xref: /netbsd-src/sys/net/if_ethersubr.c (revision eaac9e3d28859567590b7fef63a96c293f76dbf9)
1 /*	$NetBSD: if_ethersubr.c,v 1.195 2013/06/29 21:06:58 rmind 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 1.195 2013/06/29 21:06:58 rmind 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 #ifdef MBUFTRACE
215 	m_claimm(m, ifp->if_mowner);
216 #endif
217 
218 #if NCARP > 0
219 	if (ifp->if_type == IFT_CARP) {
220 		struct ifaddr *ifa;
221 
222 		/* loop back if this is going to the carp interface */
223 		if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP &&
224 		    (ifa = ifa_ifwithaddr(dst)) != NULL &&
225 		    ifa->ifa_ifp == ifp0)
226 			return looutput(ifp0, m, dst, rt0);
227 
228 		ifp = ifp->if_carpdev;
229 		/* ac = (struct arpcom *)ifp; */
230 
231 		if ((ifp0->if_flags & (IFF_UP|IFF_RUNNING)) !=
232 		    (IFF_UP|IFF_RUNNING))
233 			senderr(ENETDOWN);
234 	}
235 #endif /* NCARP > 0 */
236 
237 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
238 		senderr(ENETDOWN);
239 	if ((rt = rt0) != NULL) {
240 		if ((rt->rt_flags & RTF_UP) == 0) {
241 			if ((rt0 = rt = rtalloc1(dst, 1)) != NULL) {
242 				rt->rt_refcnt--;
243 				if (rt->rt_ifp != ifp)
244 					return (*rt->rt_ifp->if_output)
245 							(ifp, m0, dst, rt);
246 			} else
247 				senderr(EHOSTUNREACH);
248 		}
249 		if ((rt->rt_flags & RTF_GATEWAY) && dst->sa_family != AF_NS) {
250 			if (rt->rt_gwroute == NULL)
251 				goto lookup;
252 			if (((rt = rt->rt_gwroute)->rt_flags & RTF_UP) == 0) {
253 				rtfree(rt); rt = rt0;
254 			lookup: rt->rt_gwroute = rtalloc1(rt->rt_gateway, 1);
255 				if ((rt = rt->rt_gwroute) == NULL)
256 					senderr(EHOSTUNREACH);
257 				/* the "G" test below also prevents rt == rt0 */
258 				if ((rt->rt_flags & RTF_GATEWAY) ||
259 				    (rt->rt_ifp != ifp)) {
260 					rt->rt_refcnt--;
261 					rt0->rt_gwroute = NULL;
262 					senderr(EHOSTUNREACH);
263 				}
264 			}
265 		}
266 		if (rt->rt_flags & RTF_REJECT)
267 			if (rt->rt_rmx.rmx_expire == 0 ||
268 			    (u_long) time_second < rt->rt_rmx.rmx_expire)
269 				senderr(rt == rt0 ? EHOSTDOWN : EHOSTUNREACH);
270 	}
271 
272 	switch (dst->sa_family) {
273 
274 #ifdef INET
275 	case AF_INET:
276 		if (m->m_flags & M_BCAST)
277 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
278 		else if (m->m_flags & M_MCAST)
279 			ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst);
280 		else if (!arpresolve(ifp, rt, m, dst, edst))
281 			return (0);	/* if not yet resolved */
282 		/* If broadcasting on a simplex interface, loopback a copy */
283 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
284 			mcopy = m_copy(m, 0, (int)M_COPYALL);
285 		etype = htons(ETHERTYPE_IP);
286 		break;
287 
288 	case AF_ARP:
289 		ah = mtod(m, struct arphdr *);
290 		if (m->m_flags & M_BCAST)
291 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
292 		else {
293 			void *tha = ar_tha(ah);
294 
295 			if (tha == NULL) {
296 				/* fake with ARPHDR_IEEE1394 */
297 				return 0;
298 			}
299 			memcpy(edst, tha, sizeof(edst));
300 		}
301 
302 		ah->ar_hrd = htons(ARPHRD_ETHER);
303 
304 		switch (ntohs(ah->ar_op)) {
305 		case ARPOP_REVREQUEST:
306 		case ARPOP_REVREPLY:
307 			etype = htons(ETHERTYPE_REVARP);
308 			break;
309 
310 		case ARPOP_REQUEST:
311 		case ARPOP_REPLY:
312 		default:
313 			etype = htons(ETHERTYPE_ARP);
314 		}
315 
316 		break;
317 #endif
318 #ifdef INET6
319 	case AF_INET6:
320 		if (!nd6_storelladdr(ifp, rt, m, dst, edst, sizeof(edst))){
321 			/* something bad happened */
322 			return (0);
323 		}
324 		etype = htons(ETHERTYPE_IPV6);
325 		break;
326 #endif
327 #ifdef NETATALK
328     case AF_APPLETALK:
329 		if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) {
330 #ifdef NETATALKDEBUG
331 			printf("aarpresolv failed\n");
332 #endif /* NETATALKDEBUG */
333 			return (0);
334 		}
335 		/*
336 		 * ifaddr is the first thing in at_ifaddr
337 		 */
338 		aa = (struct at_ifaddr *) at_ifawithnet(
339 		    (const struct sockaddr_at *)dst, ifp);
340 		if (aa == NULL)
341 		    goto bad;
342 
343 		/*
344 		 * In the phase 2 case, we need to prepend an mbuf for the
345 		 * llc header.  Since we must preserve the value of m,
346 		 * which is passed to us by value, we m_copy() the first
347 		 * mbuf, and use it for our llc header.
