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/9.2/sys/net/if_ethersubr.c 252145 2013-06-24 09:00:18Z glebius $
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_netgraph.h"
38 #include "opt_mbuf_profiling.h"
39
40 #include <sys/param.h>
41 #include <sys/systm.h>
42 #include <sys/kernel.h>
43 #include <sys/lock.h>
44 #include <sys/malloc.h>
45 #include <sys/module.h>
46 #include <sys/mbuf.h>
47 #include <sys/random.h>
48 #include <sys/rwlock.h>
49 #include <sys/socket.h>
50 #include <sys/sockio.h>
51 #include <sys/sysctl.h>
52
53 #include <net/if.h>
54 #include <net/if_arp.h>
55 #include <net/netisr.h>
56 #include <net/route.h>
57 #include <net/if_llc.h>
58 #include <net/if_dl.h>
59 #include <net/if_types.h>
60 #include <net/bpf.h>
61 #include <net/ethernet.h>
62 #include <net/if_bridgevar.h>
63 #include <net/if_vlan_var.h>
64 #include <net/if_llatbl.h>
65 #include <net/pf_mtag.h>
66 #include <net/vnet.h>
67
68 #if defined(INET) || defined(INET6)
69 #include <netinet/in.h>
70 #include <netinet/in_var.h>
71 #include <netinet/if_ether.h>
72 #include <netinet/ip_carp.h>
73 #include <netinet/ip_var.h>
74 #include <netinet/ip_fw.h>
75 #include <netpfil/ipfw/ip_fw_private.h>
76 #endif
77 #ifdef INET6
78 #include <netinet6/nd6.h>
79 #endif
80
81 #ifdef IPX
82 #include <netipx/ipx.h>
83 #include <netipx/ipx_if.h>
84 #endif
85
86 int (*ef_inputp)(struct ifnet*, struct ether_header *eh, struct mbuf *m);
87 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp,
88 struct sockaddr *dst, short *tp, int *hlen);
89
90 #ifdef NETATALK
91 #include <netatalk/at.h>
92 #include <netatalk/at_var.h>
93 #include <netatalk/at_extern.h>
94
95 #define llc_snap_org_code llc_un.type_snap.org_code
96 #define llc_snap_ether_type llc_un.type_snap.ether_type
97
98 extern u_char at_org_code[3];
99 extern u_char aarp_org_code[3];
100 #endif /* NETATALK */
101
102 #include <security/mac/mac_framework.h>
103
104 #ifdef CTASSERT
105 CTASSERT(sizeof (struct ether_header) == ETHER_ADDR_LEN * 2 + 2);
106 CTASSERT(sizeof (struct ether_addr) == ETHER_ADDR_LEN);
107 #endif
108
109 /* netgraph node hooks for ng_ether(4) */
110 void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp);
111 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m);
112 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
113 void (*ng_ether_attach_p)(struct ifnet *ifp);
114 void (*ng_ether_detach_p)(struct ifnet *ifp);
115
116 void (*vlan_input_p)(struct ifnet *, struct mbuf *);
117
118 /* if_bridge(4) support */
119 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *);
120 int (*bridge_output_p)(struct ifnet *, struct mbuf *,
121 struct sockaddr *, struct rtentry *);
122 void (*bridge_dn_p)(struct mbuf *, struct ifnet *);
123
124 /* if_lagg(4) support */
125 struct mbuf *(*lagg_input_p)(struct ifnet *, struct mbuf *);
126
127 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] =
128 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
129
130 static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
131 struct sockaddr *);
132 #ifdef VIMAGE
133 static void ether_reassign(struct ifnet *, struct vnet *, char *);
134 #endif
135
136 /* XXX: should be in an arp support file, not here */
137 static MALLOC_DEFINE(M_ARPCOM, "arpcom", "802.* interface internals");
138
139 #define ETHER_IS_BROADCAST(addr) \
140 (bcmp(etherbroadcastaddr, (addr), ETHER_ADDR_LEN) == 0)
141
142 #define senderr(e) do { error = (e); goto bad;} while (0)
143
144 #if defined(INET) || defined(INET6)
145 int
146 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, int shared);
147 static VNET_DEFINE(int, ether_ipfw);
148 #define V_ether_ipfw VNET(ether_ipfw)
149 #endif
150
151
152 /*
153 * Ethernet output routine.
154 * Encapsulate a packet of type family for the local net.
155 * Use trailer local net encapsulation if enough data in first
156 * packet leaves a multiple of 512 bytes of data in remainder.
