xref: /freebsd/sys/net/if_ethersubr.c (revision ebbd4fa8c8427d3dd847ba33c45c996e0500e6ff)
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  * $FreeBSD$
35  */
36 
37 #include "opt_atalk.h"
38 #include "opt_inet.h"
39 #include "opt_inet6.h"
40 #include "opt_ipx.h"
41 #include "opt_bdg.h"
42 #include "opt_mac.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/mac.h>
49 #include <sys/malloc.h>
50 #include <sys/mbuf.h>
51 #include <sys/random.h>
52 #include <sys/socket.h>
53 #include <sys/sockio.h>
54 #include <sys/sysctl.h>
55 
56 #include <net/if.h>
57 #include <net/netisr.h>
58 #include <net/route.h>
59 #include <net/if_llc.h>
60 #include <net/if_dl.h>
61 #include <net/if_types.h>
62 #include <net/bpf.h>
63 #include <net/ethernet.h>
64 #include <net/bridge.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 IPX
78 #include <netipx/ipx.h>
79 #include <netipx/ipx_if.h>
80 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m);
81 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp,
82 		struct sockaddr *dst, short *tp, int *hlen);
83 #endif
84 
85 #ifdef NS
86 #include <netns/ns.h>
87 #include <netns/ns_if.h>
88 ushort ns_nettype;
89 int ether_outputdebug = 0;
90 int ether_inputdebug = 0;
91 #endif
92 
93 #ifdef NETATALK
94 #include <netatalk/at.h>
95 #include <netatalk/at_var.h>
96 #include <netatalk/at_extern.h>
97 
98 #define llc_snap_org_code llc_un.type_snap.org_code
99 #define llc_snap_ether_type llc_un.type_snap.ether_type
100 
101 extern u_char	at_org_code[3];
102 extern u_char	aarp_org_code[3];
103 #endif /* NETATALK */
104 
105 /* netgraph node hooks for ng_ether(4) */
106 void	(*ng_ether_input_p)(struct ifnet *ifp,
107 		struct mbuf **mp, struct ether_header *eh);
108 void	(*ng_ether_input_orphan_p)(struct ifnet *ifp,
109 		struct mbuf *m, struct ether_header *eh);
110 int	(*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
111 void	(*ng_ether_attach_p)(struct ifnet *ifp);
112 void	(*ng_ether_detach_p)(struct ifnet *ifp);
113 
114 int	(*vlan_input_p)(struct ether_header *eh, struct mbuf *m);
115 int	(*vlan_input_tag_p)(struct ether_header *eh, struct mbuf *m,
116 		u_int16_t t);
117 
118 /* bridge support */
119 int do_bridge;
120 bridge_in_t *bridge_in_ptr;
121 bdg_forward_t *bdg_forward_ptr;
122 bdgtakeifaces_t *bdgtakeifaces_ptr;
123 struct bdg_softc *ifp2sc;
124 
125 static	int ether_resolvemulti(struct ifnet *, struct sockaddr **,
126 		struct sockaddr *);
127 u_char	etherbroadcastaddr[6] = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
128 #define senderr(e) do { error = (e); goto bad;} while (0)
129 #define IFP2AC(IFP) ((struct arpcom *)IFP)
130 
131 int
132 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
133 	struct ip_fw **rule, struct ether_header *eh, int shared);
134 static int ether_ipfw;
135 
136 /*
137  * Ethernet output routine.
138  * Encapsulate a packet of type family for the local net.
139  * Use trailer local net encapsulation if enough data in first
140  * packet leaves a multiple of 512 bytes of data in remainder.
141  * Assumes that ifp is actually pointer to arpcom structure.
