2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1982, 1989, 1993
5 * The Regents of the University of California. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * @(#)if_ethersubr.c 8.1 (Berkeley) 6/10/93
36 #include "opt_inet6.h"
37 #include "opt_netgraph.h"
38 #include "opt_mbuf_profiling.h"
41 #include <sys/param.h>
42 #include <sys/systm.h>
44 #include <sys/eventhandler.h>
45 #include <sys/kernel.h>
47 #include <sys/malloc.h>
48 #include <sys/module.h>
50 #include <sys/random.h>
51 #include <sys/socket.h>
52 #include <sys/sockio.h>
53 #include <sys/sysctl.h>
57 #include <net/if_var.h>
58 #include <net/if_arp.h>
59 #include <net/netisr.h>
60 #include <net/route.h>
61 #include <net/if_llc.h>
62 #include <net/if_dl.h>
63 #include <net/if_types.h>
65 #include <net/ethernet.h>
66 #include <net/if_bridgevar.h>
67 #include <net/if_vlan_var.h>
68 #include <net/if_llatbl.h>
70 #include <net/rss_config.h>
73 #include <netpfil/pf/pf_mtag.h>
75 #if defined(INET) || defined(INET6)
76 #include <netinet/in.h>
77 #include <netinet/in_var.h>
78 #include <netinet/if_ether.h>
79 #include <netinet/ip_carp.h>
80 #include <netinet/ip_var.h>
83 #include <netinet6/nd6.h>
85 #include <security/mac/mac_framework.h>
88 CTASSERT(sizeof (struct ether_header) == ETHER_ADDR_LEN * 2 + 2);
89 CTASSERT(sizeof (struct ether_addr) == ETHER_ADDR_LEN);
92 VNET_DEFINE(struct pfil_head, link_pfil_hook); /* Packet filter hooks */
94 /* netgraph node hooks for ng_ether(4) */
95 void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp);
96 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m);
97 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
98 void (*ng_ether_attach_p)(struct ifnet *ifp);
99 void (*ng_ether_detach_p)(struct ifnet *ifp);
101 void (*vlan_input_p)(struct ifnet *, struct mbuf *);
103 /* if_bridge(4) support */
104 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *);
105 int (*bridge_output_p)(struct ifnet *, struct mbuf *,
106 struct sockaddr *, struct rtentry *);
107 void (*bridge_dn_p)(struct mbuf *, struct ifnet *);
109 /* if_lagg(4) support */
110 struct mbuf *(*lagg_input_p)(struct ifnet *, struct mbuf *);
112 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] =
113 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
115 static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
118 static void ether_reassign(struct ifnet *, struct vnet *, char *);
120 static int ether_requestencap(struct ifnet *, struct if_encap_req *);
123 #define senderr(e) do { error = (e); goto bad;} while (0)
126 update_mbuf_csumflags(struct mbuf *src, struct mbuf *dst)
130 if (src->m_pkthdr.csum_flags & CSUM_IP)
131 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
132 if (src->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
133 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
134 if (src->m_pkthdr.csum_flags & CSUM_SCTP)
135 csum_flags |= CSUM_SCTP_VALID;
136 dst->m_pkthdr.csum_flags |= csum_flags;
137 if (csum_flags & CSUM_DATA_VALID)
138 dst->m_pkthdr.csum_data = 0xffff;
142 * Handle link-layer encapsulation requests.
