xref: /netbsd-src/sys/net/if_ethersubr.c (revision f17b710f3d406bee67aa39c65053114ab78297c5)
1 /*	$NetBSD: if_ethersubr.c,v 1.210 2015/06/04 09:19:59 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 1.210 2015/06/04 09:19:59 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/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 #include <sys/rnd.h>
96 #include <sys/rndsource.h>
97 
98 #include <net/if.h>
99 #include <net/netisr.h>
100 #include <net/route.h>
101 #include <net/if_llc.h>
102 #include <net/if_dl.h>
103 #include <net/if_types.h>
104 
105 #include <net/if_media.h>
106 #include <dev/mii/mii.h>
107 #include <dev/mii/miivar.h>
108 
109 #if NARP == 0
110 /*
111  * XXX there should really be a way to issue this warning from within config(8)
112  */
113 #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.
114 #endif
115 
116 #include <net/bpf.h>
117 
118 #include <net/if_ether.h>
119 #include <net/if_vlanvar.h>
120 
121 #if NPPPOE > 0
122 #include <net/if_pppoe.h>
123 #endif
124 
125 #if NAGR > 0
126 #include <net/agr/ieee8023_slowprotocols.h>	/* XXX */
127 #include <net/agr/ieee8023ad.h>
128 #include <net/agr/if_agrvar.h>
129 #endif
130 
131 #if NBRIDGE > 0
132 #include <net/if_bridgevar.h>
133 #endif
134 
135 #include <netinet/in.h>
136 #ifdef INET
137 #include <netinet/in_var.h>
138 #endif
139 #include <netinet/if_inarp.h>
140 
141 #ifdef INET6
142 #ifndef INET
143 #include <netinet/in.h>
144 #endif
145 #include <netinet6/in6_var.h>
146 #include <netinet6/nd6.h>
147 #endif
148 
149 
150 #include "carp.h"
151 #if NCARP > 0
152 #include <netinet/ip_carp.h>
153 #endif
154 
155 #ifdef NETATALK
156 #include <netatalk/at.h>
157 #include <netatalk/at_var.h>
158 #include <netatalk/at_extern.h>
159 
160 #define llc_snap_org_code llc_un.type_snap.org_code
161 #define llc_snap_ether_type llc_un.type_snap.ether_type
162 
163 extern u_char	at_org_code[3];
164 extern u_char	aarp_org_code[3];
165 #endif /* NETATALK */
166 
167 #ifdef MPLS
168 #include <netmpls/mpls.h>
169 #include <netmpls/mpls_var.h>
170 #endif
171 
172 static struct timeval bigpktppslim_last;
173 static int bigpktppslim = 2;	/* XXX */
174 static int bigpktpps_count;
175 static kmutex_t bigpktpps_lock __cacheline_aligned;
176 
177 
178 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] =
179     { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
180 const uint8_t ethermulticastaddr_slowprotocols[ETHER_ADDR_LEN] =
181     { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x02 };
182 #define senderr(e) { error = (e); goto bad;}
183 
184 static	int ether_output(struct ifnet *, struct mbuf *,
185 	    const struct sockaddr *, struct rtentry *);
186 
187 /*
188  * Ethernet output routine.
189  * Encapsulate a packet of type family for the local net.
190  * Assumes that ifp is actually pointer to ethercom structure.
191  */
192 static int
193 ether_output(struct ifnet * const ifp0, struct mbuf * const m0,
194 	const struct sockaddr * const dst,
195 	struct rtentry *rt)
196 {
197 	uint16_t etype = 0;
198 	int error = 0, hdrcmplt = 0;
199  	uint8_t esrc[6], edst[6];
200 	struct mbuf *m = m0;
201 	struct mbuf *mcopy = NULL;
202 	struct ether_header *eh;
203 	struct ifnet *ifp = ifp0;
204 	ALTQ_DECL(struct altq_pktattr pktattr;)
205 #ifdef INET
206 	struct arphdr *ah;
207 #endif /* INET */
208 #ifdef NETATALK
209 	struct at_ifaddr *aa;
210 #endif /* NETATALK */
211 
212 #ifndef NET_MPSAFE
213 	KASSERT(KERNEL_LOCKED_P());
214 #endif
215 
216 #ifdef MBUFTRACE
217 	m_claimm(m, ifp->if_mowner);
218 #endif
219 
220 #if NCARP > 0
221 	if (ifp->if_type == IFT_CARP) {
222 		struct ifaddr *ifa;
223 
224 		/* loop back if this is going to the carp interface */
225 		if (dst != NULL && ifp0->if_link_state == LINK_STATE_UP &&
226 		    (ifa = ifa_ifwithaddr(dst)) != NULL &&
227 		    ifa->ifa_ifp == ifp0)
228 			return looutput(ifp0, m, dst, rt);
229 
230 		ifp = ifp->if_carpdev;
231 		/* ac = (struct arpcom *)ifp; */
232 
233 		if ((ifp0->if_flags & (IFF_UP|IFF_RUNNING)) !=
234 		    (IFF_UP|IFF_RUNNING))
235 			senderr(ENETDOWN);
236 	}
237 #endif /* NCARP > 0 */
238 
239 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
240 		senderr(ENETDOWN);
241 
242 	switch (dst->sa_family) {
243 
244 #ifdef INET
245 	case AF_INET:
246 		if (m->m_flags & M_BCAST)
247 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
248 		else if (m->m_flags & M_MCAST)
249 			ETHER_MAP_IP_MULTICAST(&satocsin(dst)->sin_addr, edst);
250 		else if (!arpresolve(ifp, rt, m, dst, edst))
251 			return (0);	/* if not yet resolved */
252 		/* If broadcasting on a simplex interface, loopback a copy */
253 		if ((m->m_flags & M_BCAST) && (ifp->if_flags & IFF_SIMPLEX))
254 			mcopy = m_copy(m, 0, (int)M_COPYALL);
255 		etype = htons(ETHERTYPE_IP);
256 		break;
257 
258 	case AF_ARP:
259 		ah = mtod(m, struct arphdr *);
260 		if (m->m_flags & M_BCAST)
261 			(void)memcpy(edst, etherbroadcastaddr, sizeof(edst));
262 		else {
263 			void *tha = ar_tha(ah);
264 
265 			if (tha == NULL) {
266 				/* fake with ARPHDR_IEEE1394 */
267 				return 0;
268 			}
269 			memcpy(edst, tha, sizeof(edst));
270 		}
271 
272 		ah->ar_hrd = htons(ARPHRD_ETHER);
273 
274 		switch (ntohs(ah->ar_op)) {
275 		case ARPOP_REVREQUEST:
276 		case ARPOP_REVREPLY:
277 			etype = htons(ETHERTYPE_REVARP);
278 			break;
279 
280 		case ARPOP_REQUEST:
281 		case ARPOP_REPLY:
282 		default:
283 			etype = htons(ETHERTYPE_ARP);
284 		}
285 
286 		break;
287 #endif
288 #ifdef INET6
289 	case AF_INET6:
290 		if (!nd6_storelladdr(ifp, rt, m, dst, edst, sizeof(edst))){
291 			/* something bad happened */
292 			return (0);
293 		}
294 		etype = htons(ETHERTYPE_IPV6);
295 		break;
296 #endif
297 #ifdef NETATALK
298     case AF_APPLETALK:
299 		if (!aarpresolve(ifp, m, (const struct sockaddr_at *)dst, edst)) {
300 #ifdef NETATALKDEBUG
301 			printf("aarpresolv failed\n");
302 #endif /* NETATALKDEBUG */
303 			return (0);
304 		}
305 		/*
306 		 * ifaddr is the first thing in at_ifaddr
307 		 */
308 		aa = (struct at_ifaddr *) at_ifawithnet(
309 		    (const struct sockaddr_at *)dst, ifp);
310 		if (aa == NULL)
311 		    goto bad;
312 
313 		/*
314 		 * In the phase 2 case, we need to prepend an mbuf for the
315 		 * llc header.  Since we must preserve the value of m,
316 		 * which is passed to us by value, we m_copy() the first
317 		 * mbuf, and use it for our llc header.
