xref: /freebsd/sys/net/if_ethersubr.c (revision 7afc53b8dfcc7d5897920ce6cc7e842fbb4ab813)
1 /*-
2  * Copyright (c) 1982, 1989, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 4. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	@(#)if_ethersubr.c	8.1 (Berkeley) 6/10/93
30  * $FreeBSD$
31  */
32 
33 #include "opt_atalk.h"
34 #include "opt_inet.h"
35 #include "opt_inet6.h"
36 #include "opt_ipx.h"
37 #include "opt_bdg.h"
38 #include "opt_mac.h"
39 #include "opt_netgraph.h"
40 #include "opt_carp.h"
41 
42 #include <sys/param.h>
43 #include <sys/systm.h>
44 #include <sys/kernel.h>
45 #include <sys/mac.h>
46 #include <sys/malloc.h>
47 #include <sys/module.h>
48 #include <sys/mbuf.h>
49 #include <sys/random.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/if_arp.h>
56 #include <net/netisr.h>
57 #include <net/route.h>
58 #include <net/if_llc.h>
59 #include <net/if_dl.h>
60 #include <net/if_types.h>
61 #include <net/bpf.h>
62 #include <net/ethernet.h>
63 #include <net/bridge.h>
64 #include <net/if_vlan_var.h>
65 
66 #if defined(INET) || defined(INET6)
67 #include <netinet/in.h>
68 #include <netinet/in_var.h>
69 #include <netinet/if_ether.h>
70 #include <netinet/ip_fw.h>
71 #include <netinet/ip_dummynet.h>
72 #endif
73 #ifdef INET6
74 #include <netinet6/nd6.h>
75 #endif
76 
77 #ifdef DEV_CARP
78 #include <netinet/ip_carp.h>
79 #endif
80 
81 #ifdef IPX
82 #include <netipx/ipx.h>
83 #include <netipx/ipx_if.h>
84 #endif
85 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m);
86 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp,
87 		struct sockaddr *dst, short *tp, int *hlen);
88 
89 #ifdef NETATALK
90 #include <netatalk/at.h>
91 #include <netatalk/at_var.h>
92 #include <netatalk/at_extern.h>
93 
94 #define llc_snap_org_code llc_un.type_snap.org_code
95 #define llc_snap_ether_type llc_un.type_snap.ether_type
96 
97 extern u_char	at_org_code[3];
98 extern u_char	aarp_org_code[3];
99 #endif /* NETATALK */
100 
101 /* netgraph node hooks for ng_ether(4) */
102 void	(*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp);
103 void	(*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m);
104 int	(*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
105 void	(*ng_ether_attach_p)(struct ifnet *ifp);
106 void	(*ng_ether_detach_p)(struct ifnet *ifp);
107 
108 void	(*vlan_input_p)(struct ifnet *, struct mbuf *);
109 
110 /* bridge support */
111 int do_bridge;
112 bridge_in_t *bridge_in_ptr;
113 bdg_forward_t *bdg_forward_ptr;
114 bdgtakeifaces_t *bdgtakeifaces_ptr;
115 struct bdg_softc *ifp2sc;
116 
117 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *);
118 int     (*bridge_output_p)(struct ifnet *, struct mbuf *,
119 		struct sockaddr *, struct rtentry *);
120 
121 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] =
122 			{ 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
123 
124 static	int ether_resolvemulti(struct ifnet *, struct sockaddr **,
125 		struct sockaddr *);
126 
127 #define senderr(e) do { error = (e); goto bad;} while (0)
128 
129 #if defined(INET) || defined(INET6)
130 int
131 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
132 	struct ip_fw **rule, int shared);
133 static int ether_ipfw;
134 #endif
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(struct ifnet *ifp, struct mbuf *m,
145 	struct sockaddr *dst, struct rtentry *rt0)
146 {
147 	short type;
148 	int error, hdrcmplt = 0;
149 	u_char esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN];
150 	struct ether_header *eh;
151 	int loop_copy = 0;
152 	int hlen;	/* link layer header length */
153 
154 #ifdef MAC
155 	error = mac_check_ifnet_transmit(ifp, m);
156 	if (error)
157 		senderr(error);
158 #endif
159 
160 	if (ifp->if_flags & IFF_MONITOR)
161 		senderr(ENETDOWN);
162 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
163 		senderr(ENETDOWN);
164 
165 	hlen = ETHER_HDR_LEN;
166 	switch (dst->sa_family) {
167 #ifdef INET
168 	case AF_INET:
169 		error = arpresolve(ifp, rt0, m, dst, edst);
170 		if (error)
171 			return (error == EWOULDBLOCK ? 0 : error);
172 		type = htons(ETHERTYPE_IP);
173 		break;
174 	case AF_ARP:
175 	{
176 		struct arphdr *ah;
177 		ah = mtod(m, struct arphdr *);
178 		ah->ar_hrd = htons(ARPHRD_ETHER);
179 
180 		loop_copy = -1; /* if this is for us, don't do it */
181 
182 		switch(ntohs(ah->ar_op)) {
183 		case ARPOP_REVREQUEST:
184 		case ARPOP_REVREPLY:
185 			type = htons(ETHERTYPE_REVARP);
186 			break;
187 		case ARPOP_REQUEST:
188 		case ARPOP_REPLY:
189 		default:
190 			type = htons(ETHERTYPE_ARP);
191 			break;
192 		}
193 
194 		if (m->m_flags & M_BCAST)
195 			bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN);
196 		else
197 			bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN);
198 
199 	}
200 	break;
201 #endif
202 #ifdef INET6
203 	case AF_INET6:
204 		error = nd6_storelladdr(ifp, rt0, m, dst, (u_char *)edst);
205 		if (error)
206 			return error;
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 		    senderr(EHOSTUNREACH); /* XXX */
229 	    if (!aarpresolve(ifp, m, (struct sockaddr_at *)dst, edst))
230 		    return (0);
231 	    /*
232 	     * In the phase 2 case, need to prepend an mbuf for the llc header.