145 ether_requestencap(struct ifnet *ifp, struct if_encap_req *req)
147 struct ether_header *eh;
150 const u_char *lladdr;
152 if (req->rtype != IFENCAP_LL)
155 if (req->bufsize < ETHER_HDR_LEN)
158 eh = (struct ether_header *)req->buf;
159 lladdr = req->lladdr;
162 switch (req->family) {
164 etype = htons(ETHERTYPE_IP);
167 etype = htons(ETHERTYPE_IPV6);
170 ah = (struct arphdr *)req->hdata;
171 ah->ar_hrd = htons(ARPHRD_ETHER);
173 switch(ntohs(ah->ar_op)) {
174 case ARPOP_REVREQUEST:
176 etype = htons(ETHERTYPE_REVARP);
181 etype = htons(ETHERTYPE_ARP);
185 if (req->flags & IFENCAP_FLAG_BROADCAST)
186 lladdr = ifp->if_broadcastaddr;
189 return (EAFNOSUPPORT);
192 memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
193 memcpy(eh->ether_dhost, lladdr, ETHER_ADDR_LEN);
194 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
195 req->bufsize = sizeof(struct ether_header);
202 ether_resolve_addr(struct ifnet *ifp, struct mbuf *m,
203 const struct sockaddr *dst, struct route *ro, u_char *phdr,
204 uint32_t *pflags, struct llentry **plle)
206 struct ether_header *eh;
207 uint32_t lleflags = 0;
209 #if defined(INET) || defined(INET6)
215 eh = (struct ether_header *)phdr;
217 switch (dst->sa_family) {
220 if ((m->m_flags & (M_BCAST | M_MCAST)) == 0)
221 error = arpresolve(ifp, 0, m, dst, phdr, &lleflags,
224 if (m->m_flags & M_BCAST)
225 memcpy(eh->ether_dhost, ifp->if_broadcastaddr,
228 const struct in_addr *a;
229 a = &(((const struct sockaddr_in *)dst)->sin_addr);
230 ETHER_MAP_IP_MULTICAST(a, eh->ether_dhost);
232 etype = htons(ETHERTYPE_IP);
233 memcpy(&eh->ether_type, &etype, sizeof(etype));
234 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
240 if ((m->m_flags & M_MCAST) == 0)
241 error = nd6_resolve(ifp, 0, m, dst, phdr, &lleflags,
244 const struct in6_addr *a6;
245 a6 = &(((const struct sockaddr_in6 *)dst)->sin6_addr);
246 ETHER_MAP_IPV6_MULTICAST(a6, eh->ether_dhost);
247 etype = htons(ETHERTYPE_IPV6);
248 memcpy(&eh->ether_type, &etype, sizeof(etype));
249 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
254 if_printf(ifp, "can't handle af%d\n", dst->sa_family);
257 return (EAFNOSUPPORT);
260 if (error == EHOSTDOWN) {
261 if (ro != NULL && (ro->ro_flags & RT_HAS_GW) != 0)
262 error = EHOSTUNREACH;
268 *pflags = RT_MAY_LOOP;
269 if (lleflags & LLE_IFADDR)
276 * Ethernet output routine.
277 * Encapsulate a packet of type family for the local net.
278 * Use trailer local net encapsulation if enough data in first
279 * packet leaves a multiple of 512 bytes of data in remainder.
282 ether_output(struct ifnet *ifp, struct mbuf *m,
283 const struct sockaddr *dst, struct route *ro)
286 char linkhdr[ETHER_HDR_LEN], *phdr;
287 struct ether_header *eh;
290 int hlen; /* link layer header length */
292 struct llentry *lle = NULL;
298 /* XXX BPF uses ro_prepend */
299 if (ro->ro_prepend != NULL) {
300 phdr = ro->ro_prepend;
302 } else if (!(m->m_flags & (M_BCAST | M_MCAST))) {
303 if ((ro->ro_flags & RT_LLE_CACHE) != 0) {
306 (lle->la_flags & LLE_VALID) == 0) {
308 lle = NULL; /* redundant */
312 /* if we lookup, keep cache */
316 * Notify LLE code that
320 llentry_mark_used(lle);
323 phdr = lle->r_linkdata;
324 hlen = lle->r_hdrlen;
325 pflags = lle->r_flags;
331 error = mac_ifnet_check_transmit(ifp, m);
337 if (ifp->if_flags & IFF_MONITOR)
339 if (!((ifp->if_flags & IFF_UP) &&
340 (ifp->if_drv_flags & IFF_DRV_RUNNING)))
344 /* No prepend data supplied. Try to calculate ourselves. */
346 hlen = ETHER_HDR_LEN;
347 error = ether_resolve_addr(ifp, m, dst, ro, phdr, &pflags,
348 addref ? &lle : NULL);
349 if (addref && lle != NULL)
352 return (error == EWOULDBLOCK ? 0 : error);
355 if ((pflags & RT_L2_ME) != 0) {
356 update_mbuf_csumflags(m, m);
357 return (if_simloop(ifp, m, dst->sa_family, 0));
359 loop_copy = pflags & RT_MAY_LOOP;
362 * Add local net header. If no space in first mbuf,
365 * Note that we do prepend regardless of RT_HAS_HEADER flag.
366 * This is done because BPF code shifts m_data pointer
367 * to the end of ethernet header prior to calling if_output().
