xref: /dflybsd-src/sys/net/if_ethersubr.c (revision bc49aa1be5400e3bdd801519c6936e8947d5d432)
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 1.17 2004/07/17 09:43:05 joerg 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,
80 		struct sockaddr *dst, 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 ether_header *eh, struct mbuf *m,
114 		u_int16_t t);
115 
116 /* bridge support */
117 int do_bridge;
118 bridge_in_t *bridge_in_ptr;
119 bdg_forward_t *bdg_forward_ptr;
120 bdgtakeifaces_t *bdgtakeifaces_ptr;
121 struct bdg_softc *ifp2sc;
122 
123 static	int ether_resolvemulti(struct ifnet *, struct sockaddr **,
124 		struct sockaddr *);
125 const uint8_t	etherbroadcastaddr[ETHER_ADDR_LEN] = {
126 	0xff, 0xff, 0xff, 0xff, 0xff, 0xff
127 };
128 
129 #define senderr(e) do { error = (e); goto bad;} while (0)
130 #define IFP2AC(IFP) ((struct arpcom *)IFP)
131 
132 int
133 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
134 	struct ip_fw **rule, struct ether_header *eh, int shared);
135 static int ether_ipfw;
136 
137 /*
138  * Ethernet output routine.
139  * Encapsulate a packet of type family for the local net.
140  * Use trailer local net encapsulation if enough data in first
141  * packet leaves a multiple of 512 bytes of data in remainder.
142  * Assumes that ifp is actually pointer to arpcom structure.
143  */
144 int
145 ether_output(ifp, m, dst, rt0)
146 	struct ifnet *ifp;
147 	struct mbuf *m;
148 	struct sockaddr *dst;
149 	struct rtentry *rt0;
150 {
151 	short type;
152 	int error = 0, hdrcmplt = 0;
153  	u_char esrc[6], edst[6];
154 	struct rtentry *rt;
155 	struct ether_header *eh;
156 	int loop_copy = 0;
157 	int hlen;	/* link layer header lenght */
158 	struct arpcom *ac = IFP2AC(ifp);
159 
160 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
161 		senderr(ENETDOWN);
162 	rt = rt0;
163 	if (rt) {
164 		if ((rt->rt_flags & RTF_UP) == 0) {
165 			rt0 = rt = rtalloc1(dst, 1, 0UL);
166 			if (rt0)
167 				rt->rt_refcnt--;
168 			else
169 				senderr(EHOSTUNREACH);
170 		}
171 		if (rt->rt_flags & RTF_GATEWAY) {
172 			if (rt->rt_gwroute == 0)
173 				goto lookup;
174 			if (((rt = rt->rt_gwroute)->rt_flags & RTF_UP) == 0) {
175 				rtfree(rt); rt = rt0;
176 			lookup: rt->rt_gwroute = rtalloc1(rt->rt_gateway, 1,
177 							  0UL);
178 				if ((rt = rt->rt_gwroute) == 0)
179 					senderr(EHOSTUNREACH);
180 			}
181 		}
182 		if (rt->rt_flags & RTF_REJECT)
183 			if (rt->rt_rmx.rmx_expire == 0 ||
184 			    time_second < rt->rt_rmx.rmx_expire)
185 				senderr(rt == rt0 ? EHOSTDOWN : EHOSTUNREACH);
186 	}
187 	hlen = ETHER_HDR_LEN;
188 	switch (dst->sa_family) {
189 #ifdef INET
190 	case AF_INET:
191 		if (!arpresolve(ifp, rt, m, dst, edst, rt0))
192 			return (0);	/* if not yet resolved */
193 		type = htons(ETHERTYPE_IP);
194 		break;
195 #endif
196 #ifdef INET6
197 	case AF_INET6:
198 		if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, (u_char *)edst)) {
199 			/* Something bad happened */
200 			return(0);
201 		}
202 		type = htons(ETHERTYPE_IPV6);
203 		break;
204 #endif
205 #ifdef IPX
206 	case AF_IPX:
207 		if (ef_outputp) {
208 		    error = ef_outputp(ifp, &m, dst, &type, &hlen);
209 		    if (error)
210 			goto bad;
211 		} else
212 		    type = htons(ETHERTYPE_IPX);
213  		bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
214 		    (caddr_t)edst, sizeof (edst));
215 		break;
216 #endif
217 #ifdef NETATALK
218 	case AF_APPLETALK:
219 	  {
220 	    struct at_ifaddr *aa;
221 
222 	    if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) {
223 		    goto bad;
224 	    }
225 	    if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst))
226 		    return (0);
227 	    /*
228 	     * In the phase 2 case, need to prepend an mbuf for the llc header.
