_at *)dst, ifp);
    342 		if (aa == NULL)
    343 		    goto bad;
    344 
    345 		/*
    346 		 * In the phase 2 case, we need to prepend an mbuf for the
    347 		 * llc header.  Since we must preserve the value of m,
    348 		 * which is passed to us by value, we m_copy() the first
    349 		 * mbuf, and use it for our llc header.
    350 		 */
    351 		if (aa->aa_flags & AFA_PHASE2) {
    352 			struct llc llc;
    353 
    354 			M_PREPEND(m, sizeof(struct llc), M_DONTWAIT);
    355 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
    356 			llc.llc_control = LLC_UI;
    357 			memcpy(llc.llc_snap_org_code, at_org_code,
    358 			    sizeof(llc.llc_snap_org_code));
    359 			llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK);
    360 			memcpy(mtod(m, void *), &llc, sizeof(struct llc));
    361 		} else {
    362 			etype = htons(ETHERTYPE_ATALK);
    363 		}
    364 		break;
    365 #endif /* NETATALK */
    366 #ifdef IPX
    367 	case AF_IPX:
    368 		etype = htons(ETHERTYPE_IPX);
    369  		memcpy(edst,
    370 		    &(((const struct sockaddr_ipx *)dst)->sipx_addr.x_host),
    371 		    sizeof(edst));
    372 		/* If broadcasting on a simplex interface, loopback a copy */
    373 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
    374 			mcopy = m_copy(m, 0, (int)M_COPYALL);
    375 		break;
    376 #endif
    377 	case pseudo_AF_HDRCMPLT:
    378 		hdrcmplt = 1;
    379 		memcpy(esrc,
    380 		    ((const struct ether_header *)dst->sa_data)->ether_shost,
    381 		    sizeof(esrc));
    382 		/* FALLTHROUGH */
    383 
    384 	case AF_UNSPEC:
    385  		memcpy(edst,
    386 		    ((const struct ether_header *)dst->sa_data)->ether_dhost,
    387 		    sizeof(edst));
    388 		/* AF_UNSPEC doesn't swap the byte order of the ether_type. */
    389 		etype = ((const struct ether_header *)dst->sa_data)->ether_type;
    390 		break;
    391 
    392 	default:
    393 		printf("%s: can't handle af%d\n", ifp->if_xname,
    394 			dst->sa_family);
    395 		senderr(EAFNOSUPPORT);
    396 	}
    397 
    398 #ifdef MPLS
    399 	if (rt0 != NULL && rt_gettag(rt0) != NULL &&
    400 	    rt_gettag(rt0)->sa_family == AF_MPLS &&
    401 	    (m->m_flags & (M_MCAST | M_BCAST)) == 0) {
    402 		union mpls_shim msh;
    403 		msh.s_addr = MPLS_GETSADDR(rt0);
    404 		if (msh.shim.label != MPLS_LABEL_IMPLNULL)
    405 			etype = htons(ETHERTYPE_MPLS);
    406 	}
    407 #endif
    408 
    409 	if (mcopy)
    410 		(void)looutput(ifp, mcopy, dst, rt);
    411 
    412 	/* If no ether type is set, this must be a 802.2 formatted packet.
    413 	 */
    414 	if (etype == 0)
    415 		etype = htons(m->m_pkthdr.len);
    416 	/*
    417 	 * Add local net header.  If no space in first mbuf,
    418 	 * allocate another.
    419 	 */
    420 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
    421 	if (m == 0)
    422 		senderr(ENOBUFS);
    423 	eh = mtod(m, struct ether_header *);
    424 	/* Note: etype is already in network byte order. */
    425 	(void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
    426  	memcpy(eh->ether_dhost, edst, sizeof(edst));
    427 	if (hdrcmplt)
    428 		memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost));
    429 	else
    430 	 	memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl),
    431 		    sizeof(eh->ether_shost));
    432 
    433 #if NCARP > 0
    434 	if (ifp0 != ifp && ifp0->if_type == IFT_CARP) {
    435 	 	memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl),
    436 		    sizeof(eh->ether_shost));
    437 	}
    438 #endif /* NCARP > 0 */
    439 
    440 	if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0)
    441 		return (error);
    442 	if (m == NULL)
    443 		return (0);
    444 
    445 #if NBRIDGE > 0
    446 	/*
    447 	 * Bridges require special output handling.
    448 	 */
    449 	if (ifp->if_bridge)
    450 		return (bridge_output(ifp, m, NULL, NULL));
    451 #endif
    452 
    453 #if NCARP > 0
    454 	if (ifp != ifp0)
    455 		ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN;
    456 #endif /* NCARP > 0 */
    457 
    458 #ifdef ALTQ
    459 	/*
    460 	 * If ALTQ is enabled on the parent interface, do
    461 	 * classification; the queueing discipline might not
    462 	 * require classification, but might require the
    463 	 * address family/header pointer in the pktattr.