348 		 */
349 		if (aa->aa_flags & AFA_PHASE2) {
350 			struct llc llc;
351 
352 			M_PREPEND(m, sizeof(struct llc), M_DONTWAIT);
353 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
354 			llc.llc_control = LLC_UI;
355 			memcpy(llc.llc_snap_org_code, at_org_code,
356 			    sizeof(llc.llc_snap_org_code));
357 			llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK);
358 			memcpy(mtod(m, void *), &llc, sizeof(struct llc));
359 		} else {
360 			etype = htons(ETHERTYPE_ATALK);
361 		}
362 		break;
363 #endif /* NETATALK */
364 #ifdef IPX
365 	case AF_IPX:
366 		etype = htons(ETHERTYPE_IPX);
367  		memcpy(edst,
368 		    &(((const struct sockaddr_ipx *)dst)->sipx_addr.x_host),
369 		    sizeof(edst));
370 		/* If broadcasting on a simplex interface, loopback a copy */
371 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
372 			mcopy = m_copy(m, 0, (int)M_COPYALL);
373 		break;
374 #endif
375 	case pseudo_AF_HDRCMPLT:
376 		hdrcmplt = 1;
377 		memcpy(esrc,
378 		    ((const struct ether_header *)dst->sa_data)->ether_shost,
379 		    sizeof(esrc));
380 		/* FALLTHROUGH */
381 
382 	case AF_UNSPEC:
383  		memcpy(edst,
384 		    ((const struct ether_header *)dst->sa_data)->ether_dhost,
385 		    sizeof(edst));
386 		/* AF_UNSPEC doesn't swap the byte order of the ether_type. */
387 		etype = ((const struct ether_header *)dst->sa_data)->ether_type;
388 		break;
389 
390 	default:
391 		printf("%s: can't handle af%d\n", ifp->if_xname,
392 			dst->sa_family);
393 		senderr(EAFNOSUPPORT);
394 	}
395 
396 #ifdef MPLS
397 	if (rt0 != NULL && rt_gettag(rt0) != NULL &&
398 	    rt_gettag(rt0)->sa_family == AF_MPLS &&
399 	    (m->m_flags & (M_MCAST | M_BCAST)) == 0) {
400 		union mpls_shim msh;
401 		msh.s_addr = MPLS_GETSADDR(rt0);
402 		if (msh.shim.label != MPLS_LABEL_IMPLNULL)
403 			etype = htons(ETHERTYPE_MPLS);
404 	}
405 #endif
406 
407 	if (mcopy)
408 		(void)looutput(ifp, mcopy, dst, rt);
409 
410 	/* If no ether type is set, this must be a 802.2 formatted packet.
411 	 */
412 	if (etype == 0)
413 		etype = htons(m->m_pkthdr.len);
414 	/*
415 	 * Add local net header.  If no space in first mbuf,
416 	 * allocate another.
417 	 */
418 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
419 	if (m == 0)
420 		senderr(ENOBUFS);
421 	eh = mtod(m, struct ether_header *);
422 	/* Note: etype is already in network byte order. */
423 	(void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
424  	memcpy(eh->ether_dhost, edst, sizeof(edst));
425 	if (hdrcmplt)
426 		memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost));
427 	else
428 	 	memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl),
429 		    sizeof(eh->ether_shost));
430 
431 #if NCARP > 0
432 	if (ifp0 != ifp && ifp0->if_type == IFT_CARP) {
433 	 	memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl),
434 		    sizeof(eh->ether_shost));
435 	}
436 #endif /* NCARP > 0 */
437 
438 	if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0)
439 		return (error);
440 	if (m == NULL)
441 		return (0);
442 
443 #if NBRIDGE > 0
444 	/*
445 	 * Bridges require special output handling.
446 	 */
447 	if (ifp->if_bridge)
448 		return (bridge_output(ifp, m, NULL, NULL));
449 #endif
450 
451 #if NCARP > 0
452 	if (ifp != ifp0)
453 		ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN;
454 #endif /* NCARP > 0 */
455 
456 #ifdef ALTQ
457 	/*
458 	 * If ALTQ is enabled on the parent interface, do
459 	 * classification; the queueing discipline might not
460 	 * require classification, but might require the
461 	 * address family/header pointer in the pktattr.
462 	 */
463 	if (ALTQ_IS_ENABLED(&ifp->if_snd))
464 		altq_etherclassify(&ifp->if_snd, m, &pktattr);
465 #endif
466 	return ifq_enqueue(ifp, m ALTQ_COMMA ALTQ_DECL(&pktattr));
467 
468 bad:
469 	if (m)
470 		m_freem(m);
471 	return (error);
472 }
473 
474 #ifdef ALTQ
475 /*
476  * This routine is a slight hack to allow a packet to be classified
477  * if the Ethernet headers are present.  It will go away when ALTQ's
478  * classification engine understands link headers.