142  */
143 int
144 ether_output(ifp, m, dst, rt0)
145 	register struct ifnet *ifp;
146 	struct mbuf *m;
147 	struct sockaddr *dst;
148 	struct rtentry *rt0;
149 {
150 	short type;
151 	int error = 0, hdrcmplt = 0;
152  	u_char esrc[6], edst[6];
153 	register struct rtentry *rt;
154 	register struct ether_header *eh;
155 	int loop_copy = 0;
156 	int hlen;	/* link layer header lenght */
157 	struct arpcom *ac = IFP2AC(ifp);
158 
159 #ifdef MAC
160 	error = mac_check_ifnet_transmit(ifp, m);
161 	if (error)
162 		senderr(error);
163 #endif
164 
165 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
166 		senderr(ENETDOWN);
167 	rt = rt0;
168 	if (rt) {
169 		if ((rt->rt_flags & RTF_UP) == 0) {
170 			rt0 = rt = rtalloc1(dst, 1, 0UL);
171 			if (rt0)
172 				rt->rt_refcnt--;
173 			else
174 				senderr(EHOSTUNREACH);
175 		}
176 		if (rt->rt_flags & RTF_GATEWAY) {
177 			if (rt->rt_gwroute == 0)
178 				goto lookup;
179 			if (((rt = rt->rt_gwroute)->rt_flags & RTF_UP) == 0) {
180 				rtfree(rt); rt = rt0;
181 			lookup: rt->rt_gwroute = rtalloc1(rt->rt_gateway, 1,
182 							  0UL);
183 				if ((rt = rt->rt_gwroute) == 0)
184 					senderr(EHOSTUNREACH);
185 			}
186 		}
187 		if (rt->rt_flags & RTF_REJECT)
188 			if (rt->rt_rmx.rmx_expire == 0 ||
189 			    time_second < rt->rt_rmx.rmx_expire)
190 				senderr(rt == rt0 ? EHOSTDOWN : EHOSTUNREACH);
191 	}
192 	hlen = ETHER_HDR_LEN;
193 	switch (dst->sa_family) {
194 #ifdef INET
195 	case AF_INET:
196 		if (!arpresolve(ifp, rt, m, dst, edst, rt0))
197 			return (0);	/* if not yet resolved */
198 		type = htons(ETHERTYPE_IP);
199 		break;
200 #endif
201 #ifdef INET6
202 	case AF_INET6:
203 		if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, (u_char *)edst)) {
204 			/* Something bad happened */
205 			return(0);
206 		}
207 		type = htons(ETHERTYPE_IPV6);
208 		break;
209 #endif
210 #ifdef IPX
211 	case AF_IPX:
212 		if (ef_outputp) {
213 		    error = ef_outputp(ifp, &m, dst, &type, &hlen);
214 		    if (error)
215 			goto bad;
216 		} else
217 		    type = htons(ETHERTYPE_IPX);
218  		bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
219 		    (caddr_t)edst, sizeof (edst));
220 		break;
221 #endif
222 #ifdef NETATALK
223 	case AF_APPLETALK:
224 	  {
225 	    struct at_ifaddr *aa;
226 
227 	    if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) {
228 		    goto bad;
229 	    }
230 	    if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst))
231 		    return (0);
232 	    /*
233 	     * In the phase 2 case, need to prepend an mbuf for the llc header.
234 	     * Since we must preserve the value of m, which is passed to us by
235 	     * value, we m_copy() the first mbuf, and use it for our llc header.
236 	     */
237 	    if ( aa->aa_flags & AFA_PHASE2 ) {
238 		struct llc llc;
239 
240 		M_PREPEND(m, sizeof(struct llc), M_TRYWAIT);
241 		llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
242 		llc.llc_control = LLC_UI;
243 		bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code));
244 		llc.llc_snap_ether_type = htons( ETHERTYPE_AT );
245 		bcopy(&llc, mtod(m, caddr_t), sizeof(struct llc));
246 		type = htons(m->m_pkthdr.len);
247 		hlen = sizeof(struct llc) + ETHER_HDR_LEN;
248 	    } else {
249 		type = htons(ETHERTYPE_AT);
250 	    }
251 	    break;
252 	  }
253 #endif /* NETATALK */
254 #ifdef NS
255 	case AF_NS:
256 		switch(ns_nettype){
257 		default:
258 		case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
259 			type = 0x8137;
260 			break;
261 		case 0x0: /* Novell 802.3 */
262 			type = htons( m->m_pkthdr.len);
263 			break;
264 		case 0xe0e0: /* Novell 802.2 and Token-Ring */
265 			M_PREPEND(m, 3, M_TRYWAIT);
266 			type = htons( m->m_pkthdr.len);
267 			cp = mtod(m, u_char *);
268 			*cp++ = 0xE0;
269 			*cp++ = 0xE0;
270 			*cp++ = 0x03;
271 			break;
272 		}
273  		bcopy((caddr_t)&(((struct sockaddr_ns *)dst)->sns_addr.x_host),
274 		    (caddr_t)edst, sizeof (edst));
275 		/*
276 		 * XXX if ns_thishost is the same as the node's ethernet
277 		 * address then just the default code will catch this anyhow.
278 		 * So I'm not sure if this next clause should be here at all?
