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 *);
116 static int ether_requestencap(struct ifnet *, struct if_encap_req *);
119 #define senderr(e) do { error = (e); goto bad;} while (0)
122 update_mbuf_csumflags(struct mbuf *src, struct mbuf *dst)
126 if (src->m_pkthdr.csum_flags & CSUM_IP)
127 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
128 if (src->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
129 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
130 if (src->m_pkthdr.csum_flags & CSUM_SCTP)
131 csum_flags |= CSUM_SCTP_VALID;
132 dst->m_pkthdr.csum_flags |= csum_flags;
133 if (csum_flags & CSUM_DATA_VALID)
134 dst->m_pkthdr.csum_data = 0xffff;
138 * Handle link-layer encapsulation requests.
141 ether_requestencap(struct ifnet *ifp, struct if_encap_req *req)
143 struct ether_header *eh;
146 const u_char *lladdr;
148 if (req->rtype != IFENCAP_LL)
151 if (req->bufsize < ETHER_HDR_LEN)
154 eh = (struct ether_header *)req->buf;
155 lladdr = req->lladdr;
158 switch (req->family) {
160 etype = htons(ETHERTYPE_IP);
163 etype = htons(ETHERTYPE_IPV6);
166 ah = (struct arphdr *)req->hdata;
167 ah->ar_hrd = htons(ARPHRD_ETHER);
169 switch(ntohs(ah->ar_op)) {
170 case ARPOP_REVREQUEST:
172 etype = htons(ETHERTYPE_REVARP);
177 etype = htons(ETHERTYPE_ARP);
181 if (req->flags & IFENCAP_FLAG_BROADCAST)
182 lladdr = ifp->if_broadcastaddr;
185 return (EAFNOSUPPORT);
188 memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
189 memcpy(eh->ether_dhost, lladdr, ETHER_ADDR_LEN);
190 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
191 req->bufsize = sizeof(struct ether_header);
198 ether_resolve_addr(struct ifnet *ifp, struct mbuf *m,
199 const struct sockaddr *dst, struct route *ro, u_char *phdr,
200 uint32_t *pflags, struct llentry **plle)
202 struct ether_header *eh;
203 uint32_t lleflags = 0;
205 #if defined(INET) || defined(INET6)
211 eh = (struct ether_header *)phdr;
213 switch (dst->sa_family) {
216 if ((m->m_flags & (M_BCAST | M_MCAST)) == 0)
217 error = arpresolve(ifp, 0, m, dst, phdr, &lleflags,
220 if (m->m_flags & M_BCAST)
221 memcpy(eh->ether_dhost, ifp->if_broadcastaddr,
224 const struct in_addr *a;
225 a = &(((const struct sockaddr_in *)dst)->sin_addr);
226 ETHER_MAP_IP_MULTICAST(a, eh->ether_dhost);
228 etype = htons(ETHERTYPE_IP);
229 memcpy(&eh->ether_type, &etype, sizeof(etype));
230 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
236 if ((m->m_flags & M_MCAST) == 0)
237 error = nd6_resolve(ifp, 0, m, dst, phdr, &lleflags,
240 const struct in6_addr *a6;
241 a6 = &(((const struct sockaddr_in6 *)dst)->sin6_addr);
242 ETHER_MAP_IPV6_MULTICAST(a6, eh->ether_dhost);
243 etype = htons(ETHERTYPE_IPV6);
244 memcpy(&eh->ether_type, &etype, sizeof(etype));
245 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
250 if_printf(ifp, "can't handle af%d\n", dst->sa_family);
253 return (EAFNOSUPPORT);
256 if (error == EHOSTDOWN) {
257 if (ro != NULL && (ro->ro_flags & RT_HAS_GW) != 0)
258 error = EHOSTUNREACH;
264 *pflags = RT_MAY_LOOP;
265 if (lleflags & LLE_IFADDR)
272 * Ethernet output routine.
273 * Encapsulate a packet of type family for the local net.