157 */
158 int
159 ether_output(struct ifnet *ifp, struct mbuf *m,
160 struct sockaddr *dst, struct route *ro)
161 {
162 short type;
163 int error = 0, hdrcmplt = 0;
164 u_char esrc[ETHER_ADDR_LEN], edst[ETHER_ADDR_LEN];
165 struct llentry *lle = NULL;
166 struct rtentry *rt0 = NULL;
167 struct ether_header *eh;
168 struct pf_mtag *t;
169 int loop_copy = 1;
170 int hlen; /* link layer header length */
171
172 if (ro != NULL) {
173 if (!(m->m_flags & (M_BCAST | M_MCAST)))
174 lle = ro->ro_lle;
175 rt0 = ro->ro_rt;
176 }
177 #ifdef MAC
178 error = mac_ifnet_check_transmit(ifp, m);
179 if (error)
180 senderr(error);
181 #endif
182
183 M_PROFILE(m);
184 if (ifp->if_flags & IFF_MONITOR)
185 senderr(ENETDOWN);
186 if (!((ifp->if_flags & IFF_UP) &&
187 (ifp->if_drv_flags & IFF_DRV_RUNNING)))
188 senderr(ENETDOWN);
189
190 hlen = ETHER_HDR_LEN;
191 switch (dst->sa_family) {
192 #ifdef INET
193 case AF_INET:
194 if (lle != NULL && (lle->la_flags & LLE_VALID))
195 memcpy(edst, &lle->ll_addr.mac16, sizeof(edst));
196 else
197 error = arpresolve(ifp, rt0, m, dst, edst, &lle);
198 if (error)
199 return (error == EWOULDBLOCK ? 0 : error);
200 type = htons(ETHERTYPE_IP);
201 break;
202 case AF_ARP:
203 {
204 struct arphdr *ah;
205 ah = mtod(m, struct arphdr *);
206 ah->ar_hrd = htons(ARPHRD_ETHER);
207
208 loop_copy = 0; /* if this is for us, don't do it */
209
210 switch(ntohs(ah->ar_op)) {
211 case ARPOP_REVREQUEST:
212 case ARPOP_REVREPLY:
213 type = htons(ETHERTYPE_REVARP);
214 break;
215 case ARPOP_REQUEST:
216 case ARPOP_REPLY:
217 default:
218 type = htons(ETHERTYPE_ARP);
219 break;
220 }
221
222 if (m->m_flags & M_BCAST)
223 bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN);
224 else
225 bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN);
226
227 }
228 break;
229 #endif
230 #ifdef INET6
231 case AF_INET6:
232 if (lle != NULL && (lle->la_flags & LLE_VALID))
233 memcpy(edst, &lle->ll_addr.mac16, sizeof(edst));
234 else
235 error = nd6_storelladdr(ifp, m, dst, (u_char *)edst, &lle);
236 if (error)
237 return error;
238 type = htons(ETHERTYPE_IPV6);
239 break;
240 #endif
241 #ifdef IPX
242 case AF_IPX:
243 if (ef_outputp) {
244 error = ef_outputp(ifp, &m, dst, &type, &hlen);
245 if (error)
246 goto bad;
247 } else
248 type = htons(ETHERTYPE_IPX);
249 bcopy((caddr_t)&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
250 (caddr_t)edst, sizeof (edst));
251 break;
252 #endif
253 #ifdef NETATALK
254 case AF_APPLETALK:
255 {
256 struct at_ifaddr *aa;
257
258 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL)
259 senderr(EHOSTUNREACH); /* XXX */
260 if (!aarpresolve(ifp, m, (struct sockaddr_at *)dst, edst)) {
261 ifa_free(&aa->aa_ifa);
262 return (0);
263 }
264 /*
265 * In the phase 2 case, need to prepend an mbuf for the llc header.
266 */
267 if ( aa->aa_flags & AFA_PHASE2 ) {
268 struct llc llc;
269
270 ifa_free(&aa->aa_ifa);
271 M_PREPEND(m, LLC_SNAPFRAMELEN, M_DONTWAIT);
272 if (m == NULL)
273 senderr(ENOBUFS);
274 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
275 llc.llc_control = LLC_UI;
276 bcopy(at_org_code, llc.llc_snap_org_code, sizeof(at_org_code));
277 llc.llc_snap_ether_type = htons( ETHERTYPE_AT );
278 bcopy(&llc, mtod(m, caddr_t), LLC_SNAPFRAMELEN);
279 type = htons(m->m_pkthdr.len);
280 hlen = LLC_SNAPFRAMELEN + ETHER_HDR_LEN;
281 } else {
282 ifa_free(&aa->aa_ifa);
283 type = htons(ETHERTYPE_AT);
284 }
285 break;
286 }
287 #endif /* NETATALK */
288
289 case pseudo_AF_HDRCMPLT:
290 hdrcmplt = 1;
291 eh = (struct ether_header *)dst->sa_data;
292 (void)memcpy(esrc, eh->ether_shost, sizeof (esrc));
293 /* FALLTHROUGH */
294
295 case AF_UNSPEC:
296 loop_copy = 0; /* if this is for us, don't do it */
297 eh = (struct ether_header *)dst->sa_data;
298 (void)memcpy(edst, eh->ether_dhost, sizeof (edst));
299 type = eh->ether_type;
300 break;
301
302 default:
303 if_printf(ifp, "can't handle af%d\n", dst->sa_family);
304 senderr(EAFNOSUPPORT);
305 }
306
307 if (lle != NULL && (lle->la_flags & LLE_IFADDR)) {
308 int csum_flags = 0;
309 if (m->m_pkthdr.csum_flags & CSUM_IP)
310 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
311 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
312 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
313 if (m->m_pkthdr.csum_flags & CSUM_SCTP)
314 csum_flags |= CSUM_SCTP_VALID;
315 m->m_pkthdr.csum_flags |= csum_flags;
316 m->m_pkthdr.csum_data = 0xffff;
317 return (if_simloop(ifp, m, dst->sa_family, 0));
318 }
319
320 /*
321 * Add local net header. If no space in first mbuf,
322 * allocate another.
323 */
324 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
325 if (m == NULL)
326 senderr(ENOBUFS);
327 eh = mtod(m, struct ether_header *);
328 (void)memcpy(&eh->ether_type, &type,
329 sizeof(eh->ether_type));
330 (void)memcpy(eh->ether_dhost, edst, sizeof (edst));
331 if (hdrcmplt)
332 (void)memcpy(eh->ether_shost, esrc,
333 sizeof(eh->ether_shost));
334 else
335 (void)memcpy(eh->ether_shost, IF_LLADDR(ifp),
336 sizeof(eh->ether_shost));
337
338 /*
339 * If a simplex interface, and the packet is being sent to our
340 * Ethernet address or a broadcast address, loopback a copy.
341 * XXX To make a simplex device behave exactly like a duplex
342 * device, we should copy in the case of sending to our own
343 * ethernet address (thus letting the original actually appear
344 * on the wire). However, we don't do that here for security
345 * reasons and compatibility with the original behavior.
