xref: /dflybsd-src/sys/net/if_ethersubr.c (revision a6d5e0d89a9c7db991a06391d6a4a6ddadf6af51)
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 1.31 2005/06/03 23:23:03 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/ifq_var.h>
61 #include <net/bpf.h>
62 #include <net/ethernet.h>
63 #include <net/bridge/bridge.h>
64 
65 #if defined(INET) || defined(INET6)
66 #include <netinet/in.h>
67 #include <netinet/in_var.h>
68 #include <netinet/if_ether.h>
69 #include <net/ipfw/ip_fw.h>
70 #include <net/dummynet/ip_dummynet.h>
71 #endif
72 #ifdef INET6
73 #include <netinet6/nd6.h>
74 #endif
75 
76 #ifdef IPX
77 #include <netproto/ipx/ipx.h>
78 #include <netproto/ipx/ipx_if.h>
79 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m);
80 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, struct sockaddr *dst,
81 		  short *tp, int *hlen);
82 #endif
83 
84 #ifdef NS
85 #include <netns/ns.h>
86 #include <netns/ns_if.h>
87 ushort ns_nettype;
88 int ether_outputdebug = 0;
89 int ether_inputdebug = 0;
90 #endif
91 
92 #ifdef NETATALK
93 #include <netproto/atalk/at.h>
94 #include <netproto/atalk/at_var.h>
95 #include <netproto/atalk/at_extern.h>
96 
97 #define	llc_snap_org_code	llc_un.type_snap.org_code
98 #define	llc_snap_ether_type	llc_un.type_snap.ether_type
99 
100 extern u_char	at_org_code[3];
101 extern u_char	aarp_org_code[3];
102 #endif /* NETATALK */
103 
104 /* netgraph node hooks for ng_ether(4) */
105 void	(*ng_ether_input_p)(struct ifnet *ifp,
106 		struct mbuf **mp, struct ether_header *eh);
107 void	(*ng_ether_input_orphan_p)(struct ifnet *ifp,
108 		struct mbuf *m, struct ether_header *eh);
109 int	(*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
110 void	(*ng_ether_attach_p)(struct ifnet *ifp);
111 void	(*ng_ether_detach_p)(struct ifnet *ifp);
112 
113 int	(*vlan_input_p)(struct ether_header *eh, struct mbuf *m);
114 int	(*vlan_input_tag_p)(struct mbuf *m, uint16_t t);
115 
116 static int ether_output(struct ifnet *, struct mbuf *, struct sockaddr *,
117 			struct rtentry *);
118 
119 /*
120  * bridge support
121  */
122 int do_bridge;
123 bridge_in_t *bridge_in_ptr;
124 bdg_forward_t *bdg_forward_ptr;
125 bdgtakeifaces_t *bdgtakeifaces_ptr;
126 struct bdg_softc *ifp2sc;
127 
128 static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
129 			      struct sockaddr *);
130 
131 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = {
132 	0xff, 0xff, 0xff, 0xff, 0xff, 0xff
133 };
134 
135 #define gotoerr(e) do { error = (e); goto bad; } while (0)
136 #define IFP2AC(ifp) ((struct arpcom *)(ifp))
137 
138 static boolean_t ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
139 				struct ip_fw **rule, struct ether_header *eh,
140 				boolean_t shared);
141 
142 static int ether_ipfw;
143 SYSCTL_DECL(_net_link);
144 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
145 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
146 	   &ether_ipfw, 0, "Pass ether pkts through firewall");
147 
148 /*
149  * Ethernet output routine.
150  * Encapsulate a packet of type family for the local net.
151  * Use trailer local net encapsulation if enough data in first
152  * packet leaves a multiple of 512 bytes of data in remainder.
153  * Assumes that ifp is actually pointer to arpcom structure.
154  */
155 static int
156 ether_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst,
157 	     struct rtentry *rt)
158 {
159 	struct ether_header *eh, *deh;
160 	u_char *edst;
161 	int loop_copy = 0;
162 	int hlen = ETHER_HDR_LEN;	/* link layer header length */
163 	struct arpcom *ac = IFP2AC(ifp);
164 	int error;
165 
166 	if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING))
167 		gotoerr(ENETDOWN);
168 
169 	M_PREPEND(m, sizeof(struct ether_header), MB_DONTWAIT);
170 	if (m == NULL)
171 		return (ENOBUFS);
172 	eh = mtod(m, struct ether_header *);
173 	edst = eh->ether_dhost;
174 
175 	/*
176 	 * Fill in the destination ethernet address and frame type.
