xref: /freebsd/sys/net/if_ethersubr.c (revision 71fe318b852b8dfb3e799cb12ef184750f7f8eac)
1 /*
2  * Copyright (c) 1982, 1989, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. All advertising materials mentioning features or use of this software
14  *    must display the following acknowledgement:
15  *	This product includes software developed by the University of
16  *	California, Berkeley and its contributors.
17  * 4. Neither the name of the University nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  *	@(#)if_ethersubr.c	8.1 (Berkeley) 6/10/93
34  * $FreeBSD$
35  */
36 
37 #include "opt_atalk.h"
38 #include "opt_inet.h"
39 #include "opt_inet6.h"
40 #include "opt_ipx.h"
41 #include "opt_bdg.h"
42 #include "opt_mac.h"
43 #include "opt_netgraph.h"
44 
45 #include <sys/param.h>
46 #include <sys/systm.h>
47 #include <sys/kernel.h>
48 #include <sys/mac.h>
49 #include <sys/malloc.h>
50 #include <sys/mbuf.h>
51 #include <sys/random.h>
52 #include <sys/socket.h>
53 #include <sys/sockio.h>
54 #include <sys/sysctl.h>
55 
56 #include <net/if.h>
57 #include <net/netisr.h>
58 #include <net/route.h>
59 #include <net/if_llc.h>
60 #include <net/if_dl.h>
61 #include <net/if_types.h>
62 #include <net/bpf.h>
63 #include <net/ethernet.h>
64 #include <net/bridge.h>
65 
66 #if defined(INET) || defined(INET6)
67 #include <netinet/in.h>
68 #include <netinet/in_var.h>
69 #include <netinet/if_ether.h>
70 #include <netinet/ip_fw.h>
71 #include <netinet/ip_dummynet.h>
72 #endif
73 #ifdef INET6
74 #include <netinet6/nd6.h>
75 #endif
76 
77 #ifdef IPX
78 #include <netipx/ipx.h>
79 #include <netipx/ipx_if.h>
80 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m);
81 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp,
82 		struct sockaddr *dst, short *tp, int *hlen);
83 #endif
84 
85 #ifdef NS
86 #include <netns/ns.h>
87 #include <netns/ns_if.h>
88 ushort ns_nettype;
89 int ether_outputdebug = 0;
90 int ether_inputdebug = 0;
91 #endif
92 
93 #ifdef NETATALK
94 #include <netatalk/at.h>
95 #include <netatalk/at_var.h>
96 #include <netatalk/at_extern.h>
97 
98 #define llc_snap_org_code llc_un.type_snap.org_code
99 #define llc_snap_ether_type llc_un.type_snap.ether_type
100 
101 extern u_char	at_org_code[3];
102 extern u_char	aarp_org_code[3];
103 #endif /* NETATALK */
104 
105 /* netgraph node hooks for ng_ether(4) */
106 void	(*ng_ether_input_p)(struct ifnet *ifp,
107 		struct mbuf **mp, struct ether_header *eh);
108 void	(*ng_ether_input_orphan_p)(struct ifnet *ifp,
109 		struct mbuf *m, struct ether_header *eh);
110 int	(*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
111 void	(*ng_ether_attach_p)(struct ifnet *ifp);
112 void	(*ng_ether_detach_p)(struct ifnet *ifp);
113 
114 int	(*vlan_input_p)(struct ether_header *eh, struct mbuf *m);
115 int	(*vlan_input_tag_p)(struct ether_header *eh, struct mbuf *m,
116 		u_int16_t t);
117 
118 /* bridge support */
119 int do_bridge;
120 bridge_in_t *bridge_in_ptr;
121 bdg_forward_t *bdg_forward_ptr;
122 bdgtakeifaces_t *bdgtakeifaces_ptr;
123 struct bdg_softc *ifp2sc;
124 
125 static	int ether_resolvemulti(struct ifnet *, struct sockaddr **,
126 		struct sockaddr *);
127 u_char	etherbroadcastaddr[6] = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
128 #define senderr(e) do { error = (e); goto bad;} while (0)
129 #define IFP2AC(IFP) ((struct arpcom *)IFP)
130 
131 int
132 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
133 	struct ip_fw **rule, struct ether_header *eh, int shared);
134 static int ether_ipfw;
135 
136 /*
137  * Ethernet output routine.
138  * Encapsulate a packet of type family for the local net.
139  * Use trailer local net encapsulation if enough data in first
140  * packet leaves a multiple of 512 bytes of data in remainder.
141  * Assumes that ifp is actually pointer to arpcom structure.
