xref: /dragonfly/sys/net/if_ethersubr.c (revision 5153f92b)
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  * 3. All advertising materials mentioning features or use of this software
14  *    must display the following acknowledgement:
15  *	This product includes software developed by the University of
16  *	California, Berkeley and its contributors.
17  * 4. Neither the name of the University nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  *	@(#)if_ethersubr.c	8.1 (Berkeley) 6/10/93
34  * $FreeBSD: src/sys/net/if_ethersubr.c,v 1.70.2.33 2003/04/28 15:45:53 archie Exp $
35  * $DragonFly: src/sys/net/if_ethersubr.c,v 1.25 2005/01/06 09:14:13 hsu Exp $
36  */
37 
38 #include "opt_atalk.h"
39 #include "opt_inet.h"
40 #include "opt_inet6.h"
41 #include "opt_ipx.h"
42 #include "opt_bdg.h"
43 #include "opt_netgraph.h"
44 
45 #include <sys/param.h>
46 #include <sys/systm.h>
47 #include <sys/kernel.h>
48 #include <sys/malloc.h>
49 #include <sys/mbuf.h>
50 #include <sys/socket.h>
51 #include <sys/sockio.h>
52 #include <sys/sysctl.h>
53 
54 #include <net/if.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/bridge/bridge.h>
63 
64 #if defined(INET) || defined(INET6)
65 #include <netinet/in.h>
66 #include <netinet/in_var.h>
67 #include <netinet/if_ether.h>
68 #include <net/ipfw/ip_fw.h>
69 #include <net/dummynet/ip_dummynet.h>
70 #endif
71 #ifdef INET6
72 #include <netinet6/nd6.h>
73 #endif
74 
75 #ifdef IPX
76 #include <netproto/ipx/ipx.h>
77 #include <netproto/ipx/ipx_if.h>
78 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m);
79 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, struct sockaddr *dst,
80 		  short *tp, int *hlen);
81 #endif
82 
83 #ifdef NS
84 #include <netns/ns.h>
85 #include <netns/ns_if.h>
86 ushort ns_nettype;
87 int ether_outputdebug = 0;
88 int ether_inputdebug = 0;
89 #endif
90 
91 #ifdef NETATALK
92 #include <netproto/atalk/at.h>
93 #include <netproto/atalk/at_var.h>
94 #include <netproto/atalk/at_extern.h>
95 
96 #define	llc_snap_org_code	llc_un.type_snap.org_code
97 #define	llc_snap_ether_type	llc_un.type_snap.ether_type
98 
99 extern u_char	at_org_code[3];
100 extern u_char	aarp_org_code[3];
101 #endif /* NETATALK */
102 
103 /* netgraph node hooks for ng_ether(4) */
104 void	(*ng_ether_input_p)(struct ifnet *ifp,
105 		struct mbuf **mp, struct ether_header *eh);
106 void	(*ng_ether_input_orphan_p)(struct ifnet *ifp,
107 		struct mbuf *m, struct ether_header *eh);
108 int	(*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
109 void	(*ng_ether_attach_p)(struct ifnet *ifp);
110 void	(*ng_ether_detach_p)(struct ifnet *ifp);
111 
112 int	(*vlan_input_p)(struct ether_header *eh, struct mbuf *m);
113 int	(*vlan_input_tag_p)(struct mbuf *m, uint16_t t);
114 
115 static int ether_output(struct ifnet *, struct mbuf *, struct sockaddr *,
116 			struct rtentry *);
117 
118 /*
119  * bridge support
120  */
121 int do_bridge;
122 bridge_in_t *bridge_in_ptr;
123 bdg_forward_t *bdg_forward_ptr;
124 bdgtakeifaces_t *bdgtakeifaces_ptr;
125 struct bdg_softc *ifp2sc;
126 
127 static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
128 			      struct sockaddr *);
129 
130 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = {
131 	0xff, 0xff, 0xff, 0xff, 0xff, 0xff
132 };
133 
134 #define gotoerr(e) do { error = (e); goto bad; } while (0)
135 #define IFP2AC(ifp) ((struct arpcom *)(ifp))
136 
137 static boolean_t ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
138 				struct ip_fw **rule, struct ether_header *eh,
139 				boolean_t shared);
140 
141 static int ether_ipfw;
142 SYSCTL_DECL(_net_link);
143 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
144 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
145 	   &ether_ipfw, 0, "Pass ether pkts through firewall");
146 
147 /*
148  * Ethernet output routine.
149  * Encapsulate a packet of type family for the local net.
150  * Use trailer local net encapsulation if enough data in first
151  * packet leaves a multiple of 512 bytes of data in remainder.
152  * Assumes that ifp is actually pointer to arpcom structure.
