xref: /netbsd-src/sys/net/if_ethersubr.c (revision ba65fde2d7fefa7d39838fa5fa855e62bd606b5e)
1 /*	$NetBSD: if_ethersubr.c,v 1.193 2012/10/31 10:17:34 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 1.193 2012/10/31 10:17:34 msaitoh Exp $");
65 
66 #include "opt_inet.h"
67 #include "opt_atalk.h"
68 #include "opt_iso.h"
69 #include "opt_ipx.h"
70 #include "opt_mbuftrace.h"
71 #include "opt_mpls.h"
72 #include "opt_gateway.h"
73 #include "opt_pfil_hooks.h"
74 #include "opt_pppoe.h"
75 #include "vlan.h"
76 #include "pppoe.h"
77 #include "bridge.h"
78 #include "arp.h"
79 #include "agr.h"
80 
81 #include <sys/param.h>
82 #include <sys/systm.h>
83 #include <sys/kernel.h>
84 #include <sys/callout.h>
85 #include <sys/malloc.h>
86 #include <sys/mbuf.h>
87 #include <sys/protosw.h>
88 #include <sys/socket.h>
89 #include <sys/ioctl.h>
90 #include <sys/errno.h>
91 #include <sys/syslog.h>
92 #include <sys/kauth.h>
93 #include <sys/cpu.h>
94 #include <sys/intr.h>
95 #include <sys/device.h>
96 
97 #include <net/if.h>
98 #include <net/netisr.h>
99 #include <net/route.h>
100 #include <net/if_llc.h>
101 #include <net/if_dl.h>
102 #include <net/if_types.h>
103 
104 #include <net/if_media.h>
105 #include <dev/mii/mii.h>
106 #include <dev/mii/miivar.h>
107 
108 #if NARP == 0
109 /*
110  * XXX there should really be a way to issue this warning from within config(8)
111  */
112 #error You have included NETATALK or a pseudo-device in your configuration that depends on the presence of ethernet interfaces, but have no such interfaces configured. Check if you really need pseudo-device bridge, pppoe, vlan or options NETATALK.
113 #endif
114 
115 #include <net/bpf.h>
116 
117 #include <net/if_ether.h>
118 #include <net/if_vlanvar.h>
119 
120 #if NPPPOE > 0
121 #include <net/if_pppoe.h>
122 #endif
123 
124 #if NAGR > 0
125 #include <net/agr/ieee8023_slowprotocols.h>	/* XXX */
126 #include <net/agr/ieee8023ad.h>
127 #include <net/agr/if_agrvar.h>
128 #endif
129 
130 #if NBRIDGE > 0
131 #include <net/if_bridgevar.h>
132 #endif
133 
134 #include <netinet/in.h>
135 #ifdef INET
136 #include <netinet/in_var.h>
137 #endif
138 #include <netinet/if_inarp.h>
139 
140 #ifdef INET6
141 #ifndef INET
142 #include <netinet/in.h>
143 #endif
144 #include <netinet6/in6_var.h>
145 #include <netinet6/nd6.h>
146 #endif
147 
148 
149 #include "carp.h"
150 #if NCARP > 0
151 #include <netinet/ip_carp.h>
152 #endif
153 
154 #ifdef IPX
155 #include <netipx/ipx.h>
156 #include <netipx/ipx_if.h>
157 #endif
158 
159 #ifdef ISO
160 #include <netiso/argo_debug.h>
161 #include <netiso/iso.h>
162 #include <netiso/iso_var.h>
163 #include <netiso/iso_snpac.h>
164 #endif
165 
166 
167 
168 #ifdef NETATALK
169 #include <netatalk/at.h>
170 #include <netatalk/at_var.h>
171 #include <netatalk/at_extern.h>
172 
173 #define llc_snap_org_code llc_un.type_snap.org_code
174 #define llc_snap_ether_type llc_un.type_snap.ether_type
175 
176 extern u_char	at_org_code[3];
177 extern u_char	aarp_org_code[3];
178 #endif /* NETATALK */
179 
180 #ifdef MPLS
181 #include <netmpls/mpls.h>
182 #include <netmpls/mpls_var.h>
183 #endif
184 
185 static struct timeval bigpktppslim_last;
186 static int bigpktppslim = 2;	/* XXX */
187 static int bigpktpps_count;
188 
189 
190 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] =
191     { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
192 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] =
193     { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 };
194 #define senderr(e) { error = (e); goto bad;}
195 
196 static	int ether_output(struct ifnet *, struct mbuf *,
197 	    const struct sockaddr *, struct rtentry *);
198 
199 /*
200  * Ethernet output routine.
201  * Encapsulate a packet of type family for the local net.
202  * Assumes that ifp is actually pointer to ethercom structure.