142  */
143 int
144 ether_output(ifp, m, dst, rt0)
145 	register struct ifnet *ifp;
146 	struct mbuf *m;
147 	struct sockaddr *dst;
148 	struct rtentry *rt0;
149 {
150 	short type;
151 	int error = 0, hdrcmplt = 0;
152  	u_char esrc[6], edst[6];
153 	register struct rtentry *rt;
154 	register struct ether_header *eh;
155 	int loop_copy = 0;
156 	int hlen;	/* link layer header lenght */
157 	struct arpcom *ac = IFP2AC(ifp);
158 
159 #ifdef MAC
160 	error = mac_check_ifnet_transmit(ifp, m);
161 	if (error)
162 		senderr(error);
163 #endif
164 
165 	if (ifp->if_flags & IFF_MONITOR)
166 		senderr(ENETDOWN);
167 	if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
168 		senderr(ENETDOWN);
169 	rt = rt0;
170 	if (rt) {
171 		if ((rt->rt_flags & RTF_UP) == 0) {
172 			rt0 = rt = rtalloc1(dst, 1, 0UL);
173 			if (rt0)
174 				rt->rt_refcnt--;
175 			else
176 				senderr(EHOSTUNREACH);
177 		}
178 		if (rt->rt_flags & RTF_GATEWAY) {
179 			if (rt->rt_gwroute == 0)
180 				goto lookup;
181 			if (((rt = rt->rt_gwroute)->rt_flags & RTF_UP) == 0) {
182 				rtfree(rt); rt = rt0;
183 			lookup: rt->rt_gwroute = rtalloc1(rt->rt_gateway, 1,
184 							  0UL);
185 				if ((rt = rt->rt_gwroute) == 0)
186 					senderr(EHOSTUNREACH);
187 			}
188 		}
189 		if (rt->rt_flags & RTF_REJECT)
190 			if (rt->rt_rmx.rmx_expire == 0 ||
191 			    time_second < rt->rt_rmx.rmx_expire)
192 				senderr(rt == rt0 ? EHOSTDOWN : EHOSTUNREACH);
193 	}
194 	hlen = ETHER_HDR_LEN;
195 	switch (dst->sa_family) {
196 #ifdef INET
197 	case AF_INET:
198 		if (!arpresolve(ifp, rt, m, dst, edst, rt0))
199 			return (0);	/* if not yet resolved */
200 		type = htons(ETHERTYPE_IP);
201 		break;
202 #endif
203 #ifdef INET6
204 	case AF_INET6:
205 		if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, (u_char *)edst)) {
206 			/* Something bad happened */
207 			return(0);
208 		}
209 		type = htons(ETHERTYPE_IPV6);
210 		break;
211 #endif
212 #ifdef IPX
213 	case AF_IPX:
214 		if (ef_outputp) {
215 		    error = ef_outputp(ifp, &m, dst, &type, &hlen);
216 		    if (error)
217 			goto bad;
218 		} else
219 		    type = htons(ETHERTYPE_IPX);
220  		bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
221 		    (caddr_t)edst, sizeof (edst));
222 		break;
223 #endif
224 #ifdef NETATALK
225 	case AF_APPLETALK:
226 	  {
227 	    struct at_ifaddr *aa;
228 
229 	    if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) {
230 		    goto bad;
231 	    }
232 	    if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst))
233 		    return (0);
234 	    /*
235 	     * In the phase 2 case, need to prepend an mbuf for the llc header.
236 	     * Since we must preserve the value of m, which is passed to us by
237 	     * value, we m_copy() the first mbuf, and use it for our llc header.
238 	     */
239 	    if ( aa->aa_flags & AFA_PHASE2 ) {
240 		struct llc llc;
241 
242 		M_PREPEND(m, sizeof(struct llc), M_TRYWAIT);
243 		llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
244 		llc.llc_control = LLC_UI;
245 		bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code));
246 		llc.llc_snap_ether_type = htons( ETHERTYPE_AT );
247 		bcopy(&llc, mtod(m, caddr_t), sizeof(struct llc));
248 		type = htons(m->m_pkthdr.len);
249 		hlen = sizeof(struct llc) + ETHER_HDR_LEN;
250 	    } else {
251 		type = htons(ETHERTYPE_AT);
252 	    }
253 	    break;
254 	  }
255 #endif /* NETATALK */
256 #ifdef NS
257 	case AF_NS:
258 		switch(ns_nettype){
259 		default:
260 		case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
261 			type = 0x8137;
262 			break;
263 		case 0x0: /* Novell 802.3 */
264 			type = htons( m->m_pkthdr.len);
265 			break;
266 		case 0xe0e0: /* Novell 802.2 and Token-Ring */
267 			M_PREPEND(m, 3, M_TRYWAIT);
268 			type = htons( m->m_pkthdr.len);
269 			cp = mtod(m, u_char *);
270 			*cp++ = 0xE0;
271 			*cp++ = 0xE0;
272 			*cp++ = 0x03;
273 			break;
274 		}
275  		bcopy((caddr_t)&(((struct sockaddr_ns *)dst)->sns_addr.x_host),
276 		    (caddr_t)edst, sizeof (edst));
277 		/*
278 		 * XXX if ns_thishost is the same as the node's ethernet
279 		 * address then just the default code will catch this anyhow.