318 		 */
319 		if (aa->aa_flags & AFA_PHASE2) {
320 			struct llc llc;
321 
322 			M_PREPEND(m, sizeof(struct llc), M_DONTWAIT);
323 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
324 			llc.llc_control = LLC_UI;
325 			memcpy(llc.llc_snap_org_code, at_org_code,
326 			    sizeof(llc.llc_snap_org_code));
327 			llc.llc_snap_ether_type = htons(ETHERTYPE_ATALK);
328 			memcpy(mtod(m, void *), &llc, sizeof(struct llc));
329 		} else {
330 			etype = htons(ETHERTYPE_ATALK);
331 		}
332 		break;
333 #endif /* NETATALK */
334 	case pseudo_AF_HDRCMPLT:
335 		hdrcmplt = 1;
336 		memcpy(esrc,
337 		    ((const struct ether_header *)dst->sa_data)->ether_shost,
338 		    sizeof(esrc));
339 		/* FALLTHROUGH */
340 
341 	case AF_UNSPEC:
342  		memcpy(edst,
343 		    ((const struct ether_header *)dst->sa_data)->ether_dhost,
344 		    sizeof(edst));
345 		/* AF_UNSPEC doesn't swap the byte order of the ether_type. */
346 		etype = ((const struct ether_header *)dst->sa_data)->ether_type;
347 		break;
348 
349 	default:
350 		printf("%s: can't handle af%d\n", ifp->if_xname,
351 			dst->sa_family);
352 		senderr(EAFNOSUPPORT);
353 	}
354 
355 #ifdef MPLS
356 	{
357 		struct m_tag *mtag;
358 		mtag = m_tag_find(m, PACKET_TAG_MPLS, NULL);
359 		if (mtag != NULL) {
360 			/* Having the tag itself indicates it's MPLS */
361 			etype = htons(ETHERTYPE_MPLS);
362 			m_tag_delete(m, mtag);
363 		}
364 	}
365 #endif
366 
367 	if (mcopy)
368 		(void)looutput(ifp, mcopy, dst, rt);
369 
370 	/* If no ether type is set, this must be a 802.2 formatted packet.
371 	 */
372 	if (etype == 0)
373 		etype = htons(m->m_pkthdr.len);
374 	/*
375 	 * Add local net header.  If no space in first mbuf,
376 	 * allocate another.
377 	 */
378 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
379 	if (m == 0)
380 		senderr(ENOBUFS);
381 	eh = mtod(m, struct ether_header *);
382 	/* Note: etype is already in network byte order. */
383 	(void)memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
384  	memcpy(eh->ether_dhost, edst, sizeof(edst));
385 	if (hdrcmplt)
386 		memcpy(eh->ether_shost, esrc, sizeof(eh->ether_shost));
387 	else
388 	 	memcpy(eh->ether_shost, CLLADDR(ifp->if_sadl),
389 		    sizeof(eh->ether_shost));
390 
391 #if NCARP > 0
392 	if (ifp0 != ifp && ifp0->if_type == IFT_CARP) {
393 	 	memcpy(eh->ether_shost, CLLADDR(ifp0->if_sadl),
394 		    sizeof(eh->ether_shost));
395 	}
396 #endif /* NCARP > 0 */
397 
398 	if ((error = pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_OUT)) != 0)
399 		return (error);
400 	if (m == NULL)
401 		return (0);
402 
403 #if NBRIDGE > 0
404 	/*
405 	 * Bridges require special output handling.
406 	 */
407 	if (ifp->if_bridge)
408 		return (bridge_output(ifp, m, NULL, NULL));
409 #endif
410 
411 #if NCARP > 0
412 	if (ifp != ifp0)
413 		ifp0->if_obytes += m->m_pkthdr.len + ETHER_HDR_LEN;
414 #endif /* NCARP > 0 */
415 
416 #ifdef ALTQ
417 	/*
418 	 * If ALTQ is enabled on the parent interface, do
419 	 * classification; the queueing discipline might not
420 	 * require classification, but might require the
421 	 * address family/header pointer in the pktattr.
422 	 */
423 	if (ALTQ_IS_ENABLED(&ifp->if_snd))
424 		altq_etherclassify(&ifp->if_snd, m, &pktattr);
425 #endif
426 	return ifq_enqueue(ifp, m ALTQ_COMMA ALTQ_DECL(&pktattr));
427 
428 bad:
429 	if (m)
430 		m_freem(m);
431 	return (error);
432 }
433 
434 #ifdef ALTQ
435 /*
436  * This routine is a slight hack to allow a packet to be classified
437  * if the Ethernet headers are present.  It will go away when ALTQ's
438  * classification engine understands link headers.
