2 * Copyright (c) 1982, 1989, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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
33 * @(#)if_ethersubr.c 8.1 (Berkeley) 6/10/93
34 * $FreeBSD: src/sys/net/if_ethersubr.c,v 1.70.2.33 2003/04/28 15:45:53 archie Exp $
35 * $DragonFly: src/sys/net/if_ethersubr.c,v 1.84 2008/08/23 04:12:23 sephe Exp $
38 #include "opt_atalk.h"
40 #include "opt_inet6.h"
43 #include "opt_netgraph.h"
45 #include "opt_ethernet.h"
47 #include <sys/param.h>
48 #include <sys/systm.h>
49 #include <sys/globaldata.h>
50 #include <sys/kernel.h>
51 #include <sys/malloc.h>
53 #include <sys/msgport.h>
54 #include <sys/socket.h>
55 #include <sys/sockio.h>
56 #include <sys/sysctl.h>
57 #include <sys/thread.h>
58 #include <sys/thread2.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>
66 #include <net/ifq_var.h>
68 #include <net/ethernet.h>
69 #include <net/vlan/if_vlan_ether.h>
70 #include <net/netmsg2.h>
72 #if defined(INET) || defined(INET6)
73 #include <netinet/in.h>
74 #include <netinet/in_var.h>
75 #include <netinet/if_ether.h>
76 #include <net/ipfw/ip_fw.h>
77 #include <net/dummynet/ip_dummynet.h>
80 #include <netinet6/nd6.h>
84 #include <netinet/ip_carp.h>
88 #include <netproto/ipx/ipx.h>
89 #include <netproto/ipx/ipx_if.h>
90 int (*ef_inputp
)(struct ifnet
*, const struct ether_header
*eh
, struct mbuf
*m
);
91 int (*ef_outputp
)(struct ifnet
*ifp
, struct mbuf
**mp
, struct sockaddr
*dst
,
92 short *tp
, int *hlen
);
97 #include <netns/ns_if.h>
99 int ether_outputdebug
= 0;
100 int ether_inputdebug
= 0;
104 #include <netproto/atalk/at.h>
105 #include <netproto/atalk/at_var.h>
106 #include <netproto/atalk/at_extern.h>
108 #define llc_snap_org_code llc_un.type_snap.org_code
109 #define llc_snap_ether_type llc_un.type_snap.ether_type
111 extern u_char at_org_code
[3];
112 extern u_char aarp_org_code
[3];
113 #endif /* NETATALK */
116 #include <netproto/mpls/mpls.h>
119 /* netgraph node hooks for ng_ether(4) */
120 void (*ng_ether_input_p
)(struct ifnet
*ifp
, struct mbuf
**mp
);
121 void (*ng_ether_input_orphan_p
)(struct ifnet
*ifp
,
122 struct mbuf
*m
, const struct ether_header
*eh
);
123 int (*ng_ether_output_p
)(struct ifnet
*ifp
, struct mbuf
**mp
);
124 void (*ng_ether_attach_p
)(struct ifnet
*ifp
);
125 void (*ng_ether_detach_p
)(struct ifnet
*ifp
);
127 void (*vlan_input2_p
)(struct mbuf
*);
129 static int ether_output(struct ifnet
*, struct mbuf
*, struct sockaddr
*,
131 static void ether_restore_header(struct mbuf
**, const struct ether_header
*,
132 const struct ether_header
*);
137 struct mbuf
*(*bridge_input_p
)(struct ifnet
*, struct mbuf
*);
138 int (*bridge_output_p
)(struct ifnet
*, struct mbuf
*);
139 void (*bridge_dn_p
)(struct mbuf
*, struct ifnet
*);
141 static int ether_resolvemulti(struct ifnet
*, struct sockaddr
**,
144 const uint8_t etherbroadcastaddr
[ETHER_ADDR_LEN
] = {
145 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
148 #define gotoerr(e) do { error = (e); goto bad; } while (0)
149 #define IFP2AC(ifp) ((struct arpcom *)(ifp))
151 static boolean_t
ether_ipfw_chk(struct mbuf
**m0
, struct ifnet
*dst
,
153 const struct ether_header
*eh
);
155 static int ether_ipfw
;
156 static u_int ether_restore_hdr
;
157 static u_int ether_prepend_hdr
;
159 SYSCTL_DECL(_net_link
);
160 SYSCTL_NODE(_net_link
, IFT_ETHER
, ether
, CTLFLAG_RW
, 0, "Ethernet");
161 SYSCTL_INT(_net_link_ether
, OID_AUTO
, ipfw
, CTLFLAG_RW
,
162 ðer_ipfw
, 0, "Pass ether pkts through firewall");
163 SYSCTL_UINT(_net_link_ether
, OID_AUTO
, restore_hdr
, CTLFLAG_RW
,
164 ðer_restore_hdr
, 0, "# of ether header restoration");
165 SYSCTL_UINT(_net_link_ether
, OID_AUTO
, prepend_hdr
, CTLFLAG_RW
,
166 ðer_prepend_hdr
, 0,
167 "# of ether header restoration which prepends mbuf");
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.
