xref: /netbsd-src/sys/net/if_ethersubr.c (revision b7b7574d3bf8eeb51a1fa3977b59142ec6434a55)
1 /*	$NetBSD: if_ethersubr.c,v 1.202 2014/06/30 10:03:41 ozaki-r 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 1.202 2014/06/30 10:03:41 ozaki-r Exp $");
65 
66 #include "opt_inet.h"
67 #include "opt_atalk.h"
68 #include "opt_ipx.h"
69 #include "opt_mbuftrace.h"
70 #include "opt_mpls.h"
71 #include "opt_gateway.h"
72 #include "opt_pppoe.h"
73 #include "vlan.h"
74 #include "pppoe.h"
75 #include "bridge.h"
76 #include "arp.h"
77 #include "agr.h"
78 
79 #include <sys/param.h>
80 #include <sys/systm.h>
81 #include <sys/sysctl.h>
82 #include <sys/kernel.h>
83 #include <sys/callout.h>
84 #include <sys/malloc.h>
85 #include <sys/mbuf.h>
86 #include <sys/protosw.h>
87 #include <sys/socket.h>
88 #include <sys/ioctl.h>
89 #include <sys/errno.h>
90 #include <sys/syslog.h>
91 #include <sys/kauth.h>
92 #include <sys/cpu.h>
93 #include <sys/intr.h>
94 #include <sys/device.h>
95 
96 #include <net/if.h>
97 #include <net/netisr.h>
98 #include <net/route.h>
99 #include <net/if_llc.h>
100 #include <net/if_dl.h>
101 #include <net/if_types.h>
102 
103 #include <net/if_media.h>
104 #include <dev/mii/mii.h>
105 #include <dev/mii/miivar.h>
106 
107 #if NARP == 0
108 /*
109  * XXX there should really be a way to issue this warning from within config(8)
110  */
111 #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.
112 #endif
113 
114 #include <net/bpf.h>
115 
116 #include <net/if_ether.h>
117 #include <net/if_vlanvar.h>
118 
119 #if NPPPOE > 0
120 #include <net/if_pppoe.h>
121 #endif
122 
123 #if NAGR > 0
124 #include <net/agr/ieee8023_slowprotocols.h>	/* XXX */
125 #include <net/agr/ieee8023ad.h>
126 #include <net/agr/if_agrvar.h>
127 #endif
128 
129 #if NBRIDGE > 0
130 #include <net/if_bridgevar.h>
131 #endif
132 
133 #include <netinet/in.h>
134 #ifdef INET
135 #include <netinet/in_var.h>
136 #endif
137 #include <netinet/if_inarp.h>
138 
139 #ifdef INET6
140 #ifndef INET
141 #include <netinet/in.h>
142 #endif
143 #include <netinet6/in6_var.h>
144 #include <netinet6/nd6.h>
145 #endif
146 
147 
148 #include "carp.h"
149 #if NCARP > 0
150 #include <netinet/ip_carp.h>
151 #endif
152 
153 #ifdef IPX
154 #include <netipx/ipx.h>
155 #include <netipx/ipx_if.h>
156 #endif
157 
158 #ifdef NETATALK
159 #include <netatalk/at.h>
160 #include <netatalk/at_var.h>
161 #include <netatalk/at_extern.h>
162 
163 #define llc_snap_org_code llc_un.type_snap.org_code
164 #define llc_snap_ether_type llc_un.type_snap.ether_type
165 
166 extern u_char	at_org_code[3];
167 extern u_char	aarp_org_code[3];
168 #endif /* NETATALK */
169 
170 #ifdef MPLS
171 #include <netmpls/mpls.h>
172 #include <netmpls/mpls_var.h>
173 #endif
174 
175 static struct timeval bigpktppslim_last;
176 static int bigpktppslim = 2;	/* XXX */
177 static int bigpktpps_count;
178 
179 
180 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] =
181     { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
182 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] =
183     { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 };
184 #define senderr(e) { error = (e); goto bad;}
185 
186 static	int ether_output(struct ifnet *, struct mbuf *,
187 	    const struct sockaddr *, struct rtentry *);
188 
189 /*
190  * Ethernet output routine.
191  * Encapsulate a packet of type family for the local net.
192  * Assumes that ifp is actually pointer to ethercom structure.
193  */
194 static int
195 ether_output(struct ifnet * const ifp0, struct mbuf * const m0,
196 	const struct sockaddr * const dst,
197 	struct rtentry *rt0)
198 {
199 	uint16_t etype = 0;
200 	int error = 0, hdrcmplt = 0;
201  	uint8_t esrc[6], edst[6];
202 	struct mbuf *m = m0;
203 	struct rtentry *rt;
204 	struct mbuf *mcopy = NULL;
205 	struct ether_header *eh;
206 	struct ifnet *ifp = ifp0;
207 	ALTQ_DECL(struct altq_pktattr pktattr;)
208 #ifdef INET
209 	struct arphdr *ah;
210 #endif /* INET */
211 #ifdef NETATALK
212 	struct at_ifaddr *aa;
213 #endif /* NETATALK */
214 
215 	KASSERT(KERNEL_LOCKED_P());
216 
217 #ifdef MBUFTRACE
218 	m_claimm(m, ifp->if_mowner);
219 #endif
220 
221 #if NCARP > 0
222 	if (ifp->if_type == IFT_CARP) {
223 		struct ifaddr *ifa;
224 
225 		/* loop back if this is going to the carp interface */
226 		if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP &&
227 		    (ifa = ifa_ifwithaddr(dst)) != NULL &&
228 		    ifa->ifa_ifp == ifp0)
229 			return looutput(ifp0, m, dst, rt0);
230 
231 		ifp = ifp->if_carpdev;
232 		/* ac = (struct arpcom *)ifp; */
233 
234 		if ((ifp0->if_flags & (IFF_UP|IFF_RUNNING)) !=
235 		    (IFF_UP|IFF_RUNNING))
236 			senderr(ENETDOWN);
237 	}
238 #endif /* NCARP > 0 */
239 
240 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
241 		senderr(ENETDOWN);
242 	if ((rt = rt0) != NULL) {
243 		if ((rt->rt_flags & RTF_UP) == 0) {
244 			if ((rt0 = rt = rtalloc1(dst, 1)) != NULL) {
245 				rt->rt_refcnt--;
246 				if (rt->rt_ifp != ifp)
247 					return (*rt->rt_ifp->if_output)
248 							(ifp, m0, dst, rt);
