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: releng/5.4/sys/net/if_ethersubr.c 145335 2005-04-20 19:11:07Z cvs2svn $
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 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] =
118 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
119
120 static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
121 struct sockaddr *);
122
123 #define senderr(e) do { error = (e); goto bad;} while (0)
124
125 #if defined(INET) || defined(INET6)
126 int
127 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
128 struct ip_fw **rule, int shared);
129 static int ether_ipfw;
130 #endif
131
132 /*
133 * Ethernet output routine.
134 * Encapsulate a packet of type family for the local net.
135 * Use trailer local net encapsulation if enough data in first
136 * packet leaves a multiple of 512 bytes of data in remainder.
137 * Assumes that ifp is actually pointer to arpcom structure.
138 */
139 int
140 ether_output(struct ifnet *ifp, struct mbuf *m,
141 struct sockaddr *dst, struct rtentry *rt0)
142 {
143 short type;
144 int error, hdrcmplt = 0;
145 u_char esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN];
146 struct ether_header *eh;
147 int loop_copy = 0;
148 int hlen; /* link layer header length */
149
150 #ifdef MAC
151 error = mac_check_ifnet_transmit(ifp, m);
152 if (error)
153 senderr(error);
154 #endif
155
156 if (ifp->if_flags & IFF_MONITOR)
157 senderr(ENETDOWN);
158 if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
159 senderr(ENETDOWN);
160
161 hlen = ETHER_HDR_LEN;
162 switch (dst->sa_family) {
163 #ifdef INET
164 case AF_INET:
165 error = arpresolve(ifp, rt0, m, dst, edst);
166 if (error)
167 return (error == EWOULDBLOCK ? 0 : error);
168 type = htons(ETHERTYPE_IP);
169 break;
170 case AF_ARP:
171 {
172 struct arphdr *ah;
173 ah = mtod(m, struct arphdr *);
174 ah->ar_hrd = htons(ARPHRD_ETHER);
175
176 loop_copy = -1; /* if this is for us, don't do it */
177
178 switch(ntohs(ah->ar_op)) {
179 case ARPOP_REVREQUEST:
180 case ARPOP_REVREPLY:
181 type = htons(ETHERTYPE_REVARP);
182 break;
183 case ARPOP_REQUEST:
184 case ARPOP_REPLY:
185 default:
186 type = htons(ETHERTYPE_ARP);
187 break;
188 }
189
190 if (m->m_flags & M_BCAST)
191 bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN);
192 else
193 bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN);
194
195 }
196 break;
197 #endif
198 #ifdef INET6
199 case AF_INET6:
200 error = nd6_storelladdr(ifp, rt0, m, dst, (u_char *)edst);
201 if (error)
202 return error;
203 type = htons(ETHERTYPE_IPV6);
204 break;
205 #endif
206 #ifdef IPX
207 case AF_IPX:
208 if (ef_outputp) {
209 error = ef_outputp(ifp, &m, dst, &type, &hlen);
210 if (error)
211 goto bad;
212 } else
213 type = htons(ETHERTYPE_IPX);
214 bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
215 (caddr_t)edst, sizeof (edst));
216 break;
217 #endif
218 #ifdef NETATALK
219 case AF_APPLETALK:
220 {
221 struct at_ifaddr *aa;
222
223 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL)
224 senderr(EHOSTUNREACH); /* XXX */
225 if (!aarpresolve(ifp, m, (struct sockaddr_at *)dst, edst))
226 return (0);
227 /*
228 * In the phase 2 case, need to prepend an mbuf for the llc header.