369 M_PREPEND(m, hlen, M_NOWAIT);
372 if ((pflags & RT_HAS_HEADER) == 0) {
373 eh = mtod(m, struct ether_header *);
374 memcpy(eh, phdr, hlen);
378 * If a simplex interface, and the packet is being sent to our
379 * Ethernet address or a broadcast address, loopback a copy.
380 * XXX To make a simplex device behave exactly like a duplex
381 * device, we should copy in the case of sending to our own
382 * ethernet address (thus letting the original actually appear
383 * on the wire). However, we don't do that here for security
384 * reasons and compatibility with the original behavior.
386 if ((m->m_flags & M_BCAST) && loop_copy && (ifp->if_flags & IFF_SIMPLEX) &&
387 ((t = pf_find_mtag(m)) == NULL || !t->routed)) {
391 * Because if_simloop() modifies the packet, we need a
392 * writable copy through m_dup() instead of a readonly
393 * one as m_copy[m] would give us. The alternative would
394 * be to modify if_simloop() to handle the readonly mbuf,
395 * but performancewise it is mostly equivalent (trading
396 * extra data copying vs. extra locking).
398 * XXX This is a local workaround. A number of less
399 * often used kernel parts suffer from the same bug.
400 * See PR kern/105943 for a proposed general solution.
402 if ((n = m_dup(m, M_NOWAIT)) != NULL) {
403 update_mbuf_csumflags(m, n);
404 (void)if_simloop(ifp, n, dst->sa_family, hlen);
406 if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1);
410 * Bridges require special output handling.
412 if (ifp->if_bridge) {
413 BRIDGE_OUTPUT(ifp, m, error);
417 #if defined(INET) || defined(INET6)
419 (error = (*carp_output_p)(ifp, m, dst)))
423 /* Handle ng_ether(4) processing, if any */
424 if (ifp->if_l2com != NULL) {
425 KASSERT(ng_ether_output_p != NULL,
426 ("ng_ether_output_p is NULL"));
427 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
436 /* Continue with link-layer output */
437 return ether_output_frame(ifp, m);
441 * Ethernet link layer output routine to send a raw frame to the device.
443 * This assumes that the 14 byte Ethernet header is present and contiguous
444 * in the first mbuf (if BRIDGE'ing).
447 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
451 if (PFIL_HOOKED(&V_link_pfil_hook)) {
452 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_OUT, NULL);
462 * Queue message on interface, update output statistics if
463 * successful, and start output if interface not yet active.
465 return ((ifp->if_transmit)(ifp, m));
469 * Process a received Ethernet packet; the packet is in the
470 * mbuf chain m with the ethernet header at the front.
473 ether_input_internal(struct ifnet *ifp, struct mbuf *m)
475 struct ether_header *eh;
478 if ((ifp->if_flags & IFF_UP) == 0) {
483 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
484 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n");
489 if (m->m_len < ETHER_HDR_LEN) {
490 /* XXX maybe should pullup? */
491 if_printf(ifp, "discard frame w/o leading ethernet "
492 "header (len %u pkt len %u)\n",
493 m->m_len, m->m_pkthdr.len);
494 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
498 eh = mtod(m, struct ether_header *);
499 etype = ntohs(eh->ether_type);
500 random_harvest_queue(m, sizeof(*m), 2, RANDOM_NET_ETHER);
502 CURVNET_SET_QUIET(ifp->if_vnet);
504 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
505 if (ETHER_IS_BROADCAST(eh->ether_dhost))
506 m->m_flags |= M_BCAST;
508 m->m_flags |= M_MCAST;
509 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
514 * Tag the mbuf with an appropriate MAC label before any other
515 * consumers can get to it.
517 mac_ifnet_create_mbuf(ifp, m);
521 * Give bpf a chance at the packet.
523 ETHER_BPF_MTAP(ifp, m);
526 * If the CRC is still on the packet, trim it off. We do this once
527 * and once only in case we are re-entered. Nothing else on the
528 * Ethernet receive path expects to see the FCS.
530 if (m->m_flags & M_HASFCS) {
531 m_adj(m, -ETHER_CRC_LEN);
532 m->m_flags &= ~M_HASFCS;
535 if (!(ifp->if_capenable & IFCAP_HWSTATS))
536 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
538 /* Allow monitor mode to claim this frame, after stats are updated. */
539 if (ifp->if_flags & IFF_MONITOR) {
545 /* Handle input from a lagg(4) port */
546 if (ifp->if_type == IFT_IEEE8023ADLAG) {
547 KASSERT(lagg_input_p != NULL,
548 ("%s: if_lagg not loaded!", __func__));
549 m = (*lagg_input_p)(ifp, m);
551 ifp = m->m_pkthdr.rcvif;
559 * If the hardware did not process an 802.1Q tag, do this now,
560 * to allow 802.1P priority frames to be passed to the main input
562 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels.