177 ether_output(struct ifnet
*ifp
, struct mbuf
*m
, struct sockaddr
*dst
,
180 struct ether_header
*eh
, *deh
;
183 int hlen
= ETHER_HDR_LEN
; /* link layer header length */
184 struct arpcom
*ac
= IFP2AC(ifp
);
187 ASSERT_NOT_SERIALIZED(ifp
->if_serializer
);
189 if (ifp
->if_flags
& IFF_MONITOR
)
191 if ((ifp
->if_flags
& (IFF_UP
| IFF_RUNNING
)) != (IFF_UP
| IFF_RUNNING
))
194 M_PREPEND(m
, sizeof(struct ether_header
), MB_DONTWAIT
);
197 eh
= mtod(m
, struct ether_header
*);
198 edst
= eh
->ether_dhost
;
201 * Fill in the destination ethernet address and frame type.
203 switch (dst
->sa_family
) {
206 if (!arpresolve(ifp
, rt
, m
, dst
, edst
))
207 return (0); /* if not yet resolved */
209 if (m
->m_flags
& M_MPLSLABELED
)
210 eh
->ether_type
= htons(ETHERTYPE_MPLS
);
213 eh
->ether_type
= htons(ETHERTYPE_IP
);
218 if (!nd6_storelladdr(&ac
->ac_if
, rt
, m
, dst
, edst
))
219 return (0); /* Something bad happenned. */
220 eh
->ether_type
= htons(ETHERTYPE_IPV6
);
225 if (ef_outputp
!= NULL
) {
226 error
= ef_outputp(ifp
, &m
, dst
, &eh
->ether_type
,
231 eh
->ether_type
= htons(ETHERTYPE_IPX
);
232 bcopy(&(((struct sockaddr_ipx
*)dst
)->sipx_addr
.x_host
),
233 edst
, ETHER_ADDR_LEN
);
239 struct at_ifaddr
*aa
;
241 if ((aa
= at_ifawithnet((struct sockaddr_at
*)dst
)) == NULL
) {
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.
251 if (aa
->aa_flags
& AFA_PHASE2
) {
254 M_PREPEND(m
, sizeof(struct llc
), MB_DONTWAIT
);
255 eh
= mtod(m
, struct ether_header
*);
256 edst
= eh
->ether_dhost
;
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
,
261 llc
.llc_snap_ether_type
= htons(ETHERTYPE_AT
);
263 mtod(m
, caddr_t
) + sizeof(struct ether_header
),
265 eh
->ether_type
= htons(m
->m_pkthdr
.len
);
266 hlen
= sizeof(struct llc
) + ETHER_HDR_LEN
;
268 eh
->ether_type
= htons(ETHERTYPE_AT
);
270 if (!aarpresolve(ac
, m
, (struct sockaddr_at
*)dst
, edst
))
279 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
280 eh
->ether_type
= 0x8137;
282 case 0x0: /* Novell 802.3 */
283 eh
->ether_type
= htons(m
->m_pkthdr
.len
);
285 case 0xe0e0: /* Novell 802.2 and Token-Ring */
286 M_PREPEND(m
, 3, MB_DONTWAIT
);
287 eh
= mtod(m
, struct ether_header
*);
288 edst
= eh
->ether_dhost
;
289 eh
->ether_type
= htons(m
->m_pkthdr
.len
);
290 cp
= mtod(m
, u_char
*) + sizeof(struct ether_header
);
296 bcopy(&(((struct sockaddr_ns
*)dst
)->sns_addr
.x_host
), edst
,
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?