229 	     * Since we must preserve the value of m, which is passed to us by
230 	     * value, we m_copy() the first mbuf, and use it for our llc header.
231 	     */
232 	    if ( aa->aa_flags & AFA_PHASE2 ) {
233 		struct llc llc;
234 
235 		M_PREPEND(m, sizeof(struct llc), MB_WAIT);
236 		llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
237 		llc.llc_control = LLC_UI;
238 		bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code));
239 		llc.llc_snap_ether_type = htons( ETHERTYPE_AT );
240 		bcopy(&llc, mtod(m, caddr_t), sizeof(struct llc));
241 		type = htons(m->m_pkthdr.len);
242 		hlen = sizeof(struct llc) + ETHER_HDR_LEN;
243 	    } else {
244 		type = htons(ETHERTYPE_AT);
245 	    }
246 	    break;
247 	  }
248 #endif /* NETATALK */
249 #ifdef NS
250 	case AF_NS:
251 		switch(ns_nettype){
252 		default:
253 		case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
254 			type = 0x8137;
255 			break;
256 		case 0x0: /* Novell 802.3 */
257 			type = htons( m->m_pkthdr.len);
258 			break;
259 		case 0xe0e0: /* Novell 802.2 and Token-Ring */
260 			M_PREPEND(m, 3, MB_WAIT);
261 			type = htons( m->m_pkthdr.len);
262 			cp = mtod(m, u_char *);
263 			*cp++ = 0xE0;
264 			*cp++ = 0xE0;
265 			*cp++ = 0x03;
266 			break;
267 		}
268  		bcopy((caddr_t)&(((struct sockaddr_ns *)dst)->sns_addr.x_host),
269 		    (caddr_t)edst, sizeof (edst));
270 		/*
271 		 * XXX if ns_thishost is the same as the node's ethernet
272 		 * address then just the default code will catch this anyhow.
273 		 * So I'm not sure if this next clause should be here at all?
274 		 * [JRE]
275 		 */
276 		if (!bcmp((caddr_t)edst, (caddr_t)&ns_thishost, sizeof(edst))){
277 			m->m_pkthdr.rcvif = ifp;
278 			netisr_dispatch(NETISR_NS, m);
279 			return (error);
280 		}
281 		if (!bcmp((caddr_t)edst, (caddr_t)&ns_broadhost, sizeof(edst))){
282 			m->m_flags |= M_BCAST;
283 		}
284 		break;
285 #endif /* NS */
286 
287 	case pseudo_AF_HDRCMPLT:
288 		hdrcmplt = 1;
289 		eh = (struct ether_header *)dst->sa_data;
290 		(void)memcpy(esrc, eh->ether_shost, sizeof (esrc));
291 		/* FALLTHROUGH */
292 
293 	case AF_UNSPEC:
294 		loop_copy = -1; /* if this is for us, don't do it */
295 		eh = (struct ether_header *)dst->sa_data;
296  		(void)memcpy(edst, eh->ether_dhost, sizeof (edst));
297 		type = eh->ether_type;
298 		break;
299 
300 	default:
301 		printf("%s: can't handle af%d\n", ifp->if_xname,
302 			dst->sa_family);
303 		senderr(EAFNOSUPPORT);
304 	}
305 
306 	/*
307 	 * Add local net header.  If no space in first mbuf,
308 	 * allocate another.
309 	 */
310 	M_PREPEND(m, sizeof (struct ether_header), MB_DONTWAIT);
311 	if (m == 0)
312 		senderr(ENOBUFS);
313 	eh = mtod(m, struct ether_header *);
314 	(void)memcpy(&eh->ether_type, &type,
315 		sizeof(eh->ether_type));
316  	(void)memcpy(eh->ether_dhost, edst, sizeof (edst));
317 	if (hdrcmplt)
318 		(void)memcpy(eh->ether_shost, esrc,
319 			sizeof(eh->ether_shost));
320 	else
321 		(void)memcpy(eh->ether_shost, ac->ac_enaddr,
322 			sizeof(eh->ether_shost));
323 
324 	/*
325 	 * If a simplex interface, and the packet is being sent to our
326 	 * Ethernet address or a broadcast address, loopback a copy.
327 	 * XXX To make a simplex device behave exactly like a duplex
328 	 * device, we should copy in the case of sending to our own
329 	 * ethernet address (thus letting the original actually appear
330 	 * on the wire). However, we don't do that here for security
331 	 * reasons and compatibility with the original behavior.
