xref: /netbsd-src/sys/net/if_ethersubr.c (revision 2c6fc41c810f5088457889d00eba558e8bc74d9e)
1 /*	$NetBSD: if_ethersubr.c,v 1.198 2014/05/15 07:35:38 msaitoh 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 1.198 2014/05/15 07:35:38 msaitoh 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 	int isr = 0;
583 #if defined (LLC) || defined(NETATALK)
584 	struct llc *l;
585 #endif
586 
587 	if ((ifp->if_flags & IFF_UP) == 0) {
588 		m_freem(m);
589 		return;
590 	}
591 
592 #ifdef MBUFTRACE
593 	m_claimm(m, &ec->ec_rx_mowner);
594 #endif
595 	eh = mtod(m, struct ether_header *);
596 	etype = ntohs(eh->ether_type);
597 	ehlen = sizeof(*eh);
598 
599 	/*
600 	 * Determine if the packet is within its size limits.
601 	 */
602 	if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len >
603 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
604 		if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count,
605 			    bigpktppslim)) {
606 			printf("%s: discarding oversize frame (len=%d)\n",
607 			    ifp->if_xname, m->m_pkthdr.len);
608 		}
609 		m_freem(m);
610 		return;
611 	}
612 
613 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
614 		/*
615 		 * If this is not a simplex interface, drop the packet
616 		 * if it came from us.
617 		 */
618 		if ((ifp->if_flags & IFF_SIMPLEX) == 0 &&
619 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost,
620 		    ETHER_ADDR_LEN) == 0) {
621 			m_freem(m);
622 			return;
623 		}
624 
625 		if (memcmp(etherbroadcastaddr,
626 		    eh->ether_dhost, ETHER_ADDR_LEN) == 0)
627 			m->m_flags |= M_BCAST;
628 		else
629 			m->m_flags |= M_MCAST;
630 		ifp->if_imcasts++;
631 	}
632 
633 	/* If the CRC is still on the packet, trim it off. */
634 	if (m->m_flags & M_HASFCS) {
635 		m_adj(m, -ETHER_CRC_LEN);
636 		m->m_flags &= ~M_HASFCS;
637 	}
638 
639 	ifp->if_ibytes += m->m_pkthdr.len;
640 
641 #if NBRIDGE > 0
642 	/*
643 	 * Tap the packet off here for a bridge.  bridge_input()
644 	 * will return NULL if it has consumed the packet, otherwise
645 	 * it gets processed as normal.  Note that bridge_input()
646 	 * will always return the original packet if we need to
647 	 * process it locally.
648 	 */
649 	if (ifp->if_bridge) {
650 		/* clear M_PROMISC, in case the packets comes from a vlan */
651 		m->m_flags &= ~M_PROMISC;
652 		m = bridge_input(ifp, m);
653 		if (m == NULL)
654 			return;
655 
656 		/*
657 		 * Bridge has determined that the packet is for us.
658 		 * Update our interface pointer -- we may have had
659 		 * to "bridge" the packet locally.
660 		 */
661 		ifp = m->m_pkthdr.rcvif;
662 	} else
663 #endif /* NBRIDGE > 0 */
664 	{
665 
666 #if NCARP > 0
667 		if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) {
668 			/*
669 			 * clear M_PROMISC, in case the packets comes from a
670 			 * vlan
671 			 */
672 			m->m_flags &= ~M_PROMISC;
673 			if (carp_input(m, (uint8_t *)&eh->ether_shost,
674 			    (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0)
675 				return;
676 		}
677 #endif /* NCARP > 0 */
678 		if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 &&
679 		    (ifp->if_flags & IFF_PROMISC) != 0 &&
680 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost,
681 			   ETHER_ADDR_LEN) != 0) {
682 			m->m_flags |= M_PROMISC;
683 		}
684 	}
685 
686 	if ((m->m_flags & M_PROMISC) == 0) {
687 		if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0)
688 			return;
689 		if (m == NULL)
690 			return;
691 
692 		eh = mtod(m, struct ether_header *);
693 		etype = ntohs(eh->ether_type);
694 		ehlen = sizeof(*eh);
695 	}
696 
697 #if NAGR > 0
698 	if (ifp->if_agrprivate &&
699 	    __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) {
700 		m->m_flags &= ~M_PROMISC;
701 		agr_input(ifp, m);
702 		return;
703 	}
704 #endif /* NAGR > 0 */
705 
706 	/*
707 	 * If VLANs are configured on the interface, check to
708 	 * see if the device performed the decapsulation and
709 	 * provided us with the tag.
