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 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * @(#)if_ethersubr.c 8.1 (Berkeley) 6/10/93
34 #include "opt_inet6.h"
35 #include "opt_netgraph.h"
36 #include "opt_mbuf_profiling.h"
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/kernel.h>
43 #include <sys/malloc.h>
44 #include <sys/module.h>
46 #include <sys/random.h>
47 #include <sys/socket.h>
48 #include <sys/sockio.h>
49 #include <sys/sysctl.h>
53 #include <net/if_var.h>
54 #include <net/if_arp.h>
55 #include <net/netisr.h>
56 #include <net/route.h>
57 #include <net/if_llc.h>
58 #include <net/if_dl.h>
59 #include <net/if_types.h>
61 #include <net/ethernet.h>
62 #include <net/if_bridgevar.h>
63 #include <net/if_vlan_var.h>
64 #include <net/if_llatbl.h>
66 #include <net/rss_config.h>
69 #include <netpfil/pf/pf_mtag.h>
71 #if defined(INET) || defined(INET6)
72 #include <netinet/in.h>
73 #include <netinet/in_var.h>
74 #include <netinet/if_ether.h>
75 #include <netinet/ip_carp.h>
76 #include <netinet/ip_var.h>
79 #include <netinet6/nd6.h>
81 #include <security/mac/mac_framework.h>
84 CTASSERT(sizeof (struct ether_header) == ETHER_ADDR_LEN * 2 + 2);
85 CTASSERT(sizeof (struct ether_addr) == ETHER_ADDR_LEN);
88 VNET_DEFINE(struct pfil_head, link_pfil_hook); /* Packet filter hooks */
90 /* netgraph node hooks for ng_ether(4) */
91 void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp);
92 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m);
93 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
94 void (*ng_ether_attach_p)(struct ifnet *ifp);
95 void (*ng_ether_detach_p)(struct ifnet *ifp);
97 void (*vlan_input_p)(struct ifnet *, struct mbuf *);
99 /* if_bridge(4) support */
100 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *);
101 int (*bridge_output_p)(struct ifnet *, struct mbuf *,
102 struct sockaddr *, struct rtentry *);
103 void (*bridge_dn_p)(struct mbuf *, struct ifnet *);
105 /* if_lagg(4) support */
106 struct mbuf *(*lagg_input_p)(struct ifnet *, struct mbuf *);
108 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] =
109 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
111 static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
114 static void ether_reassign(struct ifnet *, struct vnet *, char *);
117 #define ETHER_IS_BROADCAST(addr) \
118 (bcmp(etherbroadcastaddr, (addr), ETHER_ADDR_LEN) == 0)
120 #define senderr(e) do { error = (e); goto bad;} while (0)
123 update_mbuf_csumflags(struct mbuf *src, struct mbuf *dst)
127 if (src->m_pkthdr.csum_flags & CSUM_IP)
128 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
129 if (src->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
130 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
131 if (src->m_pkthdr.csum_flags & CSUM_SCTP)
132 csum_flags |= CSUM_SCTP_VALID;
133 dst->m_pkthdr.csum_flags |= csum_flags;
134 if (csum_flags & CSUM_DATA_VALID)
135 dst->m_pkthdr.csum_data = 0xffff;
139 * Ethernet output routine.
140 * Encapsulate a packet of type family for the local net.
141 * Use trailer local net encapsulation if enough data in first
142 * packet leaves a multiple of 512 bytes of data in remainder.