564 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) {
565 struct ether_vlan_header *evl;
567 if (m->m_len < sizeof(*evl) &&
568 (m = m_pullup(m, sizeof(*evl))) == NULL) {
570 if_printf(ifp, "cannot pullup VLAN header\n");
572 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
577 evl = mtod(m, struct ether_vlan_header *);
578 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag);
579 m->m_flags |= M_VLANTAG;
581 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN,
582 ETHER_HDR_LEN - ETHER_TYPE_LEN);
583 m_adj(m, ETHER_VLAN_ENCAP_LEN);
584 eh = mtod(m, struct ether_header *);
587 M_SETFIB(m, ifp->if_fib);
589 /* Allow ng_ether(4) to claim this frame. */
590 if (ifp->if_l2com != NULL) {
591 KASSERT(ng_ether_input_p != NULL,
592 ("%s: ng_ether_input_p is NULL", __func__));
593 m->m_flags &= ~M_PROMISC;
594 (*ng_ether_input_p)(ifp, &m);
599 eh = mtod(m, struct ether_header *);
603 * Allow if_bridge(4) to claim this frame.
604 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it
605 * and the frame should be delivered locally.
607 if (ifp->if_bridge != NULL) {
608 m->m_flags &= ~M_PROMISC;
609 BRIDGE_INPUT(ifp, m);
614 eh = mtod(m, struct ether_header *);
617 #if defined(INET) || defined(INET6)
619 * Clear M_PROMISC on frame so that carp(4) will see it when the
620 * mbuf flows up to Layer 3.
621 * FreeBSD's implementation of carp(4) uses the inprotosw
622 * to dispatch IPPROTO_CARP. carp(4) also allocates its own
623 * Ethernet addresses of the form 00:00:5e:00:01:xx, which
624 * is outside the scope of the M_PROMISC test below.
625 * TODO: Maintain a hash table of ethernet addresses other than
626 * ether_dhost which may be active on this ifp.
628 if (ifp->if_carp && (*carp_forus_p)(ifp, eh->ether_dhost)) {
629 m->m_flags &= ~M_PROMISC;
634 * If the frame received was not for our MAC address, set the
635 * M_PROMISC flag on the mbuf chain. The frame may need to
636 * be seen by the rest of the Ethernet input path in case of
637 * re-entry (e.g. bridge, vlan, netgraph) but should not be
638 * seen by upper protocol layers.
640 if (!ETHER_IS_MULTICAST(eh->ether_dhost) &&
641 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0)
642 m->m_flags |= M_PROMISC;
650 * Ethernet input dispatch; by default, direct dispatch here regardless of
651 * global configuration. However, if RSS is enabled, hook up RSS affinity
652 * so that when deferred or hybrid dispatch is enabled, we can redistribute
655 * XXXRW: Would be nice if the ifnet passed up a flag indicating whether or
656 * not it had already done work distribution via multi-queue. Then we could
657 * direct dispatch in the event load balancing was already complete and
658 * handle the case of interfaces with different capabilities better.
660 * XXXRW: Sort of want an M_DISTRIBUTED flag to avoid multiple distributions
661 * at multiple layers?