203  */
204 static int
205 ether_output(struct ifnet * const ifp0, struct mbuf * const m0,
206 	const struct sockaddr * const dst,
207 	struct rtentry *rt0)
208 {
209 	uint16_t etype = 0;
210 	int error = 0, hdrcmplt = 0;
211  	uint8_t esrc[6], edst[6];
212 	struct mbuf *m = m0;
213 	struct rtentry *rt;
214 	struct mbuf *mcopy = NULL;
215 	struct ether_header *eh;
216 	struct ifnet *ifp = ifp0;
217 	ALTQ_DECL(struct altq_pktattr pktattr;)
218 #ifdef INET
219 	struct arphdr *ah;
220 #endif /* INET */
221 #ifdef NETATALK
222 	struct at_ifaddr *aa;
223 #endif /* NETATALK */
224 
225 #ifdef MBUFTRACE
226 	m_claimm(m, ifp->if_mowner);
227 #endif
228 
229 #if NCARP > 0
230 	if (ifp->if_type == IFT_CARP) {
231 		struct ifaddr *ifa;
232 
233 		/* loop back if this is going to the carp interface */
234 		if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP &&
235 		    (ifa = ifa_ifwithaddr(dst)) != NULL &&
236 		    ifa->ifa_ifp == ifp0)
237 			return looutput(ifp0, m, dst, rt0);
238 
239 		ifp = ifp->if_carpdev;
240 		/* ac = (struct arpcom *)ifp; */
241 
242 		if ((ifp0->if_flags & (IFF_UP|IFF_RUNNING)) !=
243 		    (IFF_UP|IFF_RUNNING))
244 			senderr(ENETDOWN);
245 	}
246 #endif /* NCARP > 0 */
247 
248 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
249 		senderr(ENETDOWN);
250 	if ((rt = rt0) != NULL) {
251 		if ((rt->rt_flags & RTF_UP) == 0) {
252 			if ((rt0 = rt = rtalloc1(dst, 1)) != NULL) {
253 				rt->rt_refcnt--;
254 				if (rt->rt_ifp != ifp)
255 					return (*rt->rt_ifp->if_output)
256 							(ifp, m0, dst, rt);
257 			} else
258 				senderr(EHOSTUNREACH);
259 		}
260 		if ((rt->rt_flags & RTF_GATEWAY) && dst->sa_family != AF_NS) {
261 			if (rt->rt_gwroute == NULL)
262 				goto lookup;
263 			if (((rt = rt->rt_gwroute)->rt_flags & RTF_UP) == 0) {
264 				rtfree(rt); rt = rt0;
265 			lookup: rt->rt_gwroute = rtalloc1(rt->rt_gateway, 1);
266 				if ((rt = rt->rt_gwroute) == NULL)
267 					senderr(EHOSTUNREACH);
268 				/* the "G" test below also prevents rt == rt0 */
269 				if ((rt->rt_flags & RTF_GATEWAY) ||
270 				    (rt->rt_ifp != ifp)) {
271 					rt->rt_refcnt--;
272 					rt0->rt_gwroute = NULL;
273 					senderr(EHOSTUNREACH);
274 				}
275 			}
276 		}
277 		if (rt->rt_flags & RTF_REJECT)
278 			if (rt->rt_rmx.rmx_expire == 0 ||
279 			    (u_long) time_second < rt->rt_rmx.rmx_expire)
280 				senderr(rt == rt0 ? EHOSTDOWN : EHOSTUNREACH);
281 	}
282 
283 	switch (dst->sa_family) {
284 
285 #ifdef INET
286 	case AF_INET:
287 		if (m->m_flags & M_BCAST)
288 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
289 		else if (m->m_flags & M_MCAST)
290 			ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst);
291 		else if (!arpresolve(ifp, rt, m, dst, edst))
292 			return (0);	/* if not yet resolved */
293 		/* If broadcasting on a simplex interface, loopback a copy */
294 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
295 			mcopy = m_copy(m, 0, (int)M_COPYALL);
296 		etype = htons(ETHERTYPE_IP);
297 		break;
298 
299 	case AF_ARP:
300 		ah = mtod(m, struct arphdr *);
301 		if (m->m_flags & M_BCAST)
302 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
303 		else {
304 			void *tha = ar_tha(ah);
305 
306 			if (tha == NULL) {
307 				/* fake with ARPHDR_IEEE1394 */
308 				return 0;
309 			}
310 			memcpy(edst, tha, sizeof(edst));
311 		}
312 
313 		ah->ar_hrd = htons(ARPHRD_ETHER);
314 
315 		switch (ntohs(ah->ar_op)) {
316 		case ARPOP_REVREQUEST:
317 		case ARPOP_REVREPLY:
318 			etype = htons(ETHERTYPE_REVARP);
319 			break;
320 
321 		case ARPOP_REQUEST:
322 		case ARPOP_REPLY:
323 		default:
324 			etype = htons(ETHERTYPE_ARP);
325 		}
326 
327 		break;
328 #endif
329 #ifdef INET6
330 	case AF_INET6:
331 		if (!nd6_storelladdr(ifp, rt, m, dst, edst, sizeof(edst))){
332 			/* something bad happened */
333 			return (0);
334 		}
335 		etype = htons(ETHERTYPE_IPV6);
336 		break;
337 #endif
338 #ifdef NETATALK
339     case AF_APPLETALK:
340 		if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) {
341 #ifdef NETATALKDEBUG
342 			printf("aarpresolv failed\n");
343 #endif /* NETATALKDEBUG */
344 			return (0);
345 		}
346 		/*
347 		 * ifaddr is the first thing in at_ifaddr
348 		 */
349 		aa = (struct at_ifaddr *) at_ifawithnet(
350 		    (const struct sockaddr_at *)dst, ifp);
351 		if (aa == NULL)
352 		    goto bad;
353 
354 		/*
355 		 * In the phase 2 case, we need to prepend an mbuf for the
356 		 * llc header.  Since we must preserve the value of m,
357 		 * which is passed to us by value, we m_copy() the first
358 		 * mbuf, and use it for our llc header.
359 		 */
360 		if (aa->aa_flags & AFA_PHASE2) {
361 			struct llc llc;
362 
363 			M_PREPEND(m, sizeof(struct llc), M_DONTWAIT);
364 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
365 			llc.llc_control = LLC_UI;
366 			memcpy(llc.llc_snap_org_code, at_org_code,
367 			    sizeof(llc.llc_snap_org_code));
368 			llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK);
369 			memcpy(mtod(m, void *), &llc, sizeof(struct llc));
370 		} else {
371 			etype = htons(ETHERTYPE_ATALK);
372 		}
373 		break;
374 #endif /* NETATALK */
375 #ifdef IPX
376 	case AF_IPX:
377 		etype = htons(ETHERTYPE_IPX);
378  		memcpy(edst,
379 		    &(((const struct sockaddr_ipx *)dst)->sipx_addr.x_host),
380 		    sizeof(edst));
381 		/* If broadcasting on a simplex interface, loopback a copy */
382 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
383 			mcopy = m_copy(m, 0, (int)M_COPYALL);
384 		break;
385 #endif
386 #ifdef	ISO
387 	case AF_ISO: {
388 		int	snpalen;
389 		struct	llc *l;
390 		const struct sockaddr_dl *sdl;
391 
392 		if (rt && (sdl = satocsdl(rt->rt_gateway)) &&
393 		    sdl->sdl_family == AF_LINK && sdl->sdl_alen > 0) {
394 			memcpy(edst, CLLADDR(sdl), sizeof(edst));