332 	 */
333 	if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) {
334 		int csum_flags = 0;
335 
336 		if (m->m_pkthdr.csum_flags & CSUM_IP)
337 			csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
338 		if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
339 			csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
340 		if ((m->m_flags & M_BCAST) || (loop_copy > 0)) {
341 			struct mbuf *n;
342 
343 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
344 				n->m_pkthdr.csum_flags |= csum_flags;
345 				if (csum_flags & CSUM_DATA_VALID)
346 					n->m_pkthdr.csum_data = 0xffff;
347 				(void)if_simloop(ifp, n, dst->sa_family, hlen);
348 			} else
349 				ifp->if_iqdrops++;
350 		} else if (bcmp(eh->ether_dhost,
351 		    eh->ether_shost, ETHER_ADDR_LEN) == 0) {
352 			m->m_pkthdr.csum_flags |= csum_flags;
353 			if (csum_flags & CSUM_DATA_VALID)
354 				m->m_pkthdr.csum_data = 0xffff;
355 			(void) if_simloop(ifp, m, dst->sa_family, hlen);
356 			return (0);	/* XXX */
357 		}
358 	}
359 
360 	/* Handle ng_ether(4) processing, if any */
361 	if (ng_ether_output_p != NULL) {
362 		if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
363 bad:			if (m != NULL)
364 				m_freem(m);
365 			return (error);
366 		}
367 		if (m == NULL)
368 			return (0);
369 	}
370 
371 	/* Continue with link-layer output */
372 	return ether_output_frame(ifp, m);
373 }
374 
375 /*
376  * Ethernet link layer output routine to send a raw frame to the device.
377  *
378  * This assumes that the 14 byte Ethernet header is present and contiguous
379  * in the first mbuf (if BRIDGE'ing).
380  */
381 int
382 ether_output_frame(ifp, m)
383 	struct ifnet *ifp;
384 	struct mbuf *m;
385 {
386 	int error = 0;
387 	int s;
388 	struct ip_fw *rule = NULL;
389 
390 	/* Extract info from dummynet tag, ignore others */
391 	for (; m->m_type == MT_TAG; m = m->m_next)
392 		if (m->m_flags == PACKET_TAG_DUMMYNET)
393 			rule = ((struct dn_pkt *)m)->rule;
394 
395 	if (rule)	/* packet was already bridged */
396 		goto no_bridge;
397 
398 	if (BDG_ACTIVE(ifp) ) {
399 		struct ether_header *eh; /* a ptr suffices */
400 
401 		m->m_pkthdr.rcvif = NULL;
402 		eh = mtod(m, struct ether_header *);
403 		m_adj(m, ETHER_HDR_LEN);
404 		m = bdg_forward_ptr(m, eh, ifp);
405 		if (m != NULL)
406 			m_freem(m);
407 		return (0);
408 	}
409 
410 no_bridge:
411 	s = splimp();
412 	if (IPFW_LOADED && ether_ipfw != 0) {
413 		struct ether_header save_eh, *eh;
414 
415 		eh = mtod(m, struct ether_header *);
416 		save_eh = *eh;
417 		m_adj(m, ETHER_HDR_LEN);
418 		if (ether_ipfw_chk(&m, ifp, &rule, eh, 0) == 0) {
419 			if (m) {
420 				m_freem(m);
421 				return ENOBUFS;	/* pkt dropped */
422 			} else
423 				return 0;	/* consumed e.g. in a pipe */
424 		}
425 		/* packet was ok, restore the ethernet header */
426 		if ( (void *)(eh + 1) == (void *)m->m_data) {
427 			m->m_data -= ETHER_HDR_LEN ;
428 			m->m_len += ETHER_HDR_LEN ;
429 			m->m_pkthdr.len += ETHER_HDR_LEN ;
430 		} else {
431 			M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT);
432 			if (m == NULL) /* nope... */
433 				return ENOBUFS;
434 			bcopy(&save_eh, mtod(m, struct ether_header *),
435 			    ETHER_HDR_LEN);
436 		}
437 	}
438 
439 	/*
440 	 * Queue message on interface, update output statistics if
441 	 * successful, and start output if interface not yet active.
442 	 */
443 	if (!IF_HANDOFF(&ifp->if_snd, m, ifp))
444 		error = ENOBUFS;
445 	splx(s);
446 	return (error);
447 }
448 
449 /*
450  * ipfw processing for ethernet packets (in and out).