304 if (bcmp(edst
, &ns_thishost
, ETHER_ADDR_LEN
) == 0) {
305 m
->m_pkthdr
.rcvif
= ifp
;
306 netisr_dispatch(NETISR_NS
, m
);
309 if (bcmp(edst
, &ns_broadhost
, ETHER_ADDR_LEN
) == 0)
310 m
->m_flags
|= M_BCAST
;
313 case pseudo_AF_HDRCMPLT
:
315 loop_copy
= -1; /* if this is for us, don't do it */
316 deh
= (struct ether_header
*)dst
->sa_data
;
317 memcpy(edst
, deh
->ether_dhost
, ETHER_ADDR_LEN
);
318 eh
->ether_type
= deh
->ether_type
;
322 if_printf(ifp
, "can't handle af%d\n", dst
->sa_family
);
323 gotoerr(EAFNOSUPPORT
);
326 if (dst
->sa_family
== pseudo_AF_HDRCMPLT
) /* unlikely */
327 memcpy(eh
->ether_shost
,
328 ((struct ether_header
*)dst
->sa_data
)->ether_shost
,
331 memcpy(eh
->ether_shost
, ac
->ac_enaddr
, ETHER_ADDR_LEN
);
334 * Bridges require special output handling.
336 if (ifp
->if_bridge
) {
337 KASSERT(bridge_output_p
!= NULL
,
338 ("%s: if_bridge not loaded!", __func__
));
339 return bridge_output_p(ifp
, m
);
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.
351 if ((ifp
->if_flags
& IFF_SIMPLEX
) && (loop_copy
!= -1)) {
354 if (m
->m_pkthdr
.csum_flags
& CSUM_IP
)
355 csum_flags
|= (CSUM_IP_CHECKED
| CSUM_IP_VALID
);
356 if (m
->m_pkthdr
.csum_flags
& CSUM_DELAY_DATA
)
357 csum_flags
|= (CSUM_DATA_VALID
| CSUM_PSEUDO_HDR
);
358 if ((m
->m_flags
& M_BCAST
) || (loop_copy
> 0)) {
361 if ((n
= m_copypacket(m
, MB_DONTWAIT
)) != NULL
) {
362 n
->m_pkthdr
.csum_flags
|= csum_flags
;
363 if (csum_flags
& CSUM_DATA_VALID
)
364 n
->m_pkthdr
.csum_data
= 0xffff;
365 if_simloop(ifp
, n
, dst
->sa_family
, hlen
);
368 } else if (bcmp(eh
->ether_dhost
, eh
->ether_shost
,
369 ETHER_ADDR_LEN
) == 0) {
370 m
->m_pkthdr
.csum_flags
|= csum_flags
;
371 if (csum_flags
& CSUM_DATA_VALID
)
372 m
->m_pkthdr
.csum_data
= 0xffff;
373 if_simloop(ifp
, m
, dst
->sa_family
, hlen
);
374 return (0); /* XXX */
379 if (ifp
->if_carp
&& (error
= carp_output(ifp
, m
, dst
, NULL
)))
384 /* Handle ng_ether(4) processing, if any */
385 if (ng_ether_output_p
!= NULL
) {
386 if ((error
= (*ng_ether_output_p
)(ifp
, &m
)) != 0)
392 /* Continue with link-layer output */
393 return ether_output_frame(ifp
, m
);
401 * Ethernet link layer output routine to send a raw frame to the device.
403 * This assumes that the 14 byte Ethernet header is present and contiguous
407 ether_output_frame(struct ifnet
*ifp
, struct mbuf
*m
)
409 struct ip_fw
*rule
= NULL
;
411 struct altq_pktattr pktattr
;
413 ASSERT_NOT_SERIALIZED(ifp
->if_serializer
);
415 if (m
->m_pkthdr
.fw_flags
& DUMMYNET_MBUF_TAGGED
) {
418 /* Extract info from dummynet tag */
419 mtag
= m_tag_find(m
, PACKET_TAG_DUMMYNET
, NULL
);
420 KKASSERT(mtag
!= NULL
);
421 rule
= ((struct dn_pkt
*)m_tag_data(mtag
))->dn_priv
;
422 KKASSERT(rule
!= NULL
);
424 m_tag_delete(m
, mtag
);
425 m
->m_pkthdr
.fw_flags
&= ~DUMMYNET_MBUF_TAGGED
;
428 if (ifq_is_enabled(&ifp
->if_snd
))
429 altq_etherclassify(&ifp
->if_snd
, m
, &pktattr
);
431 if (IPFW_LOADED
&& ether_ipfw
!= 0) {
432 struct ether_header save_eh
, *eh
;
434 eh
= mtod(m
, struct ether_header
*);
436 m_adj(m
, ETHER_HDR_LEN
);
437 if (!ether_ipfw_chk(&m
, ifp
, &rule
, eh
)) {
441 return ENOBUFS
; /* pkt dropped */
443 return 0; /* consumed e.g. in a pipe */
446 /* packet was ok, restore the ethernet header */
447 ether_restore_header(&m
, eh
, &save_eh
);
456 * Queue message on interface, update output statistics if
457 * successful, and start output if interface not yet active.
