if_ethersubr.c revision 160735
1/*-
2 * Copyright (c) 1982, 1989, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 *	@(#)if_ethersubr.c	8.1 (Berkeley) 6/10/93
30 * $FreeBSD: head/sys/net/if_ethersubr.c 160735 2006-07-27 03:50:38Z avatar $
31 */
32
33#include "opt_atalk.h"
34#include "opt_inet.h"
35#include "opt_inet6.h"
36#include "opt_ipx.h"
37#include "opt_mac.h"
38#include "opt_netgraph.h"
39#include "opt_carp.h"
40
41#include <sys/param.h>
42#include <sys/systm.h>
43#include <sys/kernel.h>
44#include <sys/mac.h>
45#include <sys/malloc.h>
46#include <sys/module.h>
47#include <sys/mbuf.h>
48#include <sys/random.h>
49#include <sys/socket.h>
50#include <sys/sockio.h>
51#include <sys/sysctl.h>
52
53#include <net/if.h>
54#include <net/if_arp.h>
55#include <net/netisr.h>
56#include <net/route.h>
57#include <net/if_llc.h>
58#include <net/if_dl.h>
59#include <net/if_types.h>
60#include <net/bpf.h>
61#include <net/ethernet.h>
62#include <net/bridgestp.h>
63#include <net/if_bridgevar.h>
64#include <net/if_vlan_var.h>
65
66#if defined(INET) || defined(INET6)
67#include <netinet/in.h>
68#include <netinet/in_var.h>
69#include <netinet/if_ether.h>
70#include <netinet/ip_fw.h>
71#include <netinet/ip_dummynet.h>
72#endif
73#ifdef INET6
74#include <netinet6/nd6.h>
75#endif
76
77#ifdef DEV_CARP
78#include <netinet/ip_carp.h>
79#endif
80
81#ifdef IPX
82#include <netipx/ipx.h>
83#include <netipx/ipx_if.h>
84#endif
85int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m);
86int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp,
87		struct sockaddr *dst, short *tp, int *hlen);
88
89#ifdef NETATALK
90#include <netatalk/at.h>
91#include <netatalk/at_var.h>
92#include <netatalk/at_extern.h>
93
94#define llc_snap_org_code llc_un.type_snap.org_code
95#define llc_snap_ether_type llc_un.type_snap.ether_type
96
97extern u_char	at_org_code[3];
98extern u_char	aarp_org_code[3];
99#endif /* NETATALK */
100
101/* netgraph node hooks for ng_ether(4) */
102void	(*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp);
103void	(*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m);
104int	(*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
105void	(*ng_ether_attach_p)(struct ifnet *ifp);
106void	(*ng_ether_detach_p)(struct ifnet *ifp);
107
108void	(*vlan_input_p)(struct ifnet *, struct mbuf *);
109
110/* if_bridge(4) support */
111struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *);
112int	(*bridge_output_p)(struct ifnet *, struct mbuf *,
113		struct sockaddr *, struct rtentry *);
114void	(*bridge_dn_p)(struct mbuf *, struct ifnet *);
115
116static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] =
117			{ 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
118
119static	int ether_resolvemulti(struct ifnet *, struct sockaddr **,
120		struct sockaddr *);
121
122/* XXX: should be in an arp support file, not here */
123MALLOC_DEFINE(M_ARPCOM, "arpcom", "802.* interface internals");
124
125#define senderr(e) do { error = (e); goto bad;} while (0)
126
127#if defined(INET) || defined(INET6)
128int
129ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
130	struct ip_fw **rule, int shared);
131static int ether_ipfw;
132#endif
133
134/*
135 * Ethernet output routine.
136 * Encapsulate a packet of type family for the local net.
137 * Use trailer local net encapsulation if enough data in first
138 * packet leaves a multiple of 512 bytes of data in remainder.