    464 	 */
    465 	if (ALTQ_IS_ENABLED(&ifp->if_snd))
    466 		altq_etherclassify(&ifp->if_snd, m, &pktattr);
    467 #endif
    468 	return ifq_enqueue(ifp, m ALTQ_COMMA ALTQ_DECL(&pktattr));
    469 
    470 bad:
    471 	if (m)
    472 		m_freem(m);
    473 	return (error);
    474 }
    475 
    476 #ifdef ALTQ
    477 /*
    478  * This routine is a slight hack to allow a packet to be classified
    479  * if the Ethernet headers are present.  It will go away when ALTQ's
    480  * classification engine understands link headers.
    481  */
    482 void
    483 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
    484     struct altq_pktattr *pktattr)
    485 {
    486 	struct ether_header *eh;
    487 	uint16_t ether_type;
    488 	int hlen, af, hdrsize;
    489 	void *hdr;
    490 
    491 	hlen = ETHER_HDR_LEN;
    492 	eh = mtod(m, struct ether_header *);
    493 
    494 	ether_type = htons(eh->ether_type);
    495 
    496 	if (ether_type < ETHERMTU) {
    497 		/* LLC/SNAP */
    498 		struct llc *llc = (struct llc *)(eh + 1);
    499 		hlen += 8;
    500 
    501 		if (m->m_len < hlen ||
    502 		    llc->llc_dsap != LLC_SNAP_LSAP ||
    503 		    llc->llc_ssap != LLC_SNAP_LSAP ||
    504 		    llc->llc_control != LLC_UI) {
    505 			/* Not SNAP. */
    506 			goto bad;
    507 		}
    508 
    509 		ether_type = htons(llc->llc_un.type_snap.ether_type);
    510 	}
    511 
    512 	switch (ether_type) {
    513 	case ETHERTYPE_IP:
    514 		af = AF_INET;
    515 		hdrsize = 20;		/* sizeof(struct ip) */
    516 		break;
    517 
    518 	case ETHERTYPE_IPV6:
    519 		af = AF_INET6;
    520 		hdrsize = 40;		/* sizeof(struct ip6_hdr) */
    521 		break;
    522 
    523 	default:
    524 		af = AF_UNSPEC;
    525 		hdrsize = 0;
    526 		break;
    527 	}
    528 
    529 	while (m->m_len <= hlen) {
    530 		hlen -= m->m_len;
    531 		m = m->m_next;
    532 	}
    533 	if (m->m_len < (hlen + hdrsize)) {
    534 		/*
    535 		 * protocol header not in a single mbuf.
    536 		 * We can't cope with this situation right
    537 		 * now (but it shouldn't ever happen, really, anyhow).
    538 		 */
    539 #ifdef DEBUG
    540 		printf("altq_etherclassify: headers span multiple mbufs: "
    541 		    "%d < %d\n", m->m_len, (hlen + hdrsize));
    542 #endif
    543 		goto bad;
    544 	}
    545 
    546 	m->m_data += hlen;
    547 	m->m_len -= hlen;
    548 
    549 	hdr = mtod(m, void *);
    550 
    551 	if (ALTQ_NEEDS_CLASSIFY(ifq))
    552 		pktattr->pattr_class =
    553 		    (*ifq->altq_classify)(ifq->altq_clfier, m, af);
    554 	pktattr->pattr_af = af;
    555 	pktattr->pattr_hdr = hdr;
    556 
    557 	m->m_data -= hlen;
    558 	m->m_len += hlen;
    559 
    560 	return;
    561 
    562  bad:
    563 	pktattr->pattr_class = NULL;
    564 	pktattr->pattr_hdr = NULL;
    565 	pktattr->pattr_af = AF_UNSPEC;
    566 }
    567 #endif /* ALTQ */
    568 
    569 /*
    570  * Process a received Ethernet packet;
    571  * the packet is in the mbuf chain m with
    572  * the ether header.
    573  */
    574 void
    575 ether_input(struct ifnet *ifp, struct mbuf *m)
    576 {
    577 	struct ethercom *ec = (struct ethercom *) ifp;
    578 	struct ifqueue *inq;
    579 	uint16_t etype;
    580 	struct ether_header *eh;
    581 	size_t ehlen;
    582 #if defined (LLC) || defined(NETATALK)
    583 	struct llc *l;
    584 #endif
    585 
    586 	if ((ifp->if_flags & IFF_UP) == 0) {
    587 		m_freem(m);
    588 		return;
    589 	}
    590 
    591 #ifdef MBUFTRACE
    592 	m_claimm(m, &ec->ec_rx_mowner);
    593 #endif
    594 	eh = mtod(m, struct ether_header *);
    595 	etype = ntohs(eh->ether_type);
    596 	ehlen = sizeof(*eh);
    597 
    598 	/*
    599 	 * Determine if the packet is within its size limits.
    600 	 */
    601 	if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len >
    602 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
    603 		if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count,
    604 			    bigpktppslim)) {
    605 			printf("%s: discarding oversize frame (len=%d)\n",
    606 			    ifp->if_xname, m->m_pkthdr.len);
    607 		}
    608 		m_freem(m);
    609 		return;
    610 	}
    611 
    612 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
    613 		/*
    614 		 * If this is not a simplex interface, drop the packet
    615 		 * if it came from us.