145 ether_output(struct ifnet *ifp, struct mbuf *m,
146 const struct sockaddr *dst, struct route *ro)
149 int error = 0, hdrcmplt = 0;
150 u_char edst[ETHER_ADDR_LEN];
151 struct llentry *lle = NULL;
152 struct rtentry *rt0 = NULL;
153 struct ether_header *eh;
156 int hlen; /* link layer header length */
161 if (!(m->m_flags & (M_BCAST | M_MCAST))) {
164 pflags = lle->la_flags;
167 if (rt0 != NULL && (rt0->rt_flags & RTF_GATEWAY) != 0)
171 error = mac_ifnet_check_transmit(ifp, m);
177 if (ifp->if_flags & IFF_MONITOR)
179 if (!((ifp->if_flags & IFF_UP) &&
180 (ifp->if_drv_flags & IFF_DRV_RUNNING)))
183 hlen = ETHER_HDR_LEN;
184 switch (dst->sa_family) {
187 if (lle != NULL && (pflags & LLE_VALID) != 0)
188 memcpy(edst, &lle->ll_addr.mac16, sizeof(edst));
190 error = arpresolve(ifp, is_gw, m, dst, edst, &pflags);
192 return (error == EWOULDBLOCK ? 0 : error);
193 type = htons(ETHERTYPE_IP);
198 ah = mtod(m, struct arphdr *);
199 ah->ar_hrd = htons(ARPHRD_ETHER);
201 loop_copy = 0; /* if this is for us, don't do it */
203 switch(ntohs(ah->ar_op)) {
204 case ARPOP_REVREQUEST:
206 type = htons(ETHERTYPE_REVARP);
211 type = htons(ETHERTYPE_ARP);
215 if (m->m_flags & M_BCAST)
216 bcopy(ifp->if_broadcastaddr, edst, ETHER_ADDR_LEN);
218 bcopy(ar_tha(ah), edst, ETHER_ADDR_LEN);
225 if (lle != NULL && (pflags & LLE_VALID))
226 memcpy(edst, &lle->ll_addr.mac16, sizeof(edst));
228 error = nd6_resolve(ifp, is_gw, m, dst, (u_char *)edst,
231 return (error == EWOULDBLOCK ? 0 : error);
232 type = htons(ETHERTYPE_IPV6);
235 case pseudo_AF_HDRCMPLT:
237 const struct ether_header *eh;
243 loop_copy = 0; /* if this is for us, don't do it */
244 eh = (const struct ether_header *)dst->sa_data;
245 (void)memcpy(edst, eh->ether_dhost, sizeof (edst));
246 type = eh->ether_type;
250 if_printf(ifp, "can't handle af%d\n", dst->sa_family);
251 senderr(EAFNOSUPPORT);
254 if ((pflags & LLE_IFADDR) != 0) {
255 update_mbuf_csumflags(m, m);
256 return (if_simloop(ifp, m, dst->sa_family, 0));
260 * Add local net header. If no space in first mbuf,
263 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
266 eh = mtod(m, struct ether_header *);
268 memcpy(&eh->ether_type, &type, sizeof(eh->ether_type));
269 memcpy(eh->ether_dhost, edst, sizeof (edst));
270 memcpy(eh->ether_shost, IF_LLADDR(ifp),sizeof(eh->ether_shost));
274 * If a simplex interface, and the packet is being sent to our
275 * Ethernet address or a broadcast address, loopback a copy.
276 * XXX To make a simplex device behave exactly like a duplex
277 * device, we should copy in the case of sending to our own
278 * ethernet address (thus letting the original actually appear
279 * on the wire). However, we don't do that here for security
280 * reasons and compatibility with the original behavior.
282 if ((ifp->if_flags & IFF_SIMPLEX) && loop_copy &&
283 ((t = pf_find_mtag(m)) == NULL || !t->routed)) {
284 if (m->m_flags & M_BCAST) {
288 * Because if_simloop() modifies the packet, we need a
289 * writable copy through m_dup() instead of a readonly
290 * one as m_copy[m] would give us. The alternative would
291 * be to modify if_simloop() to handle the readonly mbuf,
292 * but performancewise it is mostly equivalent (trading
293 * extra data copying vs. extra locking).
295 * XXX This is a local workaround. A number of less
296 * often used kernel parts suffer from the same bug.
297 * See PR kern/105943 for a proposed general solution.