153  */
154 static int
155 ether_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst,
156 	     struct rtentry *rt)
157 {
158 	struct ether_header *eh, *deh;
159 	u_char *edst;
160 	int loop_copy = 0;
161 	int hlen = ETHER_HDR_LEN;	/* link layer header length */
162 	struct arpcom *ac = IFP2AC(ifp);
163 	int error;
164 
165 	if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING))
166 		gotoerr(ENETDOWN);
167 
168 	M_PREPEND(m, sizeof(struct ether_header), MB_DONTWAIT);
169 	if (m == NULL)
170 		return (ENOBUFS);
171 	eh = mtod(m, struct ether_header *);
172 	edst = eh->ether_dhost;
173 
174 	/*
175 	 * Fill in the destination ethernet address and frame type.
176 	 */
177 	switch (dst->sa_family) {
178 #ifdef INET
179 	case AF_INET:
180 		if (!arpresolve(ifp, rt, m, dst, edst))
181 			return (0);	/* if not yet resolved */
182 		eh->ether_type = htons(ETHERTYPE_IP);
183 		break;
184 #endif
185 #ifdef INET6
186 	case AF_INET6:
187 		if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, edst))
188 			return (0);		/* Something bad happenned. */
189 		eh->ether_type = htons(ETHERTYPE_IPV6);
190 		break;
191 #endif
192 #ifdef IPX
193 	case AF_IPX:
194 		if (ef_outputp != NULL) {
195 			error = ef_outputp(ifp, &m, dst, &eh->ether_type,
196 					   &hlen);
197 			if (error)
198 				goto bad;
199 		} else {
200 			eh->ether_type = htons(ETHERTYPE_IPX);
201 			bcopy(&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
202 			      edst, ETHER_ADDR_LEN);
203 		}
204 		break;
205 #endif
206 #ifdef NETATALK
207 	case AF_APPLETALK: {
208 		struct at_ifaddr *aa;
209 
210 		if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) {
211 			error = 0;	/* XXX */
212 			goto bad;
213 		}
214 		/*
215 		 * In the phase 2 case, need to prepend an mbuf for
216 		 * the llc header.  Since we must preserve the value
217 		 * of m, which is passed to us by value, we m_copy()
218 		 * the first mbuf, and use it for our llc header.
219 		 */
220 		if (aa->aa_flags & AFA_PHASE2) {
221 			struct llc llc;
222 
223 			M_PREPEND(m, sizeof(struct llc), MB_DONTWAIT);
224 			eh = mtod(m, struct ether_header *);
225 			edst = eh->ether_dhost;
226 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
227 			llc.llc_control = LLC_UI;
228 			bcopy(at_org_code, llc.llc_snap_org_code,
229 			      sizeof at_org_code);
230 			llc.llc_snap_ether_type = htons(ETHERTYPE_AT);
231 			bcopy(&llc,
232 			      mtod(m, caddr_t) + sizeof(struct ether_header),
233 			      sizeof(struct llc));
234 			eh->ether_type = htons(m->m_pkthdr.len);
235 			hlen = sizeof(struct llc) + ETHER_HDR_LEN;
236 		} else {
237 			eh->ether_type = htons(ETHERTYPE_AT);
238 		}
239 		if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst))
240 			return (0);
241 		break;
242 	  }
243 #endif
244 #ifdef NS
245 	case AF_NS:
246 		switch(ns_nettype) {
247 		default:
248 		case 0x8137:	/* Novell Ethernet_II Ethernet TYPE II */
249 			eh->ether_type = 0x8137;
250 			break;
251 		case 0x0:	/* Novell 802.3 */
252 			eh->ether_type = htons(m->m_pkthdr.len);
253 			break;
254 		case 0xe0e0:	/* Novell 802.2 and Token-Ring */
255 			M_PREPEND(m, 3, MB_DONTWAIT);
256 			eh = mtod(m, struct ether_header *);
257 			edst = eh->ether_dhost;
258 			eh->ether_type = htons(m->m_pkthdr.len);
259 			cp = mtod(m, u_char *) + sizeof(struct ether_header);
260 			*cp++ = 0xE0;
261 			*cp++ = 0xE0;
262 			*cp++ = 0x03;
263 			break;
264 		}
265 		bcopy(&(((struct sockaddr_ns *)dst)->sns_addr.x_host), edst,
266 		      ETHER_ADDR_LEN);
267 		/*
268 		 * XXX if ns_thishost is the same as the node's ethernet
269 		 * address then just the default code will catch this anyhow.
270 		 * So I'm not sure if this next clause should be here at all?