274 * Use trailer local net encapsulation if enough data in first
275 * packet leaves a multiple of 512 bytes of data in remainder.
278 ether_output(struct ifnet *ifp, struct mbuf *m,
279 const struct sockaddr *dst, struct route *ro)
282 char linkhdr[ETHER_HDR_LEN], *phdr;
283 struct ether_header *eh;
286 int hlen; /* link layer header length */
288 struct llentry *lle = NULL;
289 struct rtentry *rt0 = NULL;
295 /* XXX BPF uses ro_prepend */
296 if (ro->ro_prepend != NULL) {
297 phdr = ro->ro_prepend;
299 } else if (!(m->m_flags & (M_BCAST | M_MCAST))) {
300 if ((ro->ro_flags & RT_LLE_CACHE) != 0) {
303 (lle->la_flags & LLE_VALID) == 0) {
305 lle = NULL; /* redundant */
309 /* if we lookup, keep cache */
314 phdr = lle->r_linkdata;
315 hlen = lle->r_hdrlen;
316 pflags = lle->r_flags;
323 error = mac_ifnet_check_transmit(ifp, m);
329 if (ifp->if_flags & IFF_MONITOR)
331 if (!((ifp->if_flags & IFF_UP) &&
332 (ifp->if_drv_flags & IFF_DRV_RUNNING)))
336 /* No prepend data supplied. Try to calculate ourselves. */
338 hlen = ETHER_HDR_LEN;
339 error = ether_resolve_addr(ifp, m, dst, ro, phdr, &pflags,
340 addref ? &lle : NULL);
341 if (addref && lle != NULL)
344 return (error == EWOULDBLOCK ? 0 : error);
347 if ((pflags & RT_L2_ME) != 0) {
348 update_mbuf_csumflags(m, m);
349 return (if_simloop(ifp, m, dst->sa_family, 0));
351 loop_copy = pflags & RT_MAY_LOOP;
354 * Add local net header. If no space in first mbuf,
357 * Note that we do prepend regardless of RT_HAS_HEADER flag.
358 * This is done because BPF code shifts m_data pointer
359 * to the end of ethernet header prior to calling if_output().
361 M_PREPEND(m, hlen, M_NOWAIT);
364 if ((pflags & RT_HAS_HEADER) == 0) {
365 eh = mtod(m, struct ether_header *);
366 memcpy(eh, phdr, hlen);
370 * If a simplex interface, and the packet is being sent to our
371 * Ethernet address or a broadcast address, loopback a copy.
372 * XXX To make a simplex device behave exactly like a duplex
373 * device, we should copy in the case of sending to our own
374 * ethernet address (thus letting the original actually appear
375 * on the wire). However, we don't do that here for security
376 * reasons and compatibility with the original behavior.
378 if ((m->m_flags & M_BCAST) && loop_copy && (ifp->if_flags & IFF_SIMPLEX) &&
379 ((t = pf_find_mtag(m)) == NULL || !t->routed)) {
383 * Because if_simloop() modifies the packet, we need a
384 * writable copy through m_dup() instead of a readonly
385 * one as m_copy[m] would give us. The alternative would
386 * be to modify if_simloop() to handle the readonly mbuf,
387 * but performancewise it is mostly equivalent (trading
388 * extra data copying vs. extra locking).
390 * XXX This is a local workaround. A number of less
391 * often used kernel parts suffer from the same bug.
392 * See PR kern/105943 for a proposed general solution.
394 if ((n = m_dup(m, M_NOWAIT)) != NULL) {
395 update_mbuf_csumflags(m, n);
396 (void)if_simloop(ifp, n, dst->sa_family, hlen);
398 if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1);
402 * Bridges require special output handling.
