- Test m_pkthdr.fw_flags against DUMMYNET_MBUF_TAGGED before trying to locate
[dragonfly.git] / sys / net / if_ethersubr.c
CommitLineData
984263bc
MD
1/*
2 * Copyright (c) 1982, 1989, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * @(#)if_ethersubr.c 8.1 (Berkeley) 6/10/93
34 * $FreeBSD: src/sys/net/if_ethersubr.c,v 1.70.2.33 2003/04/28 15:45:53 archie Exp $
eb241549 35 * $DragonFly: src/sys/net/if_ethersubr.c,v 1.84 2008/08/23 04:12:23 sephe Exp $
984263bc
MD
36 */
37
38#include "opt_atalk.h"
39#include "opt_inet.h"
40#include "opt_inet6.h"
41#include "opt_ipx.h"
9b42cabe 42#include "opt_mpls.h"
984263bc 43#include "opt_netgraph.h"
0d16ba1d 44#include "opt_carp.h"
68b67450 45#include "opt_ethernet.h"
984263bc
MD
46
47#include <sys/param.h>
48#include <sys/systm.h>
68b67450 49#include <sys/globaldata.h>
984263bc
MD
50#include <sys/kernel.h>
51#include <sys/malloc.h>
52#include <sys/mbuf.h>
68b67450 53#include <sys/msgport.h>
984263bc
MD
54#include <sys/socket.h>
55#include <sys/sockio.h>
56#include <sys/sysctl.h>
68b67450
SZ
57#include <sys/thread.h>
58#include <sys/thread2.h>
984263bc
MD
59
60#include <net/if.h>
61#include <net/netisr.h>
62#include <net/route.h>
63#include <net/if_llc.h>
64#include <net/if_dl.h>
65#include <net/if_types.h>
4d723e5a 66#include <net/ifq_var.h>
984263bc
MD
67#include <net/bpf.h>
68#include <net/ethernet.h>
e6b5847c 69#include <net/vlan/if_vlan_ether.h>
29bc1092 70#include <net/netmsg2.h>
984263bc
MD
71
72#if defined(INET) || defined(INET6)
73#include <netinet/in.h>
74#include <netinet/in_var.h>
75#include <netinet/if_ether.h>
1f2de5d4
MD
76#include <net/ipfw/ip_fw.h>
77#include <net/dummynet/ip_dummynet.h>
984263bc
MD
78#endif
79#ifdef INET6
80#include <netinet6/nd6.h>
81#endif
82
0d16ba1d
MD
83#ifdef CARP
84#include <netinet/ip_carp.h>
85#endif
86
984263bc 87#ifdef IPX
d2438d69
MD
88#include <netproto/ipx/ipx.h>
89#include <netproto/ipx/ipx_if.h>
32211831 90int (*ef_inputp)(struct ifnet*, const struct ether_header *eh, struct mbuf *m);
f23061d4
JH
91int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, struct sockaddr *dst,
92 short *tp, int *hlen);
984263bc
MD
93#endif
94
95#ifdef NS
96#include <netns/ns.h>
97#include <netns/ns_if.h>
98ushort ns_nettype;
99int ether_outputdebug = 0;
100int ether_inputdebug = 0;
101#endif
102
103#ifdef NETATALK
d2438d69
MD
104#include <netproto/atalk/at.h>
105#include <netproto/atalk/at_var.h>
106#include <netproto/atalk/at_extern.h>
984263bc 107
f23061d4
JH
108#define llc_snap_org_code llc_un.type_snap.org_code
109#define llc_snap_ether_type llc_un.type_snap.ether_type
984263bc
MD
110
111extern u_char at_org_code[3];
112extern u_char aarp_org_code[3];
113#endif /* NETATALK */
114
9b42cabe
NA
115#ifdef MPLS
116#include <netproto/mpls/mpls.h>
117#endif
118
984263bc 119/* netgraph node hooks for ng_ether(4) */
601fa0f9 120void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp);
984263bc 121void (*ng_ether_input_orphan_p)(struct ifnet *ifp,
36f48e5e 122 struct mbuf *m, const struct ether_header *eh);
984263bc
MD
123int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
124void (*ng_ether_attach_p)(struct ifnet *ifp);
125void (*ng_ether_detach_p)(struct ifnet *ifp);
126
7af50411 127void (*vlan_input2_p)(struct mbuf *);
3013ac0e 128
5fe66e68
JH
129static int ether_output(struct ifnet *, struct mbuf *, struct sockaddr *,
130 struct rtentry *);
9b77ea6e
SZ
131static void ether_restore_header(struct mbuf **, const struct ether_header *,
132 const struct ether_header *);
984263bc 133
f23061d4 134/*
a8d45119 135 * if_bridge support
f23061d4 136 */
db37145f 137struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *);
eb366364 138int (*bridge_output_p)(struct ifnet *, struct mbuf *);
db37145f
SS
139void (*bridge_dn_p)(struct mbuf *, struct ifnet *);
140
5fe66e68
JH
141static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
142 struct sockaddr *);
143
144const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = {
c401f0fd
JS
145 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
146};
147
5fe66e68 148#define gotoerr(e) do { error = (e); goto bad; } while (0)
f23061d4 149#define IFP2AC(ifp) ((struct arpcom *)(ifp))
984263bc 150
5fe66e68 151static boolean_t ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
7c5bb821 152 struct ip_fw **rule,
90ca9293 153 const struct ether_header *eh);
5fe66e68 154
984263bc 155static int ether_ipfw;
9b77ea6e
SZ
156static u_int ether_restore_hdr;
157static u_int ether_prepend_hdr;
158
5fe66e68
JH
159SYSCTL_DECL(_net_link);
160SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
161SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
162 &ether_ipfw, 0, "Pass ether pkts through firewall");
9b77ea6e
SZ
163SYSCTL_UINT(_net_link_ether, OID_AUTO, restore_hdr, CTLFLAG_RW,
164 &ether_restore_hdr, 0, "# of ether header restoration");
165SYSCTL_UINT(_net_link_ether, OID_AUTO, prepend_hdr, CTLFLAG_RW,
166 &ether_prepend_hdr, 0,
167 "# of ether header restoration which prepends mbuf");
984263bc
MD
168
169/*
170 * Ethernet output routine.
171 * Encapsulate a packet of type family for the local net.
172 * Use trailer local net encapsulation if enough data in first
173 * packet leaves a multiple of 512 bytes of data in remainder.
174 * Assumes that ifp is actually pointer to arpcom structure.