139 */
140int
141ether_output(struct ifnet *ifp, struct mbuf *m,
142	struct sockaddr *dst, struct rtentry *rt0)
143{
144	short type;
145	int error, hdrcmplt = 0;
146	u_char esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN];
147	struct ether_header *eh;
148	int loop_copy = 1;
149	int hlen;	/* link layer header length */
150
151#ifdef MAC
152	error = mac_check_ifnet_transmit(ifp, m);
153	if (error)
154		senderr(error);
155#endif
156
157	if (ifp->if_flags & IFF_MONITOR)
158		senderr(ENETDOWN);
159	if (!((ifp->if_flags & IFF_UP) &&
160	    (ifp->if_drv_flags & IFF_DRV_RUNNING)))
161		senderr(ENETDOWN);
162
163	hlen = ETHER_HDR_LEN;
164	switch (dst->sa_family) {
165#ifdef INET
166	case AF_INET:
167		error = arpresolve(ifp, rt0, m, dst, edst);
168		if (error)
169			return (error == EWOULDBLOCK ? 0 : error);
170		type = htons(ETHERTYPE_IP);
171		break;
172	case AF_ARP:
173	{
174		struct arphdr *ah;
175		ah = mtod(m, struct arphdr *);
176		ah->ar_hrd = htons(ARPHRD_ETHER);
177
178		loop_copy = 0; /* if this is for us, don't do it */
179
180		switch(ntohs(ah->ar_op)) {
181		case ARPOP_REVREQUEST:
182		case ARPOP_REVREPLY:
183			type = htons(ETHERTYPE_REVARP);
184			break;
185		case ARPOP_REQUEST:
186		case ARPOP_REPLY:
187		default:
188			type = htons(ETHERTYPE_ARP);
189			break;
190		}
191
192		if (m->m_flags & M_BCAST)
193			bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN);
194		else
195			bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN);
196
197	}
198	break;
199#endif
200#ifdef INET6
201	case AF_INET6:
202		error = nd6_storelladdr(ifp, rt0, m, dst, (u_char *)edst);
203		if (error)
204			return error;
205		type = htons(ETHERTYPE_IPV6);
206		break;
207#endif
208#ifdef IPX
209	case AF_IPX:
210		if (ef_outputp) {
211		    error = ef_outputp(ifp, &m, dst, &type, &hlen);
212		    if (error)
213			goto bad;
214		} else
215		    type = htons(ETHERTYPE_IPX);
216		bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
217		    (caddr_t)edst, sizeof (edst));
218		break;
219#endif
220#ifdef NETATALK
221	case AF_APPLETALK:
222	  {
223	    struct at_ifaddr *aa;
224
225	    if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL)
226		    senderr(EHOSTUNREACH); /* XXX */
227	    if (!aarpresolve(ifp, m, (struct sockaddr_at *)dst, edst))
228		    return (0);
229	    /*
230	     * In the phase 2 case, need to prepend an mbuf for the llc header.
231	     */
232	    if ( aa->aa_flags & AFA_PHASE2 ) {
233		struct llc llc;
234
235		M_PREPEND(m, LLC_SNAPFRAMELEN, M_DONTWAIT);
236		if (m == NULL)
237			senderr(ENOBUFS);
238		llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
239		llc.llc_control = LLC_UI;
240		bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code));
241		llc.llc_snap_ether_type = htons( ETHERTYPE_AT );
242		bcopy(&llc, mtod(m, caddr_t), LLC_SNAPFRAMELEN);
243		type = htons(m->m_pkthdr.len);
244		hlen = LLC_SNAPFRAMELEN + ETHER_HDR_LEN;
245	    } else {
246		type = htons(ETHERTYPE_AT);
247	    }
248	    break;
249	  }
250#endif /* NETATALK */
251
252	case pseudo_AF_HDRCMPLT:
253		hdrcmplt = 1;
254		eh = (struct ether_header *)dst->sa_data;
255		(void)memcpy(esrc, eh->ether_shost, sizeof (esrc));
256		/* FALLTHROUGH */
257
258	case AF_UNSPEC:
259		loop_copy = 0; /* if this is for us, don't do it */
260		eh = (struct ether_header *)dst->sa_data;
261		(void)memcpy(edst, eh->ether_dhost, sizeof (edst));
262		type = eh->ether_type;
263		break;
264
265	default:
266		if_printf(ifp, "can't handle af%d\n", dst->sa_family);
267		senderr(EAFNOSUPPORT);
268	}
269
270	/*
271	 * Add local net header.  If no space in first mbuf,
272	 * allocate another.
273	 */
274	M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
275	if (m == NULL)
276		senderr(ENOBUFS);
277	eh = mtod(m, struct ether_header *);
278	(void)memcpy(&eh->ether_type, &type,
279		sizeof(eh->ether_type));
280	(void)memcpy(eh->ether_dhost, edst, sizeof (edst));
281	if (hdrcmplt)
282		(void)memcpy(eh->ether_shost, esrc,
283			sizeof(eh->ether_shost));
284	else
285		(void)memcpy(eh->ether_shost, IF_LLADDR(ifp),
286			sizeof(eh->ether_shost));
287
288       /*
289	* Bridges require special output handling.
290	*/
291	if (ifp->if_bridge) {
292		BRIDGE_OUTPUT(ifp, m, error);
293		return (error);
294	}
295
296	/*
297	 * If a simplex interface, and the packet is being sent to our
298	 * Ethernet address or a broadcast address, loopback a copy.
299	 * XXX To make a simplex device behave exactly like a duplex
300	 * device, we should copy in the case of sending to our own
301	 * ethernet address (thus letting the original actually appear
302	 * on the wire). However, we don't do that here for security
303	 * reasons and compatibility with the original behavior.