404 if (ifp->if_bridge) {
405 BRIDGE_OUTPUT(ifp, m, error);
409 #if defined(INET) || defined(INET6)
411 (error = (*carp_output_p)(ifp, m, dst)))
415 /* Handle ng_ether(4) processing, if any */
416 if (ifp->if_l2com != NULL) {
417 KASSERT(ng_ether_output_p != NULL,
418 ("ng_ether_output_p is NULL"));
419 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
428 /* Continue with link-layer output */
429 return ether_output_frame(ifp, m);
433 * Ethernet link layer output routine to send a raw frame to the device.
435 * This assumes that the 14 byte Ethernet header is present and contiguous
436 * in the first mbuf (if BRIDGE'ing).
439 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
443 if (PFIL_HOOKED(&V_link_pfil_hook)) {
444 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_OUT, NULL);
454 * Queue message on interface, update output statistics if
455 * successful, and start output if interface not yet active.
457 return ((ifp->if_transmit)(ifp, m));
461 * Process a received Ethernet packet; the packet is in the
462 * mbuf chain m with the ethernet header at the front.
465 ether_input_internal(struct ifnet *ifp, struct mbuf *m)
467 struct ether_header *eh;
470 if ((ifp->if_flags & IFF_UP) == 0) {
475 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
476 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n");
481 if (m->m_len < ETHER_HDR_LEN) {
482 /* XXX maybe should pullup? */
483 if_printf(ifp, "discard frame w/o leading ethernet "
484 "header (len %u pkt len %u)\n",
485 m->m_len, m->m_pkthdr.len);
486 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
490 eh = mtod(m, struct ether_header *);
491 etype = ntohs(eh->ether_type);
492 random_harvest_queue(m, sizeof(*m), 2, RANDOM_NET_ETHER);
494 CURVNET_SET_QUIET(ifp->if_vnet);
496 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
497 if (ETHER_IS_BROADCAST(eh->ether_dhost))
498 m->m_flags |= M_BCAST;
500 m->m_flags |= M_MCAST;
501 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
506 * Tag the mbuf with an appropriate MAC label before any other
507 * consumers can get to it.
509 mac_ifnet_create_mbuf(ifp, m);
513 * Give bpf a chance at the packet.
515 ETHER_BPF_MTAP(ifp, m);
518 * If the CRC is still on the packet, trim it off. We do this once
519 * and once only in case we are re-entered. Nothing else on the
520 * Ethernet receive path expects to see the FCS.
522 if (m->m_flags & M_HASFCS) {
523 m_adj(m, -ETHER_CRC_LEN);
524 m->m_flags &= ~M_HASFCS;
527 if (!(ifp->if_capenable & IFCAP_HWSTATS))
528 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
530 /* Allow monitor mode to claim this frame, after stats are updated. */
531 if (ifp->if_flags & IFF_MONITOR) {
537 /* Handle input from a lagg(4) port */
538 if (ifp->if_type == IFT_IEEE8023ADLAG) {
539 KASSERT(lagg_input_p != NULL,
540 ("%s: if_lagg not loaded!", __func__));
541 m = (*lagg_input_p)(ifp, m);
543 ifp = m->m_pkthdr.rcvif;
551 * If the hardware did not process an 802.1Q tag, do this now,
552 * to allow 802.1P priority frames to be passed to the main input
554 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels.
556 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) {
557 struct ether_vlan_header *evl;
559 if (m->m_len < sizeof(*evl) &&
560 (m = m_pullup(m, sizeof(*evl))) == NULL) {
562 if_printf(ifp, "cannot pullup VLAN header\n");
564 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
569 evl = mtod(m, struct ether_vlan_header *);
570 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag);
571 m->m_flags |= M_VLANTAG;
573 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN,
574 ETHER_HDR_LEN - ETHER_TYPE_LEN);
575 m_adj(m, ETHER_VLAN_ENCAP_LEN);
576 eh = mtod(m, struct ether_header *);
579 M_SETFIB(m, ifp->if_fib);
581 /* Allow ng_ether(4) to claim this frame. */
582 if (ifp->if_l2com != NULL) {
583 KASSERT(ng_ether_input_p != NULL,
584 ("%s: ng_ether_input_p is NULL", __func__));
585 m->m_flags &= ~M_PROMISC;
586 (*ng_ether_input_p)(ifp, &m);
591 eh = mtod(m, struct ether_header *);
595 * Allow if_bridge(4) to claim this frame.