395 		} else {
396 			error = iso_snparesolve(ifp,
397 			    (const struct sockaddr_iso *)dst,
398 						(char *)edst, &snpalen);
399 			if (error)
400 				goto bad; /* Not Resolved */
401 		}
402 		/* If broadcasting on a simplex interface, loopback a copy */
403 		if (*edst & 1)
404 			m->m_flags |= (M_BCAST|M_MCAST);
405 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX) &&
406 		    (mcopy = m_copy(m, 0, (int)M_COPYALL))) {
407 			M_PREPEND(mcopy, sizeof (*eh), M_DONTWAIT);
408 			if (mcopy) {
409 				eh = mtod(mcopy, struct ether_header *);
410 				memcpy(eh->ether_dhost, edst, sizeof(edst));
411 				memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl),
412 				    sizeof(edst));
413 			}
414 		}
415 		M_PREPEND(m, 3, M_DONTWAIT);
416 		if (m == NULL)
417 			return (0);
418 		l = mtod(m, struct llc *);
419 		l->llc_dsap = l->llc_ssap = LLC_ISO_LSAP;
420 		l->llc_control = LLC_UI;
421 #ifdef ARGO_DEBUG
422 		if (argo_debug[D_ETHER]) {
423 			int i;
424 			printf("unoutput: sending pkt to: ");
425 			for (i=0; i<6; i++)
426 				printf("%x ", edst[i] & 0xff);
427 			printf("\n");
428 		}
429 #endif
430 		} break;
431 #endif /* ISO */
432 
433 	case pseudo_AF_HDRCMPLT:
434 		hdrcmplt = 1;
435 		memcpy(esrc,
436 		    ((const struct ether_header *)dst->sa_data)->ether_shost,
437 		    sizeof(esrc));
438 		/* FALLTHROUGH */
439 
440 	case AF_UNSPEC:
441  		memcpy(edst,
442 		    ((const struct ether_header *)dst->sa_data)->ether_dhost,
443 		    sizeof(edst));
444 		/* AF_UNSPEC doesn't swap the byte order of the ether_type. */
445 		etype = ((const struct ether_header *)dst->sa_data)->ether_type;
446 		break;
447 
448 	default:
449 		printf("%s: can't handle af%d\n", ifp->if_xname,
450 			dst->sa_family);
451 		senderr(EAFNOSUPPORT);
452 	}
453 
454 #ifdef MPLS
455 	if (rt0 != NULL && rt_gettag(rt0) != NULL &&
456 	    rt_gettag(rt0)->sa_family == AF_MPLS &&
457 	    (m->m_flags & (M_MCAST | M_BCAST)) == 0) {
458 		union mpls_shim msh;
459 		msh.s_addr = MPLS_GETSADDR(rt0);
460 		if (msh.shim.label != MPLS_LABEL_IMPLNULL)
461 			etype = htons(ETHERTYPE_MPLS);
462 	}
463 #endif
464 
465 	if (mcopy)
466 		(void)looutput(ifp, mcopy, dst, rt);
467 
468 	/* If no ether type is set, this must be a 802.2 formatted packet.
469 	 */
470 	if (etype == 0)
471 		etype = htons(m->m_pkthdr.len);
472 	/*
473 	 * Add local net header.  If no space in first mbuf,
474 	 * allocate another.
475 	 */
476 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
477 	if (m == 0)
478 		senderr(ENOBUFS);
479 	eh = mtod(m, struct ether_header *);
480 	/* Note: etype is already in network byte order. */
481 	(void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
482  	memcpy(eh->ether_dhost, edst, sizeof(edst));
483 	if (hdrcmplt)
484 		memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost));
485 	else
486 	 	memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl),
487 		    sizeof(eh->ether_shost));
488 
489 #if NCARP > 0
490 	if (ifp0 != ifp && ifp0->if_type == IFT_CARP) {
491 	 	memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl),
492 		    sizeof(eh->ether_shost));
493 	}
494 #endif /* NCARP > 0 */
495 
496 #ifdef PFIL_HOOKS
497 	if ((error = pfil_run_hooks(&ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0)
498 		return (error);
499 	if (m == NULL)
500 		return (0);
501 #endif
502 
503 #if NBRIDGE > 0
504 	/*
505 	 * Bridges require special output handling.
506 	 */
507 	if (ifp->if_bridge)
508 		return (bridge_output(ifp, m, NULL, NULL));
509 #endif
510 
511 #if NCARP > 0
512 	if (ifp != ifp0)
513 		ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN;
514 #endif /* NCARP > 0 */
515 
516 #ifdef ALTQ
517 	/*
518 	 * If ALTQ is enabled on the parent interface, do
519 	 * classification; the queueing discipline might not
520 	 * require classification, but might require the
521 	 * address family/header pointer in the pktattr.
522 	 */
523 	if (ALTQ_IS_ENABLED(&ifp->if_snd))
524 		altq_etherclassify(&ifp->if_snd, m, &pktattr);
525 #endif
526 	return ifq_enqueue(ifp, m ALTQ_COMMA ALTQ_DECL(&pktattr));
527 
528 bad:
529 	if (m)
530 		m_freem(m);
531 	return (error);
532 }
533 
534 #ifdef ALTQ
535 /*
536  * This routine is a slight hack to allow a packet to be classified
537  * if the Ethernet headers are present.  It will go away when ALTQ's
538  * classification engine understands link headers.
539  */
540 void
541 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
542     struct altq_pktattr *pktattr)
543 {
544 	struct ether_header *eh;
545 	uint16_t ether_type;
546 	int hlen, af, hdrsize;
547 	void *hdr;
548 
549 	hlen = ETHER_HDR_LEN;
550 	eh = mtod(m, struct ether_header *);
551 
552 	ether_type = htons(eh->ether_type);
553 
554 	if (ether_type < ETHERMTU) {
555 		/* LLC/SNAP */
556 		struct llc *llc = (struct llc *)(eh + 1);
557 		hlen += 8;
558 
559 		if (m->m_len < hlen ||
560 		    llc->llc_dsap != LLC_SNAP_LSAP ||
561 		    llc->llc_ssap != LLC_SNAP_LSAP ||
562 		    llc->llc_control != LLC_UI) {
563 			/* Not SNAP. */
564 			goto bad;
565 		}
566 
567 		ether_type = htons(llc->llc_un.type_snap.ether_type);
568 	}
569 
570 	switch (ether_type) {
571 	case ETHERTYPE_IP:
572 		af = AF_INET;
573 		hdrsize = 20;		/* sizeof(struct ip) */
574 		break;
575 
576 	case ETHERTYPE_IPV6:
577 		af = AF_INET6;
578 		hdrsize = 40;		/* sizeof(struct ip6_hdr) */
579 		break;
580 
581 	default:
582 		af = AF_UNSPEC;
583 		hdrsize = 0;
584 		break;
585 	}
586 
587 	while (m->m_len <= hlen) {
588 		hlen -= m->m_len;
589 		m = m->m_next;
590 	}
591 	if (m->m_len < (hlen + hdrsize)) {
592 		/*
593 		 * protocol header not in a single mbuf.
594 		 * We can't cope with this situation right
595 		 * now (but it shouldn't ever happen, really, anyhow).