459 error
= ifq_dispatch(ifp
, m
, &pktattr
);
464 * ipfw processing for ethernet packets (in and out).
465 * The second parameter is NULL from ether_demux(), and ifp from
466 * ether_output_frame().
469 ether_ipfw_chk(struct mbuf
**m0
, struct ifnet
*dst
, struct ip_fw
**rule
,
470 const struct ether_header
*eh
)
472 struct ether_header save_eh
= *eh
; /* might be a ptr in m */
473 struct ip_fw_args args
;
477 if (*rule
!= NULL
&& fw_one_pass
)
478 return TRUE
; /* dummynet packet, already partially processed */
481 * I need some amount of data to be contiguous.
483 i
= min((*m0
)->m_pkthdr
.len
, max_protohdr
);
484 if ((*m0
)->m_len
< i
) {
485 *m0
= m_pullup(*m0
, i
);
493 if ((mtag
= m_tag_find(*m0
, PACKET_TAG_IPFW_DIVERT
, NULL
)) != NULL
)
494 m_tag_delete(*m0
, mtag
);
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
;
502 args
.m
= *m0
; /* the packet we are looking at */
503 args
.oif
= dst
; /* destination, if any */
504 args
.rule
= *rule
; /* matching rule to restart */
505 args
.eh
= &save_eh
; /* MAC header for bridged/MAC packets */
506 i
= ip_fw_chk_ptr(&args
);
510 if ((i
& IP_FW_PORT_DENY_FLAG
) || *m0
== NULL
) /* drop */
513 if (i
== 0) /* a PASS rule. */
516 if (i
& IP_FW_PORT_DYNT_FLAG
) {
518 * Pass the pkt to dummynet, which consumes it.
522 m
= *m0
; /* pass the original to dummynet */
523 *m0
= NULL
; /* and nothing back to the caller */
525 ether_restore_header(&m
, eh
, &save_eh
);
529 ip_fw_dn_io_ptr(m
, (i
& 0xffff),
530 dst
? DN_TO_ETH_OUT
: DN_TO_ETH_DEMUX
, &args
);
534 * XXX at some point add support for divert/forward actions.
535 * If none of the above matches, we have to drop the pkt.
541 ether_input(struct ifnet
*ifp
, struct mbuf
*m
)
543 ether_input_chain2(ifp
, m
, NULL
);
547 * Perform common duties while attaching to interface list
550 ether_ifattach(struct ifnet
*ifp
, uint8_t *lla
, lwkt_serialize_t serializer
)
552 ether_ifattach_bpf(ifp
, lla
, DLT_EN10MB
, sizeof(struct ether_header
),
557 ether_ifattach_bpf(struct ifnet
*ifp
, uint8_t *lla
, u_int dlt
, u_int hdrlen
,
558 lwkt_serialize_t serializer
)
560 struct sockaddr_dl
*sdl
;
562 ifp
->if_type
= IFT_ETHER
;
563 ifp
->if_addrlen
= ETHER_ADDR_LEN
;
564 ifp
->if_hdrlen
= ETHER_HDR_LEN
;
565 if_attach(ifp
, serializer
);
566 ifp
->if_mtu
= ETHERMTU
;
567 if (ifp
->if_baudrate
== 0)
568 ifp
->if_baudrate
= 10000000;
569 ifp
->if_output
= ether_output
;
570 ifp
->if_input
= ether_input
;
571 ifp
->if_resolvemulti
= ether_resolvemulti
;
572 ifp
->if_broadcastaddr
= etherbroadcastaddr
;
573 sdl
= IF_LLSOCKADDR(ifp
);
574 sdl
->sdl_type
= IFT_ETHER
;
575 sdl
->sdl_alen
= ifp
->if_addrlen
;
576 bcopy(lla
, LLADDR(sdl
), ifp
->if_addrlen
);
578 * XXX Keep the current drivers happy.