299 if ((n = m_dup(m, M_NOWAIT)) != NULL) {
300 update_mbuf_csumflags(m, n);
301 (void)if_simloop(ifp, n, dst->sa_family, hlen);
303 if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1);
304 } else if (bcmp(eh->ether_dhost, eh->ether_shost,
305 ETHER_ADDR_LEN) == 0) {
306 update_mbuf_csumflags(m, m);
307 (void) if_simloop(ifp, m, dst->sa_family, hlen);
308 return (0); /* XXX */
313 * Bridges require special output handling.
315 if (ifp->if_bridge) {
316 BRIDGE_OUTPUT(ifp, m, error);
320 #if defined(INET) || defined(INET6)
322 (error = (*carp_output_p)(ifp, m, dst)))
326 /* Handle ng_ether(4) processing, if any */
327 if (ifp->if_l2com != NULL) {
328 KASSERT(ng_ether_output_p != NULL,
329 ("ng_ether_output_p is NULL"));
330 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
339 /* Continue with link-layer output */
340 return ether_output_frame(ifp, m);
344 * Ethernet link layer output routine to send a raw frame to the device.
346 * This assumes that the 14 byte Ethernet header is present and contiguous
347 * in the first mbuf (if BRIDGE'ing).
350 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
354 if (PFIL_HOOKED(&V_link_pfil_hook)) {
355 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_OUT, NULL);
365 * Queue message on interface, update output statistics if
366 * successful, and start output if interface not yet active.
368 return ((ifp->if_transmit)(ifp, m));
371 #if defined(INET) || defined(INET6)
375 * Process a received Ethernet packet; the packet is in the
376 * mbuf chain m with the ethernet header at the front.
379 ether_input_internal(struct ifnet *ifp, struct mbuf *m)
381 struct ether_header *eh;
384 if ((ifp->if_flags & IFF_UP) == 0) {
389 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
390 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n");
395 if (m->m_len < ETHER_HDR_LEN) {
396 /* XXX maybe should pullup? */
397 if_printf(ifp, "discard frame w/o leading ethernet "
398 "header (len %u pkt len %u)\n",
399 m->m_len, m->m_pkthdr.len);
400 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
404 eh = mtod(m, struct ether_header *);
405 etype = ntohs(eh->ether_type);
406 random_harvest_queue(m, sizeof(*m), 2, RANDOM_NET_ETHER);
408 CURVNET_SET_QUIET(ifp->if_vnet);
410 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
411 if (ETHER_IS_BROADCAST(eh->ether_dhost))
412 m->m_flags |= M_BCAST;
414 m->m_flags |= M_MCAST;
415 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
420 * Tag the mbuf with an appropriate MAC label before any other
421 * consumers can get to it.
423 mac_ifnet_create_mbuf(ifp, m);
427 * Give bpf a chance at the packet.
429 ETHER_BPF_MTAP(ifp, m);
432 * If the CRC is still on the packet, trim it off. We do this once
433 * and once only in case we are re-entered. Nothing else on the
434 * Ethernet receive path expects to see the FCS.
436 if (m->m_flags & M_HASFCS) {
437 m_adj(m, -ETHER_CRC_LEN);
438 m->m_flags &= ~M_HASFCS;
441 if (!(ifp->if_capenable & IFCAP_HWSTATS))
442 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
444 /* Allow monitor mode to claim this frame, after stats are updated. */
445 if (ifp->if_flags & IFF_MONITOR) {
451 /* Handle input from a lagg(4) port */
452 if (ifp->if_type == IFT_IEEE8023ADLAG) {
453 KASSERT(lagg_input_p != NULL,
454 ("%s: if_lagg not loaded!", __func__));
455 m = (*lagg_input_p)(ifp, m);
457 ifp = m->m_pkthdr.rcvif;
465 * If the hardware did not process an 802.1Q tag, do this now,
466 * to allow 802.1P priority frames to be passed to the main input
468 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels.