663 * XXXRW: For now, enable all this only if RSS is compiled in, although it
664 * works fine without RSS. Need to characterise the performance overhead
665 * of the detour through the netisr code in the event the result is always
669 ether_nh_input(struct mbuf *m)
673 KASSERT(m->m_pkthdr.rcvif != NULL,
674 ("%s: NULL interface pointer", __func__));
675 ether_input_internal(m->m_pkthdr.rcvif, m);
678 static struct netisr_handler ether_nh = {
680 .nh_handler = ether_nh_input,
681 .nh_proto = NETISR_ETHER,
683 .nh_policy = NETISR_POLICY_CPU,
684 .nh_dispatch = NETISR_DISPATCH_DIRECT,
685 .nh_m2cpuid = rss_m2cpuid,
687 .nh_policy = NETISR_POLICY_SOURCE,
688 .nh_dispatch = NETISR_DISPATCH_DIRECT,
693 ether_init(__unused void *arg)
696 netisr_register(ðer_nh);
698 SYSINIT(ether, SI_SUB_INIT_IF, SI_ORDER_ANY, ether_init, NULL);
701 vnet_ether_init(__unused void *arg)
705 /* Initialize packet filter hooks. */
706 V_link_pfil_hook.ph_type = PFIL_TYPE_AF;
707 V_link_pfil_hook.ph_af = AF_LINK;
708 if ((i = pfil_head_register(&V_link_pfil_hook)) != 0)
709 printf("%s: WARNING: unable to register pfil link hook, "
710 "error %d\n", __func__, i);
712 netisr_register_vnet(ðer_nh);
715 VNET_SYSINIT(vnet_ether_init, SI_SUB_PROTO_IF, SI_ORDER_ANY,
716 vnet_ether_init, NULL);
720 vnet_ether_pfil_destroy(__unused void *arg)
724 if ((i = pfil_head_unregister(&V_link_pfil_hook)) != 0)
725 printf("%s: WARNING: unable to unregister pfil link hook, "
726 "error %d\n", __func__, i);
728 VNET_SYSUNINIT(vnet_ether_pfil_uninit, SI_SUB_PROTO_PFIL, SI_ORDER_ANY,
729 vnet_ether_pfil_destroy, NULL);
732 vnet_ether_destroy(__unused void *arg)
735 netisr_unregister_vnet(ðer_nh);
737 VNET_SYSUNINIT(vnet_ether_uninit, SI_SUB_PROTO_IF, SI_ORDER_ANY,
738 vnet_ether_destroy, NULL);
744 ether_input(struct ifnet *ifp, struct mbuf *m)
750 * The drivers are allowed to pass in a chain of packets linked with
751 * m_nextpkt. We split them up into separate packets here and pass
752 * them up. This allows the drivers to amortize the receive lock.
759 * We will rely on rcvif being set properly in the deferred context,
760 * so assert it is correct here.
762 KASSERT(m->m_pkthdr.rcvif == ifp, ("%s: ifnet mismatch m %p "
763 "rcvif %p ifp %p", __func__, m, m->m_pkthdr.rcvif, ifp));
764 CURVNET_SET_QUIET(ifp->if_vnet);
765 netisr_dispatch(NETISR_ETHER, m);
772 * Upper layer processing for a received Ethernet packet.
775 ether_demux(struct ifnet *ifp, struct mbuf *m)
777 struct ether_header *eh;
781 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__));
783 /* Do not grab PROMISC frames in case we are re-entered. */
784 if (PFIL_HOOKED(&V_link_pfil_hook) && !(m->m_flags & M_PROMISC)) {
785 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_IN, NULL);
787 if (i != 0 || m == NULL)
791 eh = mtod(m, struct ether_header *);
792 ether_type = ntohs(eh->ether_type);
795 * If this frame has a VLAN tag other than 0, call vlan_input()
796 * if its module is loaded. Otherwise, drop.
798 if ((m->m_flags & M_VLANTAG) &&
799 EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) {
800 if (ifp->if_vlantrunk == NULL) {
801 if_inc_counter(ifp, IFCOUNTER_NOPROTO, 1);
805 KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!",
807 /* Clear before possibly re-entering ether_input(). */
808 m->m_flags &= ~M_PROMISC;
809 (*vlan_input_p)(ifp, m);
814 * Pass promiscuously received frames to the upper layer if the user
815 * requested this by setting IFF_PPROMISC. Otherwise, drop them.
817 if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) {
823 * Reset layer specific mbuf flags to avoid confusing upper layers.
824 * Strip off Ethernet header.
826 m->m_flags &= ~M_VLANTAG;
828 m_adj(m, ETHER_HDR_LEN);
831 * Dispatch frame to upper layer.
833 switch (ether_type) {
840 if (ifp->if_flags & IFF_NOARP) {
841 /* Discard packet if ARP is disabled on interface */
856 netisr_dispatch(isr, m);
861 * Packet is to be discarded. If netgraph is present,
862 * hand the packet to it for last chance processing;
863 * otherwise dispose of it.
865 if (ifp->if_l2com != NULL) {
866 KASSERT(ng_ether_input_orphan_p != NULL,
867 ("ng_ether_input_orphan_p is NULL"));
869 * Put back the ethernet header so netgraph has a
870 * consistent view of inbound packets.