177 	 */
178 	switch (dst->sa_family) {
179 #ifdef INET
180 	case AF_INET:
181 		if (!arpresolve(ifp, rt, m, dst, edst))
182 			return (0);	/* if not yet resolved */
183 		eh->ether_type = htons(ETHERTYPE_IP);
184 		break;
185 #endif
186 #ifdef INET6
187 	case AF_INET6:
188 		if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, edst))
189 			return (0);		/* Something bad happenned. */
190 		eh->ether_type = htons(ETHERTYPE_IPV6);
191 		break;
192 #endif
193 #ifdef IPX
194 	case AF_IPX:
195 		if (ef_outputp != NULL) {
196 			error = ef_outputp(ifp, &m, dst, &eh->ether_type,
197 					   &hlen);
198 			if (error)
199 				goto bad;
200 		} else {
201 			eh->ether_type = htons(ETHERTYPE_IPX);
202 			bcopy(&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
203 			      edst, ETHER_ADDR_LEN);
204 		}
205 		break;
206 #endif
207 #ifdef NETATALK
208 	case AF_APPLETALK: {
209 		struct at_ifaddr *aa;
210 
211 		if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) {
212 			error = 0;	/* XXX */
213 			goto bad;
214 		}
215 		/*
216 		 * In the phase 2 case, need to prepend an mbuf for
217 		 * the llc header.  Since we must preserve the value
218 		 * of m, which is passed to us by value, we m_copy()
219 		 * the first mbuf, and use it for our llc header.
220 		 */
221 		if (aa->aa_flags & AFA_PHASE2) {
222 			struct llc llc;
223 
224 			M_PREPEND(m, sizeof(struct llc), MB_DONTWAIT);
225 			eh = mtod(m, struct ether_header *);
226 			edst = eh->ether_dhost;
227 			llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
228 			llc.llc_control = LLC_UI;
229 			bcopy(at_org_code, llc.llc_snap_org_code,
230 			      sizeof at_org_code);
231 			llc.llc_snap_ether_type = htons(ETHERTYPE_AT);
232 			bcopy(&llc,
233 			      mtod(m, caddr_t) + sizeof(struct ether_header),
234 			      sizeof(struct llc));
235 			eh->ether_type = htons(m->m_pkthdr.len);
236 			hlen = sizeof(struct llc) + ETHER_HDR_LEN;
237 		} else {
238 			eh->ether_type = htons(ETHERTYPE_AT);
239 		}
240 		if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst))
241 			return (0);
242 		break;
243 	  }
244 #endif
245 #ifdef NS
246 	case AF_NS:
247 		switch(ns_nettype) {
248 		default:
249 		case 0x8137:	/* Novell Ethernet_II Ethernet TYPE II */
250 			eh->ether_type = 0x8137;
251 			break;
252 		case 0x0:	/* Novell 802.3 */
253 			eh->ether_type = htons(m->m_pkthdr.len);
254 			break;
255 		case 0xe0e0:	/* Novell 802.2 and Token-Ring */
256 			M_PREPEND(m, 3, MB_DONTWAIT);
257 			eh = mtod(m, struct ether_header *);
258 			edst = eh->ether_dhost;
259 			eh->ether_type = htons(m->m_pkthdr.len);
260 			cp = mtod(m, u_char *) + sizeof(struct ether_header);
261 			*cp++ = 0xE0;
262 			*cp++ = 0xE0;
263 			*cp++ = 0x03;
264 			break;
265 		}
266 		bcopy(&(((struct sockaddr_ns *)dst)->sns_addr.x_host), edst,
267 		      ETHER_ADDR_LEN);
268 		/*
269 		 * XXX if ns_thishost is the same as the node's ethernet
270 		 * address then just the default code will catch this anyhow.
271 		 * So I'm not sure if this next clause should be here at all?
272 		 * [JRE]
273 		 */
274 		if (bcmp(edst, &ns_thishost, ETHER_ADDR_LEN) == 0) {
275 			m->m_pkthdr.rcvif = ifp;
276 			netisr_dispatch(NETISR_NS, m);
277 			return (error);
278 		}
279 		if (bcmp(edst, &ns_broadhost, ETHER_ADDR_LEN) == 0)
280 			m->m_flags |= M_BCAST;
281 		break;
282 #endif
283 	case pseudo_AF_HDRCMPLT:
284 	case AF_UNSPEC:
285 		loop_copy = -1; /* if this is for us, don't do it */
286 		deh = (struct ether_header *)dst->sa_data;
287 		memcpy(edst, deh->ether_dhost, ETHER_ADDR_LEN);
288 		eh->ether_type = deh->ether_type;
289 		break;
290 
291 	default:
292 		printf("%s: can't handle af%d\n", ifp->if_xname,
293 			dst->sa_family);
294 		gotoerr(EAFNOSUPPORT);
295 	}
296 
297 	if (dst->sa_family == pseudo_AF_HDRCMPLT)	/* unlikely */
298 		memcpy(eh->ether_shost,
299 		       ((struct ether_header *)dst->sa_data)->ether_shost,
300 		       ETHER_ADDR_LEN);
301 	else
302 		memcpy(eh->ether_shost, ac->ac_enaddr, ETHER_ADDR_LEN);
303 
304 	/*
305 	 * If a simplex interface, and the packet is being sent to our
306 	 * Ethernet address or a broadcast address, loopback a copy.
307 	 * XXX To make a simplex device behave exactly like a duplex
308 	 * device, we should copy in the case of sending to our own
309 	 * ethernet address (thus letting the original actually appear
310 	 * on the wire). However, we don't do that here for security
311 	 * reasons and compatibility with the original behavior.