470 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) {
471 struct ether_vlan_header *evl;
473 if (m->m_len < sizeof(*evl) &&
474 (m = m_pullup(m, sizeof(*evl))) == NULL) {
476 if_printf(ifp, "cannot pullup VLAN header\n");
478 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
483 evl = mtod(m, struct ether_vlan_header *);
484 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag);
485 m->m_flags |= M_VLANTAG;
487 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN,
488 ETHER_HDR_LEN - ETHER_TYPE_LEN);
489 m_adj(m, ETHER_VLAN_ENCAP_LEN);
490 eh = mtod(m, struct ether_header *);
493 M_SETFIB(m, ifp->if_fib);
495 /* Allow ng_ether(4) to claim this frame. */
496 if (ifp->if_l2com != NULL) {
497 KASSERT(ng_ether_input_p != NULL,
498 ("%s: ng_ether_input_p is NULL", __func__));
499 m->m_flags &= ~M_PROMISC;
500 (*ng_ether_input_p)(ifp, &m);
505 eh = mtod(m, struct ether_header *);
509 * Allow if_bridge(4) to claim this frame.
510 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it
511 * and the frame should be delivered locally.
513 if (ifp->if_bridge != NULL) {
514 m->m_flags &= ~M_PROMISC;
515 BRIDGE_INPUT(ifp, m);
520 eh = mtod(m, struct ether_header *);
523 #if defined(INET) || defined(INET6)
525 * Clear M_PROMISC on frame so that carp(4) will see it when the
526 * mbuf flows up to Layer 3.
527 * FreeBSD's implementation of carp(4) uses the inprotosw
528 * to dispatch IPPROTO_CARP. carp(4) also allocates its own
529 * Ethernet addresses of the form 00:00:5e:00:01:xx, which
530 * is outside the scope of the M_PROMISC test below.
531 * TODO: Maintain a hash table of ethernet addresses other than
532 * ether_dhost which may be active on this ifp.
534 if (ifp->if_carp && (*carp_forus_p)(ifp, eh->ether_dhost)) {
535 m->m_flags &= ~M_PROMISC;
540 * If the frame received was not for our MAC address, set the
541 * M_PROMISC flag on the mbuf chain. The frame may need to
542 * be seen by the rest of the Ethernet input path in case of
543 * re-entry (e.g. bridge, vlan, netgraph) but should not be
544 * seen by upper protocol layers.
546 if (!ETHER_IS_MULTICAST(eh->ether_dhost) &&
547 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0)
548 m->m_flags |= M_PROMISC;
556 * Ethernet input dispatch; by default, direct dispatch here regardless of
557 * global configuration. However, if RSS is enabled, hook up RSS affinity
558 * so that when deferred or hybrid dispatch is enabled, we can redistribute
561 * XXXRW: Would be nice if the ifnet passed up a flag indicating whether or
562 * not it had already done work distribution via multi-queue. Then we could
563 * direct dispatch in the event load balancing was already complete and
564 * handle the case of interfaces with different capabilities better.
566 * XXXRW: Sort of want an M_DISTRIBUTED flag to avoid multiple distributions
567 * at multiple layers?