596 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it
597 * and the frame should be delivered locally.
599 if (ifp->if_bridge != NULL) {
600 m->m_flags &= ~M_PROMISC;
601 BRIDGE_INPUT(ifp, m);
606 eh = mtod(m, struct ether_header *);
609 #if defined(INET) || defined(INET6)
611 * Clear M_PROMISC on frame so that carp(4) will see it when the
612 * mbuf flows up to Layer 3.
613 * FreeBSD's implementation of carp(4) uses the inprotosw
614 * to dispatch IPPROTO_CARP. carp(4) also allocates its own
615 * Ethernet addresses of the form 00:00:5e:00:01:xx, which
616 * is outside the scope of the M_PROMISC test below.
617 * TODO: Maintain a hash table of ethernet addresses other than
618 * ether_dhost which may be active on this ifp.
620 if (ifp->if_carp && (*carp_forus_p)(ifp, eh->ether_dhost)) {
621 m->m_flags &= ~M_PROMISC;
626 * If the frame received was not for our MAC address, set the
627 * M_PROMISC flag on the mbuf chain. The frame may need to
628 * be seen by the rest of the Ethernet input path in case of
629 * re-entry (e.g. bridge, vlan, netgraph) but should not be
630 * seen by upper protocol layers.
632 if (!ETHER_IS_MULTICAST(eh->ether_dhost) &&
633 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0)
634 m->m_flags |= M_PROMISC;
642 * Ethernet input dispatch; by default, direct dispatch here regardless of
643 * global configuration. However, if RSS is enabled, hook up RSS affinity
644 * so that when deferred or hybrid dispatch is enabled, we can redistribute
647 * XXXRW: Would be nice if the ifnet passed up a flag indicating whether or
648 * not it had already done work distribution via multi-queue. Then we could
649 * direct dispatch in the event load balancing was already complete and
650 * handle the case of interfaces with different capabilities better.
652 * XXXRW: Sort of want an M_DISTRIBUTED flag to avoid multiple distributions
653 * at multiple layers?
655 * XXXRW: For now, enable all this only if RSS is compiled in, although it
656 * works fine without RSS. Need to characterise the performance overhead
657 * of the detour through the netisr code in the event the result is always
661 ether_nh_input(struct mbuf *m)
665 KASSERT(m->m_pkthdr.rcvif != NULL,
666 ("%s: NULL interface pointer", __func__));
667 ether_input_internal(m->m_pkthdr.rcvif, m);
670 static struct netisr_handler ether_nh = {
672 .nh_handler = ether_nh_input,
673 .nh_proto = NETISR_ETHER,
675 .nh_policy = NETISR_POLICY_CPU,
676 .nh_dispatch = NETISR_DISPATCH_DIRECT,
677 .nh_m2cpuid = rss_m2cpuid,
679 .nh_policy = NETISR_POLICY_SOURCE,
680 .nh_dispatch = NETISR_DISPATCH_DIRECT,
685 ether_init(__unused void *arg)
688 netisr_register(ðer_nh);
690 SYSINIT(ether, SI_SUB_INIT_IF, SI_ORDER_ANY, ether_init, NULL);
693 vnet_ether_init(__unused void *arg)
697 /* Initialize packet filter hooks. */
698 V_link_pfil_hook.ph_type = PFIL_TYPE_AF;
699 V_link_pfil_hook.ph_af = AF_LINK;
700 if ((i = pfil_head_register(&V_link_pfil_hook)) != 0)
701 printf("%s: WARNING: unable to register pfil link hook, "
702 "error %d\n", __func__, i);
704 netisr_register_vnet(ðer_nh);
707 VNET_SYSINIT(vnet_ether_init, SI_SUB_PROTO_IF, SI_ORDER_ANY,
708 vnet_ether_init, NULL);
712 vnet_ether_pfil_destroy(__unused void *arg)
716 if ((i = pfil_head_unregister(&V_link_pfil_hook)) != 0)
717 printf("%s: WARNING: unable to unregister pfil link hook, "
718 "error %d\n", __func__, i);
720 VNET_SYSUNINIT(vnet_ether_pfil_uninit, SI_SUB_PROTO_PFIL, SI_ORDER_ANY,
721 vnet_ether_pfil_destroy, NULL);
724 vnet_ether_destroy(__unused void *arg)
727 netisr_unregister_vnet(ðer_nh);
729 VNET_SYSUNINIT(vnet_ether_uninit, SI_SUB_PROTO_IF, SI_ORDER_ANY,
730 vnet_ether_destroy, NULL);
736 ether_input(struct ifnet *ifp, struct mbuf *m)
742 * The drivers are allowed to pass in a chain of packets linked with
743 * m_nextpkt. We split them up into separate packets here and pass
744 * them up. This allows the drivers to amortize the receive lock.