304	 */
305	if ((ifp->if_flags & IFF_SIMPLEX) && loop_copy &&
306	    m_tag_find(m, PACKET_TAG_PF_ROUTED, NULL) == NULL) {
307		int csum_flags = 0;
308
309		if (m->m_pkthdr.csum_flags & CSUM_IP)
310			csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
311		if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
312			csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
313
314		if (m->m_flags & M_BCAST) {
315			struct mbuf *n;
316
317			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
318				n->m_pkthdr.csum_flags |= csum_flags;
319				if (csum_flags & CSUM_DATA_VALID)
320					n->m_pkthdr.csum_data = 0xffff;
321				(void)if_simloop(ifp, n, dst->sa_family, hlen);
322			} else
323				ifp->if_iqdrops++;
324		} else if (bcmp(eh->ether_dhost, eh->ether_shost,
325				ETHER_ADDR_LEN) == 0) {
326			m->m_pkthdr.csum_flags |= csum_flags;
327			if (csum_flags & CSUM_DATA_VALID)
328				m->m_pkthdr.csum_data = 0xffff;
329			(void) if_simloop(ifp, m, dst->sa_family, hlen);
330			return (0);	/* XXX */
331		}
332	}
333
334#ifdef DEV_CARP
335	if (ifp->if_carp &&
336	    (error = carp_output(ifp, m, dst, NULL)))
337		goto bad;
338#endif
339
340	/* Handle ng_ether(4) processing, if any */
341	if (IFP2AC(ifp)->ac_netgraph != NULL) {
342		KASSERT(ng_ether_output_p != NULL,
343		    ("ng_ether_output_p is NULL"));
344		if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
345bad:			if (m != NULL)
346				m_freem(m);
347			return (error);
348		}
349		if (m == NULL)
350			return (0);
351	}
352
353	/* Continue with link-layer output */
354	return ether_output_frame(ifp, m);
355}
356
357/*
358 * Ethernet link layer output routine to send a raw frame to the device.
359 *
360 * This assumes that the 14 byte Ethernet header is present and contiguous
361 * in the first mbuf (if BRIDGE'ing).
362 */
363int
364ether_output_frame(struct ifnet *ifp, struct mbuf *m)
365{
366	int error;
367#if defined(INET) || defined(INET6)
368	struct ip_fw *rule = ip_dn_claim_rule(m);
369
370	if (IPFW_LOADED && ether_ipfw != 0) {
371		if (ether_ipfw_chk(&m, ifp, &rule, 0) == 0) {
372			if (m) {
373				m_freem(m);
374				return EACCES;	/* pkt dropped */
375			} else
376				return 0;	/* consumed e.g. in a pipe */
377		}
378	}
379#endif
380
381	/*
382	 * Queue message on interface, update output statistics if
383	 * successful, and start output if interface not yet active.
384	 */
385	IFQ_HANDOFF(ifp, m, error);
386	return (error);
387}
388
389#if defined(INET) || defined(INET6)
390/*
391 * ipfw processing for ethernet packets (in and out).
392 * The second parameter is NULL from ether_demux, and ifp from
393 * ether_output_frame.
394 */
395int
396ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
397	struct ip_fw **rule, int shared)
398{
399	struct ether_header *eh;
400	struct ether_header save_eh;
401	struct mbuf *m;
402	int i;
403	struct ip_fw_args args;
404
405	if (*rule != NULL && fw_one_pass)
406		return 1; /* dummynet packet, already partially processed */
407
408	/*
409	 * I need some amt of data to be contiguous, and in case others need
410	 * the packet (shared==1) also better be in the first mbuf.
411	 */
412	m = *m0;
413	i = min( m->m_pkthdr.len, max_protohdr);
414	if ( shared || m->m_len < i) {
415		m = m_pullup(m, i);
416		if (m == NULL) {
417			*m0 = m;
418			return 0;
419		}
420	}
421	eh = mtod(m, struct ether_header *);
422	save_eh = *eh;			/* save copy for restore below */
423	m_adj(m, ETHER_HDR_LEN);	/* strip ethernet header */
424
425	args.m = m;		/* the packet we are looking at		*/
426	args.oif = dst;		/* destination, if any			*/
427	args.rule = *rule;	/* matching rule to restart		*/
428	args.next_hop = NULL;	/* we do not support forward yet	*/
429	args.eh = &save_eh;	/* MAC header for bridged/MAC packets	*/
430	args.inp = NULL;	/* used by ipfw uid/gid/jail rules	*/
431	i = ip_fw_chk_ptr(&args);
432	m = args.m;
433	if (m != NULL) {
434		/*
435		 * Restore Ethernet header, as needed, in case the
436		 * mbuf chain was replaced by ipfw.