451  * The second parameter is NULL from ether_demux, and ifp from
452  * ether_output_frame. This section of code could be used from
453  * bridge.c as well as long as we use some extra info
454  * to distinguish that case from ether_output_frame();
455  */
456 int
457 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
458 	struct ip_fw **rule, struct ether_header *eh, int shared)
459 {
460 	struct ether_header save_eh = *eh;	/* might be a ptr in m */
461 	int i;
462 	struct ip_fw_args args;
463 
464 	if (*rule != NULL && fw_one_pass)
465 		return 1; /* dummynet packet, already partially processed */
466 
467 	/*
468 	 * I need some amt of data to be contiguous, and in case others need
469 	 * the packet (shared==1) also better be in the first mbuf.
470 	 */
471 	i = min( (*m0)->m_pkthdr.len, max_protohdr);
472 	if ( shared || (*m0)->m_len < i) {
473 		*m0 = m_pullup(*m0, i);
474 		if (*m0 == NULL)
475 			return 0;
476 	}
477 
478 	args.m = *m0;		/* the packet we are looking at		*/
479 	args.oif = dst;		/* destination, if any			*/
480 	args.divert_rule = 0;	/* we do not support divert yet		*/
481 	args.rule = *rule;	/* matching rule to restart		*/
482 	args.next_hop = NULL;	/* we do not support forward yet	*/
483 	args.eh = &save_eh;	/* MAC header for bridged/MAC packets	*/
484 	i = ip_fw_chk_ptr(&args);
485 	*m0 = args.m;
486 	*rule = args.rule;
487 
488 	if ( (i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL) /* drop */
489 		return 0;
490 
491 	if (i == 0) /* a PASS rule.  */
492 		return 1;
493 
494 	if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG)) {
495 		/*
496 		 * Pass the pkt to dummynet, which consumes it.
497 		 * If shared, make a copy and keep the original.
498 		 */
499 		struct mbuf *m ;
500 
501 		if (shared) {
502 			m = m_copypacket(*m0, MB_DONTWAIT);
503 			if (m == NULL)
504 				return 0;
505 		} else {
506 			m = *m0 ; /* pass the original to dummynet */
507 			*m0 = NULL ; /* and nothing back to the caller */
508 		}
509 		/*
510 		 * Prepend the header, optimize for the common case of
511 		 * eh pointing into the mbuf.
512 		 */
513 		if ( (void *)(eh + 1) == (void *)m->m_data) {
514 			m->m_data -= ETHER_HDR_LEN ;
515 			m->m_len += ETHER_HDR_LEN ;
516 			m->m_pkthdr.len += ETHER_HDR_LEN ;
517 		} else {
518 			M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT);
519 			if (m == NULL) /* nope... */
520 				return 0;
521 			bcopy(&save_eh, mtod(m, struct ether_header *),
522 			    ETHER_HDR_LEN);
523 		}
524 		ip_dn_io_ptr(m, (i & 0xffff),
525 			dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
526 		return 0;
527 	}
528 	/*
529 	 * XXX at some point add support for divert/forward actions.
530 	 * If none of the above matches, we have to drop the pkt.
531 	 */
532 	return 0;
533 }
534 
535 /*
536  * Process a received Ethernet packet. We have two different interfaces:
537  * one (conventional) assumes the packet in the mbuf, with the ethernet
538  * header provided separately in *eh. The second one (new) has everything
539  * in the mbuf, and we can tell it because eh == NULL.
540  * The caller MUST MAKE SURE that there are at least
541  * sizeof(struct ether_header) bytes in the first mbuf.
542  *
543  * This allows us to concentrate in one place a bunch of code which
544  * is replicated in all device drivers. Also, many functions called
545  * from ether_input() try to put the eh back into the mbuf, so we
546  * can later propagate the 'contiguous packet' interface to them,
547  * and handle the old interface just here.
548  *
549  * NOTA BENE: for many drivers "eh" is a pointer into the first mbuf or
550  * cluster, right before m_data. So be very careful when working on m,
551  * as you could destroy *eh !!
552  *
553  * First we perform any link layer operations, then continue
554  * to the upper layers with ether_demux().