249 			} else
250 				senderr(EHOSTUNREACH);
251 		}
252 		if ((rt->rt_flags & RTF_GATEWAY) && dst->sa_family != AF_NS) {
253 			if (rt->rt_gwroute == NULL)
254 				goto lookup;
255 			if (((rt = rt->rt_gwroute)->rt_flags & RTF_UP) == 0) {
256 				rtfree(rt); rt = rt0;
257 			lookup: rt->rt_gwroute = rtalloc1(rt->rt_gateway, 1);
258 				if ((rt = rt->rt_gwroute) == NULL)
259 					senderr(EHOSTUNREACH);
260 				/* the "G" test below also prevents rt == rt0 */
261 				if ((rt->rt_flags & RTF_GATEWAY) ||
262 				    (rt->rt_ifp != ifp)) {
263 					rt->rt_refcnt--;
264 					rt0->rt_gwroute = NULL;
265 					senderr(EHOSTUNREACH);
266 				}
267 			}
268 		}
269 		if (rt->rt_flags & RTF_REJECT)
270 			if (rt->rt_rmx.rmx_expire == 0 ||
271 			    (u_long) time_second < rt->rt_rmx.rmx_expire)
272 				senderr(rt == rt0 ? EHOSTDOWN : EHOSTUNREACH);
273 	}
274 
275 	switch (dst->sa_family) {
276 
277 #ifdef INET
278 	case AF_INET:
279 		if (m->m_flags & M_BCAST)
280 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
281 		else if (m->m_flags & M_MCAST)
282 			ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst);
283 		else if (!arpresolve(ifp, rt, m, dst, edst))
284 			return (0);	/* if not yet resolved */
285 		/* If broadcasting on a simplex interface, loopback a copy */
286 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
287 			mcopy = m_copy(m, 0, (int)M_COPYALL);
288 		etype = htons(ETHERTYPE_IP);
289 		break;
290 
291 	case AF_ARP:
292 		ah = mtod(m, struct arphdr *);
293 		if (m->m_flags & M_BCAST)
294 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
295 		else {
296 			void *tha = ar_tha(ah);
297 
298 			if (tha == NULL) {
299 				/* fake with ARPHDR_IEEE1394 */
300 				return 0;
301 			}
302 			memcpy(edst, tha, sizeof(edst));
303 		}
304 
305 		ah->ar_hrd = htons(ARPHRD_ETHER);
306 
307 		switch (ntohs(ah->ar_op)) {
308 		case ARPOP_REVREQUEST:
309 		case ARPOP_REVREPLY:
310 			etype = htons(ETHERTYPE_REVARP);
311 			break;
312 
313 		case ARPOP_REQUEST:
314 		case ARPOP_REPLY:
315 		default:
316 			etype = htons(ETHERTYPE_ARP);
317 		}
318 
319 		break;
320 #endif
321 #ifdef INET6
322 	case AF_INET6:
323 		if (!nd6_storelladdr(ifp, rt, m, dst, edst, sizeof(edst))){
324 			/* something bad happened */
325 			return (0);
326 		}
327 		etype = htons(ETHERTYPE_IPV6);
328 		break;
329 #endif
330 #ifdef NETATALK
331     case AF_APPLETALK:
332 		if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) {
333 #ifdef NETATALKDEBUG
334 			printf("aarpresolv failed\n");
335 #endif /* NETATALKDEBUG */
336 			return (0);
337 		}
338 		/*
339 		 * ifaddr is the first thing in at_ifaddr
340 		 */
341 		aa = (struct at_ifaddr *) at_ifawithnet(
342 		    (const struct sockaddr_at *)dst, ifp);
343 		if (aa == NULL)
344 		    goto bad;
345 
346 		/*
347 		 * In the phase 2 case, we need to prepend an mbuf for the
348 		 * llc header.  Since we must preserve the value of m,
349 		 * which is passed to us by value, we m_copy() the first
350 		 * mbuf, and use it for our llc header.
351 		 */
352 		if (aa->aa_flags & AFA_PHASE2) {
353 			struct llc llc;
354 
355 			M_PREPEND(m, sizeof(struct llc), M_DONTWAIT);
356 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
357 			llc.llc_control = LLC_UI;
358 			memcpy(llc.llc_snap_org_code, at_org_code,
359 			    sizeof(llc.llc_snap_org_code));
360 			llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK);
361 			memcpy(mtod(m, void *), &llc, sizeof(struct llc));
362 		} else {
363 			etype = htons(ETHERTYPE_ATALK);
364 		}
365 		break;
366 #endif /* NETATALK */
367 #ifdef IPX
368 	case AF_IPX:
369 		etype = htons(ETHERTYPE_IPX);
370  		memcpy(edst,
371 		    &(((const struct sockaddr_ipx *)dst)->sipx_addr.x_host),
372 		    sizeof(edst));
373 		/* If broadcasting on a simplex interface, loopback a copy */
374 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
375 			mcopy = m_copy(m, 0, (int)M_COPYALL);
376 		break;
377 #endif
378 	case pseudo_AF_HDRCMPLT:
379 		hdrcmplt = 1;
380 		memcpy(esrc,
381 		    ((const struct ether_header *)dst->sa_data)->ether_shost,
382 		    sizeof(esrc));
383 		/* FALLTHROUGH */
384 
385 	case AF_UNSPEC:
386  		memcpy(edst,
387 		    ((const struct ether_header *)dst->sa_data)->ether_dhost,
388 		    sizeof(edst));
389 		/* AF_UNSPEC doesn't swap the byte order of the ether_type. */
390 		etype = ((const struct ether_header *)dst->sa_data)->ether_type;
391 		break;
392 
393 	default:
394 		printf("%s: can't handle af%d\n", ifp->if_xname,
395 			dst->sa_family);
396 		senderr(EAFNOSUPPORT);
397 	}
398 
399 #ifdef MPLS
400 	if (rt0 != NULL && rt_gettag(rt0) != NULL &&
401 	    rt_gettag(rt0)->sa_family == AF_MPLS &&
402 	    (m->m_flags & (M_MCAST | M_BCAST)) == 0) {
403 		union mpls_shim msh;
404 		msh.s_addr = MPLS_GETSADDR(rt0);
405 		if (msh.shim.label != MPLS_LABEL_IMPLNULL)
406 			etype = htons(ETHERTYPE_MPLS);
407 	}
408 #endif
409 
410 	if (mcopy)
411 		(void)looutput(ifp, mcopy, dst, rt);
412 
413 	/* If no ether type is set, this must be a 802.2 formatted packet.