346 */
347 if ((ifp->if_flags & IFF_SIMPLEX) && loop_copy &&
348 ((t = pf_find_mtag(m)) == NULL || !t->routed)) {
349 int csum_flags = 0;
350
351 if (m->m_pkthdr.csum_flags & CSUM_IP)
352 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
353 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
354 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
355 if (m->m_pkthdr.csum_flags & CSUM_SCTP)
356 csum_flags |= CSUM_SCTP_VALID;
357
358 if (m->m_flags & M_BCAST) {
359 struct mbuf *n;
360
361 /*
362 * Because if_simloop() modifies the packet, we need a
363 * writable copy through m_dup() instead of a readonly
364 * one as m_copy[m] would give us. The alternative would
365 * be to modify if_simloop() to handle the readonly mbuf,
366 * but performancewise it is mostly equivalent (trading
367 * extra data copying vs. extra locking).
368 *
369 * XXX This is a local workaround. A number of less
370 * often used kernel parts suffer from the same bug.
371 * See PR kern/105943 for a proposed general solution.
372 */
373 if ((n = m_dup(m, M_DONTWAIT)) != NULL) {
374 n->m_pkthdr.csum_flags |= csum_flags;
375 if (csum_flags & CSUM_DATA_VALID)
376 n->m_pkthdr.csum_data = 0xffff;
377 (void)if_simloop(ifp, n, dst->sa_family, hlen);
378 } else
379 ifp->if_iqdrops++;
380 } else if (bcmp(eh->ether_dhost, eh->ether_shost,
381 ETHER_ADDR_LEN) == 0) {
382 m->m_pkthdr.csum_flags |= csum_flags;
383 if (csum_flags & CSUM_DATA_VALID)
384 m->m_pkthdr.csum_data = 0xffff;
385 (void) if_simloop(ifp, m, dst->sa_family, hlen);
386 return (0); /* XXX */
387 }
388 }
389
390 /*
391 * Bridges require special output handling.
392 */
393 if (ifp->if_bridge) {
394 BRIDGE_OUTPUT(ifp, m, error);
395 return (error);
396 }
397
398 #if defined(INET) || defined(INET6)
399 if (ifp->if_carp &&
400 (error = (*carp_output_p)(ifp, m, dst, NULL)))
401 goto bad;
402 #endif
403
404 /* Handle ng_ether(4) processing, if any */
405 if (IFP2AC(ifp)->ac_netgraph != NULL) {
406 KASSERT(ng_ether_output_p != NULL,
407 ("ng_ether_output_p is NULL"));
408 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
409 bad: if (m != NULL)
410 m_freem(m);
411 return (error);
412 }
413 if (m == NULL)
414 return (0);
415 }
416
417 /* Continue with link-layer output */
418 return ether_output_frame(ifp, m);
419 }
420
421 /*
422 * Ethernet link layer output routine to send a raw frame to the device.
423 *
424 * This assumes that the 14 byte Ethernet header is present and contiguous
425 * in the first mbuf (if BRIDGE'ing).
426 */
427 int
428 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
429 {
430 #if defined(INET) || defined(INET6)
431
432 if (V_ip_fw_chk_ptr && V_ether_ipfw != 0) {
433 if (ether_ipfw_chk(&m, ifp, 0) == 0) {
434 if (m) {
435 m_freem(m);
436 return EACCES; /* pkt dropped */
437 } else
438 return 0; /* consumed e.g. in a pipe */
439 }
440 }
441 #endif
442
443 /*
444 * Queue message on interface, update output statistics if
445 * successful, and start output if interface not yet active.
446 */
447 return ((ifp->if_transmit)(ifp, m));
448 }
449
450 #if defined(INET) || defined(INET6)
451 /*
452 * ipfw processing for ethernet packets (in and out).
453 * The second parameter is NULL from ether_demux, and ifp from
454 * ether_output_frame.
455 */
456 int
457 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, int shared)
458 {
459 struct ether_header *eh;
460 struct ether_header save_eh;
461 struct mbuf *m;
462 int i;
463 struct ip_fw_args args;
464 struct m_tag *mtag;
465
466 /* fetch start point from rule, if any */
467 mtag = m_tag_locate(*m0, MTAG_IPFW_RULE, 0, NULL);
468 if (mtag == NULL) {
469 args.rule.slot = 0;
470 } else {
471 /* dummynet packet, already partially processed */
472 struct ipfw_rule_ref *r;
473
474 /* XXX can we free it after use ? */
475 mtag->m_tag_id = PACKET_TAG_NONE;
476 r = (struct ipfw_rule_ref *)(mtag + 1);
477 if (r->info & IPFW_ONEPASS)
478 return (1);
479 args.rule = *r;
480 }
481
482 /*
483 * I need some amt of data to be contiguous, and in case others need
484 * the packet (shared==1) also better be in the first mbuf.
485 */
486 m = *m0;
487 i = min( m->m_pkthdr.len, max_protohdr);
488 if ( shared || m->m_len < i) {
489 m = m_pullup(m, i);
490 if (m == NULL) {
491 *m0 = m;
492 return 0;
493 }
494 }
495 eh = mtod(m, struct ether_header *);
496 save_eh = *eh; /* save copy for restore below */
497 m_adj(m, ETHER_HDR_LEN); /* strip ethernet header */
498
499 args.m = m; /* the packet we are looking at */
500 args.oif = dst; /* destination, if any */
501 args.next_hop = NULL; /* we do not support forward yet */
502 args.next_hop6 = NULL; /* we do not support forward yet */
503 args.eh = &save_eh; /* MAC header for bridged/MAC packets */
504 args.inp = NULL; /* used by ipfw uid/gid/jail rules */
505 i = V_ip_fw_chk_ptr(&args);
506 m = args.m;
507 if (m != NULL) {
508 /*
509 * Restore Ethernet header, as needed, in case the
510 * mbuf chain was replaced by ipfw.