555  */
556 void
557 ether_input(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m)
558 {
559 	struct ether_header save_eh;
560 
561 	if (eh == NULL) {
562 		if (m->m_len < sizeof(struct ether_header)) {
563 			/* XXX error in the caller. */
564 			m_freem(m);
565 			return;
566 		}
567 		m->m_pkthdr.rcvif = ifp;
568 		eh = mtod(m, struct ether_header *);
569 		m->m_data += sizeof(struct ether_header);
570 		m->m_len -= sizeof(struct ether_header);
571 		m->m_pkthdr.len = m->m_len;
572 	}
573 
574 	/* Check for a BPF tap */
575 	if (ifp->if_bpf != NULL) {
576 		struct m_hdr mh;
577 
578 		/* This kludge is OK; BPF treats the "mbuf" as read-only */
579 		mh.mh_next = m;
580 		mh.mh_data = (char *)eh;
581 		mh.mh_len = ETHER_HDR_LEN;
582 		bpf_mtap(ifp, (struct mbuf *)&mh);
583 	}
584 
585 	ifp->if_ibytes += m->m_pkthdr.len + sizeof (*eh);
586 
587 	/* Handle ng_ether(4) processing, if any */
588 	if (ng_ether_input_p != NULL) {
589 		(*ng_ether_input_p)(ifp, &m, eh);
590 		if (m == NULL)
591 			return;
592 	}
593 
594 	/* Check for bridging mode */
595 	if (BDG_ACTIVE(ifp) ) {
596 		struct ifnet *bif;
597 
598 		/* Check with bridging code */
599 		if ((bif = bridge_in_ptr(ifp, eh)) == BDG_DROP) {
600 			m_freem(m);
601 			return;
602 		}
603 		if (bif != BDG_LOCAL) {
604 			save_eh = *eh ; /* because it might change */
605 			m = bdg_forward_ptr(m, eh, bif); /* needs forwarding */
606 			/*
607 			 * Do not continue if bdg_forward_ptr() processed our
608 			 * packet (and cleared the mbuf pointer m) or if
609 			 * it dropped (m_free'd) the packet itself.
610 			 */
611 			if (m == NULL) {
612 			    if (bif == BDG_BCAST || bif == BDG_MCAST)
613 				printf("bdg_forward drop MULTICAST PKT\n");
614 			    return;
615 			}
616 			eh = &save_eh ;
617 		}
618 		if (bif == BDG_LOCAL
619 		    || bif == BDG_BCAST
620 		    || bif == BDG_MCAST)
621 			goto recvLocal;			/* receive locally */
622 
623 		/* If not local and not multicast, just drop it */
624 		if (m != NULL)
625 			m_freem(m);
626 		return;
627        }
628 
629 recvLocal:
630 	/* Continue with upper layer processing */
631 	ether_demux(ifp, eh, m);
632 }
633 
634 /*
635  * Upper layer processing for a received Ethernet packet.
636  */
637 void
638 ether_demux(ifp, eh, m)
639 	struct ifnet *ifp;
640 	struct ether_header *eh;
641 	struct mbuf *m;
642 {
643 	int isr;
644 	u_short ether_type;
645 #if defined(NETATALK)
646 	struct llc *l;
647 #endif
648 	struct ip_fw *rule = NULL;
649 
650 	/* Extract info from dummynet tag, ignore others */
651 	for (;m->m_type == MT_TAG; m = m->m_next)
652 		if (m->m_flags == PACKET_TAG_DUMMYNET) {
653 			rule = ((struct dn_pkt *)m)->rule;
654 			ifp = m->m_next->m_pkthdr.rcvif;
655 		}
656 
657 	if (rule)	/* packet was already bridged */
658 		goto post_stats;
659 
660     if (! (BDG_ACTIVE(ifp) ) )
661 	/* Discard packet if upper layers shouldn't see it because it was
662 	   unicast to a different Ethernet address. If the driver is working
663 	   properly, then this situation can only happen when the interface
664 	   is in promiscuous mode. */
665 	if ((ifp->if_flags & IFF_PROMISC) != 0
666 	    && (eh->ether_dhost[0] & 1) == 0
667 	    && bcmp(eh->ether_dhost,
668 	      IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN) != 0
669 	    && (ifp->if_flags & IFF_PPROMISC) == 0) {
670 		m_freem(m);
671 		return;
672 	}
673 
674 	/* Discard packet if interface is not up */
675 	if ((ifp->if_flags & IFF_UP) == 0) {
676 		m_freem(m);
677 		return;
678 	}
679 	if (eh->ether_dhost[0] & 1) {
680 		if (bcmp(ifp->if_broadcastaddr, eh->ether_dhost,
681 			 ifp->if_addrlen) == 0)
682 			m->m_flags |= M_BCAST;
683 		else
684 			m->m_flags |= M_MCAST;
685 	}
686 	if (m->m_flags & (M_BCAST|M_MCAST))
687 		ifp->if_imcasts++;
688 
689 post_stats:
690 	if (IPFW_LOADED && ether_ipfw != 0) {