479  */
480 void
481 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
482     struct altq_pktattr *pktattr)
483 {
484 	struct ether_header *eh;
485 	uint16_t ether_type;
486 	int hlen, af, hdrsize;
487 	void *hdr;
488 
489 	hlen = ETHER_HDR_LEN;
490 	eh = mtod(m, struct ether_header *);
491 
492 	ether_type = htons(eh->ether_type);
493 
494 	if (ether_type < ETHERMTU) {
495 		/* LLC/SNAP */
496 		struct llc *llc = (struct llc *)(eh + 1);
497 		hlen += 8;
498 
499 		if (m->m_len < hlen ||
500 		    llc->llc_dsap != LLC_SNAP_LSAP ||
501 		    llc->llc_ssap != LLC_SNAP_LSAP ||
502 		    llc->llc_control != LLC_UI) {
503 			/* Not SNAP. */
504 			goto bad;
505 		}
506 
507 		ether_type = htons(llc->llc_un.type_snap.ether_type);
508 	}
509 
510 	switch (ether_type) {
511 	case ETHERTYPE_IP:
512 		af = AF_INET;
513 		hdrsize = 20;		/* sizeof(struct ip) */
514 		break;
515 
516 	case ETHERTYPE_IPV6:
517 		af = AF_INET6;
518 		hdrsize = 40;		/* sizeof(struct ip6_hdr) */
519 		break;
520 
521 	default:
522 		af = AF_UNSPEC;
523 		hdrsize = 0;
524 		break;
525 	}
526 
527 	while (m->m_len <= hlen) {
528 		hlen -= m->m_len;
529 		m = m->m_next;
530 	}
531 	if (m->m_len < (hlen + hdrsize)) {
532 		/*
533 		 * protocol header not in a single mbuf.
534 		 * We can't cope with this situation right
535 		 * now (but it shouldn't ever happen, really, anyhow).
536 		 */
537 #ifdef DEBUG
538 		printf("altq_etherclassify: headers span multiple mbufs: "
539 		    "%d < %d\n", m->m_len, (hlen + hdrsize));
540 #endif
541 		goto bad;
542 	}
543 
544 	m->m_data += hlen;
545 	m->m_len -= hlen;
546 
547 	hdr = mtod(m, void *);
548 
549 	if (ALTQ_NEEDS_CLASSIFY(ifq))
550 		pktattr->pattr_class =
551 		    (*ifq->altq_classify)(ifq->altq_clfier, m, af);
552 	pktattr->pattr_af = af;
553 	pktattr->pattr_hdr = hdr;
554 
555 	m->m_data -= hlen;
556 	m->m_len += hlen;
557 
558 	return;
559 
560  bad:
561 	pktattr->pattr_class = NULL;
562 	pktattr->pattr_hdr = NULL;
563 	pktattr->pattr_af = AF_UNSPEC;
564 }
565 #endif /* ALTQ */
566 
567 /*
568  * Process a received Ethernet packet;
569  * the packet is in the mbuf chain m with
570  * the ether header.
571  */
572 void
573 ether_input(struct ifnet *ifp, struct mbuf *m)
574 {
575 	struct ethercom *ec = (struct ethercom *) ifp;
576 	struct ifqueue *inq;
577 	uint16_t etype;
578 	struct ether_header *eh;
579 	size_t ehlen;
580 #if defined (LLC) || defined(NETATALK)
581 	struct llc *l;
582 #endif
583 
584 	if ((ifp->if_flags & IFF_UP) == 0) {
585 		m_freem(m);
586 		return;
587 	}
588 
589 #ifdef MBUFTRACE
590 	m_claimm(m, &ec->ec_rx_mowner);
591 #endif
592 	eh = mtod(m, struct ether_header *);
593 	etype = ntohs(eh->ether_type);
594 	ehlen = sizeof(*eh);
595 
596 	/*
597 	 * Determine if the packet is within its size limits.
598 	 */
599 	if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len >
600 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
601 		if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count,
602 			    bigpktppslim)) {
603 			printf("%s: discarding oversize frame (len=%d)\n",
604 			    ifp->if_xname, m->m_pkthdr.len);
605 		}
606 		m_freem(m);
607 		return;
608 	}
609 
610 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
611 		/*
612 		 * If this is not a simplex interface, drop the packet
613 		 * if it came from us.
614 		 */
615 		if ((ifp->if_flags & IFF_SIMPLEX) == 0 &&
616 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost,
617 		    ETHER_ADDR_LEN) == 0) {
618 			m_freem(m);
619 			return;
620 		}
621 
622 		if (memcmp(etherbroadcastaddr,
623 		    eh->ether_dhost, ETHER_ADDR_LEN) == 0)
624 			m->m_flags |= M_BCAST;
625 		else
626 			m->m_flags |= M_MCAST;
627 		ifp->if_imcasts++;
628 	}
629 
630 	/* If the CRC is still on the packet, trim it off. */
631 	if (m->m_flags & M_HASFCS) {
632 		m_adj(m, -ETHER_CRC_LEN);
633 		m->m_flags &= ~M_HASFCS;
634 	}
635 
636 	ifp->if_ibytes += m->m_pkthdr.len;
637 
638 #if NBRIDGE > 0
639 	/*
640 	 * Tap the packet off here for a bridge.  bridge_input()
641 	 * will return NULL if it has consumed the packet, otherwise
642 	 * it gets processed as normal.  Note that bridge_input()
643 	 * will always return the original packet if we need to
644 	 * process it locally.