439  */
440 void
441 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
442     struct altq_pktattr *pktattr)
443 {
444 	struct ether_header *eh;
445 	uint16_t ether_type;
446 	int hlen, af, hdrsize;
447 	void *hdr;
448 
449 	hlen = ETHER_HDR_LEN;
450 	eh = mtod(m, struct ether_header *);
451 
452 	ether_type = htons(eh->ether_type);
453 
454 	if (ether_type < ETHERMTU) {
455 		/* LLC/SNAP */
456 		struct llc *llc = (struct llc *)(eh + 1);
457 		hlen += 8;
458 
459 		if (m->m_len < hlen ||
460 		    llc->llc_dsap != LLC_SNAP_LSAP ||
461 		    llc->llc_ssap != LLC_SNAP_LSAP ||
462 		    llc->llc_control != LLC_UI) {
463 			/* Not SNAP. */
464 			goto bad;
465 		}
466 
467 		ether_type = htons(llc->llc_un.type_snap.ether_type);
468 	}
469 
470 	switch (ether_type) {
471 	case ETHERTYPE_IP:
472 		af = AF_INET;
473 		hdrsize = 20;		/* sizeof(struct ip) */
474 		break;
475 
476 	case ETHERTYPE_IPV6:
477 		af = AF_INET6;
478 		hdrsize = 40;		/* sizeof(struct ip6_hdr) */
479 		break;
480 
481 	default:
482 		af = AF_UNSPEC;
483 		hdrsize = 0;
484 		break;
485 	}
486 
487 	while (m->m_len <= hlen) {
488 		hlen -= m->m_len;
489 		m = m->m_next;
490 	}
491 	if (m->m_len < (hlen + hdrsize)) {
492 		/*
493 		 * protocol header not in a single mbuf.
494 		 * We can't cope with this situation right
495 		 * now (but it shouldn't ever happen, really, anyhow).
496 		 */
497 #ifdef DEBUG
498 		printf("altq_etherclassify: headers span multiple mbufs: "
499 		    "%d < %d\n", m->m_len, (hlen + hdrsize));
500 #endif
501 		goto bad;
502 	}
503 
504 	m->m_data += hlen;
505 	m->m_len -= hlen;
506 
507 	hdr = mtod(m, void *);
508 
509 	if (ALTQ_NEEDS_CLASSIFY(ifq))
510 		pktattr->pattr_class =
511 		    (*ifq->altq_classify)(ifq->altq_clfier, m, af);
512 	pktattr->pattr_af = af;
513 	pktattr->pattr_hdr = hdr;
514 
515 	m->m_data -= hlen;
516 	m->m_len += hlen;
517 
518 	return;
519 
520  bad:
521 	pktattr->pattr_class = NULL;
522 	pktattr->pattr_hdr = NULL;
523 	pktattr->pattr_af = AF_UNSPEC;
524 }
525 #endif /* ALTQ */
526 
527 /*
528  * Process a received Ethernet packet;
529  * the packet is in the mbuf chain m with
530  * the ether header.
531  */
532 void
533 ether_input(struct ifnet *ifp, struct mbuf *m)
534 {
535 	struct ethercom *ec = (struct ethercom *) ifp;
536 	pktqueue_t *pktq = NULL;
537 	struct ifqueue *inq = NULL;
538 	uint16_t etype;
539 	struct ether_header *eh;
540 	size_t ehlen;
541 	static int earlypkts;
542 	int isr = 0;
543 #if defined (LLC) || defined(NETATALK)
544 	struct llc *l;
545 #endif
546 
547 	if ((ifp->if_flags & IFF_UP) == 0) {
548 		m_freem(m);
549 		return;
550 	}
551 
552 #ifdef MBUFTRACE
553 	m_claimm(m, &ec->ec_rx_mowner);
554 #endif
555 	eh = mtod(m, struct ether_header *);
556 	etype = ntohs(eh->ether_type);
557 	ehlen = sizeof(*eh);
558 
559 	if(__predict_false(earlypkts < 100 || !rnd_initial_entropy)) {
560 		rnd_add_data(NULL, eh, ehlen, 0);
561 		earlypkts++;
562 	}
563 
564 	/*
565 	 * Determine if the packet is within its size limits.
566 	 */
567 	if (etype != ETHERTYPE_MPLS && m->m_pkthdr.len >
568 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
569 		mutex_enter(&bigpktpps_lock);
570 		if (ppsratecheck(&bigpktppslim_last, &bigpktpps_count,
571 			    bigpktppslim)) {
572 			printf("%s: discarding oversize frame (len=%d)\n",
573 			    ifp->if_xname, m->m_pkthdr.len);
574 		}
575 		mutex_exit(&bigpktpps_lock);
576 		m_freem(m);
577 		return;
578 	}
579 
580 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
581 		/*
582 		 * If this is not a simplex interface, drop the packet
583 		 * if it came from us.
584 		 */
585 		if ((ifp->if_flags & IFF_SIMPLEX) == 0 &&
586 		    memcmp(CLLADDR(ifp->if_sadl), eh->ether_shost,
587 		    ETHER_ADDR_LEN) == 0) {
588 			m_freem(m);
589 			return;
590 		}
591 
592 		if (memcmp(etherbroadcastaddr,
593 		    eh->ether_dhost, ETHER_ADDR_LEN) == 0)
594 			m->m_flags |= M_BCAST;
595 		else
596 			m->m_flags |= M_MCAST;
597 		ifp->if_imcasts++;
598 	}
599 
600 	/* If the CRC is still on the packet, trim it off. */
601 	if (m->m_flags & M_HASFCS) {
602 		m_adj(m, -ETHER_CRC_LEN);
603 		m->m_flags &= ~M_HASFCS;
604 	}
605 
606 	ifp->if_ibytes += m->m_pkthdr.len;
607 
608 #if NCARP > 0
609 	if (__predict_false(ifp->if_carp && ifp->if_type != IFT_CARP)) {
610 		/*
611 		 * clear M_PROMISC, in case the packets comes from a
612 		 * vlan
613 		 */
614 		m->m_flags &= ~M_PROMISC;
615 		if (carp_input(m, (uint8_t *)&eh->ether_shost,
616 		    (uint8_t *)&eh->ether_dhost, eh->ether_type) == 0)
617 			return;
618 	}
619 #endif /* NCARP > 0 */
620 	if ((m->m_flags & (M_BCAST|M_MCAST|M_PROMISC)) == 0 &&
621 	    (ifp->if_flags & IFF_PROMISC) != 0 &&
622 	    memcmp(CLLADDR(ifp->if_sadl), eh->ether_dhost,
623 		   ETHER_ADDR_LEN) != 0) {
624 		m->m_flags |= M_PROMISC;
625 	}
626 
627 	if ((m->m_flags & M_PROMISC) == 0) {
628 		if (pfil_run_hooks(ifp->if_pfil, &m, ifp, PFIL_IN) != 0)
629 			return;
630 		if (m == NULL)
631 			return;
632 
633 		eh = mtod(m, struct ether_header *);
634 		etype = ntohs(eh->ether_type);
635 		ehlen = sizeof(*eh);
636 	}
637 
638 #if NAGR > 0
639 	if (ifp->if_agrprivate &&
640 	    __predict_true(etype != ETHERTYPE_SLOWPROTOCOLS)) {
641 		m->m_flags &= ~M_PROMISC;
642 		agr_input(ifp, m);
643 		return;
644 	}
645 #endif /* NAGR > 0 */
646 
647 	/*
648 	 * If VLANs are configured on the interface, check to
649 	 * see if the device performed the decapsulation and
650 	 * provided us with the tag.