691 		if (ether_ipfw_chk(&m, NULL, &rule, eh, 0 ) == 0) {
692 			if (m)
693 				m_freem(m);
694 			return;
695 		}
696 	}
697 
698 	ether_type = ntohs(eh->ether_type);
699 
700 	switch (ether_type) {
701 #ifdef INET
702 	case ETHERTYPE_IP:
703 		if (ipflow_fastforward(m))
704 			return;
705 		isr = NETISR_IP;
706 		break;
707 
708 	case ETHERTYPE_ARP:
709 		if (ifp->if_flags & IFF_NOARP) {
710 			/* Discard packet if ARP is disabled on interface */
711 			m_freem(m);
712 			return;
713 		}
714 		isr = NETISR_ARP;
715 		break;
716 #endif
717 #ifdef IPX
718 	case ETHERTYPE_IPX:
719 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
720 			return;
721 		isr = NETISR_IPX;
722 		break;
723 #endif
724 #ifdef INET6
725 	case ETHERTYPE_IPV6:
726 		isr = NETISR_IPV6;
727 		break;
728 #endif
729 #ifdef NS
730 	case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
731 		isr = NETISR_NS;
732 		break;
733 
734 #endif /* NS */
735 #ifdef NETATALK
736         case ETHERTYPE_AT:
737 		isr = NETISR_ATALK1;
738                 break;
739         case ETHERTYPE_AARP:
740 		isr = NETISR_AARP;
741                 break;
742 #endif /* NETATALK */
743 	case ETHERTYPE_VLAN:
744 		/* XXX lock ? */
745 		if (vlan_input_p != NULL)
746 			(*vlan_input_p)(eh, m);
747 		else {
748 			m->m_pkthdr.rcvif->if_noproto++;
749 			m_freem(m);
750 		}
751 		/* XXX unlock ? */
752 		return;
753 	default:
754 #ifdef IPX
755 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
756 			return;
757 #endif /* IPX */
758 #ifdef NS
759 		checksum = mtod(m, ushort *);
760 		/* Novell 802.3 */
761 		if ((ether_type <= ETHERMTU) &&
762 		    ((*checksum == 0xffff) || (*checksum == 0xE0E0))) {
763 			if (*checksum == 0xE0E0) {
764 				m->m_pkthdr.len -= 3;
765 				m->m_len -= 3;
766 				m->m_data += 3;
767 			}
768 			isr = NETISR_NS;
769 			break;
770 		}
771 #endif /* NS */
772 #ifdef NETATALK
773 		if (ether_type > ETHERMTU)
774 			goto dropanyway;
775 		l = mtod(m, struct llc *);
776 		if (l->llc_dsap == LLC_SNAP_LSAP &&
777 		    l->llc_ssap == LLC_SNAP_LSAP &&
778 		    l->llc_control == LLC_UI) {
779                         if (Bcmp(&(l->llc_snap_org_code)[0], at_org_code,
780 			    sizeof(at_org_code)) == 0 &&
781 			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
782 				m_adj(m, sizeof(struct llc));
783 				isr = NETISR_ATALK2;
784 				break;
785 			}
786 			if (Bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
787 			    sizeof(aarp_org_code)) == 0 &&
788 			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
789 				m_adj(m, sizeof(struct llc));
790 				isr = NETISR_AARP;
791 				break;
792 			}
793 		}
794 dropanyway:
795 #endif /* NETATALK */
796 		if (ng_ether_input_orphan_p != NULL)
797 			(*ng_ether_input_orphan_p)(ifp, m, eh);
798 		else
799 			m_freem(m);
800 		return;
801 	}
802 	netisr_dispatch(isr, m);
803 }
804 
805 /*
806  * Perform common duties while attaching to interface list
807  */
808 
809 void
810 ether_ifattach(struct ifnet *ifp, uint8_t *lla)
811 {
812 	ether_ifattach_bpf(ifp, lla, DLT_EN10MB, sizeof(struct ether_header));
813 }
814 
815 void
816 ether_ifattach_bpf(struct ifnet *ifp, uint8_t *lla, u_int dlt, u_int hdrlen)
817 {
818 	struct ifaddr *ifa;
819 	struct sockaddr_dl *sdl;
820 
821 	ifp->if_type = IFT_ETHER;
822 	ifp->if_addrlen = ETHER_ADDR_LEN;
823 	ifp->if_broadcastaddr = etherbroadcastaddr;
824 	ifp->if_hdrlen = 14;
825 	if_attach(ifp);
826 	ifp->if_mtu = ETHERMTU;
827 	ifp->if_resolvemulti = ether_resolvemulti;
828 	if (ifp->if_baudrate == 0)
829 	    ifp->if_baudrate = 10000000;
830 	ifa = ifnet_addrs[ifp->if_index - 1];
831 	KASSERT(ifa != NULL, ("%s: no lladdr!\n", __FUNCTION__));
832 	sdl = (struct sockaddr_dl *)ifa->ifa_addr;
833 	sdl->sdl_type = IFT_ETHER;
834 	sdl->sdl_alen = ifp->if_addrlen;
835 	bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
836 	/*
837 	 * XXX Keep the current drivers happy.