233 	     */
234 	    if ( aa->aa_flags & AFA_PHASE2 ) {
235 		struct llc llc;
236 
237 		M_PREPEND(m, LLC_SNAPFRAMELEN, M_DONTWAIT);
238 		if (m == NULL)
239 			senderr(ENOBUFS);
240 		llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
241 		llc.llc_control = LLC_UI;
242 		bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code));
243 		llc.llc_snap_ether_type = htons( ETHERTYPE_AT );
244 		bcopy(&llc, mtod(m, caddr_t), LLC_SNAPFRAMELEN);
245 		type = htons(m->m_pkthdr.len);
246 		hlen = LLC_SNAPFRAMELEN + ETHER_HDR_LEN;
247 	    } else {
248 		type = htons(ETHERTYPE_AT);
249 	    }
250 	    break;
251 	  }
252 #endif /* NETATALK */
253 
254 	case pseudo_AF_HDRCMPLT:
255 		hdrcmplt = 1;
256 		eh = (struct ether_header *)dst->sa_data;
257 		(void)memcpy(esrc, eh->ether_shost, sizeof (esrc));
258 		/* FALLTHROUGH */
259 
260 	case AF_UNSPEC:
261 		loop_copy = -1; /* if this is for us, don't do it */
262 		eh = (struct ether_header *)dst->sa_data;
263 		(void)memcpy(edst, eh->ether_dhost, sizeof (edst));
264 		type = eh->ether_type;
265 		break;
266 
267 	default:
268 		if_printf(ifp, "can't handle af%d\n", dst->sa_family);
269 		senderr(EAFNOSUPPORT);
270 	}
271 
272 	/*
273 	 * Add local net header.  If no space in first mbuf,
274 	 * allocate another.
275 	 */
276 	M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
277 	if (m == NULL)
278 		senderr(ENOBUFS);
279 	eh = mtod(m, struct ether_header *);
280 	(void)memcpy(&eh->ether_type, &type,
281 		sizeof(eh->ether_type));
282 	(void)memcpy(eh->ether_dhost, edst, sizeof (edst));
283 	if (hdrcmplt)
284 		(void)memcpy(eh->ether_shost, esrc,
285 			sizeof(eh->ether_shost));
286 	else
287 		(void)memcpy(eh->ether_shost, IFP2AC(ifp)->ac_enaddr,
288 			sizeof(eh->ether_shost));
289 
290        /*
291 	* Bridges require special output handling.
292 	*/
293 	if (ifp->if_bridge) {
294 		KASSERT(bridge_output_p != NULL,
295 		    ("ether_input: if_bridge not loaded!"));
296 		return ((*bridge_output_p)(ifp, m, NULL, NULL));
297 	}
298 
299 	/*
300 	 * If a simplex interface, and the packet is being sent to our
301 	 * Ethernet address or a broadcast address, loopback a copy.
302 	 * XXX To make a simplex device behave exactly like a duplex
303 	 * device, we should copy in the case of sending to our own
304 	 * ethernet address (thus letting the original actually appear
305 	 * on the wire). However, we don't do that here for security
306 	 * reasons and compatibility with the original behavior.