    616 		 */
    617 		if ((ifp->if_flags & IFF_SIMPLEX) == 0 &&
    618 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost,
    619 		    ETHER_ADDR_LEN) == 0) {
    620 			m_freem(m);
    621 			return;
    622 		}
    623 
    624 		if (memcmp(etherbroadcastaddr,
    625 		    eh->ether_dhost, ETHER_ADDR_LEN) == 0)
    626 			m->m_flags |= M_BCAST;
    627 		else
    628 			m->m_flags |= M_MCAST;
    629 		ifp->if_imcasts++;
    630 	}
    631 
    632 	/* If the CRC is still on the packet, trim it off. */
    633 	if (m->m_flags & M_HASFCS) {
    634 		m_adj(m, -ETHER_CRC_LEN);
    635 		m->m_flags &= ~M_HASFCS;
    636 	}
    637 
    638 	ifp->if_ibytes += m->m_pkthdr.len;
    639 
    640 #if NBRIDGE > 0
    641 	/*
    642 	 * Tap the packet off here for a bridge.  bridge_input()
    643 	 * will return NULL if it has consumed the packet, otherwise
    644 	 * it gets processed as normal.  Note that bridge_input()
    645 	 * will always return the original packet if we need to
    646 	 * process it locally.
    647 	 */
    648 	if (ifp->if_bridge) {
    649 		/* clear M_PROMISC, in case the packets comes from a vlan */
    650 		m->m_flags &= ~M_PROMISC;
    651 		m = bridge_input(ifp, m);
    652 		if (m == NULL)
    653 			return;
    654 
    655 		/*
    656 		 * Bridge has determined that the packet is for us.
    657 		 * Update our interface pointer -- we may have had
    658 		 * to "bridge" the packet locally.
    659 		 */
    660 		ifp = m->m_pkthdr.rcvif;
    661 	} else
    662 #endif /* NBRIDGE > 0 */
    663 	{
    664 
    665 #if NCARP > 0
    666 		if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) {
    667 			/*
    668 			 * clear M_PROMISC, in case the packets comes from a
    669 			 * vlan
    670 			 */
    671 			m->m_flags &= ~M_PROMISC;
    672 			if (carp_input(m, (uint8_t *)&eh->ether_shost,
    673 			    (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0)
    674 				return;
    675 		}
    676 #endif /* NCARP > 0 */
    677 		if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 &&
    678 		    (ifp->if_flags & IFF_PROMISC) != 0 &&
    679 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost,
    680 			   ETHER_ADDR_LEN) != 0) {
    681 			m->m_flags |= M_PROMISC;
    682 		}
    683 	}
    684 
    685 	if ((m->m_flags & M_PROMISC) == 0) {
    686 		if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0)
    687 			return;
    688 		if (m == NULL)
    689 			return;
    690 
    691 		eh = mtod(m, struct ether_header *);
    692 		etype = ntohs(eh->ether_type);
    693 		ehlen = sizeof(*eh);
    694 	}
    695 
    696 #if NAGR > 0
    697 	if (ifp->if_agrprivate &&
    698 	    __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) {
    699 		m->m_flags &= ~M_PROMISC;
    700 		agr_input(ifp, m);
    701 		return;
    702 	}
    703 #endif /* NAGR > 0 */
    704 
    705 	/*
    706 	 * If VLANs are configured on the interface, check to
    707 	 * see if the device performed the decapsulation and
    708 	 * provided us with the tag.
    709 	 */
    710 	if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) {
    711 #if NVLAN > 0
    712 		/*
    713 		 * vlan_input() will either recursively call ether_input()
    714 		 * or drop the packet.
    715 		 */
    716 		vlan_input(ifp, m);
    717 #else
    718 		m_freem(m);
    719 #endif
    720 		return;
    721 	}
    722 
    723 	/*
    724 	 * Handle protocols that expect to have the Ethernet header
    725 	 * (and possibly FCS) intact.
    726 	 */
    727 	switch (etype) {
    728 	case ETHERTYPE_VLAN: {
    729 		struct ether_vlan_header *evl = (void *)eh;
    730 		/*
    731 		 * If there is a tag of 0, then the VLAN header was probably
    732 		 * just being used to store the priority.  Extract the ether
    733 		 * type, and if IP or IPV6, let them deal with it.
    734 		 */
    735 		if (m->m_len <= sizeof(*evl)
    736 		    && EVL_VLANOFTAG(evl->evl_tag) == 0) {
    737 			etype = ntohs(evl->evl_proto);
    738 			ehlen = sizeof(*evl);
    739 			if ((m->m_flags & M_PROMISC) == 0
    740 			    && (etype == ETHERTYPE_IP
    741 				|| etype == ETHERTYPE_IPV6))
    742 				break;
    743 		}
    744 #if NVLAN > 0
    745 		/*
    746 		 * vlan_input() will either recursively call ether_input()
    747 		 * or drop the packet.