414 	 */
415 	if (etype == 0)
416 		etype = htons(m->m_pkthdr.len);
417 	/*
418 	 * Add local net header.  If no space in first mbuf,
419 	 * allocate another.
420 	 */
421 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
422 	if (m == 0)
423 		senderr(ENOBUFS);
424 	eh = mtod(m, struct ether_header *);
425 	/* Note: etype is already in network byte order. */
426 	(void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
427  	memcpy(eh->ether_dhost, edst, sizeof(edst));
428 	if (hdrcmplt)
429 		memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost));
430 	else
431 	 	memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl),
432 		    sizeof(eh->ether_shost));
433 
434 #if NCARP > 0
435 	if (ifp0 != ifp && ifp0->if_type == IFT_CARP) {
436 	 	memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl),
437 		    sizeof(eh->ether_shost));
438 	}
439 #endif /* NCARP > 0 */
440 
441 	if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0)
442 		return (error);
443 	if (m == NULL)
444 		return (0);
445 
446 #if NBRIDGE > 0
447 	/*
448 	 * Bridges require special output handling.
449 	 */
450 	if (ifp->if_bridge)
451 		return (bridge_output(ifp, m, NULL, NULL));
452 #endif
453 
454 #if NCARP > 0
455 	if (ifp != ifp0)
456 		ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN;
457 #endif /* NCARP > 0 */
458 
459 #ifdef ALTQ
460 	/*
461 	 * If ALTQ is enabled on the parent interface, do
462 	 * classification; the queueing discipline might not
463 	 * require classification, but might require the
464 	 * address family/header pointer in the pktattr.
465 	 */
466 	if (ALTQ_IS_ENABLED(&ifp->if_snd))
467 		altq_etherclassify(&ifp->if_snd, m, &pktattr);
468 #endif
469 	return ifq_enqueue(ifp, m ALTQ_COMMA ALTQ_DECL(&pktattr));
470 
471 bad:
472 	if (m)
473 		m_freem(m);
474 	return (error);
475 }
476 
477 #ifdef ALTQ
478 /*
479  * This routine is a slight hack to allow a packet to be classified
480  * if the Ethernet headers are present.  It will go away when ALTQ's
481  * classification engine understands link headers.
482  */
483 void
484 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
485     struct altq_pktattr *pktattr)
486 {
487 	struct ether_header *eh;
488 	uint16_t ether_type;
489 	int hlen, af, hdrsize;
490 	void *hdr;
491 
492 	hlen = ETHER_HDR_LEN;
493 	eh = mtod(m, struct ether_header *);
494 
495 	ether_type = htons(eh->ether_type);
496 
497 	if (ether_type < ETHERMTU) {
498 		/* LLC/SNAP */
499 		struct llc *llc = (struct llc *)(eh + 1);
500 		hlen += 8;
501 
502 		if (m->m_len < hlen ||
503 		    llc->llc_dsap != LLC_SNAP_LSAP ||
504 		    llc->llc_ssap != LLC_SNAP_LSAP ||
505 		    llc->llc_control != LLC_UI) {
506 			/* Not SNAP. */
507 			goto bad;
508 		}
509 
510 		ether_type = htons(llc->llc_un.type_snap.ether_type);
511 	}
512 
513 	switch (ether_type) {
514 	case ETHERTYPE_IP:
515 		af = AF_INET;
516 		hdrsize = 20;		/* sizeof(struct ip) */
517 		break;
518 
519 	case ETHERTYPE_IPV6:
520 		af = AF_INET6;
521 		hdrsize = 40;		/* sizeof(struct ip6_hdr) */
522 		break;
523 
524 	default:
525 		af = AF_UNSPEC;
526 		hdrsize = 0;
527 		break;
528 	}
529 
530 	while (m->m_len <= hlen) {
531 		hlen -= m->m_len;
532 		m = m->m_next;
533 	}
534 	if (m->m_len < (hlen + hdrsize)) {
535 		/*
536 		 * protocol header not in a single mbuf.
537 		 * We can't cope with this situation right
538 		 * now (but it shouldn't ever happen, really, anyhow).
539 		 */
540 #ifdef DEBUG
541 		printf("altq_etherclassify: headers span multiple mbufs: "
542 		    "%d < %d\n", m->m_len, (hlen + hdrsize));
543 #endif
544 		goto bad;
545 	}
546 
547 	m->m_data += hlen;
548 	m->m_len -= hlen;
549 
550 	hdr = mtod(m, void *);
551 
552 	if (ALTQ_NEEDS_CLASSIFY(ifq))
553 		pktattr->pattr_class =
554 		    (*ifq->altq_classify)(ifq->altq_clfier, m, af);
555 	pktattr->pattr_af = af;
556 	pktattr->pattr_hdr = hdr;
557 
558 	m->m_data -= hlen;
559 	m->m_len += hlen;
560 
561 	return;
562 
563  bad:
564 	pktattr->pattr_class = NULL;
565 	pktattr->pattr_hdr = NULL;
566 	pktattr->pattr_af = AF_UNSPEC;
567 }
568 #endif /* ALTQ */
569 
570 /*
571  * Process a received Ethernet packet;
572  * the packet is in the mbuf chain m with
573  * the ether header.