307 	 */
308 	if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) {
309 		int csum_flags = 0;
310 
311 		if (m->m_pkthdr.csum_flags & CSUM_IP)
312 			csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
313 		if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
314 			csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
315 
316 		if ((m->m_flags & M_BCAST) || (loop_copy > 0)) {
317 			struct mbuf *n;
318 
319 			if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
320 				n->m_pkthdr.csum_flags |= csum_flags;
321 				if (csum_flags & CSUM_DATA_VALID)
322 					n->m_pkthdr.csum_data = 0xffff;
323 				(void)if_simloop(ifp, n, dst->sa_family, hlen);
324 			} else
325 				ifp->if_iqdrops++;
326 		} else if (bcmp(eh->ether_dhost, eh->ether_shost,
327 				ETHER_ADDR_LEN) == 0) {
328 			m->m_pkthdr.csum_flags |= csum_flags;
329 			if (csum_flags & CSUM_DATA_VALID)
330 				m->m_pkthdr.csum_data = 0xffff;
331 			(void) if_simloop(ifp, m, dst->sa_family, hlen);
332 			return (0);	/* XXX */
333 		}
334 	}
335 
336 #ifdef DEV_CARP
337 	if (ifp->if_carp &&
338 	    (error = carp_output(ifp, m, dst, NULL)))
339 		goto bad;
340 #endif
341 
342 	/* Handle ng_ether(4) processing, if any */
343 	if (IFP2AC(ifp)->ac_netgraph != NULL) {
344 		if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
345 bad:			if (m != NULL)
346 				m_freem(m);
347 			return (error);
348 		}
349 		if (m == NULL)
350 			return (0);
351 	}
352 
353 	/* Continue with link-layer output */
354 	return ether_output_frame(ifp, m);
355 }
356 
357 /*
358  * Ethernet link layer output routine to send a raw frame to the device.
359  *
360  * This assumes that the 14 byte Ethernet header is present and contiguous
361  * in the first mbuf (if BRIDGE'ing).
362  */
363 int
364 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
365 {
366 #if defined(INET) || defined(INET6)
367 	struct ip_fw *rule = ip_dn_claim_rule(m);
368 #else
369 	void *rule = NULL;
370 #endif
371 	int error;
372 
373 	if (rule == NULL && BDG_ACTIVE(ifp)) {
374 		/*
375 		 * Beware, the bridge code notices the null rcvif and
376 		 * uses that identify that it's being called from
377 		 * ether_output as opposd to ether_input.  Yech.
378 		 */
379 		m->m_pkthdr.rcvif = NULL;
380 		m = bdg_forward_ptr(m, ifp);
381 		if (m != NULL)
382 			m_freem(m);
383 		return (0);
384 	}
385 #if defined(INET) || defined(INET6)
386 	if (IPFW_LOADED && ether_ipfw != 0) {
387 		if (ether_ipfw_chk(&m, ifp, &rule, 0) == 0) {
388 			if (m) {
389 				m_freem(m);
390 				return EACCES;	/* pkt dropped */
391 			} else
392 				return 0;	/* consumed e.g. in a pipe */
393 		}
394 	}
395 #endif
396 
397 	/*
398 	 * Queue message on interface, update output statistics if
399 	 * successful, and start output if interface not yet active.
400 	 */
401 	IFQ_HANDOFF(ifp, m, error);
402 	return (error);
403 }
404 
405 #if defined(INET) || defined(INET6)
406 /*
407  * ipfw processing for ethernet packets (in and out).
408  * The second parameter is NULL from ether_demux, and ifp from
409  * ether_output_frame. This section of code could be used from
410  * bridge.c as well as long as we use some extra info
411  * to distinguish that case from ether_output_frame();
412  */
413 int
414 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
415 	struct ip_fw **rule, int shared)
416 {
417 	struct ether_header *eh;
418 	struct ether_header save_eh;
419 	struct mbuf *m;
420 	int i;
421 	struct ip_fw_args args;
422 
423 	if (*rule != NULL && fw_one_pass)
424 		return 1; /* dummynet packet, already partially processed */
425 
426 	/*
427 	 * I need some amt of data to be contiguous, and in case others need
428 	 * the packet (shared==1) also better be in the first mbuf.
429 	 */
430 	m = *m0;
431 	i = min( m->m_pkthdr.len, max_protohdr);
432 	if ( shared || m->m_len < i) {
433 		m = m_pullup(m, i);
434 		if (m == NULL) {
435 			*m0 = m;
436 			return 0;
437 		}
438 	}
439 	eh = mtod(m, struct ether_header *);
440 	save_eh = *eh;			/* save copy for restore below */
441 	m_adj(m, ETHER_HDR_LEN);	/* strip ethernet header */
442 
443 	args.m = m;		/* the packet we are looking at		*/
444 	args.oif = dst;		/* destination, if any			*/
445 	args.rule = *rule;	/* matching rule to restart		*/
446 	args.next_hop = NULL;	/* we do not support forward yet	*/
447 	args.eh = &save_eh;	/* MAC header for bridged/MAC packets	*/
448 	i = ip_fw_chk_ptr(&args);
449 	m = args.m;
450 	if (m != NULL) {
451 		/*
452 		 * Restore Ethernet header, as needed, in case the
453 		 * mbuf chain was replaced by ipfw.
454 		 */
455 		M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
456 		if (m == NULL) {
457 			*m0 = m;
458 			return 0;
459 		}
460 		if (eh != mtod(m, struct ether_header *))
461 			bcopy(&save_eh, mtod(m, struct ether_header *),
462 				ETHER_HDR_LEN);
463 	}
464 	*m0 = m;
465 	*rule = args.rule;
466 
467 	if (i == IP_FW_DENY) /* drop */
468 		return 0;
469 
470 	KASSERT(m != NULL, ("ether_ipfw_chk: m is NULL"));
471 
472 	if (i == IP_FW_PASS) /* a PASS rule.  */
473 		return 1;
474 
475 	if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) {
476 		/*
477 		 * Pass the pkt to dummynet, which consumes it.