579 * XXX Remove once all drivers have been cleaned up
581 if (lla
!= IFP2AC(ifp
)->ac_enaddr
)
582 bcopy(lla
, IFP2AC(ifp
)->ac_enaddr
, ifp
->if_addrlen
);
583 bpfattach(ifp
, dlt
, hdrlen
);
584 if (ng_ether_attach_p
!= NULL
)
585 (*ng_ether_attach_p
)(ifp
);
587 if_printf(ifp
, "MAC address: %6D\n", lla
, ":");
591 * Perform common duties while detaching an Ethernet interface
594 ether_ifdetach(struct ifnet
*ifp
)
598 if (ng_ether_detach_p
!= NULL
)
599 (*ng_ether_detach_p
)(ifp
);
605 ether_ioctl(struct ifnet
*ifp
, int command
, caddr_t data
)
607 struct ifaddr
*ifa
= (struct ifaddr
*) data
;
608 struct ifreq
*ifr
= (struct ifreq
*) data
;
611 #define IF_INIT(ifp) \
613 if (((ifp)->if_flags & IFF_UP) == 0) { \
614 (ifp)->if_flags |= IFF_UP; \
615 (ifp)->if_init((ifp)->if_softc); \
619 ASSERT_SERIALIZED(ifp
->if_serializer
);
623 switch (ifa
->ifa_addr
->sa_family
) {
626 IF_INIT(ifp
); /* before arpwhohas */
627 arp_ifinit(ifp
, ifa
);
632 * XXX - This code is probably wrong
636 struct ipx_addr
*ina
= &IA_SIPX(ifa
)->sipx_addr
;
637 struct arpcom
*ac
= IFP2AC(ifp
);
639 if (ipx_nullhost(*ina
))
640 ina
->x_host
= *(union ipx_host
*) ac
->ac_enaddr
;
642 bcopy(ina
->x_host
.c_host
, ac
->ac_enaddr
,
643 sizeof ac
->ac_enaddr
);
645 IF_INIT(ifp
); /* Set new address. */
651 * XXX - This code is probably wrong
655 struct ns_addr
*ina
= &(IA_SNS(ifa
)->sns_addr
);
656 struct arpcom
*ac
= IFP2AC(ifp
);
658 if (ns_nullhost(*ina
))
659 ina
->x_host
= *(union ns_host
*)(ac
->ac_enaddr
);
661 bcopy(ina
->x_host
.c_host
, ac
->ac_enaddr
,
662 sizeof ac
->ac_enaddr
);
678 bcopy(IFP2AC(ifp
)->ac_enaddr
,
679 ((struct sockaddr
*)ifr
->ifr_data
)->sa_data
,
685 * Set the interface MTU.
687 if (ifr
->ifr_mtu
> ETHERMTU
) {
690 ifp
->if_mtu
= ifr
->ifr_mtu
;
705 struct sockaddr
**llsa
,
708 struct sockaddr_dl
*sdl
;
709 struct sockaddr_in
*sin
;
711 struct sockaddr_in6
*sin6
;
715 switch(sa
->sa_family
) {
718 * No mapping needed. Just check that it's a valid MC address.
720 sdl
= (struct sockaddr_dl
*)sa
;
721 e_addr
= LLADDR(sdl
);
722 if ((e_addr
[0] & 1) != 1)
723 return EADDRNOTAVAIL
;
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
,
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
;
746 sin6
= (struct sockaddr_in6
*)sa
;
747 if (IN6_IS_ADDR_UNSPECIFIED(&sin6
->sin6_addr
)) {
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.)
753 ifp
->if_flags
|= IFF_ALLMULTI
;
757 if (!IN6_IS_ADDR_MULTICAST(&sin6
->sin6_addr
))
758 return EADDRNOTAVAIL
;
759 MALLOC(sdl
, struct sockaddr_dl
*, sizeof *sdl
, M_IFMADDR
,
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
;
774 * Well, the text isn't quite right, but it's the name
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.