710 	 */
711 	if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) {
712 #if NVLAN > 0
713 		/*
714 		 * vlan_input() will either recursively call ether_input()
715 		 * or drop the packet.
716 		 */
717 		vlan_input(ifp, m);
718 #else
719 		m_freem(m);
720 #endif
721 		return;
722 	}
723 
724 	/*
725 	 * Handle protocols that expect to have the Ethernet header
726 	 * (and possibly FCS) intact.
727 	 */
728 	switch (etype) {
729 	case ETHERTYPE_VLAN: {
730 		struct ether_vlan_header *evl = (void *)eh;
731 		/*
732 		 * If there is a tag of 0, then the VLAN header was probably
733 		 * just being used to store the priority.  Extract the ether
734 		 * type, and if IP or IPV6, let them deal with it.
735 		 */
736 		if (m->m_len <= sizeof(*evl)
737 		    && EVL_VLANOFTAG(evl->evl_tag) == 0) {
738 			etype = ntohs(evl->evl_proto);
739 			ehlen = sizeof(*evl);
740 			if ((m->m_flags & M_PROMISC) == 0
741 			    && (etype == ETHERTYPE_IP
742 				|| etype == ETHERTYPE_IPV6))
743 				break;
744 		}
745 #if NVLAN > 0
746 		/*
747 		 * vlan_input() will either recursively call ether_input()
748 		 * or drop the packet.
749 		 */
750 		if (((struct ethercom *)ifp)->ec_nvlans != 0)
751 			vlan_input(ifp, m);
752 		else
753 #endif /* NVLAN > 0 */
754 			m_freem(m);
755 		return;
756 	}
757 #if NPPPOE > 0
758 	case ETHERTYPE_PPPOEDISC:
759 	case ETHERTYPE_PPPOE:
760 		if (m->m_flags & M_PROMISC) {
761 			m_freem(m);
762 			return;
763 		}
764 #ifndef PPPOE_SERVER
765 		if (m->m_flags & (M_MCAST | M_BCAST)) {
766 			m_freem(m);
767 			return;
768 		}
769 #endif
770 
771 		if (etype == ETHERTYPE_PPPOEDISC)
772 			inq = &ppoediscinq;
773 		else
774 			inq = &ppoeinq;
775 		if (IF_QFULL(inq)) {
776 			IF_DROP(inq);
777 			m_freem(m);
778 		} else
779 			IF_ENQUEUE(inq, m);
780 		softint_schedule(pppoe_softintr);
781 		return;
782 #endif /* NPPPOE > 0 */
783 	case ETHERTYPE_SLOWPROTOCOLS: {
784 		uint8_t subtype;
785 
786 #if defined(DIAGNOSTIC)
787 		if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) {
788 			panic("ether_input: too short slow protocol packet");
789 		}
790 #endif
791 		m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype);
792 		switch (subtype) {
793 #if NAGR > 0
794 		case SLOWPROTOCOLS_SUBTYPE_LACP:
795 			if (ifp->if_agrprivate) {
796 				ieee8023ad_lacp_input(ifp, m);
797 				return;
798 			}
799 			break;
800 
801 		case SLOWPROTOCOLS_SUBTYPE_MARKER:
802 			if (ifp->if_agrprivate) {
803 				ieee8023ad_marker_input(ifp, m);
804 				return;
805 			}
806 			break;
807 #endif /* NAGR > 0 */
808 		default:
809 			if (subtype == 0 || subtype > 10) {
810 				/* illegal value */
811 				m_freem(m);
812 				return;
813 			}
814 			/* unknown subtype */
815 			break;
816 		}
817 		/* FALLTHROUGH */
818 	}
819 	default:
820 		if (m->m_flags & M_PROMISC) {
821 			m_freem(m);
822 			return;
823 		}
824 	}
825 
826 	/* If the CRC is still on the packet, trim it off. */
827 	if (m->m_flags & M_HASFCS) {