478 		 * If shared, make a copy and keep the original.
479 		 */
480 		if (shared) {
481 			m = m_copypacket(m, M_DONTWAIT);
482 			if (m == NULL)
483 				return 0;
484 		} else {
485 			/*
486 			 * Pass the original to dummynet and
487 			 * nothing back to the caller
488 			 */
489 			*m0 = NULL ;
490 		}
491 		ip_dn_io_ptr(m, dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
492 		return 0;
493 	}
494 	/*
495 	 * XXX at some point add support for divert/forward actions.
496 	 * If none of the above matches, we have to drop the pkt.
497 	 */
498 	return 0;
499 }
500 #endif
501 
502 /*
503  * Process a received Ethernet packet; the packet is in the
504  * mbuf chain m with the ethernet header at the front.
505  */
506 static void
507 ether_input(struct ifnet *ifp, struct mbuf *m)
508 {
509 	struct ether_header *eh;
510 	u_short etype;
511 
512 	/*
513 	 * Do consistency checks to verify assumptions
514 	 * made by code past this point.
515 	 */
516 	if ((m->m_flags & M_PKTHDR) == 0) {
517 		if_printf(ifp, "discard frame w/o packet header\n");
518 		ifp->if_ierrors++;
519 		m_freem(m);
520 		return;
521 	}
522 	if (m->m_len < ETHER_HDR_LEN) {
523 		/* XXX maybe should pullup? */
524 		if_printf(ifp, "discard frame w/o leading ethernet "
525 				"header (len %u pkt len %u)\n",
526 				m->m_len, m->m_pkthdr.len);
527 		ifp->if_ierrors++;
528 		m_freem(m);
529 		return;
530 	}
531 	eh = mtod(m, struct ether_header *);
532 	etype = ntohs(eh->ether_type);
533 	if (m->m_pkthdr.len >
534 	    ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
535 		if_printf(ifp, "discard oversize frame "
536 				"(ether type %x flags %x len %u > max %lu)\n",
537 				etype, m->m_flags, m->m_pkthdr.len,
538 				ETHER_MAX_FRAME(ifp, etype,
539 						m->m_flags & M_HASFCS));
540 		ifp->if_ierrors++;
541 		m_freem(m);
542 		return;
543 	}
544 	if (m->m_pkthdr.rcvif == NULL) {
545 		if_printf(ifp, "discard frame w/o interface pointer\n");
546 		ifp->if_ierrors++;
547 		m_freem(m);
548 		return;
549 	}
550 #ifdef DIAGNOSTIC
551 	if (m->m_pkthdr.rcvif != ifp) {
552 		if_printf(ifp, "Warning, frame marked as received on %s\n",
553 			m->m_pkthdr.rcvif->if_xname);
554 	}
555 #endif
556 
557 #ifdef MAC
558 	/*
559 	 * Tag the mbuf with an appropriate MAC label before any other
560 	 * consumers can get to it.
561 	 */
562 	mac_create_mbuf_from_ifnet(ifp, m);
563 #endif
564 
565 	/*
566 	 * Give bpf a chance at the packet.
567 	 */
568 	BPF_MTAP(ifp, m);
569 
570 	if (ifp->if_flags & IFF_MONITOR) {
571 		/*
572 		 * Interface marked for monitoring; discard packet.
573 		 */
574 		m_freem(m);
575 		return;
576 	}
577 
578 	/* If the CRC is still on the packet, trim it off. */
579 	if (m->m_flags & M_HASFCS) {
580 		m_adj(m, -ETHER_CRC_LEN);
581 		m->m_flags &= ~M_HASFCS;
582 	}
583 
584 	ifp->if_ibytes += m->m_pkthdr.len;
585 
586 	/* Handle ng_ether(4) processing, if any */
587 	if (IFP2AC(ifp)->ac_netgraph != NULL) {
588 		(*ng_ether_input_p)(ifp, &m);
589 		if (m == NULL)
590 			return;
591 	}
592 
593 	/*
594 	* Tap the packet off here for a bridge.  bridge_input()
595 	* will return NULL if it has consumed the packet, otherwise
596 	* it gets processed as normal.  Note that bridge_input()
597 	* will always return the original packet if we need to
598 	* process it locally.
599 	*/
600 	if (ifp->if_bridge) {
601 		KASSERT(bridge_input_p != NULL,
602 		    ("ether_input: if_bridge not loaded!"));
603 
604 		/* Mark the packet as broadcast or multicast. This is also set
605 		 * further down the code in ether_demux() but since the bridge
606 		 * input routine rarely returns a mbuf for further processing,
607 		 * it is an acceptable duplication.