437		 */
438		M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
439		if (m == NULL) {
440			*m0 = m;
441			return 0;
442		}
443		if (eh != mtod(m, struct ether_header *))
444			bcopy(&save_eh, mtod(m, struct ether_header *),
445				ETHER_HDR_LEN);
446	}
447	*m0 = m;
448	*rule = args.rule;
449
450	if (i == IP_FW_DENY) /* drop */
451		return 0;
452
453	KASSERT(m != NULL, ("ether_ipfw_chk: m is NULL"));
454
455	if (i == IP_FW_PASS) /* a PASS rule.  */
456		return 1;
457
458	if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) {
459		/*
460		 * Pass the pkt to dummynet, which consumes it.
461		 * If shared, make a copy and keep the original.
462		 */
463		if (shared) {
464			m = m_copypacket(m, M_DONTWAIT);
465			if (m == NULL)
466				return 0;
467		} else {
468			/*
469			 * Pass the original to dummynet and
470			 * nothing back to the caller
471			 */
472			*m0 = NULL ;
473		}
474		ip_dn_io_ptr(m, dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
475		return 0;
476	}
477	/*
478	 * XXX at some point add support for divert/forward actions.
479	 * If none of the above matches, we have to drop the pkt.
480	 */
481	return 0;
482}
483#endif
484
485/*
486 * Process a received Ethernet packet; the packet is in the
487 * mbuf chain m with the ethernet header at the front.
488 */
489static void
490ether_input(struct ifnet *ifp, struct mbuf *m)
491{
492	struct ether_header *eh;
493	u_short etype;
494
495	/*
496	 * Do consistency checks to verify assumptions
497	 * made by code past this point.
498	 */
499	if ((m->m_flags & M_PKTHDR) == 0) {
500		if_printf(ifp, "discard frame w/o packet header\n");
501		ifp->if_ierrors++;
502		m_freem(m);
503		return;
504	}
505	if (m->m_len < ETHER_HDR_LEN) {
506		/* XXX maybe should pullup? */
507		if_printf(ifp, "discard frame w/o leading ethernet "
508				"header (len %u pkt len %u)\n",
509				m->m_len, m->m_pkthdr.len);
510		ifp->if_ierrors++;
511		m_freem(m);
512		return;
513	}
514	eh = mtod(m, struct ether_header *);
515	etype = ntohs(eh->ether_type);
516	if (m->m_pkthdr.len >
517	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
518		if_printf(ifp, "discard oversize frame "
519				"(ether type %x flags %x len %u > max %lu)\n",
520				etype, m->m_flags, m->m_pkthdr.len,
521				ETHER_MAX_FRAME(ifp, etype,
522						m->m_flags & M_HASFCS));
523		ifp->if_ierrors++;
524		m_freem(m);
525		return;
526	}
527	if (m->m_pkthdr.rcvif == NULL) {
528		if_printf(ifp, "discard frame w/o interface pointer\n");
529		ifp->if_ierrors++;
530		m_freem(m);
531		return;
532	}
533#ifdef DIAGNOSTIC
534	if (m->m_pkthdr.rcvif != ifp) {
535		if_printf(ifp, "Warning, frame marked as received on %s\n",
536			m->m_pkthdr.rcvif->if_xname);
537	}
538#endif
539
540#ifdef MAC
541	/*
542	 * Tag the mbuf with an appropriate MAC label before any other
543	 * consumers can get to it.
544	 */
545	mac_create_mbuf_from_ifnet(ifp, m);
546#endif
547
548	/*
549	 * Give bpf a chance at the packet.
550	 */
551	BPF_MTAP(ifp, m);
552
553	/* If the CRC is still on the packet, trim it off. */
554	if (m->m_flags & M_HASFCS) {
555		m_adj(m, -ETHER_CRC_LEN);
556		m->m_flags &= ~M_HASFCS;
557	}
558
559	ifp->if_ibytes += m->m_pkthdr.len;
560
561	if (ifp->if_flags & IFF_MONITOR) {
562		/*
563		 * Interface marked for monitoring; discard packet.
564		 */
565		m_freem(m);
566		return;
567	}
568
569	/* Handle ng_ether(4) processing, if any */
570	if (IFP2AC(ifp)->ac_netgraph != NULL) {
571		KASSERT(ng_ether_input_p != NULL,
572		    ("ng_ether_input_p is NULL"));
573		(*ng_ether_input_p)(ifp, &m);
574		if (m == NULL)
575			return;
576	}
577
578	/*
579	 * Tap the packet off here for a bridge.  bridge_input()
580	 * will return NULL if it has consumed the packet, otherwise
581	 * it gets processed as normal.  Note that bridge_input()
582	 * will always return the original packet if we need to
583	 * process it locally.