574  */
575 void
576 ether_input(struct ifnet *ifp, struct mbuf *m)
577 {
578 	struct ethercom *ec = (struct ethercom *) ifp;
579 	pktqueue_t *pktq = NULL;
580 	struct ifqueue *inq = NULL;
581 	uint16_t etype;
582 	struct ether_header *eh;
583 	size_t ehlen;
584 	int isr = 0;
585 #if defined (LLC) || defined(NETATALK)
586 	struct llc *l;
587 #endif
588 
589 	if ((ifp->if_flags & IFF_UP) == 0) {
590 		m_freem(m);
591 		return;
592 	}
593 
594 #ifdef MBUFTRACE
595 	m_claimm(m, &ec->ec_rx_mowner);
596 #endif
597 	eh = mtod(m, struct ether_header *);
598 	etype = ntohs(eh->ether_type);
599 	ehlen = sizeof(*eh);
600 
601 	/*
602 	 * Determine if the packet is within its size limits.
603 	 */
604 	if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len >
605 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
606 		if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count,
607 			    bigpktppslim)) {
608 			printf("%s: discarding oversize frame (len=%d)\n",
609 			    ifp->if_xname, m->m_pkthdr.len);
610 		}
611 		m_freem(m);
612 		return;
613 	}
614 
615 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
616 		/*
617 		 * If this is not a simplex interface, drop the packet
618 		 * if it came from us.
619 		 */
620 		if ((ifp->if_flags & IFF_SIMPLEX) == 0 &&
621 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost,
622 		    ETHER_ADDR_LEN) == 0) {
623 			m_freem(m);
624 			return;
625 		}
626 
627 		if (memcmp(etherbroadcastaddr,
628 		    eh->ether_dhost, ETHER_ADDR_LEN) == 0)
629 			m->m_flags |= M_BCAST;
630 		else
631 			m->m_flags |= M_MCAST;
632 		ifp->if_imcasts++;
633 	}
634 
635 	/* If the CRC is still on the packet, trim it off. */
636 	if (m->m_flags & M_HASFCS) {
637 		m_adj(m, -ETHER_CRC_LEN);
638 		m->m_flags &= ~M_HASFCS;
639 	}
640 
641 	ifp->if_ibytes += m->m_pkthdr.len;
642 
643 #if NCARP > 0
644 	if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) {
645 		/*
646 		 * clear M_PROMISC, in case the packets comes from a
647 		 * vlan
648 		 */
649 		m->m_flags &= ~M_PROMISC;
650 		if (carp_input(m, (uint8_t *)&eh->ether_shost,
651 		    (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0)
652 			return;
653 	}
654 #endif /* NCARP > 0 */
655 	if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 &&
656 	    (ifp->if_flags & IFF_PROMISC) != 0 &&
657 	    memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost,
658 		   ETHER_ADDR_LEN) != 0) {
659 		m->m_flags |= M_PROMISC;
660 	}
661 
662 	if ((m->m_flags & M_PROMISC) == 0) {
663 		if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0)
664 			return;
665 		if (m == NULL)
666 			return;
667 
668 		eh = mtod(m, struct ether_header *);
669 		etype = ntohs(eh->ether_type);
670 		ehlen = sizeof(*eh);
671 	}
672 
673 #if NAGR > 0
674 	if (ifp->if_agrprivate &&
675 	    __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) {
676 		m->m_flags &= ~M_PROMISC;
677 		agr_input(ifp, m);
678 		return;
679 	}
680 #endif /* NAGR > 0 */
681 
682 	/*
683 	 * If VLANs are configured on the interface, check to
684 	 * see if the device performed the decapsulation and
685 	 * provided us with the tag.
686 	 */
687 	if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) {
688 #if NVLAN > 0
689 		/*
690 		 * vlan_input() will either recursively call ether_input()
691 		 * or drop the packet.
692 		 */
693 		vlan_input(ifp, m);
694 #else
695 		m_freem(m);
696 #endif
697 		return;
698 	}
699 
700 	/*
701 	 * Handle protocols that expect to have the Ethernet header
702 	 * (and possibly FCS) intact.
703 	 */
704 	switch (etype) {
705 	case ETHERTYPE_VLAN: {
706 		struct ether_vlan_header *evl = (void *)eh;
707 		/*
708 		 * If there is a tag of 0, then the VLAN header was probably
709 		 * just being used to store the priority.  Extract the ether
710 		 * type, and if IP or IPV6, let them deal with it.
711 		 */
712 		if (m->m_len <= sizeof(*evl)
713 		    && EVL_VLANOFTAG(evl->evl_tag) == 0) {
714 			etype = ntohs(evl->evl_proto);
715 			ehlen = sizeof(*evl);
716 			if ((m->m_flags & M_PROMISC) == 0
717 			    && (etype == ETHERTYPE_IP
718 				|| etype == ETHERTYPE_IPV6))
719 				break;
720 		}
721 #if NVLAN > 0
722 		/*
723 		 * vlan_input() will either recursively call ether_input()
724 		 * or drop the packet.