596 		 */
597 #ifdef DEBUG
598 		printf("altq_etherclassify: headers span multiple mbufs: "
599 		    "%d < %d\n", m->m_len, (hlen + hdrsize));
600 #endif
601 		goto bad;
602 	}
603 
604 	m->m_data += hlen;
605 	m->m_len -= hlen;
606 
607 	hdr = mtod(m, void *);
608 
609 	if (ALTQ_NEEDS_CLASSIFY(ifq))
610 		pktattr->pattr_class =
611 		    (*ifq->altq_classify)(ifq->altq_clfier, m, af);
612 	pktattr->pattr_af = af;
613 	pktattr->pattr_hdr = hdr;
614 
615 	m->m_data -= hlen;
616 	m->m_len += hlen;
617 
618 	return;
619 
620  bad:
621 	pktattr->pattr_class = NULL;
622 	pktattr->pattr_hdr = NULL;
623 	pktattr->pattr_af = AF_UNSPEC;
624 }
625 #endif /* ALTQ */
626 
627 /*
628  * Process a received Ethernet packet;
629  * the packet is in the mbuf chain m with
630  * the ether header.
631  */
632 void
633 ether_input(struct ifnet *ifp, struct mbuf *m)
634 {
635 	struct ethercom *ec = (struct ethercom *) ifp;
636 	struct ifqueue *inq;
637 	uint16_t etype;
638 	struct ether_header *eh;
639 	size_t ehlen;
640 #if defined (ISO) || defined (LLC) || defined(NETATALK)
641 	struct llc *l;
642 #endif
643 
644 	if ((ifp->if_flags & IFF_UP) == 0) {
645 		m_freem(m);
646 		return;
647 	}
648 
649 #ifdef MBUFTRACE
650 	m_claimm(m, &ec->ec_rx_mowner);
651 #endif
652 	eh = mtod(m, struct ether_header *);
653 	etype = ntohs(eh->ether_type);
654 	ehlen = sizeof(*eh);
655 
656 	/*
657 	 * Determine if the packet is within its size limits.
658 	 */
659 	if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len >
660 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
661 		if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count,
662 			    bigpktppslim)) {
663 			printf("%s: discarding oversize frame (len=%d)\n",
664 			    ifp->if_xname, m->m_pkthdr.len);
665 		}
666 		m_freem(m);
667 		return;
668 	}
669 
670 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
671 		/*
672 		 * If this is not a simplex interface, drop the packet
673 		 * if it came from us.
674 		 */
675 		if ((ifp->if_flags & IFF_SIMPLEX) == 0 &&
676 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost,
677 		    ETHER_ADDR_LEN) == 0) {
678 			m_freem(m);
679 			return;
680 		}
681 
682 		if (memcmp(etherbroadcastaddr,
683 		    eh->ether_dhost, ETHER_ADDR_LEN) == 0)
684 			m->m_flags |= M_BCAST;
685 		else
686 			m->m_flags |= M_MCAST;
687 		ifp->if_imcasts++;
688 	}
689 
690 	/* If the CRC is still on the packet, trim it off. */
691 	if (m->m_flags & M_HASFCS) {
692 		m_adj(m, -ETHER_CRC_LEN);
693 		m->m_flags &= ~M_HASFCS;
694 	}
695 
696 	ifp->if_ibytes += m->m_pkthdr.len;
697 
698 #if NBRIDGE > 0
699 	/*
700 	 * Tap the packet off here for a bridge.  bridge_input()
701 	 * will return NULL if it has consumed the packet, otherwise
702 	 * it gets processed as normal.  Note that bridge_input()
703 	 * will always return the original packet if we need to
704 	 * process it locally.
705 	 */
706 	if (ifp->if_bridge) {
707 		/* clear M_PROMISC, in case the packets comes from a vlan */
708 		m->m_flags &= ~M_PROMISC;
709 		m = bridge_input(ifp, m);
710 		if (m == NULL)
711 			return;
712 
713 		/*
714 		 * Bridge has determined that the packet is for us.
715 		 * Update our interface pointer -- we may have had
716 		 * to "bridge" the packet locally.
717 		 */
718 		ifp = m->m_pkthdr.rcvif;
719 	} else
720 #endif /* NBRIDGE > 0 */
721 	{
722 
723 #if NCARP > 0
724 		if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) {
725 			/*
726 			 * clear M_PROMISC, in case the packets comes from a
727 			 * vlan
728 			 */
729 			m->m_flags &= ~M_PROMISC;
730 			if (carp_input(m, (uint8_t *)&eh->ether_shost,
731 			    (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0)
732 				return;
733 		}
734 #endif /* NCARP > 0 */
735 		if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 &&
736 		    (ifp->if_flags & IFF_PROMISC) != 0 &&
737 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost,
738 			   ETHER_ADDR_LEN) != 0) {
739 			m->m_flags |= M_PROMISC;
740 		}
741 	}
742 
743 #ifdef PFIL_HOOKS
744 	if ((m->m_flags & M_PROMISC) == 0) {
745 		if (pfil_run_hooks(&ifp->if_pfil, &m, ifp, PFIL_IN) != 0)
746 			return;
747 		if (m == NULL)
748 			return;
749 
750 		eh = mtod(m, struct ether_header *);
751 		etype = ntohs(eh->ether_type);
752 		ehlen = sizeof(*eh);
753 	}
754 #endif
755 
756 #if NAGR > 0
757 	if (ifp->if_agrprivate &&
758 	    __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) {
759 		m->m_flags &= ~M_PROMISC;
760 		agr_input(ifp, m);
761 		return;
762 	}
763 #endif /* NAGR > 0 */
764 
765 	/*
766 	 * If VLANs are configured on the interface, check to
767 	 * see if the device performed the decapsulation and
768 	 * provided us with the tag.
769 	 */
770 	if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) {
771 #if NVLAN > 0
772 		/*
773 		 * vlan_input() will either recursively call ether_input()
774 		 * or drop the packet.
775 		 */
776 		vlan_input(ifp, m);
777 #else
778 		m_freem(m);
779 #endif
780 		return;
781 	}
782 
783 	/*
784 	 * Handle protocols that expect to have the Ethernet header
785 	 * (and possibly FCS) intact.
786 	 */
787 	switch (etype) {
788 	case ETHERTYPE_VLAN: {
789 		struct ether_vlan_header *evl = (void *)eh;
790 		/*
791 		 * If there is a tag of 0, then the VLAN header was probably
792 		 * just being used to store the priority.  Extract the ether
793 		 * type, and if IP or IPV6, let them deal with it.
794 		 */
795 		if (m->m_len <= sizeof(*evl)
796 		    && EVL_VLANOFTAG(evl->evl_tag) == 0) {
797 			etype = ntohs(evl->evl_proto);
798 			ehlen = sizeof(*evl);
799 			if ((m->m_flags & M_PROMISC) == 0
800 			    && (etype == ETHERTYPE_IP
801 				|| etype == ETHERTYPE_IPV6))
802 				break;
803 		}
804 #if NVLAN > 0
805 		/*
806 		 * vlan_input() will either recursively call ether_input()
807 		 * or drop the packet.