280 		 * So I'm not sure if this next clause should be here at all?
281 		 * [JRE]
282 		 */
283 		if (!bcmp((caddr_t)edst, (caddr_t)&ns_thishost, sizeof(edst))){
284 			m->m_pkthdr.rcvif = ifp;
285 			inq = &nsintrq;
286 			if (IF_HANDOFF(inq, m, NULL))
287 				schednetisr(NETISR_NS);
288 			return (error);
289 		}
290 		if (!bcmp((caddr_t)edst, (caddr_t)&ns_broadhost, sizeof(edst))){
291 			m->m_flags |= M_BCAST;
292 		}
293 		break;
294 #endif /* NS */
295 
296 	case pseudo_AF_HDRCMPLT:
297 		hdrcmplt = 1;
298 		eh = (struct ether_header *)dst->sa_data;
299 		(void)memcpy(esrc, eh->ether_shost, sizeof (esrc));
300 		/* FALLTHROUGH */
301 
302 	case AF_UNSPEC:
303 		loop_copy = -1; /* if this is for us, don't do it */
304 		eh = (struct ether_header *)dst->sa_data;
305  		(void)memcpy(edst, eh->ether_dhost, sizeof (edst));
306 		type = eh->ether_type;
307 		break;
308 
309 	default:
310 		if_printf(ifp, "can't handle af%d\n", dst->sa_family);
311 		senderr(EAFNOSUPPORT);
312 	}
313 
314 	/*
315 	 * Add local net header.  If no space in first mbuf,
316 	 * allocate another.
317 	 */
318 	M_PREPEND(m, sizeof (struct ether_header), M_DONTWAIT);
319 	if (m == 0)
320 		senderr(ENOBUFS);
321 	eh = mtod(m, struct ether_header *);
322 	(void)memcpy(&eh->ether_type, &type,
323 		sizeof(eh->ether_type));
324  	(void)memcpy(eh->ether_dhost, edst, sizeof (edst));
325 	if (hdrcmplt)
326 		(void)memcpy(eh->ether_shost, esrc,
327 			sizeof(eh->ether_shost));
328 	else
329 		(void)memcpy(eh->ether_shost, ac->ac_enaddr,
330 			sizeof(eh->ether_shost));
331 
332 	/*
333 	 * If a simplex interface, and the packet is being sent to our
334 	 * Ethernet address or a broadcast address, loopback a copy.
335 	 * XXX To make a simplex device behave exactly like a duplex
336 	 * device, we should copy in the case of sending to our own
337 	 * ethernet address (thus letting the original actually appear
338 	 * on the wire). However, we don't do that here for security
339 	 * reasons and compatibility with the original behavior.
340 	 */
341 	if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) {
342 		int csum_flags = 0;
343 
344 		if (m->m_pkthdr.csum_flags & CSUM_IP)
345 			csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
346 		if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
347 			csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
348 		if ((m->m_flags & M_BCAST) || (loop_copy > 0)) {
349 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
350 
351 			n->m_pkthdr.csum_flags |= csum_flags;
352 			if (csum_flags & CSUM_DATA_VALID)
353 				n->m_pkthdr.csum_data = 0xffff;
354 
355 			(void) if_simloop(ifp, n, dst->sa_family, hlen);
356 		} else if (bcmp(eh->ether_dhost,
357 		    eh->ether_shost, ETHER_ADDR_LEN) == 0) {
358 			m->m_pkthdr.csum_flags |= csum_flags;
359 			if (csum_flags & CSUM_DATA_VALID)
360 				m->m_pkthdr.csum_data = 0xffff;
361 			(void) if_simloop(ifp, m, dst->sa_family, hlen);
362 			return (0);	/* XXX */
363 		}
364 	}
365 
366 	/* Handle ng_ether(4) processing, if any */
367 	if (ng_ether_output_p != NULL) {
368 		if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
369 bad:			if (m != NULL)
370 				m_freem(m);
371 			return (error);
372 		}
373 		if (m == NULL)
374 			return (0);
375 	}
376 
377 	/* Continue with link-layer output */
378 	return ether_output_frame(ifp, m);
379 }
380 
381 /*
382  * Ethernet link layer output routine to send a raw frame to the device.
383  *
384  * This assumes that the 14 byte Ethernet header is present and contiguous
385  * in the first mbuf (if BRIDGE'ing).