725 		 */
726 		if (((struct ethercom *)ifp)->ec_nvlans != 0)
727 			vlan_input(ifp, m);
728 		else
729 #endif /* NVLAN > 0 */
730 			m_freem(m);
731 		return;
732 	}
733 #if NPPPOE > 0
734 	case ETHERTYPE_PPPOEDISC:
735 	case ETHERTYPE_PPPOE:
736 		if (m->m_flags & M_PROMISC) {
737 			m_freem(m);
738 			return;
739 		}
740 #ifndef PPPOE_SERVER
741 		if (m->m_flags & (M_MCAST | M_BCAST)) {
742 			m_freem(m);
743 			return;
744 		}
745 #endif
746 
747 		if (etype == ETHERTYPE_PPPOEDISC)
748 			inq = &ppoediscinq;
749 		else
750 			inq = &ppoeinq;
751 		if (IF_QFULL(inq)) {
752 			IF_DROP(inq);
753 			m_freem(m);
754 		} else {
755 			IF_ENQUEUE(inq, m);
756 			softint_schedule(pppoe_softintr);
757 		}
758 		return;
759 #endif /* NPPPOE > 0 */
760 	case ETHERTYPE_SLOWPROTOCOLS: {
761 		uint8_t subtype;
762 
763 #if defined(DIAGNOSTIC)
764 		if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) {
765 			panic("ether_input: too short slow protocol packet");
766 		}
767 #endif
768 		m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype);
769 		switch (subtype) {
770 #if NAGR > 0
771 		case SLOWPROTOCOLS_SUBTYPE_LACP:
772 			if (ifp->if_agrprivate) {
773 				ieee8023ad_lacp_input(ifp, m);
774 				return;
775 			}
776 			break;
777 
778 		case SLOWPROTOCOLS_SUBTYPE_MARKER:
779 			if (ifp->if_agrprivate) {
780 				ieee8023ad_marker_input(ifp, m);
781 				return;
782 			}
783 			break;
784 #endif /* NAGR > 0 */
785 		default:
786 			if (subtype == 0 || subtype > 10) {
787 				/* illegal value */
788 				m_freem(m);
789 				return;
790 			}
791 			/* unknown subtype */
792 			break;
793 		}
794 		/* FALLTHROUGH */
795 	}
796 	default:
797 		if (m->m_flags & M_PROMISC) {
798 			m_freem(m);
799 			return;
800 		}
801 	}
802 
803 	/* If the CRC is still on the packet, trim it off. */
804 	if (m->m_flags & M_HASFCS) {
805 		m_adj(m, -ETHER_CRC_LEN);
806 		m->m_flags &= ~M_HASFCS;
807 	}
808 
809 	if (etype > ETHERMTU + sizeof (struct ether_header)) {
810 		/* Strip off the Ethernet header. */
811 		m_adj(m, ehlen);
812 
813 		switch (etype) {
814 #ifdef INET
815 		case ETHERTYPE_IP:
816 #ifdef GATEWAY
817 			if (ipflow_fastforward(m))
818 				return;
819 #endif
820 			pktq = ip_pktq;
821 			break;
822 
823 		case ETHERTYPE_ARP:
824 			isr = NETISR_ARP;
825 			inq = &arpintrq;
826 			break;
827 
828 		case ETHERTYPE_REVARP:
829 			revarpinput(m);	/* XXX queue? */
830 			return;
831 #endif
832 #ifdef INET6
833 		case ETHERTYPE_IPV6:
834 			if (__predict_false(!in6_present)) {
835 				m_freem(m);
836 				return;
837 			}
838 #ifdef GATEWAY
839 			if (ip6flow_fastforward(&m))
840 				return;
841 #endif
842 			pktq = ip6_pktq;
843 			break;
844 #endif
845 #ifdef IPX
846 		case ETHERTYPE_IPX:
847 			isr = NETISR_IPX;
848 			inq = &ipxintrq;
849 			break;
850 #endif
851 #ifdef NETATALK
852 		case ETHERTYPE_ATALK:
853 			isr = NETISR_ATALK;
854 			inq = &atintrq1;
855 			break;
856 		case ETHERTYPE_AARP:
857 			/* probably this should be done with a NETISR as well */
858 			aarpinput(ifp, m); /* XXX */
859 			return;
860 #endif /* NETATALK */
861 #ifdef MPLS
862 		case ETHERTYPE_MPLS:
863 			isr = NETISR_MPLS;
864 			inq = &mplsintrq;
865 			break;
866 #endif
867 		default:
868 			m_freem(m);
869 			return;
870 		}
871 	} else {
872 #if defined (LLC) || defined (NETATALK)
873 		l = (struct llc *)(eh+1);
874 		switch (l->llc_dsap) {
875 #ifdef NETATALK
876 		case LLC_SNAP_LSAP:
877 			switch (l->llc_control) {
878 			case LLC_UI:
879 				if (l->llc_ssap != LLC_SNAP_LSAP) {
880 					goto dropanyway;
881 				}
882 
883 				if (memcmp(&(l->llc_snap_org_code)[0],
884 				    at_org_code, sizeof(at_org_code)) == 0 &&
885 				    ntohs(l->llc_snap_ether_type) ==
886 				    ETHERTYPE_ATALK) {
887 					inq = &atintrq2;
888 					m_adj(m, sizeof(struct ether_header)
889 					    + sizeof(struct llc));
890 					isr = NETISR_ATALK;
891 					break;
892 				}
893 
894 				if (memcmp(&(l->llc_snap_org_code)[0],
895 				    aarp_org_code,
896 				    sizeof(aarp_org_code)) == 0 &&
897 				    ntohs(l->llc_snap_ether_type) ==
898 				    ETHERTYPE_AARP) {
899 					m_adj( m, sizeof(struct ether_header)
900 					    + sizeof(struct llc));
901 					aarpinput(ifp, m); /* XXX */
902 				    return;
903 				}
904 
905 			default:
906 				goto dropanyway;
907 			}
908 			break;
909 		dropanyway:
910 #endif
911 		default:
912 			m_freem(m);
913 			return;
914 		}
915 #else /* ISO || LLC || NETATALK*/
916 		m_freem(m);
917 		return;
918 #endif /* ISO || LLC || NETATALK*/
919 	}
920 
921 	if (__predict_true(pktq)) {
922 		const uint32_t h = pktq_rps_hash(m);
923 		if (__predict_false(!pktq_enqueue(pktq, m, h))) {
924 			m_freem(m);
925 		}
926 		return;
927 	}
928 
929 	if (__predict_false(!inq)) {
930 		/* Should not happen. */
931 		m_freem(m);
932 		return;
933 	}
934 	if (IF_QFULL(inq)) {
935 		IF_DROP(inq);
936 		m_freem(m);
937 	} else {
938 		IF_ENQUEUE(inq, m);
939 		schednetisr(isr);
940 	}
941 }
942 
943 /*
944  * Convert Ethernet address to printable (loggable) representation.
