xref: /netbsd-src/sys/net/if_ethersubr.c (revision b04b4ec44f02436f5b5851e3b56f4dc4b018e051)
1 /*	$NetBSD: if_ethersubr.c,v 1.211 2015/08/12 02:20:31 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 1.211 2015/08/12 02:20:31 ozaki-r Exp $");
65 
66 #include "opt_inet.h"
67 #include "opt_atalk.h"
68 #include "opt_mbuftrace.h"
69 #include "opt_mpls.h"
70 #include "opt_gateway.h"
71 #include "opt_pppoe.h"
72 #include "opt_net_mpsafe.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/sysctl.h>
80 #include <sys/malloc.h>
81 #include <sys/mbuf.h>
82 #include <sys/mutex.h>
83 #include <sys/ioctl.h>
84 #include <sys/errno.h>
85 #include <sys/device.h>
86 #include <sys/rnd.h>
87 #include <sys/rndsource.h>
88 
89 #include <net/if.h>
90 #include <net/netisr.h>
91 #include <net/route.h>
92 #include <net/if_llc.h>
93 #include <net/if_dl.h>
94 #include <net/if_types.h>
95 #include <net/pktqueue.h>
96 
97 #include <net/if_media.h>
98 #include <dev/mii/mii.h>
99 #include <dev/mii/miivar.h>
100 
101 #if NARP == 0
102 /*
103  * XXX there should really be a way to issue this warning from within config(8)
104  */
105 #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.
106 #endif
107 
108 #include <net/bpf.h>
109 
110 #include <net/if_ether.h>
111 #include <net/if_vlanvar.h>
112 
113 #if NPPPOE > 0
114 #include <net/if_pppoe.h>
115 #endif
116 
117 #if NAGR > 0
118 #include <net/agr/ieee8023_slowprotocols.h>	/* XXX */
119 #include <net/agr/ieee8023ad.h>
120 #include <net/agr/if_agrvar.h>
121 #endif
122 
123 #if NBRIDGE > 0
124 #include <net/if_bridgevar.h>
125 #endif
126 
127 #include <netinet/in.h>
128 #ifdef INET
129 #include <netinet/in_var.h>
130 #endif
131 #include <netinet/if_inarp.h>
132 
133 #ifdef INET6
134 #ifndef INET
135 #include <netinet/in.h>
136 #endif
137 #include <netinet6/in6_var.h>
138 #include <netinet6/nd6.h>
139 #endif
140 
141 
142 #include "carp.h"
143 #if NCARP > 0
144 #include <netinet/ip_carp.h>
145 #endif
146 
147 #ifdef NETATALK
148 #include <netatalk/at.h>
149 #include <netatalk/at_var.h>
150 #include <netatalk/at_extern.h>
151 
152 #define llc_snap_org_code llc_un.type_snap.org_code
153 #define llc_snap_ether_type llc_un.type_snap.ether_type
154 
155 extern u_char	at_org_code[3];
156 extern u_char	aarp_org_code[3];
157 #endif /* NETATALK */
158 
159 #ifdef MPLS
160 #include <netmpls/mpls.h>
161 #include <netmpls/mpls_var.h>
162 #endif
163 
164 static struct timeval bigpktppslim_last;
165 static int bigpktppslim = 2;	/* XXX */
166 static int bigpktpps_count;
167 static kmutex_t bigpktpps_lock __cacheline_aligned;
168 
169 
170 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] =
171     { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
172 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] =
173     { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 };
174 #define senderr(e) { error = (e); goto bad;}
175 
176 static	int ether_output(struct ifnet *, struct mbuf *,
177 	    const struct sockaddr *, struct rtentry *);
178 
179 /*
180  * Ethernet output routine.
181  * Encapsulate a packet of type family for the local net.
182  * Assumes that ifp is actually pointer to ethercom structure.
183  */
184 static int
185 ether_output(struct ifnet * const ifp0, struct mbuf * const m0,
186 	const struct sockaddr * const dst,
187 	struct rtentry *rt)
188 {
189 	uint16_t etype = 0;
190 	int error = 0, hdrcmplt = 0;
191  	uint8_t esrc[6], edst[6];
192 	struct mbuf *m = m0;
193 	struct mbuf *mcopy = NULL;
194 	struct ether_header *eh;
195 	struct ifnet *ifp = ifp0;
196 	ALTQ_DECL(struct altq_pktattr pktattr;)
197 #ifdef INET
198 	struct arphdr *ah;
199 #endif /* INET */
200 #ifdef NETATALK
201 	struct at_ifaddr *aa;
202 #endif /* NETATALK */
203 
204 #ifndef NET_MPSAFE
205 	KASSERT(KERNEL_LOCKED_P());
206 #endif
207 
208 #ifdef MBUFTRACE
209 	m_claimm(m, ifp->if_mowner);
210 #endif
211 
212 #if NCARP > 0
213 	if (ifp->if_type == IFT_CARP) {
214 		struct ifaddr *ifa;
215 
216 		/* loop back if this is going to the carp interface */
217 		if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP &&
218 		    (ifa = ifa_ifwithaddr(dst)) != NULL &&
219 		    ifa->ifa_ifp == ifp0)
220 			return looutput(ifp0, m, dst, rt);
221 
222 		ifp = ifp->if_carpdev;
223 		/* ac = (struct arpcom *)ifp; */
224 
225 		if ((ifp0->if_flags & (IFF_UP|IFF_RUNNING)) !=
226 		    (IFF_UP|IFF_RUNNING))
227 			senderr(ENETDOWN);
228 	}
229 #endif /* NCARP > 0 */
230 
231 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
232 		senderr(ENETDOWN);
233 
234 	switch (dst->sa_family) {
235 
236 #ifdef INET
237 	case AF_INET:
238 		if (m->m_flags & M_BCAST)
239 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
240 		else if (m->m_flags & M_MCAST)
241 			ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst);
242 		else if (!arpresolve(ifp, rt, m, dst, edst))
243 			return (0);	/* if not yet resolved */
244 		/* If broadcasting on a simplex interface, loopback a copy */