569 * XXXRW: For now, enable all this only if RSS is compiled in, although it
570 * works fine without RSS. Need to characterise the performance overhead
571 * of the detour through the netisr code in the event the result is always
575 ether_nh_input(struct mbuf *m)
579 KASSERT(m->m_pkthdr.rcvif != NULL,
580 ("%s: NULL interface pointer", __func__));
581 ether_input_internal(m->m_pkthdr.rcvif, m);
584 static struct netisr_handler ether_nh = {
586 .nh_handler = ether_nh_input,
587 .nh_proto = NETISR_ETHER,
589 .nh_policy = NETISR_POLICY_CPU,
590 .nh_dispatch = NETISR_DISPATCH_DIRECT,
591 .nh_m2cpuid = rss_m2cpuid,
593 .nh_policy = NETISR_POLICY_SOURCE,
594 .nh_dispatch = NETISR_DISPATCH_DIRECT,
599 ether_init(__unused void *arg)
602 netisr_register(ðer_nh);
604 SYSINIT(ether, SI_SUB_INIT_IF, SI_ORDER_ANY, ether_init, NULL);
607 vnet_ether_init(__unused void *arg)
611 /* Initialize packet filter hooks. */
612 V_link_pfil_hook.ph_type = PFIL_TYPE_AF;
613 V_link_pfil_hook.ph_af = AF_LINK;
614 if ((i = pfil_head_register(&V_link_pfil_hook)) != 0)
615 printf("%s: WARNING: unable to register pfil link hook, "
616 "error %d\n", __func__, i);
618 VNET_SYSINIT(vnet_ether_init, SI_SUB_PROTO_IF, SI_ORDER_ANY,
619 vnet_ether_init, NULL);
622 vnet_ether_destroy(__unused void *arg)
626 if ((i = pfil_head_unregister(&V_link_pfil_hook)) != 0)
627 printf("%s: WARNING: unable to unregister pfil link hook, "
628 "error %d\n", __func__, i);
630 VNET_SYSUNINIT(vnet_ether_uninit, SI_SUB_PROTO_IF, SI_ORDER_ANY,
631 vnet_ether_destroy, NULL);
636 ether_input(struct ifnet *ifp, struct mbuf *m)
642 * The drivers are allowed to pass in a chain of packets linked with
643 * m_nextpkt. We split them up into separate packets here and pass
644 * them up. This allows the drivers to amortize the receive lock.
651 * We will rely on rcvif being set properly in the deferred context,
652 * so assert it is correct here.
654 KASSERT(m->m_pkthdr.rcvif == ifp, ("%s: ifnet mismatch", __func__));
655 netisr_dispatch(NETISR_ETHER, m);
661 * Upper layer processing for a received Ethernet packet.
664 ether_demux(struct ifnet *ifp, struct mbuf *m)
666 struct ether_header *eh;
670 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__));
672 /* Do not grab PROMISC frames in case we are re-entered. */
673 if (PFIL_HOOKED(&V_link_pfil_hook) && !(m->m_flags & M_PROMISC)) {
674 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_IN, NULL);
676 if (i != 0 || m == NULL)
680 eh = mtod(m, struct ether_header *);
681 ether_type = ntohs(eh->ether_type);
684 * If this frame has a VLAN tag other than 0, call vlan_input()
685 * if its module is loaded. Otherwise, drop.
687 if ((m->m_flags & M_VLANTAG) &&
688 EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) {
689 if (ifp->if_vlantrunk == NULL) {
690 if_inc_counter(ifp, IFCOUNTER_NOPROTO, 1);
694 KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!",
696 /* Clear before possibly re-entering ether_input(). */
697 m->m_flags &= ~M_PROMISC;
698 (*vlan_input_p)(ifp, m);
703 * Pass promiscuously received frames to the upper layer if the user
704 * requested this by setting IFF_PPROMISC. Otherwise, drop them.
706 if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) {
712 * Reset layer specific mbuf flags to avoid confusing upper layers.
713 * Strip off Ethernet header.
715 m->m_flags &= ~M_VLANTAG;
717 m_adj(m, ETHER_HDR_LEN);
720 * Dispatch frame to upper layer.
722 switch (ether_type) {
729 if (ifp->if_flags & IFF_NOARP) {
730 /* Discard packet if ARP is disabled on interface */
745 netisr_dispatch(isr, m);
750 * Packet is to be discarded. If netgraph is present,
751 * hand the packet to it for last chance processing;
752 * otherwise dispose of it.
754 if (ifp->if_l2com != NULL) {
755 KASSERT(ng_ether_input_orphan_p != NULL,
756 ("ng_ether_input_orphan_p is NULL"));
758 * Put back the ethernet header so netgraph has a
759 * consistent view of inbound packets.
761 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
762 (*ng_ether_input_orphan_p)(ifp, m);
769 * Convert Ethernet address to printable (loggable) representation.