751 * We will rely on rcvif being set properly in the deferred context,
752 * so assert it is correct here.
754 KASSERT(m->m_pkthdr.rcvif == ifp, ("%s: ifnet mismatch m %p "
755 "rcvif %p ifp %p", __func__, m, m->m_pkthdr.rcvif, ifp));
756 CURVNET_SET_QUIET(ifp->if_vnet);
757 netisr_dispatch(NETISR_ETHER, m);
764 * Upper layer processing for a received Ethernet packet.
767 ether_demux(struct ifnet *ifp, struct mbuf *m)
769 struct ether_header *eh;
773 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__));
775 /* Do not grab PROMISC frames in case we are re-entered. */
776 if (PFIL_HOOKED(&V_link_pfil_hook) && !(m->m_flags & M_PROMISC)) {
777 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_IN, NULL);
779 if (i != 0 || m == NULL)
783 eh = mtod(m, struct ether_header *);
784 ether_type = ntohs(eh->ether_type);
787 * If this frame has a VLAN tag other than 0, call vlan_input()
788 * if its module is loaded. Otherwise, drop.
790 if ((m->m_flags & M_VLANTAG) &&
791 EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) {
792 if (ifp->if_vlantrunk == NULL) {
793 if_inc_counter(ifp, IFCOUNTER_NOPROTO, 1);
797 KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!",
799 /* Clear before possibly re-entering ether_input(). */
800 m->m_flags &= ~M_PROMISC;
801 (*vlan_input_p)(ifp, m);
806 * Pass promiscuously received frames to the upper layer if the user
807 * requested this by setting IFF_PPROMISC. Otherwise, drop them.
809 if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) {
815 * Reset layer specific mbuf flags to avoid confusing upper layers.
816 * Strip off Ethernet header.
818 m->m_flags &= ~M_VLANTAG;
820 m_adj(m, ETHER_HDR_LEN);
823 * Dispatch frame to upper layer.
825 switch (ether_type) {
832 if (ifp->if_flags & IFF_NOARP) {
833 /* Discard packet if ARP is disabled on interface */
848 netisr_dispatch(isr, m);
853 * Packet is to be discarded. If netgraph is present,
854 * hand the packet to it for last chance processing;
855 * otherwise dispose of it.
857 if (ifp->if_l2com != NULL) {
858 KASSERT(ng_ether_input_orphan_p != NULL,
859 ("ng_ether_input_orphan_p is NULL"));
861 * Put back the ethernet header so netgraph has a
862 * consistent view of inbound packets.
864 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
865 (*ng_ether_input_orphan_p)(ifp, m);
872 * Convert Ethernet address to printable (loggable) representation.
873 * This routine is for compatibility; it's better to just use
875 * printf("%6D", <pointer to address>, ":");
877 * since there's no static buffer involved.