808 		 */
809 		if (((struct ethercom *)ifp)->ec_nvlans != 0)
810 			vlan_input(ifp, m);
811 		else
812 #endif /* NVLAN > 0 */
813 			m_freem(m);
814 		return;
815 	}
816 #if NPPPOE > 0
817 	case ETHERTYPE_PPPOEDISC:
818 	case ETHERTYPE_PPPOE:
819 		if (m->m_flags & M_PROMISC) {
820 			m_freem(m);
821 			return;
822 		}
823 #ifndef PPPOE_SERVER
824 		if (m->m_flags & (M_MCAST | M_BCAST)) {
825 			m_freem(m);
826 			return;
827 		}
828 #endif
829 
830 		if (etype == ETHERTYPE_PPPOEDISC)
831 			inq = &ppoediscinq;
832 		else
833 			inq = &ppoeinq;
834 		if (IF_QFULL(inq)) {
835 			IF_DROP(inq);
836 			m_freem(m);
837 		} else
838 			IF_ENQUEUE(inq, m);
839 		softint_schedule(pppoe_softintr);
840 		return;
841 #endif /* NPPPOE > 0 */
842 	case ETHERTYPE_SLOWPROTOCOLS: {
843 		uint8_t subtype;
844 
845 #if defined(DIAGNOSTIC)
846 		if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) {
847 			panic("ether_input: too short slow protocol packet");
848 		}
849 #endif
850 		m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype);
851 		switch (subtype) {
852 #if NAGR > 0
853 		case SLOWPROTOCOLS_SUBTYPE_LACP:
854 			if (ifp->if_agrprivate) {
855 				ieee8023ad_lacp_input(ifp, m);
856 				return;
857 			}
858 			break;
859 
860 		case SLOWPROTOCOLS_SUBTYPE_MARKER:
861 			if (ifp->if_agrprivate) {
862 				ieee8023ad_marker_input(ifp, m);
863 				return;
864 			}
865 			break;
866 #endif /* NAGR > 0 */
867 		default:
868 			if (subtype == 0 || subtype > 10) {
869 				/* illegal value */
870 				m_freem(m);
871 				return;
872 			}
873 			/* unknown subtype */
874 			break;
875 		}
876 		/* FALLTHROUGH */
877 	}
878 	default:
879 		if (m->m_flags & M_PROMISC) {
880 			m_freem(m);
881 			return;
882 		}
883 	}
884 
885 	/* If the CRC is still on the packet, trim it off. */
886 	if (m->m_flags & M_HASFCS) {
887 		m_adj(m, -ETHER_CRC_LEN);
888 		m->m_flags &= ~M_HASFCS;
889 	}
890 
891 	if (etype > ETHERMTU + sizeof (struct ether_header)) {
892 		/* Strip off the Ethernet header. */
893 		m_adj(m, ehlen);
894 
895 		switch (etype) {
896 #ifdef INET
897 		case ETHERTYPE_IP:
898 #ifdef GATEWAY
899 			if (ipflow_fastforward(m))
900 				return;
901 #endif
902 			schednetisr(NETISR_IP);
903 			inq = &ipintrq;
904 			break;
905 
906 		case ETHERTYPE_ARP:
907 			schednetisr(NETISR_ARP);
908 			inq = &arpintrq;
909 			break;
910 
911 		case ETHERTYPE_REVARP:
912 			revarpinput(m);	/* XXX queue? */
913 			return;
914 #endif
915 #ifdef INET6
916 		case ETHERTYPE_IPV6:
917 #ifdef GATEWAY
918 			if (ip6flow_fastforward(&m))
919 				return;
920 #endif
921 			schednetisr(NETISR_IPV6);
922 			inq = &ip6intrq;
923 			break;
924 #endif
925 #ifdef IPX
926 		case ETHERTYPE_IPX:
927 			schednetisr(NETISR_IPX);
928 			inq = &ipxintrq;
929 			break;
930 #endif
931 #ifdef NETATALK
932 		case ETHERTYPE_ATALK:
933 			schednetisr(NETISR_ATALK);
934 			inq = &atintrq1;
935 			break;
936 		case ETHERTYPE_AARP:
937 			/* probably this should be done with a NETISR as well */
938 			aarpinput(ifp, m); /* XXX */
939 			return;
940 #endif /* NETATALK */
941 #ifdef MPLS
942 		case ETHERTYPE_MPLS:
943 			schednetisr(NETISR_MPLS);
944 			inq = &mplsintrq;
945 			break;
946 #endif
947 		default:
948 			m_freem(m);
949 			return;
950 		}
951 	} else {
952 #if defined (ISO) || defined (LLC) || defined (NETATALK)
953 		l = (struct llc *)(eh+1);
954 		switch (l->llc_dsap) {
955 #ifdef NETATALK
956 		case LLC_SNAP_LSAP:
957 			switch (l->llc_control) {
958 			case LLC_UI:
959 				if (l->llc_ssap != LLC_SNAP_LSAP) {
960 					goto dropanyway;
961 				}
962 
963 				if (memcmp(&(l->llc_snap_org_code)[0],
964 				    at_org_code, sizeof(at_org_code)) == 0 &&
965 				    ntohs(l->llc_snap_ether_type) ==
966 				    ETHERTYPE_ATALK) {
967 					inq = &atintrq2;
968 					m_adj(m, sizeof(struct ether_header)
969 					    + sizeof(struct llc));
970 					schednetisr(NETISR_ATALK);
971 					break;
972 				}
973 
974 				if (memcmp(&(l->llc_snap_org_code)[0],
975 				    aarp_org_code,
976 				    sizeof(aarp_org_code)) == 0 &&
977 				    ntohs(l->llc_snap_ether_type) ==
978 				    ETHERTYPE_AARP) {
979 					m_adj( m, sizeof(struct ether_header)
980 					    + sizeof(struct llc));
981 					aarpinput(ifp, m); /* XXX */
982 				    return;
983 				}
984 
985 			default:
986 				goto dropanyway;
987 			}
988 			break;
989 #endif /* NETATALK */
990 #ifdef	ISO
991 		case LLC_ISO_LSAP:
992 			switch (l->llc_control) {
993 			case LLC_UI:
994 				/* LLC_UI_P forbidden in class 1 service */
995 				if ((l->llc_dsap == LLC_ISO_LSAP) &&	/* XXX? case tested */
996 				    (l->llc_ssap == LLC_ISO_LSAP)) {
997 					/* LSAP for ISO */
998 					/* XXX length computation?? */
999 					if (m->m_pkthdr.len > etype + sizeof(struct ether_header))
1000 						m_adj(m, etype - m->m_pkthdr.len);
1001 
1002 #ifdef ARGO_DEBUG
1003 					if (argo_debug[D_ETHER])
1004 						printf("clnp packet");
1005 #endif
1006 					schednetisr(NETISR_ISO);
1007 					inq = &clnlintrq;
1008 					break;
1009 				}
1010 				goto dropanyway;