279 		 * [JRE]
280 		 */
281 		if (!bcmp((caddr_t)edst, (caddr_t)&ns_thishost, sizeof(edst))){
282 			m->m_pkthdr.rcvif = ifp;
283 			inq = &nsintrq;
284 			if (IF_HANDOFF(inq, m, NULL))
285 				schednetisr(NETISR_NS);
286 			return (error);
287 		}
288 		if (!bcmp((caddr_t)edst, (caddr_t)&ns_broadhost, sizeof(edst))){
289 			m->m_flags |= M_BCAST;
290 		}
291 		break;
292 #endif /* NS */
293 
294 	case pseudo_AF_HDRCMPLT:
295 		hdrcmplt = 1;
296 		eh = (struct ether_header *)dst->sa_data;
297 		(void)memcpy(esrc, eh->ether_shost, sizeof (esrc));
298 		/* FALLTHROUGH */
299 
300 	case AF_UNSPEC:
301 		loop_copy = -1; /* if this is for us, don't do it */
302 		eh = (struct ether_header *)dst->sa_data;
303  		(void)memcpy(edst, eh->ether_dhost, sizeof (edst));
304 		type = eh->ether_type;
305 		break;
306 
307 	default:
308 		printf("%s%d: can't handle af%d\n", ifp->if_name, ifp->if_unit,
309 			dst->sa_family);
310 		senderr(EAFNOSUPPORT);
311 	}
312 
313 	/*
314 	 * Add local net header.  If no space in first mbuf,
315 	 * allocate another.
316 	 */
317 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
318 	if (m == 0)
319 		senderr(ENOBUFS);
320 	eh = mtod(m, struct ether_header *);
321 	(void)memcpy(&eh->ether_type, &type,
322 		sizeof(eh->ether_type));
323  	(void)memcpy(eh->ether_dhost, edst, sizeof (edst));
324 	if (hdrcmplt)
325 		(void)memcpy(eh->ether_shost, esrc,
326 			sizeof(eh->ether_shost));
327 	else
328 		(void)memcpy(eh->ether_shost, ac->ac_enaddr,
329 			sizeof(eh->ether_shost));
330 
331 	/*
332 	 * If a simplex interface, and the packet is being sent to our
333 	 * Ethernet address or a broadcast address, loopback a copy.
334 	 * XXX To make a simplex device behave exactly like a duplex
335 	 * device, we should copy in the case of sending to our own
336 	 * ethernet address (thus letting the original actually appear
337 	 * on the wire). However, we don't do that here for security
338 	 * reasons and compatibility with the original behavior.
339 	 */
340 	if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) {
341 		int csum_flags = 0;
342 
343 		if (m->m_pkthdr.csum_flags & CSUM_IP)
344 			csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
345 		if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
346 			csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
347 		if ((m->m_flags & M_BCAST) || (loop_copy > 0)) {
348 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
349 
350 			n->m_pkthdr.csum_flags |= csum_flags;
351 			if (csum_flags & CSUM_DATA_VALID)
352 				n->m_pkthdr.csum_data = 0xffff;
353 
354 			(void) if_simloop(ifp, n, dst->sa_family, hlen);
355 		} else if (bcmp(eh->ether_dhost,
356 		    eh->ether_shost, ETHER_ADDR_LEN) == 0) {
357 			m->m_pkthdr.csum_flags |= csum_flags;
358 			if (csum_flags & CSUM_DATA_VALID)
359 				m->m_pkthdr.csum_data = 0xffff;
360 			(void) if_simloop(ifp, m, dst->sa_family, hlen);
361 			return (0);	/* XXX */
362 		}
363 	}
364 
365 	/* Handle ng_ether(4) processing, if any */
366 	if (ng_ether_output_p != NULL) {
367 		if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
368 bad:			if (m != NULL)
369 				m_freem(m);
370 			return (error);
371 		}
372 		if (m == NULL)
373 			return (0);
374 	}
375 
376 	/* Continue with link-layer output */
377 	return ether_output_frame(ifp, m);
378 }
379 
380 /*
381  * Ethernet link layer output routine to send a raw frame to the device.
382  *
383  * This assumes that the 14 byte Ethernet header is present and contiguous
384  * in the first mbuf (if BRIDGE'ing).