872 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
873 (*ng_ether_input_orphan_p)(ifp, m);
880 * Convert Ethernet address to printable (loggable) representation.
881 * This routine is for compatibility; it's better to just use
883 * printf("%6D", <pointer to address>, ":");
885 * since there's no static buffer involved.
888 ether_sprintf(const u_char *ap)
890 static char etherbuf[18];
891 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
896 * Perform common duties while attaching to interface list
899 ether_ifattach(struct ifnet *ifp, const u_int8_t *lla)
903 struct sockaddr_dl *sdl;
905 ifp->if_addrlen = ETHER_ADDR_LEN;
906 ifp->if_hdrlen = ETHER_HDR_LEN;
908 ifp->if_mtu = ETHERMTU;
909 ifp->if_output = ether_output;
910 ifp->if_input = ether_input;
911 ifp->if_resolvemulti = ether_resolvemulti;
912 ifp->if_requestencap = ether_requestencap;
914 ifp->if_reassign = ether_reassign;
916 if (ifp->if_baudrate == 0)
917 ifp->if_baudrate = IF_Mbps(10); /* just a default */
918 ifp->if_broadcastaddr = etherbroadcastaddr;
921 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
922 sdl = (struct sockaddr_dl *)ifa->ifa_addr;
923 sdl->sdl_type = IFT_ETHER;
924 sdl->sdl_alen = ifp->if_addrlen;
925 bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
927 if (ifp->if_hw_addr != NULL)
928 bcopy(lla, ifp->if_hw_addr, ifp->if_addrlen);
930 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
931 if (ng_ether_attach_p != NULL)
932 (*ng_ether_attach_p)(ifp);
934 /* Announce Ethernet MAC address if non-zero. */
935 for (i = 0; i < ifp->if_addrlen; i++)
938 if (i != ifp->if_addrlen)
939 if_printf(ifp, "Ethernet address: %6D\n", lla, ":");
941 uuid_ether_add(LLADDR(sdl));
943 /* Add necessary bits are setup; announce it now. */
944 EVENTHANDLER_INVOKE(ether_ifattach_event, ifp);
945 if (IS_DEFAULT_VNET(curvnet))
946 devctl_notify("ETHERNET", ifp->if_xname, "IFATTACH", NULL);
950 * Perform common duties while detaching an Ethernet interface
953 ether_ifdetach(struct ifnet *ifp)
955 struct sockaddr_dl *sdl;
957 sdl = (struct sockaddr_dl *)(ifp->if_addr->ifa_addr);
958 uuid_ether_del(LLADDR(sdl));
960 if (ifp->if_l2com != NULL) {
961 KASSERT(ng_ether_detach_p != NULL,
962 ("ng_ether_detach_p is NULL"));
963 (*ng_ether_detach_p)(ifp);
972 ether_reassign(struct ifnet *ifp, struct vnet *new_vnet, char *unused __unused)
975 if (ifp->if_l2com != NULL) {
976 KASSERT(ng_ether_detach_p != NULL,
977 ("ng_ether_detach_p is NULL"));
978 (*ng_ether_detach_p)(ifp);
981 if (ng_ether_attach_p != NULL) {
982 CURVNET_SET_QUIET(new_vnet);
983 (*ng_ether_attach_p)(ifp);
989 SYSCTL_DECL(_net_link);
990 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
994 * This is for reference. We have a table-driven version
995 * of the little-endian crc32 generator, which is faster
996 * than the double-loop.
999 ether_crc32_le(const uint8_t *buf, size_t len)
1006 crc = 0xffffffff; /* initial value */
1008 for (i = 0; i < len; i++) {
1009 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
1010 carry = (crc ^ data) & 1;
1013 crc = (crc ^ ETHER_CRC_POLY_LE);
1021 ether_crc32_le(const uint8_t *buf, size_t len)
1023 static const uint32_t crctab[] = {
1024 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1025 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1026 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1027 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1032 crc = 0xffffffff; /* initial value */
1034 for (i = 0; i < len; i++) {
1036 crc = (crc >> 4) ^ crctab[crc & 0xf];
1037 crc = (crc >> 4) ^ crctab[crc & 0xf];
1045 ether_crc32_be(const uint8_t *buf, size_t len)
1048 uint32_t crc, carry;
1052 crc = 0xffffffff; /* initial value */
1054 for (i = 0; i < len; i++) {
1055 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
1056 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
1059 crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1067 ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
1069 struct ifaddr *ifa = (struct ifaddr *) data;
1070 struct ifreq *ifr = (struct ifreq *) data;
1075 ifp->if_flags |= IFF_UP;
1077 switch (ifa->ifa_addr->sa_family) {
1080 ifp->if_init(ifp->if_softc); /* before arpwhohas */
1081 arp_ifinit(ifp, ifa);
1085 ifp->if_init(ifp->if_softc);
1092 struct sockaddr *sa;
1094 sa = (struct sockaddr *) & ifr->ifr_data;
1095 bcopy(IF_LLADDR(ifp),
1096 (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
1102 * Set the interface MTU.