1011 
1012 			case LLC_XID:
1013 			case LLC_XID_P:
1014 				if(m->m_len < LLC_XID_BASIC_MINLEN + sizeof(struct ether_header))
1015 					/* XXX m_pullup? */
1016 					goto dropanyway;
1017 				l->llc_window = 0;
1018 				l->llc_fid = LLC_XID_FORMAT_BASIC;
1019 				l->llc_class = LLC_XID_CLASS_I;
1020 				l->llc_dsap = l->llc_ssap = 0;
1021 				/* Fall through to */
1022 			case LLC_TEST:
1023 			case LLC_TEST_P:
1024 			{
1025 				struct sockaddr sa;
1026 				struct ether_header *eh2;
1027 				int i;
1028 				u_char c = l->llc_dsap;
1029 
1030 				l->llc_dsap = l->llc_ssap;
1031 				l->llc_ssap = c;
1032 				m_adj(m, sizeof(struct ether_header));
1033 				/* XXX we can optimize here? */
1034 				if (m->m_flags & (M_BCAST | M_MCAST))
1035 					memcpy(eh->ether_dhost,
1036 					    CLLADDR(ifp->if_sadl),
1037 					    ETHER_ADDR_LEN);
1038 				sa.sa_family = AF_UNSPEC;
1039 				sa.sa_len = sizeof(sa);
1040 				eh2 = (struct ether_header *)sa.sa_data;
1041 				for (i = 0; i < 6; i++) {
1042 					eh2->ether_shost[i] = c =
1043 					    eh->ether_dhost[i];
1044 					eh2->ether_dhost[i] =
1045 					    eh->ether_dhost[i] =
1046 					    eh->ether_shost[i];
1047 					eh->ether_shost[i] = c;
1048 				}
1049 				ifp->if_output(ifp, m, &sa, NULL);
1050 				return;
1051 			}
1052 			default:
1053 				m_freem(m);
1054 				return;
1055 			}
1056 			break;
1057 #endif /* ISO */
1058 #if defined (ISO) || defined (NETATALK)
1059 		dropanyway:
1060 #endif
1061 		default:
1062 			m_freem(m);
1063 			return;
1064 		}
1065 #else /* ISO || LLC || NETATALK*/
1066 		m_freem(m);
1067 		return;
1068 #endif /* ISO || LLC || NETATALK*/
1069 	}
1070 
1071 	if (IF_QFULL(inq)) {
1072 		IF_DROP(inq);
1073 		m_freem(m);
1074 	} else
1075 		IF_ENQUEUE(inq, m);
1076 }
1077 
1078 /*
1079  * Convert Ethernet address to printable (loggable) representation.
1080  */
1081 char *
1082 ether_sprintf(const u_char *ap)
1083 {
1084 	static char etherbuf[3 * ETHER_ADDR_LEN];
1085 	return ether_snprintf(etherbuf, sizeof(etherbuf), ap);
1086 }
1087 
1088 char *
1089 ether_snprintf(char *buf, size_t len, const u_char *ap)
1090 {
1091 	char *cp = buf;
1092 	size_t i;
1093 
1094 	for (i = 0; i < len / 3; i++) {
1095 		*cp++ = hexdigits[*ap >> 4];
1096 		*cp++ = hexdigits[*ap++ & 0xf];
1097 		*cp++ = ':';
1098 	}
1099 	*--cp = '\0';
1100 	return buf;
1101 }
1102 
1103 /*
1104  * Perform common duties while attaching to interface list
1105  */
1106 void
1107 ether_ifattach(struct ifnet *ifp, const uint8_t *lla)
1108 {
1109 	struct ethercom *ec = (struct ethercom *)ifp;
1110 
1111 	ifp->if_type = IFT_ETHER;
1112 	ifp->if_hdrlen = ETHER_HDR_LEN;
1113 	ifp->if_dlt = DLT_EN10MB;
1114 	ifp->if_mtu = ETHERMTU;
1115 	ifp->if_output = ether_output;
1116 	ifp->if_input = ether_input;
1117 	if (ifp->if_baudrate == 0)
1118 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
1119 
1120 	if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla));
1121 
1122 	LIST_INIT(&ec->ec_multiaddrs);
1123 	ifp->if_broadcastaddr = etherbroadcastaddr;
1124 	bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header));
1125 #ifdef MBUFTRACE
1126 	strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname,
1127 	    sizeof(ec->ec_tx_mowner.mo_name));
1128 	strlcpy(ec->ec_tx_mowner.mo_descr, "tx",
1129 	    sizeof(ec->ec_tx_mowner.mo_descr));
1130 	strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname,
1131 	    sizeof(ec->ec_rx_mowner.mo_name));
1132 	strlcpy(ec->ec_rx_mowner.mo_descr, "rx",
1133 	    sizeof(ec->ec_rx_mowner.mo_descr));
1134 	MOWNER_ATTACH(&ec->ec_tx_mowner);
1135 	MOWNER_ATTACH(&ec->ec_rx_mowner);
1136 	ifp->if_mowner = &ec->ec_tx_mowner;
1137 #endif
1138 }
1139 
1140 void
1141 ether_ifdetach(struct ifnet *ifp)
1142 {
1143 	struct ethercom *ec = (void *) ifp;
1144 	struct ether_multi *enm;
1145 	int s;
1146 
1147 	/*
1148 	 * Prevent further calls to ioctl (for example turning off
1149 	 * promiscuous mode from the bridge code), which eventually can
1150 	 * call if_init() which can cause panics because the interface
1151 	 * is in the process of being detached. Return device not configured
1152 	 * instead.
1153 	 */
1154 	ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio;
1155 
1156 #if NBRIDGE > 0
1157 	if (ifp->if_bridge)
1158 		bridge_ifdetach(ifp);
1159 #endif
1160 
1161 	bpf_detach(ifp);
1162 
1163 #if NVLAN > 0
1164 	if (ec->ec_nvlans)
1165 		vlan_ifdetach(ifp);
1166 #endif
1167 
1168 	s = splnet();
1169 	while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) {
1170 		LIST_REMOVE(enm, enm_list);
1171 		free(enm, M_IFMADDR);
1172 		ec->ec_multicnt--;
1173 	}
1174 	splx(s);
1175 
1176 #if 0	/* done in if_detach() */
1177 	if_free_sadl(ifp);
1178 #endif
1179 
1180 	ifp->if_mowner = NULL;
1181 	MOWNER_DETACH(&ec->ec_rx_mowner);
1182 	MOWNER_DETACH(&ec->ec_tx_mowner);
1183 }
1184 
1185 #if 0
1186 /*
1187  * This is for reference.  We have a table-driven version
1188  * of the little-endian crc32 generator, which is faster
1189  * than the double-loop.