385  */
386 int
387 ether_output_frame(ifp, m)
388 	struct ifnet *ifp;
389 	struct mbuf *m;
390 {
391 	int error = 0;
392 	struct ip_fw *rule = NULL;
393 
394 	/* Extract info from dummynet tag, ignore others */
395 	for (; m->m_type == MT_TAG; m = m->m_next)
396 		if (m->m_flags == PACKET_TAG_DUMMYNET)
397 			rule = ((struct dn_pkt *)m)->rule;
398 
399 	if (rule)	/* packet was already bridged */
400 		goto no_bridge;
401 
402 	if (BDG_ACTIVE(ifp) ) {
403 		struct ether_header *eh; /* a ptr suffices */
404 
405 		m->m_pkthdr.rcvif = NULL;
406 		eh = mtod(m, struct ether_header *);
407 		m_adj(m, ETHER_HDR_LEN);
408 		m = bdg_forward_ptr(m, eh, ifp);
409 		if (m != NULL)
410 			m_freem(m);
411 		return (0);
412 	}
413 
414 no_bridge:
415 	if (IPFW_LOADED && ether_ipfw != 0) {
416 		struct ether_header save_eh, *eh;
417 
418 		eh = mtod(m, struct ether_header *);
419 		save_eh = *eh;
420 		m_adj(m, ETHER_HDR_LEN);
421 		if (ether_ipfw_chk(&m, ifp, &rule, eh, 0) == 0) {
422 			if (m) {
423 				m_freem(m);
424 				return ENOBUFS;	/* pkt dropped */
425 			} else
426 				return 0;	/* consumed e.g. in a pipe */
427 		}
428 		/* packet was ok, restore the ethernet header */
429 		if ( (void *)(eh + 1) == (void *)m->m_data) {
430 			m->m_data -= ETHER_HDR_LEN ;
431 			m->m_len += ETHER_HDR_LEN ;
432 			m->m_pkthdr.len += ETHER_HDR_LEN ;
433 		} else {
434 			M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
435 			if (m == NULL) /* nope... */
436 				return ENOBUFS;
437 			bcopy(&save_eh, mtod(m, struct ether_header *),
438 			    ETHER_HDR_LEN);
439 		}
440 	}
441 
442 	/*
443 	 * Queue message on interface, update output statistics if
444 	 * successful, and start output if interface not yet active.
445 	 */
446 	if (! IF_HANDOFF(&ifp->if_snd, m, ifp))
447 		return (ENOBUFS);
448 	return (error);
449 }
450 
451 /*
452  * ipfw processing for ethernet packets (in and out).
453  * The second parameter is NULL from ether_demux, and ifp from
454  * ether_output_frame. This section of code could be used from
455  * bridge.c as well as long as we use some extra info
456  * to distinguish that case from ether_output_frame();
457  */
458 int
459 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
460 	struct ip_fw **rule, struct ether_header *eh, int shared)
461 {
462 	struct ether_header save_eh = *eh;	/* might be a ptr in m */
463 	int i;
464 	struct ip_fw_args args;
465 
466 	if (*rule != NULL) /* dummynet packet, already partially processed */
467 		return 1; /* HACK! I should obey the fw_one_pass */
468 
469 	/*
470 	 * I need some amt of data to be contiguous, and in case others need
471 	 * the packet (shared==1) also better be in the first mbuf.
472 	 */
473 	i = min( (*m0)->m_pkthdr.len, max_protohdr);
474 	if ( shared || (*m0)->m_len < i) {
475 		*m0 = m_pullup(*m0, i);
476 		if (*m0 == NULL)
477 			return 0;
478 	}
479 
480 	args.m = *m0;		/* the packet we are looking at		*/
481 	args.oif = dst;		/* destination, if any			*/
482 	args.divert_rule = 0;	/* we do not support divert yet		*/
483 	args.rule = *rule;	/* matching rule to restart		*/
484 	args.next_hop = NULL;	/* we do not support forward yet	*/
485 	args.eh = &save_eh;	/* MAC header for bridged/MAC packets	*/
486 	i = ip_fw_chk_ptr(&args);
487 	*m0 = args.m;
488 	*rule = args.rule;
489 
490 	if ( (i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL) /* drop */
491 		return 0;
492 
493 	if (i == 0) /* a PASS rule.  */
494 		return 1;
495 
496 	if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG)) {
497 		/*
498 		 * Pass the pkt to dummynet, which consumes it.
499 		 * If shared, make a copy and keep the original.
500 		 */
501 		struct mbuf *m ;
502 
503 		if (shared) {
504 			m = m_copypacket(*m0, M_DONTWAIT);
505 			if (m == NULL)
506 				return 0;
507 		} else {
508 			m = *m0 ; /* pass the original to dummynet */
509 			*m0 = NULL ; /* and nothing back to the caller */
510 		}
511 		/*
512 		 * Prepend the header, optimize for the common case of
513 		 * eh pointing into the mbuf.