511 */
512 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
513 if (m == NULL) {
514 *m0 = m;
515 return 0;
516 }
517 if (eh != mtod(m, struct ether_header *))
518 bcopy(&save_eh, mtod(m, struct ether_header *),
519 ETHER_HDR_LEN);
520 }
521 *m0 = m;
522
523 if (i == IP_FW_DENY) /* drop */
524 return 0;
525
526 KASSERT(m != NULL, ("ether_ipfw_chk: m is NULL"));
527
528 if (i == IP_FW_PASS) /* a PASS rule. */
529 return 1;
530
531 if (ip_dn_io_ptr && (i == IP_FW_DUMMYNET)) {
532 int dir;
533 /*
534 * Pass the pkt to dummynet, which consumes it.
535 * If shared, make a copy and keep the original.
536 */
537 if (shared) {
538 m = m_copypacket(m, M_DONTWAIT);
539 if (m == NULL)
540 return 0;
541 } else {
542 /*
543 * Pass the original to dummynet and
544 * nothing back to the caller
545 */
546 *m0 = NULL ;
547 }
548 dir = PROTO_LAYER2 | (dst ? DIR_OUT : DIR_IN);
549 ip_dn_io_ptr(&m, dir, &args);
550 return 0;
551 }
552 /*
553 * XXX at some point add support for divert/forward actions.
554 * If none of the above matches, we have to drop the pkt.
555 */
556 return 0;
557 }
558 #endif
559
560 /*
561 * Process a received Ethernet packet; the packet is in the
562 * mbuf chain m with the ethernet header at the front.
563 */
564 static void
565 ether_input_internal(struct ifnet *ifp, struct mbuf *m)
566 {
567 struct ether_header *eh;
568 u_short etype;
569
570 if ((ifp->if_flags & IFF_UP) == 0) {
571 m_freem(m);
572 return;
573 }
574 #ifdef DIAGNOSTIC
575 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
576 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n");
577 m_freem(m);
578 return;
579 }
580 #endif
581 /*
582 * Do consistency checks to verify assumptions
583 * made by code past this point.
584 */
585 if ((m->m_flags & M_PKTHDR) == 0) {
586 if_printf(ifp, "discard frame w/o packet header\n");
587 ifp->if_ierrors++;
588 m_freem(m);
589 return;
590 }
591 if (m->m_len < ETHER_HDR_LEN) {
592 /* XXX maybe should pullup? */
593 if_printf(ifp, "discard frame w/o leading ethernet "
594 "header (len %u pkt len %u)\n",
595 m->m_len, m->m_pkthdr.len);
596 ifp->if_ierrors++;
597 m_freem(m);
598 return;
599 }
600 eh = mtod(m, struct ether_header *);
601 etype = ntohs(eh->ether_type);
602 if (m->m_pkthdr.rcvif == NULL) {
603 if_printf(ifp, "discard frame w/o interface pointer\n");
604 ifp->if_ierrors++;
605 m_freem(m);
606 return;
607 }
608 #ifdef DIAGNOSTIC
609 if (m->m_pkthdr.rcvif != ifp) {
610 if_printf(ifp, "Warning, frame marked as received on %s\n",
611 m->m_pkthdr.rcvif->if_xname);
612 }
613 #endif
614
615 CURVNET_SET_QUIET(ifp->if_vnet);
616
617 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
618 if (ETHER_IS_BROADCAST(eh->ether_dhost))
619 m->m_flags |= M_BCAST;
620 else
621 m->m_flags |= M_MCAST;
622 ifp->if_imcasts++;
623 }
624
625 #ifdef MAC
626 /*
627 * Tag the mbuf with an appropriate MAC label before any other
628 * consumers can get to it.
629 */
630 mac_ifnet_create_mbuf(ifp, m);
631 #endif
632
633 /*
634 * Give bpf a chance at the packet.
635 */
636 ETHER_BPF_MTAP(ifp, m);
637
638 /*
639 * If the CRC is still on the packet, trim it off. We do this once
640 * and once only in case we are re-entered. Nothing else on the
641 * Ethernet receive path expects to see the FCS.
642 */
643 if (m->m_flags & M_HASFCS) {
644 m_adj(m, -ETHER_CRC_LEN);
645 m->m_flags &= ~M_HASFCS;
646 }
647
648 ifp->if_ibytes += m->m_pkthdr.len;
649
650 /* Allow monitor mode to claim this frame, after stats are updated. */
651 if (ifp->if_flags & IFF_MONITOR) {
652 m_freem(m);
653 CURVNET_RESTORE();
654 return;
655 }
656
657 /* Handle input from a lagg(4) port */
658 if (ifp->if_type == IFT_IEEE8023ADLAG) {
659 KASSERT(lagg_input_p != NULL,
660 ("%s: if_lagg not loaded!", __func__));
661 m = (*lagg_input_p)(ifp, m);
662 if (m != NULL)
663 ifp = m->m_pkthdr.rcvif;
664 else {
665 CURVNET_RESTORE();
666 return;
667 }
668 }
669
670 /*
671 * If the hardware did not process an 802.1Q tag, do this now,
672 * to allow 802.1P priority frames to be passed to the main input
673 * path correctly.
674 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels.