229 */
230 if ( aa->aa_flags & AFA_PHASE2 ) {
231 struct llc llc;
232
233 M_PREPEND(m, LLC_SNAPFRAMELEN, M_DONTWAIT);
234 if (m == NULL)
235 senderr(ENOBUFS);
236 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
237 llc.llc_control = LLC_UI;
238 bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code));
239 llc.llc_snap_ether_type = htons( ETHERTYPE_AT );
240 bcopy(&llc, mtod(m, caddr_t), LLC_SNAPFRAMELEN);
241 type = htons(m->m_pkthdr.len);
242 hlen = LLC_SNAPFRAMELEN + ETHER_HDR_LEN;
243 } else {
244 type = htons(ETHERTYPE_AT);
245 }
246 break;
247 }
248 #endif /* NETATALK */
249
250 case pseudo_AF_HDRCMPLT:
251 hdrcmplt = 1;
252 eh = (struct ether_header *)dst->sa_data;
253 (void)memcpy(esrc, eh->ether_shost, sizeof (esrc));
254 /* FALLTHROUGH */
255
256 case AF_UNSPEC:
257 loop_copy = -1; /* if this is for us, don't do it */
258 eh = (struct ether_header *)dst->sa_data;
259 (void)memcpy(edst, eh->ether_dhost, sizeof (edst));
260 type = eh->ether_type;
261 break;
262
263 default:
264 if_printf(ifp, "can't handle af%d\n", dst->sa_family);
265 senderr(EAFNOSUPPORT);
266 }
267
268 /*
269 * Add local net header. If no space in first mbuf,
270 * allocate another.
271 */
272 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
273 if (m == NULL)
274 senderr(ENOBUFS);
275 eh = mtod(m, struct ether_header *);
276 (void)memcpy(&eh->ether_type, &type,
277 sizeof(eh->ether_type));
278 (void)memcpy(eh->ether_dhost, edst, sizeof (edst));
279 if (hdrcmplt)
280 (void)memcpy(eh->ether_shost, esrc,
281 sizeof(eh->ether_shost));
282 else
283 (void)memcpy(eh->ether_shost, IFP2AC(ifp)->ac_enaddr,
284 sizeof(eh->ether_shost));
285
286 /*
287 * If a simplex interface, and the packet is being sent to our
288 * Ethernet address or a broadcast address, loopback a copy.
289 * XXX To make a simplex device behave exactly like a duplex
290 * device, we should copy in the case of sending to our own
291 * ethernet address (thus letting the original actually appear
292 * on the wire). However, we don't do that here for security
293 * reasons and compatibility with the original behavior.
294 */
295 if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) {
296 int csum_flags = 0;
297
298 if (m->m_pkthdr.csum_flags & CSUM_IP)
299 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
300 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
301 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
302
303 if ((m->m_flags & M_BCAST) || (loop_copy > 0)) {
304 struct mbuf *n;
305
306 if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) {
307 n->m_pkthdr.csum_flags |= csum_flags;
308 if (csum_flags & CSUM_DATA_VALID)
309 n->m_pkthdr.csum_data = 0xffff;
310 (void)if_simloop(ifp, n, dst->sa_family, hlen);
311 } else
312 ifp->if_iqdrops++;
313 } else if (bcmp(eh->ether_dhost, eh->ether_shost,
314 ETHER_ADDR_LEN) == 0) {
315 m->m_pkthdr.csum_flags |= csum_flags;
316 if (csum_flags & CSUM_DATA_VALID)
317 m->m_pkthdr.csum_data = 0xffff;
318 (void) if_simloop(ifp, m, dst->sa_family, hlen);
319 return (0); /* XXX */
320 }
321 }
322
323 #ifdef DEV_CARP
324 if (ifp->if_carp &&
325 (error = carp_output(ifp, m, dst, NULL)))
326 goto bad;
327 #endif
328
329 /* Handle ng_ether(4) processing, if any */
330 if (ng_ether_output_p != NULL) {
331 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
332 bad: if (m != NULL)
333 m_freem(m);
334 return (error);
335 }
336 if (m == NULL)
337 return (0);
338 }
339
340 /* Continue with link-layer output */
341 return ether_output_frame(ifp, m);
342 }
343
344 /*
345 * Ethernet link layer output routine to send a raw frame to the device.
346 *
347 * This assumes that the 14 byte Ethernet header is present and contiguous
348 * in the first mbuf (if BRIDGE'ing).
349 */
350 int
351 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
352 {
353 #if defined(INET) || defined(INET6)
354 struct ip_fw *rule = ip_dn_claim_rule(m);
355 #else
356 void *rule = NULL;
357 #endif
358 int error;
359
360 if (rule == NULL && BDG_ACTIVE(ifp)) {
361 /*
362 * Beware, the bridge code notices the null rcvif and
363 * uses that identify that it's being called from
364 * ether_output as opposd to ether_input. Yech.