880 ether_sprintf(const u_char *ap)
882 static char etherbuf[18];
883 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
888 * Perform common duties while attaching to interface list
891 ether_ifattach(struct ifnet *ifp, const u_int8_t *lla)
895 struct sockaddr_dl *sdl;
897 ifp->if_addrlen = ETHER_ADDR_LEN;
898 ifp->if_hdrlen = ETHER_HDR_LEN;
900 ifp->if_mtu = ETHERMTU;
901 ifp->if_output = ether_output;
902 ifp->if_input = ether_input;
903 ifp->if_resolvemulti = ether_resolvemulti;
904 ifp->if_requestencap = ether_requestencap;
906 ifp->if_reassign = ether_reassign;
908 if (ifp->if_baudrate == 0)
909 ifp->if_baudrate = IF_Mbps(10); /* just a default */
910 ifp->if_broadcastaddr = etherbroadcastaddr;
913 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
914 sdl = (struct sockaddr_dl *)ifa->ifa_addr;
915 sdl->sdl_type = IFT_ETHER;
916 sdl->sdl_alen = ifp->if_addrlen;
917 bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
919 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
920 if (ng_ether_attach_p != NULL)
921 (*ng_ether_attach_p)(ifp);
923 /* Announce Ethernet MAC address if non-zero. */
924 for (i = 0; i < ifp->if_addrlen; i++)
927 if (i != ifp->if_addrlen)
928 if_printf(ifp, "Ethernet address: %6D\n", lla, ":");
930 uuid_ether_add(LLADDR(sdl));
934 * Perform common duties while detaching an Ethernet interface
937 ether_ifdetach(struct ifnet *ifp)
939 struct sockaddr_dl *sdl;
941 sdl = (struct sockaddr_dl *)(ifp->if_addr->ifa_addr);
942 uuid_ether_del(LLADDR(sdl));
944 if (ifp->if_l2com != NULL) {
945 KASSERT(ng_ether_detach_p != NULL,
946 ("ng_ether_detach_p is NULL"));
947 (*ng_ether_detach_p)(ifp);
956 ether_reassign(struct ifnet *ifp, struct vnet *new_vnet, char *unused __unused)
959 if (ifp->if_l2com != NULL) {
960 KASSERT(ng_ether_detach_p != NULL,
961 ("ng_ether_detach_p is NULL"));
962 (*ng_ether_detach_p)(ifp);
965 if (ng_ether_attach_p != NULL) {
966 CURVNET_SET_QUIET(new_vnet);
967 (*ng_ether_attach_p)(ifp);
973 SYSCTL_DECL(_net_link);
974 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
978 * This is for reference. We have a table-driven version
979 * of the little-endian crc32 generator, which is faster
980 * than the double-loop.
983 ether_crc32_le(const uint8_t *buf, size_t len)
990 crc = 0xffffffff; /* initial value */
992 for (i = 0; i < len; i++) {
993 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
994 carry = (crc ^ data) & 1;
997 crc = (crc ^ ETHER_CRC_POLY_LE);
1005 ether_crc32_le(const uint8_t *buf, size_t len)
1007 static const uint32_t crctab[] = {
1008 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1009 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1010 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1011 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1016 crc = 0xffffffff; /* initial value */
1018 for (i = 0; i < len; i++) {
1020 crc = (crc >> 4) ^ crctab[crc & 0xf];
1021 crc = (crc >> 4) ^ crctab[crc & 0xf];
1029 ether_crc32_be(const uint8_t *buf, size_t len)
1032 uint32_t crc, carry;
1036 crc = 0xffffffff; /* initial value */
1038 for (i = 0; i < len; i++) {
1039 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
1040 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
1043 crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1051 ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
1053 struct ifaddr *ifa = (struct ifaddr *) data;
1054 struct ifreq *ifr = (struct ifreq *) data;
1059 ifp->if_flags |= IFF_UP;
1061 switch (ifa->ifa_addr->sa_family) {
1064 ifp->if_init(ifp->if_softc); /* before arpwhohas */
1065 arp_ifinit(ifp, ifa);
1069 ifp->if_init(ifp->if_softc);
1076 struct sockaddr *sa;
1078 sa = (struct sockaddr *) & ifr->ifr_data;
1079 bcopy(IF_LLADDR(ifp),
1080 (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
1086 * Set the interface MTU.