788 ether_crc32_le(const uint8_t *buf
, size_t len
)
790 uint32_t c
, crc
, carry
;
793 crc
= 0xffffffffU
; /* initial value */
795 for (i
= 0; i
< len
; i
++) {
797 for (j
= 0; j
< 8; j
++) {
798 carry
= ((crc
& 0x01) ? 1 : 0) ^ (c
& 0x01);
802 crc
= (crc
^ ETHER_CRC_POLY_LE
);
810 ether_crc32_le(const uint8_t *buf
, size_t len
)
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
821 crc
= 0xffffffffU
; /* initial value */
823 for (i
= 0; i
< len
; i
++) {
825 crc
= (crc
>> 4) ^ crctab
[crc
& 0xf];
826 crc
= (crc
>> 4) ^ crctab
[crc
& 0xf];
834 ether_crc32_be(const uint8_t *buf
, size_t len
)
836 uint32_t c
, crc
, carry
;
839 crc
= 0xffffffffU
; /* initial value */
841 for (i
= 0; i
< len
; i
++) {
843 for (j
= 0; j
< 8; j
++) {
844 carry
= ((crc
& 0x80000000U
) ? 1 : 0) ^ (c
& 0x01);
848 crc
= (crc
^ ETHER_CRC_POLY_BE
) | carry
;
856 * find the size of ethernet header, and call classifier
859 altq_etherclassify(struct ifaltq
*ifq
, struct mbuf
*m
,
860 struct altq_pktattr
*pktattr
)
862 struct ether_header
*eh
;
864 int hlen
, af
, hdrsize
;
867 hlen
= sizeof(struct ether_header
);
868 eh
= mtod(m
, struct ether_header
*);
870 ether_type
= ntohs(eh
->ether_type
);
871 if (ether_type
< ETHERMTU
) {
873 struct llc
*llc
= (struct llc
*)(eh
+ 1);
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! */
882 ether_type
= ntohs(llc
->llc_un
.type_snap
.ether_type
);
885 if (ether_type
== ETHERTYPE_IP
) {
887 hdrsize
= 20; /* sizeof(struct ip) */
889 } else if (ether_type
== ETHERTYPE_IPV6
) {
891 hdrsize
= 40; /* sizeof(struct ip6_hdr) */
896 while (m
->m_len
<= hlen
) {
900 hdr
= m
->m_data
+ hlen
;
901 if (m
->m_len
< hlen
+ hdrsize
) {
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)
911 ifq_classify(ifq
, m
, af
, pktattr
);
918 pktattr
->pattr_class
= NULL
;
919 pktattr
->pattr_hdr
= NULL
;
920 pktattr
->pattr_af
= AF_UNSPEC
;
924 ether_restore_header(struct mbuf
**m0
, const struct ether_header
*eh
,
925 const struct ether_header
*save_eh
)
927 struct mbuf
*m
= *m0
;
932 * Prepend the header, optimize for the common case of
933 * eh pointing into the mbuf.
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
;
942 M_PREPEND(m
, ETHER_HDR_LEN
, MB_DONTWAIT
);
944 bcopy(save_eh
, mtod(m
, struct ether_header
*),
951 #ifdef ETHER_INPUT_CHAIN
954 ether_input_ipifunc(void *arg
)
956 struct mbuf
*m
, *next
;
964 port
= m
->m_pkthdr
.header
;
965 m
->m_pkthdr
.header
= NULL
;
968 &m
->m_hdr
.mh_netmsg
.nm_netmsg
.nm_lmsg
);
975 ether_input_dispatch(struct mbuf_chain
*chain
)
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
);
987 if (chain
->mc_head
!= NULL
)
988 ether_input_ipifunc(chain
->mc_head
);
993 ether_input_chain_init(struct mbuf_chain
*chain
)
998 for (i
= 0; i
< ncpus
; ++i
)
999 chain
[i
].mc_head
= chain
[i
].mc_tail
= NULL
;
1001 chain
->mc_head
= chain
->mc_tail
= NULL
;
1005 #endif /* ETHER_INPUT_CHAIN */
1008 * Upper layer processing for a received Ethernet packet.
1011 ether_demux_oncpu(struct ifnet
*ifp
, struct mbuf
*m
)
1013 struct ether_header
*eh
;
1014 int isr
, redispatch
;
1016 struct ip_fw
*rule
= NULL
;
1022 KASSERT(m
->m_len
>= ETHER_HDR_LEN
,
1023 ("ether header is no contiguous!\n"));
1025 eh
= mtod(m
, struct ether_header
*);
1027 if (m
->m_pkthdr
.fw_flags
& DUMMYNET_MBUF_TAGGED
) {
1030 /* Extract info from dummynet tag */
1031 mtag
= m_tag_find(m
, PACKET_TAG_DUMMYNET
, NULL
);
1032 KKASSERT(mtag
!= NULL
);
1033 rule
= ((struct dn_pkt
*)m_tag_data(mtag
))->dn_priv
;
1034 KKASSERT(rule
!= NULL
);
1036 KKASSERT(ifp
== NULL
);
1037 ifp
= m
->m_pkthdr
.rcvif
;
1039 m_tag_delete(m
, mtag
);
1040 m
->m_pkthdr
.fw_flags
&= ~DUMMYNET_MBUF_TAGGED
;
1042 /* packet is passing the second time */
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
1056 if (ifp
->if_carp
&& carp_forus(ifp
->if_carp
, eh
->ether_dhost
))
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.