    748 		 */
    749 		if (((struct ethercom *)ifp)->ec_nvlans != 0)
    750 			vlan_input(ifp, m);
    751 		else
    752 #endif /* NVLAN > 0 */
    753 			m_freem(m);
    754 		return;
    755 	}
    756 #if NPPPOE > 0
    757 	case ETHERTYPE_PPPOEDISC:
    758 	case ETHERTYPE_PPPOE:
    759 		if (m->m_flags & M_PROMISC) {
    760 			m_freem(m);
    761 			return;
    762 		}
    763 #ifndef PPPOE_SERVER
    764 		if (m->m_flags & (M_MCAST | M_BCAST)) {
    765 			m_freem(m);
    766 			return;
    767 		}
    768 #endif
    769 
    770 		if (etype == ETHERTYPE_PPPOEDISC)
    771 			inq = &ppoediscinq;
    772 		else
    773 			inq = &ppoeinq;
    774 		if (IF_QFULL(inq)) {
    775 			IF_DROP(inq);
    776 			m_freem(m);
    777 		} else
    778 			IF_ENQUEUE(inq, m);
    779 		softint_schedule(pppoe_softintr);
    780 		return;
    781 #endif /* NPPPOE > 0 */
    782 	case ETHERTYPE_SLOWPROTOCOLS: {
    783 		uint8_t subtype;
    784 
    785 #if defined(DIAGNOSTIC)
    786 		if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) {
    787 			panic("ether_input: too short slow protocol packet");
    788 		}
    789 #endif
    790 		m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype);
    791 		switch (subtype) {
    792 #if NAGR > 0
    793 		case SLOWPROTOCOLS_SUBTYPE_LACP:
    794 			if (ifp->if_agrprivate) {
    795 				ieee8023ad_lacp_input(ifp, m);
    796 				return;
    797 			}
    798 			break;
    799 
    800 		case SLOWPROTOCOLS_SUBTYPE_MARKER:
    801 			if (ifp->if_agrprivate) {
    802 				ieee8023ad_marker_input(ifp, m);
    803 				return;
    804 			}
    805 			break;
    806 #endif /* NAGR > 0 */
    807 		default:
    808 			if (subtype == 0 || subtype > 10) {
    809 				/* illegal value */
    810 				m_freem(m);
    811 				return;
    812 			}
    813 			/* unknown subtype */
    814 			break;
    815 		}
    816 		/* FALLTHROUGH */
    817 	}
    818 	default:
    819 		if (m->m_flags & M_PROMISC) {
    820 			m_freem(m);
    821 			return;
    822 		}
    823 	}
    824 
    825 	/* If the CRC is still on the packet, trim it off. */
    826 	if (m->m_flags & M_HASFCS) {
    827 		m_adj(m, -ETHER_CRC_LEN);
    828 		m->m_flags &= ~M_HASFCS;
    829 	}
    830 
    831 	if (etype > ETHERMTU + sizeof (struct ether_header)) {
    832 		/* Strip off the Ethernet header. */
    833 		m_adj(m, ehlen);
    834 
    835 		switch (etype) {
    836 #ifdef INET
    837 		case ETHERTYPE_IP:
    838 #ifdef GATEWAY
    839 			if (ipflow_fastforward(m))
    840 				return;
    841 #endif
    842 			schednetisr(NETISR_IP);
    843 			inq = &ipintrq;
    844 			break;
    845 
    846 		case ETHERTYPE_ARP:
    847 			schednetisr(NETISR_ARP);
    848 			inq = &arpintrq;
    849 			break;
    850 
    851 		case ETHERTYPE_REVARP:
    852 			revarpinput(m);	/* XXX queue? */
    853 			return;
    854 #endif
    855 #ifdef INET6
    856 		case ETHERTYPE_IPV6:
    857 			if (__predict_false(!in6_present)) {
    858 				m_freem(m);
    859 				return;
    860 			}
    861 #ifdef GATEWAY
    862 			if (ip6flow_fastforward(&m))
    863 				return;
    864 #endif
    865 			schednetisr(NETISR_IPV6);
    866 			inq = &ip6intrq;
    867 			break;
    868 #endif
    869 #ifdef IPX
    870 		case ETHERTYPE_IPX:
    871 			schednetisr(NETISR_IPX);
    872 			inq = &ipxintrq;
    873 			break;
    874 #endif
    875 #ifdef NETATALK
    876 		case ETHERTYPE_ATALK:
    877 			schednetisr(NETISR_ATALK);
    878 			inq = &atintrq1;
    879 			break;
    880 		case ETHERTYPE_AARP:
    881 			/* probably this should be done with a NETISR as well */
    882 			aarpinput(ifp, m); /* XXX */
    883 			return;
    884 #endif /* NETATALK */
    885 #ifdef MPLS
    886 		case ETHERTYPE_MPLS:
    887 			schednetisr(NETISR_MPLS);
    888 			inq = &mplsintrq;
    889 			break;
    890 #endif
    891 		default:
    892 			m_freem(m);
    893 			return;
    894 		}
    895 	} else {
    896 #if defined (LLC) || defined (NETATALK)
    897 		l = (struct llc *)(eh+1);
    898 		switch (l->llc_dsap) {