675 */
676 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) {
677 struct ether_vlan_header *evl;
678
679 if (m->m_len < sizeof(*evl) &&
680 (m = m_pullup(m, sizeof(*evl))) == NULL) {
681 #ifdef DIAGNOSTIC
682 if_printf(ifp, "cannot pullup VLAN header\n");
683 #endif
684 ifp->if_ierrors++;
685 m_freem(m);
686 CURVNET_RESTORE();
687 return;
688 }
689
690 evl = mtod(m, struct ether_vlan_header *);
691 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag);
692 m->m_flags |= M_VLANTAG;
693
694 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN,
695 ETHER_HDR_LEN - ETHER_TYPE_LEN);
696 m_adj(m, ETHER_VLAN_ENCAP_LEN);
697 eh = mtod(m, struct ether_header *);
698 }
699
700 M_SETFIB(m, ifp->if_fib);
701
702 /* Allow ng_ether(4) to claim this frame. */
703 if (IFP2AC(ifp)->ac_netgraph != NULL) {
704 KASSERT(ng_ether_input_p != NULL,
705 ("%s: ng_ether_input_p is NULL", __func__));
706 m->m_flags &= ~M_PROMISC;
707 (*ng_ether_input_p)(ifp, &m);
708 if (m == NULL) {
709 CURVNET_RESTORE();
710 return;
711 }
712 eh = mtod(m, struct ether_header *);
713 }
714
715 /*
716 * Allow if_bridge(4) to claim this frame.
717 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it
718 * and the frame should be delivered locally.
719 */
720 if (ifp->if_bridge != NULL) {
721 m->m_flags &= ~M_PROMISC;
722 BRIDGE_INPUT(ifp, m);
723 if (m == NULL) {
724 CURVNET_RESTORE();
725 return;
726 }
727 eh = mtod(m, struct ether_header *);
728 }
729
730 #if defined(INET) || defined(INET6)
731 /*
732 * Clear M_PROMISC on frame so that carp(4) will see it when the
733 * mbuf flows up to Layer 3.
734 * FreeBSD's implementation of carp(4) uses the inprotosw
735 * to dispatch IPPROTO_CARP. carp(4) also allocates its own
736 * Ethernet addresses of the form 00:00:5e:00:01:xx, which
737 * is outside the scope of the M_PROMISC test below.
738 * TODO: Maintain a hash table of ethernet addresses other than
739 * ether_dhost which may be active on this ifp.
740 */
741 if (ifp->if_carp && (*carp_forus_p)(ifp, eh->ether_dhost)) {
742 m->m_flags &= ~M_PROMISC;
743 } else
744 #endif
745 {
746 /*
747 * If the frame received was not for our MAC address, set the
748 * M_PROMISC flag on the mbuf chain. The frame may need to
749 * be seen by the rest of the Ethernet input path in case of
750 * re-entry (e.g. bridge, vlan, netgraph) but should not be
751 * seen by upper protocol layers.
752 */
753 if (!ETHER_IS_MULTICAST(eh->ether_dhost) &&
754 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0)
755 m->m_flags |= M_PROMISC;
756 }
757
758 /* First chunk of an mbuf contains good entropy */
759 if (harvest.ethernet)
760 random_harvest(m, 16, 3, 0, RANDOM_NET);
761
762 ether_demux(ifp, m);
763 CURVNET_RESTORE();
764 }
765
766 /*
767 * Ethernet input dispatch; by default, direct dispatch here regardless of
768 * global configuration.
769 */
770 static void
771 ether_nh_input(struct mbuf *m)
772 {
773
774 ether_input_internal(m->m_pkthdr.rcvif, m);
775 }
776
777 static struct netisr_handler ether_nh = {
778 .nh_name = "ether",
779 .nh_handler = ether_nh_input,
780 .nh_proto = NETISR_ETHER,
781 .nh_policy = NETISR_POLICY_SOURCE,
782 .nh_dispatch = NETISR_DISPATCH_DIRECT,
783 };
784
785 static void
786 ether_init(__unused void *arg)
787 {
788
789 netisr_register(ðer_nh);
790 }
791 SYSINIT(ether, SI_SUB_INIT_IF, SI_ORDER_ANY, ether_init, NULL);
792
793 static void
794 ether_input(struct ifnet *ifp, struct mbuf *m)
795 {
796
797 /*
798 * We will rely on rcvif being set properly in the deferred context,
799 * so assert it is correct here.
800 */
801 KASSERT(m->m_pkthdr.rcvif == ifp, ("%s: ifnet mismatch", __func__));
802
803 netisr_dispatch(NETISR_ETHER, m);
804 }
805
806 /*
807 * Upper layer processing for a received Ethernet packet.
808 */
809 void
810 ether_demux(struct ifnet *ifp, struct mbuf *m)
811 {
812 struct ether_header *eh;
813 int isr;
814 u_short ether_type;
815 #if defined(NETATALK)
816 struct llc *l;
817 #endif
818
819 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__));
820
821 #if defined(INET) || defined(INET6)
822 /*
823 * Allow dummynet and/or ipfw to claim the frame.
824 * Do not do this for PROMISC frames in case we are re-entered.
825 */
826 if (V_ip_fw_chk_ptr && V_ether_ipfw != 0 && !(m->m_flags & M_PROMISC)) {
827 if (ether_ipfw_chk(&m, NULL, 0) == 0) {
828 if (m)
829 m_freem(m); /* dropped; free mbuf chain */
830 return; /* consumed */
831 }
832 }
833 #endif
834 eh = mtod(m, struct ether_header *);
835 ether_type = ntohs(eh->ether_type);
836
837 /*
838 * If this frame has a VLAN tag other than 0, call vlan_input()
839 * if its module is loaded. Otherwise, drop.
840 */
841 if ((m->m_flags & M_VLANTAG) &&
842 EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) {
843 if (ifp->if_vlantrunk == NULL) {
844 ifp->if_noproto++;
845 m_freem(m);
846 return;
847 }
848 KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!",
849 __func__));
850 /* Clear before possibly re-entering ether_input(). */
851 m->m_flags &= ~M_PROMISC;
852 (*vlan_input_p)(ifp, m);
853 return;
854 }
855
856 /*
857 * Pass promiscuously received frames to the upper layer if the user
858 * requested this by setting IFF_PPROMISC. Otherwise, drop them.
859 */
860 if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) {
861 m_freem(m);
862 return;
863 }
864
865 /*
866 * Reset layer specific mbuf flags to avoid confusing upper layers.