1104 if (ifr->ifr_mtu > ETHERMTU) {
1107 ifp->if_mtu = ifr->ifr_mtu;
1111 error = EINVAL; /* XXX netbsd has ENOTTY??? */
1118 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
1119 struct sockaddr *sa)
1121 struct sockaddr_dl *sdl;
1123 struct sockaddr_in *sin;
1126 struct sockaddr_in6 *sin6;
1130 switch(sa->sa_family) {
1133 * No mapping needed. Just check that it's a valid MC address.
1135 sdl = (struct sockaddr_dl *)sa;
1136 e_addr = LLADDR(sdl);
1137 if (!ETHER_IS_MULTICAST(e_addr))
1138 return EADDRNOTAVAIL;
1144 sin = (struct sockaddr_in *)sa;
1145 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1146 return EADDRNOTAVAIL;
1147 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1148 sdl->sdl_alen = ETHER_ADDR_LEN;
1149 e_addr = LLADDR(sdl);
1150 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1151 *llsa = (struct sockaddr *)sdl;
1156 sin6 = (struct sockaddr_in6 *)sa;
1157 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1159 * An IP6 address of 0 means listen to all
1160 * of the Ethernet multicast address used for IP6.
1161 * (This is used for multicast routers.)
1163 ifp->if_flags |= IFF_ALLMULTI;
1167 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1168 return EADDRNOTAVAIL;
1169 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1170 sdl->sdl_alen = ETHER_ADDR_LEN;
1171 e_addr = LLADDR(sdl);
1172 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1173 *llsa = (struct sockaddr *)sdl;
1179 * Well, the text isn't quite right, but it's the name
1182 return EAFNOSUPPORT;
1186 static moduledata_t ether_mod = {
1191 ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen)
1193 struct ether_vlan_header vlan;
1196 KASSERT((m->m_flags & M_VLANTAG) != 0,
1197 ("%s: vlan information not present", __func__));
1198 KASSERT(m->m_len >= sizeof(struct ether_header),
1199 ("%s: mbuf not large enough for header", __func__));
1200 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header));
1201 vlan.evl_proto = vlan.evl_encap_proto;
1202 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN);
1203 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag);
1204 m->m_len -= sizeof(struct ether_header);
1205 m->m_data += sizeof(struct ether_header);
1207 * If a data link has been supplied by the caller, then we will need to
1208 * re-create a stack allocated mbuf chain with the following structure:
1210 * (1) mbuf #1 will contain the supplied data link
1211 * (2) mbuf #2 will contain the vlan header
1212 * (3) mbuf #3 will contain the original mbuf's packet data
1214 * Otherwise, submit the packet and vlan header via bpf_mtap2().
1218 mv.m_data = (caddr_t)&vlan;
1219 mv.m_len = sizeof(vlan);
1225 bpf_mtap2(bp, &vlan, sizeof(vlan), m);
1226 m->m_len += sizeof(struct ether_header);
1227 m->m_data -= sizeof(struct ether_header);
1231 ether_vlanencap(struct mbuf *m, uint16_t tag)
1233 struct ether_vlan_header *evl;
1235 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_NOWAIT);
1238 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */
1240 if (m->m_len < sizeof(*evl)) {
1241 m = m_pullup(m, sizeof(*evl));
1247 * Transform the Ethernet header into an Ethernet header
1248 * with 802.1Q encapsulation.
1250 evl = mtod(m, struct ether_vlan_header *);
1251 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN,
1252 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN);
1253 evl->evl_encap_proto = htons(ETHERTYPE_VLAN);
1254 evl->evl_tag = htons(tag);
1258 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY);
1259 MODULE_VERSION(ether, 1);