386  */
387 int
388 ether_output_frame(ifp, m)
389 	struct ifnet *ifp;
390 	struct mbuf *m;
391 {
392 	int error = 0;
393 	struct ip_fw *rule = NULL;
394 
395 	/* Extract info from dummynet tag, ignore others */
396 	for (; m->m_type == MT_TAG; m = m->m_next)
397 		if (m->m_flags == PACKET_TAG_DUMMYNET)
398 			rule = ((struct dn_pkt *)m)->rule;
399 
400 	if (rule)	/* packet was already bridged */
401 		goto no_bridge;
402 
403 	if (BDG_ACTIVE(ifp) ) {
404 		struct ether_header *eh; /* a ptr suffices */
405 
406 		m->m_pkthdr.rcvif = NULL;
407 		eh = mtod(m, struct ether_header *);
408 		m_adj(m, ETHER_HDR_LEN);
409 		m = bdg_forward_ptr(m, eh, ifp);
410 		if (m != NULL)
411 			m_freem(m);
412 		return (0);
413 	}
414 
415 no_bridge:
416 	if (IPFW_LOADED && ether_ipfw != 0) {
417 		struct ether_header save_eh, *eh;
418 
419 		eh = mtod(m, struct ether_header *);
420 		save_eh = *eh;
421 		m_adj(m, ETHER_HDR_LEN);
422 		if (ether_ipfw_chk(&m, ifp, &rule, eh, 0) == 0) {
423 			if (m) {
424 				m_freem(m);
425 				return ENOBUFS;	/* pkt dropped */
426 			} else
427 				return 0;	/* consumed e.g. in a pipe */
428 		}
429 		/* packet was ok, restore the ethernet header */
430 		if ( (void *)(eh + 1) == (void *)m->m_data) {
431 			m->m_data -= ETHER_HDR_LEN ;
432 			m->m_len += ETHER_HDR_LEN ;
433 			m->m_pkthdr.len += ETHER_HDR_LEN ;
434 		} else {
435 			M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
436 			if (m == NULL) /* nope... */
437 				return ENOBUFS;
438 			bcopy(&save_eh, mtod(m, struct ether_header *),
439 			    ETHER_HDR_LEN);
440 		}
441 	}
442 
443 	/*
444 	 * Queue message on interface, update output statistics if
445 	 * successful, and start output if interface not yet active.
446 	 */
447 	if (! IF_HANDOFF(&ifp->if_snd, m, ifp))
448 		return (ENOBUFS);
449 	return (error);
450 }
451 
452 /*
453  * ipfw processing for ethernet packets (in and out).
454  * The second parameter is NULL from ether_demux, and ifp from
455  * ether_output_frame. This section of code could be used from
456  * bridge.c as well as long as we use some extra info
457  * to distinguish that case from ether_output_frame();
458  */
459 int
460 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
461 	struct ip_fw **rule, struct ether_header *eh, int shared)
462 {
463 	struct ether_header save_eh = *eh;	/* might be a ptr in m */
464 	int i;
465 	struct ip_fw_args args;
466 
467 	if (*rule != NULL /*&& fw_one_pass*/)	/* HACK! need to obey fw_one_pass */
468 		return 1; /* dummynet packet, already partially processed */
469 
470 	/*
471 	 * I need some amt of data to be contiguous, and in case others need
472 	 * the packet (shared==1) also better be in the first mbuf.
473 	 */
474 	i = min( (*m0)->m_pkthdr.len, max_protohdr);
475 	if ( shared || (*m0)->m_len < i) {
476 		*m0 = m_pullup(*m0, i);
477 		if (*m0 == NULL)
478 			return 0;
479 	}
480 
481 	args.m = *m0;		/* the packet we are looking at		*/
482 	args.oif = dst;		/* destination, if any			*/
483 	args.divert_rule = 0;	/* we do not support divert yet		*/
484 	args.rule = *rule;	/* matching rule to restart		*/
485 	args.next_hop = NULL;	/* we do not support forward yet	*/
486 	args.eh = &save_eh;	/* MAC header for bridged/MAC packets	*/
487 	i = ip_fw_chk_ptr(&args);
488 	*m0 = args.m;
489 	*rule = args.rule;
490 
491 	if ( (i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL) /* drop */
492 		return 0;
493 
494 	if (i == 0) /* a PASS rule.  */
495 		return 1;
496 
497 	if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG)) {
498 		/*
499 		 * Pass the pkt to dummynet, which consumes it.
500 		 * If shared, make a copy and keep the original.
501 		 */
502 		struct mbuf *m ;
503 
504 		if (shared) {
505 			m = m_copypacket(*m0, M_DONTWAIT);
506 			if (m == NULL)
507 				return 0;
508 		} else {
509 			m = *m0 ; /* pass the original to dummynet */
510 			*m0 = NULL ; /* and nothing back to the caller */
511 		}
512 		/*
513 		 * Prepend the header, optimize for the common case of
514 		 * eh pointing into the mbuf.
515 		 */
516 		if ( (void *)(eh + 1) == (void *)m->m_data) {
517 			m->m_data -= ETHER_HDR_LEN ;
518 			m->m_len += ETHER_HDR_LEN ;
519 			m->m_pkthdr.len += ETHER_HDR_LEN ;
520 		} else {
521 			M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
522 			if (m == NULL) /* nope... */
523 				return 0;
524 			bcopy(&save_eh, mtod(m, struct ether_header *),
525 			    ETHER_HDR_LEN);
526 		}
527 		ip_dn_io_ptr(m, (i & 0xffff),
528 			dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
529 		return 0;
530 	}
531 	/*
532 	 * XXX at some point add support for divert/forward actions.