245 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
246 			mcopy = m_copy(m, 0, (int)M_COPYALL);
247 		etype = htons(ETHERTYPE_IP);
248 		break;
249 
250 	case AF_ARP:
251 		ah = mtod(m, struct arphdr *);
252 		if (m->m_flags & M_BCAST)
253 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
254 		else {
255 			void *tha = ar_tha(ah);
256 
257 			if (tha == NULL) {
258 				/* fake with ARPHDR_IEEE1394 */
259 				return 0;
260 			}
261 			memcpy(edst, tha, sizeof(edst));
262 		}
263 
264 		ah->ar_hrd = htons(ARPHRD_ETHER);
265 
266 		switch (ntohs(ah->ar_op)) {
267 		case ARPOP_REVREQUEST:
268 		case ARPOP_REVREPLY:
269 			etype = htons(ETHERTYPE_REVARP);
270 			break;
271 
272 		case ARPOP_REQUEST:
273 		case ARPOP_REPLY:
274 		default:
275 			etype = htons(ETHERTYPE_ARP);
276 		}
277 
278 		break;
279 #endif
280 #ifdef INET6
281 	case AF_INET6:
282 		if (!nd6_storelladdr(ifp, rt, m, dst, edst, sizeof(edst))){
283 			/* something bad happened */
284 			return (0);
285 		}
286 		etype = htons(ETHERTYPE_IPV6);
287 		break;
288 #endif
289 #ifdef NETATALK
290     case AF_APPLETALK:
291 		if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) {
292 #ifdef NETATALKDEBUG
293 			printf("aarpresolv failed\n");
294 #endif /* NETATALKDEBUG */
295 			return (0);
296 		}
297 		/*
298 		 * ifaddr is the first thing in at_ifaddr
299 		 */
300 		aa = (struct at_ifaddr *) at_ifawithnet(
301 		    (const struct sockaddr_at *)dst, ifp);
302 		if (aa == NULL)
303 		    goto bad;
304 
305 		/*
306 		 * In the phase 2 case, we need to prepend an mbuf for the
307 		 * llc header.  Since we must preserve the value of m,
308 		 * which is passed to us by value, we m_copy() the first
309 		 * mbuf, and use it for our llc header.
310 		 */
311 		if (aa->aa_flags & AFA_PHASE2) {
312 			struct llc llc;
313 
314 			M_PREPEND(m, sizeof(struct llc), M_DONTWAIT);
315 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
316 			llc.llc_control = LLC_UI;
317 			memcpy(llc.llc_snap_org_code, at_org_code,
318 			    sizeof(llc.llc_snap_org_code));
319 			llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK);
320 			memcpy(mtod(m, void *), &llc, sizeof(struct llc));
321 		} else {
322 			etype = htons(ETHERTYPE_ATALK);
323 		}
324 		break;
325 #endif /* NETATALK */
326 	case pseudo_AF_HDRCMPLT:
327 		hdrcmplt = 1;
328 		memcpy(esrc,
329 		    ((const struct ether_header *)dst->sa_data)->ether_shost,
330 		    sizeof(esrc));
331 		/* FALLTHROUGH */
332 
333 	case AF_UNSPEC:
334  		memcpy(edst,
335 		    ((const struct ether_header *)dst->sa_data)->ether_dhost,
336 		    sizeof(edst));
337 		/* AF_UNSPEC doesn't swap the byte order of the ether_type. */
338 		etype = ((const struct ether_header *)dst->sa_data)->ether_type;
339 		break;
340 
341 	default:
342 		printf("%s: can't handle af%d\n", ifp->if_xname,
343 			dst->sa_family);
344 		senderr(EAFNOSUPPORT);
345 	}
346 
347 #ifdef MPLS
348 	{
349 		struct m_tag *mtag;
350 		mtag = m_tag_find(m, PACKET_TAG_MPLS, NULL);
351 		if (mtag != NULL) {
352 			/* Having the tag itself indicates it's MPLS */
353 			etype = htons(ETHERTYPE_MPLS);
354 			m_tag_delete(m, mtag);
355 		}
356 	}
357 #endif
358 
359 	if (mcopy)
360 		(void)looutput(ifp, mcopy, dst, rt);
361 
362 	/* If no ether type is set, this must be a 802.2 formatted packet.
363 	 */
364 	if (etype == 0)
365 		etype = htons(m->m_pkthdr.len);
366 	/*
367 	 * Add local net header.  If no space in first mbuf,
368 	 * allocate another.
369 	 */
370 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
371 	if (m == 0)
372 		senderr(ENOBUFS);
373 	eh = mtod(m, struct ether_header *);
374 	/* Note: etype is already in network byte order. */
375 	(void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
376  	memcpy(eh->ether_dhost, edst, sizeof(edst));
377 	if (hdrcmplt)
378 		memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost));
379 	else
380 	 	memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl),
381 		    sizeof(eh->ether_shost));
382 
383 #if NCARP > 0
384 	if (ifp0 != ifp && ifp0->if_type == IFT_CARP) {
385 	 	memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl),
386 		    sizeof(eh->ether_shost));
387 	}
388 #endif /* NCARP > 0 */
389 
390 	if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0)
391 		return (error);
392 	if (m == NULL)
393 		return (0);
394 
395 #if NBRIDGE > 0
396 	/*
397 	 * Bridges require special output handling.
398 	 */
399 	if (ifp->if_bridge)
400 		return (bridge_output(ifp, m, NULL, NULL));
401 #endif
402 
403 #if NCARP > 0
404 	if (ifp != ifp0)
405 		ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN;
406 #endif /* NCARP > 0 */
407 
408 #ifdef ALTQ
409 	/*
410 	 * If ALTQ is enabled on the parent interface, do
411 	 * classification; the queueing discipline might not
412 	 * require classification, but might require the
413 	 * address family/header pointer in the pktattr.