514 		 */
515 		if ( (void *)(eh + 1) == (void *)m->m_data) {
516 			m->m_data -= ETHER_HDR_LEN ;
517 			m->m_len += ETHER_HDR_LEN ;
518 			m->m_pkthdr.len += ETHER_HDR_LEN ;
519 		} else {
520 			M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
521 			if (m == NULL) /* nope... */
522 				return 0;
523 			bcopy(&save_eh, mtod(m, struct ether_header *),
524 			    ETHER_HDR_LEN);
525 		}
526 		ip_dn_io_ptr(m, (i & 0xffff),
527 			dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
528 		return 0;
529 	}
530 	/*
531 	 * XXX at some point add support for divert/forward actions.
532 	 * If none of the above matches, we have to drop the pkt.
533 	 */
534 	return 0;
535 }
536 
537 /*
538  * Process a received Ethernet packet;
539  * the packet is in the mbuf chain m without
540  * the ether header, which is provided separately.
541  *
542  * NOTA BENE: for many drivers "eh" is a pointer into the first mbuf or
543  * cluster, right before m_data. So be very careful when working on m,
544  * as you could destroy *eh !!
545  * A (probably) more convenient and efficient interface to ether_input
546  * is to have the whole packet (with the ethernet header) into the mbuf:
547  * modules which do not need the ethernet header can easily drop it, while
548  * others (most noticeably bridge and ng_ether) do not need to do additional
549  * work to put the ethernet header back into the mbuf.
550  *
551  * First we perform any link layer operations, then continue
552  * to the upper layers with ether_demux().
553  */
554 void
555 ether_input(ifp, eh, m)
556 	struct ifnet *ifp;
557 	struct ether_header *eh;
558 	struct mbuf *m;
559 {
560 	struct ether_header save_eh;
561 
562 #ifdef MAC
563 	mac_create_mbuf_from_ifnet(ifp, m);
564 #endif
565 
566 	/* Check for a BPF tap */
567 	if (ifp->if_bpf != NULL) {
568 		struct m_hdr mh;
569 
570 		/* This kludge is OK; BPF treats the "mbuf" as read-only */
571 		mh.mh_next = m;
572 		mh.mh_data = (char *)eh;
573 		mh.mh_len = ETHER_HDR_LEN;
574 		bpf_mtap(ifp, (struct mbuf *)&mh);
575 	}
576 
577 	ifp->if_ibytes += m->m_pkthdr.len + sizeof (*eh);
578 
579 	/* Handle ng_ether(4) processing, if any */
580 	if (ng_ether_input_p != NULL) {
581 		(*ng_ether_input_p)(ifp, &m, eh);
582 		if (m == NULL)
583 			return;
584 	}
585 
586 	/* Check for bridging mode */
587 	if (BDG_ACTIVE(ifp) ) {
588 		struct ifnet *bif;
589 
590 		/* Check with bridging code */
591 		if ((bif = bridge_in_ptr(ifp, eh)) == BDG_DROP) {
592 			m_freem(m);
593 			return;
594 		}
595 		if (bif != BDG_LOCAL) {
596 			struct mbuf *oldm = m ;
597 
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 			if (m != oldm) /* m changed! */
611 			    eh = &save_eh ;
612 		}
613 		if (bif == BDG_LOCAL
614 		    || bif == BDG_BCAST
615 		    || bif == BDG_MCAST)
616 			goto recvLocal;			/* receive locally */
617 
618 		/* If not local and not multicast, just drop it */
619 		if (m != NULL)
620 			m_freem(m);
621 		return;
622        }
623 
624 recvLocal:
625 	/* Continue with upper layer processing */
626 	ether_demux(ifp, eh, m);
627 	/* First chunk of an mbuf contains good junk */
628 	if (harvest.ethernet)
629 		random_harvest(m, 16, 3, 0, RANDOM_NET);
630 }
631 
632 /*
633  * Upper layer processing for a received Ethernet packet.