867 * Strip off Ethernet header.
868 */
869 m->m_flags &= ~M_VLANTAG;
870 m->m_flags &= ~(M_PROTOFLAGS);
871 m_adj(m, ETHER_HDR_LEN);
872
873 /*
874 * Dispatch frame to upper layer.
875 */
876 switch (ether_type) {
877 #ifdef INET
878 case ETHERTYPE_IP:
879 if ((m = ip_fastforward(m)) == NULL)
880 return;
881 isr = NETISR_IP;
882 break;
883
884 case ETHERTYPE_ARP:
885 if (ifp->if_flags & IFF_NOARP) {
886 /* Discard packet if ARP is disabled on interface */
887 m_freem(m);
888 return;
889 }
890 isr = NETISR_ARP;
891 break;
892 #endif
893 #ifdef IPX
894 case ETHERTYPE_IPX:
895 if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
896 return;
897 isr = NETISR_IPX;
898 break;
899 #endif
900 #ifdef INET6
901 case ETHERTYPE_IPV6:
902 isr = NETISR_IPV6;
903 break;
904 #endif
905 #ifdef NETATALK
906 case ETHERTYPE_AT:
907 isr = NETISR_ATALK1;
908 break;
909 case ETHERTYPE_AARP:
910 isr = NETISR_AARP;
911 break;
912 #endif /* NETATALK */
913 default:
914 #ifdef IPX
915 if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
916 return;
917 #endif /* IPX */
918 #if defined(NETATALK)
919 if (ether_type > ETHERMTU)
920 goto discard;
921 l = mtod(m, struct llc *);
922 if (l->llc_dsap == LLC_SNAP_LSAP &&
923 l->llc_ssap == LLC_SNAP_LSAP &&
924 l->llc_control == LLC_UI) {
925 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code,
926 sizeof(at_org_code)) == 0 &&
927 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
928 m_adj(m, LLC_SNAPFRAMELEN);
929 isr = NETISR_ATALK2;
930 break;
931 }
932 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
933 sizeof(aarp_org_code)) == 0 &&
934 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
935 m_adj(m, LLC_SNAPFRAMELEN);
936 isr = NETISR_AARP;
937 break;
938 }
939 }
940 #endif /* NETATALK */
941 goto discard;
942 }
943 netisr_dispatch(isr, m);
944 return;
945
946 discard:
947 /*
948 * Packet is to be discarded. If netgraph is present,
949 * hand the packet to it for last chance processing;
950 * otherwise dispose of it.
951 */
952 if (IFP2AC(ifp)->ac_netgraph != NULL) {
953 KASSERT(ng_ether_input_orphan_p != NULL,
954 ("ng_ether_input_orphan_p is NULL"));
955 /*
956 * Put back the ethernet header so netgraph has a
957 * consistent view of inbound packets.
958 */
959 M_PREPEND(m, ETHER_HDR_LEN, M_DONTWAIT);
960 (*ng_ether_input_orphan_p)(ifp, m);
961 return;
962 }
963 m_freem(m);
964 }
965
966 /*
967 * Convert Ethernet address to printable (loggable) representation.
968 * This routine is for compatibility; it's better to just use
969 *
970 * printf("%6D", <pointer to address>, ":");
971 *
972 * since there's no static buffer involved.
973 */
974 char *
975 ether_sprintf(const u_char *ap)
976 {
977 static char etherbuf[18];
978 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
979 return (etherbuf);
980 }
981
982 /*
983 * Perform common duties while attaching to interface list
984 */
985 void
986 ether_ifattach(struct ifnet *ifp, const u_int8_t *lla)
987 {
988 int i;
989 struct ifaddr *ifa;
990 struct sockaddr_dl *sdl;
991
992 ifp->if_addrlen = ETHER_ADDR_LEN;
993 ifp->if_hdrlen = ETHER_HDR_LEN;
994 if_attach(ifp);
995 ifp->if_mtu = ETHERMTU;
996 ifp->if_output = ether_output;
997 ifp->if_input = ether_input;
998 ifp->if_resolvemulti = ether_resolvemulti;
999 #ifdef VIMAGE
1000 ifp->if_reassign = ether_reassign;
1001 #endif
1002 if (ifp->if_baudrate == 0)
1003 ifp->if_baudrate = IF_Mbps(10); /* just a default */
1004 ifp->if_broadcastaddr = etherbroadcastaddr;
1005
1006 ifa = ifp->if_addr;
1007 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
1008 sdl = (struct sockaddr_dl *)ifa->ifa_addr;
1009 sdl->sdl_type = IFT_ETHER;
1010 sdl->sdl_alen = ifp->if_addrlen;
1011 bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
1012
1013 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
1014 if (ng_ether_attach_p != NULL)
1015 (*ng_ether_attach_p)(ifp);
1016
1017 /* Announce Ethernet MAC address if non-zero. */
1018 for (i = 0; i < ifp->if_addrlen; i++)
1019 if (lla[i] != 0)
1020 break;
1021 if (i != ifp->if_addrlen)
1022 if_printf(ifp, "Ethernet address: %6D\n", lla, ":");
1023 }
1024
1025 /*
1026 * Perform common duties while detaching an Ethernet interface
1027 */
1028 void
1029 ether_ifdetach(struct ifnet *ifp)
1030 {
1031 if (IFP2AC(ifp)->ac_netgraph != NULL) {
1032 KASSERT(ng_ether_detach_p != NULL,
1033 ("ng_ether_detach_p is NULL"));
1034 (*ng_ether_detach_p)(ifp);
1035 }
1036
1037 bpfdetach(ifp);
1038 if_detach(ifp);
1039 }
1040
1041 #ifdef VIMAGE
1042 void
1043 ether_reassign(struct ifnet *ifp, struct vnet *new_vnet, char *unused __unused)
1044 {
1045
1046 if (IFP2AC(ifp)->ac_netgraph != NULL) {
1047 KASSERT(ng_ether_detach_p != NULL,
1048 ("ng_ether_detach_p is NULL"));
1049 (*ng_ether_detach_p)(ifp);
1050 }
1051
1052 if (ng_ether_attach_p != NULL) {
1053 CURVNET_SET_QUIET(new_vnet);
1054 (*ng_ether_attach_p)(ifp);
1055 CURVNET_RESTORE();
1056 }
1057 }
1058 #endif
1059
1060 SYSCTL_DECL(_net_link);
1061 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
1062 #if defined(INET) || defined(INET6)
1063 SYSCTL_VNET_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
1064 &VNET_NAME(ether_ipfw), 0, "Pass ether pkts through firewall");
1065 #endif
1066
1067 #if 0
1068 /*
1069 * This is for reference. We have a table-driven version
1070 * of the little-endian crc32 generator, which is faster
1071 * than the double-loop.