175 */
3013ac0e
JS
176static int
177ether_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst,
f23061d4 178 struct rtentry *rt)
984263bc 179{
f23061d4
JH
180 struct ether_header *eh, *deh;
181 u_char *edst;
984263bc 182 int loop_copy = 0;
f23061d4 183 int hlen = ETHER_HDR_LEN; /* link layer header length */
984263bc 184 struct arpcom *ac = IFP2AC(ifp);
f23061d4 185 int error;
984263bc 186
57dff79c
SZ
187 ASSERT_NOT_SERIALIZED(ifp->if_serializer);
188
3a593c54
MD
189 if (ifp->if_flags & IFF_MONITOR)
190 gotoerr(ENETDOWN);
f23061d4
JH
191 if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING))
192 gotoerr(ENETDOWN);
193
194 M_PREPEND(m, sizeof(struct ether_header), MB_DONTWAIT);
195 if (m == NULL)
5fe66e68 196 return (ENOBUFS);
f23061d4
JH
197 eh = mtod(m, struct ether_header *);
198 edst = eh->ether_dhost;
199
5fe66e68
JH
200 /*
201 * Fill in the destination ethernet address and frame type.
202 */
984263bc
MD
203 switch (dst->sa_family) {
204#ifdef INET
205 case AF_INET:
f23061d4 206 if (!arpresolve(ifp, rt, m, dst, edst))
984263bc 207 return (0); /* if not yet resolved */
cb8d752c
NA
208#ifdef MPLS
209 if (m->m_flags & M_MPLSLABELED)
210 eh->ether_type = htons(ETHERTYPE_MPLS);
211 else
212#endif
213 eh->ether_type = htons(ETHERTYPE_IP);
984263bc
MD
214 break;
215#endif
216#ifdef INET6
217 case AF_INET6:
f23061d4 218 if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, edst))
5fe66e68 219 return (0); /* Something bad happenned. */
f23061d4 220 eh->ether_type = htons(ETHERTYPE_IPV6);
984263bc
MD
221 break;
222#endif
223#ifdef IPX
224 case AF_IPX:
f23061d4
JH
225 if (ef_outputp != NULL) {
226 error = ef_outputp(ifp, &m, dst, &eh->ether_type,
227 &hlen);
228 if (error)
229 goto bad;
230 } else {
231 eh->ether_type = htons(ETHERTYPE_IPX);
232 bcopy(&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
233 edst, ETHER_ADDR_LEN);
234 }
984263bc
MD
235 break;
236#endif
237#ifdef NETATALK
f23061d4
JH
238 case AF_APPLETALK: {
239 struct at_ifaddr *aa;
240
241 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) {
242 error = 0; /* XXX */
243 goto bad;
244 }
245 /*
246 * In the phase 2 case, need to prepend an mbuf for
247 * the llc header. Since we must preserve the value
248 * of m, which is passed to us by value, we m_copy()
249 * the first mbuf, and use it for our llc header.
250 */
251 if (aa->aa_flags & AFA_PHASE2) {
252 struct llc llc;
253
254 M_PREPEND(m, sizeof(struct llc), MB_DONTWAIT);
5fe66e68
JH
255 eh = mtod(m, struct ether_header *);
256 edst = eh->ether_dhost;
f23061d4
JH
257 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
258 llc.llc_control = LLC_UI;
259 bcopy(at_org_code, llc.llc_snap_org_code,
260 sizeof at_org_code);
261 llc.llc_snap_ether_type = htons(ETHERTYPE_AT);
5fe66e68
JH
262 bcopy(&llc,
263 mtod(m, caddr_t) + sizeof(struct ether_header),
264 sizeof(struct llc));
f23061d4
JH
265 eh->ether_type = htons(m->m_pkthdr.len);
266 hlen = sizeof(struct llc) + ETHER_HDR_LEN;
267 } else {
268 eh->ether_type = htons(ETHERTYPE_AT);
269 }
270 if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst))
271 return (0);
272 break;
984263bc 273 }
5fe66e68 274#endif
984263bc
MD
275#ifdef NS
276 case AF_NS:
f23061d4 277 switch(ns_nettype) {
984263bc 278 default:
f23061d4
JH
279 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
280 eh->ether_type = 0x8137;
984263bc 281 break;
f23061d4
JH
282 case 0x0: /* Novell 802.3 */
283 eh->ether_type = htons(m->m_pkthdr.len);
984263bc 284 break;
f23061d4
JH
285 case 0xe0e0: /* Novell 802.2 and Token-Ring */
286 M_PREPEND(m, 3, MB_DONTWAIT);
5fe66e68
JH
287 eh = mtod(m, struct ether_header *);
288 edst = eh->ether_dhost;
f23061d4 289 eh->ether_type = htons(m->m_pkthdr.len);
5fe66e68 290 cp = mtod(m, u_char *) + sizeof(struct ether_header);
984263bc
MD
291 *cp++ = 0xE0;
292 *cp++ = 0xE0;
293 *cp++ = 0x03;
294 break;
295 }
f23061d4
JH
296 bcopy(&(((struct sockaddr_ns *)dst)->sns_addr.x_host), edst,
297 ETHER_ADDR_LEN);
984263bc
MD
298 /*
299 * XXX if ns_thishost is the same as the node's ethernet
300 * address then just the default code will catch this anyhow.
301 * So I'm not sure if this next clause should be here at all?
302 * [JRE]
303 */
f23061d4 304 if (bcmp(edst, &ns_thishost, ETHER_ADDR_LEN) == 0) {
984263bc 305 m->m_pkthdr.rcvif = ifp;
8bde602d 306 netisr_dispatch(NETISR_NS, m);
984263bc
MD
307 return (error);
308 }
f23061d4 309 if (bcmp(edst, &ns_broadhost, ETHER_ADDR_LEN) == 0)
984263bc 310 m->m_flags |= M_BCAST;
984263bc 311 break;
5fe66e68 312#endif
984263bc 313 case pseudo_AF_HDRCMPLT:
984263bc
MD
314 case AF_UNSPEC:
315 loop_copy = -1; /* if this is for us, don't do it */
f23061d4
JH
316 deh = (struct ether_header *)dst->sa_data;
317 memcpy(edst, deh->ether_dhost, ETHER_ADDR_LEN);
318 eh->ether_type = deh->ether_type;
984263bc
MD
319 break;
320
321 default:
8f0777ca 322 if_printf(ifp, "can't handle af%d\n", dst->sa_family);
f23061d4 323 gotoerr(EAFNOSUPPORT);
984263bc
MD
324 }
325
f23061d4
JH
326 if (dst->sa_family == pseudo_AF_HDRCMPLT) /* unlikely */
327 memcpy(eh->ether_shost,
328 ((struct ether_header *)dst->sa_data)->ether_shost,
329 ETHER_ADDR_LEN);
984263bc 330 else
f23061d4 331 memcpy(eh->ether_shost, ac->ac_enaddr, ETHER_ADDR_LEN);
984263bc 332
db37145f
SS
333 /*
334 * Bridges require special output handling.
335 */
336 if (ifp->if_bridge) {
8f0777ca
SZ
337 KASSERT(bridge_output_p != NULL,
338 ("%s: if_bridge not loaded!", __func__));
ad8c8b44 339 return bridge_output_p(ifp, m);
db37145f
SS
340 }
341
984263bc
MD
342 /*
343 * If a simplex interface, and the packet is being sent to our
344 * Ethernet address or a broadcast address, loopback a copy.
345 * XXX To make a simplex device behave exactly like a duplex
346 * device, we should copy in the case of sending to our own
347 * ethernet address (thus letting the original actually appear
348 * on the wire). However, we don't do that here for security
349 * reasons and compatibility with the original behavior.