    899 #ifdef NETATALK
    900 		case LLC_SNAP_LSAP:
    901 			switch (l->llc_control) {
    902 			case LLC_UI:
    903 				if (l->llc_ssap != LLC_SNAP_LSAP) {
    904 					goto dropanyway;
    905 				}
    906 
    907 				if (memcmp(&(l->llc_snap_org_code)[0],
    908 				    at_org_code, sizeof(at_org_code)) == 0 &&
    909 				    ntohs(l->llc_snap_ether_type) ==
    910 				    ETHERTYPE_ATALK) {
    911 					inq = &atintrq2;
    912 					m_adj(m, sizeof(struct ether_header)
    913 					    + sizeof(struct llc));
    914 					schednetisr(NETISR_ATALK);
    915 					break;
    916 				}
    917 
    918 				if (memcmp(&(l->llc_snap_org_code)[0],
    919 				    aarp_org_code,
    920 				    sizeof(aarp_org_code)) == 0 &&
    921 				    ntohs(l->llc_snap_ether_type) ==
    922 				    ETHERTYPE_AARP) {
    923 					m_adj( m, sizeof(struct ether_header)
    924 					    + sizeof(struct llc));
    925 					aarpinput(ifp, m); /* XXX */
    926 				    return;
    927 				}
    928 
    929 			default:
    930 				goto dropanyway;
    931 			}
    932 			break;
    933 		dropanyway:
    934 #endif
    935 		default:
    936 			m_freem(m);
    937 			return;
    938 		}
    939 #else /* ISO || LLC || NETATALK*/
    940 		m_freem(m);
    941 		return;
    942 #endif /* ISO || LLC || NETATALK*/
    943 	}
    944 
    945 	if (IF_QFULL(inq)) {
    946 		IF_DROP(inq);
    947 		m_freem(m);
    948 	} else
    949 		IF_ENQUEUE(inq, m);
    950 }
    951 
    952 /*
    953  * Convert Ethernet address to printable (loggable) representation.
    954  */
    955 char *
    956 ether_sprintf(const u_char *ap)
    957 {
    958 	static char etherbuf[3 * ETHER_ADDR_LEN];
    959 	return ether_snprintf(etherbuf, sizeof(etherbuf), ap);
    960 }
    961 
    962 char *
    963 ether_snprintf(char *buf, size_t len, const u_char *ap)
    964 {
    965 	char *cp = buf;
    966 	size_t i;
    967 
    968 	for (i = 0; i < len / 3; i++) {
    969 		*cp++ = hexdigits[*ap >> 4];
    970 		*cp++ = hexdigits[*ap++ & 0xf];
    971 		*cp++ = ':';
    972 	}
    973 	*--cp = '\0';
    974 	return buf;
    975 }
    976 
    977 /*
    978  * Perform common duties while attaching to interface list
    979  */
    980 void
    981 ether_ifattach(struct ifnet *ifp, const uint8_t *lla)
    982 {
    983 	struct ethercom *ec = (struct ethercom *)ifp;
    984 
    985 	ifp->if_type = IFT_ETHER;
    986 	ifp->if_hdrlen = ETHER_HDR_LEN;
    987 	ifp->if_dlt = DLT_EN10MB;
    988 	ifp->if_mtu = ETHERMTU;
    989 	ifp->if_output = ether_output;
    990 	ifp->if_input = ether_input;
    991 	if (ifp->if_baudrate == 0)
    992 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
    993 
    994 	if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla));
    995 
    996 	LIST_INIT(&ec->ec_multiaddrs);
    997 	ifp->if_broadcastaddr = etherbroadcastaddr;
    998 	bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header));
    999 #ifdef MBUFTRACE
   1000 	strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname,
   1001 	    sizeof(ec->ec_tx_mowner.mo_name));
   1002 	strlcpy(ec->ec_tx_mowner.mo_descr, "tx",
   1003 	    sizeof(ec->ec_tx_mowner.mo_descr));
   1004 	strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname,
   1005 	    sizeof(ec->ec_rx_mowner.mo_name));
   1006 	strlcpy(ec->ec_rx_mowner.mo_descr, "rx",
   1007 	    sizeof(ec->ec_rx_mowner.mo_descr));
   1008 	MOWNER_ATTACH(&ec->ec_tx_mowner);
   1009 	MOWNER_ATTACH(&ec->ec_rx_mowner);
   1010 	ifp->if_mowner = &ec->ec_tx_mowner;
   1011 #endif
   1012 }
   1013 
   1014 void
   1015 ether_ifdetach(struct ifnet *ifp)
   1016 {
   1017 	struct ethercom *ec = (void *) ifp;
   1018 	struct ether_multi *enm;
   1019 	int s;
   1020 
   1021 	/*
   1022 	 * Prevent further calls to ioctl (for example turning off
   1023 	 * promiscuous mode from the bridge code), which eventually can
   1024 	 * call if_init() which can cause panics because the interface
   1025 	 * is in the process of being detached. Return device not configured
   1026 	 * instead.