828 		m_adj(m, -ETHER_CRC_LEN);
829 		m->m_flags &= ~M_HASFCS;
830 	}
831 
832 	if (etype > ETHERMTU + sizeof (struct ether_header)) {
833 		/* Strip off the Ethernet header. */
834 		m_adj(m, ehlen);
835 
836 		switch (etype) {
837 #ifdef INET
838 		case ETHERTYPE_IP:
839 #ifdef GATEWAY
840 			if (ipflow_fastforward(m))
841 				return;
842 #endif
843 			isr = NETISR_IP;
844 			inq = &ipintrq;
845 			break;
846 
847 		case ETHERTYPE_ARP:
848 			isr = NETISR_ARP;
849 			inq = &arpintrq;
850 			break;
851 
852 		case ETHERTYPE_REVARP:
853 			revarpinput(m);	/* XXX queue? */
854 			return;
855 #endif
856 #ifdef INET6
857 		case ETHERTYPE_IPV6:
858 			if (__predict_false(!in6_present)) {
859 				m_freem(m);
860 				return;
861 			}
862 #ifdef GATEWAY
863 			if (ip6flow_fastforward(&m))
864 				return;
865 #endif
866 			isr = NETISR_IPV6;
867 			inq = &ip6intrq;
868 			break;
869 #endif
870 #ifdef IPX
871 		case ETHERTYPE_IPX:
872 			isr = NETISR_IPX;
873 			inq = &ipxintrq;
874 			break;
875 #endif
876 #ifdef NETATALK
877 		case ETHERTYPE_ATALK:
878 			isr = NETISR_ATALK;
879 			inq = &atintrq1;
880 			break;
881 		case ETHERTYPE_AARP:
882 			/* probably this should be done with a NETISR as well */
883 			aarpinput(ifp, m); /* XXX */
884 			return;
885 #endif /* NETATALK */
886 #ifdef MPLS
887 		case ETHERTYPE_MPLS:
888 			isr = NETISR_MPLS;
889 			inq = &mplsintrq;
890 			break;
891 #endif
892 		default:
893 			m_freem(m);
894 			return;
895 		}
896 	} else {
897 #if defined (LLC) || defined (NETATALK)
898 		l = (struct llc *)(eh+1);
899 		switch (l->llc_dsap) {
900 #ifdef NETATALK
901 		case LLC_SNAP_LSAP:
902 			switch (l->llc_control) {
903 			case LLC_UI:
904 				if (l->llc_ssap != LLC_SNAP_LSAP) {
905 					goto dropanyway;
906 				}
907 
908 				if (memcmp(&(l->llc_snap_org_code)[0],
909 				    at_org_code, sizeof(at_org_code)) == 0 &&
910 				    ntohs(l->llc_snap_ether_type) ==
911 				    ETHERTYPE_ATALK) {
912 					inq = &atintrq2;
913 					m_adj(m, sizeof(struct ether_header)
914 					    + sizeof(struct llc));
915 					isr = NETISR_ATALK;
916 					break;
917 				}
918 
919 				if (memcmp(&(l->llc_snap_org_code)[0],
920 				    aarp_org_code,
921 				    sizeof(aarp_org_code)) == 0 &&
922 				    ntohs(l->llc_snap_ether_type) ==
923 				    ETHERTYPE_AARP) {
924 					m_adj( m, sizeof(struct ether_header)
925 					    + sizeof(struct llc));
926 					aarpinput(ifp, m); /* XXX */
927 				    return;
928 				}
929 
930 			default:
931 				goto dropanyway;
932 			}
933 			break;
934 		dropanyway:
935 #endif
936 		default:
937 			m_freem(m);
938 			return;
939 		}
940 #else /* ISO || LLC || NETATALK*/
941 		m_freem(m);
942 		return;
943 #endif /* ISO || LLC || NETATALK*/
944 	}
945 
946 	if (IF_QFULL(inq)) {
947 		IF_DROP(inq);
948 		m_freem(m);
949 	} else {
950 		IF_ENQUEUE(inq, m);
951 		schednetisr(isr);
952 	}
953 }
954 
955 /*
956  * Convert Ethernet address to printable (loggable) representation.