271 		 * [JRE]
272 		 */
273 		if (bcmp(edst, &ns_thishost, ETHER_ADDR_LEN) == 0) {
274 			m->m_pkthdr.rcvif = ifp;
275 			netisr_dispatch(NETISR_NS, m);
276 			return (error);
277 		}
278 		if (bcmp(edst, &ns_broadhost, ETHER_ADDR_LEN) == 0)
279 			m->m_flags |= M_BCAST;
280 		break;
281 #endif
282 	case pseudo_AF_HDRCMPLT:
283 	case AF_UNSPEC:
284 		loop_copy = -1; /* if this is for us, don't do it */
285 		deh = (struct ether_header *)dst->sa_data;
286 		memcpy(edst, deh->ether_dhost, ETHER_ADDR_LEN);
287 		eh->ether_type = deh->ether_type;
288 		break;
289 
290 	default:
291 		printf("%s: can't handle af%d\n", ifp->if_xname,
292 			dst->sa_family);
293 		gotoerr(EAFNOSUPPORT);
294 	}
295 
296 	if (dst->sa_family == pseudo_AF_HDRCMPLT)	/* unlikely */
297 		memcpy(eh->ether_shost,
298 		       ((struct ether_header *)dst->sa_data)->ether_shost,
299 		       ETHER_ADDR_LEN);
300 	else
301 		memcpy(eh->ether_shost, ac->ac_enaddr, ETHER_ADDR_LEN);
302 
303 	/*
304 	 * If a simplex interface, and the packet is being sent to our
305 	 * Ethernet address or a broadcast address, loopback a copy.
306 	 * XXX To make a simplex device behave exactly like a duplex
307 	 * device, we should copy in the case of sending to our own
308 	 * ethernet address (thus letting the original actually appear
309 	 * on the wire). However, we don't do that here for security
310 	 * reasons and compatibility with the original behavior.
311 	 */
312 	if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) {
313 		int csum_flags = 0;
314 
315 		if (m->m_pkthdr.csum_flags & CSUM_IP)
316 			csum_flags |= (CSUM_IP_CHECKED | CSUM_IP_VALID);
317 		if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
318 			csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR);
319 		if ((m->m_flags & M_BCAST) || (loop_copy > 0)) {
320 			struct mbuf *n;
321 
322 			if ((n = m_copypacket(m, MB_DONTWAIT)) != NULL) {
323 				n->m_pkthdr.csum_flags |= csum_flags;
324 				if (csum_flags & CSUM_DATA_VALID)
325 					n->m_pkthdr.csum_data = 0xffff;
326 				if_simloop(ifp, n, dst->sa_family, hlen);
327 			} else
328 				ifp->if_iqdrops++;
329 		} else if (bcmp(eh->ether_dhost, eh->ether_shost,
330 				ETHER_ADDR_LEN) == 0) {
331 			m->m_pkthdr.csum_flags |= csum_flags;
332 			if (csum_flags & CSUM_DATA_VALID)
333 				m->m_pkthdr.csum_data = 0xffff;
334 			if_simloop(ifp, m, dst->sa_family, hlen);
335 			return (0);	/* XXX */
336 		}
337 	}
338 
339 	/* Handle ng_ether(4) processing, if any */
340 	if (ng_ether_output_p != NULL) {
341 		if ((error = (*ng_ether_output_p)(ifp, &m)) != 0)
342 			goto bad;
343 		if (m == NULL)
344 			return (0);
345 	}
346 
347 	/* Continue with link-layer output */
348 	return ether_output_frame(ifp, m);
349 
350 bad:
351 	m_freem(m);
352 	return (error);
353 }
354 
355 /*
356  * Ethernet link layer output routine to send a raw frame to the device.
357  *
358  * This assumes that the 14 byte Ethernet header is present and contiguous
359  * in the first mbuf (if BRIDGE'ing).
360  */
361 int
362 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
363 {
364 	struct ip_fw *rule = NULL;
365 	int error = 0;
366 	int s;
367 
368 	/* Extract info from dummynet tag, ignore others */
369 	while (m->m_type == MT_TAG) {
370 		if (m->m_flags == PACKET_TAG_DUMMYNET) {
371 			rule = ((struct dn_pkt *)m)->rule;
372 			break;
373 		}
374 		m = m->m_next;
375 	}
376 	if (rule != NULL)		/* packet was already bridged */
377 		goto no_bridge;
378 
379 	if (BDG_ACTIVE(ifp)) {
380 		struct ether_header *eh;	/* a pointer suffices */
381 
382 		m->m_pkthdr.rcvif = NULL;
383 		eh = mtod(m, struct ether_header *);
384 		m_adj(m, ETHER_HDR_LEN);
385 		m = bdg_forward_ptr(m, eh, ifp);
386 		m_freem(m);
387 		return (0);
388 	}
389 
390 no_bridge:
391 	s = splimp();
392 	if (IPFW_LOADED && ether_ipfw != 0) {
393 		struct ether_header save_eh, *eh;
394 
395 		eh = mtod(m, struct ether_header *);
396 		save_eh = *eh;
397 		m_adj(m, ETHER_HDR_LEN);
398 		if (!ether_ipfw_chk(&m, ifp, &rule, eh, FALSE)) {
399 			if (m != NULL) {
400 				m_freem(m);
401 				return ENOBUFS; /* pkt dropped */
402 			} else
403 				return 0;	/* consumed e.g. in a pipe */
404 		}
405 		eh = mtod(m, struct ether_header *);
406 		/* packet was ok, restore the ethernet header */
407 		if ((void *)(eh + 1) == (void *)m->m_data) {
408 			m->m_data -= ETHER_HDR_LEN ;
409 			m->m_len += ETHER_HDR_LEN ;
410 			m->m_pkthdr.len += ETHER_HDR_LEN ;
411 		} else {
412 			M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT);
413 			if (m == NULL) /* nope... */
414 				return ENOBUFS;
415 			bcopy(&save_eh, mtod(m, struct ether_header *),
416 			      ETHER_HDR_LEN);
417 		}
418 	}
419 
420 	/*
421 	 * Queue message on interface, update output statistics if
422 	 * successful, and start output if interface not yet active.