645 	 */
646 	if (ifp->if_bridge) {
647 		/* clear M_PROMISC, in case the packets comes from a vlan */
648 		m->m_flags &= ~M_PROMISC;
649 		m = bridge_input(ifp, m);
650 		if (m == NULL)
651 			return;
652 
653 		/*
654 		 * Bridge has determined that the packet is for us.
655 		 * Update our interface pointer -- we may have had
656 		 * to "bridge" the packet locally.
657 		 */
658 		ifp = m->m_pkthdr.rcvif;
659 	} else
660 #endif /* NBRIDGE > 0 */
661 	{
662 
663 #if NCARP > 0
664 		if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) {
665 			/*
666 			 * clear M_PROMISC, in case the packets comes from a
667 			 * vlan
668 			 */
669 			m->m_flags &= ~M_PROMISC;
670 			if (carp_input(m, (uint8_t *)&eh->ether_shost,
671 			    (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0)
672 				return;
673 		}
674 #endif /* NCARP > 0 */
675 		if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 &&
676 		    (ifp->if_flags & IFF_PROMISC) != 0 &&
677 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost,
678 			   ETHER_ADDR_LEN) != 0) {
679 			m->m_flags |= M_PROMISC;
680 		}
681 	}
682 
683 	if ((m->m_flags & M_PROMISC) == 0) {
684 		if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0)
685 			return;
686 		if (m == NULL)
687 			return;
688 
689 		eh = mtod(m, struct ether_header *);
690 		etype = ntohs(eh->ether_type);
691 		ehlen = sizeof(*eh);
692 	}
693 
694 #if NAGR > 0
695 	if (ifp->if_agrprivate &&
696 	    __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) {
697 		m->m_flags &= ~M_PROMISC;
698 		agr_input(ifp, m);
699 		return;
700 	}
701 #endif /* NAGR > 0 */
702 
703 	/*
704 	 * If VLANs are configured on the interface, check to
705 	 * see if the device performed the decapsulation and
706 	 * provided us with the tag.
707 	 */
708 	if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) {
709 #if NVLAN > 0
710 		/*
711 		 * vlan_input() will either recursively call ether_input()
712 		 * or drop the packet.
713 		 */
714 		vlan_input(ifp, m);
715 #else
716 		m_freem(m);
717 #endif
718 		return;
719 	}
720 
721 	/*
722 	 * Handle protocols that expect to have the Ethernet header
723 	 * (and possibly FCS) intact.
724 	 */
725 	switch (etype) {
726 	case ETHERTYPE_VLAN: {
727 		struct ether_vlan_header *evl = (void *)eh;
728 		/*
729 		 * If there is a tag of 0, then the VLAN header was probably
730 		 * just being used to store the priority.  Extract the ether
731 		 * type, and if IP or IPV6, let them deal with it.
732 		 */
733 		if (m->m_len <= sizeof(*evl)
734 		    && EVL_VLANOFTAG(evl->evl_tag) == 0) {
735 			etype = ntohs(evl->evl_proto);
736 			ehlen = sizeof(*evl);
737 			if ((m->m_flags & M_PROMISC) == 0
738 			    && (etype == ETHERTYPE_IP
739 				|| etype == ETHERTYPE_IPV6))
740 				break;
741 		}
742 #if NVLAN > 0
743 		/*
744 		 * vlan_input() will either recursively call ether_input()
745 		 * or drop the packet.
746 		 */
747 		if (((struct ethercom *)ifp)->ec_nvlans != 0)
748 			vlan_input(ifp, m);
749 		else
750 #endif /* NVLAN > 0 */
751 			m_freem(m);
752 		return;
753 	}
754 #if NPPPOE > 0
755 	case ETHERTYPE_PPPOEDISC:
756 	case ETHERTYPE_PPPOE:
757 		if (m->m_flags & M_PROMISC) {
758 			m_freem(m);
759 			return;
760 		}
761 #ifndef PPPOE_SERVER
762 		if (m->m_flags & (M_MCAST | M_BCAST)) {
763 			m_freem(m);
764 			return;
765 		}
766 #endif
767 
768 		if (etype == ETHERTYPE_PPPOEDISC)
769 			inq = &ppoediscinq;
770 		else
771 			inq = &ppoeinq;
772 		if (IF_QFULL(inq)) {
773 			IF_DROP(inq);
774 			m_freem(m);
775 		} else
776 			IF_ENQUEUE(inq, m);
777 		softint_schedule(pppoe_softintr);
778 		return;
779 #endif /* NPPPOE > 0 */
780 	case ETHERTYPE_SLOWPROTOCOLS: {
781 		uint8_t subtype;
782 
783 #if defined(DIAGNOSTIC)
784 		if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) {