312 	 */
313 	if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) {
314 		int csum_flags = 0;
315 
316 		if (m->m_pkthdr.csum_flags & CSUM_IP)
317 			csum_flags |= (CSUM_IP_CHECKED | CSUM_IP_VALID);
318 		if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
319 			csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR);
320 		if ((m->m_flags & M_BCAST) || (loop_copy > 0)) {
321 			struct mbuf *n;
322 
323 			if ((n = m_copypacket(m, MB_DONTWAIT)) != NULL) {
324 				n->m_pkthdr.csum_flags |= csum_flags;
325 				if (csum_flags & CSUM_DATA_VALID)
326 					n->m_pkthdr.csum_data = 0xffff;
327 				if_simloop(ifp, n, dst->sa_family, hlen);
328 			} else
329 				ifp->if_iqdrops++;
330 		} else if (bcmp(eh->ether_dhost, eh->ether_shost,
331 				ETHER_ADDR_LEN) == 0) {
332 			m->m_pkthdr.csum_flags |= csum_flags;
333 			if (csum_flags & CSUM_DATA_VALID)
334 				m->m_pkthdr.csum_data = 0xffff;
335 			if_simloop(ifp, m, dst->sa_family, hlen);
336 			return (0);	/* XXX */
337 		}
338 	}
339 
340 	/* Handle ng_ether(4) processing, if any */
341 	if (ng_ether_output_p != NULL) {
342 		if ((error = (*ng_ether_output_p)(ifp, &m)) != 0)
343 			goto bad;
344 		if (m == NULL)
345 			return (0);
346 	}
347 
348 	/* Continue with link-layer output */
349 	return ether_output_frame(ifp, m);
350 
351 bad:
352 	m_freem(m);
353 	return (error);
354 }
355 
356 /*
357  * Ethernet link layer output routine to send a raw frame to the device.
358  *
359  * This assumes that the 14 byte Ethernet header is present and contiguous
360  * in the first mbuf (if BRIDGE'ing).
361  */
362 int
363 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
364 {
365 	struct ip_fw *rule = NULL;
366 	int error = 0;
367 	int s;
368 	struct altq_pktattr pktattr;
369 
370 	/* Extract info from dummynet tag, ignore others */
371 	while (m->m_type == MT_TAG) {
372 		if (m->m_flags == PACKET_TAG_DUMMYNET) {
373 			rule = ((struct dn_pkt *)m)->rule;
374 			break;
375 		}
376 		m = m->m_next;
377 	}
378 	if (rule != NULL)		/* packet was already bridged */
379 		goto no_bridge;
380 
381 	if (BDG_ACTIVE(ifp)) {
382 		struct ether_header *eh;	/* a pointer suffices */
383 
384 		m->m_pkthdr.rcvif = NULL;
385 		eh = mtod(m, struct ether_header *);
386 		m_adj(m, ETHER_HDR_LEN);
387 		m = bdg_forward_ptr(m, eh, ifp);
388 		m_freem(m);
389 		return (0);
390 	}
391 
392 no_bridge:
393 	if (ifq_is_enabled(&ifp->if_snd))
394 		altq_etherclassify(&ifp->if_snd, m, &pktattr);
395 	s = splimp();
396 	if (IPFW_LOADED && ether_ipfw != 0) {
397 		struct ether_header save_eh, *eh;
398 
399 		eh = mtod(m, struct ether_header *);
400 		save_eh = *eh;
401 		m_adj(m, ETHER_HDR_LEN);
402 		if (!ether_ipfw_chk(&m, ifp, &rule, eh, FALSE)) {
403 			if (m != NULL) {
404 				m_freem(m);
405 				return ENOBUFS; /* pkt dropped */
406 			} else
407 				return 0;	/* consumed e.g. in a pipe */
408 		}
409 		eh = mtod(m, struct ether_header *);
410 		/* packet was ok, restore the ethernet header */
411 		if ((void *)(eh + 1) == (void *)m->m_data) {
412 			m->m_data -= ETHER_HDR_LEN ;
413 			m->m_len += ETHER_HDR_LEN ;
414 			m->m_pkthdr.len += ETHER_HDR_LEN ;
415 		} else {
416 			M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT);
417 			if (m == NULL) /* nope... */
418 				return ENOBUFS;
419 			bcopy(&save_eh, mtod(m, struct ether_header *),
420 			      ETHER_HDR_LEN);
421 		}
422 	}
423 
424 	/*
425 	 * Queue message on interface, update output statistics if
426 	 * successful, and start output if interface not yet active.
427 	 */
428 	error = ifq_handoff(ifp, m, &pktattr);
429 	splx(s);
430 	return (error);
431 }
432 
433 /*
434  * ipfw processing for ethernet packets (in and out).
435  * The second parameter is NULL from ether_demux(), and ifp from
436  * ether_output_frame(). This section of code could be used from
437  * bridge.c as well as long as we use some extra info
438  * to distinguish that case from ether_output_frame().