608 		 */
609 		if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
610 			if (bcmp(etherbroadcastaddr, eh->ether_dhost,
611 				sizeof(etherbroadcastaddr)) == 0)
612 				m->m_flags |= M_BCAST;
613 			else
614 				m->m_flags |= M_MCAST;
615 		}
616 
617 		m = (*bridge_input_p)(ifp, m);
618 		if (m == NULL)
619 			return;
620 		/*
621 		* Bridge has determined that the packet is for us.
622 		* Update our interface pointer -- we may have had
623 		* to "bridge" the packet locally.
624 		*/
625 		ifp = m->m_pkthdr.rcvif;
626 	}
627 
628 	/* Check for bridging mode */
629 	if (BDG_ACTIVE(ifp) )
630 		if ((m = bridge_in_ptr(ifp, m)) == NULL)
631 			return;
632 
633 	/* First chunk of an mbuf contains good entropy */
634 	if (harvest.ethernet)
635 		random_harvest(m, 16, 3, 0, RANDOM_NET);
636 	ether_demux(ifp, m);
637 }
638 
639 /*
640  * Upper layer processing for a received Ethernet packet.
641  */
642 void
643 ether_demux(struct ifnet *ifp, struct mbuf *m)
644 {
645 	struct ether_header *eh;
646 	int isr;
647 	u_short ether_type;
648 #if defined(NETATALK)
649 	struct llc *l;
650 #endif
651 #if defined(INET) || defined(INET6)
652 	struct ip_fw *rule = ip_dn_claim_rule(m);
653 #endif
654 
655 	KASSERT(ifp != NULL, ("ether_demux: NULL interface pointer"));
656 
657 	eh = mtod(m, struct ether_header *);
658 	ether_type = ntohs(eh->ether_type);
659 
660 #if defined(INET) || defined(INET6)
661 	if (rule)	/* packet was already bridged */
662 		goto post_stats;
663 #endif
664 
665 	if (!(BDG_ACTIVE(ifp)) &&
666 	    !((ether_type == ETHERTYPE_VLAN || m->m_flags & M_VLANTAG) &&
667 	    ifp->if_nvlans > 0)) {
668 #ifdef DEV_CARP
669 		/*
670 		 * XXX: Okay, we need to call carp_forus() and - if it is for
671 		 * us jump over code that does the normal check
672 		 * "ac_enaddr == ether_dhost". The check sequence is a bit
673 		 * different from OpenBSD, so we jump over as few code as
674 		 * possible, to catch _all_ sanity checks. This needs
675 		 * evaluation, to see if the carp ether_dhost values break any
676 		 * of these checks!
677 		 */
678 		if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost))
679 			goto pre_stats;
680 #endif
681 		/*
682 		 * Discard packet if upper layers shouldn't see it because it
683 		 * was unicast to a different Ethernet address. If the driver
684 		 * is working properly, then this situation can only happen
685 		 * when the interface is in promiscuous mode.
686 		 *
687 		 * If VLANs are active, and this packet has a VLAN tag, do
688 		 * not drop it here but pass it on to the VLAN layer, to
689 		 * give them a chance to consider it as well (e. g. in case
690 		 * bridging is only active on a VLAN).  They will drop it if
691 		 * it's undesired.
692 		 */
693 		if ((ifp->if_flags & IFF_PROMISC) != 0
694 		    && !ETHER_IS_MULTICAST(eh->ether_dhost)
695 		    && bcmp(eh->ether_dhost,
696 		      IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN) != 0
697 		    && (ifp->if_flags & IFF_PPROMISC) == 0) {
698 			    m_freem(m);
699 			    return;
700 		}
701 	}
702 
703 #ifdef DEV_CARP
704 pre_stats:
705 #endif
706 	/* Discard packet if interface is not up */
707 	if ((ifp->if_flags & IFF_UP) == 0) {
708 		m_freem(m);
709 		return;
710 	}
711 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
712 		if (bcmp(etherbroadcastaddr, eh->ether_dhost,
713 		    sizeof(etherbroadcastaddr)) == 0)
714 			m->m_flags |= M_BCAST;
715 		else
716 			m->m_flags |= M_MCAST;
717 	}
718 	if (m->m_flags & (M_BCAST|M_MCAST))
719 		ifp->if_imcasts++;
720 
721 #if defined(INET) || defined(INET6)
722 post_stats:
723 	if (IPFW_LOADED && ether_ipfw != 0) {
724 		if (ether_ipfw_chk(&m, NULL, &rule, 0) == 0) {
725 			if (m)
726 				m_freem(m);
727 			return;
728 		}
729 	}
730 #endif
731 
732 	/*
733 	 * Check to see if the device performed the VLAN decapsulation and
734 	 * provided us with the tag.
735 	 */
736 	if (m->m_flags & M_VLANTAG) {
737 		/*
738 		 * If no VLANs are configured, drop.