770 * This routine is for compatibility; it's better to just use
772 * printf("%6D", <pointer to address>, ":");
774 * since there's no static buffer involved.
777 ether_sprintf(const u_char *ap)
779 static char etherbuf[18];
780 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
785 * Perform common duties while attaching to interface list
788 ether_ifattach(struct ifnet *ifp, const u_int8_t *lla)
792 struct sockaddr_dl *sdl;
794 ifp->if_addrlen = ETHER_ADDR_LEN;
795 ifp->if_hdrlen = ETHER_HDR_LEN;
797 ifp->if_mtu = ETHERMTU;
798 ifp->if_output = ether_output;
799 ifp->if_input = ether_input;
800 ifp->if_resolvemulti = ether_resolvemulti;
802 ifp->if_reassign = ether_reassign;
804 if (ifp->if_baudrate == 0)
805 ifp->if_baudrate = IF_Mbps(10); /* just a default */
806 ifp->if_broadcastaddr = etherbroadcastaddr;
809 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
810 sdl = (struct sockaddr_dl *)ifa->ifa_addr;
811 sdl->sdl_type = IFT_ETHER;
812 sdl->sdl_alen = ifp->if_addrlen;
813 bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
815 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
816 if (ng_ether_attach_p != NULL)
817 (*ng_ether_attach_p)(ifp);
819 /* Announce Ethernet MAC address if non-zero. */
820 for (i = 0; i < ifp->if_addrlen; i++)
823 if (i != ifp->if_addrlen)
824 if_printf(ifp, "Ethernet address: %6D\n", lla, ":");
826 uuid_ether_add(LLADDR(sdl));
830 * Perform common duties while detaching an Ethernet interface
833 ether_ifdetach(struct ifnet *ifp)
835 struct sockaddr_dl *sdl;
837 sdl = (struct sockaddr_dl *)(ifp->if_addr->ifa_addr);
838 uuid_ether_del(LLADDR(sdl));
840 if (ifp->if_l2com != NULL) {
841 KASSERT(ng_ether_detach_p != NULL,
842 ("ng_ether_detach_p is NULL"));
843 (*ng_ether_detach_p)(ifp);
852 ether_reassign(struct ifnet *ifp, struct vnet *new_vnet, char *unused __unused)
855 if (ifp->if_l2com != NULL) {
856 KASSERT(ng_ether_detach_p != NULL,
857 ("ng_ether_detach_p is NULL"));
858 (*ng_ether_detach_p)(ifp);
861 if (ng_ether_attach_p != NULL) {
862 CURVNET_SET_QUIET(new_vnet);
863 (*ng_ether_attach_p)(ifp);
869 SYSCTL_DECL(_net_link);
870 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
874 * This is for reference. We have a table-driven version
875 * of the little-endian crc32 generator, which is faster
876 * than the double-loop.
879 ether_crc32_le(const uint8_t *buf, size_t len)
886 crc = 0xffffffff; /* initial value */
888 for (i = 0; i < len; i++) {
889 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
890 carry = (crc ^ data) & 1;
893 crc = (crc ^ ETHER_CRC_POLY_LE);
901 ether_crc32_le(const uint8_t *buf, size_t len)
903 static const uint32_t crctab[] = {
904 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
905 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
906 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
907 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
912 crc = 0xffffffff; /* initial value */
914 for (i = 0; i < len; i++) {
916 crc = (crc >> 4) ^ crctab[crc & 0xf];
917 crc = (crc >> 4) ^ crctab[crc & 0xf];
925 ether_crc32_be(const uint8_t *buf, size_t len)
932 crc = 0xffffffff; /* initial value */
934 for (i = 0; i < len; i++) {
935 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
936 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
939 crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
947 ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
949 struct ifaddr *ifa = (struct ifaddr *) data;
950 struct ifreq *ifr = (struct ifreq *) data;
955 ifp->if_flags |= IFF_UP;
957 switch (ifa->ifa_addr->sa_family) {
960 ifp->if_init(ifp->if_softc); /* before arpwhohas */
961 arp_ifinit(ifp, ifa);
965 ifp->if_init(ifp->if_softc);
974 sa = (struct sockaddr *) & ifr->ifr_data;
975 bcopy(IF_LLADDR(ifp),
976 (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
982 * Set the interface MTU.