350 */
351 if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) {
352 int csum_flags = 0;
353
354 if (m->m_pkthdr.csum_flags & CSUM_IP)
f23061d4 355 csum_flags |= (CSUM_IP_CHECKED | CSUM_IP_VALID);
984263bc 356 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
f23061d4 357 csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR);
984263bc
MD
358 if ((m->m_flags & M_BCAST) || (loop_copy > 0)) {
359 struct mbuf *n;
360
f23061d4 361 if ((n = m_copypacket(m, MB_DONTWAIT)) != NULL) {
984263bc
MD
362 n->m_pkthdr.csum_flags |= csum_flags;
363 if (csum_flags & CSUM_DATA_VALID)
364 n->m_pkthdr.csum_data = 0xffff;
f23061d4 365 if_simloop(ifp, n, dst->sa_family, hlen);
984263bc
MD
366 } else
367 ifp->if_iqdrops++;
f23061d4
JH
368 } else if (bcmp(eh->ether_dhost, eh->ether_shost,
369 ETHER_ADDR_LEN) == 0) {
984263bc
MD
370 m->m_pkthdr.csum_flags |= csum_flags;
371 if (csum_flags & CSUM_DATA_VALID)
372 m->m_pkthdr.csum_data = 0xffff;
f23061d4 373 if_simloop(ifp, m, dst->sa_family, hlen);
984263bc
MD
374 return (0); /* XXX */
375 }
376 }
377
0d16ba1d
MD
378#ifdef CARP
379 if (ifp->if_carp && (error = carp_output(ifp, m, dst, NULL)))
380 goto bad;
381#endif
382
383
984263bc
MD
384 /* Handle ng_ether(4) processing, if any */
385 if (ng_ether_output_p != NULL) {
0c3c561c
JH
386 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0)
387 goto bad;
984263bc
MD
388 if (m == NULL)
389 return (0);
390 }
391
392 /* Continue with link-layer output */
393 return ether_output_frame(ifp, m);
0c3c561c
JH
394
395bad:
396 m_freem(m);
397 return (error);
984263bc
MD
398}
399
400/*
401 * Ethernet link layer output routine to send a raw frame to the device.
402 *
403 * This assumes that the 14 byte Ethernet header is present and contiguous
a8d45119 404 * in the first mbuf.
984263bc
MD
405 */
406int
f23061d4 407ether_output_frame(struct ifnet *ifp, struct mbuf *m)
984263bc 408{
f23061d4 409 struct ip_fw *rule = NULL;
984263bc 410 int error = 0;
4d723e5a 411 struct altq_pktattr pktattr;
984263bc 412
57dff79c
SZ
413 ASSERT_NOT_SERIALIZED(ifp->if_serializer);
414
eb241549
SZ
415 if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
416 struct m_tag *mtag;
417
418 /* Extract info from dummynet tag */
419 mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL);
420 KKASSERT(mtag != NULL);
84a3e25a 421 rule = ((struct dn_pkt *)m_tag_data(mtag))->dn_priv;
eb241549 422 KKASSERT(rule != NULL);
4c7020ad
SZ
423
424 m_tag_delete(m, mtag);
eb241549 425 m->m_pkthdr.fw_flags &= ~DUMMYNET_MBUF_TAGGED;
5fe66e68 426 }
984263bc 427
4d723e5a
JS
428 if (ifq_is_enabled(&ifp->if_snd))
429 altq_etherclassify(&ifp->if_snd, m, &pktattr);
4986965b 430 crit_enter();
984263bc
MD
431 if (IPFW_LOADED && ether_ipfw != 0) {
432 struct ether_header save_eh, *eh;
433
434 eh = mtod(m, struct ether_header *);
435 save_eh = *eh;
436 m_adj(m, ETHER_HDR_LEN);
90ca9293 437 if (!ether_ipfw_chk(&m, ifp, &rule, eh)) {
4986965b 438 crit_exit();
5fe66e68 439 if (m != NULL) {
984263bc 440 m_freem(m);
f23061d4 441 return ENOBUFS; /* pkt dropped */
984263bc
MD
442 } else
443 return 0; /* consumed e.g. in a pipe */
444 }
9b77ea6e 445
984263bc 446 /* packet was ok, restore the ethernet header */
9b77ea6e
SZ
447 ether_restore_header(&m, eh, &save_eh);
448 if (m == NULL) {
449 crit_exit();
450 return ENOBUFS;
984263bc
MD
451 }
452 }
78195a76 453 crit_exit();
984263bc
MD
454
455 /*
456 * Queue message on interface, update output statistics if
457 * successful, and start output if interface not yet active.
458 */
9db4b353 459 error = ifq_dispatch(ifp, m, &pktattr);
984263bc
MD
460 return (error);
461}
462
463/*
464 * ipfw processing for ethernet packets (in and out).
5fe66e68 465 * The second parameter is NULL from ether_demux(), and ifp from
a8d45119 466 * ether_output_frame().
984263bc 467 */
5fe66e68 468static boolean_t
90ca9293
SZ
469ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, struct ip_fw **rule,
470 const struct ether_header *eh)
984263bc
MD
471{
472 struct ether_header save_eh = *eh; /* might be a ptr in m */
984263bc 473 struct ip_fw_args args;
e5ecc832 474 struct m_tag *mtag;
f23061d4 475 int i;
984263bc
MD
476
477 if (*rule != NULL && fw_one_pass)
5fe66e68 478 return TRUE; /* dummynet packet, already partially processed */
984263bc
MD
479
480 /*
90ca9293 481 * I need some amount of data to be contiguous.
984263bc 482 */
f23061d4 483 i = min((*m0)->m_pkthdr.len, max_protohdr);
90ca9293 484 if ((*m0)->m_len < i) {
984263bc
MD
485 *m0 = m_pullup(*m0, i);
486 if (*m0 == NULL)
5fe66e68 487 return FALSE;
984263bc
MD
488 }
489
5de23090
SZ
490 /*
491 * Clean up tags
492 */
e5ecc832
JS
493 if ((mtag = m_tag_find(*m0, PACKET_TAG_IPFW_DIVERT, NULL)) != NULL)
494 m_tag_delete(*m0, mtag);
5de23090
SZ
495 if ((*m0)->m_pkthdr.fw_flags & IPFORWARD_MBUF_TAGGED) {
496 mtag = m_tag_find(*m0, PACKET_TAG_IPFORWARD, NULL);
497 KKASSERT(mtag != NULL);
498 m_tag_delete(*m0, mtag);
499 (*m0)->m_pkthdr.fw_flags &= ~IPFORWARD_MBUF_TAGGED;
500 }
501
502 args.m = *m0; /* the packet we are looking at */
503 args.oif = dst; /* destination, if any */
984263bc 504 args.rule = *rule; /* matching rule to restart */
984263bc
MD
505 args.eh = &save_eh; /* MAC header for bridged/MAC packets */
506 i = ip_fw_chk_ptr(&args);
507 *m0 = args.m;
508 *rule = args.rule;
509
5fe66e68
JH
510 if ((i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL) /* drop */
511 return FALSE;
984263bc 512
5fe66e68
JH
513 if (i == 0) /* a PASS rule. */
514 return TRUE;
984263bc 515
84a3e25a 516 if (i & IP_FW_PORT_DYNT_FLAG) {
984263bc
MD
517 /*
518 * Pass the pkt to dummynet, which consumes it.