423 	 */
424 	if (!IF_HANDOFF(&ifp->if_snd, m, ifp))
425 		error = ENOBUFS;
426 	splx(s);
427 	return (error);
428 }
429 
430 /*
431  * ipfw processing for ethernet packets (in and out).
432  * The second parameter is NULL from ether_demux(), and ifp from
433  * ether_output_frame(). This section of code could be used from
434  * bridge.c as well as long as we use some extra info
435  * to distinguish that case from ether_output_frame().
436  */
437 static boolean_t
438 ether_ipfw_chk(
439 	struct mbuf **m0,
440 	struct ifnet *dst,
441 	struct ip_fw **rule,
442 	struct ether_header *eh,
443 	boolean_t shared)
444 {
445 	struct ether_header save_eh = *eh;	/* might be a ptr in m */
446 	struct ip_fw_args args;
447 	int i;
448 
449 	if (*rule != NULL && fw_one_pass)
450 		return TRUE; /* dummynet packet, already partially processed */
451 
452 	/*
453 	 * I need some amount of data to be contiguous, and in case others
454 	 * need the packet (shared==TRUE), it also better be in the first mbuf.
455 	 */
456 	i = min((*m0)->m_pkthdr.len, max_protohdr);
457 	if (shared || (*m0)->m_len < i) {
458 		*m0 = m_pullup(*m0, i);
459 		if (*m0 == NULL)
460 			return FALSE;
461 	}
462 
463 	args.m = *m0;		/* the packet we are looking at		*/
464 	args.oif = dst;		/* destination, if any			*/
465 	args.divert_rule = 0;	/* we do not support divert yet		*/
466 	args.rule = *rule;	/* matching rule to restart		*/
467 	args.next_hop = NULL;	/* we do not support forward yet	*/
468 	args.eh = &save_eh;	/* MAC header for bridged/MAC packets	*/
469 	i = ip_fw_chk_ptr(&args);
470 	*m0 = args.m;
471 	*rule = args.rule;
472 
473 	if ((i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL)	/* drop */
474 		return FALSE;
475 
476 	if (i == 0)					/* a PASS rule.  */
477 		return TRUE;
478 
479 	if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG)) {
480 		/*
481 		 * Pass the pkt to dummynet, which consumes it.
482 		 * If shared, make a copy and keep the original.
483 		 */
484 		struct mbuf *m ;
485 
486 		if (shared) {
487 			m = m_copypacket(*m0, MB_DONTWAIT);
488 			if (m == NULL)
489 				return FALSE;
490 		} else {
491 			m = *m0 ;	/* pass the original to dummynet */
492 			*m0 = NULL ;	/* and nothing back to the caller */
493 		}
494 		/*
495 		 * Prepend the header, optimize for the common case of
496 		 * eh pointing into the mbuf.
497 		 */
498 		if ((void *)(eh + 1) == (void *)m->m_data) {
499 			m->m_data -= ETHER_HDR_LEN ;
500 			m->m_len += ETHER_HDR_LEN ;
501 			m->m_pkthdr.len += ETHER_HDR_LEN ;
502 		} else {
503 			M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT);
504 			if (m == NULL)
505 				return FALSE;
506 			bcopy(&save_eh, mtod(m, struct ether_header *),
507 			      ETHER_HDR_LEN);
508 		}
509 		ip_dn_io_ptr(m, (i & 0xffff),
510 			     dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
511 		return FALSE;
512 	}
513 	/*
514 	 * XXX at some point add support for divert/forward actions.
515 	 * If none of the above matches, we have to drop the pkt.
516 	 */
517 	return FALSE;
518 }
519 
520 /*
521  * XXX merge this function with ether_input.
522  */
523 static void
524 ether_input_internal(struct ifnet *ifp, struct mbuf *m)
525 {
526 	ether_input(ifp, NULL, m);
527 }
528 
529 /*
530  * Process a received Ethernet packet. We have two different interfaces:
531  * one (conventional) assumes the packet in the mbuf, with the ethernet
532  * header provided separately in *eh. The second one (new) has everything
533  * in the mbuf, and we can tell it because eh == NULL.