533 	 * If none of the above matches, we have to drop the pkt.
534 	 */
535 	return 0;
536 }
537 
538 /*
539  * Process a received Ethernet packet. We have two different interfaces:
540  * one (conventional) assumes the packet in the mbuf, with the ethernet
541  * header provided separately in *eh. The second one (new) has everything
542  * in the mbuf, and we can tell it because eh == NULL.
543  * The caller MUST MAKE SURE that there are at least
544  * sizeof(struct ether_header) bytes in the first mbuf.
545  *
546  * This allows us to concentrate in one place a bunch of code which
547  * is replicated in all device drivers. Also, many functions called
548  * from ether_input() try to put the eh back into the mbuf, so we
549  * can later propagate the 'contiguous packet' interface to them,
550  * and handle the old interface just here.
551  *
552  * NOTA BENE: for many drivers "eh" is a pointer into the first mbuf or
553  * cluster, right before m_data. So be very careful when working on m,
554  * as you could destroy *eh !!
555  *
556  * First we perform any link layer operations, then continue
557  * to the upper layers with ether_demux().
558  */
559 void
560 ether_input(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m)
561 {
562 	struct ether_header save_eh;
563 
564 	if (eh == NULL) {
565 		if (m->m_len < sizeof(struct ether_header)) {
566 			/* XXX error in the caller. */
567 			m_freem(m);
568 			return;
569 		}
570 		if (ifp->if_bpf != NULL)
571 			bpf_mtap(ifp, m);
572 		m->m_pkthdr.rcvif = ifp;
573 		eh = mtod(m, struct ether_header *);
574 		m->m_data += sizeof(struct ether_header);
575 		m->m_len -= sizeof(struct ether_header);
576 		m->m_pkthdr.len = m->m_len;
577 	} else if (ifp->if_bpf != NULL) {
578 		struct m_hdr mh;
579 
580 		/* This kludge is OK; BPF treats the "mbuf" as read-only */
581 		mh.mh_next = m;
582 		mh.mh_data = (char *)eh;
583 		mh.mh_len = ETHER_HDR_LEN;
584 		bpf_mtap(ifp, (struct mbuf *)&mh);
585 	}
586 
587 	if (ifp->if_flags & IFF_MONITOR) {
588 		m_freem(m);
589 		return;
590 	}
591 
592 #ifdef MAC
593 	mac_create_mbuf_from_ifnet(ifp, m);
594 #endif
595 
596 	ifp->if_ibytes += m->m_pkthdr.len + sizeof (*eh);
597 
598 	/* Handle ng_ether(4) processing, if any */
599 	if (ng_ether_input_p != NULL) {
600 		(*ng_ether_input_p)(ifp, &m, eh);
601 		if (m == NULL)
602 			return;
603 	}
604 
605 	/* Check for bridging mode */
606 	if (BDG_ACTIVE(ifp) ) {
607 		struct ifnet *bif;
608 
609 		/* Check with bridging code */
610 		if ((bif = bridge_in_ptr(ifp, eh)) == BDG_DROP) {
611 			m_freem(m);
612 			return;
613 		}
614 		if (bif != BDG_LOCAL) {
615 			save_eh = *eh ; /* because it might change */
616 			m = bdg_forward_ptr(m, eh, bif); /* needs forwarding */
617 			/*
618 			 * Do not continue if bdg_forward_ptr() processed our
619 			 * packet (and cleared the mbuf pointer m) or if
620 			 * it dropped (m_free'd) the packet itself.
621 			 */
622 			if (m == NULL) {
623 			    if (bif == BDG_BCAST || bif == BDG_MCAST)
624 				printf("bdg_forward drop MULTICAST PKT\n");
625 			    return;
626 			}
627 			eh = &save_eh ;
628 		}
629 		if (bif == BDG_LOCAL
630 		    || bif == BDG_BCAST
631 		    || bif == BDG_MCAST)
632 			goto recvLocal;			/* receive locally */
633 
634 		/* If not local and not multicast, just drop it */
635 		if (m != NULL)
636 			m_freem(m);
637 		return;
638 	}
639 
640 recvLocal:
641 	/* Continue with upper layer processing */
642 	ether_demux(ifp, eh, m);
643 	/* First chunk of an mbuf contains good entropy */
644 	if (harvest.ethernet)
645 		random_harvest(m, 16, 3, 0, RANDOM_NET);
646 }
647 
648 /*
649  * Upper layer processing for a received Ethernet packet.