785 			panic("ether_input: too short slow protocol packet");
786 		}
787 #endif
788 		m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype);
789 		switch (subtype) {
790 #if NAGR > 0
791 		case SLOWPROTOCOLS_SUBTYPE_LACP:
792 			if (ifp->if_agrprivate) {
793 				ieee8023ad_lacp_input(ifp, m);
794 				return;
795 			}
796 			break;
797 
798 		case SLOWPROTOCOLS_SUBTYPE_MARKER:
799 			if (ifp->if_agrprivate) {
800 				ieee8023ad_marker_input(ifp, m);
801 				return;
802 			}
803 			break;
804 #endif /* NAGR > 0 */
805 		default:
806 			if (subtype == 0 || subtype > 10) {
807 				/* illegal value */
808 				m_freem(m);
809 				return;
810 			}
811 			/* unknown subtype */
812 			break;
813 		}
814 		/* FALLTHROUGH */
815 	}
816 	default:
817 		if (m->m_flags & M_PROMISC) {
818 			m_freem(m);
819 			return;
820 		}
821 	}
822 
823 	/* If the CRC is still on the packet, trim it off. */
824 	if (m->m_flags & M_HASFCS) {
825 		m_adj(m, -ETHER_CRC_LEN);
826 		m->m_flags &= ~M_HASFCS;
827 	}
828 
829 	if (etype > ETHERMTU + sizeof (struct ether_header)) {
830 		/* Strip off the Ethernet header. */
831 		m_adj(m, ehlen);
832 
833 		switch (etype) {
834 #ifdef INET
835 		case ETHERTYPE_IP:
836 #ifdef GATEWAY
837 			if (ipflow_fastforward(m))
838 				return;
839 #endif
840 			schednetisr(NETISR_IP);
841 			inq = &ipintrq;
842 			break;
843 
844 		case ETHERTYPE_ARP:
845 			schednetisr(NETISR_ARP);
846 			inq = &arpintrq;
847 			break;
848 
849 		case ETHERTYPE_REVARP:
850 			revarpinput(m);	/* XXX queue? */
851 			return;
852 #endif
853 #ifdef INET6
854 		case ETHERTYPE_IPV6:
855 #ifdef GATEWAY
856 			if (ip6flow_fastforward(&m))
857 				return;
858 #endif
859 			schednetisr(NETISR_IPV6);
860 			inq = &ip6intrq;
861 			break;
862 #endif
863 #ifdef IPX
864 		case ETHERTYPE_IPX:
865 			schednetisr(NETISR_IPX);
866 			inq = &ipxintrq;
867 			break;
868 #endif
869 #ifdef NETATALK
870 		case ETHERTYPE_ATALK:
871 			schednetisr(NETISR_ATALK);
872 			inq = &atintrq1;
873 			break;
874 		case ETHERTYPE_AARP:
875 			/* probably this should be done with a NETISR as well */
876 			aarpinput(ifp, m); /* XXX */
877 			return;
878 #endif /* NETATALK */
879 #ifdef MPLS
880 		case ETHERTYPE_MPLS:
881 			schednetisr(NETISR_MPLS);
882 			inq = &mplsintrq;
883 			break;
884 #endif
885 		default:
886 			m_freem(m);
887 			return;
888 		}
889 	} else {
890 #if defined (LLC) || defined (NETATALK)
891 		l = (struct llc *)(eh+1);
892 		switch (l->llc_dsap) {
893 #ifdef NETATALK
894 		case LLC_SNAP_LSAP:
895 			switch (l->llc_control) {
896 			case LLC_UI:
897 				if (l->llc_ssap != LLC_SNAP_LSAP) {
898 					goto dropanyway;
899 				}
900 
901 				if (memcmp(&(l->llc_snap_org_code)[0],
902 				    at_org_code, sizeof(at_org_code)) == 0 &&
903 				    ntohs(l->llc_snap_ether_type) ==
904 				    ETHERTYPE_ATALK) {
905 					inq = &atintrq2;
906 					m_adj(m, sizeof(struct ether_header)
907 					    + sizeof(struct llc));
908 					schednetisr(NETISR_ATALK);
909 					break;
910 				}
911 
912 				if (memcmp(&(l->llc_snap_org_code)[0],
913 				    aarp_org_code,
914 				    sizeof(aarp_org_code)) == 0 &&
915 				    ntohs(l->llc_snap_ether_type) ==
916 				    ETHERTYPE_AARP) {
917 					m_adj( m, sizeof(struct ether_header)
918 					    + sizeof(struct llc));
919 					aarpinput(ifp, m); /* XXX */
920 				    return;
921 				}
922 
923 			default:
924 				goto dropanyway;
925 			}
926 			break;
927 		dropanyway:
928 #endif
929 		default:
930 			m_freem(m);
931 			return;
932 		}
933 #else /* ISO || LLC || NETATALK*/
934 		m_freem(m);
935 		return;
936 #endif /* ISO || LLC || NETATALK*/
937 	}
938 
939 	if (IF_QFULL(inq)) {
940 		IF_DROP(inq);
941 		m_freem(m);
942 	} else
943 		IF_ENQUEUE(inq, m);
944 }
945 
946 /*
947  * Convert Ethernet address to printable (loggable) representation.