957  */
958 char *
959 ether_sprintf(const u_char *ap)
960 {
961 	static char etherbuf[3 * ETHER_ADDR_LEN];
962 	return ether_snprintf(etherbuf, sizeof(etherbuf), ap);
963 }
964 
965 char *
966 ether_snprintf(char *buf, size_t len, const u_char *ap)
967 {
968 	char *cp = buf;
969 	size_t i;
970 
971 	for (i = 0; i < len / 3; i++) {
972 		*cp++ = hexdigits[*ap >> 4];
973 		*cp++ = hexdigits[*ap++ & 0xf];
974 		*cp++ = ':';
975 	}
976 	*--cp = '\0';
977 	return buf;
978 }
979 
980 /*
981  * Perform common duties while attaching to interface list
982  */
983 void
984 ether_ifattach(struct ifnet *ifp, const uint8_t *lla)
985 {
986 	struct ethercom *ec = (struct ethercom *)ifp;
987 
988 	ifp->if_type = IFT_ETHER;
989 	ifp->if_hdrlen = ETHER_HDR_LEN;
990 	ifp->if_dlt = DLT_EN10MB;
991 	ifp->if_mtu = ETHERMTU;
992 	ifp->if_output = ether_output;
993 	ifp->if_input = ether_input;
994 	if (ifp->if_baudrate == 0)
995 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
996 
997 	if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla));
998 
999 	LIST_INIT(&ec->ec_multiaddrs);
1000 	ifp->if_broadcastaddr = etherbroadcastaddr;
1001 	bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header));
1002 #ifdef MBUFTRACE
1003 	strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname,
1004 	    sizeof(ec->ec_tx_mowner.mo_name));
1005 	strlcpy(ec->ec_tx_mowner.mo_descr, "tx",
1006 	    sizeof(ec->ec_tx_mowner.mo_descr));
1007 	strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname,
1008 	    sizeof(ec->ec_rx_mowner.mo_name));
1009 	strlcpy(ec->ec_rx_mowner.mo_descr, "rx",
1010 	    sizeof(ec->ec_rx_mowner.mo_descr));
1011 	MOWNER_ATTACH(&ec->ec_tx_mowner);
1012 	MOWNER_ATTACH(&ec->ec_rx_mowner);
1013 	ifp->if_mowner = &ec->ec_tx_mowner;
1014 #endif
1015 }
1016 
1017 void
1018 ether_ifdetach(struct ifnet *ifp)
1019 {
1020 	struct ethercom *ec = (void *) ifp;
1021 	struct ether_multi *enm;
1022 	int s;
1023 
1024 	/*
1025 	 * Prevent further calls to ioctl (for example turning off
1026 	 * promiscuous mode from the bridge code), which eventually can
1027 	 * call if_init() which can cause panics because the interface
1028 	 * is in the process of being detached. Return device not configured
1029 	 * instead.
1030 	 */
1031 	ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio;
1032 
1033 #if NBRIDGE > 0
1034 	if (ifp->if_bridge)
1035 		bridge_ifdetach(ifp);
1036 #endif
1037 
1038 	bpf_detach(ifp);
1039 
1040 #if NVLAN > 0
1041 	if (ec->ec_nvlans)
1042 		vlan_ifdetach(ifp);
1043 #endif
1044 
1045 	s = splnet();
1046 	while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) {
1047 		LIST_REMOVE(enm, enm_list);
1048 		free(enm, M_IFMADDR);
1049 		ec->ec_multicnt--;
1050 	}
1051 	splx(s);
1052 
1053 #if 0	/* done in if_detach() */
1054 	if_free_sadl(ifp);
1055 #endif
1056 
1057 	ifp->if_mowner = NULL;
1058 	MOWNER_DETACH(&ec->ec_rx_mowner);
1059 	MOWNER_DETACH(&ec->ec_tx_mowner);
1060 }
1061 
1062 #if 0
1063 /*
1064  * This is for reference.  We have a table-driven version
1065  * of the little-endian crc32 generator, which is faster
1066  * than the double-loop.