414 	 */
415 	if (ALTQ_IS_ENABLED(&ifp->if_snd))
416 		altq_etherclassify(&ifp->if_snd, m, &pktattr);
417 #endif
418 	return ifq_enqueue(ifp, m ALTQ_COMMA ALTQ_DECL(&pktattr));
419 
420 bad:
421 	if (m)
422 		m_freem(m);
423 	return (error);
424 }
425 
426 #ifdef ALTQ
427 /*
428  * This routine is a slight hack to allow a packet to be classified
429  * if the Ethernet headers are present.  It will go away when ALTQ's
430  * classification engine understands link headers.
431  */
432 void
433 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
434     struct altq_pktattr *pktattr)
435 {
436 	struct ether_header *eh;
437 	uint16_t ether_type;
438 	int hlen, af, hdrsize;
439 	void *hdr;
440 
441 	hlen = ETHER_HDR_LEN;
442 	eh = mtod(m, struct ether_header *);
443 
444 	ether_type = htons(eh->ether_type);
445 
446 	if (ether_type < ETHERMTU) {
447 		/* LLC/SNAP */
448 		struct llc *llc = (struct llc *)(eh + 1);
449 		hlen += 8;
450 
451 		if (m->m_len < hlen ||
452 		    llc->llc_dsap != LLC_SNAP_LSAP ||
453 		    llc->llc_ssap != LLC_SNAP_LSAP ||
454 		    llc->llc_control != LLC_UI) {
455 			/* Not SNAP. */
456 			goto bad;
457 		}
458 
459 		ether_type = htons(llc->llc_un.type_snap.ether_type);
460 	}
461 
462 	switch (ether_type) {
463 	case ETHERTYPE_IP:
464 		af = AF_INET;
465 		hdrsize = 20;		/* sizeof(struct ip) */
466 		break;
467 
468 	case ETHERTYPE_IPV6:
469 		af = AF_INET6;
470 		hdrsize = 40;		/* sizeof(struct ip6_hdr) */
471 		break;
472 
473 	default:
474 		af = AF_UNSPEC;
475 		hdrsize = 0;
476 		break;
477 	}
478 
479 	while (m->m_len <= hlen) {
480 		hlen -= m->m_len;
481 		m = m->m_next;
482 	}
483 	if (m->m_len < (hlen + hdrsize)) {
484 		/*
485 		 * protocol header not in a single mbuf.
486 		 * We can't cope with this situation right
487 		 * now (but it shouldn't ever happen, really, anyhow).
488 		 */
489 #ifdef DEBUG
490 		printf("altq_etherclassify: headers span multiple mbufs: "
491 		    "%d < %d\n", m->m_len, (hlen + hdrsize));
492 #endif
493 		goto bad;
494 	}
495 
496 	m->m_data += hlen;
497 	m->m_len -= hlen;
498 
499 	hdr = mtod(m, void *);
500 
501 	if (ALTQ_NEEDS_CLASSIFY(ifq))
502 		pktattr->pattr_class =
503 		    (*ifq->altq_classify)(ifq->altq_clfier, m, af);
504 	pktattr->pattr_af = af;
505 	pktattr->pattr_hdr = hdr;
506 
507 	m->m_data -= hlen;
508 	m->m_len += hlen;
509 
510 	return;
511 
512  bad:
513 	pktattr->pattr_class = NULL;
514 	pktattr->pattr_hdr = NULL;
515 	pktattr->pattr_af = AF_UNSPEC;
516 }
517 #endif /* ALTQ */
518 
519 /*
520  * Process a received Ethernet packet;
521  * the packet is in the mbuf chain m with
522  * the ether header.
523  */
524 void
525 ether_input(struct ifnet *ifp, struct mbuf *m)
526 {
527 	struct ethercom *ec = (struct ethercom *) ifp;
528 	pktqueue_t *pktq = NULL;
529 	struct ifqueue *inq = NULL;
530 	uint16_t etype;
531 	struct ether_header *eh;
532 	size_t ehlen;
533 	static int earlypkts;
534 	int isr = 0;
535 #if defined (LLC) || defined(NETATALK)
536 	struct llc *l;
537 #endif
538 
539 	if ((ifp->if_flags & IFF_UP) == 0) {
540 		m_freem(m);
541 		return;
542 	}
543 
544 #ifdef MBUFTRACE
545 	m_claimm(m, &ec->ec_rx_mowner);
546 #endif
547 	eh = mtod(m, struct ether_header *);
548 	etype = ntohs(eh->ether_type);
549 	ehlen = sizeof(*eh);
550 
551 	if(__predict_false(earlypkts < 100 || !rnd_initial_entropy)) {
552 		rnd_add_data(NULL, eh, ehlen, 0);
553 		earlypkts++;
554 	}
555 
556 	/*
557 	 * Determine if the packet is within its size limits.
558 	 */
559 	if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len >
560 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
561 		mutex_enter(&bigpktpps_lock);
562 		if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count,
563 			    bigpktppslim)) {
564 			printf("%s: discarding oversize frame (len=%d)\n",
565 			    ifp->if_xname, m->m_pkthdr.len);
566 		}
567 		mutex_exit(&bigpktpps_lock);
568 		m_freem(m);
569 		return;
570 	}
571 
572 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
573 		/*
574 		 * If this is not a simplex interface, drop the packet
575 		 * if it came from us.