439  */
440 static boolean_t
441 ether_ipfw_chk(
442 	struct mbuf **m0,
443 	struct ifnet *dst,
444 	struct ip_fw **rule,
445 	struct ether_header *eh,
446 	boolean_t shared)
447 {
448 	struct ether_header save_eh = *eh;	/* might be a ptr in m */
449 	struct ip_fw_args args;
450 	struct m_tag *mtag;
451 	int i;
452 
453 	if (*rule != NULL && fw_one_pass)
454 		return TRUE; /* dummynet packet, already partially processed */
455 
456 	/*
457 	 * I need some amount of data to be contiguous, and in case others
458 	 * need the packet (shared==TRUE), it also better be in the first mbuf.
459 	 */
460 	i = min((*m0)->m_pkthdr.len, max_protohdr);
461 	if (shared || (*m0)->m_len < i) {
462 		*m0 = m_pullup(*m0, i);
463 		if (*m0 == NULL)
464 			return FALSE;
465 	}
466 
467 	args.m = *m0;		/* the packet we are looking at		*/
468 	args.oif = dst;		/* destination, if any			*/
469 	if ((mtag = m_tag_find(*m0, PACKET_TAG_IPFW_DIVERT, NULL)) != NULL)
470 		m_tag_delete(*m0, mtag);
471 	args.rule = *rule;	/* matching rule to restart		*/
472 	args.next_hop = NULL;	/* we do not support forward yet	*/
473 	args.eh = &save_eh;	/* MAC header for bridged/MAC packets	*/
474 	i = ip_fw_chk_ptr(&args);
475 	*m0 = args.m;
476 	*rule = args.rule;
477 
478 	if ((i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL)	/* drop */
479 		return FALSE;
480 
481 	if (i == 0)					/* a PASS rule.  */
482 		return TRUE;
483 
484 	if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG)) {
485 		/*
486 		 * Pass the pkt to dummynet, which consumes it.
487 		 * If shared, make a copy and keep the original.
488 		 */
489 		struct mbuf *m ;
490 
491 		if (shared) {
492 			m = m_copypacket(*m0, MB_DONTWAIT);
493 			if (m == NULL)
494 				return FALSE;
495 		} else {
496 			m = *m0 ;	/* pass the original to dummynet */
497 			*m0 = NULL ;	/* and nothing back to the caller */
498 		}
499 		/*
500 		 * Prepend the header, optimize for the common case of
501 		 * eh pointing into the mbuf.
502 		 */
503 		if ((void *)(eh + 1) == (void *)m->m_data) {
504 			m->m_data -= ETHER_HDR_LEN ;
505 			m->m_len += ETHER_HDR_LEN ;
506 			m->m_pkthdr.len += ETHER_HDR_LEN ;
507 		} else {
508 			M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT);
509 			if (m == NULL)
510 				return FALSE;
511 			bcopy(&save_eh, mtod(m, struct ether_header *),
512 			      ETHER_HDR_LEN);
513 		}
514 		ip_dn_io_ptr(m, (i & 0xffff),
515 			     dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
516 		return FALSE;
517 	}
518 	/*
519 	 * XXX at some point add support for divert/forward actions.
520 	 * If none of the above matches, we have to drop the pkt.
521 	 */
522 	return FALSE;
523 }
524 
525 /*
526  * XXX merge this function with ether_input.
527  */
528 static void
529 ether_input_internal(struct ifnet *ifp, struct mbuf *m)
530 {
531 	ether_input(ifp, NULL, m);
532 }
533 
534 /*
535  * Process a received Ethernet packet. We have two different interfaces:
536  * one (conventional) assumes the packet in the mbuf, with the ethernet
537  * header provided separately in *eh. The second one (new) has everything
538  * in the mbuf, and we can tell it because eh == NULL.
539  * The caller MUST MAKE SURE that there are at least
540  * sizeof(struct ether_header) bytes in the first mbuf.
541  *
542  * This allows us to concentrate in one place a bunch of code which
543  * is replicated in all device drivers. Also, many functions called
544  * from ether_input() try to put the eh back into the mbuf, so we
545  * can later propagate the 'contiguous packet' interface to them,
546  * and handle the old interface just here.
547  *
548  * NOTA BENE: for many drivers "eh" is a pointer into the first mbuf or
549  * cluster, right before m_data. So be very careful when working on m,
550  * as you could destroy *eh !!
551  *
552  * First we perform any link layer operations, then continue
553  * to the upper layers with ether_demux().