365 */
366 m->m_pkthdr.rcvif = NULL;
367 m = bdg_forward_ptr(m, ifp);
368 if (m != NULL)
369 m_freem(m);
370 return (0);
371 }
372 #if defined(INET) || defined(INET6)
373 if (IPFW_LOADED && ether_ipfw != 0) {
374 if (ether_ipfw_chk(&m, ifp, &rule, 0) == 0) {
375 if (m) {
376 m_freem(m);
377 return EACCES; /* pkt dropped */
378 } else
379 return 0; /* consumed e.g. in a pipe */
380 }
381 }
382 #endif
383
384 /*
385 * Queue message on interface, update output statistics if
386 * successful, and start output if interface not yet active.
387 */
388 IFQ_HANDOFF(ifp, m, error);
389 return (error);
390 }
391
392 #if defined(INET) || defined(INET6)
393 /*
394 * ipfw processing for ethernet packets (in and out).
395 * The second parameter is NULL from ether_demux, and ifp from
396 * ether_output_frame. This section of code could be used from
397 * bridge.c as well as long as we use some extra info
398 * to distinguish that case from ether_output_frame();
399 */
400 int
401 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
402 struct ip_fw **rule, int shared)
403 {
404 struct ether_header *eh;
405 struct ether_header save_eh;
406 struct mbuf *m;
407 int i;
408 struct ip_fw_args args;
409
410 if (*rule != NULL && fw_one_pass)
411 return 1; /* dummynet packet, already partially processed */
412
413 /*
414 * I need some amt of data to be contiguous, and in case others need
415 * the packet (shared==1) also better be in the first mbuf.
416 */
417 m = *m0;
418 i = min( m->m_pkthdr.len, max_protohdr);
419 if ( shared || m->m_len < i) {
420 m = m_pullup(m, i);
421 if (m == NULL) {
422 *m0 = m;
423 return 0;
424 }
425 }
426 eh = mtod(m, struct ether_header *);
427 save_eh = *eh; /* save copy for restore below */
428 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */
429
430 args.m = m; /* the packet we are looking at */
431 args.oif = dst; /* destination, if any */
432 args.rule = *rule; /* matching rule to restart */
433 args.next_hop = NULL; /* we do not support forward yet */
434 args.eh = &save_eh; /* MAC header for bridged/MAC packets */
435 i = ip_fw_chk_ptr(&args);
436 m = args.m;
437 if (m != NULL) {
438 /*
439 * Restore Ethernet header, as needed, in case the
440 * mbuf chain was replaced by ipfw.
441 */
442 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
443 if (m == NULL) {
444 *m0 = m;
445 return 0;
446 }
447 if (eh != mtod(m, struct ether_header *))
448 bcopy(&save_eh, mtod(m, struct ether_header *),
449 ETHER_HDR_LEN);
450 }
451 *m0 = m;
452 *rule = args.rule;
453
454 if ( (i & IP_FW_PORT_DENY_FLAG) || m == NULL) /* drop */
455 return 0;
456
457 if (i == 0) /* a PASS rule. */
458 return 1;
459
460 if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG)) {
461 /*
462 * Pass the pkt to dummynet, which consumes it.
463 * If shared, make a copy and keep the original.
464 */
465 if (shared) {
466 m = m_copypacket(m, M_DONTWAIT);
467 if (m == NULL)
468 return 0;
469 } else {
470 /*
471 * Pass the original to dummynet and
472 * nothing back to the caller
473 */
474 *m0 = NULL ;
475 }
476 ip_dn_io_ptr(m, (i & 0xffff),
477 dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
478 return 0;
479 }
480 /*
481 * XXX at some point add support for divert/forward actions.
482 * If none of the above matches, we have to drop the pkt.
483 */
484 return 0;
485 }
486 #endif
487
488 /*
489 * Process a received Ethernet packet; the packet is in the
490 * mbuf chain m with the ethernet header at the front.
491 */
492 static void
493 ether_input(struct ifnet *ifp, struct mbuf *m)
494 {
495 struct ether_header *eh;
496 u_short etype;
497
498 /*
499 * Do consistency checks to verify assumptions
500 * made by code past this point.