634  */
635 void
636 ether_demux(ifp, eh, m)
637 	struct ifnet *ifp;
638 	struct ether_header *eh;
639 	struct mbuf *m;
640 {
641 	struct ifqueue *inq;
642 	u_short ether_type;
643 #if defined(NETATALK)
644 	register struct llc *l;
645 #endif
646 	struct ip_fw *rule = NULL;
647 
648 	/* Extract info from dummynet tag, ignore others */
649 	for (;m->m_type == MT_TAG; m = m->m_next)
650 		if (m->m_flags == PACKET_TAG_DUMMYNET) {
651 			rule = ((struct dn_pkt *)m)->rule;
652 			ifp = m->m_next->m_pkthdr.rcvif;
653 		}
654 
655 	if (rule)	/* packet was already bridged */
656 		goto post_stats;
657 
658     if (! (BDG_ACTIVE(ifp) ) )
659 	/* Discard packet if upper layers shouldn't see it because it was
660 	   unicast to a different Ethernet address. If the driver is working
661 	   properly, then this situation can only happen when the interface
662 	   is in promiscuous mode. */
663 	if ((ifp->if_flags & IFF_PROMISC) != 0
664 	    && (eh->ether_dhost[0] & 1) == 0
665 	    && bcmp(eh->ether_dhost,
666 	      IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN) != 0) {
667 		m_freem(m);
668 		return;
669 	}
670 
671 	/* Discard packet if interface is not up */
672 	if ((ifp->if_flags & IFF_UP) == 0) {
673 		m_freem(m);
674 		return;
675 	}
676 	if (eh->ether_dhost[0] & 1) {
677 		if (bcmp((caddr_t)etherbroadcastaddr, (caddr_t)eh->ether_dhost,
678 			 sizeof(etherbroadcastaddr)) == 0)
679 			m->m_flags |= M_BCAST;
680 		else
681 			m->m_flags |= M_MCAST;
682 	}
683 	if (m->m_flags & (M_BCAST|M_MCAST))
684 		ifp->if_imcasts++;
685 
686 post_stats:
687 	if (IPFW_LOADED && ether_ipfw != 0) {
688 		if (ether_ipfw_chk(&m, NULL, &rule, eh, 0 ) == 0) {
689 			if (m)
690 				m_freem(m);
691 			return;
692 		}
693 	}
694 
695 	ether_type = ntohs(eh->ether_type);
696 
697 	switch (ether_type) {
698 #ifdef INET
699 	case ETHERTYPE_IP:
700 		if (ipflow_fastforward(m))
701 			return;
702 		schednetisr(NETISR_IP);
703 		inq = &ipintrq;
704 		break;
705 
706 	case ETHERTYPE_ARP:
707 		if (ifp->if_flags & IFF_NOARP) {
708 			/* Discard packet if ARP is disabled on interface */
709 			m_freem(m);
710 			return;
711 		}
712 		schednetisr(NETISR_ARP);
713 		inq = &arpintrq;
714 		break;
715 #endif
716 #ifdef IPX
717 	case ETHERTYPE_IPX:
718 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
719 			return;
720 		schednetisr(NETISR_IPX);
721 		inq = &ipxintrq;
722 		break;
723 #endif
724 #ifdef INET6
725 	case ETHERTYPE_IPV6:
726 		schednetisr(NETISR_IPV6);
727 		inq = &ip6intrq;
728 		break;
729 #endif
730 #ifdef NS
731 	case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
732 		schednetisr(NETISR_NS);
733 		inq = &nsintrq;
734 		break;
735 
736 #endif /* NS */
737 #ifdef NETATALK
738         case ETHERTYPE_AT:
739                 schednetisr(NETISR_ATALK);
740                 inq = &atintrq1;
741                 break;
742         case ETHERTYPE_AARP:
743 		/* probably this should be done with a NETISR as well */
744                 aarpinput(IFP2AC(ifp), m); /* XXX */
745                 return;
746 #endif /* NETATALK */
747 	case ETHERTYPE_VLAN:
748 		/* XXX lock ? */
749 		if (vlan_input_p != NULL)
750 			(*vlan_input_p)(eh, m);
751 		else {
752 			m->m_pkthdr.rcvif->if_noproto++;
753 			m_freem(m);
754 		}
755 		/* XXX unlock ? */
756 		return;
757 	default:
758 #ifdef IPX
759 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
760 			return;
761 #endif /* IPX */
762 #ifdef NS
763 		checksum = mtod(m, ushort *);
764 		/* Novell 802.3 */
765 		if ((ether_type <= ETHERMTU) &&
766 			((*checksum == 0xffff) || (*checksum == 0xE0E0))){
767 			if(*checksum == 0xE0E0) {
768 				m->m_pkthdr.len -= 3;
769 				m->m_len -= 3;
770 				m->m_data += 3;
771 			}
772 				schednetisr(NETISR_NS);
773 				inq = &nsintrq;
774 				break;
775 		}