945  */
946 char *
947 ether_sprintf(const u_char *ap)
948 {
949 	static char etherbuf[3 * ETHER_ADDR_LEN];
950 	return ether_snprintf(etherbuf, sizeof(etherbuf), ap);
951 }
952 
953 char *
954 ether_snprintf(char *buf, size_t len, const u_char *ap)
955 {
956 	char *cp = buf;
957 	size_t i;
958 
959 	for (i = 0; i < len / 3; i++) {
960 		*cp++ = hexdigits[*ap >> 4];
961 		*cp++ = hexdigits[*ap++ & 0xf];
962 		*cp++ = ':';
963 	}
964 	*--cp = '\0';
965 	return buf;
966 }
967 
968 /*
969  * Perform common duties while attaching to interface list
970  */
971 void
972 ether_ifattach(struct ifnet *ifp, const uint8_t *lla)
973 {
974 	struct ethercom *ec = (struct ethercom *)ifp;
975 
976 	ifp->if_type = IFT_ETHER;
977 	ifp->if_hdrlen = ETHER_HDR_LEN;
978 	ifp->if_dlt = DLT_EN10MB;
979 	ifp->if_mtu = ETHERMTU;
980 	ifp->if_output = ether_output;
981 	ifp->if_input = ether_input;
982 	if (ifp->if_baudrate == 0)
983 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
984 
985 	if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla));
986 
987 	LIST_INIT(&ec->ec_multiaddrs);
988 	ifp->if_broadcastaddr = etherbroadcastaddr;
989 	bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header));
990 #ifdef MBUFTRACE
991 	strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname,
992 	    sizeof(ec->ec_tx_mowner.mo_name));
993 	strlcpy(ec->ec_tx_mowner.mo_descr, "tx",
994 	    sizeof(ec->ec_tx_mowner.mo_descr));
995 	strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname,
996 	    sizeof(ec->ec_rx_mowner.mo_name));
997 	strlcpy(ec->ec_rx_mowner.mo_descr, "rx",
998 	    sizeof(ec->ec_rx_mowner.mo_descr));
999 	MOWNER_ATTACH(&ec->ec_tx_mowner);
1000 	MOWNER_ATTACH(&ec->ec_rx_mowner);
1001 	ifp->if_mowner = &ec->ec_tx_mowner;
1002 #endif
1003 }
1004 
1005 void
1006 ether_ifdetach(struct ifnet *ifp)
1007 {
1008 	struct ethercom *ec = (void *) ifp;
1009 	struct ether_multi *enm;
1010 	int s;
1011 
1012 	/*
1013 	 * Prevent further calls to ioctl (for example turning off
1014 	 * promiscuous mode from the bridge code), which eventually can
1015 	 * call if_init() which can cause panics because the interface
1016 	 * is in the process of being detached. Return device not configured
1017 	 * instead.
1018 	 */
1019 	ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio;
1020 
1021 #if NBRIDGE > 0
1022 	if (ifp->if_bridge)
1023 		bridge_ifdetach(ifp);
1024 #endif
1025 
1026 	bpf_detach(ifp);
1027 
1028 #if NVLAN > 0
1029 	if (ec->ec_nvlans)
1030 		vlan_ifdetach(ifp);
1031 #endif
1032 
1033 	s = splnet();
1034 	while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) {
1035 		LIST_REMOVE(enm, enm_list);
1036 		free(enm, M_IFMADDR);
1037 		ec->ec_multicnt--;
1038 	}
1039 	splx(s);
1040 
1041 #if 0	/* done in if_detach() */
1042 	if_free_sadl(ifp);
1043 #endif
1044 
1045 	ifp->if_mowner = NULL;
1046 	MOWNER_DETACH(&ec->ec_rx_mowner);
1047 	MOWNER_DETACH(&ec->ec_tx_mowner);
1048 }
1049 
1050 #if 0
1051 /*
1052  * This is for reference.  We have a table-driven version
1053  * of the little-endian crc32 generator, which is faster
1054  * than the double-loop.
1055  */
1056 uint32_t
1057 ether_crc32_le(const uint8_t *buf, size_t len)
1058 {
1059 	uint32_t c, crc, carry;
1060 	size_t i, j;
1061 
1062 	crc = 0xffffffffU;	/* initial value */
1063 
1064 	for (i = 0; i < len; i++) {
1065 		c = buf[i];
1066 		for (j = 0; j < 8; j++) {
1067 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
1068 			crc >>= 1;
1069 			c >>= 1;
1070 			if (carry)
1071 				crc = (crc ^ ETHER_CRC_POLY_LE);
1072 		}
1073 	}
1074 
1075 	return (crc);
1076 }
1077 #else
1078 uint32_t
1079 ether_crc32_le(const uint8_t *buf, size_t len)
1080 {
1081 	static const uint32_t crctab[] = {
1082 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1083 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1084 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1085 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1086 	};
1087 	uint32_t crc;
1088 	size_t i;
1089 
1090 	crc = 0xffffffffU;	/* initial value */
1091 
1092 	for (i = 0; i < len; i++) {
1093 		crc ^= buf[i];
1094 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1095 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1096 	}
1097 
1098 	return (crc);
1099 }
1100 #endif
1101 
1102 uint32_t
1103 ether_crc32_be(const uint8_t *buf, size_t len)
1104 {
1105 	uint32_t c, crc, carry;
1106 	size_t i, j;
1107 
1108 	crc = 0xffffffffU;	/* initial value */
1109 
1110 	for (i = 0; i < len; i++) {
1111 		c = buf[i];
1112 		for (j = 0; j < 8; j++) {
1113 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
1114 			crc <<= 1;
1115 			c >>= 1;
1116 			if (carry)
1117 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1118 		}
1119 	}
1120 
1121 	return (crc);
1122 }
1123 
1124 #ifdef INET
1125 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] =
1126     { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 };
1127 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] =
1128     { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff };
1129 #endif
1130 #ifdef INET6
1131 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] =
1132     { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 };
1133 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] =
1134     { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff };
1135 #endif
1136 
1137 /*
1138  * ether_aton implementation, not using a static buffer.
1139  */
1140 int
1141 ether_aton_r(u_char *dest, size_t len, const char *str)
1142 {
1143         const u_char *cp = (const void *)str;
1144 	u_char *ep;
1145 
1146 #define atox(c)	(((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10))
1147 
1148 	if (len < ETHER_ADDR_LEN)
1149 		return ENOSPC;
1150 
1151 	ep = dest + ETHER_ADDR_LEN;
1152 
1153 	while (*cp) {
1154                 if (!isxdigit(*cp))
1155                         return EINVAL;
1156 		*dest = atox(*cp);
1157 		cp++;
1158                 if (isxdigit(*cp)) {
1159                         *dest = (*dest << 4) | atox(*cp);
1160 			dest++;
1161 			cp++;
1162                 } else
1163 			dest++;
1164 		if (dest == ep)
1165 			return *cp == '\0' ? 0 : ENAMETOOLONG;
1166 		switch (*cp) {
1167 		case ':':
1168 		case '-':
1169 		case '.':
1170 			cp++;
1171 			break;
1172 		}
1173         }
1174 	return ENOBUFS;
1175 }
1176 
1177 /*
1178  * Convert a sockaddr into an Ethernet address or range of Ethernet
1179  * addresses.