501 */
502 if ((m->m_flags & M_PKTHDR) == 0) {
503 if_printf(ifp, "discard frame w/o packet header\n");
504 ifp->if_ierrors++;
505 m_freem(m);
506 return;
507 }
508 if (m->m_len < ETHER_HDR_LEN) {
509 /* XXX maybe should pullup? */
510 if_printf(ifp, "discard frame w/o leading ethernet "
511 "header (len %u pkt len %u)\n",
512 m->m_len, m->m_pkthdr.len);
513 ifp->if_ierrors++;
514 m_freem(m);
515 return;
516 }
517 eh = mtod(m, struct ether_header *);
518 etype = ntohs(eh->ether_type);
519 if (m->m_pkthdr.len >
520 ETHER_MAX_FRAME(ifp, etype, m->m_flags & M_HASFCS)) {
521 if_printf(ifp, "discard oversize frame "
522 "(ether type %x flags %x len %u > max %lu)\n",
523 etype, m->m_flags, m->m_pkthdr.len,
524 ETHER_MAX_FRAME(ifp, etype,
525 m->m_flags & M_HASFCS));
526 ifp->if_ierrors++;
527 m_freem(m);
528 return;
529 }
530 if (m->m_pkthdr.rcvif == NULL) {
531 if_printf(ifp, "discard frame w/o interface pointer\n");
532 ifp->if_ierrors++;
533 m_freem(m);
534 return;
535 }
536 #ifdef DIAGNOSTIC
537 if (m->m_pkthdr.rcvif != ifp) {
538 if_printf(ifp, "Warning, frame marked as received on %s\n",
539 m->m_pkthdr.rcvif->if_xname);
540 }
541 #endif
542
543 #ifdef MAC
544 /*
545 * Tag the mbuf with an appropriate MAC label before any other
546 * consumers can get to it.
547 */
548 mac_create_mbuf_from_ifnet(ifp, m);
549 #endif
550
551 /*
552 * Give bpf a chance at the packet.
553 */
554 BPF_MTAP(ifp, m);
555
556 if (ifp->if_flags & IFF_MONITOR) {
557 /*
558 * Interface marked for monitoring; discard packet.
559 */
560 m_freem(m);
561 return;
562 }
563
564 /* If the CRC is still on the packet, trim it off. */
565 if (m->m_flags & M_HASFCS) {
566 m_adj(m, -ETHER_CRC_LEN);
567 m->m_flags &= ~M_HASFCS;
568 }
569
570 ifp->if_ibytes += m->m_pkthdr.len;
571
572 /* Handle ng_ether(4) processing, if any */
573 if (ng_ether_input_p != NULL) {
574 (*ng_ether_input_p)(ifp, &m);
575 if (m == NULL)
576 return;
577 }
578
579 /* Check for bridging mode */
580 if (BDG_ACTIVE(ifp) ) {
581 struct ifnet *bif;
582
583 /*
584 * Check with bridging code to see how the packet
585 * should be handled. Possibilities are:
586 *
587 * BDG_BCAST broadcast
588 * BDG_MCAST multicast
589 * BDG_LOCAL for local address, don't forward
590 * BDG_DROP discard
591 * ifp forward only to specified interface(s)
592 *
593 * Non-local destinations are handled by passing the
594 * packet back to the bridge code.
595 */
596 bif = bridge_in_ptr(ifp, eh);
597 if (bif == BDG_DROP) { /* discard packet */
598 m_freem(m);
599 return;
600 }
601 if (bif != BDG_LOCAL) { /* non-local, forward */
602 m = bdg_forward_ptr(m, bif);
603 /*
604 * The bridge may consume the packet if it's not
605 * supposed to be passed up or if a problem occurred
606 * while doing its job. This is reflected by it
607 * returning a NULL mbuf pointer.
608 */
609 if (m == NULL) {
610 if (bif == BDG_BCAST || bif == BDG_MCAST)
611 if_printf(ifp,
612 "bridge dropped %s packet\n",
613 bif == BDG_BCAST ? "broadcast" :
614 "multicast");
615 return;
616 }
617 /*
618 * But in some cases the bridge may return the
619 * packet for us to free; sigh.
620 */
621 if (bif != BDG_BCAST && bif != BDG_MCAST) {
622 m_freem(m);
623 return;
624 }
625 }
626 }
627
628 /* First chunk of an mbuf contains good entropy */
629 if (harvest.ethernet)
630 random_harvest(m, 16, 3, 0, RANDOM_NET);
631 ether_demux(ifp, m);
632 }
633
634 /*
635 * Upper layer processing for a received Ethernet packet.