984263bc 519 */
90ca9293 520 struct mbuf *m;
984263bc 521
90ca9293
SZ
522 m = *m0; /* pass the original to dummynet */
523 *m0 = NULL; /* and nothing back to the caller */
9b77ea6e
SZ
524
525 ether_restore_header(&m, eh, &save_eh);
526 if (m == NULL)
527 return FALSE;
528
84a3e25a
SZ
529 ip_fw_dn_io_ptr(m, (i & 0xffff),
530 dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
5fe66e68 531 return FALSE;
984263bc
MD
532 }
533 /*
534 * XXX at some point add support for divert/forward actions.
535 * If none of the above matches, we have to drop the pkt.
536 */
5fe66e68 537 return FALSE;
984263bc
MD
538}
539
4853cd0f 540static void
68b67450
SZ
541ether_input(struct ifnet *ifp, struct mbuf *m)
542{
297c8124 543 ether_input_chain2(ifp, m, NULL);
984263bc
MD
544}
545
546/*
547 * Perform common duties while attaching to interface list
548 */
549void
78195a76 550ether_ifattach(struct ifnet *ifp, uint8_t *lla, lwkt_serialize_t serializer)
c0f6c904 551{
78195a76
MD
552 ether_ifattach_bpf(ifp, lla, DLT_EN10MB, sizeof(struct ether_header),
553 serializer);
c0f6c904
JS
554}
555
556void
78195a76
MD
557ether_ifattach_bpf(struct ifnet *ifp, uint8_t *lla, u_int dlt, u_int hdrlen,
558 lwkt_serialize_t serializer)
984263bc 559{
82ed7fc2 560 struct sockaddr_dl *sdl;
984263bc
MD
561
562 ifp->if_type = IFT_ETHER;
c401f0fd 563 ifp->if_addrlen = ETHER_ADDR_LEN;
5fe66e68 564 ifp->if_hdrlen = ETHER_HDR_LEN;
78195a76 565 if_attach(ifp, serializer);
984263bc 566 ifp->if_mtu = ETHERMTU;
984263bc 567 if (ifp->if_baudrate == 0)
f23061d4 568 ifp->if_baudrate = 10000000;
5fe66e68 569 ifp->if_output = ether_output;
0b076e92 570 ifp->if_input = ether_input;
5fe66e68
JH
571 ifp->if_resolvemulti = ether_resolvemulti;
572 ifp->if_broadcastaddr = etherbroadcastaddr;
f2682cb9 573 sdl = IF_LLSOCKADDR(ifp);
984263bc
MD
574 sdl->sdl_type = IFT_ETHER;
575 sdl->sdl_alen = ifp->if_addrlen;
0a8b5977 576 bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
c568d5be
JS
577 /*
578 * XXX Keep the current drivers happy.
579 * XXX Remove once all drivers have been cleaned up
580 */
581 if (lla != IFP2AC(ifp)->ac_enaddr)
582 bcopy(lla, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen);
c0f6c904 583 bpfattach(ifp, dlt, hdrlen);
984263bc
MD
584 if (ng_ether_attach_p != NULL)
585 (*ng_ether_attach_p)(ifp);
267caeeb
JS
586
587 if_printf(ifp, "MAC address: %6D\n", lla, ":");
984263bc
MD
588}
589
590/*
591 * Perform common duties while detaching an Ethernet interface
592 */
593void
0a8b5977 594ether_ifdetach(struct ifnet *ifp)
984263bc 595{
45b8be9e 596 if_down(ifp);
45b8be9e 597
984263bc
MD
598 if (ng_ether_detach_p != NULL)
599 (*ng_ether_detach_p)(ifp);
0a8b5977 600 bpfdetach(ifp);
984263bc 601 if_detach(ifp);
984263bc
MD
602}
603
984263bc 604int
f23061d4 605ether_ioctl(struct ifnet *ifp, int command, caddr_t data)
984263bc
MD
606{
607 struct ifaddr *ifa = (struct ifaddr *) data;
608 struct ifreq *ifr = (struct ifreq *) data;
609 int error = 0;
610
cbf2eda6
SZ
611#define IF_INIT(ifp) \
612do { \
613 if (((ifp)->if_flags & IFF_UP) == 0) { \
614 (ifp)->if_flags |= IFF_UP; \
615 (ifp)->if_init((ifp)->if_softc); \
616 } \
617} while (0)
618
78195a76
MD
619 ASSERT_SERIALIZED(ifp->if_serializer);
620
984263bc
MD
621 switch (command) {
622 case SIOCSIFADDR:
984263bc
MD
623 switch (ifa->ifa_addr->sa_family) {
624#ifdef INET
625 case AF_INET:
cbf2eda6 626 IF_INIT(ifp); /* before arpwhohas */
984263bc
MD
627 arp_ifinit(ifp, ifa);
628 break;
629#endif
630#ifdef IPX
631 /*
632 * XXX - This code is probably wrong
633 */
634 case AF_IPX:
635 {
f23061d4 636 struct ipx_addr *ina = &IA_SIPX(ifa)->sipx_addr;
984263bc
MD
637 struct arpcom *ac = IFP2AC(ifp);
638
639 if (ipx_nullhost(*ina))
f23061d4
JH
640 ina->x_host = *(union ipx_host *) ac->ac_enaddr;
641 else
642 bcopy(ina->x_host.c_host, ac->ac_enaddr,
643 sizeof ac->ac_enaddr);
984263bc 644
cbf2eda6 645 IF_INIT(ifp); /* Set new address. */
984263bc
MD
646 break;
647 }
648#endif
649#ifdef NS
650 /*
651 * XXX - This code is probably wrong
652 */
653 case AF_NS:
654 {
82ed7fc2 655 struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr);
984263bc
MD
656 struct arpcom *ac = IFP2AC(ifp);
657
658 if (ns_nullhost(*ina))
f23061d4
JH
659 ina->x_host = *(union ns_host *)(ac->ac_enaddr);
660 else
661 bcopy(ina->x_host.c_host, ac->ac_enaddr,
662 sizeof ac->ac_enaddr);
984263bc
MD
663
664 /*
665 * Set new address
666 */
cbf2eda6 667 IF_INIT(ifp);
984263bc
MD
668 break;
669 }
670#endif
671 default:
cbf2eda6 672 IF_INIT(ifp);
984263bc
MD
673 break;
674 }
675 break;
676
677 case SIOCGIFADDR:
f23061d4
JH
678 bcopy(IFP2AC(ifp)->ac_enaddr,
679 ((struct sockaddr *)ifr->ifr_data)->sa_data,
680 ETHER_ADDR_LEN);
984263bc
MD
681 break;
682
683 case SIOCSIFMTU:
684 /*
685 * Set the interface MTU.