1067  */
1068 uint32_t
1069 ether_crc32_le(const uint8_t *buf, size_t len)
1070 {
1071 	uint32_t c, crc, carry;
1072 	size_t i, j;
1073 
1074 	crc = 0xffffffffU;	/* initial value */
1075 
1076 	for (i = 0; i < len; i++) {
1077 		c = buf[i];
1078 		for (j = 0; j < 8; j++) {
1079 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
1080 			crc >>= 1;
1081 			c >>= 1;
1082 			if (carry)
1083 				crc = (crc ^ ETHER_CRC_POLY_LE);
1084 		}
1085 	}
1086 
1087 	return (crc);
1088 }
1089 #else
1090 uint32_t
1091 ether_crc32_le(const uint8_t *buf, size_t len)
1092 {
1093 	static const uint32_t crctab[] = {
1094 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1095 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1096 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1097 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1098 	};
1099 	uint32_t crc;
1100 	size_t i;
1101 
1102 	crc = 0xffffffffU;	/* initial value */
1103 
1104 	for (i = 0; i < len; i++) {
1105 		crc ^= buf[i];
1106 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1107 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1108 	}
1109 
1110 	return (crc);
1111 }
1112 #endif
1113 
1114 uint32_t
1115 ether_crc32_be(const uint8_t *buf, size_t len)
1116 {
1117 	uint32_t c, crc, carry;
1118 	size_t i, j;
1119 
1120 	crc = 0xffffffffU;	/* initial value */
1121 
1122 	for (i = 0; i < len; i++) {
1123 		c = buf[i];
1124 		for (j = 0; j < 8; j++) {
1125 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
1126 			crc <<= 1;
1127 			c >>= 1;
1128 			if (carry)
1129 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1130 		}
1131 	}
1132 
1133 	return (crc);
1134 }
1135 
1136 #ifdef INET
1137 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] =
1138     { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 };
1139 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] =
1140     { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff };
1141 #endif
1142 #ifdef INET6
1143 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] =
1144     { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 };
1145 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] =
1146     { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff };
1147 #endif
1148 
1149 /*
1150  * ether_aton implementation, not using a static buffer.
1151  */
1152 int
1153 ether_aton_r(u_char *dest, size_t len, const char *str)
1154 {
1155         const u_char *cp = (const void *)str;
1156 	u_char *ep;
1157 
1158 #define atox(c)	(((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10))
1159 
1160 	if (len < ETHER_ADDR_LEN)
1161 		return ENOSPC;
1162 
1163 	ep = dest + ETHER_ADDR_LEN;
1164 
1165 	while (*cp) {
1166                 if (!isxdigit(*cp))
1167                         return EINVAL;
1168 		*dest = atox(*cp);
1169 		cp++;
1170                 if (isxdigit(*cp)) {
1171                         *dest = (*dest << 4) | atox(*cp);
1172 			dest++;
1173 			cp++;
1174                 } else
1175 			dest++;
1176 		if (dest == ep)
1177 			return *cp == '\0' ? 0 : ENAMETOOLONG;
1178 		switch (*cp) {
1179 		case ':':
1180 		case '-':
1181 		case '.':
1182 			cp++;
1183 			break;
1184 		}
1185         }
1186 	return ENOBUFS;
1187 }
1188 
1189 /*
1190  * Convert a sockaddr into an Ethernet address or range of Ethernet
1191  * addresses.
1192  */
1193 int
1194 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN],
1195     uint8_t addrhi[ETHER_ADDR_LEN])
1196 {
1197 #ifdef INET
1198 	const struct sockaddr_in *sin;
1199 #endif /* INET */
1200 #ifdef INET6
1201 	const struct sockaddr_in6 *sin6;
1202 #endif /* INET6 */
1203 
1204 	switch (sa->sa_family) {
1205 
1206 	case AF_UNSPEC:
1207 		memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN);
1208 		memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1209 		break;
1210 
1211 #ifdef INET
1212 	case AF_INET:
1213 		sin = satocsin(sa);
1214 		if (sin->sin_addr.s_addr == INADDR_ANY) {
1215 			/*
1216 			 * An IP address of INADDR_ANY means listen to
1217 			 * or stop listening to all of the Ethernet
1218 			 * multicast addresses used for IP.
1219 			 * (This is for the sake of IP multicast routers.)