636 */
637 void
638 ether_demux(struct ifnet *ifp, struct mbuf *m)
639 {
640 struct ether_header *eh;
641 int isr;
642 u_short ether_type;
643 #if defined(NETATALK)
644 struct llc *l;
645 #endif
646 #if defined(INET) || defined(INET6)
647 struct ip_fw *rule = ip_dn_claim_rule(m);
648 #endif
649
650 KASSERT(ifp != NULL, ("ether_demux: NULL interface pointer"));
651
652 eh = mtod(m, struct ether_header *);
653 ether_type = ntohs(eh->ether_type);
654
655 #if defined(INET) || defined(INET6)
656 if (rule) /* packet was already bridged */
657 goto post_stats;
658 #endif
659
660 if (!(BDG_ACTIVE(ifp)) &&
661 !(ether_type == ETHERTYPE_VLAN && ifp->if_nvlans > 0)) {
662 #ifdef DEV_CARP
663 /*
664 * XXX: Okay, we need to call carp_forus() and - if it is for
665 * us jump over code that does the normal check
666 * "ac_enaddr == ether_dhost". The check sequence is a bit
667 * different from OpenBSD, so we jump over as few code as
668 * possible, to catch _all_ sanity checks. This needs
669 * evaluation, to see if the carp ether_dhost values break any
670 * of these checks!
671 */
672 if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost))
673 goto pre_stats;
674 #endif
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 VLANs are active, and this packet has a VLAN tag, do
682 * not drop it here but pass it on to the VLAN layer, to
683 * give them a chance to consider it as well (e. g. in case
684 * bridging is only active on a VLAN). They will drop it if
685 * it's undesired.
686 */
687 if ((ifp->if_flags & IFF_PROMISC) != 0
688 && (eh->ether_dhost[0] & 1) == 0
689 && bcmp(eh->ether_dhost,
690 IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN) != 0
691 && (ifp->if_flags & IFF_PPROMISC) == 0) {
692 m_freem(m);
693 return;
694 }
695 }
696
697 #ifdef DEV_CARP
698 pre_stats:
699 #endif
700 /* Discard packet if interface is not up */
701 if ((ifp->if_flags & IFF_UP) == 0) {
702 m_freem(m);
703 return;
704 }
705 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
706 if (bcmp(etherbroadcastaddr, eh->ether_dhost,
707 sizeof(etherbroadcastaddr)) == 0)
708 m->m_flags |= M_BCAST;
709 else
710 m->m_flags |= M_MCAST;
711 }
712 if (m->m_flags & (M_BCAST|M_MCAST))
713 ifp->if_imcasts++;
714
715 #if defined(INET) || defined(INET6)
716 post_stats:
717 if (IPFW_LOADED && ether_ipfw != 0) {
718 if (ether_ipfw_chk(&m, NULL, &rule, 0) == 0) {
719 if (m)
720 m_freem(m);
721 return;
722 }
723 }
724 #endif
725
726 /*
727 * If VLANs are configured on the interface, check to
728 * see if the device performed the decapsulation and
729 * provided us with the tag.
730 */
731 if (ifp->if_nvlans &&
732 m_tag_locate(m, MTAG_VLAN, MTAG_VLAN_TAG, NULL) != NULL) {
733 /*
734 * vlan_input() will either recursively call ether_input()
735 * or drop the packet.
736 */
737 KASSERT(vlan_input_p != NULL,("ether_input: VLAN not loaded!"));
738 (*vlan_input_p)(ifp, m);
739 return;
740 }
741
742 /*
743 * Handle protocols that expect to have the Ethernet header
744 * (and possibly FCS) intact.