650  */
651 void
652 ether_demux(ifp, eh, m)
653 	struct ifnet *ifp;
654 	struct ether_header *eh;
655 	struct mbuf *m;
656 {
657 	struct ifqueue *inq;
658 	u_short ether_type;
659 #if defined(NETATALK)
660 	register struct llc *l;
661 #endif
662 	struct ip_fw *rule = NULL;
663 
664 	/* Extract info from dummynet tag, ignore others */
665 	for (;m->m_type == MT_TAG; m = m->m_next)
666 		if (m->m_flags == PACKET_TAG_DUMMYNET) {
667 			rule = ((struct dn_pkt *)m)->rule;
668 			ifp = m->m_next->m_pkthdr.rcvif;
669 		}
670 
671 	if (rule)	/* packet was already bridged */
672 		goto post_stats;
673 
674 	if (!(BDG_ACTIVE(ifp))) {
675 		/*
676 		 * Discard packet if upper layers shouldn't see it because it
677 		 * was unicast to a different Ethernet address. If the driver
678 		 * is working properly, then this situation can only happen
679 		 * when the interface is in promiscuous mode.
680 		 */
681 		if ((ifp->if_flags & IFF_PROMISC) != 0
682 		    && (eh->ether_dhost[0] & 1) == 0
683 		    && bcmp(eh->ether_dhost,
684 		      IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN) != 0
685 		    && (ifp->if_flags & IFF_PPROMISC) == 0) {
686 			    m_freem(m);
687 			    return;
688 		}
689 	}
690 
691 	/* Discard packet if interface is not up */
692 	if ((ifp->if_flags & IFF_UP) == 0) {
693 		m_freem(m);
694 		return;
695 	}
696 	if (eh->ether_dhost[0] & 1) {
697 		if (bcmp((caddr_t)etherbroadcastaddr, (caddr_t)eh->ether_dhost,
698 			 sizeof(etherbroadcastaddr)) == 0)
699 			m->m_flags |= M_BCAST;
700 		else
701 			m->m_flags |= M_MCAST;
702 	}
703 	if (m->m_flags & (M_BCAST|M_MCAST))
704 		ifp->if_imcasts++;
705 
706 post_stats:
707 	if (IPFW_LOADED && ether_ipfw != 0) {
708 		if (ether_ipfw_chk(&m, NULL, &rule, eh, 0 ) == 0) {
709 			if (m)
710 				m_freem(m);
711 			return;
712 		}
713 	}
714 
715 	ether_type = ntohs(eh->ether_type);
716 
717 	switch (ether_type) {
718 #ifdef INET
719 	case ETHERTYPE_IP:
720 		if (ipflow_fastforward(m))
721 			return;
722 		schednetisr(NETISR_IP);
723 		inq = &ipintrq;
724 		break;
725 
726 	case ETHERTYPE_ARP:
727 		if (ifp->if_flags & IFF_NOARP) {
728 			/* Discard packet if ARP is disabled on interface */
729 			m_freem(m);
730 			return;
731 		}
732 		schednetisr(NETISR_ARP);
733 		inq = &arpintrq;
734 		break;
735 #endif
736 #ifdef IPX
737 	case ETHERTYPE_IPX:
738 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
739 			return;
740 		schednetisr(NETISR_IPX);
741 		inq = &ipxintrq;
742 		break;
743 #endif
744 #ifdef INET6
745 	case ETHERTYPE_IPV6:
746 		schednetisr(NETISR_IPV6);
747 		inq = &ip6intrq;
748 		break;
749 #endif
750 #ifdef NS
751 	case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
752 		schednetisr(NETISR_NS);
753 		inq = &nsintrq;
754 		break;
755 
756 #endif /* NS */
757 #ifdef NETATALK
758         case ETHERTYPE_AT:
759                 schednetisr(NETISR_ATALK);
760                 inq = &atintrq1;
761                 break;
762         case ETHERTYPE_AARP:
763 		/* probably this should be done with a NETISR as well */
764                 aarpinput(IFP2AC(ifp), m); /* XXX */
765                 return;
766 #endif /* NETATALK */
767 	case ETHERTYPE_VLAN:
768 		/* XXX lock ? */
769 		if (vlan_input_p != NULL)
770 			(*vlan_input_p)(eh, m);
771 		else {
772 			m->m_pkthdr.rcvif->if_noproto++;
773 			m_freem(m);
774 		}
775 		/* XXX unlock ? */
776 		return;
777 	default:
778 #ifdef IPX
779 		if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
780 			return;
781 #endif /* IPX */
782 #ifdef NS
783 		checksum = mtod(m, ushort *);
784 		/* Novell 802.3 */
785 		if ((ether_type <= ETHERMTU) &&
786 			((*checksum == 0xffff) || (*checksum == 0xE0E0))){
787 			if(*checksum == 0xE0E0) {
788 				m->m_pkthdr.len -= 3;
789 				m->m_len -= 3;
790 				m->m_data += 3;
791 			}
792 				schednetisr(NETISR_NS);