651 	 */
652 	if (ec->ec_nvlans && m_tag_find(m, PACKET_TAG_VLAN, NULL) != NULL) {
653 #if NVLAN > 0
654 		/*
655 		 * vlan_input() will either recursively call ether_input()
656 		 * or drop the packet.
657 		 */
658 		vlan_input(ifp, m);
659 #else
660 		m_freem(m);
661 #endif
662 		return;
663 	}
664 
665 	/*
666 	 * Handle protocols that expect to have the Ethernet header
667 	 * (and possibly FCS) intact.
668 	 */
669 	switch (etype) {
670 	case ETHERTYPE_VLAN: {
671 		struct ether_vlan_header *evl = (void *)eh;
672 		/*
673 		 * If there is a tag of 0, then the VLAN header was probably
674 		 * just being used to store the priority.  Extract the ether
675 		 * type, and if IP or IPV6, let them deal with it.
676 		 */
677 		if (m->m_len <= sizeof(*evl)
678 		    && EVL_VLANOFTAG(evl->evl_tag) == 0) {
679 			etype = ntohs(evl->evl_proto);
680 			ehlen = sizeof(*evl);
681 			if ((m->m_flags & M_PROMISC) == 0
682 			    && (etype == ETHERTYPE_IP
683 				|| etype == ETHERTYPE_IPV6))
684 				break;
685 		}
686 #if NVLAN > 0
687 		/*
688 		 * vlan_input() will either recursively call ether_input()
689 		 * or drop the packet.
690 		 */
691 		if (((struct ethercom *)ifp)->ec_nvlans != 0)
692 			vlan_input(ifp, m);
693 		else
694 #endif /* NVLAN > 0 */
695 			m_freem(m);
696 		return;
697 	}
698 #if NPPPOE > 0
699 	case ETHERTYPE_PPPOEDISC:
700 	case ETHERTYPE_PPPOE:
701 		if (m->m_flags & M_PROMISC) {
702 			m_freem(m);
703 			return;
704 		}
705 #ifndef PPPOE_SERVER
706 		if (m->m_flags & (M_MCAST | M_BCAST)) {
707 			m_freem(m);
708 			return;
709 		}
710 #endif
711 
712 		if (etype == ETHERTYPE_PPPOEDISC)
713 			inq = &ppoediscinq;
714 		else
715 			inq = &ppoeinq;
716 		if (IF_QFULL(inq)) {
717 			IF_DROP(inq);
718 			m_freem(m);
719 		} else {
720 			IF_ENQUEUE(inq, m);
721 			softint_schedule(pppoe_softintr);
722 		}
723 		return;
724 #endif /* NPPPOE > 0 */
725 	case ETHERTYPE_SLOWPROTOCOLS: {
726 		uint8_t subtype;
727 
728 #if defined(DIAGNOSTIC)
729 		if (m->m_pkthdr.len < sizeof(*eh) + sizeof(subtype)) {
730 			panic("ether_input: too short slow protocol packet");
731 		}
732 #endif
733 		m_copydata(m, sizeof(*eh), sizeof(subtype), &subtype);
734 		switch (subtype) {
735 #if NAGR > 0
736 		case SLOWPROTOCOLS_SUBTYPE_LACP:
737 			if (ifp->if_agrprivate) {
738 				ieee8023ad_lacp_input(ifp, m);
739 				return;
740 			}
741 			break;
742 
743 		case SLOWPROTOCOLS_SUBTYPE_MARKER:
744 			if (ifp->if_agrprivate) {
745 				ieee8023ad_marker_input(ifp, m);
746 				return;
747 			}
748 			break;
749 #endif /* NAGR > 0 */
750 		default:
751 			if (subtype == 0 || subtype > 10) {
752 				/* illegal value */
753 				m_freem(m);
754 				return;
755 			}
756 			/* unknown subtype */
757 			break;
758 		}
759 		/* FALLTHROUGH */
760 	}
761 	default:
762 		if (m->m_flags & M_PROMISC) {
763 			m_freem(m);
764 			return;
765 		}
766 	}
767 
768 	/* If the CRC is still on the packet, trim it off. */
769 	if (m->m_flags & M_HASFCS) {
770 		m_adj(m, -ETHER_CRC_LEN);
771 		m->m_flags &= ~M_HASFCS;
772 	}
773 
774 	if (etype > ETHERMTU + sizeof (struct ether_header)) {
775 		/* Strip off the Ethernet header. */
776 		m_adj(m, ehlen);
777 
778 		switch (etype) {
779 #ifdef INET
780 		case ETHERTYPE_IP:
781 #ifdef GATEWAY
782 			if (ipflow_fastforward(m))
783 				return;
784 #endif
785 			pktq = ip_pktq;
786 			break;
787 
788 		case ETHERTYPE_ARP:
789 			isr = NETISR_ARP;
790 			inq = &arpintrq;
791 			break;
792 
793 		case ETHERTYPE_REVARP:
794 			revarpinput(m);	/* XXX queue? */
795 			return;
796 #endif
797 #ifdef INET6
798 		case ETHERTYPE_IPV6:
799 			if (__predict_false(!in6_present)) {
800 				m_freem(m);
801 				return;
802 			}
803 #ifdef GATEWAY
804 			if (ip6flow_fastforward(&m))
805 				return;
806 #endif
807 			pktq = ip6_pktq;
808 			break;
809 #endif
810 #ifdef NETATALK
811 		case ETHERTYPE_ATALK:
812 			isr = NETISR_ATALK;
813 			inq = &atintrq1;
814 			break;
815 		case ETHERTYPE_AARP:
816 			/* probably this should be done with a NETISR as well */
817 			aarpinput(ifp, m); /* XXX */
818 			return;
819 #endif /* NETATALK */
820 #ifdef MPLS