838 	 * XXX Remove once all drivers have been cleaned up
839 	 */
840 	if (lla != IFP2AC(ifp)->ac_enaddr)
841 		bcopy(lla, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen);
842 	bpfattach(ifp, dlt, hdrlen);
843 	if (ng_ether_attach_p != NULL)
844 		(*ng_ether_attach_p)(ifp);
845 	if (BDG_LOADED)
846 		bdgtakeifaces_ptr();
847 
848 	if_printf(ifp, "MAC address: %6D\n", lla, ":");
849 }
850 
851 /*
852  * Perform common duties while detaching an Ethernet interface
853  */
854 void
855 ether_ifdetach(struct ifnet *ifp)
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 SYSCTL_DECL(_net_link);
866 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
867 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
868 	    &ether_ipfw,0,"Pass ether pkts through firewall");
869 
870 int
871 ether_ioctl(ifp, command, data)
872 	struct ifnet *ifp;
873 	int command;
874 	caddr_t data;
875 {
876 	struct ifaddr *ifa = (struct ifaddr *) data;
877 	struct ifreq *ifr = (struct ifreq *) data;
878 	int error = 0;
879 
880 	switch (command) {
881 	case SIOCSIFADDR:
882 		ifp->if_flags |= IFF_UP;
883 
884 		switch (ifa->ifa_addr->sa_family) {
885 #ifdef INET
886 		case AF_INET:
887 			ifp->if_init(ifp->if_softc);	/* before arpwhohas */
888 			arp_ifinit(ifp, ifa);
889 			break;
890 #endif
891 #ifdef IPX
892 		/*
893 		 * XXX - This code is probably wrong
894 		 */
895 		case AF_IPX:
896 			{
897 			struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr);
898 			struct arpcom *ac = IFP2AC(ifp);
899 
900 			if (ipx_nullhost(*ina))
901 				ina->x_host =
902 				    *(union ipx_host *)
903 			            ac->ac_enaddr;
904 			else {
905 				bcopy((caddr_t) ina->x_host.c_host,
906 				      (caddr_t) ac->ac_enaddr,
907 				      sizeof(ac->ac_enaddr));
908 			}
909 
910 			/*
911 			 * Set new address
912 			 */
913 			ifp->if_init(ifp->if_softc);
914 			break;
915 			}
916 #endif
917 #ifdef NS
918 		/*
919 		 * XXX - This code is probably wrong
920 		 */
921 		case AF_NS:
922 		{
923 			struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr);
924 			struct arpcom *ac = IFP2AC(ifp);
925 
926 			if (ns_nullhost(*ina))
927 				ina->x_host =
928 				    *(union ns_host *) (ac->ac_enaddr);
929 			else {
930 				bcopy((caddr_t) ina->x_host.c_host,
931 				      (caddr_t) ac->ac_enaddr,
932 				      sizeof(ac->ac_enaddr));
933 			}
934 
935 			/*
936 			 * Set new address
937 			 */
938 			ifp->if_init(ifp->if_softc);
939 			break;
940 		}
941 #endif
942 		default:
943 			ifp->if_init(ifp->if_softc);
944 			break;
945 		}
946 		break;
947 
948 	case SIOCGIFADDR:
949 		{
950 			struct sockaddr *sa;
951 
952 			sa = (struct sockaddr *) & ifr->ifr_data;
953 			bcopy(IFP2AC(ifp)->ac_enaddr,
954 			      (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
955 		}
956 		break;
957 
958 	case SIOCSIFMTU:
959 		/*
960 		 * Set the interface MTU.
961 		 */
962 		if (ifr->ifr_mtu > ETHERMTU) {
963 			error = EINVAL;
964 		} else {
965 			ifp->if_mtu = ifr->ifr_mtu;
966 		}
967 		break;
968 	default:
969 		error = EINVAL;
970 		break;
971 	}
972 	return (error);
973 }
974 
975 int
976 ether_resolvemulti(ifp, llsa, sa)
977 	struct ifnet *ifp;
978 	struct sockaddr **llsa;
979 	struct sockaddr *sa;
980 {
981 	struct sockaddr_dl *sdl;
982 	struct sockaddr_in *sin;
983 #ifdef INET6
984 	struct sockaddr_in6 *sin6;
985 #endif
986 	u_char *e_addr;
987 
988 	switch(sa->sa_family) {
989 	case AF_LINK:
990 		/*
991 		 * No mapping needed. Just check that it's a valid MC address.