1220 			 */
1221 			memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN);
1222 			memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN);
1223 		}
1224 		else {
1225 			ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo);
1226 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1227 		}
1228 		break;
1229 #endif
1230 #ifdef INET6
1231 	case AF_INET6:
1232 		sin6 = satocsin6(sa);
1233 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1234 			/*
1235 			 * An IP6 address of 0 means listen to or stop
1236 			 * listening to all of the Ethernet multicast
1237 			 * address used for IP6.
1238 			 * (This is used for multicast routers.)
1239 			 */
1240 			memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN);
1241 			memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN);
1242 		} else {
1243 			ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo);
1244 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1245 		}
1246 		break;
1247 #endif
1248 
1249 	default:
1250 		return EAFNOSUPPORT;
1251 	}
1252 	return 0;
1253 }
1254 
1255 /*
1256  * Add an Ethernet multicast address or range of addresses to the list for a
1257  * given interface.
1258  */
1259 int
1260 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec)
1261 {
1262 	struct ether_multi *enm;
1263 	u_char addrlo[ETHER_ADDR_LEN];
1264 	u_char addrhi[ETHER_ADDR_LEN];
1265 	int s = splnet(), error;
1266 
1267 	error = ether_multiaddr(sa, addrlo, addrhi);
1268 	if (error != 0) {
1269 		splx(s);
1270 		return error;
1271 	}
1272 
1273 	/*
1274 	 * Verify that we have valid Ethernet multicast addresses.
1275 	 */
1276 	if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) {
1277 		splx(s);
1278 		return EINVAL;
1279 	}
1280 	/*
1281 	 * See if the address range is already in the list.
1282 	 */
1283 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1284 	if (enm != NULL) {
1285 		/*
1286 		 * Found it; just increment the reference count.
1287 		 */
1288 		++enm->enm_refcount;
1289 		splx(s);
1290 		return 0;
1291 	}
1292 	/*
1293 	 * New address or range; malloc a new multicast record
1294 	 * and link it into the interface's multicast list.
1295 	 */
1296 	enm = (struct ether_multi *)malloc(sizeof(*enm), M_IFMADDR, M_NOWAIT);
1297 	if (enm == NULL) {
1298 		splx(s);
1299 		return ENOBUFS;
1300 	}
1301 	memcpy(enm->enm_addrlo, addrlo, 6);
1302 	memcpy(enm->enm_addrhi, addrhi, 6);
1303 	enm->enm_refcount = 1;
1304 	LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list);
1305 	ec->ec_multicnt++;
1306 	splx(s);
1307 	/*
1308 	 * Return ENETRESET to inform the driver that the list has changed
1309 	 * and its reception filter should be adjusted accordingly.
1310 	 */
1311 	return ENETRESET;
1312 }
1313 
1314 /*
1315  * Delete a multicast address record.
1316  */
1317 int
1318 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec)
1319 {
1320 	struct ether_multi *enm;
1321 	u_char addrlo[ETHER_ADDR_LEN];
1322 	u_char addrhi[ETHER_ADDR_LEN];
1323 	int s = splnet(), error;
1324 
1325 	error = ether_multiaddr(sa, addrlo, addrhi);
1326 	if (error != 0) {
1327 		splx(s);
1328 		return (error);
1329 	}
1330 
1331 	/*
1332 	 * Look ur the address in our list.
1333 	 */
1334 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1335 	if (enm == NULL) {
1336 		splx(s);
1337 		return (ENXIO);
1338 	}
1339 	if (--enm->enm_refcount != 0) {
1340 		/*
1341 		 * Still some claims to this record.
1342 		 */
1343 		splx(s);
1344 		return (0);
1345 	}
1346 	/*
1347 	 * No remaining claims to this record; unlink and free it.
1348 	 */
1349 	LIST_REMOVE(enm, enm_list);
1350 	free(enm, M_IFMADDR);
1351 	ec->ec_multicnt--;
1352 	splx(s);
1353 	/*
1354 	 * Return ENETRESET to inform the driver that the list has changed
1355 	 * and its reception filter should be adjusted accordingly.
1356 	 */
1357 	return (ENETRESET);
1358 }
1359 
1360 void
1361 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb)
1362 {
1363 	ec->ec_ifflags_cb = cb;
1364 }
1365 
1366 /*
1367  * Common ioctls for Ethernet interfaces.  Note, we must be
1368  * called at splnet().