776 #endif /* NS */
777 #if defined(NETATALK)
778 		if (ether_type > ETHERMTU)
779 			goto dropanyway;
780 		l = mtod(m, struct llc *);
781 		switch (l->llc_dsap) {
782 		case LLC_SNAP_LSAP:
783 		    switch (l->llc_control) {
784 		    case LLC_UI:
785 			if (l->llc_ssap != LLC_SNAP_LSAP)
786 			    goto dropanyway;
787 
788 			if (Bcmp(&(l->llc_snap_org_code)[0], at_org_code,
789 				   sizeof(at_org_code)) == 0 &&
790 			     ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
791 			    inq = &atintrq2;
792 			    m_adj( m, sizeof( struct llc ));
793 			    schednetisr(NETISR_ATALK);
794 			    break;
795 			}
796 
797 			if (Bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
798 				   sizeof(aarp_org_code)) == 0 &&
799 			     ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
800 			    m_adj( m, sizeof( struct llc ));
801 			    aarpinput(IFP2AC(ifp), m); /* XXX */
802 			    return;
803 			}
804 
805 		    default:
806 			goto dropanyway;
807 		    }
808 		    break;
809 		dropanyway:
810 		default:
811 			if (ng_ether_input_orphan_p != NULL)
812 				(*ng_ether_input_orphan_p)(ifp, m, eh);
813 			else
814 				m_freem(m);
815 			return;
816 		}
817 #else /* NETATALK */
818 		if (ng_ether_input_orphan_p != NULL)
819 			(*ng_ether_input_orphan_p)(ifp, m, eh);
820 		else
821 			m_freem(m);
822 		return;
823 #endif /* NETATALK */
824 	}
825 
826 	(void) IF_HANDOFF(inq, m, NULL);
827 }
828 
829 /*
830  * Perform common duties while attaching to interface list
831  */
832 void
833 ether_ifattach(ifp, bpf)
834 	register struct ifnet *ifp;
835 	int bpf;
836 {
837 	register struct ifaddr *ifa;
838 	register struct sockaddr_dl *sdl;
839 
840 	ifp->if_type = IFT_ETHER;
841 	ifp->if_addrlen = 6;
842 	ifp->if_hdrlen = 14;
843 	if_attach(ifp);
844 	ifp->if_mtu = ETHERMTU;
845 	ifp->if_resolvemulti = ether_resolvemulti;
846 	if (ifp->if_baudrate == 0)
847 	    ifp->if_baudrate = 10000000;
848 	ifp->if_broadcastaddr = etherbroadcastaddr;
849 	ifa = ifaddr_byindex(ifp->if_index);
850 	KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
851 	sdl = (struct sockaddr_dl *)ifa->ifa_addr;
852 	sdl->sdl_type = IFT_ETHER;
853 	sdl->sdl_alen = ifp->if_addrlen;
854 	bcopy((IFP2AC(ifp))->ac_enaddr, LLADDR(sdl), ifp->if_addrlen);
855 	if (bpf)
856 		bpfattach(ifp, DLT_EN10MB, sizeof(struct ether_header));
857 	if (ng_ether_attach_p != NULL)
858 		(*ng_ether_attach_p)(ifp);
859 	if (BDG_LOADED)
860 		bdgtakeifaces_ptr();
861 }
862 
863 /*
864  * Perform common duties while detaching an Ethernet interface
865  */
866 void
867 ether_ifdetach(ifp, bpf)
868 	struct ifnet *ifp;
869 	int bpf;
870 {
871 	if (ng_ether_detach_p != NULL)
872 		(*ng_ether_detach_p)(ifp);
873 	if (bpf)
874 		bpfdetach(ifp);
875 	if_detach(ifp);
876 	if (BDG_LOADED)
877 		bdgtakeifaces_ptr();
878 }
879 
880 SYSCTL_DECL(_net_link);
881 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
882 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
883 	    &ether_ipfw,0,"Pass ether pkts through firewall");
884 
885 int
886 ether_ioctl(ifp, command, data)
887 	struct ifnet *ifp;
888 	int command;
889 	caddr_t data;
890 {
891 	struct ifaddr *ifa = (struct ifaddr *) data;
892 	struct ifreq *ifr = (struct ifreq *) data;
893 	int error = 0;
894 
895 	switch (command) {
896 	case SIOCSIFADDR:
897 		ifp->if_flags |= IFF_UP;
898 
899 		switch (ifa->ifa_addr->sa_family) {
900 #ifdef INET
901 		case AF_INET:
902 			ifp->if_init(ifp->if_softc);	/* before arpwhohas */
903 			arp_ifinit(ifp, ifa);
904 			break;
905 #endif
906 #ifdef IPX
907 		/*
908 		 * XXX - This code is probably wrong
909 		 */
910 		case AF_IPX:
911 			{
912 			register struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr);
913 			struct arpcom *ac = IFP2AC(ifp);
914 
915 			if (ipx_nullhost(*ina))
916 				ina->x_host =