534  * The caller MUST MAKE SURE that there are at least
535  * sizeof(struct ether_header) bytes in the first mbuf.
536  *
537  * This allows us to concentrate in one place a bunch of code which
538  * is replicated in all device drivers. Also, many functions called
539  * from ether_input() try to put the eh back into the mbuf, so we
540  * can later propagate the 'contiguous packet' interface to them,
541  * and handle the old interface just here.
542  *
543  * NOTA BENE: for many drivers "eh" is a pointer into the first mbuf or
544  * cluster, right before m_data. So be very careful when working on m,
545  * as you could destroy *eh !!
546  *
547  * First we perform any link layer operations, then continue
548  * to the upper layers with ether_demux().
549  */
550 void
551 ether_input(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m)
552 {
553 	struct ether_header save_eh;
554 
555 	if (eh == NULL) {
556 		if (m->m_len < sizeof(struct ether_header)) {
557 			/* XXX error in the caller. */
558 			m_freem(m);
559 			return;
560 		}
561 		m->m_pkthdr.rcvif = ifp;
562 		eh = mtod(m, struct ether_header *);
563 		m_adj(m, sizeof(struct ether_header));
564 		/* XXX */
565 		/* m->m_pkthdr.len = m->m_len; */
566 	}
567 
568 	/* Check for a BPF tap */
569 	if (ifp->if_bpf != NULL) {
570 		struct m_hdr mh;
571 
572 		/* This kludge is OK; BPF treats the "mbuf" as read-only */
573 		mh.mh_next = m;
574 		mh.mh_data = (char *)eh;
575 		mh.mh_len = ETHER_HDR_LEN;
576 		bpf_mtap(ifp, (struct mbuf *)&mh);
577 	}
578 
579 	ifp->if_ibytes += m->m_pkthdr.len + (sizeof *eh);
580 
581 	/* Handle ng_ether(4) processing, if any */
582 	if (ng_ether_input_p != NULL) {
583 		(*ng_ether_input_p)(ifp, &m, eh);
584 		if (m == NULL)
585 			return;
586 	}
587 
588 	/* Check for bridging mode */
589 	if (BDG_ACTIVE(ifp)) {
590 		struct ifnet *bif;
591 
592 		/* Check with bridging code */
593 		if ((bif = bridge_in_ptr(ifp, eh)) == BDG_DROP) {
594 			m_freem(m);
595 			return;
596 		}
597 		if (bif != BDG_LOCAL) {
598 			save_eh = *eh ; /* because it might change */
599 			m = bdg_forward_ptr(m, eh, bif); /* needs forwarding */
600 			/*
601 			 * Do not continue if bdg_forward_ptr() processed our
602 			 * packet (and cleared the mbuf pointer m) or if
603 			 * it dropped (m_free'd) the packet itself.
604 			 */
605 			if (m == NULL) {
606 			    if (bif == BDG_BCAST || bif == BDG_MCAST)
607 				printf("bdg_forward drop MULTICAST PKT\n");
608 			    return;
609 			}
610 			eh = &save_eh ;
611 		}
612 		if (bif == BDG_LOCAL || bif == BDG_BCAST || bif == BDG_MCAST)
613 			goto recvLocal;		/* receive locally */
614 
615 		/* If not local and not multicast, just drop it */
616 		m_freem(m);
617 		return;
618 	}
619 
620 recvLocal:
621 	/* Continue with upper layer processing */
622 	ether_demux(ifp, eh, m);
623 }
624 
625 /*
626  * Upper layer processing for a received Ethernet packet.
627  */
628 void
629 ether_demux(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m)
630 {
631 	int isr;
632 	u_short ether_type;
633 	struct ip_fw *rule = NULL;
634 #ifdef NETATALK
635 	struct llc *l;
636 #endif
637 
638 	/* Extract info from dummynet tag, ignore others */
639 	while (m->m_type == MT_TAG) {
640 		if (m->m_flags == PACKET_TAG_DUMMYNET) {
641 			rule = ((struct dn_pkt *)m)->rule;
642 			ifp = m->m_next->m_pkthdr.rcvif;
643 			break;
644 		}
645 		m = m->m_next;
646 	}
647 	if (rule)	/* packet was already bridged */
648 		goto post_stats;
649 
650 	/*
651 	 * Discard packet if upper layers shouldn't see it because
652 	 * it was unicast to a different Ethernet address.  If the
653 	 * driver is working properly, then this situation can only
654 	 * happen when the interface is in promiscuous mode.