745 */
746 switch (ether_type) {
747 case ETHERTYPE_VLAN:
748 if (ifp->if_nvlans != 0) {
749 KASSERT(vlan_input_p,("ether_input: VLAN not loaded!"));
750 (*vlan_input_p)(ifp, m);
751 } else {
752 ifp->if_noproto++;
753 m_freem(m);
754 }
755 return;
756 }
757
758 /* Strip off Ethernet header. */
759 m_adj(m, ETHER_HDR_LEN);
760
761 /* If the CRC is still on the packet, trim it off. */
762 if (m->m_flags & M_HASFCS) {
763 m_adj(m, -ETHER_CRC_LEN);
764 m->m_flags &= ~M_HASFCS;
765 }
766
767 switch (ether_type) {
768 #ifdef INET
769 case ETHERTYPE_IP:
770 if (ip_fastforward(m))
771 return;
772 isr = NETISR_IP;
773 break;
774
775 case ETHERTYPE_ARP:
776 if (ifp->if_flags & IFF_NOARP) {
777 /* Discard packet if ARP is disabled on interface */
778 m_freem(m);
779 return;
780 }
781 isr = NETISR_ARP;
782 break;
783 #endif
784 #ifdef IPX
785 case ETHERTYPE_IPX:
786 if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
787 return;
788 isr = NETISR_IPX;
789 break;
790 #endif
791 #ifdef INET6
792 case ETHERTYPE_IPV6:
793 isr = NETISR_IPV6;
794 break;
795 #endif
796 #ifdef NETATALK
797 case ETHERTYPE_AT:
798 isr = NETISR_ATALK1;
799 break;
800 case ETHERTYPE_AARP:
801 isr = NETISR_AARP;
802 break;
803 #endif /* NETATALK */
804 default:
805 #ifdef IPX
806 if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
807 return;
808 #endif /* IPX */
809 #if defined(NETATALK)
810 if (ether_type > ETHERMTU)
811 goto discard;
812 l = mtod(m, struct llc *);
813 if (l->llc_dsap == LLC_SNAP_LSAP &&
814 l->llc_ssap == LLC_SNAP_LSAP &&
815 l->llc_control == LLC_UI) {
816 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code,
817 sizeof(at_org_code)) == 0 &&
818 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
819 m_adj(m, LLC_SNAPFRAMELEN);
820 isr = NETISR_ATALK2;
821 break;
822 }
823 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
824 sizeof(aarp_org_code)) == 0 &&
825 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
826 m_adj(m, LLC_SNAPFRAMELEN);
827 isr = NETISR_AARP;
828 break;
829 }
830 }
831 #endif /* NETATALK */
832 goto discard;
833 }
834 netisr_dispatch(isr, m);
835 return;
836
837 discard:
838 /*
839 * Packet is to be discarded. If netgraph is present,
840 * hand the packet to it for last chance processing;
841 * otherwise dispose of it.
842 */
843 if (ng_ether_input_orphan_p != NULL) {
844 /*
845 * Put back the ethernet header so netgraph has a
846 * consistent view of inbound packets.
847 */
848 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
849 (*ng_ether_input_orphan_p)(ifp, m);
850 return;
851 }
852 m_freem(m);
853 }
854
855 /*
856 * Convert Ethernet address to printable (loggable) representation.
857 * This routine is for compatibility; it's better to just use
858 *
859 * printf("%6D", <pointer to address>, ":");
860 *
861 * since there's no static buffer involved.
862 */
863 char *
864 ether_sprintf(const u_char *ap)
865 {
866 static char etherbuf[18];
867 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
868 return (etherbuf);
869 }
870
871 /*
872 * Perform common duties while attaching to interface list
873 */
874 void
875 ether_ifattach(struct ifnet *ifp, const u_int8_t *llc)
876 {
877 int i;
878 struct ifaddr *ifa;
879 struct sockaddr_dl *sdl;
880
881 ifp->if_type = IFT_ETHER;
882 ifp->if_addrlen = ETHER_ADDR_LEN;
883 ifp->if_hdrlen = ETHER_HDR_LEN;
884 if_attach(ifp);
885 ifp->if_mtu = ETHERMTU;
886 ifp->if_output = ether_output;
887 ifp->if_input = ether_input;
888 ifp->if_resolvemulti = ether_resolvemulti;
889 if (ifp->if_baudrate == 0)
890 ifp->if_baudrate = IF_Mbps(10); /* just a default */
891 ifp->if_broadcastaddr = etherbroadcastaddr;
892
893 ifa = ifaddr_byindex(ifp->if_index);
894 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
895 sdl = (struct sockaddr_dl *)ifa->ifa_addr;
896 sdl->sdl_type = IFT_ETHER;
897 sdl->sdl_alen = ifp->if_addrlen;
898 bcopy(llc, LLADDR(sdl), ifp->if_addrlen);
899 /*
900 * XXX: This doesn't belong here; we do it until
901 * XXX: all drivers are cleaned up
902 */
903 if (llc != IFP2AC(ifp)->ac_enaddr)
904 bcopy(llc, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen);
905