1088 if (ifr->ifr_mtu > ETHERMTU) {
1091 ifp->if_mtu = ifr->ifr_mtu;
1095 error = EINVAL; /* XXX netbsd has ENOTTY??? */
1102 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
1103 struct sockaddr *sa)
1105 struct sockaddr_dl *sdl;
1107 struct sockaddr_in *sin;
1110 struct sockaddr_in6 *sin6;
1114 switch(sa->sa_family) {
1117 * No mapping needed. Just check that it's a valid MC address.
1119 sdl = (struct sockaddr_dl *)sa;
1120 e_addr = LLADDR(sdl);
1121 if (!ETHER_IS_MULTICAST(e_addr))
1122 return EADDRNOTAVAIL;
1128 sin = (struct sockaddr_in *)sa;
1129 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1130 return EADDRNOTAVAIL;
1131 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1132 sdl->sdl_alen = ETHER_ADDR_LEN;
1133 e_addr = LLADDR(sdl);
1134 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1135 *llsa = (struct sockaddr *)sdl;
1140 sin6 = (struct sockaddr_in6 *)sa;
1141 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1143 * An IP6 address of 0 means listen to all
1144 * of the Ethernet multicast address used for IP6.
1145 * (This is used for multicast routers.)
1147 ifp->if_flags |= IFF_ALLMULTI;
1151 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1152 return EADDRNOTAVAIL;
1153 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1154 sdl->sdl_alen = ETHER_ADDR_LEN;
1155 e_addr = LLADDR(sdl);
1156 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1157 *llsa = (struct sockaddr *)sdl;
1163 * Well, the text isn't quite right, but it's the name
1166 return EAFNOSUPPORT;
1170 static moduledata_t ether_mod = {
1175 ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen)
1177 struct ether_vlan_header vlan;
1180 KASSERT((m->m_flags & M_VLANTAG) != 0,
1181 ("%s: vlan information not present", __func__));
1182 KASSERT(m->m_len >= sizeof(struct ether_header),
1183 ("%s: mbuf not large enough for header", __func__));
1184 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header));
1185 vlan.evl_proto = vlan.evl_encap_proto;
1186 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN);
1187 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag);
1188 m->m_len -= sizeof(struct ether_header);
1189 m->m_data += sizeof(struct ether_header);
1191 * If a data link has been supplied by the caller, then we will need to
1192 * re-create a stack allocated mbuf chain with the following structure:
1194 * (1) mbuf #1 will contain the supplied data link
1195 * (2) mbuf #2 will contain the vlan header
1196 * (3) mbuf #3 will contain the original mbuf's packet data
1198 * Otherwise, submit the packet and vlan header via bpf_mtap2().
1202 mv.m_data = (caddr_t)&vlan;
1203 mv.m_len = sizeof(vlan);
1209 bpf_mtap2(bp, &vlan, sizeof(vlan), m);
1210 m->m_len += sizeof(struct ether_header);
1211 m->m_data -= sizeof(struct ether_header);
1215 ether_vlanencap(struct mbuf *m, uint16_t tag)
1217 struct ether_vlan_header *evl;
1219 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_NOWAIT);
1222 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */
1224 if (m->m_len < sizeof(*evl)) {
1225 m = m_pullup(m, sizeof(*evl));
1231 * Transform the Ethernet header into an Ethernet header
1232 * with 802.1Q encapsulation.
1234 evl = mtod(m, struct ether_vlan_header *);
1235 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN,
1236 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN);
1237 evl->evl_encap_proto = htons(ETHERTYPE_VLAN);
1238 evl->evl_tag = htons(tag);
1242 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY);
1243 MODULE_VERSION(ether, 1);