1190  */
1191 uint32_t
1192 ether_crc32_le(const uint8_t *buf, size_t len)
1193 {
1194 	uint32_t c, crc, carry;
1195 	size_t i, j;
1196 
1197 	crc = 0xffffffffU;	/* initial value */
1198 
1199 	for (i = 0; i < len; i++) {
1200 		c = buf[i];
1201 		for (j = 0; j < 8; j++) {
1202 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
1203 			crc >>= 1;
1204 			c >>= 1;
1205 			if (carry)
1206 				crc = (crc ^ ETHER_CRC_POLY_LE);
1207 		}
1208 	}
1209 
1210 	return (crc);
1211 }
1212 #else
1213 uint32_t
1214 ether_crc32_le(const uint8_t *buf, size_t len)
1215 {
1216 	static const uint32_t crctab[] = {
1217 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1218 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1219 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1220 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1221 	};
1222 	uint32_t crc;
1223 	size_t i;
1224 
1225 	crc = 0xffffffffU;	/* initial value */
1226 
1227 	for (i = 0; i < len; i++) {
1228 		crc ^= buf[i];
1229 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1230 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1231 	}
1232 
1233 	return (crc);
1234 }
1235 #endif
1236 
1237 uint32_t
1238 ether_crc32_be(const uint8_t *buf, size_t len)
1239 {
1240 	uint32_t c, crc, carry;
1241 	size_t i, j;
1242 
1243 	crc = 0xffffffffU;	/* initial value */
1244 
1245 	for (i = 0; i < len; i++) {
1246 		c = buf[i];
1247 		for (j = 0; j < 8; j++) {
1248 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
1249 			crc <<= 1;
1250 			c >>= 1;
1251 			if (carry)
1252 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1253 		}
1254 	}
1255 
1256 	return (crc);
1257 }
1258 
1259 #ifdef INET
1260 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] =
1261     { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 };
1262 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] =
1263     { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff };
1264 #endif
1265 #ifdef INET6
1266 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] =
1267     { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 };
1268 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] =
1269     { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff };
1270 #endif
1271 
1272 /*
1273  * ether_aton implementation, not using a static buffer.
1274  */
1275 int
1276 ether_aton_r(u_char *dest, size_t len, const char *str)
1277 {
1278         const u_char *cp = (const void *)str;
1279 	u_char *ep;
1280 
1281 #define atox(c)	(((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10))
1282 
1283 	if (len < ETHER_ADDR_LEN)
1284 		return ENOSPC;
1285 
1286 	ep = dest + ETHER_ADDR_LEN;
1287 
1288 	while (*cp) {
1289                 if (!isxdigit(*cp))
1290                         return EINVAL;
1291 		*dest = atox(*cp);
1292 		cp++;
1293                 if (isxdigit(*cp)) {
1294                         *dest = (*dest << 4) | atox(*cp);
1295 			dest++;
1296 			cp++;
1297                 } else
1298 			dest++;
1299 		if (dest == ep)
1300 			return *cp == '\0' ? 0 : ENAMETOOLONG;
1301 		switch (*cp) {
1302 		case ':':
1303 		case '-':
1304 		case '.':
1305 			cp++;
1306 			break;
1307 		}
1308         }
1309 	return ENOBUFS;
1310 }
1311 
1312 /*
1313  * Convert a sockaddr into an Ethernet address or range of Ethernet
1314  * addresses.
1315  */
1316 int
1317 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN],
1318     uint8_t addrhi[ETHER_ADDR_LEN])
1319 {
1320 #ifdef INET
1321 	const struct sockaddr_in *sin;
1322 #endif /* INET */
1323 #ifdef INET6
1324 	const struct sockaddr_in6 *sin6;
1325 #endif /* INET6 */
1326 
1327 	switch (sa->sa_family) {
1328 
1329 	case AF_UNSPEC:
1330 		memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN);
1331 		memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1332 		break;
1333 
1334 #ifdef INET
1335 	case AF_INET:
1336 		sin = satocsin(sa);
1337 		if (sin->sin_addr.s_addr == INADDR_ANY) {
1338 			/*
1339 			 * An IP address of INADDR_ANY means listen to
1340 			 * or stop listening to all of the Ethernet
1341 			 * multicast addresses used for IP.
1342 			 * (This is for the sake of IP multicast routers.)
1343 			 */
1344 			memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN);
1345 			memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN);
1346 		}
1347 		else {
1348 			ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo);
1349 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1350 		}
1351 		break;
1352 #endif
1353 #ifdef INET6
1354 	case AF_INET6:
1355 		sin6 = satocsin6(sa);
1356 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1357 			/*
1358 			 * An IP6 address of 0 means listen to or stop
1359 			 * listening to all of the Ethernet multicast
1360 			 * address used for IP6.
1361 			 * (This is used for multicast routers.)
1362 			 */
1363 			memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN);
1364 			memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN);
1365 		} else {
1366 			ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo);
1367 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1368 		}
1369 		break;
1370 #endif
1371 
1372 	default:
1373 		return EAFNOSUPPORT;
1374 	}
1375 	return 0;
1376 }
1377 
1378 /*
1379  * Add an Ethernet multicast address or range of addresses to the list for a
1380  * given interface.
1381  */
1382 int
1383 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec)
1384 {
1385 	struct ether_multi *enm;
1386 	u_char addrlo[ETHER_ADDR_LEN];
1387 	u_char addrhi[ETHER_ADDR_LEN];
1388 	int s = splnet(), error;
1389 
1390 	error = ether_multiaddr(sa, addrlo, addrhi);
1391 	if (error != 0) {
1392 		splx(s);
1393 		return error;
1394 	}
1395 
1396 	/*
1397 	 * Verify that we have valid Ethernet multicast addresses.
1398 	 */
1399 	if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) {
1400 		splx(s);
1401 		return EINVAL;
1402 	}
1403 	/*
1404 	 * See if the address range is already in the list.
1405 	 */
1406 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1407 	if (enm != NULL) {
1408 		/*
1409 		 * Found it; just increment the reference count.