821 		case ETHERTYPE_MPLS:
822 			isr = NETISR_MPLS;
823 			inq = &mplsintrq;
824 			break;
825 #endif
826 		default:
827 			m_freem(m);
828 			return;
829 		}
830 	} else {
831 #if defined (LLC) || defined (NETATALK)
832 		l = (struct llc *)(eh+1);
833 		switch (l->llc_dsap) {
834 #ifdef NETATALK
835 		case LLC_SNAP_LSAP:
836 			switch (l->llc_control) {
837 			case LLC_UI:
838 				if (l->llc_ssap != LLC_SNAP_LSAP) {
839 					goto dropanyway;
840 				}
841 
842 				if (memcmp(&(l->llc_snap_org_code)[0],
843 				    at_org_code, sizeof(at_org_code)) == 0 &&
844 				    ntohs(l->llc_snap_ether_type) ==
845 				    ETHERTYPE_ATALK) {
846 					inq = &atintrq2;
847 					m_adj(m, sizeof(struct ether_header)
848 					    + sizeof(struct llc));
849 					isr = NETISR_ATALK;
850 					break;
851 				}
852 
853 				if (memcmp(&(l->llc_snap_org_code)[0],
854 				    aarp_org_code,
855 				    sizeof(aarp_org_code)) == 0 &&
856 				    ntohs(l->llc_snap_ether_type) ==
857 				    ETHERTYPE_AARP) {
858 					m_adj( m, sizeof(struct ether_header)
859 					    + sizeof(struct llc));
860 					aarpinput(ifp, m); /* XXX */
861 				    return;
862 				}
863 
864 			default:
865 				goto dropanyway;
866 			}
867 			break;
868 		dropanyway:
869 #endif
870 		default:
871 			m_freem(m);
872 			return;
873 		}
874 #else /* ISO || LLC || NETATALK*/
875 		m_freem(m);
876 		return;
877 #endif /* ISO || LLC || NETATALK*/
878 	}
879 
880 	if (__predict_true(pktq)) {
881 		const uint32_t h = pktq_rps_hash(m);
882 		if (__predict_false(!pktq_enqueue(pktq, m, h))) {
883 			m_freem(m);
884 		}
885 		return;
886 	}
887 
888 	if (__predict_false(!inq)) {
889 		/* Should not happen. */
890 		m_freem(m);
891 		return;
892 	}
893 	if (IF_QFULL(inq)) {
894 		IF_DROP(inq);
895 		m_freem(m);
896 	} else {
897 		IF_ENQUEUE(inq, m);
898 		schednetisr(isr);
899 	}
900 }
901 
902 /*
903  * Convert Ethernet address to printable (loggable) representation.
904  */
905 char *
906 ether_sprintf(const u_char *ap)
907 {
908 	static char etherbuf[3 * ETHER_ADDR_LEN];
909 	return ether_snprintf(etherbuf, sizeof(etherbuf), ap);
910 }
911 
912 char *
913 ether_snprintf(char *buf, size_t len, const u_char *ap)
914 {
915 	char *cp = buf;
916 	size_t i;
917 
918 	for (i = 0; i < len / 3; i++) {
919 		*cp++ = hexdigits[*ap >> 4];
920 		*cp++ = hexdigits[*ap++ & 0xf];
921 		*cp++ = ':';
922 	}
923 	*--cp = '\0';
924 	return buf;
925 }
926 
927 /*
928  * Perform common duties while attaching to interface list
929  */
930 void
931 ether_ifattach(struct ifnet *ifp, const uint8_t *lla)
932 {
933 	struct ethercom *ec = (struct ethercom *)ifp;
934 
935 	ifp->if_type = IFT_ETHER;
936 	ifp->if_hdrlen = ETHER_HDR_LEN;
937 	ifp->if_dlt = DLT_EN10MB;
938 	ifp->if_mtu = ETHERMTU;
939 	ifp->if_output = ether_output;
940 	ifp->if_input = ether_input;
941 	if (ifp->if_baudrate == 0)
942 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
943 
944 	if_set_sadl(ifp, lla, ETHER_ADDR_LEN, !ETHER_IS_LOCAL(lla));
945 
946 	LIST_INIT(&ec->ec_multiaddrs);
947 	ifp->if_broadcastaddr = etherbroadcastaddr;
948 	bpf_attach(ifp, DLT_EN10MB, sizeof(struct ether_header));
949 #ifdef MBUFTRACE
950 	strlcpy(ec->ec_tx_mowner.mo_name, ifp->if_xname,
951 	    sizeof(ec->ec_tx_mowner.mo_name));
952 	strlcpy(ec->ec_tx_mowner.mo_descr, "tx",
953 	    sizeof(ec->ec_tx_mowner.mo_descr));
954 	strlcpy(ec->ec_rx_mowner.mo_name, ifp->if_xname,
955 	    sizeof(ec->ec_rx_mowner.mo_name));
956 	strlcpy(ec->ec_rx_mowner.mo_descr, "rx",
957 	    sizeof(ec->ec_rx_mowner.mo_descr));
958 	MOWNER_ATTACH(&ec->ec_tx_mowner);
959 	MOWNER_ATTACH(&ec->ec_rx_mowner);
960 	ifp->if_mowner = &ec->ec_tx_mowner;
961 #endif
962 }
963 
964 void
965 ether_ifdetach(struct ifnet *ifp)
966 {
967 	struct ethercom *ec = (void *) ifp;
968 	struct ether_multi *enm;
969 	int s;
970 
971 	/*
972 	 * Prevent further calls to ioctl (for example turning off
973 	 * promiscuous mode from the bridge code), which eventually can
974 	 * call if_init() which can cause panics because the interface
975 	 * is in the process of being detached. Return device not configured
976 	 * instead.