584	 *
585	 * XXX: When changing the below block, please also look
586	 * at the src/sys/netgraph/ng_ether.c:ng_ether_rcv_upper()
587	 */
588	if (ifp->if_bridge) {
589		BRIDGE_INPUT(ifp, m);
590		if (m == NULL)
591			return;
592	}
593
594	/* First chunk of an mbuf contains good entropy */
595	if (harvest.ethernet)
596		random_harvest(m, 16, 3, 0, RANDOM_NET);
597	ether_demux(ifp, m);
598}
599
600/*
601 * Upper layer processing for a received Ethernet packet.
602 */
603void
604ether_demux(struct ifnet *ifp, struct mbuf *m)
605{
606	struct ether_header *eh;
607	int isr;
608	u_short ether_type;
609#if defined(NETATALK)
610	struct llc *l;
611#endif
612#if defined(INET) || defined(INET6)
613	struct ip_fw *rule = ip_dn_claim_rule(m);
614#endif
615
616	KASSERT(ifp != NULL, ("ether_demux: NULL interface pointer"));
617
618	eh = mtod(m, struct ether_header *);
619	ether_type = ntohs(eh->ether_type);
620
621#if defined(INET) || defined(INET6)
622	if (rule)	/* packet was already bridged */
623		goto post_stats;
624#endif
625
626	if (!(ifp->if_bridge) &&
627	    !((ether_type == ETHERTYPE_VLAN || m->m_flags & M_VLANTAG) &&
628	    ifp->if_vlantrunk != NULL)) {
629#ifdef DEV_CARP
630		/*
631		 * XXX: Okay, we need to call carp_forus() and - if it is for
632		 * us jump over code that does the normal check
633		 * "IF_LLADDR(ifp) == ether_dhost". The check sequence is a bit
634		 * different from OpenBSD, so we jump over as few code as
635		 * possible, to catch _all_ sanity checks. This needs
636		 * evaluation, to see if the carp ether_dhost values break any
637		 * of these checks!
638		 */
639		if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost))
640			goto pre_stats;
641#endif
642		/*
643		 * Discard packet if upper layers shouldn't see it because it
644		 * was unicast to a different Ethernet address. If the driver
645		 * is working properly, then this situation can only happen
646		 * when the interface is in promiscuous mode.
647		 *
648		 * If VLANs are active, and this packet has a VLAN tag, do
649		 * not drop it here but pass it on to the VLAN layer, to
650		 * give them a chance to consider it as well (e. g. in case
651		 * bridging is only active on a VLAN).  They will drop it if
652		 * it's undesired.
653		 */
654		if ((ifp->if_flags & IFF_PROMISC) != 0
655		    && !ETHER_IS_MULTICAST(eh->ether_dhost)
656		    && bcmp(eh->ether_dhost,
657		      IF_LLADDR(ifp), ETHER_ADDR_LEN) != 0
658		    && (ifp->if_flags & IFF_PPROMISC) == 0) {
659			    m_freem(m);
660			    return;
661		}
662	}
663
664#ifdef DEV_CARP
665pre_stats:
666#endif
667	/* Discard packet if interface is not up */
668	if ((ifp->if_flags & IFF_UP) == 0) {
669		m_freem(m);
670		return;
671	}
672	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
673		if (bcmp(etherbroadcastaddr, eh->ether_dhost,
674		    sizeof(etherbroadcastaddr)) == 0)
675			m->m_flags |= M_BCAST;
676		else
677			m->m_flags |= M_MCAST;
678	}
679	if (m->m_flags & (M_BCAST|M_MCAST))
680		ifp->if_imcasts++;
681
682#if defined(INET) || defined(INET6)
683post_stats:
684	if (IPFW_LOADED && ether_ipfw != 0) {
685		if (ether_ipfw_chk(&m, NULL, &rule, 0) == 0) {
686			if (m)
687				m_freem(m);
688			return;
689		}
690	}
691#endif
692
693	/*
694	 * Check to see if the device performed the VLAN decapsulation and
695	 * provided us with the tag.
696	 */
697	if (m->m_flags & M_VLANTAG) {
698		/*
699		 * If no VLANs are configured, drop.
700		 */
701		if (ifp->if_vlantrunk == NULL) {
702			ifp->if_noproto++;
703			m_freem(m);
704			return;
705		}
706		/*
707		 * vlan_input() will either recursively call ether_input()
708		 * or drop the packet.
709		 */
710		KASSERT(vlan_input_p != NULL,("ether_input: VLAN not loaded!"));
711		(*vlan_input_p)(ifp, m);
712		return;
713	}
714
715	/*
716	 * Handle protocols that expect to have the Ethernet header
717	 * (and possibly FCS) intact.