554  */
555 void
556 ether_input(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m)
557 {
558 	struct ether_header save_eh;
559 
560 	if (eh == NULL) {
561 		if (m->m_len < sizeof(struct ether_header)) {
562 			/* XXX error in the caller. */
563 			m_freem(m);
564 			return;
565 		}
566 		m->m_pkthdr.rcvif = ifp;
567 		eh = mtod(m, struct ether_header *);
568 		m_adj(m, sizeof(struct ether_header));
569 		/* XXX */
570 		/* m->m_pkthdr.len = m->m_len; */
571 	}
572 
573 	if (ifp->if_bpf)
574 		bpf_ptap(ifp->if_bpf, m, eh, ETHER_HDR_LEN);
575 
576 	ifp->if_ibytes += m->m_pkthdr.len + (sizeof *eh);
577 
578 	/* Handle ng_ether(4) processing, if any */
579 	if (ng_ether_input_p != NULL) {
580 		(*ng_ether_input_p)(ifp, &m, eh);
581 		if (m == NULL)
582 			return;
583 	}
584 
585 	/* Check for bridging mode */
586 	if (BDG_ACTIVE(ifp)) {
587 		struct ifnet *bif;
588 
589 		/* Check with bridging code */
590 		if ((bif = bridge_in_ptr(ifp, eh)) == BDG_DROP) {
591 			m_freem(m);
592 			return;
593 		}
594 		if (bif != BDG_LOCAL) {
595 			save_eh = *eh ; /* because it might change */
596 			m = bdg_forward_ptr(m, eh, bif); /* needs forwarding */
597 			/*
598 			 * Do not continue if bdg_forward_ptr() processed our
599 			 * packet (and cleared the mbuf pointer m) or if
600 			 * it dropped (m_free'd) the packet itself.
601 			 */
602 			if (m == NULL) {
603 			    if (bif == BDG_BCAST || bif == BDG_MCAST)
604 				printf("bdg_forward drop MULTICAST PKT\n");
605 			    return;
606 			}
607 			eh = &save_eh ;
608 		}
609 		if (bif == BDG_LOCAL || bif == BDG_BCAST || bif == BDG_MCAST)
610 			goto recvLocal;		/* receive locally */
611 
612 		/* If not local and not multicast, just drop it */
613 		m_freem(m);
614 		return;
615 	}
616 
617 recvLocal:
618 	/* Continue with upper layer processing */
619 	ether_demux(ifp, eh, m);
620 }
621 
622 /*
623  * Upper layer processing for a received Ethernet packet.
624  */
625 void
626 ether_demux(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m)
627 {
628 	int isr;
629 	u_short ether_type;
630 	struct ip_fw *rule = NULL;
631 #ifdef NETATALK
632 	struct llc *l;
633 #endif
634 
635 	/* Extract info from dummynet tag, ignore others */
636 	while (m->m_type == MT_TAG) {
637 		if (m->m_flags == PACKET_TAG_DUMMYNET) {
638 			rule = ((struct dn_pkt *)m)->rule;
639 			ifp = m->m_next->m_pkthdr.rcvif;
640 			break;
641 		}
642 		m = m->m_next;
643 	}
644 	if (rule)	/* packet was already bridged */
645 		goto post_stats;
646 
647 	/*
648 	 * Discard packet if upper layers shouldn't see it because
649 	 * it was unicast to a different Ethernet address.  If the
650 	 * driver is working properly, then this situation can only
651 	 * happen when the interface is in promiscuous mode.
652 	 */
653 	if (!BDG_ACTIVE(ifp) &&
654 	    ((ifp->if_flags & (IFF_PROMISC | IFF_PPROMISC)) == IFF_PROMISC) &&
655 	    (eh->ether_dhost[0] & 1) == 0 &&
656 	    bcmp(eh->ether_dhost, IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN)) {
657 		m_freem(m);
658 		return;
659 	}
660 	/* Discard packet if interface is not up */
661 	if (!(ifp->if_flags & IFF_UP)) {
662 		m_freem(m);
663 		return;
664 	}
665 	if (eh->ether_dhost[0] & 1) {
666 		if (bcmp(ifp->if_broadcastaddr, eh->ether_dhost,
667 			 ifp->if_addrlen) == 0)
668 			m->m_flags |= M_BCAST;
669 		else
670 			m->m_flags |= M_MCAST;
671 		ifp->if_imcasts++;
672 	}
673 
674 post_stats:
675 	if (IPFW_LOADED && ether_ipfw != 0) {
676 		if (!ether_ipfw_chk(&m, NULL, &rule, eh, FALSE)) {
677 			m_freem(m);
678 			return;
679 		}
680 		eh = mtod(m, struct ether_header *);