686 */
687 if (ifr->ifr_mtu > ETHERMTU) {
688 error = EINVAL;
689 } else {
690 ifp->if_mtu = ifr->ifr_mtu;
691 }
692 break;
c2d9fd91
JS
693 default:
694 error = EINVAL;
695 break;
984263bc
MD
696 }
697 return (error);
cbf2eda6
SZ
698
699#undef IF_INIT
984263bc
MD
700}
701
702int
f23061d4
JH
703ether_resolvemulti(
704 struct ifnet *ifp,
705 struct sockaddr **llsa,
706 struct sockaddr *sa)
984263bc
MD
707{
708 struct sockaddr_dl *sdl;
709 struct sockaddr_in *sin;
710#ifdef INET6
711 struct sockaddr_in6 *sin6;
712#endif
713 u_char *e_addr;
714
715 switch(sa->sa_family) {
716 case AF_LINK:
717 /*
718 * No mapping needed. Just check that it's a valid MC address.
719 */
720 sdl = (struct sockaddr_dl *)sa;
721 e_addr = LLADDR(sdl);
722 if ((e_addr[0] & 1) != 1)
723 return EADDRNOTAVAIL;
724 *llsa = 0;
725 return 0;
726
727#ifdef INET
728 case AF_INET:
729 sin = (struct sockaddr_in *)sa;
730 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
731 return EADDRNOTAVAIL;
732 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
5fe66e68 733 M_WAITOK | M_ZERO);
984263bc
MD
734 sdl->sdl_len = sizeof *sdl;
735 sdl->sdl_family = AF_LINK;
736 sdl->sdl_index = ifp->if_index;
737 sdl->sdl_type = IFT_ETHER;
738 sdl->sdl_alen = ETHER_ADDR_LEN;
739 e_addr = LLADDR(sdl);
740 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
741 *llsa = (struct sockaddr *)sdl;
742 return 0;
743#endif
744#ifdef INET6
745 case AF_INET6:
746 sin6 = (struct sockaddr_in6 *)sa;
747 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
748 /*
749 * An IP6 address of 0 means listen to all
750 * of the Ethernet multicast address used for IP6.
751 * (This is used for multicast routers.)
752 */
753 ifp->if_flags |= IFF_ALLMULTI;
754 *llsa = 0;
755 return 0;
756 }
757 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
758 return EADDRNOTAVAIL;
759 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
5fe66e68 760 M_WAITOK | M_ZERO);
984263bc
MD
761 sdl->sdl_len = sizeof *sdl;
762 sdl->sdl_family = AF_LINK;
763 sdl->sdl_index = ifp->if_index;
764 sdl->sdl_type = IFT_ETHER;
765 sdl->sdl_alen = ETHER_ADDR_LEN;
766 e_addr = LLADDR(sdl);
767 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
768 *llsa = (struct sockaddr *)sdl;
769 return 0;
770#endif
771
772 default:
773 /*
774 * Well, the text isn't quite right, but it's the name
775 * that counts...
776 */
777 return EAFNOSUPPORT;
778 }
779}
d6018c31
JS
780
781#if 0
782/*
783 * This is for reference. We have a table-driven version
784 * of the little-endian crc32 generator, which is faster
785 * than the double-loop.
786 */
787uint32_t
788ether_crc32_le(const uint8_t *buf, size_t len)
789{
790 uint32_t c, crc, carry;
791 size_t i, j;
792
793 crc = 0xffffffffU; /* initial value */
794
795 for (i = 0; i < len; i++) {
796 c = buf[i];
797 for (j = 0; j < 8; j++) {
798 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
799 crc >>= 1;
800 c >>= 1;
801 if (carry)
802 crc = (crc ^ ETHER_CRC_POLY_LE);
803 }
804 }
805
f23061d4 806 return (crc);
d6018c31
JS
807}
808#else
809uint32_t
810ether_crc32_le(const uint8_t *buf, size_t len)
811{
812 static const uint32_t crctab[] = {
813 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
814 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
815 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
816 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
817 };
818 uint32_t crc;
819 size_t i;
820
821 crc = 0xffffffffU; /* initial value */
822
823 for (i = 0; i < len; i++) {
824 crc ^= buf[i];
825 crc = (crc >> 4) ^ crctab[crc & 0xf];
826 crc = (crc >> 4) ^ crctab[crc & 0xf];
827 }
828
f23061d4 829 return (crc);
d6018c31
JS
830}
831#endif
832
833uint32_t
834ether_crc32_be(const uint8_t *buf, size_t len)
835{
836 uint32_t c, crc, carry;
837 size_t i, j;
838
839 crc = 0xffffffffU; /* initial value */
840
841 for (i = 0; i < len; i++) {
842 c = buf[i];
843 for (j = 0; j < 8; j++) {
844 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
845 crc <<= 1;
846 c >>= 1;
847 if (carry)
848 crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
849 }
850 }
851
f23061d4 852 return (crc);
d6018c31 853}
4d723e5a 854
4d723e5a
JS
855/*
856 * find the size of ethernet header, and call classifier
857 */
858void
859altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
860 struct altq_pktattr *pktattr)
861{
862 struct ether_header *eh;
863 uint16_t ether_type;
864 int hlen, af, hdrsize;
865 caddr_t hdr;
866
867 hlen = sizeof(struct ether_header);
868 eh = mtod(m, struct ether_header *);
869
870 ether_type = ntohs(eh->ether_type);
871 if (ether_type < ETHERMTU) {
872 /* ick! LLC/SNAP */
873 struct llc *llc = (struct llc *)(eh + 1);
874 hlen += 8;
875
876 if (m->m_len < hlen ||
877 llc->llc_dsap != LLC_SNAP_LSAP ||
878 llc->llc_ssap != LLC_SNAP_LSAP ||
879 llc->llc_control != LLC_UI)
880 goto bad; /* not snap! */
881
882 ether_type = ntohs(llc->llc_un.type_snap.ether_type);
883 }
884
885 if (ether_type == ETHERTYPE_IP) {
886 af = AF_INET;
887 hdrsize = 20; /* sizeof(struct ip) */
888#ifdef INET6
889 } else if (ether_type == ETHERTYPE_IPV6) {
890 af = AF_INET6;
891 hdrsize = 40; /* sizeof(struct ip6_hdr) */
892#endif
893 } else
894 goto bad;
895
896 while (m->m_len <= hlen) {
897 hlen -= m->m_len;
898 m = m->m_next;
899 }
900 hdr = m->m_data + hlen;
901 if (m->m_len < hlen + hdrsize) {
902 /*
903 * ip header is not in a single mbuf. this should not
904 * happen in the current code.