718	 */
719	switch (ether_type) {
720	case ETHERTYPE_VLAN:
721		if (ifp->if_vlantrunk != NULL) {
722			KASSERT(vlan_input_p,("ether_input: VLAN not loaded!"));
723			(*vlan_input_p)(ifp, m);
724		} else {
725			ifp->if_noproto++;
726			m_freem(m);
727		}
728		return;
729	}
730
731	/* Strip off Ethernet header. */
732	m_adj(m, ETHER_HDR_LEN);
733
734	/* If the CRC is still on the packet, trim it off. */
735	if (m->m_flags & M_HASFCS) {
736		m_adj(m, -ETHER_CRC_LEN);
737		m->m_flags &= ~M_HASFCS;
738	}
739
740	/* Reset layer specific mbuf flags to avoid confusing upper layers. */
741	m->m_flags &= ~(M_PROTOFLAGS);
742
743	switch (ether_type) {
744#ifdef INET
745	case ETHERTYPE_IP:
746		if ((m = ip_fastforward(m)) == NULL)
747			return;
748		isr = NETISR_IP;
749		break;
750
751	case ETHERTYPE_ARP:
752		if (ifp->if_flags & IFF_NOARP) {
753			/* Discard packet if ARP is disabled on interface */
754			m_freem(m);
755			return;
756		}
757		isr = NETISR_ARP;
758		break;
759#endif
760#ifdef IPX
761	case ETHERTYPE_IPX:
762		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
763			return;
764		isr = NETISR_IPX;
765		break;
766#endif
767#ifdef INET6
768	case ETHERTYPE_IPV6:
769		isr = NETISR_IPV6;
770		break;
771#endif
772#ifdef NETATALK
773	case ETHERTYPE_AT:
774		isr = NETISR_ATALK1;
775		break;
776	case ETHERTYPE_AARP:
777		isr = NETISR_AARP;
778		break;
779#endif /* NETATALK */
780	default:
781#ifdef IPX
782		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
783			return;
784#endif /* IPX */
785#if defined(NETATALK)
786		if (ether_type > ETHERMTU)
787			goto discard;
788		l = mtod(m, struct llc *);
789		if (l->llc_dsap == LLC_SNAP_LSAP &&
790		    l->llc_ssap == LLC_SNAP_LSAP &&
791		    l->llc_control == LLC_UI) {
792			if (bcmp(&(l->llc_snap_org_code)[0], at_org_code,
793			    sizeof(at_org_code)) == 0 &&
794			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
795				m_adj(m, LLC_SNAPFRAMELEN);
796				isr = NETISR_ATALK2;
797				break;
798			}
799			if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
800			    sizeof(aarp_org_code)) == 0 &&
801			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
802				m_adj(m, LLC_SNAPFRAMELEN);
803				isr = NETISR_AARP;
804				break;
805			}
806		}
807#endif /* NETATALK */
808		goto discard;
809	}
810	netisr_dispatch(isr, m);
811	return;
812
813discard:
814	/*
815	 * Packet is to be discarded.  If netgraph is present,
816	 * hand the packet to it for last chance processing;
817	 * otherwise dispose of it.
818	 */
819	if (IFP2AC(ifp)->ac_netgraph != NULL) {
820		KASSERT(ng_ether_input_orphan_p != NULL,
821		    ("ng_ether_input_orphan_p is NULL"));
822		/*
823		 * Put back the ethernet header so netgraph has a
824		 * consistent view of inbound packets.
825		 */
826		M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
827		(*ng_ether_input_orphan_p)(ifp, m);
828		return;
829	}
830	m_freem(m);
831}
832
833/*
834 * Convert Ethernet address to printable (loggable) representation.
835 * This routine is for compatibility; it's better to just use
836 *
837 *	printf("%6D", <pointer to address>, ":");
838 *
839 * since there's no static buffer involved.