948  */
949 char *
950 ether_sprintf(const u_char *ap)
951 {
952 	static char etherbuf[3 * ETHER_ADDR_LEN];
953 	return ether_snprintf(etherbuf, sizeof(etherbuf), ap);
954 }
955 
956 char *
957 ether_snprintf(char *buf, size_t len, const u_char *ap)
958 {
959 	char *cp = buf;
960 	size_t i;
961 
962 	for (i = 0; i < len / 3; i++) {
963 		*cp++ = hexdigits[*ap >> 4];
964 		*cp++ = hexdigits[*ap++ & 0xf];
965 		*cp++ = ':';
966 	}
967 	*--cp = '\0';
968 	return buf;
969 }
970 
971 /*
972  * Perform common duties while attaching to interface list
973  */
974 void
975 ether_ifattach(struct ifnet *ifp, const uint8_t *lla)
976 {
977 	struct ethercom *ec = (struct ethercom *)ifp;
978 
979 	ifp->if_type = IFT_ETHER;
980 	ifp->if_hdrlen = ETHER_HDR_LEN;
981 	ifp->if_dlt = DLT_EN10MB;
982 	ifp->if_mtu = ETHERMTU;
983 	ifp->if_output = ether_output;
984 	ifp->if_input = ether_input;
985 	if (ifp->if_baudrate == 0)
986 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
987 
988 	if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla));
989 
990 	LIST_INIT(&ec->ec_multiaddrs);
991 	ifp->if_broadcastaddr = etherbroadcastaddr;
992 	bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header));
993 #ifdef MBUFTRACE
994 	strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname,
995 	    sizeof(ec->ec_tx_mowner.mo_name));
996 	strlcpy(ec->ec_tx_mowner.mo_descr, "tx",
997 	    sizeof(ec->ec_tx_mowner.mo_descr));
998 	strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname,
999 	    sizeof(ec->ec_rx_mowner.mo_name));
1000 	strlcpy(ec->ec_rx_mowner.mo_descr, "rx",
1001 	    sizeof(ec->ec_rx_mowner.mo_descr));
1002 	MOWNER_ATTACH(&ec->ec_tx_mowner);
1003 	MOWNER_ATTACH(&ec->ec_rx_mowner);
1004 	ifp->if_mowner = &ec->ec_tx_mowner;
1005 #endif
1006 }
1007 
1008 void
1009 ether_ifdetach(struct ifnet *ifp)
1010 {
1011 	struct ethercom *ec = (void *) ifp;
1012 	struct ether_multi *enm;
1013 	int s;
1014 
1015 	/*
1016 	 * Prevent further calls to ioctl (for example turning off
1017 	 * promiscuous mode from the bridge code), which eventually can
1018 	 * call if_init() which can cause panics because the interface
1019 	 * is in the process of being detached. Return device not configured
1020 	 * instead.
1021 	 */
1022 	ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio;
1023 
1024 #if NBRIDGE > 0
1025 	if (ifp->if_bridge)
1026 		bridge_ifdetach(ifp);
1027 #endif
1028 
1029 	bpf_detach(ifp);
1030 
1031 #if NVLAN > 0
1032 	if (ec->ec_nvlans)
1033 		vlan_ifdetach(ifp);
1034 #endif
1035 
1036 	s = splnet();
1037 	while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) {
1038 		LIST_REMOVE(enm, enm_list);
1039 		free(enm, M_IFMADDR);
1040 		ec->ec_multicnt--;
1041 	}
1042 	splx(s);
1043 
1044 #if 0	/* done in if_detach() */
1045 	if_free_sadl(ifp);
1046 #endif
1047 
1048 	ifp->if_mowner = NULL;
1049 	MOWNER_DETACH(&ec->ec_rx_mowner);
1050 	MOWNER_DETACH(&ec->ec_tx_mowner);
1051 }
1052 
1053 #if 0
1054 /*
1055  * This is for reference.  We have a table-driven version
1056  * of the little-endian crc32 generator, which is faster
1057  * than the double-loop.
1058  */
1059 uint32_t
1060 ether_crc32_le(const uint8_t *buf, size_t len)
1061 {
1062 	uint32_t c, crc, carry;
1063 	size_t i, j;
1064 
1065 	crc = 0xffffffffU;	/* initial value */
1066 
1067 	for (i = 0; i < len; i++) {
1068 		c = buf[i];
1069 		for (j = 0; j < 8; j++) {
1070 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
1071 			crc >>= 1;
1072 			c >>= 1;
1073 			if (carry)
1074 				crc = (crc ^ ETHER_CRC_POLY_LE);
1075 		}
1076 	}
1077 
1078 	return (crc);
1079 }
1080 #else
1081 uint32_t
1082 ether_crc32_le(const uint8_t *buf, size_t len)
1083 {
1084 	static const uint32_t crctab[] = {
1085 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1086 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1087 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1088 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1089 	};
1090 	uint32_t crc;
1091 	size_t i;
1092 
1093 	crc = 0xffffffffU;	/* initial value */
1094 
1095 	for (i = 0; i < len; i++) {
1096 		crc ^= buf[i];
1097 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1098 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1099 	}
1100 
1101 	return (crc);
1102 }
1103 #endif
1104 
1105 uint32_t
1106 ether_crc32_be(const uint8_t *buf, size_t len)
1107 {
1108 	uint32_t c, crc, carry;
1109 	size_t i, j;
1110 
1111 	crc = 0xffffffffU;	/* initial value */
1112 
1113 	for (i = 0; i < len; i++) {
1114 		c = buf[i];
1115 		for (j = 0; j < 8; j++) {
1116 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
1117 			crc <<= 1;
1118 			c >>= 1;
1119 			if (carry)
1120 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1121 		}
1122 	}
1123 
1124 	return (crc);
1125 }
1126 
1127 #ifdef INET
1128 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] =
1129     { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 };
1130 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] =
1131     { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff };
1132 #endif
1133 #ifdef INET6
1134 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] =
1135     { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 };
1136 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] =
1137     { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff };
1138 #endif
1139 
1140 /*
1141  * ether_aton implementation, not using a static buffer.