977 	 */
978 	ifp->if_ioctl = (int (*)(struct ifnet *, u_long, void *))enxio;
979 
980 #if NBRIDGE > 0
981 	if (ifp->if_bridge)
982 		bridge_ifdetach(ifp);
983 #endif
984 
985 	bpf_detach(ifp);
986 
987 #if NVLAN > 0
988 	if (ec->ec_nvlans)
989 		vlan_ifdetach(ifp);
990 #endif
991 
992 	s = splnet();
993 	while ((enm = LIST_FIRST(&ec->ec_multiaddrs)) != NULL) {
994 		LIST_REMOVE(enm, enm_list);
995 		free(enm, M_IFMADDR);
996 		ec->ec_multicnt--;
997 	}
998 	splx(s);
999 
1000 	ifp->if_mowner = NULL;
1001 	MOWNER_DETACH(&ec->ec_rx_mowner);
1002 	MOWNER_DETACH(&ec->ec_tx_mowner);
1003 }
1004 
1005 #if 0
1006 /*
1007  * This is for reference.  We have a table-driven version
1008  * of the little-endian crc32 generator, which is faster
1009  * than the double-loop.
1010  */
1011 uint32_t
1012 ether_crc32_le(const uint8_t *buf, size_t len)
1013 {
1014 	uint32_t c, crc, carry;
1015 	size_t i, j;
1016 
1017 	crc = 0xffffffffU;	/* initial value */
1018 
1019 	for (i = 0; i < len; i++) {
1020 		c = buf[i];
1021 		for (j = 0; j < 8; j++) {
1022 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
1023 			crc >>= 1;
1024 			c >>= 1;
1025 			if (carry)
1026 				crc = (crc ^ ETHER_CRC_POLY_LE);
1027 		}
1028 	}
1029 
1030 	return (crc);
1031 }
1032 #else
1033 uint32_t
1034 ether_crc32_le(const uint8_t *buf, size_t len)
1035 {
1036 	static const uint32_t crctab[] = {
1037 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1038 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1039 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1040 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1041 	};
1042 	uint32_t crc;
1043 	size_t i;
1044 
1045 	crc = 0xffffffffU;	/* initial value */
1046 
1047 	for (i = 0; i < len; i++) {
1048 		crc ^= buf[i];
1049 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1050 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1051 	}
1052 
1053 	return (crc);
1054 }
1055 #endif
1056 
1057 uint32_t
1058 ether_crc32_be(const uint8_t *buf, size_t len)
1059 {
1060 	uint32_t c, crc, carry;
1061 	size_t i, j;
1062 
1063 	crc = 0xffffffffU;	/* initial value */
1064 
1065 	for (i = 0; i < len; i++) {
1066 		c = buf[i];
1067 		for (j = 0; j < 8; j++) {
1068 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
1069 			crc <<= 1;
1070 			c >>= 1;
1071 			if (carry)
1072 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1073 		}
1074 	}
1075 
1076 	return (crc);
1077 }
1078 
1079 #ifdef INET
1080 const uint8_t ether_ipmulticast_min[ETHER_ADDR_LEN] =
1081     { 0x01, 0x00, 0x5e, 0x00, 0x00, 0x00 };
1082 const uint8_t ether_ipmulticast_max[ETHER_ADDR_LEN] =
1083     { 0x01, 0x00, 0x5e, 0x7f, 0xff, 0xff };
1084 #endif
1085 #ifdef INET6
1086 const uint8_t ether_ip6multicast_min[ETHER_ADDR_LEN] =
1087     { 0x33, 0x33, 0x00, 0x00, 0x00, 0x00 };
1088 const uint8_t ether_ip6multicast_max[ETHER_ADDR_LEN] =
1089     { 0x33, 0x33, 0xff, 0xff, 0xff, 0xff };
1090 #endif
1091 
1092 /*
1093  * ether_aton implementation, not using a static buffer.
1094  */
1095 int
1096 ether_aton_r(u_char *dest, size_t len, const char *str)
1097 {
1098         const u_char *cp = (const void *)str;
1099 	u_char *ep;
1100 
1101 #define atox(c)	(((c) <= '9') ? ((c) - '0') : ((toupper(c) - 'A') + 10))
1102 
1103 	if (len < ETHER_ADDR_LEN)
1104 		return ENOSPC;
1105 
1106 	ep = dest + ETHER_ADDR_LEN;
1107 
1108 	while (*cp) {
1109                 if (!isxdigit(*cp))
1110                         return EINVAL;
1111 		*dest = atox(*cp);
1112 		cp++;
1113                 if (isxdigit(*cp)) {
1114                         *dest = (*dest << 4) | atox(*cp);
1115 			dest++;
1116 			cp++;
1117                 } else
1118 			dest++;
1119 		if (dest == ep)
1120 			return *cp == '\0' ? 0 : ENAMETOOLONG;
1121 		switch (*cp) {
1122 		case ':':
1123 		case '-':
1124 		case '.':
1125 			cp++;
1126 			break;
1127 		}
1128         }
1129 	return ENOBUFS;
1130 }
1131 
1132 /*
1133  * Convert a sockaddr into an Ethernet address or range of Ethernet
1134  * addresses.
1135  */
1136 int
1137 ether_multiaddr(const struct sockaddr *sa, uint8_t addrlo[ETHER_ADDR_LEN],
1138     uint8_t addrhi[ETHER_ADDR_LEN])
1139 {
1140 #ifdef INET
1141 	const struct sockaddr_in *sin;
1142 #endif /* INET */
1143 #ifdef INET6
1144 	const struct sockaddr_in6 *sin6;
1145 #endif /* INET6 */
1146 
1147 	switch (sa->sa_family) {
1148 
1149 	case AF_UNSPEC:
1150 		memcpy(addrlo, sa->sa_data, ETHER_ADDR_LEN);
1151 		memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1152 		break;
1153 
1154 #ifdef INET
1155 	case AF_INET:
1156 		sin = satocsin(sa);
1157 		if (sin->sin_addr.s_addr == INADDR_ANY) {
1158 			/*
1159 			 * An IP address of INADDR_ANY means listen to
1160 			 * or stop listening to all of the Ethernet
1161 			 * multicast addresses used for IP.
1162 			 * (This is for the sake of IP multicast routers.)