1410 		 */
1411 		++enm->enm_refcount;
1412 		splx(s);
1413 		return 0;
1414 	}
1415 	/*
1416 	 * New address or range; malloc a new multicast record
1417 	 * and link it into the interface's multicast list.
1418 	 */
1419 	enm = (struct ether_multi *)malloc(sizeof(*enm), M_IFMADDR, M_NOWAIT);
1420 	if (enm == NULL) {
1421 		splx(s);
1422 		return ENOBUFS;
1423 	}
1424 	memcpy(enm->enm_addrlo, addrlo, 6);
1425 	memcpy(enm->enm_addrhi, addrhi, 6);
1426 	enm->enm_refcount = 1;
1427 	LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list);
1428 	ec->ec_multicnt++;
1429 	splx(s);
1430 	/*
1431 	 * Return ENETRESET to inform the driver that the list has changed
1432 	 * and its reception filter should be adjusted accordingly.
1433 	 */
1434 	return ENETRESET;
1435 }
1436 
1437 /*
1438  * Delete a multicast address record.
1439  */
1440 int
1441 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec)
1442 {
1443 	struct ether_multi *enm;
1444 	u_char addrlo[ETHER_ADDR_LEN];
1445 	u_char addrhi[ETHER_ADDR_LEN];
1446 	int s = splnet(), error;
1447 
1448 	error = ether_multiaddr(sa, addrlo, addrhi);
1449 	if (error != 0) {
1450 		splx(s);
1451 		return (error);
1452 	}
1453 
1454 	/*
1455 	 * Look ur the address in our list.
1456 	 */
1457 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1458 	if (enm == NULL) {
1459 		splx(s);
1460 		return (ENXIO);
1461 	}
1462 	if (--enm->enm_refcount != 0) {
1463 		/*
1464 		 * Still some claims to this record.
1465 		 */
1466 		splx(s);
1467 		return (0);
1468 	}
1469 	/*
1470 	 * No remaining claims to this record; unlink and free it.
1471 	 */
1472 	LIST_REMOVE(enm, enm_list);
1473 	free(enm, M_IFMADDR);
1474 	ec->ec_multicnt--;
1475 	splx(s);
1476 	/*
1477 	 * Return ENETRESET to inform the driver that the list has changed
1478 	 * and its reception filter should be adjusted accordingly.
1479 	 */
1480 	return (ENETRESET);
1481 }
1482 
1483 void
1484 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb)
1485 {
1486 	ec->ec_ifflags_cb = cb;
1487 }
1488 
1489 /*
1490  * Common ioctls for Ethernet interfaces.  Note, we must be
1491  * called at splnet().
1492  */
1493 int
1494 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data)
1495 {
1496 	struct ethercom *ec = (void *) ifp;
1497 	struct eccapreq *eccr;
1498 	struct ifreq *ifr = (struct ifreq *)data;
1499 	struct if_laddrreq *iflr = data;
1500 	const struct sockaddr_dl *sdl;
1501 	static const uint8_t zero[ETHER_ADDR_LEN];
1502 	int error;
1503 
1504 	switch (cmd) {
1505 	case SIOCINITIFADDR:
1506 	    {
1507 		struct ifaddr *ifa = (struct ifaddr *)data;
1508 		if (ifa->ifa_addr->sa_family != AF_LINK
1509 		    && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) !=
1510 		       (IFF_UP|IFF_RUNNING)) {
1511 			ifp->if_flags |= IFF_UP;
1512 			if ((error = (*ifp->if_init)(ifp)) != 0)
1513 				return error;
1514 		}
1515 #ifdef INET
1516 		if (ifa->ifa_addr->sa_family == AF_INET)
1517 			arp_ifinit(ifp, ifa);
1518 #endif /* INET */
1519 		return 0;
1520 	    }
1521 
1522 	case SIOCSIFMTU:
1523 	    {
1524 		int maxmtu;
1525 
1526 		if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU)
1527 			maxmtu = ETHERMTU_JUMBO;
1528 		else
1529 			maxmtu = ETHERMTU;
1530 
1531 		if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu)
1532 			return EINVAL;
1533 		else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET)
1534 			return error;
1535 		else if (ifp->if_flags & IFF_UP) {
1536 			/* Make sure the device notices the MTU change. */
1537 			return (*ifp->if_init)(ifp);
1538 		} else
1539 			return 0;
1540 	    }
1541 
1542 	case SIOCSIFFLAGS:
1543 		if ((error = ifioctl_common(ifp, cmd, data)) != 0)
1544 			return error;
1545 		switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) {
1546 		case IFF_RUNNING:
1547 			/*
1548 			 * If interface is marked down and it is running,
1549 			 * then stop and disable it.
1550 			 */
1551 			(*ifp->if_stop)(ifp, 1);
1552 			break;
1553 		case IFF_UP:
1554 			/*
1555 			 * If interface is marked up and it is stopped, then
1556 			 * start it.
1557 			 */
1558 			return (*ifp->if_init)(ifp);
1559 		case IFF_UP|IFF_RUNNING:
1560 			error = 0;
1561 			if (ec->ec_ifflags_cb == NULL ||
1562 			    (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) {
1563 				/*
1564 				 * Reset the interface to pick up
1565 				 * changes in any other flags that
1566 				 * affect the hardware state.
1567 				 */
1568 				return (*ifp->if_init)(ifp);
1569 			} else
1570 				return error;
1571 		case 0:
1572 			break;
1573 		}
1574 		return 0;
1575 	case SIOCGETHERCAP:
1576 		eccr = (struct eccapreq *)data;
1577 		eccr->eccr_capabilities = ec->ec_capabilities;
1578 		eccr->eccr_capenable = ec->ec_capenable;
1579 		return 0;
1580 	case SIOCADDMULTI:
1581 		return ether_addmulti(ifreq_getaddr(cmd, ifr), ec);
1582 	case SIOCDELMULTI:
1583 		return ether_delmulti(ifreq_getaddr(cmd, ifr), ec);
1584 	case SIOCSIFMEDIA:
1585 	case SIOCGIFMEDIA:
1586 		if (ec->ec_mii == NULL)
1587 			return ENOTTY;
1588 		return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd);
1589 	case SIOCALIFADDR:
1590 		sdl = satocsdl(sstocsa(&iflr->addr));
1591 		if (sdl->sdl_family != AF_LINK)
1592 			;
1593 		else if (ETHER_IS_MULTICAST(CLLADDR(sdl)))
1594 			return EINVAL;
1595 		else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0)
1596 			return EINVAL;
1597 		/*FALLTHROUGH*/
1598 	default:
1599 		return ifioctl_common(ifp, cmd, data);
1600 	}
1601 	return 0;
1602 }
1603