840 */
841char *
842ether_sprintf(const u_char *ap)
843{
844	static char etherbuf[18];
845	snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
846	return (etherbuf);
847}
848
849/*
850 * Perform common duties while attaching to interface list
851 */
852void
853ether_ifattach(struct ifnet *ifp, const u_int8_t *lla)
854{
855	int i;
856	struct ifaddr *ifa;
857	struct sockaddr_dl *sdl;
858
859	ifp->if_addrlen = ETHER_ADDR_LEN;
860	ifp->if_hdrlen = ETHER_HDR_LEN;
861	if_attach(ifp);
862	ifp->if_mtu = ETHERMTU;
863	ifp->if_output = ether_output;
864	ifp->if_input = ether_input;
865	ifp->if_resolvemulti = ether_resolvemulti;
866	if (ifp->if_baudrate == 0)
867		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
868	ifp->if_broadcastaddr = etherbroadcastaddr;
869
870	ifa = ifp->if_addr;
871	KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
872	sdl = (struct sockaddr_dl *)ifa->ifa_addr;
873	sdl->sdl_type = IFT_ETHER;
874	sdl->sdl_alen = ifp->if_addrlen;
875	bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
876
877	bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
878	if (ng_ether_attach_p != NULL)
879		(*ng_ether_attach_p)(ifp);
880
881	/* Announce Ethernet MAC address if non-zero. */
882	for (i = 0; i < ifp->if_addrlen; i++)
883		if (lla[i] != 0)
884			break;
885	if (i != ifp->if_addrlen)
886		if_printf(ifp, "Ethernet address: %6D\n", lla, ":");
887	if (debug_mpsafenet && (ifp->if_flags & IFF_NEEDSGIANT) != 0)
888		if_printf(ifp, "if_start running deferred for Giant\n");
889}
890
891/*
892 * Perform common duties while detaching an Ethernet interface
893 */
894void
895ether_ifdetach(struct ifnet *ifp)
896{
897	if (IFP2AC(ifp)->ac_netgraph != NULL) {
898		KASSERT(ng_ether_detach_p != NULL,
899		    ("ng_ether_detach_p is NULL"));
900		(*ng_ether_detach_p)(ifp);
901	}
902
903	bpfdetach(ifp);
904	if_detach(ifp);
905}
906
907SYSCTL_DECL(_net_link);
908SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
909#if defined(INET) || defined(INET6)
910SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
911	    &ether_ipfw,0,"Pass ether pkts through firewall");
912#endif
913
914#if 0
915/*
916 * This is for reference.  We have a table-driven version
917 * of the little-endian crc32 generator, which is faster
918 * than the double-loop.
919 */
920uint32_t
921ether_crc32_le(const uint8_t *buf, size_t len)
922{
923	size_t i;
924	uint32_t crc;
925	int bit;
926	uint8_t data;
927
928	crc = 0xffffffff;	/* initial value */
929
930	for (i = 0; i < len; i++) {
931		for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1)
932			carry = (crc ^ data) & 1;
933			crc >>= 1;
934			if (carry)
935				crc = (crc ^ ETHER_CRC_POLY_LE);
936	}
937
938	return (crc);
939}
940#else
941uint32_t
942ether_crc32_le(const uint8_t *buf, size_t len)
943{
944	static const uint32_t crctab[] = {
945		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
946		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
947		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
948		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
949	};
950	size_t i;
951	uint32_t crc;
952
953	crc = 0xffffffff;	/* initial value */
954
955	for (i = 0; i < len; i++) {
956		crc ^= buf[i];
957		crc = (crc >> 4) ^ crctab[crc & 0xf];
958		crc = (crc >> 4) ^ crctab[crc & 0xf];
959	}
960
961	return (crc);
962}
963#endif
964
965uint32_t
966ether_crc32_be(const uint8_t *buf, size_t len)
967{
968	size_t i;
969	uint32_t crc, carry;
970	int bit;
971	uint8_t data;
972
973	crc = 0xffffffff;	/* initial value */
974
975	for (i = 0; i < len; i++) {
976		for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
977			carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
978			crc <<= 1;
979			if (carry)
980				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
981		}
982	}
983
984	return (crc);
985}
986
987int
988ether_ioctl(struct ifnet *ifp, int command, caddr_t data)
989{
990	struct ifaddr *ifa = (struct ifaddr *) data;
991	struct ifreq *ifr = (struct ifreq *) data;
992	int error = 0;
993
994	switch (command) {
995	case SIOCSIFADDR:
996		ifp->if_flags |= IFF_UP;
997
998		switch (ifa->ifa_addr->sa_family) {
999#ifdef INET
1000		case AF_INET:
1001			ifp->if_init(ifp->if_softc);	/* before arpwhohas */
1002			arp_ifinit(ifp, ifa);
1003			break;
1004#endif
1005#ifdef IPX
1006		/*
1007		 * XXX - This code is probably wrong
1008		 */
1009		case AF_IPX:
1010			{
1011			struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr);
1012
1013			if (ipx_nullhost(*ina))
1014				ina->x_host =
1015				    *(union ipx_host *)
1016				    IF_LLADDR(ifp);
1017			else {
1018				bcopy((caddr_t) ina->x_host.c_host,
1019				      (caddr_t) IF_LLADDR(ifp),
1020				      ETHER_ADDR_LEN);
1021			}
1022
1023			/*
1024			 * Set new address
1025			 */
1026			ifp->if_init(ifp->if_softc);
1027			break;
1028			}
1029#endif
1030		default:
1031			ifp->if_init(ifp->if_softc);
1032			break;
1033		}
1034		break;
1035
1036	case SIOCGIFADDR:
1037		{
1038			struct sockaddr *sa;
1039
1040			sa = (struct sockaddr *) & ifr->ifr_data;
1041			bcopy(IF_LLADDR(ifp),
1042			      (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
1043		}
1044		break;
1045
1046	case SIOCSIFMTU:
1047		/*
1048		 * Set the interface MTU.