681 	}
682 
683 	ether_type = ntohs(eh->ether_type);
684 
685 	switch (ether_type) {
686 #ifdef INET
687 	case ETHERTYPE_IP:
688 		if (ipflow_fastforward(m))
689 			return;
690 		isr = NETISR_IP;
691 		break;
692 
693 	case ETHERTYPE_ARP:
694 		if (ifp->if_flags & IFF_NOARP) {
695 			/* Discard packet if ARP is disabled on interface */
696 			m_freem(m);
697 			return;
698 		}
699 		isr = NETISR_ARP;
700 		break;
701 #endif
702 
703 #ifdef INET6
704 	case ETHERTYPE_IPV6:
705 		isr = NETISR_IPV6;
706 		break;
707 #endif
708 
709 #ifdef IPX
710 	case ETHERTYPE_IPX:
711 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
712 			return;
713 		isr = NETISR_IPX;
714 		break;
715 #endif
716 
717 #ifdef NS
718 	case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
719 		isr = NETISR_NS;
720 		break;
721 
722 #endif
723 
724 #ifdef NETATALK
725 	case ETHERTYPE_AT:
726 		isr = NETISR_ATALK1;
727 		break;
728 	case ETHERTYPE_AARP:
729 		isr = NETISR_AARP;
730 		break;
731 #endif
732 
733 	case ETHERTYPE_VLAN:
734 		if (vlan_input_p != NULL)
735 			(*vlan_input_p)(eh, m);
736 		else {
737 			m->m_pkthdr.rcvif->if_noproto++;
738 			m_freem(m);
739 		}
740 		return;
741 
742 	default:
743 #ifdef IPX
744 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
745 			return;
746 #endif
747 #ifdef NS
748 		checksum = mtod(m, ushort *);
749 		/* Novell 802.3 */
750 		if ((ether_type <= ETHERMTU) &&
751 		    ((*checksum == 0xffff) || (*checksum == 0xE0E0))) {
752 			if (*checksum == 0xE0E0) {
753 				m->m_pkthdr.len -= 3;
754 				m->m_len -= 3;
755 				m->m_data += 3;
756 			}
757 			isr = NETISR_NS;
758 			break;
759 		}
760 #endif
761 #ifdef NETATALK
762 		if (ether_type > ETHERMTU)
763 			goto dropanyway;
764 		l = mtod(m, struct llc *);
765 		if (l->llc_dsap == LLC_SNAP_LSAP &&
766 		    l->llc_ssap == LLC_SNAP_LSAP &&
767 		    l->llc_control == LLC_UI) {
768 			if (bcmp(&(l->llc_snap_org_code)[0], at_org_code,
769 				 sizeof at_org_code) == 0 &&
770 			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
771 				m_adj(m, sizeof(struct llc));
772 				isr = NETISR_ATALK2;
773 				break;
774 			}
775 			if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
776 				 sizeof aarp_org_code) == 0 &&
777 			    ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
778 				m_adj(m, sizeof(struct llc));
779 				isr = NETISR_AARP;
780 				break;
781 			}
782 		}
783 dropanyway:
784 #endif
785 		if (ng_ether_input_orphan_p != NULL)
786 			(*ng_ether_input_orphan_p)(ifp, m, eh);
787 		else
788 			m_freem(m);
789 		return;
790 	}
791 	netisr_dispatch(isr, m);
792 }
793 
794 /*
795  * Perform common duties while attaching to interface list
796  */
797 
798 void
799 ether_ifattach(struct ifnet *ifp, uint8_t *lla)
800 {
801 	ether_ifattach_bpf(ifp, lla, DLT_EN10MB, sizeof(struct ether_header));
802 }
803 
804 void
805 ether_ifattach_bpf(struct ifnet *ifp, uint8_t *lla, u_int dlt, u_int hdrlen)
806 {
807 	struct sockaddr_dl *sdl;
808 
809 	ifp->if_type = IFT_ETHER;
810 	ifp->if_addrlen = ETHER_ADDR_LEN;
811 	ifp->if_hdrlen = ETHER_HDR_LEN;
812 	if_attach(ifp);
813 	ifp->if_mtu = ETHERMTU;
814 	if (ifp->if_baudrate == 0)
815 		ifp->if_baudrate = 10000000;
816 	ifp->if_output = ether_output;
817 	ifp->if_input = ether_input_internal;
818 	ifp->if_resolvemulti = ether_resolvemulti;
819 	ifp->if_broadcastaddr = etherbroadcastaddr;
820 	sdl = IF_LLSOCKADDR(ifp);
821 	sdl->sdl_type = IFT_ETHER;
822 	sdl->sdl_alen = ifp->if_addrlen;
823 	bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
824 	/*
825 	 * XXX Keep the current drivers happy.