   1027 	 */
   1028 	ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio;
   1029 
   1030 #if NBRIDGE > 0
   1031 	if (ifp->if_bridge)
   1032 		bridge_ifdetach(ifp);
   1033 #endif
   1034 
   1035 	bpf_detach(ifp);
   1036 
   1037 #if NVLAN > 0
   1038 	if (ec->ec_nvlans)
   1039 		vlan_ifdetach(ifp);
   1040 #endif
   1041 
   1042 	s = splnet();
   1043 	while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) {
   1044 		LIST_REMOVE(enm, enm_list);
   1045 		free(enm, M_IFMADDR);
   1046 		ec->ec_multicnt--;
   1047 	}
   1048 	splx(s);
   1049 
   1050 #if 0	/* done in if_detach() */
   1051 	if_free_sadl(ifp);
   1052 #endif
   1053 
   1054 	ifp->if_mowner = NULL;
   1055 	MOWNER_DETACH(&ec->ec_rx_mowner);
   1056 	MOWNER_DETACH(&ec->ec_tx_mowner);
   1057 }
   1058 
   1059 #if 0
   1060 /*
   1061  * This is for reference.  We have a table-driven version
   1062  * of the little-endian crc32 generator, which is faster
   1063  * than the double-loop.
   1064  */
   1065 uint32_t
   1066 ether_crc32_le(const uint8_t *buf, size_t len)
   1067 {
   1068 	uint32_t c, crc, carry;
   1069 	size_t i, j;
   1070 
   1071 	crc = 0xffffffffU;	/* initial value */
   1072 
   1073 	for (i = 0; i < len; i++) {
   1074 		c = buf[i];
   1075 		for (j = 0; j < 8; j++) {
   1076 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
   1077 			crc >>= 1;
   1078 			c >>= 1;
   1079 			if (carry)
   1080 				crc = (crc ^ ETHER_CRC_POLY_LE);
   1081 		}
   1082 	}
   1083 
   1084 	return (crc);
   1085 }
   1086 #else
   1087 uint32_t
   1088 ether_crc32_le(const uint8_t *buf, size_t len)
   1089 {
   1090 	static const uint32_t crctab[] = {
   1091 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
   1092 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
   1093 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
   1094 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
   1095 	};
   1096 	uint32_t crc;
   1097 	size_t i;
   1098 
   1099 	crc = 0xffffffffU;	/* initial value */
   1100 
   1101 	for (i = 0; i < len; i++) {
   1102 		crc ^= buf[i];
   1103 		crc = (crc >> 4) ^ crctab[crc & 0xf];
   1104 		crc = (crc >> 4) ^ crctab[crc & 0xf];
   1105 	}
   1106 
   1107 	return (crc);
   1108 }
   1109 #endif
   1110 
   1111 uint32_t
   1112 ether_crc32_be(const uint8_t *buf, size_t len)
   1113 {
   1114 	uint32_t c, crc, carry;
   1115 	size_t i, j;
   1116 
   1117 	crc = 0xffffffffU;	/* initial value */
   1118 
   1119 	for (i = 0; i < len; i++) {
   1120 		c = buf[i];
   1121 		for (j = 0; j < 8; j++) {
   1122 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
   1123 			crc <<= 1;
   1124 			c >>= 1;
   1125 			if (carry)
   1126 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
   1127 		}
   1128 	}
   1129 
   1130 	return (crc);
   1131 }
   1132 
   1133 #ifdef INET
   1134 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] =
   1135     { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 };
   1136 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] =
   1137     { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff };
   1138 #endif
   1139 #ifdef INET6
   1140 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] =
   1141     { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 };
   1142 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] =
   1143     { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff };
   1144 #endif
   1145 
   1146 /*
   1147  * ether_aton implementation, not using a static buffer.
   1148  */
   1149 int
   1150 ether_aton_r(u_char *dest, size_t len, const char *str)
   1151 {
   1152         const u_char *cp = (const void *)str;
   1153 	u_char *ep;
   1154 
   1155 #define atox(c)	(((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10))
   1156 
   1157 	if (len < ETHER_ADDR_LEN)
   1158 		return ENOSPC;
   1159 
   1160 	ep = dest + ETHER_ADDR_LEN;
   1161 
   1162 	while (*cp) {
   1163                 if (!isxdigit(*cp))
   1164                         return EINVAL;
   1165 		*dest = atox(*cp);
   1166 		cp++;
   1167                 if (isxdigit(*cp)) {
   1168                         *dest = (*dest << 4) | atox(*cp);
   1169 			dest++;
   1170 			cp++;
   1171                 } else
   1172 			dest++;
   1173 		if (dest == ep)
   1174 			return *cp == '\0' ? 0 : ENAMETOOLONG;
   1175 		switch (*cp) {
   1176 		case ':':
   1177 		case '-':
   1178 		case '.':
   1179 			cp++;
   1180 			break;
   1181 		}
   1182         }
   1183 	return ENOBUFS;
   1184 }
   1185 
   1186 /*
   1187  * Convert a sockaddr into an Ethernet address or range of Ethernet
   1188  * addresses.