1180  */
1181 int
1182 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN],
1183     uint8_t addrhi[ETHER_ADDR_LEN])
1184 {
1185 #ifdef INET
1186 	const struct sockaddr_in *sin;
1187 #endif /* INET */
1188 #ifdef INET6
1189 	const struct sockaddr_in6 *sin6;
1190 #endif /* INET6 */
1191 
1192 	switch (sa->sa_family) {
1193 
1194 	case AF_UNSPEC:
1195 		memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN);
1196 		memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1197 		break;
1198 
1199 #ifdef INET
1200 	case AF_INET:
1201 		sin = satocsin(sa);
1202 		if (sin->sin_addr.s_addr == INADDR_ANY) {
1203 			/*
1204 			 * An IP address of INADDR_ANY means listen to
1205 			 * or stop listening to all of the Ethernet
1206 			 * multicast addresses used for IP.
1207 			 * (This is for the sake of IP multicast routers.)
1208 			 */
1209 			memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN);
1210 			memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN);
1211 		}
1212 		else {
1213 			ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo);
1214 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1215 		}
1216 		break;
1217 #endif
1218 #ifdef INET6
1219 	case AF_INET6:
1220 		sin6 = satocsin6(sa);
1221 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1222 			/*
1223 			 * An IP6 address of 0 means listen to or stop
1224 			 * listening to all of the Ethernet multicast
1225 			 * address used for IP6.
1226 			 * (This is used for multicast routers.)
1227 			 */
1228 			memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN);
1229 			memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN);
1230 		} else {
1231 			ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo);
1232 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1233 		}
1234 		break;
1235 #endif
1236 
1237 	default:
1238 		return EAFNOSUPPORT;
1239 	}
1240 	return 0;
1241 }
1242 
1243 /*
1244  * Add an Ethernet multicast address or range of addresses to the list for a
1245  * given interface.
1246  */
1247 int
1248 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec)
1249 {
1250 	struct ether_multi *enm;
1251 	u_char addrlo[ETHER_ADDR_LEN];
1252 	u_char addrhi[ETHER_ADDR_LEN];
1253 	int s = splnet(), error;
1254 
1255 	error = ether_multiaddr(sa, addrlo, addrhi);
1256 	if (error != 0) {
1257 		splx(s);
1258 		return error;
1259 	}
1260 
1261 	/*
1262 	 * Verify that we have valid Ethernet multicast addresses.
1263 	 */
1264 	if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) {
1265 		splx(s);
1266 		return EINVAL;
1267 	}
1268 	/*
1269 	 * See if the address range is already in the list.
1270 	 */
1271 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1272 	if (enm != NULL) {
1273 		/*
1274 		 * Found it; just increment the reference count.
1275 		 */
1276 		++enm->enm_refcount;
1277 		splx(s);
1278 		return 0;
1279 	}
1280 	/*
1281 	 * New address or range; malloc a new multicast record
1282 	 * and link it into the interface's multicast list.
1283 	 */
1284 	enm = (struct ether_multi *)malloc(sizeof(*enm), M_IFMADDR, M_NOWAIT);
1285 	if (enm == NULL) {
1286 		splx(s);
1287 		return ENOBUFS;
1288 	}
1289 	memcpy(enm->enm_addrlo, addrlo, 6);
1290 	memcpy(enm->enm_addrhi, addrhi, 6);
1291 	enm->enm_refcount = 1;
1292 	LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list);
1293 	ec->ec_multicnt++;
1294 	splx(s);
1295 	/*
1296 	 * Return ENETRESET to inform the driver that the list has changed
1297 	 * and its reception filter should be adjusted accordingly.
1298 	 */
1299 	return ENETRESET;
1300 }
1301 
1302 /*
1303  * Delete a multicast address record.
1304  */
1305 int
1306 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec)
1307 {
1308 	struct ether_multi *enm;
1309 	u_char addrlo[ETHER_ADDR_LEN];
1310 	u_char addrhi[ETHER_ADDR_LEN];
1311 	int s = splnet(), error;
1312 
1313 	error = ether_multiaddr(sa, addrlo, addrhi);
1314 	if (error != 0) {
1315 		splx(s);
1316 		return (error);
1317 	}
1318 
1319 	/*
1320 	 * Look ur the address in our list.
1321 	 */
1322 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1323 	if (enm == NULL) {
1324 		splx(s);
1325 		return (ENXIO);
1326 	}
1327 	if (--enm->enm_refcount != 0) {
1328 		/*
1329 		 * Still some claims to this record.
1330 		 */
1331 		splx(s);
1332 		return (0);
1333 	}
1334 	/*
1335 	 * No remaining claims to this record; unlink and free it.
1336 	 */
1337 	LIST_REMOVE(enm, enm_list);
1338 	free(enm, M_IFMADDR);
1339 	ec->ec_multicnt--;
1340 	splx(s);
1341 	/*
1342 	 * Return ENETRESET to inform the driver that the list has changed
1343 	 * and its reception filter should be adjusted accordingly.
1344 	 */
1345 	return (ENETRESET);
1346 }
1347 
1348 void
1349 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb)
1350 {
1351 	ec->ec_ifflags_cb = cb;
1352 }
1353 
1354 /*
1355  * Common ioctls for Ethernet interfaces.  Note, we must be
1356  * called at splnet().
