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. 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>
42 #include <sys/eventhandler.h>
43 #include <sys/kernel.h>
45 #include <sys/malloc.h>
46 #include <sys/module.h>
48 #include <sys/random.h>
49 #include <sys/socket.h>
50 #include <sys/sockio.h>
51 #include <sys/sysctl.h>
55 #include <net/if_var.h>
56 #include <net/if_arp.h>
57 #include <net/netisr.h>
58 #include <net/route.h>
59 #include <net/if_llc.h>
60 #include <net/if_dl.h>
61 #include <net/if_types.h>
63 #include <net/ethernet.h>
64 #include <net/if_bridgevar.h>
65 #include <net/if_vlan_var.h>
66 #include <net/if_llatbl.h>
68 #include <net/rss_config.h>
71 #include <netpfil/pf/pf_mtag.h>
73 #if defined(INET) || defined(INET6)
74 #include <netinet/in.h>
75 #include <netinet/in_var.h>
76 #include <netinet/if_ether.h>
77 #include <netinet/ip_carp.h>
78 #include <netinet/ip_var.h>
81 #include <netinet6/nd6.h>
83 #include <security/mac/mac_framework.h>
86 CTASSERT(sizeof (struct ether_header) == ETHER_ADDR_LEN * 2 + 2);
87 CTASSERT(sizeof (struct ether_addr) == ETHER_ADDR_LEN);
90 VNET_DEFINE(struct pfil_head, link_pfil_hook); /* Packet filter hooks */
92 /* netgraph node hooks for ng_ether(4) */
93 void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp);
94 void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m);
95 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
96 void (*ng_ether_attach_p)(struct ifnet *ifp);
97 void (*ng_ether_detach_p)(struct ifnet *ifp);
99 void (*vlan_input_p)(struct ifnet *, struct mbuf *);
101 /* if_bridge(4) support */
102 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *);
103 int (*bridge_output_p)(struct ifnet *, struct mbuf *,
104 struct sockaddr *, struct rtentry *);
105 void (*bridge_dn_p)(struct mbuf *, struct ifnet *);
107 /* if_lagg(4) support */
108 struct mbuf *(*lagg_input_p)(struct ifnet *, struct mbuf *);
110 static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] =
111 { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
113 static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
116 static void ether_reassign(struct ifnet *, struct vnet *, char *);
118 static int ether_requestencap(struct ifnet *, struct if_encap_req *);
121 #define senderr(e) do { error = (e); goto bad;} while (0)
124 update_mbuf_csumflags(struct mbuf *src, struct mbuf *dst)
128 if (src->m_pkthdr.csum_flags & CSUM_IP)
129 csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
130 if (src->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
131 csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
132 if (src->m_pkthdr.csum_flags & CSUM_SCTP)
133 csum_flags |= CSUM_SCTP_VALID;
134 dst->m_pkthdr.csum_flags |= csum_flags;
135 if (csum_flags & CSUM_DATA_VALID)
136 dst->m_pkthdr.csum_data = 0xffff;
140 * Handle link-layer encapsulation requests.
143 ether_requestencap(struct ifnet *ifp, struct if_encap_req *req)
145 struct ether_header *eh;
148 const u_char *lladdr;
150 if (req->rtype != IFENCAP_LL)
153 if (req->bufsize < ETHER_HDR_LEN)
156 eh = (struct ether_header *)req->buf;
157 lladdr = req->lladdr;
160 switch (req->family) {
162 etype = htons(ETHERTYPE_IP);
165 etype = htons(ETHERTYPE_IPV6);
168 ah = (struct arphdr *)req->hdata;
169 ah->ar_hrd = htons(ARPHRD_ETHER);
171 switch(ntohs(ah->ar_op)) {
172 case ARPOP_REVREQUEST:
174 etype = htons(ETHERTYPE_REVARP);
179 etype = htons(ETHERTYPE_ARP);
183 if (req->flags & IFENCAP_FLAG_BROADCAST)
184 lladdr = ifp->if_broadcastaddr;
187 return (EAFNOSUPPORT);
190 memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