576 		 */
577 		if ((ifp->if_flags & IFF_SIMPLEX) == 0 &&
578 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost,
579 		    ETHER_ADDR_LEN) == 0) {
580 			m_freem(m);
581 			return;
582 		}
583 
584 		if (memcmp(etherbroadcastaddr,
585 		    eh->ether_dhost, ETHER_ADDR_LEN) == 0)
586 			m->m_flags |= M_BCAST;
587 		else
588 			m->m_flags |= M_MCAST;
589 		ifp->if_imcasts++;
590 	}
591 
592 	/* If the CRC is still on the packet, trim it off. */
593 	if (m->m_flags & M_HASFCS) {
594 		m_adj(m, -ETHER_CRC_LEN);
595 		m->m_flags &= ~M_HASFCS;
596 	}
597 
598 	ifp->if_ibytes += m->m_pkthdr.len;
599 
600 #if NCARP > 0
601 	if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) {
602 		/*
603 		 * clear M_PROMISC, in case the packets comes from a
604 		 * vlan
605 		 */
606 		m->m_flags &= ~M_PROMISC;
607 		if (carp_input(m, (uint8_t *)&eh->ether_shost,
608 		    (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0)
609 			return;
610 	}
611 #endif /* NCARP > 0 */
612 	if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 &&
613 	    (ifp->if_flags & IFF_PROMISC) != 0 &&
614 	    memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost,
615 		   ETHER_ADDR_LEN) != 0) {
616 		m->m_flags |= M_PROMISC;
617 	}
618 
619 	if ((m->m_flags & M_PROMISC) == 0) {
620 		if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0)
621 			return;
622 		if (m == NULL)
623 			return;
624 
625 		eh = mtod(m, struct ether_header *);
626 		etype = ntohs(eh->ether_type);
627 		ehlen = sizeof(*eh);
628 	}
629 
630 #if NAGR > 0
631 	if (ifp->if_agrprivate &&
632 	    __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) {
633 		m->m_flags &= ~M_PROMISC;
634 		agr_input(ifp, m);
635 		return;
636 	}
637 #endif /* NAGR > 0 */
638 
639 	/*
640 	 * If VLANs are configured on the interface, check to
641 	 * see if the device performed the decapsulation and
642 	 * provided us with the tag.
643 	 */
644 	if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) {
645 #if NVLAN > 0
646 		/*
647 		 * vlan_input() will either recursively call ether_input()
648 		 * or drop the packet.
649 		 */
650 		vlan_input(ifp, m);
651 #else
652 		m_freem(m);
653 #endif
654 		return;
655 	}
656 
657 	/*
658 	 * Handle protocols that expect to have the Ethernet header
659 	 * (and possibly FCS) intact.
660 	 */
661 	switch (etype) {
662 	case ETHERTYPE_VLAN: {
663 		struct ether_vlan_header *evl = (void *)eh;
664 		/*
665 		 * If there is a tag of 0, then the VLAN header was probably
666 		 * just being used to store the priority.  Extract the ether
667 		 * type, and if IP or IPV6, let them deal with it.
668 		 */
669 		if (m->m_len <= sizeof(*evl)
670 		    && EVL_VLANOFTAG(evl->evl_tag) == 0) {
671 			etype = ntohs(evl->evl_proto);
672 			ehlen = sizeof(*evl);
673 			if ((m->m_flags & M_PROMISC) == 0
674 			    && (etype == ETHERTYPE_IP
675 				|| etype == ETHERTYPE_IPV6))
676 				break;
677 		}
678 #if NVLAN > 0
679 		/*
680 		 * vlan_input() will either recursively call ether_input()
681 		 * or drop the packet.
682 		 */
683 		if (((struct ethercom *)ifp)->ec_nvlans != 0)
684 			vlan_input(ifp, m);
685 		else
686 #endif /* NVLAN > 0 */
687 			m_freem(m);
688 		return;
689 	}
690 #if NPPPOE > 0
691 	case ETHERTYPE_PPPOEDISC:
692 	case ETHERTYPE_PPPOE:
693 		if (m->m_flags & M_PROMISC) {
694 			m_freem(m);
695 			return;
696 		}
697 #ifndef PPPOE_SERVER
698 		if (m->m_flags & (M_MCAST | M_BCAST)) {
699 			m_freem(m);
700 			return;
701 		}
702 #endif
703 
704 		if (etype == ETHERTYPE_PPPOEDISC)
705 			inq = &ppoediscinq;
706 		else
707 			inq = &ppoeinq;
708 		if (IF_QFULL(inq)) {
709 			IF_DROP(inq);
710 			m_freem(m);
711 		} else {
712 			IF_ENQUEUE(inq, m);
713 			softint_schedule(pppoe_softintr);
714 		}
715 		return;
716 #endif /* NPPPOE > 0 */
717 	case ETHERTYPE_SLOWPROTOCOLS: {
718 		uint8_t subtype;
719 
720 #if defined(DIAGNOSTIC)
721 		if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) {
722 			panic("ether_input: too short slow protocol packet");
723 		}
724 #endif
725 		m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype);
726 		switch (subtype) {
727 #if NAGR > 0
728 		case SLOWPROTOCOLS_SUBTYPE_LACP:
729 			if (ifp->if_agrprivate) {
730 				ieee8023ad_lacp_input(ifp, m);
731 				return;
732 			}
733 			break;
734 
735 		case SLOWPROTOCOLS_SUBTYPE_MARKER:
736 			if (ifp->if_agrprivate) {
737 				ieee8023ad_marker_input(ifp, m);
738 				return;
739 			}
740 			break;
741 #endif /* NAGR > 0 */
742 		default:
743 			if (subtype == 0 || subtype > 10) {
744 				/* illegal value */
745 				m_freem(m);
746 				return;
747 			}
748 			/* unknown subtype */
749 			break;
750 		}
751 		/* FALLTHROUGH */
752 	}
753 	default:
754 		if (m->m_flags & M_PROMISC) {
755 			m_freem(m);
756 			return;
757 		}
758 	}
759 
760 	/* If the CRC is still on the packet, trim it off. */
761 	if (m->m_flags & M_HASFCS) {
762 		m_adj(m, -ETHER_CRC_LEN);
763 		m->m_flags &= ~M_HASFCS;
764 	}
765 
766 	if (etype > ETHERMTU + sizeof (struct ether_header)) {
767 		/* Strip off the Ethernet header. */
768 		m_adj(m, ehlen);
769 