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
)) {
1074 if (IPFW_LOADED
&& ether_ipfw
!= 0) {
1075 struct ether_header save_eh
= *eh
;
1077 /* XXX old crufty stuff, needs to be removed */
1078 m_adj(m
, sizeof(struct ether_header
));
1080 if (!ether_ipfw_chk(&m
, NULL
, &rule
, eh
)) {
1085 ether_restore_header(&m
, eh
, &save_eh
);
1088 eh
= mtod(m
, struct ether_header
*);
1091 ether_type
= ntohs(eh
->ether_type
);
1092 KKASSERT(ether_type
!= ETHERTYPE_VLAN
);
1094 if (m
->m_flags
& M_VLANTAG
) {
1095 if (vlan_input2_p
!= NULL
) {
1098 m
->m_pkthdr
.rcvif
->if_noproto
++;
1104 m_adj(m
, sizeof(struct ether_header
));
1107 switch (ether_type
) {
1110 if (ipflow_fastforward(m
))
1116 if (ifp
->if_flags
& IFF_NOARP
) {
1117 /* Discard packet if ARP is disabled on interface */
1126 case ETHERTYPE_IPV6
:
1133 if (ef_inputp
&& ef_inputp(ifp
, eh
, m
) == 0)
1140 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
1148 isr
= NETISR_ATALK1
;
1150 case ETHERTYPE_AARP
:
1156 case ETHERTYPE_MPLS
:
1157 case ETHERTYPE_MPLS_MCAST
:
1158 /* Should have been set by ether_input_chain2(). */
1159 KKASSERT(m
->m_flags
& M_MPLSLABELED
);
1166 * The accurate msgport is not determined before
1167 * we reach here, so redo the dispatching
1171 if (ef_inputp
&& ef_inputp(ifp
, eh
, m
) == 0)
1175 checksum
= mtod(m
, ushort
*);
1177 if ((ether_type
<= ETHERMTU
) &&
1178 ((*checksum
== 0xffff) || (*checksum
== 0xE0E0))) {
1179 if (*checksum
== 0xE0E0) {
1180 m
->m_pkthdr
.len
-= 3;
1189 if (ether_type
> ETHERMTU
)
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
;
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
));
1212 if (ng_ether_input_orphan_p
!= NULL
)
1213 ng_ether_input_orphan_p(ifp
, m
, eh
);
1222 netisr_dispatch(isr
, m
);
1226 * First we perform any link layer operations, then continue to the
1227 * upper layers with ether_demux_oncpu().
1230 ether_input_oncpu(struct ifnet
*ifp
, struct mbuf
*m
)
1232 if ((ifp
->if_flags
& (IFF_UP
| IFF_MONITOR
)) != IFF_UP
) {
1234 * Receiving interface's flags are changed, when this
1235 * packet is waiting for processing; discard it.
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.
1248 if (ifp
->if_bridge
) {
1249 KASSERT(bridge_input_p
!= NULL
,
1250 ("%s: if_bridge not loaded!", __func__
));
1252 if(m
->m_flags
& M_PROTO1
) {
1253 m
->m_flags
&= ~M_PROTO1
;
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
);
1261 KASSERT(ifp
== m
->m_pkthdr
.rcvif
,
1262 ("bridge_input_p changed rcvif\n"));
1266 /* Handle ng_ether(4) processing, if any */
1267 if (ng_ether_input_p
!= NULL
) {
1268 ng_ether_input_p(ifp
, &m
);
1273 /* Continue with upper layer processing */
1274 ether_demux_oncpu(ifp
, m
);
1278 ether_input_handler(struct netmsg
*nmsg
)
1280 struct netmsg_packet
*nmp
= (struct netmsg_packet
*)nmsg
;
1286 ifp
= m
->m_pkthdr
.rcvif
;
1288 ether_input_oncpu(ifp
, m
);
1291 static __inline
void
1292 ether_init_netpacket(int num
, struct mbuf
*m
)
1294 struct netmsg_packet
*pmsg
;
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
;
1303 static __inline
struct lwkt_port
*
1304 ether_mport(int num
, struct mbuf
**m
)
1306 if (num
== NETISR_MAX
) {
1308 * All packets whose target msgports can't be
1309 * determined here are dispatched to netisr0,
1310 * where further dispatching may happen.