191 memcpy(eh->ether_dhost, lladdr, ETHER_ADDR_LEN);
192 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
193 req->bufsize = sizeof(struct ether_header);
200 ether_resolve_addr(struct ifnet *ifp, struct mbuf *m,
201 const struct sockaddr *dst, struct route *ro, u_char *phdr,
202 uint32_t *pflags, struct llentry **plle)
204 struct ether_header *eh;
205 uint32_t lleflags = 0;
207 #if defined(INET) || defined(INET6)
213 eh = (struct ether_header *)phdr;
215 switch (dst->sa_family) {
218 if ((m->m_flags & (M_BCAST | M_MCAST)) == 0)
219 error = arpresolve(ifp, 0, m, dst, phdr, &lleflags,
222 if (m->m_flags & M_BCAST)
223 memcpy(eh->ether_dhost, ifp->if_broadcastaddr,
226 const struct in_addr *a;
227 a = &(((const struct sockaddr_in *)dst)->sin_addr);
228 ETHER_MAP_IP_MULTICAST(a, eh->ether_dhost);
230 etype = htons(ETHERTYPE_IP);
231 memcpy(&eh->ether_type, &etype, sizeof(etype));
232 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
238 if ((m->m_flags & M_MCAST) == 0)
239 error = nd6_resolve(ifp, 0, m, dst, phdr, &lleflags,
242 const struct in6_addr *a6;
243 a6 = &(((const struct sockaddr_in6 *)dst)->sin6_addr);
244 ETHER_MAP_IPV6_MULTICAST(a6, eh->ether_dhost);
245 etype = htons(ETHERTYPE_IPV6);
246 memcpy(&eh->ether_type, &etype, sizeof(etype));
247 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
252 if_printf(ifp, "can't handle af%d\n", dst->sa_family);
255 return (EAFNOSUPPORT);
258 if (error == EHOSTDOWN) {
259 if (ro != NULL && (ro->ro_flags & RT_HAS_GW) != 0)
260 error = EHOSTUNREACH;
266 *pflags = RT_MAY_LOOP;
267 if (lleflags & LLE_IFADDR)
274 * Ethernet output routine.
275 * Encapsulate a packet of type family for the local net.
276 * Use trailer local net encapsulation if enough data in first
277 * packet leaves a multiple of 512 bytes of data in remainder.
280 ether_output(struct ifnet *ifp, struct mbuf *m,
281 const struct sockaddr *dst, struct route *ro)
284 char linkhdr[ETHER_HDR_LEN], *phdr;
285 struct ether_header *eh;
288 int hlen; /* link layer header length */
290 struct llentry *lle = NULL;
291 struct rtentry *rt0 = NULL;
297 /* XXX BPF uses ro_prepend */
298 if (ro->ro_prepend != NULL) {
299 phdr = ro->ro_prepend;
301 } else if (!(m->m_flags & (M_BCAST | M_MCAST))) {
302 if ((ro->ro_flags & RT_LLE_CACHE) != 0) {
305 (lle->la_flags & LLE_VALID) == 0) {
307 lle = NULL; /* redundant */
311 /* if we lookup, keep cache */
316 phdr = lle->r_linkdata;
317 hlen = lle->r_hdrlen;
318 pflags = lle->r_flags;
325 error = mac_ifnet_check_transmit(ifp, m);
331 if (ifp->if_flags & IFF_MONITOR)
333 if (!((ifp->if_flags & IFF_UP) &&
334 (ifp->if_drv_flags & IFF_DRV_RUNNING)))
338 /* No prepend data supplied. Try to calculate ourselves. */
340 hlen = ETHER_HDR_LEN;
341 error = ether_resolve_addr(ifp, m, dst, ro, phdr, &pflags,
342 addref ? &lle : NULL);
343 if (addref && lle != NULL)
346 return (error == EWOULDBLOCK ? 0 : error);
349 if ((pflags & RT_L2_ME) != 0) {
350 update_mbuf_csumflags(m, m);
351 return (if_simloop(ifp, m, dst->sa_family, 0));
353 loop_copy = pflags & RT_MAY_LOOP;
356 * Add local net header. If no space in first mbuf,
359 * Note that we do prepend regardless of RT_HAS_HEADER flag.
360 * This is done because BPF code shifts m_data pointer
361 * to the end of ethernet header prior to calling if_output().
363 M_PREPEND(m, hlen, M_NOWAIT);
366 if ((pflags & RT_HAS_HEADER) == 0) {
367 eh = mtod(m, struct ether_header *);
368 memcpy(eh, phdr, hlen);
372 * If a simplex interface, and the packet is being sent to our
373 * Ethernet address or a broadcast address, loopback a copy.