770 		switch (etype) {
771 #ifdef INET
772 		case ETHERTYPE_IP:
773 #ifdef GATEWAY
774 			if (ipflow_fastforward(m))
775 				return;
776 #endif
777 			pktq = ip_pktq;
778 			break;
779 
780 		case ETHERTYPE_ARP:
781 			isr = NETISR_ARP;
782 			inq = &arpintrq;
783 			break;
784 
785 		case ETHERTYPE_REVARP:
786 			revarpinput(m);	/* XXX queue? */
787 			return;
788 #endif
789 #ifdef INET6
790 		case ETHERTYPE_IPV6:
791 			if (__predict_false(!in6_present)) {
792 				m_freem(m);
793 				return;
794 			}
795 #ifdef GATEWAY
796 			if (ip6flow_fastforward(&m))
797 				return;
798 #endif
799 			pktq = ip6_pktq;
800 			break;
801 #endif
802 #ifdef NETATALK
803 		case ETHERTYPE_ATALK:
804 			isr = NETISR_ATALK;
805 			inq = &atintrq1;
806 			break;
807 		case ETHERTYPE_AARP:
808 			/* probably this should be done with a NETISR as well */
809 			aarpinput(ifp, m); /* XXX */
810 			return;
811 #endif /* NETATALK */
812 #ifdef MPLS
813 		case ETHERTYPE_MPLS:
814 			isr = NETISR_MPLS;
815 			inq = &mplsintrq;
816 			break;
817 #endif
818 		default:
819 			m_freem(m);
820 			return;
821 		}
822 	} else {
823 #if defined (LLC) || defined (NETATALK)
824 		l = (struct llc *)(eh+1);
825 		switch (l->llc_dsap) {
826 #ifdef NETATALK
827 		case LLC_SNAP_LSAP:
828 			switch (l->llc_control) {
829 			case LLC_UI:
830 				if (l->llc_ssap != LLC_SNAP_LSAP) {
831 					goto dropanyway;
832 				}
833 
834 				if (memcmp(&(l->llc_snap_org_code)[0],
835 				    at_org_code, sizeof(at_org_code)) == 0 &&
836 				    ntohs(l->llc_snap_ether_type) ==
837 				    ETHERTYPE_ATALK) {
838 					inq = &atintrq2;
839 					m_adj(m, sizeof(struct ether_header)
840 					    + sizeof(struct llc));
841 					isr = NETISR_ATALK;
842 					break;
843 				}
844 
845 				if (memcmp(&(l->llc_snap_org_code)[0],
846 				    aarp_org_code,
847 				    sizeof(aarp_org_code)) == 0 &&
848 				    ntohs(l->llc_snap_ether_type) ==
849 				    ETHERTYPE_AARP) {
850 					m_adj( m, sizeof(struct ether_header)
851 					    + sizeof(struct llc));
852 					aarpinput(ifp, m); /* XXX */
853 				    return;
854 				}
855 
856 			default:
857 				goto dropanyway;
858 			}
859 			break;
860 		dropanyway:
861 #endif
862 		default:
863 			m_freem(m);
864 			return;
865 		}
866 #else /* ISO || LLC || NETATALK*/
867 		m_freem(m);
868 		return;
869 #endif /* ISO || LLC || NETATALK*/
870 	}
871 
872 	if (__predict_true(pktq)) {
873 		const uint32_t h = pktq_rps_hash(m);
874 		if (__predict_false(!pktq_enqueue(pktq, m, h))) {
875 			m_freem(m);
876 		}
877 		return;
878 	}
879 
880 	if (__predict_false(!inq)) {
881 		/* Should not happen. */
882 		m_freem(m);
883 		return;
884 	}
885 	if (IF_QFULL(inq)) {
886 		IF_DROP(inq);
887 		m_freem(m);
888 	} else {
889 		IF_ENQUEUE(inq, m);
890 		schednetisr(isr);
891 	}
892 }
893 
894 /*
895  * Convert Ethernet address to printable (loggable) representation.
896  */
897 char *
898 ether_sprintf(const u_char *ap)
899 {
900 	static char etherbuf[3 * ETHER_ADDR_LEN];
901 	return ether_snprintf(etherbuf, sizeof(etherbuf), ap);
902 }
903 
904 char *
905 ether_snprintf(char *buf, size_t len, const u_char *ap)
906 {
907 	char *cp = buf;
908 	size_t i;
909 
910 	for (i = 0; i < len / 3; i++) {
911 		*cp++ = hexdigits[*ap >> 4];
912 		*cp++ = hexdigits[*ap++ & 0xf];
913 		*cp++ = ':';
914 	}
915 	*--cp = '\0';
916 	return buf;
917 }
918 
919 /*
920  * Perform common duties while attaching to interface list
921  */
922 void
923 ether_ifattach(struct ifnet *ifp, const uint8_t *lla)
924 {
925 	struct ethercom *ec = (struct ethercom *)ifp;
926 
927 	ifp->if_type = IFT_ETHER;
928 	ifp->if_hdrlen = ETHER_HDR_LEN;
929 	ifp->if_dlt = DLT_EN10MB;
930 	ifp->if_mtu = ETHERMTU;
931 	ifp->if_output = ether_output;
932 	ifp->if_input = ether_input;
933 	if (ifp->if_baudrate == 0)
934 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
935 
936 	if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla));
937 
938 	LIST_INIT(&ec->ec_multiaddrs);
939 	ifp->if_broadcastaddr = etherbroadcastaddr;
940 	bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header));
941 #ifdef MBUFTRACE
942 	strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname,
943 	    sizeof(ec->ec_tx_mowner.mo_name));
944 	strlcpy(ec->ec_tx_mowner.mo_descr, "tx",
945 	    sizeof(ec->ec_tx_mowner.mo_descr));
946 	strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname,
947 	    sizeof(ec->ec_rx_mowner.mo_name));
948 	strlcpy(ec->ec_rx_mowner.mo_descr, "rx",
949 	    sizeof(ec->ec_rx_mowner.mo_descr));
950 	MOWNER_ATTACH(&ec->ec_tx_mowner);
951 	MOWNER_ATTACH(&ec->ec_rx_mowner);
952 	ifp->if_mowner = &ec->ec_tx_mowner;
953 #endif
954 }
955 
956 void
957 ether_ifdetach(struct ifnet *ifp)
958 {
959 	struct ethercom *ec = (void *) ifp;
960 	struct ether_multi *enm;
961 	int s;
962 
963 	/*
964 	 * Prevent further calls to ioctl (for example turning off
965 	 * promiscuous mode from the bridge code), which eventually can
966 	 * call if_init() which can cause panics because the interface
967 	 * is in the process of being detached. Return device not configured
968 	 * instead.