655 	 */
656 	if (!BDG_ACTIVE(ifp) &&
657 	    ((ifp->if_flags & (IFF_PROMISC | IFF_PPROMISC)) == IFF_PROMISC) &&
658 	    (eh->ether_dhost[0] & 1) == 0 &&
659 	    bcmp(eh->ether_dhost, IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN)) {
660 		m_freem(m);
661 		return;
662 	}
663 	/* Discard packet if interface is not up */
664 	if (!(ifp->if_flags & IFF_UP)) {
665 		m_freem(m);
666 		return;
667 	}
668 	if (eh->ether_dhost[0] & 1) {
669 		if (bcmp(ifp->if_broadcastaddr, eh->ether_dhost,
670 			 ifp->if_addrlen) == 0)
671 			m->m_flags |= M_BCAST;
672 		else
673 			m->m_flags |= M_MCAST;
674 		ifp->if_imcasts++;
675 	}
676 
677 post_stats:
678 	if (IPFW_LOADED && ether_ipfw != 0) {
679 		if (!ether_ipfw_chk(&m, NULL, &rule, eh, FALSE)) {
680 			m_freem(m);
681 			return;
682 		}
683 		eh = mtod(m, struct ether_header *);
684 	}
685 
686 	ether_type = ntohs(eh->ether_type);
687 
688 	switch (ether_type) {
689 #ifdef INET
690 	case ETHERTYPE_IP:
691 		if (ipflow_fastforward(m))
692 			return;
693 		isr = NETISR_IP;
694 		break;
695 
696 	case ETHERTYPE_ARP:
697 		if (ifp->if_flags & IFF_NOARP) {
698 			/* Discard packet if ARP is disabled on interface */
699 			m_freem(m);
700 			return;
701 		}
702 		isr = NETISR_ARP;
703 		break;
704 #endif
705 
706 #ifdef INET6
707 	case ETHERTYPE_IPV6:
708 		isr = NETISR_IPV6;
709 		break;
710 #endif
711 
712 #ifdef IPX
713 	case ETHERTYPE_IPX:
714 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
715 			return;
716 		isr = NETISR_IPX;
717 		break;
718 #endif
719 
720 #ifdef NS
721 	case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
722 		isr = NETISR_NS;
723 		break;
724 
725 #endif
726 
727 #ifdef NETATALK
728 	case ETHERTYPE_AT:
729 		isr = NETISR_ATALK1;
730 		break;
731 	case ETHERTYPE_AARP:
732 		isr = NETISR_AARP;
733 		break;
734 #endif
735 
736 	case ETHERTYPE_VLAN:
737 		if (vlan_input_p != NULL)
738 			(*vlan_input_p)(eh, m);
739 		else {
740 			m->m_pkthdr.rcvif->if_noproto++;
741 			m_freem(m);
742 		}
743 		return;
744 
745 	default:
746 #ifdef IPX
747 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
748 			return;
749 #endif
750 #ifdef NS
751 		checksum = mtod(m, ushort *);
752 		/* Novell 802.3 */
753 		if ((ether_type <= ETHERMTU) &&
754 		    ((*checksum == 0xffff) || (*checksum == 0xE0E0))) {
755 			if (*checksum == 0xE0E0) {
756 				m->m_pkthdr.len -= 3;
757 				m->m_len -= 3;
758 				m->m_data += 3;
759 			}
760 			isr = NETISR_NS;
761 			break;
762 		}
763 #endif
764 #ifdef NETATALK
765 		if (ether_type > ETHERMTU)
766 			goto dropanyway;
767 		l = mtod(m, struct llc *);
768 		if (l->llc_dsap == LLC_SNAP_LSAP &&
769 		    l->llc_ssap == LLC_SNAP_LSAP &&
770 		    l->llc_control == LLC_UI) {
771 			if (bcmp(&(l->llc_snap_org_code)[0], at_org_code,
772 				 sizeof at_org_code) == 0 &&
773 			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
774 				m_adj(m, sizeof(struct llc));
775 				isr = NETISR_ATALK2;
776 				break;
777 			}
778 			if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
779 				 sizeof aarp_org_code) == 0 &&
780 			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
781 				m_adj(m, sizeof(struct llc));
782 				isr = NETISR_AARP;
783 				break;
784 			}
785 		}
786 dropanyway:
787 #endif
788 		if (ng_ether_input_orphan_p != NULL)
789 			(*ng_ether_input_orphan_p)(ifp, m, eh);
790 		else
791 			m_freem(m);
792 		return;
793 	}
794 	netisr_dispatch(isr, m);
795 }
796 
797 /*
798  * Perform common duties while attaching to interface list
799  */
800 
801 void
802 ether_ifattach(struct ifnet *ifp, uint8_t *lla)
803 {
804 	ether_ifattach_bpf(ifp, lla, DLT_EN10MB, sizeof(struct ether_header));
805 }
806 
807 void
808 ether_ifattach_bpf(struct ifnet *ifp, uint8_t *lla, u_int dlt, u_int hdrlen)
809 {
810 	struct ifaddr *ifa;
811 	struct sockaddr_dl *sdl;
812 
813 	ifp->if_type = IFT_ETHER;
814 	ifp->if_addrlen = ETHER_ADDR_LEN;
815 	ifp->if_hdrlen = ETHER_HDR_LEN;
816 	if_attach(ifp);
817 	ifp->if_mtu = ETHERMTU;
818 	if (ifp->if_baudrate == 0)
819 		ifp->if_baudrate = 10000000;
820 	ifp->if_output = ether_output;
821 	ifp->if_input = ether_input_internal;
822 	ifp->if_resolvemulti = ether_resolvemulti;
823 	ifp->if_broadcastaddr = etherbroadcastaddr;
824 	ifa = ifnet_addrs[ifp->if_index - 1];
825 	KASSERT(ifa != NULL, ("%s: no lladdr!\n", __FUNCTION__));
826 	sdl = (struct sockaddr_dl *)ifa->ifa_addr;
827 	sdl->sdl_type = IFT_ETHER;
828 	sdl->sdl_alen = ifp->if_addrlen;
829 	bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
830 	/*
831 	 * XXX Keep the current drivers happy.