1072 */
1073 uint32_t
1074 ether_crc32_le(const uint8_t *buf, size_t len)
1075 {
1076 size_t i;
1077 uint32_t crc;
1078 int bit;
1079 uint8_t data;
1080
1081 crc = 0xffffffff; /* initial value */
1082
1083 for (i = 0; i < len; i++) {
1084 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
1085 carry = (crc ^ data) & 1;
1086 crc >>= 1;
1087 if (carry)
1088 crc = (crc ^ ETHER_CRC_POLY_LE);
1089 }
1090 }
1091
1092 return (crc);
1093 }
1094 #else
1095 uint32_t
1096 ether_crc32_le(const uint8_t *buf, size_t len)
1097 {
1098 static const uint32_t crctab[] = {
1099 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1100 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1101 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1102 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1103 };
1104 size_t i;
1105 uint32_t crc;
1106
1107 crc = 0xffffffff; /* initial value */
1108
1109 for (i = 0; i < len; i++) {
1110 crc ^= buf[i];
1111 crc = (crc >> 4) ^ crctab[crc & 0xf];
1112 crc = (crc >> 4) ^ crctab[crc & 0xf];
1113 }
1114
1115 return (crc);
1116 }
1117 #endif
1118
1119 uint32_t
1120 ether_crc32_be(const uint8_t *buf, size_t len)
1121 {
1122 size_t i;
1123 uint32_t crc, carry;
1124 int bit;
1125 uint8_t data;
1126
1127 crc = 0xffffffff; /* initial value */
1128
1129 for (i = 0; i < len; i++) {
1130 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
1131 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
1132 crc <<= 1;
1133 if (carry)
1134 crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1135 }
1136 }
1137
1138 return (crc);
1139 }
1140
1141 int
1142 ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
1143 {
1144 struct ifaddr *ifa = (struct ifaddr *) data;
1145 struct ifreq *ifr = (struct ifreq *) data;
1146 int error = 0;
1147
1148 switch (command) {
1149 case SIOCSIFADDR:
1150 ifp->if_flags |= IFF_UP;
1151
1152 switch (ifa->ifa_addr->sa_family) {
1153 #ifdef INET
1154 case AF_INET:
1155 ifp->if_init(ifp->if_softc); /* before arpwhohas */
1156 arp_ifinit(ifp, ifa);
1157 break;
1158 #endif
1159 #ifdef IPX
1160 /*
1161 * XXX - This code is probably wrong
1162 */
1163 case AF_IPX:
1164 {
1165 struct ipx_addr *ina = &(IA_SIPX(ifa)->sipx_addr);
1166
1167 if (ipx_nullhost(*ina))
1168 ina->x_host =
1169 *(union ipx_host *)
1170 IF_LLADDR(ifp);
1171 else {
1172 bcopy((caddr_t) ina->x_host.c_host,
1173 (caddr_t) IF_LLADDR(ifp),
1174 ETHER_ADDR_LEN);
1175 }
1176
1177 /*
1178 * Set new address
1179 */
1180 ifp->if_init(ifp->if_softc);
1181 break;
1182 }
1183 #endif
1184 default:
1185 ifp->if_init(ifp->if_softc);
1186 break;
1187 }
1188 break;
1189
1190 case SIOCGIFADDR:
1191 {
1192 struct sockaddr *sa;
1193
1194 sa = (struct sockaddr *) & ifr->ifr_data;
1195 bcopy(IF_LLADDR(ifp),
1196 (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
1197 }
1198 break;
1199
1200 case SIOCSIFMTU:
1201 /*
1202 * Set the interface MTU.
1203 */
1204 if (ifr->ifr_mtu > ETHERMTU) {
1205 error = EINVAL;
1206 } else {
1207 ifp->if_mtu = ifr->ifr_mtu;
1208 }
1209 break;
1210 default:
1211 error = EINVAL; /* XXX netbsd has ENOTTY??? */
1212 break;
1213 }
1214 return (error);
1215 }
1216
1217 static int
1218 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
1219 struct sockaddr *sa)
1220 {
1221 struct sockaddr_dl *sdl;
1222 #ifdef INET
1223 struct sockaddr_in *sin;
1224 #endif
1225 #ifdef INET6
1226 struct sockaddr_in6 *sin6;
1227 #endif
1228 u_char *e_addr;
1229
1230 switch(sa->sa_family) {
1231 case AF_LINK:
1232 /*
1233 * No mapping needed. Just check that it's a valid MC address.