969 	 */
970 	ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio;
971 
972 #if NBRIDGE > 0
973 	if (ifp->if_bridge)
974 		bridge_ifdetach(ifp);
975 #endif
976 
977 	bpf_detach(ifp);
978 
979 #if NVLAN > 0
980 	if (ec->ec_nvlans)
981 		vlan_ifdetach(ifp);
982 #endif
983 
984 	s = splnet();
985 	while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) {
986 		LIST_REMOVE(enm, enm_list);
987 		free(enm, M_IFMADDR);
988 		ec->ec_multicnt--;
989 	}
990 	splx(s);
991 
992 	ifp->if_mowner = NULL;
993 	MOWNER_DETACH(&ec->ec_rx_mowner);
994 	MOWNER_DETACH(&ec->ec_tx_mowner);
995 }
996 
997 #if 0
998 /*
999  * This is for reference.  We have a table-driven version
1000  * of the little-endian crc32 generator, which is faster
1001  * than the double-loop.
1002  */
1003 uint32_t
1004 ether_crc32_le(const uint8_t *buf, size_t len)
1005 {
1006 	uint32_t c, crc, carry;
1007 	size_t i, j;
1008 
1009 	crc = 0xffffffffU;	/* initial value */
1010 
1011 	for (i = 0; i < len; i++) {
1012 		c = buf[i];
1013 		for (j = 0; j < 8; j++) {
1014 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
1015 			crc >>= 1;
1016 			c >>= 1;
1017 			if (carry)
1018 				crc = (crc ^ ETHER_CRC_POLY_LE);
1019 		}
1020 	}
1021 
1022 	return (crc);
1023 }
1024 #else
1025 uint32_t
1026 ether_crc32_le(const uint8_t *buf, size_t len)
1027 {
1028 	static const uint32_t crctab[] = {
1029 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1030 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1031 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1032 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1033 	};
1034 	uint32_t crc;
1035 	size_t i;
1036 
1037 	crc = 0xffffffffU;	/* initial value */
1038 
1039 	for (i = 0; i < len; i++) {
1040 		crc ^= buf[i];
1041 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1042 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1043 	}
1044 
1045 	return (crc);
1046 }
1047 #endif
1048 
1049 uint32_t
1050 ether_crc32_be(const uint8_t *buf, size_t len)
1051 {
1052 	uint32_t c, crc, carry;
1053 	size_t i, j;
1054 
1055 	crc = 0xffffffffU;	/* initial value */
1056 
1057 	for (i = 0; i < len; i++) {
1058 		c = buf[i];
1059 		for (j = 0; j < 8; j++) {
1060 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
1061 			crc <<= 1;
1062 			c >>= 1;
1063 			if (carry)
1064 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1065 		}
1066 	}
1067 
1068 	return (crc);
1069 }
1070 
1071 #ifdef INET
1072 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] =
1073     { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 };
1074 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] =
1075     { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff };
1076 #endif
1077 #ifdef INET6
1078 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] =
1079     { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 };
1080 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] =
1081     { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff };
1082 #endif
1083 
1084 /*
1085  * ether_aton implementation, not using a static buffer.
1086  */
1087 int
1088 ether_aton_r(u_char *dest, size_t len, const char *str)
1089 {
1090         const u_char *cp = (const void *)str;
1091 	u_char *ep;
1092 
1093 #define atox(c)	(((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10))
1094 
1095 	if (len < ETHER_ADDR_LEN)
1096 		return ENOSPC;
1097 
1098 	ep = dest + ETHER_ADDR_LEN;
1099 
1100 	while (*cp) {
1101                 if (!isxdigit(*cp))
1102                         return EINVAL;
1103 		*dest = atox(*cp);
1104 		cp++;
1105                 if (isxdigit(*cp)) {
1106                         *dest = (*dest << 4) | atox(*cp);
1107 			dest++;
1108 			cp++;
1109                 } else
1110 			dest++;
1111 		if (dest == ep)
1112 			return *cp == '\0' ? 0 : ENAMETOOLONG;
1113 		switch (*cp) {
1114 		case ':':
1115 		case '-':
1116 		case '.':
1117 			cp++;
1118 			break;
1119 		}
1120         }
1121 	return ENOBUFS;
1122 }
1123 
1124 /*
1125  * Convert a sockaddr into an Ethernet address or range of Ethernet
1126  * addresses.