906 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
907 if (ng_ether_attach_p != NULL)
908 (*ng_ether_attach_p)(ifp);
909 if (BDG_LOADED)
910 bdgtakeifaces_ptr();
911
912 /* Announce Ethernet MAC address if non-zero. */
913 for (i = 0; i < ifp->if_addrlen; i++)
914 if (llc[i] != 0)
915 break;
916 if (i != ifp->if_addrlen)
917 if_printf(ifp, "Ethernet address: %6D\n", llc, ":");
918 if (debug_mpsafenet && (ifp->if_flags & IFF_NEEDSGIANT) != 0)
919 if_printf(ifp, "if_start running deferred for Giant\n");
920 }
921
922 /*
923 * Perform common duties while detaching an Ethernet interface
924 */
925 void
926 ether_ifdetach(struct ifnet *ifp)
927 {
928 if (ng_ether_detach_p != NULL)
929 (*ng_ether_detach_p)(ifp);
930 bpfdetach(ifp);
931 if_detach(ifp);
932 if (BDG_LOADED)
933 bdgtakeifaces_ptr();
934 }
935
936 SYSCTL_DECL(_net_link);
937 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
938 #if defined(INET) || defined(INET6)
939 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
940 ðer_ipfw,0,"Pass ether pkts through firewall");
941 #endif
942
943 #if 0
944 /*
945 * This is for reference. We have a table-driven version
946 * of the little-endian crc32 generator, which is faster
947 * than the double-loop.
948 */
949 uint32_t
950 ether_crc32_le(const uint8_t *buf, size_t len)
951 {
952 size_t i;
953 uint32_t crc;
954 int bit;
955 uint8_t data;
956
957 crc = 0xffffffff; /* initial value */
958
959 for (i = 0; i < len; i++) {
960 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1)
961 carry = (crc ^ data) & 1;
962 crc >>= 1;
963 if (carry)
964 crc = (crc ^ ETHER_CRC_POLY_LE);
965 }
966
967 return (crc);
968 }
969 #else
970 uint32_t
971 ether_crc32_le(const uint8_t *buf, size_t len)
972 {
973 static const uint32_t crctab[] = {
974 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
975 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
976 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
977 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
978 };
979 size_t i;
980 uint32_t crc;
981
982 crc = 0xffffffff; /* initial value */
983
984 for (i = 0; i < len; i++) {
985 crc ^= buf[i];
986 crc = (crc >> 4) ^ crctab[crc & 0xf];
987 crc = (crc >> 4) ^ crctab[crc & 0xf];
988 }
989
990 return (crc);
991 }
992 #endif
993
994 uint32_t
995 ether_crc32_be(const uint8_t *buf, size_t len)
996 {
997 size_t i;
998 uint32_t crc, carry;
999 int bit;
1000 uint8_t data;
1001
1002 crc = 0xffffffff; /* initial value */
1003
1004 for (i = 0; i < len; i++) {
1005 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
1006 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
1007 crc <<= 1;
1008 if (carry)
1009 crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1010 }
1011 }
1012
1013 return (crc);
1014 }
1015
1016 int
1017 ether_ioctl(struct ifnet *ifp, int command, caddr_t data)
1018 {
1019 struct ifaddr *ifa = (struct ifaddr *) data;
1020 struct ifreq *ifr = (struct ifreq *) data;
1021 int error = 0;
1022
1023 switch (command) {
1024 case SIOCSIFADDR:
1025 ifp->if_flags |= IFF_UP;
1026
1027 switch (ifa->ifa_addr->sa_family) {
1028 #ifdef INET
1029 case AF_INET:
1030 ifp->if_init(ifp->if_softc); /* before arpwhohas */
1031 arp_ifinit(ifp, ifa);
1032 break;
1033 #endif
1034 #ifdef IPX
1035 /*
1036 * XXX - This code is probably wrong
1037 */
1038 case AF_IPX:
1039 {
1040 struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr);
1041 struct arpcom *ac = IFP2AC(ifp);
1042
1043 if (ipx_nullhost(*ina))
1044 ina->x_host =
1045 *(union ipx_host *)
1046 ac->ac_enaddr;
1047 else {
1048 bcopy((caddr_t) ina->x_host.c_host,
1049 (caddr_t) ac->ac_enaddr,
1050 sizeof(ac->ac_enaddr));
1051 }
1052
1053 /*
1054 * Set new address
1055 */
1056 ifp->if_init(ifp->if_softc);
1057 break;
1058 }
1059 #endif
1060 default:
1061 ifp->if_init(ifp->if_softc);
1062 break;
1063 }
1064 break;
1065
1066 case SIOCGIFADDR:
1067 {
1068 struct sockaddr *sa;
1069
1070 sa = (struct sockaddr *) & ifr->ifr_data;
1071 bcopy(IFP2AC(ifp)->ac_enaddr,
1072 (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
1073 }
1074 break;
1075
1076 case SIOCSIFMTU:
1077 /*
1078 * Set the interface MTU.