1312 return cpu_portfn(0);
1314 return netisr_find_port(num
, m
);
1318 * Process a received Ethernet packet.
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.
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
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.
1339 ether_input_chain2(struct ifnet
*ifp
, struct mbuf
*m
, struct mbuf_chain
*chain
)
1341 struct ether_header
*eh
, *save_eh
, save_eh0
;
1342 struct lwkt_port
*port
;
1343 uint16_t ether_type
;
1346 ASSERT_SERIALIZED(ifp
->if_serializer
);
1349 /* Discard packet if interface is not up */
1350 if (!(ifp
->if_flags
& IFF_UP
)) {
1355 if (m
->m_len
< sizeof(struct ether_header
)) {
1356 /* XXX error in the caller. */
1360 eh
= mtod(m
, struct ether_header
*);
1362 m
->m_pkthdr
.rcvif
= ifp
;
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
;
1369 m
->m_flags
|= M_MCAST
;
1373 ETHER_BPF_MTAP(ifp
, m
);
1375 ifp
->if_ibytes
+= m
->m_pkthdr
.len
;
1377 if (ifp
->if_flags
& IFF_MONITOR
) {
1379 * Interface marked for monitoring; discard packet.
1385 if (ntohs(eh
->ether_type
) == ETHERTYPE_VLAN
&&
1386 (m
->m_flags
& M_VLANTAG
) == 0) {
1388 * Extract vlan tag if hardware does not do it for us
1390 vlan_ether_decap(&m
);
1393 eh
= mtod(m
, struct ether_header
*);
1395 ether_type
= ntohs(eh
->ether_type
);
1397 if ((m
->m_flags
& M_VLANTAG
) && ether_type
== ETHERTYPE_VLAN
) {
1399 * To prevent possible dangerous recursion,
1400 * we don't do vlan-in-vlan
1406 KKASSERT(ether_type
!= ETHERTYPE_VLAN
);
1409 * Map ether type to netisr id.
1411 switch (ether_type
) {
1423 case ETHERTYPE_IPV6
:
1435 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
1442 isr
= NETISR_ATALK1
;
1444 case ETHERTYPE_AARP
:
1450 case ETHERTYPE_MPLS
:
1451 case ETHERTYPE_MPLS_MCAST
:
1452 m
->m_flags
|= M_MPLSLABELED
;
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.
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.
1473 if (m
->m_pkthdr
.len
!= m
->m_len
) {
1475 save_eh
= &save_eh0
;
1481 * Temporarily remove ether header; ether_mport()
1482 * expects a packet without ether header.
1484 m_adj(m
, sizeof(struct ether_header
));
1487 * Find the packet's target msgport.
1489 port
= ether_mport(isr
, &m
);
1491 KKASSERT(m
== NULL
);
1496 * Restore ether header.
1498 if (save_eh
!= NULL
) {
1499 ether_restore_header(&m
, eh
, save_eh
);
1503 m
->m_data
-= ETHER_HDR_LEN
;
1504 m
->m_len
+= ETHER_HDR_LEN
;
1505 m
->m_pkthdr
.len
+= ETHER_HDR_LEN
;
1509 * Initialize mbuf's netmsg packet _after_ possible
1510 * ether header restoration, else the initialized
1511 * netmsg packet may be lost during ether header
1514 ether_init_netpacket(isr
, m
);
1516 #ifdef ETHER_INPUT_CHAIN
1517 if (chain
!= NULL
) {
1518 struct mbuf_chain
*c
;
1521 m
->m_pkthdr
.header
= port
; /* XXX */
1522 cpuid
= port
->mpu_td
->td_gd
->gd_cpuid
;
1525 if (c
->mc_head
== NULL
) {
1526 c
->mc_head
= c
->mc_tail
= m
;
1528 c
->mc_tail
->m_nextpkt
= m
;
1531 m
->m_nextpkt
= NULL
;
1533 #endif /* ETHER_INPUT_CHAIN */
1534 lwkt_sendmsg(port
, &m
->m_hdr
.mh_netmsg
.nm_netmsg
.nm_lmsg
);