1127  */
1128 int
1129 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN],
1130     uint8_t addrhi[ETHER_ADDR_LEN])
1131 {
1132 #ifdef INET
1133 	const struct sockaddr_in *sin;
1134 #endif /* INET */
1135 #ifdef INET6
1136 	const struct sockaddr_in6 *sin6;
1137 #endif /* INET6 */
1138 
1139 	switch (sa->sa_family) {
1140 
1141 	case AF_UNSPEC:
1142 		memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN);
1143 		memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1144 		break;
1145 
1146 #ifdef INET
1147 	case AF_INET:
1148 		sin = satocsin(sa);
1149 		if (sin->sin_addr.s_addr == INADDR_ANY) {
1150 			/*
1151 			 * An IP address of INADDR_ANY means listen to
1152 			 * or stop listening to all of the Ethernet
1153 			 * multicast addresses used for IP.
1154 			 * (This is for the sake of IP multicast routers.)
1155 			 */
1156 			memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN);
1157 			memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN);
1158 		}
1159 		else {
1160 			ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo);
1161 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1162 		}
1163 		break;
1164 #endif
1165 #ifdef INET6
1166 	case AF_INET6:
1167 		sin6 = satocsin6(sa);
1168 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1169 			/*
1170 			 * An IP6 address of 0 means listen to or stop
1171 			 * listening to all of the Ethernet multicast
1172 			 * address used for IP6.
1173 			 * (This is used for multicast routers.)
1174 			 */
1175 			memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN);
1176 			memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN);
1177 		} else {
1178 			ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo);
1179 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1180 		}
1181 		break;
1182 #endif
1183 
1184 	default:
1185 		return EAFNOSUPPORT;
1186 	}
1187 	return 0;
1188 }
1189 
1190 /*
1191  * Add an Ethernet multicast address or range of addresses to the list for a
1192  * given interface.
1193  */
1194 int
1195 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec)
1196 {
1197 	struct ether_multi *enm;
1198 	u_char addrlo[ETHER_ADDR_LEN];
1199 	u_char addrhi[ETHER_ADDR_LEN];
1200 	int s = splnet(), error;
1201 
1202 	error = ether_multiaddr(sa, addrlo, addrhi);
1203 	if (error != 0) {
1204 		splx(s);
1205 		return error;
1206 	}
1207 
1208 	/*
1209 	 * Verify that we have valid Ethernet multicast addresses.
1210 	 */
1211 	if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) {
1212 		splx(s);
1213 		return EINVAL;
1214 	}
1215 	/*
1216 	 * See if the address range is already in the list.
1217 	 */
1218 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1219 	if (enm != NULL) {
1220 		/*
1221 		 * Found it; just increment the reference count.
1222 		 */
1223 		++enm->enm_refcount;
1224 		splx(s);
1225 		return 0;
1226 	}
1227 	/*
1228 	 * New address or range; malloc a new multicast record
1229 	 * and link it into the interface's multicast list.
1230 	 */
1231 	enm = (struct ether_multi *)malloc(sizeof(*enm), M_IFMADDR, M_NOWAIT);
1232 	if (enm == NULL) {
1233 		splx(s);
1234 		return ENOBUFS;
1235 	}
1236 	memcpy(enm->enm_addrlo, addrlo, 6);
1237 	memcpy(enm->enm_addrhi, addrhi, 6);
1238 	enm->enm_refcount = 1;
1239 	LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list);
1240 	ec->ec_multicnt++;
1241 	splx(s);
1242 	/*
1243 	 * Return ENETRESET to inform the driver that the list has changed
1244 	 * and its reception filter should be adjusted accordingly.
1245 	 */
1246 	return ENETRESET;
1247 }
1248 
1249 /*
1250  * Delete a multicast address record.
1251  */
1252 int
1253 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec)
1254 {
1255 	struct ether_multi *enm;
1256 	u_char addrlo[ETHER_ADDR_LEN];
1257 	u_char addrhi[ETHER_ADDR_LEN];
1258 	int s = splnet(), error;
1259 
1260 	error = ether_multiaddr(sa, addrlo, addrhi);
1261 	if (error != 0) {
1262 		splx(s);
1263 		return (error);
1264 	}
1265 
1266 	/*
1267 	 * Look ur the address in our list.
1268 	 */
1269 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1270 	if (enm == NULL) {
1271 		splx(s);
1272 		return (ENXIO);
1273 	}
1274 	if (--enm->enm_refcount != 0) {
1275 		/*
1276 		 * Still some claims to this record.
1277 		 */
1278 		splx(s);
1279 		return (0);
1280 	}
1281 	/*
1282 	 * No remaining claims to this record; unlink and free it.
1283 	 */
1284 	LIST_REMOVE(enm, enm_list);
1285 	free(enm, M_IFMADDR);
1286 	ec->ec_multicnt--;
1287 	splx(s);
1288 	/*
1289 	 * Return ENETRESET to inform the driver that the list has changed
1290 	 * and its reception filter should be adjusted accordingly.
1291 	 */
1292 	return (ENETRESET);
1293 }
1294 
1295 void
1296 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb)
1297 {
1298 	ec->ec_ifflags_cb = cb;
1299 }
1300 
1301 /*
1302  * Common ioctls for Ethernet interfaces.  Note, we must be
1303  * called at splnet().