905 * (todo: use m_pulldown in the future)
906 */
907 goto bad;
908 }
909 m->m_data += hlen;
910 m->m_len -= hlen;
911 ifq_classify(ifq, m, af, pktattr);
912 m->m_data -= hlen;
913 m->m_len += hlen;
914
915 return;
916
917bad:
918 pktattr->pattr_class = NULL;
919 pktattr->pattr_hdr = NULL;
920 pktattr->pattr_af = AF_UNSPEC;
921}
9b77ea6e
SZ
922
923static void
924ether_restore_header(struct mbuf **m0, const struct ether_header *eh,
925 const struct ether_header *save_eh)
926{
927 struct mbuf *m = *m0;
928
929 ether_restore_hdr++;
930
931 /*
932 * Prepend the header, optimize for the common case of
933 * eh pointing into the mbuf.
934 */
935 if ((const void *)(eh + 1) == (void *)m->m_data) {
936 m->m_data -= ETHER_HDR_LEN;
937 m->m_len += ETHER_HDR_LEN;
938 m->m_pkthdr.len += ETHER_HDR_LEN;
939 } else {
940 ether_prepend_hdr++;
941
942 M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT);
943 if (m != NULL) {
3d26a382 944 bcopy(save_eh, mtod(m, struct ether_header *),
9b77ea6e
SZ
945 ETHER_HDR_LEN);
946 }
947 }
948 *m0 = m;
949}
68b67450
SZ
950
951#ifdef ETHER_INPUT_CHAIN
952
953static void
954ether_input_ipifunc(void *arg)
955{
956 struct mbuf *m, *next;
957 lwkt_port_t port;
958
959 m = arg;
960 do {
961 next = m->m_nextpkt;
962 m->m_nextpkt = NULL;
963
964 port = m->m_pkthdr.header;
965 m->m_pkthdr.header = NULL;
966
967 lwkt_sendmsg(port,
968 &m->m_hdr.mh_netmsg.nm_netmsg.nm_lmsg);
969
970 m = next;
971 } while (m != NULL);
972}
973
974void
975ether_input_dispatch(struct mbuf_chain *chain)
976{
977#ifdef SMP
978 int i;
979
980 for (i = 0; i < ncpus; ++i) {
981 if (chain[i].mc_head != NULL) {
982 lwkt_send_ipiq(globaldata_find(i),
983 ether_input_ipifunc, chain[i].mc_head);
984 }
985 }
986#else
b358f907
SZ
987 if (chain->mc_head != NULL)
988 ether_input_ipifunc(chain->mc_head);
68b67450
SZ
989#endif
990}
991
57ccf5a2
SZ
992void
993ether_input_chain_init(struct mbuf_chain *chain)
994{
995#ifdef SMP
996 int i;
997
998 for (i = 0; i < ncpus; ++i)
999 chain[i].mc_head = chain[i].mc_tail = NULL;
1000#else
1001 chain->mc_head = chain->mc_tail = NULL;
1002#endif
1003}
1004
68b67450 1005#endif /* ETHER_INPUT_CHAIN */
29bc1092 1006
5f32d321
SZ
1007/*
1008 * Upper layer processing for a received Ethernet packet.
1009 */
297c8124 1010void
29bc1092
SZ
1011ether_demux_oncpu(struct ifnet *ifp, struct mbuf *m)
1012{
1013 struct ether_header *eh;
1014 int isr, redispatch;
1015 u_short ether_type;
1016 struct ip_fw *rule = NULL;
29bc1092
SZ
1017#ifdef NETATALK
1018 struct llc *l;
1019#endif
1020
1021 M_ASSERTPKTHDR(m);
1022 KASSERT(m->m_len >= ETHER_HDR_LEN,
1023 ("ether header is no contiguous!\n"));
1024
1025 eh = mtod(m, struct ether_header *);
1026
eb241549
SZ
1027 if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
1028 struct m_tag *mtag;
1029
1030 /* Extract info from dummynet tag */
1031 mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL);
1032 KKASSERT(mtag != NULL);
29bc1092 1033 rule = ((struct dn_pkt *)m_tag_data(mtag))->dn_priv;
eb241549
SZ
1034 KKASSERT(rule != NULL);
1035
29bc1092
SZ
1036 KKASSERT(ifp == NULL);
1037 ifp = m->m_pkthdr.rcvif;
1038
1039 m_tag_delete(m, mtag);
eb241549
SZ
1040 m->m_pkthdr.fw_flags &= ~DUMMYNET_MBUF_TAGGED;
1041
1042 /* packet is passing the second time */
29bc1092 1043 goto post_stats;
eb241549 1044 }
29bc1092
SZ
1045
1046#ifdef CARP
1047 /*
1048 * XXX: Okay, we need to call carp_forus() and - if it is for
1049 * us jump over code that does the normal check
1050 * "ac_enaddr == ether_dhost". The check sequence is a bit
1051 * different from OpenBSD, so we jump over as few code as
1052 * possible, to catch _all_ sanity checks. This needs
1053 * evaluation, to see if the carp ether_dhost values break any
1054 * of these checks!
1055 */
1056 if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost))
1057 goto post_stats;
1058#endif
1059
1060 /*
1061 * Discard packet if upper layers shouldn't see it because
1062 * it was unicast to a different Ethernet address. If the
1063 * driver is working properly, then this situation can only
1064 * happen when the interface is in promiscuous mode.