1163 			 */
1164 			memcpy(addrlo, ether_ipmulticast_min, ETHER_ADDR_LEN);
1165 			memcpy(addrhi, ether_ipmulticast_max, ETHER_ADDR_LEN);
1166 		}
1167 		else {
1168 			ETHER_MAP_IP_MULTICAST(&sin->sin_addr, addrlo);
1169 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1170 		}
1171 		break;
1172 #endif
1173 #ifdef INET6
1174 	case AF_INET6:
1175 		sin6 = satocsin6(sa);
1176 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1177 			/*
1178 			 * An IP6 address of 0 means listen to or stop
1179 			 * listening to all of the Ethernet multicast
1180 			 * address used for IP6.
1181 			 * (This is used for multicast routers.)
1182 			 */
1183 			memcpy(addrlo, ether_ip6multicast_min, ETHER_ADDR_LEN);
1184 			memcpy(addrhi, ether_ip6multicast_max, ETHER_ADDR_LEN);
1185 		} else {
1186 			ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, addrlo);
1187 			memcpy(addrhi, addrlo, ETHER_ADDR_LEN);
1188 		}
1189 		break;
1190 #endif
1191 
1192 	default:
1193 		return EAFNOSUPPORT;
1194 	}
1195 	return 0;
1196 }
1197 
1198 /*
1199  * Add an Ethernet multicast address or range of addresses to the list for a
1200  * given interface.
1201  */
1202 int
1203 ether_addmulti(const struct sockaddr *sa, struct ethercom *ec)
1204 {
1205 	struct ether_multi *enm;
1206 	u_char addrlo[ETHER_ADDR_LEN];
1207 	u_char addrhi[ETHER_ADDR_LEN];
1208 	int s = splnet(), error;
1209 
1210 	error = ether_multiaddr(sa, addrlo, addrhi);
1211 	if (error != 0) {
1212 		splx(s);
1213 		return error;
1214 	}
1215 
1216 	/*
1217 	 * Verify that we have valid Ethernet multicast addresses.
1218 	 */
1219 	if (!ETHER_IS_MULTICAST(addrlo) || !ETHER_IS_MULTICAST(addrhi)) {
1220 		splx(s);
1221 		return EINVAL;
1222 	}
1223 	/*
1224 	 * See if the address range is already in the list.
1225 	 */
1226 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1227 	if (enm != NULL) {
1228 		/*
1229 		 * Found it; just increment the reference count.
1230 		 */
1231 		++enm->enm_refcount;
1232 		splx(s);
1233 		return 0;
1234 	}
1235 	/*
1236 	 * New address or range; malloc a new multicast record
1237 	 * and link it into the interface's multicast list.
1238 	 */
1239 	enm = (struct ether_multi *)malloc(sizeof(*enm), M_IFMADDR, M_NOWAIT);
1240 	if (enm == NULL) {
1241 		splx(s);
1242 		return ENOBUFS;
1243 	}
1244 	memcpy(enm->enm_addrlo, addrlo, 6);
1245 	memcpy(enm->enm_addrhi, addrhi, 6);
1246 	enm->enm_refcount = 1;
1247 	LIST_INSERT_HEAD(&ec->ec_multiaddrs, enm, enm_list);
1248 	ec->ec_multicnt++;
1249 	splx(s);
1250 	/*
1251 	 * Return ENETRESET to inform the driver that the list has changed
1252 	 * and its reception filter should be adjusted accordingly.
1253 	 */
1254 	return ENETRESET;
1255 }
1256 
1257 /*
1258  * Delete a multicast address record.
1259  */
1260 int
1261 ether_delmulti(const struct sockaddr *sa, struct ethercom *ec)
1262 {
1263 	struct ether_multi *enm;
1264 	u_char addrlo[ETHER_ADDR_LEN];
1265 	u_char addrhi[ETHER_ADDR_LEN];
1266 	int s = splnet(), error;
1267 
1268 	error = ether_multiaddr(sa, addrlo, addrhi);
1269 	if (error != 0) {
1270 		splx(s);
1271 		return (error);
1272 	}
1273 
1274 	/*
1275 	 * Look ur the address in our list.
1276 	 */
1277 	ETHER_LOOKUP_MULTI(addrlo, addrhi, ec, enm);
1278 	if (enm == NULL) {
1279 		splx(s);
1280 		return (ENXIO);
1281 	}
1282 	if (--enm->enm_refcount != 0) {
1283 		/*
1284 		 * Still some claims to this record.
1285 		 */
1286 		splx(s);
1287 		return (0);
1288 	}
1289 	/*
1290 	 * No remaining claims to this record; unlink and free it.
1291 	 */
1292 	LIST_REMOVE(enm, enm_list);
1293 	free(enm, M_IFMADDR);
1294 	ec->ec_multicnt--;
1295 	splx(s);
1296 	/*
1297 	 * Return ENETRESET to inform the driver that the list has changed
1298 	 * and its reception filter should be adjusted accordingly.
1299 	 */
1300 	return (ENETRESET);
1301 }
1302 
1303 void
1304 ether_set_ifflags_cb(struct ethercom *ec, ether_cb_t cb)
1305 {
1306 	ec->ec_ifflags_cb = cb;
1307 }
1308 
1309 /*
1310  * Common ioctls for Ethernet interfaces.  Note, we must be
1311  * called at splnet().
1312  */
1313 int
1314 ether_ioctl(struct ifnet *ifp, u_long cmd, void *data)
1315 {
1316 	struct ethercom *ec = (void *) ifp;
1317 	struct eccapreq *eccr;
1318 	struct ifreq *ifr = (struct ifreq *)data;
1319 	struct if_laddrreq *iflr = data;
1320 	const struct sockaddr_dl *sdl;
1321 	static const uint8_t zero[ETHER_ADDR_LEN];
1322 	int error;
1323 
1324 	switch (cmd) {
1325 	case SIOCINITIFADDR:
1326 	    {
1327 		struct ifaddr *ifa = (struct ifaddr *)data;
1328 		if (ifa->ifa_addr->sa_family != AF_LINK
1329 		    && (ifp->if_flags & (IFF_UP|IFF_RUNNING)) !=
1330 		       (IFF_UP|IFF_RUNNING)) {
1331 			ifp->if_flags |= IFF_UP;
1332 			if ((error = (*ifp->if_init)(ifp)) != 0)
1333 				return error;
1334 		}
1335 #ifdef INET
1336 		if (ifa->ifa_addr->sa_family == AF_INET)
1337 			arp_ifinit(ifp, ifa);
1338 #endif /* INET */
1339 		return 0;
1340 	    }
1341 
1342 	case SIOCSIFMTU:
1343 	    {
1344 		int maxmtu;
1345 
1346 		if (ec->ec_capabilities & ETHERCAP_JUMBO_MTU)
1347 			maxmtu = ETHERMTU_JUMBO;
1348 		else
1349 			maxmtu = ETHERMTU;
1350 
1351 		if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > maxmtu)
1352 			return EINVAL;
1353 		else if ((error = ifioctl_common(ifp, cmd, data)) != ENETRESET)
1354 			return error;
1355 		else if (ifp->if_flags & IFF_UP) {
1356 			/* Make sure the device notices the MTU change. */
1357 			return (*ifp->if_init)(ifp);
1358 		} else
1359 			return 0;
1360 	    }
1361 
1362 	case SIOCSIFFLAGS:
1363 		if ((error = ifioctl_common(ifp, cmd, data)) != 0)
1364 			return error;
1365 		switch (ifp->if_flags & (IFF_UP|IFF_RUNNING)) {
1366 		case IFF_RUNNING:
1367 			/*
1368 			 * If interface is marked down and it is running,
1369 			 * then stop and disable it.