793 				inq = &nsintrq;
794 				break;
795 		}
796 #endif /* NS */
797 #if defined(NETATALK)
798 		if (ether_type > ETHERMTU)
799 			goto dropanyway;
800 		l = mtod(m, struct llc *);
801 		switch (l->llc_dsap) {
802 		case LLC_SNAP_LSAP:
803 		    switch (l->llc_control) {
804 		    case LLC_UI:
805 			if (l->llc_ssap != LLC_SNAP_LSAP)
806 			    goto dropanyway;
807 
808 			if (Bcmp(&(l->llc_snap_org_code)[0], at_org_code,
809 				   sizeof(at_org_code)) == 0 &&
810 			     ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
811 			    inq = &atintrq2;
812 			    m_adj( m, sizeof( struct llc ));
813 			    schednetisr(NETISR_ATALK);
814 			    break;
815 			}
816 
817 			if (Bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
818 				   sizeof(aarp_org_code)) == 0 &&
819 			     ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
820 			    m_adj( m, sizeof( struct llc ));
821 			    aarpinput(IFP2AC(ifp), m); /* XXX */
822 			    return;
823 			}
824 
825 		    default:
826 			goto dropanyway;
827 		    }
828 		    break;
829 		dropanyway:
830 		default:
831 			if (ng_ether_input_orphan_p != NULL)
832 				(*ng_ether_input_orphan_p)(ifp, m, eh);
833 			else
834 				m_freem(m);
835 			return;
836 		}
837 #else /* NETATALK */
838 		if (ng_ether_input_orphan_p != NULL)
839 			(*ng_ether_input_orphan_p)(ifp, m, eh);
840 		else
841 			m_freem(m);
842 		return;
843 #endif /* NETATALK */
844 	}
845 
846 	(void) IF_HANDOFF(inq, m, NULL);
847 }
848 
849 /*
850  * Perform common duties while attaching to interface list
851  */
852 void
853 ether_ifattach(ifp, bpf)
854 	register struct ifnet *ifp;
855 	int bpf;
856 {
857 	register struct ifaddr *ifa;
858 	register struct sockaddr_dl *sdl;
859 
860 	ifp->if_type = IFT_ETHER;
861 	ifp->if_addrlen = 6;
862 	ifp->if_hdrlen = 14;
863 	if_attach(ifp);
864 	ifp->if_mtu = ETHERMTU;
865 	ifp->if_resolvemulti = ether_resolvemulti;
866 	if (ifp->if_baudrate == 0)
867 	    ifp->if_baudrate = 10000000;
868 	ifp->if_broadcastaddr = etherbroadcastaddr;
869 	ifa = ifaddr_byindex(ifp->if_index);
870 	KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
871 	sdl = (struct sockaddr_dl *)ifa->ifa_addr;
872 	sdl->sdl_type = IFT_ETHER;
873 	sdl->sdl_alen = ifp->if_addrlen;
874 	bcopy((IFP2AC(ifp))->ac_enaddr, LLADDR(sdl), ifp->if_addrlen);
875 	if (bpf)
876 		bpfattach(ifp, DLT_EN10MB, sizeof(struct ether_header));
877 	if (ng_ether_attach_p != NULL)
878 		(*ng_ether_attach_p)(ifp);
879 	if (BDG_LOADED)
880 		bdgtakeifaces_ptr();
881 }
882 
883 /*
884  * Perform common duties while detaching an Ethernet interface
885  */
886 void
887 ether_ifdetach(ifp, bpf)
888 	struct ifnet *ifp;
889 	int bpf;
890 {
891 	if (ng_ether_detach_p != NULL)
892 		(*ng_ether_detach_p)(ifp);
893 	if (bpf)
894 		bpfdetach(ifp);
895 	if_detach(ifp);
896 	if (BDG_LOADED)
897 		bdgtakeifaces_ptr();
898 }
899 
900 SYSCTL_DECL(_net_link);
901 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
902 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
903 	    &ether_ipfw,0,"Pass ether pkts through firewall");
904 
905 int
906 ether_ioctl(ifp, command, data)
907 	struct ifnet *ifp;
908 	int command;
909 	caddr_t data;
910 {
911 	struct ifaddr *ifa = (struct ifaddr *) data;
912 	struct ifreq *ifr = (struct ifreq *) data;
913 	int error = 0;
914 
915 	switch (command) {
916 	case SIOCSIFADDR:
917 		ifp->if_flags |= IFF_UP;
918 
919 		switch (ifa->ifa_addr->sa_family) {
920 #ifdef INET
921 		case AF_INET:
922 			ifp->if_init(ifp->if_softc);	/* before arpwhohas */
923 			arp_ifinit(ifp, ifa);
924 			break;
925 #endif
926 #ifdef IPX
927 		/*
928 		 * XXX - This code is probably wrong
929 		 */
930 		case AF_IPX:
931 			{
932 			register struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr);
933 			struct arpcom *ac = IFP2AC(ifp);
934 
935 			if (ipx_nullhost(*ina))