1304  */
1305 int
1306 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data)
1307 {
1308 	struct ethercom *ec = (void *) ifp;
1309 	struct eccapreq *eccr;
1310 	struct ifreq *ifr = (struct ifreq *)data;
1311 	struct if_laddrreq *iflr = data;
1312 	const struct sockaddr_dl *sdl;
1313 	static const uint8_t zero[ETHER_ADDR_LEN];
1314 	int error;
1315 
1316 	switch (cmd) {
1317 	case SIOCINITIFADDR:
1318 	    {
1319 		struct ifaddr *ifa = (struct ifaddr *)data;
1320 		if (ifa->ifa_addr->sa_family != AF_LINK
1321 		    && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) !=
1322 		       (IFF_UP|IFF_RUNNING)) {
1323 			ifp->if_flags |= IFF_UP;
1324 			if ((error = (*ifp->if_init)(ifp)) != 0)
1325 				return error;
1326 		}
1327 #ifdef INET
1328 		if (ifa->ifa_addr->sa_family == AF_INET)
1329 			arp_ifinit(ifp, ifa);
1330 #endif /* INET */
1331 		return 0;
1332 	    }
1333 
1334 	case SIOCSIFMTU:
1335 	    {
1336 		int maxmtu;
1337 
1338 		if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU)
1339 			maxmtu = ETHERMTU_JUMBO;
1340 		else
1341 			maxmtu = ETHERMTU;
1342 
1343 		if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu)
1344 			return EINVAL;
1345 		else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET)
1346 			return error;
1347 		else if (ifp->if_flags & IFF_UP) {
1348 			/* Make sure the device notices the MTU change. */
1349 			return (*ifp->if_init)(ifp);
1350 		} else
1351 			return 0;
1352 	    }
1353 
1354 	case SIOCSIFFLAGS:
1355 		if ((error = ifioctl_common(ifp, cmd, data)) != 0)
1356 			return error;
1357 		switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) {
1358 		case IFF_RUNNING:
1359 			/*
1360 			 * If interface is marked down and it is running,
1361 			 * then stop and disable it.
1362 			 */
1363 			(*ifp->if_stop)(ifp, 1);
1364 			break;
1365 		case IFF_UP:
1366 			/*
1367 			 * If interface is marked up and it is stopped, then
1368 			 * start it.
1369 			 */
1370 			return (*ifp->if_init)(ifp);
1371 		case IFF_UP|IFF_RUNNING:
1372 			error = 0;
1373 			if (ec->ec_ifflags_cb == NULL ||
1374 			    (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) {
1375 				/*
1376 				 * Reset the interface to pick up
1377 				 * changes in any other flags that
1378 				 * affect the hardware state.
1379 				 */
1380 				return (*ifp->if_init)(ifp);
1381 			} else
1382 				return error;
1383 		case 0:
1384 			break;
1385 		}
1386 		return 0;
1387 	case SIOCGETHERCAP:
1388 		eccr = (struct eccapreq *)data;
1389 		eccr->eccr_capabilities = ec->ec_capabilities;
1390 		eccr->eccr_capenable = ec->ec_capenable;
1391 		return 0;
1392 	case SIOCADDMULTI:
1393 		return ether_addmulti(ifreq_getaddr(cmd, ifr), ec);
1394 	case SIOCDELMULTI:
1395 		return ether_delmulti(ifreq_getaddr(cmd, ifr), ec);
1396 	case SIOCSIFMEDIA:
1397 	case SIOCGIFMEDIA:
1398 		if (ec->ec_mii == NULL)
1399 			return ENOTTY;
1400 		return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd);
1401 	case SIOCALIFADDR:
1402 		sdl = satocsdl(sstocsa(&iflr->addr));
1403 		if (sdl->sdl_family != AF_LINK)
1404 			;
1405 		else if (ETHER_IS_MULTICAST(CLLADDR(sdl)))
1406 			return EINVAL;
1407 		else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0)
1408 			return EINVAL;
1409 		/*FALLTHROUGH*/
1410 	default:
1411 		return ifioctl_common(ifp, cmd, data);
1412 	}
1413 	return 0;
1414 }
1415 
1416 static int
1417 ether_multicast_sysctl(SYSCTLFN_ARGS)
1418 {
1419 	struct ether_multi *enm;
1420 	struct ether_multi_sysctl addr;
1421 	struct ifnet *ifp;
1422 	struct ethercom *ec;
1423 	int error;
1424 	size_t written;
1425 
1426 	if (namelen != 1)
1427 		return EINVAL;
1428 
1429 	ifp = if_byindex(name[0]);
1430 	if (ifp == NULL)
1431 		return ENODEV;
1432 	if (ifp->if_type != IFT_ETHER) {
1433 		*oldlenp = 0;
1434 		return 0;
1435 	}
1436 	ec = (struct ethercom *)ifp;
1437 
1438 	if (oldp == NULL) {
1439 		*oldlenp = ec->ec_multicnt * sizeof(addr);
1440 		return 0;
1441 	}
1442 
1443 	memset(&addr, 0, sizeof(addr));
1444 	error = 0;
1445 	written = 0;
1446 
1447 	LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) {
1448 		if (written + sizeof(addr) > *oldlenp)
1449 			break;
1450 		addr.enm_refcount = enm->enm_refcount;
1451 		memcpy(addr.enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN);
1452 		memcpy(addr.enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN);
1453 		error = sysctl_copyout(l, &addr, oldp, sizeof(addr));
1454 		if (error)
1455 			break;
1456 		written += sizeof(addr);
1457 		oldp = (char *)oldp + sizeof(addr);
1458 	}
1459 
1460 	*oldlenp = written;
1461 	return error;
1462 }
1463 
1464 SYSCTL_SETUP(sysctl_net_ether_setup, "sysctl net.ether subtree setup")
1465 {
1466 	const struct sysctlnode *rnode = NULL;
1467 
1468 	sysctl_createv(clog, 0, NULL, &rnode,
1469 		       CTLFLAG_PERMANENT,
1470 		       CTLTYPE_NODE, "ether",
1471 		       SYSCTL_DESCR("Ethernet-specific information"),
1472 		       NULL, 0, NULL, 0,
1473 		       CTL_NET, CTL_CREATE, CTL_EOL);
1474 
1475 	sysctl_createv(clog, 0, &rnode, NULL,
1476 		       CTLFLAG_PERMANENT,
1477 		       CTLTYPE_NODE, "multicast",
1478 		       SYSCTL_DESCR("multicast addresses"),
1479 		       ether_multicast_sysctl, 0, NULL, 0,
1480 		       CTL_CREATE, CTL_EOL);
1481 }
1482 
1483 void
1484 etherinit(void)
1485 {
1486 	mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET);
1487 }
1488