1065 */
1066 if (((ifp->if_flags & (IFF_PROMISC | IFF_PPROMISC)) == IFF_PROMISC) &&
1067 (eh->ether_dhost[0] & 1) == 0 &&
1068 bcmp(eh->ether_dhost, IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN)) {
1069 m_freem(m);
1070 return;
1071 }
1072
1073post_stats:
1074 if (IPFW_LOADED && ether_ipfw != 0) {
1075 struct ether_header save_eh = *eh;
1076
1077 /* XXX old crufty stuff, needs to be removed */
1078 m_adj(m, sizeof(struct ether_header));
1079
1080 if (!ether_ipfw_chk(&m, NULL, &rule, eh)) {
1081 m_freem(m);
1082 return;
1083 }
1084
1085 ether_restore_header(&m, eh, &save_eh);
1086 if (m == NULL)
1087 return;
1088 eh = mtod(m, struct ether_header *);
1089 }
1090
1091 ether_type = ntohs(eh->ether_type);
1092 KKASSERT(ether_type != ETHERTYPE_VLAN);
1093
1094 if (m->m_flags & M_VLANTAG) {
7af50411
SZ
1095 if (vlan_input2_p != NULL) {
1096 vlan_input2_p(m);
29bc1092
SZ
1097 } else {
1098 m->m_pkthdr.rcvif->if_noproto++;
1099 m_freem(m);
1100 }
29bc1092
SZ
1101 return;
1102 }
1103
1104 m_adj(m, sizeof(struct ether_header));
1105 redispatch = 0;
1106
1107 switch (ether_type) {
1108#ifdef INET
1109 case ETHERTYPE_IP:
297c8124 1110 if (ipflow_fastforward(m))
29bc1092 1111 return;
29bc1092
SZ
1112 isr = NETISR_IP;
1113 break;
1114
1115 case ETHERTYPE_ARP:
1116 if (ifp->if_flags & IFF_NOARP) {
1117 /* Discard packet if ARP is disabled on interface */
1118 m_freem(m);
1119 return;
1120 }
1121 isr = NETISR_ARP;
1122 break;
1123#endif
1124
1125#ifdef INET6
1126 case ETHERTYPE_IPV6:
1127 isr = NETISR_IPV6;
1128 break;
1129#endif
1130
1131#ifdef IPX
1132 case ETHERTYPE_IPX:
1133 if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
1134 return;
1135 isr = NETISR_IPX;
1136 break;
1137#endif
1138
1139#ifdef NS
1140 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
1141 isr = NETISR_NS;
1142 break;
1143
1144#endif
1145
1146#ifdef NETATALK
1147 case ETHERTYPE_AT:
1148 isr = NETISR_ATALK1;
1149 break;
1150 case ETHERTYPE_AARP:
1151 isr = NETISR_AARP;
1152 break;
1153#endif
1154
a020e9d5
SZ
1155#ifdef MPLS
1156 case ETHERTYPE_MPLS:
1157 case ETHERTYPE_MPLS_MCAST:
cb8d752c
NA
1158 /* Should have been set by ether_input_chain2(). */
1159 KKASSERT(m->m_flags & M_MPLSLABELED);
a020e9d5
SZ
1160 isr = NETISR_MPLS;
1161 break;
1162#endif
1163
29bc1092
SZ
1164 default:
1165 /*
1166 * The accurate msgport is not determined before
1167 * we reach here, so redo the dispatching
1168 */
1169 redispatch = 1;
1170#ifdef IPX
1171 if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
1172 return;
1173#endif
1174#ifdef NS
1175 checksum = mtod(m, ushort *);
1176 /* Novell 802.3 */
1177 if ((ether_type <= ETHERMTU) &&
1178 ((*checksum == 0xffff) || (*checksum == 0xE0E0))) {
1179 if (*checksum == 0xE0E0) {
1180 m->m_pkthdr.len -= 3;
1181 m->m_len -= 3;
1182 m->m_data += 3;
1183 }
1184 isr = NETISR_NS;
1185 break;
1186 }
1187#endif
1188#ifdef NETATALK
1189 if (ether_type > ETHERMTU)
1190 goto dropanyway;
1191 l = mtod(m, struct llc *);
1192 if (l->llc_dsap == LLC_SNAP_LSAP &&
1193 l->llc_ssap == LLC_SNAP_LSAP &&
1194 l->llc_control == LLC_UI) {
1195 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code,
1196 sizeof at_org_code) == 0 &&
1197 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
1198 m_adj(m, sizeof(struct llc));
1199 isr = NETISR_ATALK2;
1200 break;
1201 }
1202 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
1203 sizeof aarp_org_code) == 0 &&
1204 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
1205 m_adj(m, sizeof(struct llc));
1206 isr = NETISR_AARP;
1207 break;
1208 }
1209 }
1210dropanyway:
1211#endif
1212 if (ng_ether_input_orphan_p != NULL)
1213 ng_ether_input_orphan_p(ifp, m, eh);
1214 else
1215 m_freem(m);
1216 return;
1217 }
1218
1219 if (!redispatch)
1220 netisr_run(isr, m);
1221 else
1222 netisr_dispatch(isr, m);
1223}
1224
5f32d321
SZ
1225/*
1226 * First we perform any link layer operations, then continue to the
1227 * upper layers with ether_demux_oncpu().
1228 */
72b87a91 1229void
29bc1092
SZ
1230ether_input_oncpu(struct ifnet *ifp, struct mbuf *m)
1231{
160af078
SZ
1232 if ((ifp->if_flags & (IFF_UP | IFF_MONITOR)) != IFF_UP) {
1233 /*
1234 * Receiving interface's flags are changed, when this
1235 * packet is waiting for processing; discard it.
1236 */
1237 m_freem(m);
1238 return;
1239 }
1240
29bc1092
SZ
1241 /*
1242 * Tap the packet off here for a bridge. bridge_input()
1243 * will return NULL if it has consumed the packet, otherwise
1244 * it gets processed as normal. Note that bridge_input()
1245 * will always return the original packet if we need to
1246 * process it locally.
1247 */
1248 if (ifp->if_bridge) {
1249 KASSERT(bridge_input_p != NULL,
1250 ("%s: if_bridge not loaded!", __func__));
1251
1252 if(m->m_flags & M_PROTO1) {
1253 m->m_flags &= ~M_PROTO1;
1254 } else {
1255 /* clear M_PROMISC, in case the packets comes from a vlan */
1256 /* m->m_flags &= ~M_PROMISC; */
1257 m = bridge_input_p(ifp, m);
1258 if (m == NULL)
1259 return;
1260
1261 KASSERT(ifp == m->m_pkthdr.rcvif,
1262 ("bridge_input_p changed rcvif\n"));
1263 }
1264 }
1265
1266 /* Handle ng_ether(4) processing, if any */
1267 if (ng_ether_input_p != NULL) {
1268 ng_ether_input_p(ifp, &m);
1269 if (m == NULL)
1270 return;
1271 }
1272
1273 /* Continue with upper layer processing */
1274 ether_demux_oncpu(ifp, m);
1275}
1276
1277static void
1278ether_input_handler(struct netmsg *nmsg)
1279{
1280 struct netmsg_packet *nmp = (struct netmsg_packet *)nmsg;
1281 struct ifnet *ifp;
1282 struct mbuf *m;
1283
1284 m = nmp->nm_packet;
1285 M_ASSERTPKTHDR(m);
1286 ifp = m->m_pkthdr.rcvif;
1287
1288 ether_input_oncpu(ifp, m);
1289}
1290
1291static __inline void
1292ether_init_netpacket(int num, struct mbuf *m)
1293{
1294 struct netmsg_packet *pmsg;
1295
1296 pmsg = &m->m_hdr.mh_netmsg;
1297 netmsg_init(&pmsg->nm_netmsg, &netisr_apanic_rport, 0,
1298 ether_input_handler);
1299 pmsg->nm_packet = m;
1300 pmsg->nm_netmsg.nm_lmsg.u.ms_result = num;
1301}
1302
1303static __inline struct lwkt_port *
e420e7d5 1304ether_mport(int num, struct mbuf **m)
29bc1092 1305{
29bc1092
SZ
1306 if (num == NETISR_MAX) {
1307 /*
1308 * All packets whose target msgports can't be
1309 * determined here are dispatched to netisr0,
1310 * where further dispatching may happen.
1311 */
1312 return cpu_portfn(0);
1313 }
e420e7d5 1314 return netisr_find_port(num, m);
29bc1092
SZ
1315}
1316
62f35c44
SZ
1317/*
1318 * Process a received Ethernet packet.
1319 *
1320 * The ethernet header is assumed to be in the mbuf so the caller
1321 * MUST MAKE SURE that there are at least sizeof(struct ether_header)
1322 * bytes in the first mbuf.
1323 *
1324 * We first try to find the target msgport for this ether frame, if
1325 * there is no target msgport for it, this ether frame is discarded,
1326 * else we do following processing according to whether 'chain' is
1327 * NULL or not:
1328 * - If 'chain' is NULL, this ether frame is sent to the target msgport
1329 * immediately. This situation happens when ether_input_chain2 is
1330 * accessed through ifnet.if_input.