936 				ina->x_host =
937 				    *(union ipx_host *)
938 			            ac->ac_enaddr;
939 			else {
940 				bcopy((caddr_t) ina->x_host.c_host,
941 				      (caddr_t) ac->ac_enaddr,
942 				      sizeof(ac->ac_enaddr));
943 			}
944 
945 			/*
946 			 * Set new address
947 			 */
948 			ifp->if_init(ifp->if_softc);
949 			break;
950 			}
951 #endif
952 #ifdef NS
953 		/*
954 		 * XXX - This code is probably wrong
955 		 */
956 		case AF_NS:
957 		{
958 			register struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr);
959 			struct arpcom *ac = IFP2AC(ifp);
960 
961 			if (ns_nullhost(*ina))
962 				ina->x_host =
963 				    *(union ns_host *) (ac->ac_enaddr);
964 			else {
965 				bcopy((caddr_t) ina->x_host.c_host,
966 				      (caddr_t) ac->ac_enaddr,
967 				      sizeof(ac->ac_enaddr));
968 			}
969 
970 			/*
971 			 * Set new address
972 			 */
973 			ifp->if_init(ifp->if_softc);
974 			break;
975 		}
976 #endif
977 		default:
978 			ifp->if_init(ifp->if_softc);
979 			break;
980 		}
981 		break;
982 
983 	case SIOCGIFADDR:
984 		{
985 			struct sockaddr *sa;
986 
987 			sa = (struct sockaddr *) & ifr->ifr_data;
988 			bcopy(IFP2AC(ifp)->ac_enaddr,
989 			      (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
990 		}
991 		break;
992 
993 	case SIOCSIFMTU:
994 		/*
995 		 * Set the interface MTU.
996 		 */
997 		if (ifr->ifr_mtu > ETHERMTU) {
998 			error = EINVAL;
999 		} else {
1000 			ifp->if_mtu = ifr->ifr_mtu;
1001 		}
1002 		break;
1003 	}
1004 	return (error);
1005 }
1006 
1007 static int
1008 ether_resolvemulti(ifp, llsa, sa)
1009 	struct ifnet *ifp;
1010 	struct sockaddr **llsa;
1011 	struct sockaddr *sa;
1012 {
1013 	struct sockaddr_dl *sdl;
1014 	struct sockaddr_in *sin;
1015 #ifdef INET6
1016 	struct sockaddr_in6 *sin6;
1017 #endif
1018 	u_char *e_addr;
1019 
1020 	switch(sa->sa_family) {
1021 	case AF_LINK:
1022 		/*
1023 		 * No mapping needed. Just check that it's a valid MC address.
1024 		 */
1025 		sdl = (struct sockaddr_dl *)sa;
1026 		e_addr = LLADDR(sdl);
1027 		if ((e_addr[0] & 1) != 1)
1028 			return EADDRNOTAVAIL;
1029 		*llsa = 0;
1030 		return 0;
1031 
1032 #ifdef INET
1033 	case AF_INET:
1034 		sin = (struct sockaddr_in *)sa;
1035 		if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1036 			return EADDRNOTAVAIL;
1037 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1038 		       M_WAITOK|M_ZERO);
1039 		sdl->sdl_len = sizeof *sdl;
1040 		sdl->sdl_family = AF_LINK;
1041 		sdl->sdl_index = ifp->if_index;
1042 		sdl->sdl_type = IFT_ETHER;
1043 		sdl->sdl_alen = ETHER_ADDR_LEN;
1044 		e_addr = LLADDR(sdl);
1045 		ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1046 		*llsa = (struct sockaddr *)sdl;
1047 		return 0;
1048 #endif
1049 #ifdef INET6
1050 	case AF_INET6:
1051 		sin6 = (struct sockaddr_in6 *)sa;
1052 		if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1053 			/*
1054 			 * An IP6 address of 0 means listen to all
1055 			 * of the Ethernet multicast address used for IP6.
1056 			 * (This is used for multicast routers.)
1057 			 */
1058 			ifp->if_flags |= IFF_ALLMULTI;
1059 			*llsa = 0;
1060 			return 0;
1061 		}
1062 		if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1063 			return EADDRNOTAVAIL;
1064 		MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1065 		       M_WAITOK|M_ZERO);
1066 		sdl->sdl_len = sizeof *sdl;
1067 		sdl->sdl_family = AF_LINK;
1068 		sdl->sdl_index = ifp->if_index;
1069 		sdl->sdl_type = IFT_ETHER;
1070 		sdl->sdl_alen = ETHER_ADDR_LEN;
1071 		e_addr = LLADDR(sdl);
1072 		ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1073 		*llsa = (struct sockaddr *)sdl;
1074 		return 0;
1075 #endif
1076 
1077 	default:
1078 		/*
1079 		 * Well, the text isn't quite right, but it's the name
1080 		 * that counts...
1081 		 */
1082 		return EAFNOSUPPORT;
1083 	}
1084 }
1085