739 		 */
740 		if (ifp->if_nvlans == 0) {
741 			ifp->if_noproto++;
742 			m_freem(m);
743 			return;
744 		}
745 		/*
746 		 * vlan_input() will either recursively call ether_input()
747 		 * or drop the packet.
748 		 */
749 		KASSERT(vlan_input_p != NULL,("ether_input: VLAN not loaded!"));
750 		(*vlan_input_p)(ifp, m);
751 		return;
752 	}
753 
754 	/*
755 	 * Handle protocols that expect to have the Ethernet header
756 	 * (and possibly FCS) intact.
757 	 */
758 	switch (ether_type) {
759 	case ETHERTYPE_VLAN:
760 		if (ifp->if_nvlans != 0) {
761 			KASSERT(vlan_input_p,("ether_input: VLAN not loaded!"));
762 			(*vlan_input_p)(ifp, m);
763 		} else {
764 			ifp->if_noproto++;
765 			m_freem(m);
766 		}
767 		return;
768 	}
769 
770 	/* Strip off Ethernet header. */
771 	m_adj(m, ETHER_HDR_LEN);
772 
773 	/* If the CRC is still on the packet, trim it off. */
774 	if (m->m_flags & M_HASFCS) {
775 		m_adj(m, -ETHER_CRC_LEN);
776 		m->m_flags &= ~M_HASFCS;
777 	}
778 
779 	switch (ether_type) {
780 #ifdef INET
781 	case ETHERTYPE_IP:
782 		if (ip_fastforward(m))
783 			return;
784 		isr = NETISR_IP;
785 		break;
786 
787 	case ETHERTYPE_ARP:
788 		if (ifp->if_flags & IFF_NOARP) {
789 			/* Discard packet if ARP is disabled on interface */
790 			m_freem(m);
791 			return;
792 		}
793 		isr = NETISR_ARP;
794 		break;
795 #endif
796 #ifdef IPX
797 	case ETHERTYPE_IPX:
798 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
799 			return;
800 		isr = NETISR_IPX;
801 		break;
802 #endif
803 #ifdef INET6
804 	case ETHERTYPE_IPV6:
805 		isr = NETISR_IPV6;
806 		break;
807 #endif
808 #ifdef NETATALK
809 	case ETHERTYPE_AT:
810 		isr = NETISR_ATALK1;
811 		break;
812 	case ETHERTYPE_AARP:
813 		isr = NETISR_AARP;
814 		break;
815 #endif /* NETATALK */
816 	default:
817 #ifdef IPX
818 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
819 			return;
820 #endif /* IPX */
821 #if defined(NETATALK)
822 		if (ether_type > ETHERMTU)
823 			goto discard;
824 		l = mtod(m, struct llc *);
825 		if (l->llc_dsap == LLC_SNAP_LSAP &&
826 		    l->llc_ssap == LLC_SNAP_LSAP &&
827 		    l->llc_control == LLC_UI) {
828 			if (bcmp(&(l->llc_snap_org_code)[0], at_org_code,
829 			    sizeof(at_org_code)) == 0 &&
830 			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
831 				m_adj(m, LLC_SNAPFRAMELEN);
832 				isr = NETISR_ATALK2;
833 				break;
834 			}
835 			if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
836 			    sizeof(aarp_org_code)) == 0 &&
837 			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
838 				m_adj(m, LLC_SNAPFRAMELEN);
839 				isr = NETISR_AARP;
840 				break;
841 			}
842 		}
843 #endif /* NETATALK */
844 		goto discard;
845 	}
846 	netisr_dispatch(isr, m);
847 	return;
848 
849 discard:
850 	/*
851 	 * Packet is to be discarded.  If netgraph is present,
852 	 * hand the packet to it for last chance processing;
853 	 * otherwise dispose of it.
854 	 */
855 	if (IFP2AC(ifp)->ac_netgraph != NULL) {
856 		/*
857 		 * Put back the ethernet header so netgraph has a
858 		 * consistent view of inbound packets.
859 		 */
860 		M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
861 		(*ng_ether_input_orphan_p)(ifp, m);
862 		return;
863 	}
864 	m_freem(m);
865 }
866 
867 /*
868  * Convert Ethernet address to printable (loggable) representation.
869  * This routine is for compatibility; it's better to just use
870  *
871  *	printf("%6D", <pointer to address>, ":");
872  *
873  * since there's no static buffer involved.