917 				    *(union ipx_host *)
918 			            ac->ac_enaddr;
919 			else {
920 				bcopy((caddr_t) ina->x_host.c_host,
921 				      (caddr_t) ac->ac_enaddr,
922 				      sizeof(ac->ac_enaddr));
923 			}
924 
925 			/*
926 			 * Set new address
927 			 */
928 			ifp->if_init(ifp->if_softc);
929 			break;
930 			}
931 #endif
932 #ifdef NS
933 		/*
934 		 * XXX - This code is probably wrong
935 		 */
936 		case AF_NS:
937 		{
938 			register struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr);
939 			struct arpcom *ac = IFP2AC(ifp);
940 
941 			if (ns_nullhost(*ina))
942 				ina->x_host =
943 				    *(union ns_host *) (ac->ac_enaddr);
944 			else {
945 				bcopy((caddr_t) ina->x_host.c_host,
946 				      (caddr_t) ac->ac_enaddr,
947 				      sizeof(ac->ac_enaddr));
948 			}
949 
950 			/*
951 			 * Set new address
952 			 */
953 			ifp->if_init(ifp->if_softc);
954 			break;
955 		}
956 #endif
957 		default:
958 			ifp->if_init(ifp->if_softc);
959 			break;
960 		}
961 		break;
962 
963 	case SIOCGIFADDR:
964 		{
965 			struct sockaddr *sa;
966 
967 			sa = (struct sockaddr *) & ifr->ifr_data;
968 			bcopy(IFP2AC(ifp)->ac_enaddr,
969 			      (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
970 		}
971 		break;
972 
973 	case SIOCSIFMTU:
974 		/*
975 		 * Set the interface MTU.
976 		 */
977 		if (ifr->ifr_mtu > ETHERMTU) {
978 			error = EINVAL;
979 		} else {
980 			ifp->if_mtu = ifr->ifr_mtu;
981 		}
982 		break;
983 	}
984 	return (error);
985 }
986 
987 int
988 ether_resolvemulti(ifp, llsa, sa)
989 	struct ifnet *ifp;
990 	struct sockaddr **llsa;
991 	struct sockaddr *sa;
992 {
993 	struct sockaddr_dl *sdl;
994 	struct sockaddr_in *sin;
995 #ifdef INET6
996 	struct sockaddr_in6 *sin6;
997 #endif
998 	u_char *e_addr;
999 
1000 	switch(sa->sa_family) {
1001 	case AF_LINK:
1002 		/*
1003 		 * No mapping needed. Just check that it's a valid MC address.
1004 		 */
1005 		sdl = (struct sockaddr_dl *)sa;
1006 		e_addr = LLADDR(sdl);
1007 		if ((e_addr[0] & 1) != 1)
1008 			return EADDRNOTAVAIL;
1009 		*llsa = 0;
1010 		return 0;
1011 
1012 #ifdef INET
1013 	case AF_INET:
1014 		sin = (struct sockaddr_in *)sa;
1015 		if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1016 			return EADDRNOTAVAIL;
1017 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1018 		       M_WAITOK|M_ZERO);
1019 		sdl->sdl_len = sizeof *sdl;
1020 		sdl->sdl_family = AF_LINK;
1021 		sdl->sdl_index = ifp->if_index;
1022 		sdl->sdl_type = IFT_ETHER;
1023 		sdl->sdl_alen = ETHER_ADDR_LEN;
1024 		e_addr = LLADDR(sdl);
1025 		ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1026 		*llsa = (struct sockaddr *)sdl;
1027 		return 0;
1028 #endif
1029 #ifdef INET6
1030 	case AF_INET6:
1031 		sin6 = (struct sockaddr_in6 *)sa;
1032 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1033 			/*
1034 			 * An IP6 address of 0 means listen to all
1035 			 * of the Ethernet multicast address used for IP6.
1036 			 * (This is used for multicast routers.)
1037 			 */
1038 			ifp->if_flags |= IFF_ALLMULTI;
1039 			*llsa = 0;
1040 			return 0;
1041 		}
1042 		if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1043 			return EADDRNOTAVAIL;
1044 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1045 		       M_WAITOK|M_ZERO);
1046 		sdl->sdl_len = sizeof *sdl;
1047 		sdl->sdl_family = AF_LINK;
1048 		sdl->sdl_index = ifp->if_index;
1049 		sdl->sdl_type = IFT_ETHER;
1050 		sdl->sdl_alen = ETHER_ADDR_LEN;
1051 		e_addr = LLADDR(sdl);
1052 		ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1053 		*llsa = (struct sockaddr *)sdl;
1054 		return 0;
1055 #endif
1056 
1057 	default:
1058 		/*
1059 		 * Well, the text isn't quite right, but it's the name
1060 		 * that counts...
1061 		 */
1062 		return EAFNOSUPPORT;
1063 	}
1064 }
1065 
1066