1234 */
1235 sdl = (struct sockaddr_dl *)sa;
1236 e_addr = LLADDR(sdl);
1237 if (!ETHER_IS_MULTICAST(e_addr))
1238 return EADDRNOTAVAIL;
1239 *llsa = 0;
1240 return 0;
1241
1242 #ifdef INET
1243 case AF_INET:
1244 sin = (struct sockaddr_in *)sa;
1245 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1246 return EADDRNOTAVAIL;
1247 sdl = malloc(sizeof *sdl, M_IFMADDR,
1248 M_NOWAIT|M_ZERO);
1249 if (sdl == NULL)
1250 return ENOMEM;
1251 sdl->sdl_len = sizeof *sdl;
1252 sdl->sdl_family = AF_LINK;
1253 sdl->sdl_index = ifp->if_index;
1254 sdl->sdl_type = IFT_ETHER;
1255 sdl->sdl_alen = ETHER_ADDR_LEN;
1256 e_addr = LLADDR(sdl);
1257 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1258 *llsa = (struct sockaddr *)sdl;
1259 return 0;
1260 #endif
1261 #ifdef INET6
1262 case AF_INET6:
1263 sin6 = (struct sockaddr_in6 *)sa;
1264 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1265 /*
1266 * An IP6 address of 0 means listen to all
1267 * of the Ethernet multicast address used for IP6.
1268 * (This is used for multicast routers.)
1269 */
1270 ifp->if_flags |= IFF_ALLMULTI;
1271 *llsa = 0;
1272 return 0;
1273 }
1274 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1275 return EADDRNOTAVAIL;
1276 sdl = malloc(sizeof *sdl, M_IFMADDR,
1277 M_NOWAIT|M_ZERO);
1278 if (sdl == NULL)
1279 return (ENOMEM);
1280 sdl->sdl_len = sizeof *sdl;
1281 sdl->sdl_family = AF_LINK;
1282 sdl->sdl_index = ifp->if_index;
1283 sdl->sdl_type = IFT_ETHER;
1284 sdl->sdl_alen = ETHER_ADDR_LEN;
1285 e_addr = LLADDR(sdl);
1286 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1287 *llsa = (struct sockaddr *)sdl;
1288 return 0;
1289 #endif
1290
1291 default:
1292 /*
1293 * Well, the text isn't quite right, but it's the name
1294 * that counts...
1295 */
1296 return EAFNOSUPPORT;
1297 }
1298 }
1299
1300 static void*
1301 ether_alloc(u_char type, struct ifnet *ifp)
1302 {
1303 struct arpcom *ac;
1304
1305 ac = malloc(sizeof(struct arpcom), M_ARPCOM, M_WAITOK | M_ZERO);
1306 ac->ac_ifp = ifp;
1307
1308 return (ac);
1309 }
1310
1311 static void
1312 ether_free(void *com, u_char type)
1313 {
1314
1315 free(com, M_ARPCOM);
1316 }
1317
1318 static int
1319 ether_modevent(module_t mod, int type, void *data)
1320 {
1321
1322 switch (type) {
1323 case MOD_LOAD:
1324 if_register_com_alloc(IFT_ETHER, ether_alloc, ether_free);
1325 break;
1326 case MOD_UNLOAD:
1327 if_deregister_com_alloc(IFT_ETHER);
1328 break;
1329 default:
1330 return EOPNOTSUPP;
1331 }
1332
1333 return (0);
1334 }
1335
1336 static moduledata_t ether_mod = {
1337 "ether",
1338 ether_modevent,
1339 0
1340 };
1341
1342 void
1343 ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen)
1344 {
1345 struct ether_vlan_header vlan;
1346 struct mbuf mv, mb;
1347
1348 KASSERT((m->m_flags & M_VLANTAG) != 0,
1349 ("%s: vlan information not present", __func__));
1350 KASSERT(m->m_len >= sizeof(struct ether_header),
1351 ("%s: mbuf not large enough for header", __func__));
1352 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header));
1353 vlan.evl_proto = vlan.evl_encap_proto;
1354 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN);
1355 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag);
1356 m->m_len -= sizeof(struct ether_header);
1357 m->m_data += sizeof(struct ether_header);
1358 /*
1359 * If a data link has been supplied by the caller, then we will need to
1360 * re-create a stack allocated mbuf chain with the following structure:
1361 *
1362 * (1) mbuf #1 will contain the supplied data link
1363 * (2) mbuf #2 will contain the vlan header
1364 * (3) mbuf #3 will contain the original mbuf's packet data
1365 *
1366 * Otherwise, submit the packet and vlan header via bpf_mtap2().
1367 */
1368 if (data != NULL) {
1369 mv.m_next = m;
1370 mv.m_data = (caddr_t)&vlan;
1371 mv.m_len = sizeof(vlan);
1372 mb.m_next = &mv;
1373 mb.m_data = data;
1374 mb.m_len = dlen;
1375 bpf_mtap(bp, &mb);
1376 } else
1377 bpf_mtap2(bp, &vlan, sizeof(vlan), m);
1378 m->m_len += sizeof(struct ether_header);
1379 m->m_data -= sizeof(struct ether_header);
1380 }
1381
1382 struct mbuf *
1383 ether_vlanencap(struct mbuf *m, uint16_t tag)
1384 {
1385 struct ether_vlan_header *evl;
1386
1387 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_DONTWAIT);
1388 if (m == NULL)
1389 return (NULL);
1390 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */
1391
1392 if (m->m_len < sizeof(*evl)) {
1393 m = m_pullup(m, sizeof(*evl));
1394 if (m == NULL)
1395 return (NULL);
1396 }
1397
1398 /*
1399 * Transform the Ethernet header into an Ethernet header
1400 * with 802.1Q encapsulation.
1401 */
1402 evl = mtod(m, struct ether_vlan_header *);
1403 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN,
1404 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN);
1405 evl->evl_encap_proto = htons(ETHERTYPE_VLAN);
1406 evl->evl_tag = htons(tag);
1407 return (m);
1408 }
1409
1410 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY);
1411 MODULE_VERSION(ether, 1);
Cache object: 3567670855543f0c00c7c4f4ca9bd4a3
|