826 	 * XXX Remove once all drivers have been cleaned up
827 	 */
828 	if (lla != IFP2AC(ifp)->ac_enaddr)
829 		bcopy(lla, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen);
830 	bpfattach(ifp, dlt, hdrlen);
831 	if (ng_ether_attach_p != NULL)
832 		(*ng_ether_attach_p)(ifp);
833 	if (BDG_LOADED)
834 		bdgtakeifaces_ptr();
835 
836 	if_printf(ifp, "MAC address: %6D\n", lla, ":");
837 }
838 
839 /*
840  * Perform common duties while detaching an Ethernet interface
841  */
842 void
843 ether_ifdetach(struct ifnet *ifp)
844 {
845 	int s;
846 
847 	s = splnet();
848 	if_down(ifp);
849 	splx(s);
850 
851 	if (ng_ether_detach_p != NULL)
852 		(*ng_ether_detach_p)(ifp);
853 	bpfdetach(ifp);
854 	if_detach(ifp);
855 	if (BDG_LOADED)
856 		bdgtakeifaces_ptr();
857 }
858 
859 int
860 ether_ioctl(struct ifnet *ifp, int command, caddr_t data)
861 {
862 	struct ifaddr *ifa = (struct ifaddr *) data;
863 	struct ifreq *ifr = (struct ifreq *) data;
864 	int error = 0;
865 
866 	switch (command) {
867 	case SIOCSIFADDR:
868 		ifp->if_flags |= IFF_UP;
869 
870 		switch (ifa->ifa_addr->sa_family) {
871 #ifdef INET
872 		case AF_INET:
873 			ifp->if_init(ifp->if_softc);	/* before arpwhohas */
874 			arp_ifinit(ifp, ifa);
875 			break;
876 #endif
877 #ifdef IPX
878 		/*
879 		 * XXX - This code is probably wrong
880 		 */
881 		case AF_IPX:
882 			{
883 			struct ipx_addr *ina = &IA_SIPX(ifa)->sipx_addr;
884 			struct arpcom *ac = IFP2AC(ifp);
885 
886 			if (ipx_nullhost(*ina))
887 				ina->x_host = *(union ipx_host *) ac->ac_enaddr;
888 			else
889 				bcopy(ina->x_host.c_host, ac->ac_enaddr,
890 				      sizeof ac->ac_enaddr);
891 
892 			ifp->if_init(ifp->if_softc);	/* Set new address. */
893 			break;
894 			}
895 #endif
896 #ifdef NS
897 		/*
898 		 * XXX - This code is probably wrong
899 		 */
900 		case AF_NS:
901 		{
902 			struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr);
903 			struct arpcom *ac = IFP2AC(ifp);
904 
905 			if (ns_nullhost(*ina))
906 				ina->x_host = *(union ns_host *)(ac->ac_enaddr);
907 			else
908 				bcopy(ina->x_host.c_host, ac->ac_enaddr,
909 				      sizeof ac->ac_enaddr);
910 
911 			/*
912 			 * Set new address
913 			 */
914 			ifp->if_init(ifp->if_softc);
915 			break;
916 		}
917 #endif
918 		default:
919 			ifp->if_init(ifp->if_softc);
920 			break;
921 		}
922 		break;
923 
924 	case SIOCGIFADDR:
925 		bcopy(IFP2AC(ifp)->ac_enaddr,
926 		      ((struct sockaddr *)ifr->ifr_data)->sa_data,
927 		      ETHER_ADDR_LEN);
928 		break;
929 
930 	case SIOCSIFMTU:
931 		/*
932 		 * Set the interface MTU.
933 		 */
934 		if (ifr->ifr_mtu > ETHERMTU) {
935 			error = EINVAL;
936 		} else {
937 			ifp->if_mtu = ifr->ifr_mtu;
938 		}
939 		break;
940 	default:
941 		error = EINVAL;
942 		break;
943 	}
944 	return (error);
945 }
946 
947 int
948 ether_resolvemulti(
949 	struct ifnet *ifp,
950 	struct sockaddr **llsa,
951 	struct sockaddr *sa)
952 {
953 	struct sockaddr_dl *sdl;
954 	struct sockaddr_in *sin;
955 #ifdef INET6
956 	struct sockaddr_in6 *sin6;
957 #endif
958 	u_char *e_addr;
959 
960 	switch(sa->sa_family) {
961 	case AF_LINK:
962 		/*
963 		 * No mapping needed. Just check that it's a valid MC address.
964 		 */
965 		sdl = (struct sockaddr_dl *)sa;
966 		e_addr = LLADDR(sdl);
967 		if ((e_addr[0] & 1) != 1)
968 			return EADDRNOTAVAIL;
969 		*llsa = 0;
970 		return 0;
971 
972 #ifdef INET
973 	case AF_INET:
974 		sin = (struct sockaddr_in *)sa;
975 		if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
976 			return EADDRNOTAVAIL;
977 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
978 		       M_WAITOK | M_ZERO);
979 		sdl->sdl_len = sizeof *sdl;
980 		sdl->sdl_family = AF_LINK;
981 		sdl->sdl_index = ifp->if_index;
982 		sdl->sdl_type = IFT_ETHER;
983 		sdl->sdl_alen = ETHER_ADDR_LEN;
984 		e_addr = LLADDR(sdl);
985 		ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
986 		*llsa = (struct sockaddr *)sdl;
987 		return 0;
988 #endif
989 #ifdef INET6
990 	case AF_INET6:
991 		sin6 = (struct sockaddr_in6 *)sa;
992 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
993 			/*
994 			 * An IP6 address of 0 means listen to all
995 			 * of the Ethernet multicast address used for IP6.
996 			 * (This is used for multicast routers.)
997 			 */
998 			ifp->if_flags |= IFF_ALLMULTI;
999 			*llsa = 0;
1000 			return 0;
1001 		}
1002 		if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1003 			return EADDRNOTAVAIL;
1004 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1005 		       M_WAITOK | M_ZERO);
1006 		sdl->sdl_len = sizeof *sdl;
1007 		sdl->sdl_family = AF_LINK;
1008 		sdl->sdl_index = ifp->if_index;
1009 		sdl->sdl_type = IFT_ETHER;
1010 		sdl->sdl_alen = ETHER_ADDR_LEN;
1011 		e_addr = LLADDR(sdl);
1012 		ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1013 		*llsa = (struct sockaddr *)sdl;
1014 		return 0;
1015 #endif
1016 
1017 	default:
1018 		/*
1019 		 * Well, the text isn't quite right, but it's the name
1020 		 * that counts...