1049		 */
1050		if (ifr->ifr_mtu > ETHERMTU) {
1051			error = EINVAL;
1052		} else {
1053			ifp->if_mtu = ifr->ifr_mtu;
1054		}
1055		break;
1056	default:
1057		error = EINVAL;			/* XXX netbsd has ENOTTY??? */
1058		break;
1059	}
1060	return (error);
1061}
1062
1063static int
1064ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
1065	struct sockaddr *sa)
1066{
1067	struct sockaddr_dl *sdl;
1068#ifdef INET
1069	struct sockaddr_in *sin;
1070#endif
1071#ifdef INET6
1072	struct sockaddr_in6 *sin6;
1073#endif
1074	u_char *e_addr;
1075
1076	switch(sa->sa_family) {
1077	case AF_LINK:
1078		/*
1079		 * No mapping needed. Just check that it's a valid MC address.
1080		 */
1081		sdl = (struct sockaddr_dl *)sa;
1082		e_addr = LLADDR(sdl);
1083		if (!ETHER_IS_MULTICAST(e_addr))
1084			return EADDRNOTAVAIL;
1085		*llsa = 0;
1086		return 0;
1087
1088#ifdef INET
1089	case AF_INET:
1090		sin = (struct sockaddr_in *)sa;
1091		if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1092			return EADDRNOTAVAIL;
1093		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1094		       M_NOWAIT|M_ZERO);
1095		if (sdl == NULL)
1096			return ENOMEM;
1097		sdl->sdl_len = sizeof *sdl;
1098		sdl->sdl_family = AF_LINK;
1099		sdl->sdl_index = ifp->if_index;
1100		sdl->sdl_type = IFT_ETHER;
1101		sdl->sdl_alen = ETHER_ADDR_LEN;
1102		e_addr = LLADDR(sdl);
1103		ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1104		*llsa = (struct sockaddr *)sdl;
1105		return 0;
1106#endif
1107#ifdef INET6
1108	case AF_INET6:
1109		sin6 = (struct sockaddr_in6 *)sa;
1110		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1111			/*
1112			 * An IP6 address of 0 means listen to all
1113			 * of the Ethernet multicast address used for IP6.
1114			 * (This is used for multicast routers.)
1115			 */
1116			ifp->if_flags |= IFF_ALLMULTI;
1117			*llsa = 0;
1118			return 0;
1119		}
1120		if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1121			return EADDRNOTAVAIL;
1122		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1123		       M_NOWAIT|M_ZERO);
1124		if (sdl == NULL)
1125			return (ENOMEM);
1126		sdl->sdl_len = sizeof *sdl;
1127		sdl->sdl_family = AF_LINK;
1128		sdl->sdl_index = ifp->if_index;
1129		sdl->sdl_type = IFT_ETHER;
1130		sdl->sdl_alen = ETHER_ADDR_LEN;
1131		e_addr = LLADDR(sdl);
1132		ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1133		*llsa = (struct sockaddr *)sdl;
1134		return 0;
1135#endif
1136
1137	default:
1138		/*
1139		 * Well, the text isn't quite right, but it's the name
1140		 * that counts...
1141		 */
1142		return EAFNOSUPPORT;
1143	}
1144}
1145
1146static void*
1147ether_alloc(u_char type, struct ifnet *ifp)
1148{
1149	struct arpcom	*ac;
1150
1151	ac = malloc(sizeof(struct arpcom), M_ARPCOM, M_WAITOK | M_ZERO);
1152	ac->ac_ifp = ifp;
1153
1154	return (ac);
1155}
1156
1157static void
1158ether_free(void *com, u_char type)
1159{
1160
1161	free(com, M_ARPCOM);
1162}
1163
1164static int
1165ether_modevent(module_t mod, int type, void *data)
1166{
1167
1168	switch (type) {
1169	case MOD_LOAD:
1170		if_register_com_alloc(IFT_ETHER, ether_alloc, ether_free);
1171		break;
1172	case MOD_UNLOAD:
1173		if_deregister_com_alloc(IFT_ETHER);
1174		break;
1175	default:
1176		return EOPNOTSUPP;
1177	}
1178
1179	return (0);
1180}
1181
1182static moduledata_t ether_mod = {
1183	"ether",
1184	ether_modevent,
1185	0
1186};
1187
1188DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY);
1189MODULE_VERSION(ether, 1);
1190