874  */
875 char *
876 ether_sprintf(const u_char *ap)
877 {
878 	static char etherbuf[18];
879 	snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
880 	return (etherbuf);
881 }
882 
883 /*
884  * Perform common duties while attaching to interface list
885  */
886 void
887 ether_ifattach(struct ifnet *ifp, const u_int8_t *llc)
888 {
889 	int i;
890 	struct ifaddr *ifa;
891 	struct sockaddr_dl *sdl;
892 
893 	ifp->if_type = IFT_ETHER;
894 	ifp->if_addrlen = ETHER_ADDR_LEN;
895 	ifp->if_hdrlen = ETHER_HDR_LEN;
896 	if_attach(ifp);
897 	ifp->if_mtu = ETHERMTU;
898 	ifp->if_output = ether_output;
899 	ifp->if_input = ether_input;
900 	ifp->if_resolvemulti = ether_resolvemulti;
901 	if (ifp->if_baudrate == 0)
902 		ifp->if_baudrate = IF_Mbps(10);		/* just a default */
903 	ifp->if_broadcastaddr = etherbroadcastaddr;
904 
905 	ifa = ifaddr_byindex(ifp->if_index);
906 	KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
907 	sdl = (struct sockaddr_dl *)ifa->ifa_addr;
908 	sdl->sdl_type = IFT_ETHER;
909 	sdl->sdl_alen = ifp->if_addrlen;
910 	bcopy(llc, LLADDR(sdl), ifp->if_addrlen);
911 	/*
912 	 * XXX: This doesn't belong here; we do it until
913 	 * XXX:  all drivers are cleaned up
914 	 */
915 	if (llc != IFP2AC(ifp)->ac_enaddr)
916 		bcopy(llc, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen);
917 
918 	bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
919 	if (ng_ether_attach_p != NULL)
920 		(*ng_ether_attach_p)(ifp);
921 	if (BDG_LOADED)
922 		bdgtakeifaces_ptr();
923 
924 	/* Announce Ethernet MAC address if non-zero. */
925 	for (i = 0; i < ifp->if_addrlen; i++)
926 		if (llc[i] != 0)
927 			break;
928 	if (i != ifp->if_addrlen)
929 		if_printf(ifp, "Ethernet address: %6D\n", llc, ":");
930 	if (debug_mpsafenet && (ifp->if_flags & IFF_NEEDSGIANT) != 0)
931 		if_printf(ifp, "if_start running deferred for Giant\n");
932 }
933 
934 /*
935  * Perform common duties while detaching an Ethernet interface
936  */
937 void
938 ether_ifdetach(struct ifnet *ifp)
939 {
940 	if (IFP2AC(ifp)->ac_netgraph != NULL)
941 		(*ng_ether_detach_p)(ifp);
942 	bpfdetach(ifp);
943 	if_detach(ifp);
944 	if (BDG_LOADED)
945 		bdgtakeifaces_ptr();
946 }
947 
948 SYSCTL_DECL(_net_link);
949 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
950 #if defined(INET) || defined(INET6)
951 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
952 	    &ether_ipfw,0,"Pass ether pkts through firewall");
953 #endif
954 
955 #if 0
956 /*
957  * This is for reference.  We have a table-driven version
958  * of the little-endian crc32 generator, which is faster
959  * than the double-loop.
960  */
961 uint32_t
962 ether_crc32_le(const uint8_t *buf, size_t len)
963 {
964 	size_t i;
965 	uint32_t crc;
966 	int bit;
967 	uint8_t data;
968 
969 	crc = 0xffffffff;	/* initial value */
970 
971 	for (i = 0; i < len; i++) {
972 		for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1)
973 			carry = (crc ^ data) & 1;
974 			crc >>= 1;
975 			if (carry)
976 				crc = (crc ^ ETHER_CRC_POLY_LE);
977 	}
978 
979 	return (crc);
980 }
981 #else
982 uint32_t
983 ether_crc32_le(const uint8_t *buf, size_t len)
984 {
985 	static const uint32_t crctab[] = {
986 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
987 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
988 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
989 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
990 	};
991 	size_t i;
992 	uint32_t crc;
993 
994 	crc = 0xffffffff;	/* initial value */
995 
996 	for (i = 0; i < len; i++) {
997 		crc ^= buf[i];
998 		crc = (crc >> 4) ^ crctab[crc & 0xf];
999 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1000 	}
1001 
1002 	return (crc);
1003 }
1004 #endif
1005 
1006 uint32_t
1007 ether_crc32_be(const uint8_t *buf, size_t len)
1008 {
1009 	size_t i;
1010 	uint32_t crc, carry;
1011 	int bit;
1012 	uint8_t data;
1013 
1014 	crc = 0xffffffff;	/* initial value */
1015 
1016 	for (i = 0; i < len; i++) {
1017 		for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
1018 			carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
1019 			crc <<= 1;
1020 			if (carry)
1021 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1022 		}
1023 	}
1024 
1025 	return (crc);
1026 }
1027 
1028 int
1029 ether_ioctl(struct ifnet *ifp, int command, caddr_t data)
1030 {
1031 	struct ifaddr *ifa = (struct ifaddr *) data;
1032 	struct ifreq *ifr = (struct ifreq *) data;
1033 	int error = 0;
1034 
1035 	switch (command) {
1036 	case SIOCSIFADDR:
1037 		ifp->if_flags |= IFF_UP;
1038 
1039 		switch (ifa->ifa_addr->sa_family) {
1040 #ifdef INET
1041 		case AF_INET:
1042 			ifp->if_init(ifp->if_softc);	/* before arpwhohas */
1043 			arp_ifinit(ifp, ifa);
1044 			break;
1045 #endif
1046 #ifdef IPX
1047 		/*
1048 		 * XXX - This code is probably wrong
1049 		 */
1050 		case AF_IPX:
1051 			{
1052 			struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr);
1053 			struct arpcom *ac = IFP2AC(ifp);
1054 
1055 			if (ipx_nullhost(*ina))
1056 				ina->x_host =
1057 				    *(union ipx_host *)
1058 				    ac->ac_enaddr;
1059 			else {
1060 				bcopy((caddr_t) ina->x_host.c_host,
1061 				      (caddr_t) ac->ac_enaddr,
1062 				      sizeof(ac->ac_enaddr));
1063 			}
1064 
1065 			/*
1066 			 * Set new address
1067 			 */
1068 			ifp->if_init(ifp->if_softc);
1069 			break;
1070 			}
1071 #endif
1072 		default:
1073 			ifp->if_init(ifp->if_softc);
1074 			break;
1075 		}
1076 		break;
1077 
1078 	case SIOCGIFADDR:
1079 		{
1080 			struct sockaddr *sa;
1081 
1082 			sa = (struct sockaddr *) & ifr->ifr_data;
1083 			bcopy(IFP2AC(ifp)->ac_enaddr,
1084 			      (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
1085 		}
1086 		break;
1087 
1088 	case SIOCSIFMTU:
1089 		/*
1090 		 * Set the interface MTU.