1142  */
1143 int
1144 ether_aton_r(u_char *dest, size_t len, const char *str)
1145 {
1146         const u_char *cp = (const void *)str;
1147 	u_char *ep;
1148 
1149 #define atox(c)	(((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10))
1150 
1151 	if (len < ETHER_ADDR_LEN)
1152 		return ENOSPC;
1153 
1154 	ep = dest + ETHER_ADDR_LEN;
1155 
1156 	while (*cp) {
1157                 if (!isxdigit(*cp))
1158                         return EINVAL;
1159 		*dest = atox(*cp);
1160 		cp++;
1161                 if (isxdigit(*cp)) {
1162                         *dest = (*dest << 4) | atox(*cp);
1163 			dest++;
1164 			cp++;
1165                 } else
1166 			dest++;
1167 		if (dest == ep)
1168 			return *cp == '\0' ? 0 : ENAMETOOLONG;
1169 		switch (*cp) {
1170 		case ':':
1171 		case '-':
1172 		case '.':
1173 			cp++;
1174 			break;
1175 		}
1176         }
1177 	return ENOBUFS;
1178 }
1179 
1180 /*
1181  * Convert a sockaddr into an Ethernet address or range of Ethernet
1182  * addresses.
1183  */
1184 int
1185 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN],
1186     uint8_t addrhi[ETHER_ADDR_LEN])
1187 {
1188 #ifdef INET
1189 	const struct sockaddr_in *sin;
1190 #endif /* INET */
1191 #ifdef INET6
1192 	const struct sockaddr_in6 *sin6;
1193 #endif /* INET6 */
1194 
1195 	switch (sa->sa_family) {
1196 
1197 	case AF_UNSPEC:
1198 		memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN);
1199 		memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1200 		break;
1201 
1202 #ifdef INET
1203 	case AF_INET:
1204 		sin = satocsin(sa);
1205 		if (sin->sin_addr.s_addr == INADDR_ANY) {
1206 			/*
1207 			 * An IP address of INADDR_ANY means listen to
1208 			 * or stop listening to all of the Ethernet
1209 			 * multicast addresses used for IP.
1210 			 * (This is for the sake of IP multicast routers.)
1211 			 */
1212 			memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN);
1213 			memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN);
1214 		}
1215 		else {
1216 			ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo);
1217 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1218 		}
1219 		break;
1220 #endif
1221 #ifdef INET6
1222 	case AF_INET6:
1223 		sin6 = satocsin6(sa);
1224 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1225 			/*
1226 			 * An IP6 address of 0 means listen to or stop
1227 			 * listening to all of the Ethernet multicast
1228 			 * address used for IP6.
1229 			 * (This is used for multicast routers.)
1230 			 */
1231 			memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN);
1232 			memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN);
1233 		} else {
1234 			ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo);
1235 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1236 		}
1237 		break;
1238 #endif
1239 
1240 	default:
1241 		return EAFNOSUPPORT;
1242 	}
1243 	return 0;
1244 }
1245 
1246 /*
1247  * Add an Ethernet multicast address or range of addresses to the list for a
1248  * given interface.
1249  */
1250 int
1251 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec)
1252 {
1253 	struct ether_multi *enm;
1254 	u_char addrlo[ETHER_ADDR_LEN];
1255 	u_char addrhi[ETHER_ADDR_LEN];
1256 	int s = splnet(), error;
1257 
1258 	error = ether_multiaddr(sa, addrlo, addrhi);
1259 	if (error != 0) {
1260 		splx(s);
1261 		return error;
1262 	}
1263 
1264 	/*
1265 	 * Verify that we have valid Ethernet multicast addresses.
1266 	 */
1267 	if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) {
1268 		splx(s);
1269 		return EINVAL;
1270 	}
1271 	/*
1272 	 * See if the address range is already in the list.
1273 	 */
1274 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1275 	if (enm != NULL) {
1276 		/*
1277 		 * Found it; just increment the reference count.
1278 		 */
1279 		++enm->enm_refcount;
1280 		splx(s);
1281 		return 0;
1282 	}
1283 	/*
1284 	 * New address or range; malloc a new multicast record
1285 	 * and link it into the interface's multicast list.
1286 	 */
1287 	enm = (struct ether_multi *)malloc(sizeof(*enm), M_IFMADDR, M_NOWAIT);
1288 	if (enm == NULL) {
1289 		splx(s);
1290 		return ENOBUFS;
1291 	}
1292 	memcpy(enm->enm_addrlo, addrlo, 6);
1293 	memcpy(enm->enm_addrhi, addrhi, 6);
1294 	enm->enm_refcount = 1;
1295 	LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list);
1296 	ec->ec_multicnt++;
1297 	splx(s);
1298 	/*
1299 	 * Return ENETRESET to inform the driver that the list has changed
1300 	 * and its reception filter should be adjusted accordingly.
1301 	 */
1302 	return ENETRESET;
1303 }
1304 
1305 /*
1306  * Delete a multicast address record.
1307  */
1308 int
1309 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec)
1310 {
1311 	struct ether_multi *enm;
1312 	u_char addrlo[ETHER_ADDR_LEN];
1313 	u_char addrhi[ETHER_ADDR_LEN];
1314 	int s = splnet(), error;
1315 
1316 	error = ether_multiaddr(sa, addrlo, addrhi);
1317 	if (error != 0) {
1318 		splx(s);
1319 		return (error);
1320 	}
1321 
1322 	/*
1323 	 * Look ur the address in our list.