832 	 * XXX Remove once all drivers have been cleaned up
833 	 */
834 	if (lla != IFP2AC(ifp)->ac_enaddr)
835 		bcopy(lla, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen);
836 	bpfattach(ifp, dlt, hdrlen);
837 	if (ng_ether_attach_p != NULL)
838 		(*ng_ether_attach_p)(ifp);
839 	if (BDG_LOADED)
840 		bdgtakeifaces_ptr();
841 
842 	if_printf(ifp, "MAC address: %6D\n", lla, ":");
843 }
844 
845 /*
846  * Perform common duties while detaching an Ethernet interface
847  */
848 void
849 ether_ifdetach(struct ifnet *ifp)
850 {
851 	int s;
852 
853 	s = splnet();
854 	if_down(ifp);
855 	splx(s);
856 
857 	if (ng_ether_detach_p != NULL)
858 		(*ng_ether_detach_p)(ifp);
859 	bpfdetach(ifp);
860 	if_detach(ifp);
861 	if (BDG_LOADED)
862 		bdgtakeifaces_ptr();
863 }
864 
865 int
866 ether_ioctl(struct ifnet *ifp, int command, caddr_t data)
867 {
868 	struct ifaddr *ifa = (struct ifaddr *) data;
869 	struct ifreq *ifr = (struct ifreq *) data;
870 	int error = 0;
871 
872 	switch (command) {
873 	case SIOCSIFADDR:
874 		ifp->if_flags |= IFF_UP;
875 
876 		switch (ifa->ifa_addr->sa_family) {
877 #ifdef INET
878 		case AF_INET:
879 			ifp->if_init(ifp->if_softc);	/* before arpwhohas */
880 			arp_ifinit(ifp, ifa);
881 			break;
882 #endif
883 #ifdef IPX
884 		/*
885 		 * XXX - This code is probably wrong
886 		 */
887 		case AF_IPX:
888 			{
889 			struct ipx_addr *ina = &IA_SIPX(ifa)->sipx_addr;
890 			struct arpcom *ac = IFP2AC(ifp);
891 
892 			if (ipx_nullhost(*ina))
893 				ina->x_host = *(union ipx_host *) ac->ac_enaddr;
894 			else
895 				bcopy(ina->x_host.c_host, ac->ac_enaddr,
896 				      sizeof ac->ac_enaddr);
897 
898 			ifp->if_init(ifp->if_softc);	/* Set new address. */
899 			break;
900 			}
901 #endif
902 #ifdef NS
903 		/*
904 		 * XXX - This code is probably wrong
905 		 */
906 		case AF_NS:
907 		{
908 			struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr);
909 			struct arpcom *ac = IFP2AC(ifp);
910 
911 			if (ns_nullhost(*ina))
912 				ina->x_host = *(union ns_host *)(ac->ac_enaddr);
913 			else
914 				bcopy(ina->x_host.c_host, ac->ac_enaddr,
915 				      sizeof ac->ac_enaddr);
916 
917 			/*
918 			 * Set new address
919 			 */
920 			ifp->if_init(ifp->if_softc);
921 			break;
922 		}
923 #endif
924 		default:
925 			ifp->if_init(ifp->if_softc);
926 			break;
927 		}
928 		break;
929 
930 	case SIOCGIFADDR:
931 		bcopy(IFP2AC(ifp)->ac_enaddr,
932 		      ((struct sockaddr *)ifr->ifr_data)->sa_data,
933 		      ETHER_ADDR_LEN);
934 		break;
935 
936 	case SIOCSIFMTU:
937 		/*
938 		 * Set the interface MTU.