1357  */
1358 int
1359 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data)
1360 {
1361 	struct ethercom *ec = (void *) ifp;
1362 	struct eccapreq *eccr;
1363 	struct ifreq *ifr = (struct ifreq *)data;
1364 	struct if_laddrreq *iflr = data;
1365 	const struct sockaddr_dl *sdl;
1366 	static const uint8_t zero[ETHER_ADDR_LEN];
1367 	int error;
1368 
1369 	switch (cmd) {
1370 	case SIOCINITIFADDR:
1371 	    {
1372 		struct ifaddr *ifa = (struct ifaddr *)data;
1373 		if (ifa->ifa_addr->sa_family != AF_LINK
1374 		    && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) !=
1375 		       (IFF_UP|IFF_RUNNING)) {
1376 			ifp->if_flags |= IFF_UP;
1377 			if ((error = (*ifp->if_init)(ifp)) != 0)
1378 				return error;
1379 		}
1380 #ifdef INET
1381 		if (ifa->ifa_addr->sa_family == AF_INET)
1382 			arp_ifinit(ifp, ifa);
1383 #endif /* INET */
1384 		return 0;
1385 	    }
1386 
1387 	case SIOCSIFMTU:
1388 	    {
1389 		int maxmtu;
1390 
1391 		if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU)
1392 			maxmtu = ETHERMTU_JUMBO;
1393 		else
1394 			maxmtu = ETHERMTU;
1395 
1396 		if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu)
1397 			return EINVAL;
1398 		else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET)
1399 			return error;
1400 		else if (ifp->if_flags & IFF_UP) {
1401 			/* Make sure the device notices the MTU change. */
1402 			return (*ifp->if_init)(ifp);
1403 		} else
1404 			return 0;
1405 	    }
1406 
1407 	case SIOCSIFFLAGS:
1408 		if ((error = ifioctl_common(ifp, cmd, data)) != 0)
1409 			return error;
1410 		switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) {
1411 		case IFF_RUNNING:
1412 			/*
1413 			 * If interface is marked down and it is running,
1414 			 * then stop and disable it.
1415 			 */
1416 			(*ifp->if_stop)(ifp, 1);
1417 			break;
1418 		case IFF_UP:
1419 			/*
1420 			 * If interface is marked up and it is stopped, then
1421 			 * start it.
1422 			 */
1423 			return (*ifp->if_init)(ifp);
1424 		case IFF_UP|IFF_RUNNING:
1425 			error = 0;
1426 			if (ec->ec_ifflags_cb == NULL ||
1427 			    (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) {
1428 				/*
1429 				 * Reset the interface to pick up
1430 				 * changes in any other flags that
1431 				 * affect the hardware state.
1432 				 */
1433 				return (*ifp->if_init)(ifp);
1434 			} else
1435 				return error;
1436 		case 0:
1437 			break;
1438 		}
1439 		return 0;
1440 	case SIOCGETHERCAP:
1441 		eccr = (struct eccapreq *)data;
1442 		eccr->eccr_capabilities = ec->ec_capabilities;
1443 		eccr->eccr_capenable = ec->ec_capenable;
1444 		return 0;
1445 	case SIOCADDMULTI:
1446 		return ether_addmulti(ifreq_getaddr(cmd, ifr), ec);
1447 	case SIOCDELMULTI:
1448 		return ether_delmulti(ifreq_getaddr(cmd, ifr), ec);
1449 	case SIOCSIFMEDIA:
1450 	case SIOCGIFMEDIA:
1451 		if (ec->ec_mii == NULL)
1452 			return ENOTTY;
1453 		return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd);
1454 	case SIOCALIFADDR:
1455 		sdl = satocsdl(sstocsa(&iflr->addr));
1456 		if (sdl->sdl_family != AF_LINK)
1457 			;
1458 		else if (ETHER_IS_MULTICAST(CLLADDR(sdl)))
1459 			return EINVAL;
1460 		else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0)
1461 			return EINVAL;
1462 		/*FALLTHROUGH*/
1463 	default:
1464 		return ifioctl_common(ifp, cmd, data);
1465 	}
1466 	return 0;
1467 }
1468 
1469 static int
1470 ether_multicast_sysctl(SYSCTLFN_ARGS)
1471 {
1472 	struct ether_multi *enm;
1473 	struct ether_multi_sysctl addr;
1474 	struct ifnet *ifp;
1475 	struct ethercom *ec;
1476 	int error;
1477 	size_t written;
1478 
1479 	if (namelen != 1)
1480 		return EINVAL;
1481 
1482 	ifp = if_byindex(name[0]);
1483 	if (ifp == NULL)
1484 		return ENODEV;
1485 	if (ifp->if_type != IFT_ETHER) {
1486 		*oldlenp = 0;
1487 		return 0;
1488 	}
1489 	ec = (struct ethercom *)ifp;
1490 
1491 	if (oldp == NULL) {
1492 		*oldlenp = ec->ec_multicnt * sizeof(addr);
1493 		return 0;
1494 	}
1495 
1496 	memset(&addr, 0, sizeof(addr));
1497 	error = 0;
1498 	written = 0;
1499 
1500 	LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) {
1501 		if (written + sizeof(addr) > *oldlenp)
1502 			break;
1503 		addr.enm_refcount = enm->enm_refcount;
1504 		memcpy(addr.enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN);
1505 		memcpy(addr.enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN);
1506 		error = sysctl_copyout(l, &addr, oldp, sizeof(addr));
1507 		if (error)
1508 			break;
1509 		written += sizeof(addr);
1510 		oldp = (char *)oldp + sizeof(addr);
1511 	}
1512 
1513 	*oldlenp = written;
1514 	return error;
1515 }
1516 
1517 SYSCTL_SETUP(sysctl_net_ether_setup, "sysctl net.ether subtree setup")
1518 {
1519 	const struct sysctlnode *rnode = NULL;
1520 
1521 	sysctl_createv(clog, 0, NULL, &rnode,
1522 		       CTLFLAG_PERMANENT,
1523 		       CTLTYPE_NODE, "ether",
1524 		       SYSCTL_DESCR("Ethernet-specific information"),
1525 		       NULL, 0, NULL, 0,
1526 		       CTL_NET, CTL_CREATE, CTL_EOL);
1527 
1528 	sysctl_createv(clog, 0, &rnode, NULL,
1529 		       CTLFLAG_PERMANENT,
1530 		       CTLTYPE_NODE, "multicast",
1531 		       SYSCTL_DESCR("multicast addresses"),
1532 		       ether_multicast_sysctl, 0, NULL, 0,
1533 		       CTL_CREATE, CTL_EOL);
1534 }
1535