1324 	 */
1325 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1326 	if (enm == NULL) {
1327 		splx(s);
1328 		return (ENXIO);
1329 	}
1330 	if (--enm->enm_refcount != 0) {
1331 		/*
1332 		 * Still some claims to this record.
1333 		 */
1334 		splx(s);
1335 		return (0);
1336 	}
1337 	/*
1338 	 * No remaining claims to this record; unlink and free it.
1339 	 */
1340 	LIST_REMOVE(enm, enm_list);
1341 	free(enm, M_IFMADDR);
1342 	ec->ec_multicnt--;
1343 	splx(s);
1344 	/*
1345 	 * Return ENETRESET to inform the driver that the list has changed
1346 	 * and its reception filter should be adjusted accordingly.
1347 	 */
1348 	return (ENETRESET);
1349 }
1350 
1351 void
1352 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb)
1353 {
1354 	ec->ec_ifflags_cb = cb;
1355 }
1356 
1357 /*
1358  * Common ioctls for Ethernet interfaces.  Note, we must be
1359  * called at splnet().
1360  */
1361 int
1362 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data)
1363 {
1364 	struct ethercom *ec = (void *) ifp;
1365 	struct eccapreq *eccr;
1366 	struct ifreq *ifr = (struct ifreq *)data;
1367 	struct if_laddrreq *iflr = data;
1368 	const struct sockaddr_dl *sdl;
1369 	static const uint8_t zero[ETHER_ADDR_LEN];
1370 	int error;
1371 
1372 	switch (cmd) {
1373 	case SIOCINITIFADDR:
1374 	    {
1375 		struct ifaddr *ifa = (struct ifaddr *)data;
1376 		if (ifa->ifa_addr->sa_family != AF_LINK
1377 		    && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) !=
1378 		       (IFF_UP|IFF_RUNNING)) {
1379 			ifp->if_flags |= IFF_UP;
1380 			if ((error = (*ifp->if_init)(ifp)) != 0)
1381 				return error;
1382 		}
1383 #ifdef INET
1384 		if (ifa->ifa_addr->sa_family == AF_INET)
1385 			arp_ifinit(ifp, ifa);
1386 #endif /* INET */
1387 		return 0;
1388 	    }
1389 
1390 	case SIOCSIFMTU:
1391 	    {
1392 		int maxmtu;
1393 
1394 		if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU)
1395 			maxmtu = ETHERMTU_JUMBO;
1396 		else
1397 			maxmtu = ETHERMTU;
1398 
1399 		if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu)
1400 			return EINVAL;
1401 		else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET)
1402 			return error;
1403 		else if (ifp->if_flags & IFF_UP) {
1404 			/* Make sure the device notices the MTU change. */
1405 			return (*ifp->if_init)(ifp);
1406 		} else
1407 			return 0;
1408 	    }
1409 
1410 	case SIOCSIFFLAGS:
1411 		if ((error = ifioctl_common(ifp, cmd, data)) != 0)
1412 			return error;
1413 		switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) {
1414 		case IFF_RUNNING:
1415 			/*
1416 			 * If interface is marked down and it is running,
1417 			 * then stop and disable it.
1418 			 */
1419 			(*ifp->if_stop)(ifp, 1);
1420 			break;
1421 		case IFF_UP:
1422 			/*
1423 			 * If interface is marked up and it is stopped, then
1424 			 * start it.
1425 			 */
1426 			return (*ifp->if_init)(ifp);
1427 		case IFF_UP|IFF_RUNNING:
1428 			error = 0;
1429 			if (ec->ec_ifflags_cb == NULL ||
1430 			    (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) {
1431 				/*
1432 				 * Reset the interface to pick up
1433 				 * changes in any other flags that
1434 				 * affect the hardware state.
1435 				 */
1436 				return (*ifp->if_init)(ifp);
1437 			} else
1438 				return error;
1439 		case 0:
1440 			break;
1441 		}
1442 		return 0;
1443 	case SIOCGETHERCAP:
1444 		eccr = (struct eccapreq *)data;
1445 		eccr->eccr_capabilities = ec->ec_capabilities;
1446 		eccr->eccr_capenable = ec->ec_capenable;
1447 		return 0;
1448 	case SIOCADDMULTI:
1449 		return ether_addmulti(ifreq_getaddr(cmd, ifr), ec);
1450 	case SIOCDELMULTI:
1451 		return ether_delmulti(ifreq_getaddr(cmd, ifr), ec);
1452 	case SIOCSIFMEDIA:
1453 	case SIOCGIFMEDIA:
1454 		if (ec->ec_mii == NULL)
1455 			return ENOTTY;
1456 		return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd);
1457 	case SIOCALIFADDR:
1458 		sdl = satocsdl(sstocsa(&iflr->addr));
1459 		if (sdl->sdl_family != AF_LINK)
1460 			;
1461 		else if (ETHER_IS_MULTICAST(CLLADDR(sdl)))
1462 			return EINVAL;
1463 		else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0)
1464 			return EINVAL;
1465 		/*FALLTHROUGH*/
1466 	default:
1467 		return ifioctl_common(ifp, cmd, data);
1468 	}
1469 	return 0;
1470 }
1471