1021 		 */
1022 		return EAFNOSUPPORT;
1023 	}
1024 }
1025 
1026 #if 0
1027 /*
1028  * This is for reference.  We have a table-driven version
1029  * of the little-endian crc32 generator, which is faster
1030  * than the double-loop.
1031  */
1032 uint32_t
1033 ether_crc32_le(const uint8_t *buf, size_t len)
1034 {
1035 	uint32_t c, crc, carry;
1036 	size_t i, j;
1037 
1038 	crc = 0xffffffffU;	/* initial value */
1039 
1040 	for (i = 0; i < len; i++) {
1041 		c = buf[i];
1042 		for (j = 0; j < 8; j++) {
1043 			carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
1044 			crc >>= 1;
1045 			c >>= 1;
1046 			if (carry)
1047 				crc = (crc ^ ETHER_CRC_POLY_LE);
1048 		}
1049 	}
1050 
1051 	return (crc);
1052 }
1053 #else
1054 uint32_t
1055 ether_crc32_le(const uint8_t *buf, size_t len)
1056 {
1057 	static const uint32_t crctab[] = {
1058 		0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1059 		0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1060 		0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1061 		0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1062 	};
1063 	uint32_t crc;
1064 	size_t i;
1065 
1066 	crc = 0xffffffffU;	/* initial value */
1067 
1068 	for (i = 0; i < len; i++) {
1069 		crc ^= buf[i];
1070 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1071 		crc = (crc >> 4) ^ crctab[crc & 0xf];
1072 	}
1073 
1074 	return (crc);
1075 }
1076 #endif
1077 
1078 uint32_t
1079 ether_crc32_be(const uint8_t *buf, size_t len)
1080 {
1081 	uint32_t c, crc, carry;
1082 	size_t i, j;
1083 
1084 	crc = 0xffffffffU;	/* initial value */
1085 
1086 	for (i = 0; i < len; i++) {
1087 		c = buf[i];
1088 		for (j = 0; j < 8; j++) {
1089 			carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
1090 			crc <<= 1;
1091 			c >>= 1;
1092 			if (carry)
1093 				crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1094 		}
1095 	}
1096 
1097 	return (crc);
1098 }
1099 
1100 /*
1101  * find the size of ethernet header, and call classifier
1102  */
1103 void
1104 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
1105 		   struct altq_pktattr *pktattr)
1106 {
1107 	struct ether_header *eh;
1108 	uint16_t ether_type;
1109 	int hlen, af, hdrsize;
1110 	caddr_t hdr;
1111 
1112 	hlen = sizeof(struct ether_header);
1113 	eh = mtod(m, struct ether_header *);
1114 
1115 	ether_type = ntohs(eh->ether_type);
1116 	if (ether_type < ETHERMTU) {
1117 		/* ick! LLC/SNAP */
1118 		struct llc *llc = (struct llc *)(eh + 1);
1119 		hlen += 8;
1120 
1121 		if (m->m_len < hlen ||
1122 		    llc->llc_dsap != LLC_SNAP_LSAP ||
1123 		    llc->llc_ssap != LLC_SNAP_LSAP ||
1124 		    llc->llc_control != LLC_UI)
1125 			goto bad;  /* not snap! */
1126 
1127 		ether_type = ntohs(llc->llc_un.type_snap.ether_type);
1128 	}
1129 
1130 	if (ether_type == ETHERTYPE_IP) {
1131 		af = AF_INET;
1132 		hdrsize = 20;  /* sizeof(struct ip) */
1133 #ifdef INET6
1134 	} else if (ether_type == ETHERTYPE_IPV6) {
1135 		af = AF_INET6;
1136 		hdrsize = 40;  /* sizeof(struct ip6_hdr) */
1137 #endif
1138 	} else
1139 		goto bad;
1140 
1141 	while (m->m_len <= hlen) {
1142 		hlen -= m->m_len;
1143 		m = m->m_next;
1144 	}
1145 	hdr = m->m_data + hlen;
1146 	if (m->m_len < hlen + hdrsize) {
1147 		/*
1148 		 * ip header is not in a single mbuf.  this should not
1149 		 * happen in the current code.
1150 		 * (todo: use m_pulldown in the future)
1151 		 */
1152 		goto bad;
1153 	}
1154 	m->m_data += hlen;
1155 	m->m_len -= hlen;
1156 	ifq_classify(ifq, m, af, pktattr);
1157 	m->m_data -= hlen;
1158 	m->m_len += hlen;
1159 
1160 	return;
1161 
1162 bad:
1163 	pktattr->pattr_class = NULL;
1164 	pktattr->pattr_hdr = NULL;
1165 	pktattr->pattr_af = AF_UNSPEC;
1166 }
1167