1370 			 */
1371 			(*ifp->if_stop)(ifp, 1);
1372 			break;
1373 		case IFF_UP:
1374 			/*
1375 			 * If interface is marked up and it is stopped, then
1376 			 * start it.
1377 			 */
1378 			return (*ifp->if_init)(ifp);
1379 		case IFF_UP|IFF_RUNNING:
1380 			error = 0;
1381 			if (ec->ec_ifflags_cb == NULL ||
1382 			    (error = (*ec->ec_ifflags_cb)(ec)) == ENETRESET) {
1383 				/*
1384 				 * Reset the interface to pick up
1385 				 * changes in any other flags that
1386 				 * affect the hardware state.
1387 				 */
1388 				return (*ifp->if_init)(ifp);
1389 			} else
1390 				return error;
1391 		case 0:
1392 			break;
1393 		}
1394 		return 0;
1395 	case SIOCGETHERCAP:
1396 		eccr = (struct eccapreq *)data;
1397 		eccr->eccr_capabilities = ec->ec_capabilities;
1398 		eccr->eccr_capenable = ec->ec_capenable;
1399 		return 0;
1400 	case SIOCADDMULTI:
1401 		return ether_addmulti(ifreq_getaddr(cmd, ifr), ec);
1402 	case SIOCDELMULTI:
1403 		return ether_delmulti(ifreq_getaddr(cmd, ifr), ec);
1404 	case SIOCSIFMEDIA:
1405 	case SIOCGIFMEDIA:
1406 		if (ec->ec_mii == NULL)
1407 			return ENOTTY;
1408 		return ifmedia_ioctl(ifp, ifr, &ec->ec_mii->mii_media, cmd);
1409 	case SIOCALIFADDR:
1410 		sdl = satocsdl(sstocsa(&iflr->addr));
1411 		if (sdl->sdl_family != AF_LINK)
1412 			;
1413 		else if (ETHER_IS_MULTICAST(CLLADDR(sdl)))
1414 			return EINVAL;
1415 		else if (memcmp(zero, CLLADDR(sdl), sizeof(zero)) == 0)
1416 			return EINVAL;
1417 		/*FALLTHROUGH*/
1418 	default:
1419 		return ifioctl_common(ifp, cmd, data);
1420 	}
1421 	return 0;
1422 }
1423 
1424 static int
1425 ether_multicast_sysctl(SYSCTLFN_ARGS)
1426 {
1427 	struct ether_multi *enm;
1428 	struct ether_multi_sysctl addr;
1429 	struct ifnet *ifp;
1430 	struct ethercom *ec;
1431 	int error;
1432 	size_t written;
1433 
1434 	if (namelen != 1)
1435 		return EINVAL;
1436 
1437 	ifp = if_byindex(name[0]);
1438 	if (ifp == NULL)
1439 		return ENODEV;
1440 	if (ifp->if_type != IFT_ETHER) {
1441 		*oldlenp = 0;
1442 		return 0;
1443 	}
1444 	ec = (struct ethercom *)ifp;
1445 
1446 	if (oldp == NULL) {
1447 		*oldlenp = ec->ec_multicnt * sizeof(addr);
1448 		return 0;
1449 	}
1450 
1451 	memset(&addr, 0, sizeof(addr));
1452 	error = 0;
1453 	written = 0;
1454 
1455 	LIST_FOREACH(enm, &ec->ec_multiaddrs, enm_list) {
1456 		if (written + sizeof(addr) > *oldlenp)
1457 			break;
1458 		addr.enm_refcount = enm->enm_refcount;
1459 		memcpy(addr.enm_addrlo, enm->enm_addrlo, ETHER_ADDR_LEN);
1460 		memcpy(addr.enm_addrhi, enm->enm_addrhi, ETHER_ADDR_LEN);
1461 		error = sysctl_copyout(l, &addr, oldp, sizeof(addr));
1462 		if (error)
1463 			break;
1464 		written += sizeof(addr);
1465 		oldp = (char *)oldp + sizeof(addr);
1466 	}
1467 
1468 	*oldlenp = written;
1469 	return error;
1470 }
1471 
1472 SYSCTL_SETUP(sysctl_net_ether_setup, "sysctl net.ether subtree setup")
1473 {
1474 	const struct sysctlnode *rnode = NULL;
1475 
1476 	sysctl_createv(clog, 0, NULL, &rnode,
1477 		       CTLFLAG_PERMANENT,
1478 		       CTLTYPE_NODE, "ether",
1479 		       SYSCTL_DESCR("Ethernet-specific information"),
1480 		       NULL, 0, NULL, 0,
1481 		       CTL_NET, CTL_CREATE, CTL_EOL);
1482 
1483 	sysctl_createv(clog, 0, &rnode, NULL,
1484 		       CTLFLAG_PERMANENT,
1485 		       CTLTYPE_NODE, "multicast",
1486 		       SYSCTL_DESCR("multicast addresses"),
1487 		       ether_multicast_sysctl, 0, NULL, 0,
1488 		       CTL_CREATE, CTL_EOL);
1489 }
1490 
1491 void
1492 etherinit(void)
1493 {
1494 	mutex_init(&bigpktpps_lock, MUTEX_DEFAULT, IPL_NET);
1495 }
1496