992 		 */
993 		sdl = (struct sockaddr_dl *)sa;
994 		e_addr = LLADDR(sdl);
995 		if ((e_addr[0] & 1) != 1)
996 			return EADDRNOTAVAIL;
997 		*llsa = 0;
998 		return 0;
999 
1000 #ifdef INET
1001 	case AF_INET:
1002 		sin = (struct sockaddr_in *)sa;
1003 		if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1004 			return EADDRNOTAVAIL;
1005 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1006 		       M_WAITOK|M_ZERO);
1007 		sdl->sdl_len = sizeof *sdl;
1008 		sdl->sdl_family = AF_LINK;
1009 		sdl->sdl_index = ifp->if_index;
1010 		sdl->sdl_type = IFT_ETHER;
1011 		sdl->sdl_alen = ETHER_ADDR_LEN;
1012 		e_addr = LLADDR(sdl);
1013 		ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1014 		*llsa = (struct sockaddr *)sdl;
1015 		return 0;
1016 #endif
1017 #ifdef INET6
1018 	case AF_INET6:
1019 		sin6 = (struct sockaddr_in6 *)sa;
1020 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1021 			/*
1022 			 * An IP6 address of 0 means listen to all
1023 			 * of the Ethernet multicast address used for IP6.
1024 			 * (This is used for multicast routers.)
1025 			 */
1026 			ifp->if_flags |= IFF_ALLMULTI;
1027 			*llsa = 0;
1028 			return 0;
1029 		}
1030 		if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1031 			return EADDRNOTAVAIL;
1032 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1033 		       M_WAITOK|M_ZERO);
1034 		sdl->sdl_len = sizeof *sdl;
1035 		sdl->sdl_family = AF_LINK;
1036 		sdl->sdl_index = ifp->if_index;
1037 		sdl->sdl_type = IFT_ETHER;
1038 		sdl->sdl_alen = ETHER_ADDR_LEN;
1039 		e_addr = LLADDR(sdl);
1040 		ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1041 		*llsa = (struct sockaddr *)sdl;
1042 		return 0;
1043 #endif
1044 
1045 	default:
1046 		/*
1047 		 * Well, the text isn't quite right, but it's the name
1048 		 * that counts...
1049 		 */
1050 		return EAFNOSUPPORT;
1051 	}
1052 }
1053 
1054 #if 0
1055 /*
1056  * This is for reference.  We have a table-driven version
1057  * of the little-endian crc32 generator, which is faster
1058  * than the double-loop.
1059  */
1060 uint32_t
1061 ether_crc32_le(const uint8_t *buf, size_t len)
1062 {
1063 	uint32_t c, crc, carry;
1064 	size_t i, j;
1065 
1066 	crc = 0xffffffffU;	/* initial value */
1067 
1068 	for (i = 0; i < len; i++) {
1069 		c = buf[i];
1070 		for (j = 0; j < 8; j++) {
1071 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
1072 			crc >>= 1;
1073 			c >>= 1;
1074 			if (carry)
1075 				crc = (crc ^ ETHER_CRC_POLY_LE);
1076 		}
1077 	}
1078 
1079 	return(crc);
1080 }
1081 #else
1082 uint32_t
1083 ether_crc32_le(const uint8_t *buf, size_t len)
1084 {
1085 	static const uint32_t crctab[] = {
1086 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1087 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1088 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1089 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1090 	};
1091 	uint32_t crc;
1092 	size_t i;
1093 
1094 	crc = 0xffffffffU;	/* initial value */
1095 
1096 	for (i = 0; i < len; i++) {
1097 		crc ^= buf[i];
1098 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1099 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1100 	}
1101 
1102 	return(crc);
1103 }
1104 #endif
1105 
1106 uint32_t
1107 ether_crc32_be(const uint8_t *buf, size_t len)
1108 {
1109 	uint32_t c, crc, carry;
1110 	size_t i, j;
1111 
1112 	crc = 0xffffffffU;	/* initial value */
1113 
1114 	for (i = 0; i < len; i++) {
1115 		c = buf[i];
1116 		for (j = 0; j < 8; j++) {
1117 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
1118 			crc <<= 1;
1119 			c >>= 1;
1120 			if (carry)
1121 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1122 		}
1123 	}
1124 
1125 	return(crc);
1126 }
1127