984 if (ifr->ifr_mtu > ETHERMTU) {
987 ifp->if_mtu = ifr->ifr_mtu;
991 error = EINVAL; /* XXX netbsd has ENOTTY??? */
998 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
1001 struct sockaddr_dl *sdl;
1003 struct sockaddr_in *sin;
1006 struct sockaddr_in6 *sin6;
1010 switch(sa->sa_family) {
1013 * No mapping needed. Just check that it's a valid MC address.
1015 sdl = (struct sockaddr_dl *)sa;
1016 e_addr = LLADDR(sdl);
1017 if (!ETHER_IS_MULTICAST(e_addr))
1018 return EADDRNOTAVAIL;
1024 sin = (struct sockaddr_in *)sa;
1025 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1026 return EADDRNOTAVAIL;
1027 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1028 sdl->sdl_alen = ETHER_ADDR_LEN;
1029 e_addr = LLADDR(sdl);
1030 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1031 *llsa = (struct sockaddr *)sdl;
1036 sin6 = (struct sockaddr_in6 *)sa;
1037 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1039 * An IP6 address of 0 means listen to all
1040 * of the Ethernet multicast address used for IP6.
1041 * (This is used for multicast routers.)
1043 ifp->if_flags |= IFF_ALLMULTI;
1047 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1048 return EADDRNOTAVAIL;
1049 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1050 sdl->sdl_alen = ETHER_ADDR_LEN;
1051 e_addr = LLADDR(sdl);
1052 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1053 *llsa = (struct sockaddr *)sdl;
1059 * Well, the text isn't quite right, but it's the name
1062 return EAFNOSUPPORT;
1066 static moduledata_t ether_mod = {
1071 ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen)
1073 struct ether_vlan_header vlan;
1076 KASSERT((m->m_flags & M_VLANTAG) != 0,
1077 ("%s: vlan information not present", __func__));
1078 KASSERT(m->m_len >= sizeof(struct ether_header),
1079 ("%s: mbuf not large enough for header", __func__));
1080 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header));
1081 vlan.evl_proto = vlan.evl_encap_proto;
1082 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN);
1083 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag);
1084 m->m_len -= sizeof(struct ether_header);
1085 m->m_data += sizeof(struct ether_header);
1087 * If a data link has been supplied by the caller, then we will need to
1088 * re-create a stack allocated mbuf chain with the following structure:
1090 * (1) mbuf #1 will contain the supplied data link
1091 * (2) mbuf #2 will contain the vlan header
1092 * (3) mbuf #3 will contain the original mbuf's packet data
1094 * Otherwise, submit the packet and vlan header via bpf_mtap2().
1098 mv.m_data = (caddr_t)&vlan;
1099 mv.m_len = sizeof(vlan);
1105 bpf_mtap2(bp, &vlan, sizeof(vlan), m);
1106 m->m_len += sizeof(struct ether_header);
1107 m->m_data -= sizeof(struct ether_header);
1111 ether_vlanencap(struct mbuf *m, uint16_t tag)
1113 struct ether_vlan_header *evl;
1115 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_NOWAIT);
1118 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */
1120 if (m->m_len < sizeof(*evl)) {
1121 m = m_pullup(m, sizeof(*evl));
1127 * Transform the Ethernet header into an Ethernet header
1128 * with 802.1Q encapsulation.
1130 evl = mtod(m, struct ether_vlan_header *);
1131 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN,
1132 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN);
1133 evl->evl_encap_proto = htons(ETHERTYPE_VLAN);
1134 evl->evl_tag = htons(tag);
1138 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY);
1139 MODULE_VERSION(ether, 1);