   1189  */
   1190 int
   1191 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN],
   1192     uint8_t addrhi[ETHER_ADDR_LEN])
   1193 {
   1194 #ifdef INET
   1195 	const struct sockaddr_in *sin;
   1196 #endif /* INET */
   1197 #ifdef INET6
   1198 	const struct sockaddr_in6 *sin6;
   1199 #endif /* INET6 */
   1200 
   1201 	switch (sa->sa_family) {
   1202 
   1203 	case AF_UNSPEC:
   1204 		memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN);
   1205 		memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
   1206 		break;
   1207 
   1208 #ifdef INET
   1209 	case AF_INET:
   1210 		sin = satocsin(sa);
   1211 		if (sin->sin_addr.s_addr == INADDR_ANY) {
   1212 			/*
   1213 			 * An IP address of INADDR_ANY means listen to
   1214 			 * or stop listening to all of the Ethernet
   1215 			 * multicast addresses used for IP.
   1216 			 * (This is for the sake of IP multicast routers.)
   1217 			 */
   1218 			memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN);
   1219 			memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN);
   1220 		}
   1221 		else {
   1222 			ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo);
   1223 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
   1224 		}
   1225 		break;
   1226 #endif
   1227 #ifdef INET6
   1228 	case AF_INET6:
   1229 		sin6 = satocsin6(sa);
   1230 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
   1231 			/*
   1232 			 * An IP6 address of 0 means listen to or stop
   1233 			 * listening to all of the Ethernet multicast
   1234 			 * address used for IP6.
   1235 			 * (This is used for multicast routers.)
   1236 			 */
   1237 			memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN);
   1238 			memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN);
   1239 		} else {
   1240 			ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo);
   1241 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
   1242 		}
   1243 		break;
   1244 #endif
   1245 
   1246 	default:
   1247 		return EAFNOSUPPORT;
   1248 	}
   1249 	return 0;
   1250 }
   1251 
   1252 /*
   1253  * Add an Ethernet multicast address or range of addresses to the list for a
   1254  * given interface.
   1255  */
   1256 int
   1257 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec)
   1258 {
   1259 	struct ether_multi *enm;
   1260 	u_char addrlo[ETHER_ADDR_LEN];
   1261 	u_char addrhi[ETHER_ADDR_LEN];
   1262 	int s = splnet(), error;
   1263 
   1264 	error = ether_multiaddr(sa, addrlo, addrhi);
   1265 	if (error != 0) {
   1266 		splx(s);
   1267 		return error;
   1268 	}
   1269 
   1270 	/*
   1271 	 * Verify that we have valid Ethernet multicast addresses.
   1272 	 */
   1273 	if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) {
   1274 		splx(s);
   1275 		return EINVAL;
   1276 	}
   1277 	/*
   1278 	 * See if the address range is already in the list.
   1279 	 */
   1280 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
   1281 	if (enm != NULL) {
   1282 		/*
   1283 		 * Found it; just increment the reference count.
   1284 		 */
   1285 		++enm->enm_refcount;
   1286 		splx(s);
   1287 		return 0;
   1288 	}
   1289 	/*
   1290 	 * New address or range; malloc a new multicast record
   1291 	 * and link it into the interface's multicast list.
   1292 	 */
   1293 	enm = (struct ether_multi *)malloc(sizeof(*enm), M_IFMADDR, M_NOWAIT);
   1294 	if (enm == NULL) {
   1295 		splx(s);
   1296 		return ENOBUFS;
   1297 	}
   1298 	memcpy(enm->enm_addrlo, addrlo, 6);
   1299 	memcpy(enm->enm_addrhi, addrhi, 6);
   1300 	enm->enm_refcount = 1;
   1301 	LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list);
   1302 	ec->ec_multicnt++;
   1303 	splx(s);
   1304 	/*
   1305 	 * Return ENETRESET to inform the driver that the list has changed
   1306 	 * and its reception filter should be adjusted accordingly.
   1307 	 */
   1308 	return ENETRESET;
   1309 }
   1310 
   1311 /*
   1312  * Delete a multicast address record.
   1313  */
   1314 int
   1315 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec)
   1316 {
   1317 	struct ether_multi *enm;
   1318 	u_char addrlo[ETHER_ADDR_LEN];
   1319 	u_char addrhi[ETHER_ADDR_LEN];
   1320 	int s = splnet(), error;
   1321 
   1322 	error = ether_multiaddr(sa, addrlo, addrhi);
   1323 	if (error != 0) {
   1324 		splx(s);
   1325 		return (error);
   1326 	}
   1327 
   1328 	/*
   1329 	 * Look ur the address in our list.
   1330 	 */
   1331 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
   1332 	if (enm == NULL) {
   1333 		splx(s);
   1334 		return (ENXIO);
   1335 	}
   1336 	if (--enm->enm_refcount != 0) {
   1337 		/*
   1338 		 * Still some claims to this record.
   1339 		 */
   1340 		splx(s);
   1341 		return (0);
   1342 	}
   1343 	/*
   1344 	 * No remaining claims to this record; unlink and free it.