1369  */
1370 int
1371 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data)
1372 {
1373 	struct ethercom *ec = (void *) ifp;
1374 	struct eccapreq *eccr;
1375 	struct ifreq *ifr = (struct ifreq *)data;
1376 	struct if_laddrreq *iflr = data;
1377 	const struct sockaddr_dl *sdl;
1378 	static const uint8_t zero[ETHER_ADDR_LEN];
1379 	int error;
1380 
1381 	switch (cmd) {
1382 	case SIOCINITIFADDR:
1383 	    {
1384 		struct ifaddr *ifa = (struct ifaddr *)data;
1385 		if (ifa->ifa_addr->sa_family != AF_LINK
1386 		    && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) !=
1387 		       (IFF_UP|IFF_RUNNING)) {
1388 			ifp->if_flags |= IFF_UP;
1389 			if ((error = (*ifp->if_init)(ifp)) != 0)
1390 				return error;
1391 		}
1392 #ifdef INET
1393 		if (ifa->ifa_addr->sa_family == AF_INET)
1394 			arp_ifinit(ifp, ifa);
1395 #endif /* INET */
1396 		return 0;
1397 	    }
1398 
1399 	case SIOCSIFMTU:
1400 	    {
1401 		int maxmtu;
1402 
1403 		if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU)
1404 			maxmtu = ETHERMTU_JUMBO;
1405 		else
1406 			maxmtu = ETHERMTU;
1407 
1408 		if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu)
1409 			return EINVAL;
1410 		else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET)
1411 			return error;
1412 		else if (ifp->if_flags & IFF_UP) {
1413 			/* Make sure the device notices the MTU change. */
1414 			return (*ifp->if_init)(ifp);
1415 		} else
1416 			return 0;
1417 	    }
1418 
1419 	case SIOCSIFFLAGS:
1420 		if ((error = ifioctl_common(ifp, cmd, data)) != 0)
1421 			return error;
1422 		switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) {
1423 		case IFF_RUNNING:
1424 			/*
1425 			 * If interface is marked down and it is running,
1426 			 * then stop and disable it.
1427 			 */
1428 			(*ifp->if_stop)(ifp, 1);
1429 			break;
1430 		case IFF_UP:
1431 			/*
1432 			 * If interface is marked up and it is stopped, then
1433 			 * start it.
1434 			 */
1435 			return (*ifp->if_init)(ifp);
1436 		case IFF_UP|IFF_RUNNING:
1437 			error = 0;
1438 			if (ec->ec_ifflags_cb == NULL ||
1439 			    (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) {
1440 				/*
1441 				 * Reset the interface to pick up
1442 				 * changes in any other flags that
1443 				 * affect the hardware state.
1444 				 */
1445 				return (*ifp->if_init)(ifp);
1446 			} else
1447 				return error;
1448 		case 0:
1449 			break;
1450 		}
1451 		return 0;
1452 	case SIOCGETHERCAP:
1453 		eccr = (struct eccapreq *)data;
1454 		eccr->eccr_capabilities = ec->ec_capabilities;
1455 		eccr->eccr_capenable = ec->ec_capenable;
1456 		return 0;
1457 	case SIOCADDMULTI:
1458 		return ether_addmulti(ifreq_getaddr(cmd, ifr), ec);
1459 	case SIOCDELMULTI:
1460 		return ether_delmulti(ifreq_getaddr(cmd, ifr), ec);
1461 	case SIOCSIFMEDIA:
1462 	case SIOCGIFMEDIA:
1463 		if (ec->ec_mii == NULL)
1464 			return ENOTTY;
1465 		return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd);
1466 	case SIOCALIFADDR:
1467 		sdl = satocsdl(sstocsa(&iflr->addr));
1468 		if (sdl->sdl_family != AF_LINK)
1469 			;
1470 		else if (ETHER_IS_MULTICAST(CLLADDR(sdl)))
1471 			return EINVAL;
1472 		else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0)
1473 			return EINVAL;
1474 		/*FALLTHROUGH*/
1475 	default:
1476 		return ifioctl_common(ifp, cmd, data);
1477 	}
1478 	return 0;
1479 }
1480