939 		 */
940 		if (ifr->ifr_mtu > ETHERMTU) {
941 			error = EINVAL;
942 		} else {
943 			ifp->if_mtu = ifr->ifr_mtu;
944 		}
945 		break;
946 	default:
947 		error = EINVAL;
948 		break;
949 	}
950 	return (error);
951 }
952 
953 int
954 ether_resolvemulti(
955 	struct ifnet *ifp,
956 	struct sockaddr **llsa,
957 	struct sockaddr *sa)
958 {
959 	struct sockaddr_dl *sdl;
960 	struct sockaddr_in *sin;
961 #ifdef INET6
962 	struct sockaddr_in6 *sin6;
963 #endif
964 	u_char *e_addr;
965 
966 	switch(sa->sa_family) {
967 	case AF_LINK:
968 		/*
969 		 * No mapping needed. Just check that it's a valid MC address.
970 		 */
971 		sdl = (struct sockaddr_dl *)sa;
972 		e_addr = LLADDR(sdl);
973 		if ((e_addr[0] & 1) != 1)
974 			return EADDRNOTAVAIL;
975 		*llsa = 0;
976 		return 0;
977 
978 #ifdef INET
979 	case AF_INET:
980 		sin = (struct sockaddr_in *)sa;
981 		if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
982 			return EADDRNOTAVAIL;
983 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
984 		       M_WAITOK | M_ZERO);
985 		sdl->sdl_len = sizeof *sdl;
986 		sdl->sdl_family = AF_LINK;
987 		sdl->sdl_index = ifp->if_index;
988 		sdl->sdl_type = IFT_ETHER;
989 		sdl->sdl_alen = ETHER_ADDR_LEN;
990 		e_addr = LLADDR(sdl);
991 		ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
992 		*llsa = (struct sockaddr *)sdl;
993 		return 0;
994 #endif
995 #ifdef INET6
996 	case AF_INET6:
997 		sin6 = (struct sockaddr_in6 *)sa;
998 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
999 			/*
1000 			 * An IP6 address of 0 means listen to all
1001 			 * of the Ethernet multicast address used for IP6.
1002 			 * (This is used for multicast routers.)
1003 			 */
1004 			ifp->if_flags |= IFF_ALLMULTI;
1005 			*llsa = 0;
1006 			return 0;
1007 		}
1008 		if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1009 			return EADDRNOTAVAIL;
1010 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1011 		       M_WAITOK | M_ZERO);
1012 		sdl->sdl_len = sizeof *sdl;
1013 		sdl->sdl_family = AF_LINK;
1014 		sdl->sdl_index = ifp->if_index;
1015 		sdl->sdl_type = IFT_ETHER;
1016 		sdl->sdl_alen = ETHER_ADDR_LEN;
1017 		e_addr = LLADDR(sdl);
1018 		ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1019 		*llsa = (struct sockaddr *)sdl;
1020 		return 0;
1021 #endif
1022 
1023 	default:
1024 		/*
1025 		 * Well, the text isn't quite right, but it's the name
1026 		 * that counts...
1027 		 */
1028 		return EAFNOSUPPORT;
1029 	}
1030 }
1031 
1032 #if 0
1033 /*
1034  * This is for reference.  We have a table-driven version
1035  * of the little-endian crc32 generator, which is faster
1036  * than the double-loop.
1037  */
1038 uint32_t
1039 ether_crc32_le(const uint8_t *buf, size_t len)
1040 {
1041 	uint32_t c, crc, carry;
1042 	size_t i, j;
1043 
1044 	crc = 0xffffffffU;	/* initial value */
1045 
1046 	for (i = 0; i < len; i++) {
1047 		c = buf[i];
1048 		for (j = 0; j < 8; j++) {
1049 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
1050 			crc >>= 1;
1051 			c >>= 1;
1052 			if (carry)
1053 				crc = (crc ^ ETHER_CRC_POLY_LE);
1054 		}
1055 	}
1056 
1057 	return (crc);
1058 }
1059 #else
1060 uint32_t
1061 ether_crc32_le(const uint8_t *buf, size_t len)
1062 {
1063 	static const uint32_t crctab[] = {
1064 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1065 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1066 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1067 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1068 	};
1069 	uint32_t crc;
1070 	size_t i;
1071 
1072 	crc = 0xffffffffU;	/* initial value */
1073 
1074 	for (i = 0; i < len; i++) {
1075 		crc ^= buf[i];
1076 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1077 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1078 	}
1079 
1080 	return (crc);
1081 }
1082 #endif
1083 
1084 uint32_t
1085 ether_crc32_be(const uint8_t *buf, size_t len)
1086 {
1087 	uint32_t c, crc, carry;
1088 	size_t i, j;
1089 
1090 	crc = 0xffffffffU;	/* initial value */
1091 
1092 	for (i = 0; i < len; i++) {
1093 		c = buf[i];
1094 		for (j = 0; j < 8; j++) {
1095 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
1096 			crc <<= 1;
1097 			c >>= 1;
1098 			if (carry)
1099 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1100 		}
1101 	}
1102 
1103 	return (crc);
1104 }
1105