1079 */
1080 if (ifr->ifr_mtu > ETHERMTU) {
1081 error = EINVAL;
1082 } else {
1083 ifp->if_mtu = ifr->ifr_mtu;
1084 }
1085 break;
1086 default:
1087 error = EINVAL; /* XXX netbsd has ENOTTY??? */
1088 break;
1089 }
1090 return (error);
1091 }
1092
1093 static int
1094 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
1095 struct sockaddr *sa)
1096 {
1097 struct sockaddr_dl *sdl;
1098 #ifdef INET
1099 struct sockaddr_in *sin;
1100 #endif
1101 #ifdef INET6
1102 struct sockaddr_in6 *sin6;
1103 #endif
1104 u_char *e_addr;
1105
1106 switch(sa->sa_family) {
1107 case AF_LINK:
1108 /*
1109 * No mapping needed. Just check that it's a valid MC address.
1110 */
1111 sdl = (struct sockaddr_dl *)sa;
1112 e_addr = LLADDR(sdl);
1113 if (!ETHER_IS_MULTICAST(e_addr))
1114 return EADDRNOTAVAIL;
1115 *llsa = 0;
1116 return 0;
1117
1118 #ifdef INET
1119 case AF_INET:
1120 sin = (struct sockaddr_in *)sa;
1121 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1122 return EADDRNOTAVAIL;
1123 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1124 M_WAITOK|M_ZERO);
1125 sdl->sdl_len = sizeof *sdl;
1126 sdl->sdl_family = AF_LINK;
1127 sdl->sdl_index = ifp->if_index;
1128 sdl->sdl_type = IFT_ETHER;
1129 sdl->sdl_alen = ETHER_ADDR_LEN;
1130 e_addr = LLADDR(sdl);
1131 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1132 *llsa = (struct sockaddr *)sdl;
1133 return 0;
1134 #endif
1135 #ifdef INET6
1136 case AF_INET6:
1137 sin6 = (struct sockaddr_in6 *)sa;
1138 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1139 /*
1140 * An IP6 address of 0 means listen to all
1141 * of the Ethernet multicast address used for IP6.
1142 * (This is used for multicast routers.)
1143 */
1144 ifp->if_flags |= IFF_ALLMULTI;
1145 *llsa = 0;
1146 return 0;
1147 }
1148 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1149 return EADDRNOTAVAIL;
1150 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
1151 M_WAITOK|M_ZERO);
1152 sdl->sdl_len = sizeof *sdl;
1153 sdl->sdl_family = AF_LINK;
1154 sdl->sdl_index = ifp->if_index;
1155 sdl->sdl_type = IFT_ETHER;
1156 sdl->sdl_alen = ETHER_ADDR_LEN;
1157 e_addr = LLADDR(sdl);
1158 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1159 *llsa = (struct sockaddr *)sdl;
1160 return 0;
1161 #endif
1162
1163 default:
1164 /*
1165 * Well, the text isn't quite right, but it's the name
1166 * that counts...
1167 */
1168 return EAFNOSUPPORT;
1169 }
1170 }
1171
1172 static moduledata_t ether_mod = {
1173 "ether",
1174 NULL,
1175 0
1176 };
1177
1178 DECLARE_MODULE(ether, ether_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
1179 MODULE_VERSION(ether, 1);
Cache object: 0191d652a910796cd3114615e12aab0f
|