1331 * - If 'chain' is not NULL, this ether frame is queued to the 'chain'
1332 * bucket indexed by the target msgport's cpuid and the target msgport
1333 * is saved in mbuf's m_pkthdr.m_head. Caller of ether_input_chain2
1334 * must initialize 'chain' by calling ether_input_chain_init().
1335 * ether_input_dispatch must be called later to send ether frames
1336 * queued on 'chain' to their target msgport.
1337 */
29bc1092
SZ
1338void
1339ether_input_chain2(struct ifnet *ifp, struct mbuf *m, struct mbuf_chain *chain)
1340{
1341 struct ether_header *eh, *save_eh, save_eh0;
1342 struct lwkt_port *port;
1343 uint16_t ether_type;
1344 int isr;
1345
1346 ASSERT_SERIALIZED(ifp->if_serializer);
1347 M_ASSERTPKTHDR(m);
1348
1349 /* Discard packet if interface is not up */
1350 if (!(ifp->if_flags & IFF_UP)) {
1351 m_freem(m);
1352 return;
1353 }
1354
1355 if (m->m_len < sizeof(struct ether_header)) {
1356 /* XXX error in the caller. */
1357 m_freem(m);
1358 return;
1359 }
1360 eh = mtod(m, struct ether_header *);
1361
1362 m->m_pkthdr.rcvif = ifp;
1363
1364 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
1365 if (bcmp(ifp->if_broadcastaddr, eh->ether_dhost,
1366 ifp->if_addrlen) == 0)
1367 m->m_flags |= M_BCAST;
1368 else
1369 m->m_flags |= M_MCAST;
1370 ifp->if_imcasts++;
1371 }
1372
1373 ETHER_BPF_MTAP(ifp, m);
1374
1375 ifp->if_ibytes += m->m_pkthdr.len;
1376
1377 if (ifp->if_flags & IFF_MONITOR) {
1378 /*
1379 * Interface marked for monitoring; discard packet.
1380 */
b5a65047
SZ
1381 m_freem(m);
1382 return;
29bc1092
SZ
1383 }
1384
1385 if (ntohs(eh->ether_type) == ETHERTYPE_VLAN &&
1386 (m->m_flags & M_VLANTAG) == 0) {
1387 /*
1388 * Extract vlan tag if hardware does not do it for us
1389 */
1390 vlan_ether_decap(&m);
1391 if (m == NULL)
1392 return;
1393 eh = mtod(m, struct ether_header *);
1394 }
1395 ether_type = ntohs(eh->ether_type);
1396
1397 if ((m->m_flags & M_VLANTAG) && ether_type == ETHERTYPE_VLAN) {
1398 /*
1399 * To prevent possible dangerous recursion,
1400 * we don't do vlan-in-vlan
1401 */
1402 ifp->if_noproto++;
1403 m_freem(m);
1404 return;
1405 }
1406 KKASSERT(ether_type != ETHERTYPE_VLAN);
1407
1408 /*
1409 * Map ether type to netisr id.
1410 */
1411 switch (ether_type) {
1412#ifdef INET
1413 case ETHERTYPE_IP:
1414 isr = NETISR_IP;
1415 break;
1416
1417 case ETHERTYPE_ARP:
1418 isr = NETISR_ARP;
1419 break;
1420#endif
1421
1422#ifdef INET6
1423 case ETHERTYPE_IPV6:
1424 isr = NETISR_IPV6;
1425 break;
1426#endif
1427
1428#ifdef IPX
1429 case ETHERTYPE_IPX:
1430 isr = NETISR_IPX;
1431 break;
1432#endif
1433
1434#ifdef NS
1435 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
1436 isr = NETISR_NS;
1437 break;
1438#endif
1439
1440#ifdef NETATALK
1441 case ETHERTYPE_AT:
1442 isr = NETISR_ATALK1;
1443 break;
1444 case ETHERTYPE_AARP:
1445 isr = NETISR_AARP;
1446 break;
1447#endif
1448
a020e9d5
SZ
1449#ifdef MPLS
1450 case ETHERTYPE_MPLS:
1451 case ETHERTYPE_MPLS_MCAST:
cb8d752c 1452 m->m_flags |= M_MPLSLABELED;
a020e9d5
SZ
1453 isr = NETISR_MPLS;
1454 break;
1455#endif
1456
29bc1092
SZ
1457 default:
1458 /*
1459 * NETISR_MAX is an invalid value; it is chosen to let
1460 * ether_mport() know that we are not able to decide
1461 * this packet's msgport here.
1462 */
1463 isr = NETISR_MAX;
1464 break;
1465 }
1466
1467 /*
1468 * If the packet is in contiguous memory, following
1469 * m_adj() could ensure that the hidden ether header
1470 * will not be destroyed, else we will have to save
1471 * the ether header for the later restoration.
1472 */
1473 if (m->m_pkthdr.len != m->m_len) {
1474 save_eh0 = *eh;
1475 save_eh = &save_eh0;
1476 } else {
1477 save_eh = NULL;
1478 }
1479
1480 /*
1481 * Temporarily remove ether header; ether_mport()
1482 * expects a packet without ether header.
1483 */
1484 m_adj(m, sizeof(struct ether_header));
1485
1486 /*
1487 * Find the packet's target msgport.
1488 */
1489 port = ether_mport(isr, &m);
1490 if (port == NULL) {
1491 KKASSERT(m == NULL);
1492 return;
1493 }
1494
1495 /*
1496 * Restore ether header.
1497 */
1498 if (save_eh != NULL) {
1499 ether_restore_header(&m, eh, save_eh);
1500 if (m == NULL)
1501 return;
1502 } else {
1503 m->m_data -= ETHER_HDR_LEN;
1504 m->m_len += ETHER_HDR_LEN;
1505 m->m_pkthdr.len += ETHER_HDR_LEN;
1506 }
1507
1508 /*
1509 * Initialize mbuf's netmsg packet _after_ possible
1510 * ether header restoration, else the initialized
1511 * netmsg packet may be lost during ether header
1512 * restoration.
1513 */
1514 ether_init_netpacket(isr, m);
1515
1516#ifdef ETHER_INPUT_CHAIN
1517 if (chain != NULL) {
1518 struct mbuf_chain *c;
1519 int cpuid;
1520
1521 m->m_pkthdr.header = port; /* XXX */
1522 cpuid = port->mpu_td->td_gd->gd_cpuid;
1523
1524 c = &chain[cpuid];
1525 if (c->mc_head == NULL) {
1526 c->mc_head = c->mc_tail = m;
1527 } else {
1528 c->mc_tail->m_nextpkt = m;
1529 c->mc_tail = m;
1530 }
1531 m->m_nextpkt = NULL;
1532 } else
1533#endif /* ETHER_INPUT_CHAIN */
1534 lwkt_sendmsg(port, &m->m_hdr.mh_netmsg.nm_netmsg.nm_lmsg);
1535}