1091 		 */
1092 		if (ifr->ifr_mtu > ETHERMTU) {
1093 			error = EINVAL;
1094 		} else {
1095 			ifp->if_mtu = ifr->ifr_mtu;
1096 		}
1097 		break;
1098 	default:
1099 		error = EINVAL;			/* XXX netbsd has ENOTTY??? */
1100 		break;
1101 	}
1102 	return (error);
1103 }
1104 
1105 static int
1106 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
1107 	struct sockaddr *sa)
1108 {
1109 	struct sockaddr_dl *sdl;
1110 #ifdef INET
1111 	struct sockaddr_in *sin;
1112 #endif
1113 #ifdef INET6
1114 	struct sockaddr_in6 *sin6;
1115 #endif
1116 	u_char *e_addr;
1117 
1118 	switch(sa->sa_family) {
1119 	case AF_LINK:
1120 		/*
1121 		 * No mapping needed. Just check that it's a valid MC address.
1122 		 */
1123 		sdl = (struct sockaddr_dl *)sa;
1124 		e_addr = LLADDR(sdl);
1125 		if (!ETHER_IS_MULTICAST(e_addr))
1126 			return EADDRNOTAVAIL;
1127 		*llsa = 0;
1128 		return 0;
1129 
1130 #ifdef INET
1131 	case AF_INET:
1132 		sin = (struct sockaddr_in *)sa;
1133 		if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1134 			return EADDRNOTAVAIL;
1135 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1136 		       M_WAITOK|M_ZERO);
1137 		sdl->sdl_len = sizeof *sdl;
1138 		sdl->sdl_family = AF_LINK;
1139 		sdl->sdl_index = ifp->if_index;
1140 		sdl->sdl_type = IFT_ETHER;
1141 		sdl->sdl_alen = ETHER_ADDR_LEN;
1142 		e_addr = LLADDR(sdl);
1143 		ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1144 		*llsa = (struct sockaddr *)sdl;
1145 		return 0;
1146 #endif
1147 #ifdef INET6
1148 	case AF_INET6:
1149 		sin6 = (struct sockaddr_in6 *)sa;
1150 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1151 			/*
1152 			 * An IP6 address of 0 means listen to all
1153 			 * of the Ethernet multicast address used for IP6.
1154 			 * (This is used for multicast routers.)
1155 			 */
1156 			ifp->if_flags |= IFF_ALLMULTI;
1157 			*llsa = 0;
1158 			return 0;
1159 		}
1160 		if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1161 			return EADDRNOTAVAIL;
1162 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1163 		       M_WAITOK|M_ZERO);
1164 		sdl->sdl_len = sizeof *sdl;
1165 		sdl->sdl_family = AF_LINK;
1166 		sdl->sdl_index = ifp->if_index;
1167 		sdl->sdl_type = IFT_ETHER;
1168 		sdl->sdl_alen = ETHER_ADDR_LEN;
1169 		e_addr = LLADDR(sdl);
1170 		ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1171 		*llsa = (struct sockaddr *)sdl;
1172 		return 0;
1173 #endif
1174 
1175 	default:
1176 		/*
1177 		 * Well, the text isn't quite right, but it's the name
1178 		 * that counts...
1179 		 */
1180 		return EAFNOSUPPORT;
1181 	}
1182 }
1183 
1184 static moduledata_t ether_mod = {
1185 	"ether",
1186 	NULL,
1187 	0
1188 };
1189 
1190 DECLARE_MODULE(ether, ether_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
1191 MODULE_VERSION(ether, 1);
1192