   1345 	 */
   1346 	LIST_REMOVE(enm, enm_list);
   1347 	free(enm, M_IFMADDR);
   1348 	ec->ec_multicnt--;
   1349 	splx(s);
   1350 	/*
   1351 	 * Return ENETRESET to inform the driver that the list has changed
   1352 	 * and its reception filter should be adjusted accordingly.
   1353 	 */
   1354 	return (ENETRESET);
   1355 }
   1356 
   1357 void
   1358 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb)
   1359 {
   1360 	ec->ec_ifflags_cb = cb;
   1361 }
   1362 
   1363 /*
   1364  * Common ioctls for Ethernet interfaces.  Note, we must be
   1365  * called at splnet().
   1366  */
   1367 int
   1368 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data)
   1369 {
   1370 	struct ethercom *ec = (void *) ifp;
   1371 	struct eccapreq *eccr;
   1372 	struct ifreq *ifr = (struct ifreq *)data;
   1373 	struct if_laddrreq *iflr = data;
   1374 	const struct sockaddr_dl *sdl;
   1375 	static const uint8_t zero[ETHER_ADDR_LEN];
   1376 	int error;
   1377 
   1378 	switch (cmd) {
   1379 	case SIOCINITIFADDR:
   1380 	    {
   1381 		struct ifaddr *ifa = (struct ifaddr *)data;
   1382 		if (ifa->ifa_addr->sa_family != AF_LINK
   1383 		    && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) !=
   1384 		       (IFF_UP|IFF_RUNNING)) {
   1385 			ifp->if_flags |= IFF_UP;
   1386 			if ((error = (*ifp->if_init)(ifp)) != 0)
   1387 				return error;
   1388 		}
   1389 #ifdef INET
   1390 		if (ifa->ifa_addr->sa_family == AF_INET)
   1391 			arp_ifinit(ifp, ifa);
   1392 #endif /* INET */
   1393 		return 0;
   1394 	    }
   1395 
   1396 	case SIOCSIFMTU:
   1397 	    {
   1398 		int maxmtu;
   1399 
   1400 		if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU)
   1401 			maxmtu = ETHERMTU_JUMBO;
   1402 		else
   1403 			maxmtu = ETHERMTU;
   1404 
   1405 		if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu)
   1406 			return EINVAL;
   1407 		else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET)
   1408 			return error;
   1409 		else if (ifp->if_flags & IFF_UP) {
   1410 			/* Make sure the device notices the MTU change. */
   1411 			return (*ifp->if_init)(ifp);
   1412 		} else
   1413 			return 0;
   1414 	    }
   1415 
   1416 	case SIOCSIFFLAGS:
   1417 		if ((error = ifioctl_common(ifp, cmd, data)) != 0)
   1418 			return error;
   1419 		switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) {
   1420 		case IFF_RUNNING:
   1421 			/*
   1422 			 * If interface is marked down and it is running,
   1423 			 * then stop and disable it.
   1424 			 */
   1425 			(*ifp->if_stop)(ifp, 1);
   1426 			break;
   1427 		case IFF_UP:
   1428 			/*
   1429 			 * If interface is marked up and it is stopped, then
   1430 			 * start it.
   1431 			 */
   1432 			return (*ifp->if_init)(ifp);
   1433 		case IFF_UP|IFF_RUNNING:
   1434 			error = 0;
   1435 			if (ec->ec_ifflags_cb == NULL ||
   1436 			    (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) {
   1437 				/*
   1438 				 * Reset the interface to pick up
   1439 				 * changes in any other flags that
   1440 				 * affect the hardware state.
   1441 				 */
   1442 				return (*ifp->if_init)(ifp);
   1443 			} else
   1444 				return error;
   1445 		case 0:
   1446 			break;
   1447 		}
   1448 		return 0;
   1449 	case SIOCGETHERCAP:
   1450 		eccr = (struct eccapreq *)data;
   1451 		eccr->eccr_capabilities = ec->ec_capabilities;
   1452 		eccr->eccr_capenable = ec->ec_capenable;
   1453 		return 0;
   1454 	case SIOCADDMULTI:
   1455 		return ether_addmulti(ifreq_getaddr(cmd, ifr), ec);
   1456 	case SIOCDELMULTI:
   1457 		return ether_delmulti(ifreq_getaddr(cmd, ifr), ec);
   1458 	case SIOCSIFMEDIA:
   1459 	case SIOCGIFMEDIA:
   1460 		if (ec->ec_mii == NULL)
   1461 			return ENOTTY;
   1462 		return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd);
   1463 	case SIOCALIFADDR:
   1464 		sdl = satocsdl(sstocsa(&iflr->addr));
   1465 		if (sdl->sdl_family != AF_LINK)
   1466 			;
   1467 		else if (ETHER_IS_MULTICAST(CLLADDR(sdl)))
   1468 			return EINVAL;
   1469 		else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0)
   1470 			return EINVAL;
   1471 		/*FALLTHROUGH*/
   1472 	default:
   1473 		return ifioctl_common(ifp, cmd, data);
   1474 	}
   1475 	return 0;
   1476 }
   1477