374 * XXX To make a simplex device behave exactly like a duplex
375 * device, we should copy in the case of sending to our own
376 * ethernet address (thus letting the original actually appear
377 * on the wire). However, we don't do that here for security
378 * reasons and compatibility with the original behavior.
380 if ((m->m_flags & M_BCAST) && loop_copy && (ifp->if_flags & IFF_SIMPLEX) &&
381 ((t = pf_find_mtag(m)) == NULL || !t->routed)) {
385 * Because if_simloop() modifies the packet, we need a
386 * writable copy through m_dup() instead of a readonly
387 * one as m_copy[m] would give us. The alternative would
388 * be to modify if_simloop() to handle the readonly mbuf,
389 * but performancewise it is mostly equivalent (trading
390 * extra data copying vs. extra locking).
392 * XXX This is a local workaround. A number of less
393 * often used kernel parts suffer from the same bug.
394 * See PR kern/105943 for a proposed general solution.
396 if ((n = m_dup(m, M_NOWAIT)) != NULL) {
397 update_mbuf_csumflags(m, n);
398 (void)if_simloop(ifp, n, dst->sa_family, hlen);
400 if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1);
404 * Bridges require special output handling.
406 if (ifp->if_bridge) {
407 BRIDGE_OUTPUT(ifp, m, error);
411 #if defined(INET) || defined(INET6)
413 (error = (*carp_output_p)(ifp, m, dst)))
417 /* Handle ng_ether(4) processing, if any */
418 if (ifp->if_l2com != NULL) {
419 KASSERT(ng_ether_output_p != NULL,
420 ("ng_ether_output_p is NULL"));
421 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
430 /* Continue with link-layer output */
431 return ether_output_frame(ifp, m);
435 * Ethernet link layer output routine to send a raw frame to the device.
437 * This assumes that the 14 byte Ethernet header is present and contiguous
438 * in the first mbuf (if BRIDGE'ing).
441 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
445 if (PFIL_HOOKED(&V_link_pfil_hook)) {
446 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_OUT, NULL);
456 * Queue message on interface, update output statistics if
457 * successful, and start output if interface not yet active.
459 return ((ifp->if_transmit)(ifp, m));
463 * Process a received Ethernet packet; the packet is in the
464 * mbuf chain m with the ethernet header at the front.
467 ether_input_internal(struct ifnet *ifp, struct mbuf *m)
469 struct ether_header *eh;
472 if ((ifp->if_flags & IFF_UP) == 0) {
477 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
478 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n");
483 if (m->m_len < ETHER_HDR_LEN) {
484 /* XXX maybe should pullup? */
485 if_printf(ifp, "discard frame w/o leading ethernet "
486 "header (len %u pkt len %u)\n",
487 m->m_len, m->m_pkthdr.len);
488 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
492 eh = mtod(m, struct ether_header *);
493 etype = ntohs(eh->ether_type);
494 random_harvest_queue(m, sizeof(*m), 2, RANDOM_NET_ETHER);
496 CURVNET_SET_QUIET(ifp->if_vnet);
498 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
499 if (ETHER_IS_BROADCAST(eh->ether_dhost))
500 m->m_flags |= M_BCAST;
502 m->m_flags |= M_MCAST;
503 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
508 * Tag the mbuf with an appropriate MAC label before any other
509 * consumers can get to it.
511 mac_ifnet_create_mbuf(ifp, m);
515 * Give bpf a chance at the packet.
517 ETHER_BPF_MTAP(ifp, m);
520 * If the CRC is still on the packet, trim it off. We do this once
521 * and once only in case we are re-entered. Nothing else on the
522 * Ethernet receive path expects to see the FCS.
524 if (m->m_flags & M_HASFCS) {
525 m_adj(m, -ETHER_CRC_LEN);
526 m->m_flags &= ~M_HASFCS;
529 if (!(ifp->if_capenable & IFCAP_HWSTATS))
530 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
532 /* Allow monitor mode to claim this frame, after stats are updated. */
533 if (ifp->if_flags & IFF_MONITOR) {
539 /* Handle input from a lagg(4) port */
540 if (ifp->if_type == IFT_IEEE8023ADLAG) {
541 KASSERT(lagg_input_p != NULL,
542 ("%s: if_lagg not loaded!", __func__));
543 m = (*lagg_input_p)(ifp, m);
545 ifp = m->m_pkthdr.rcvif;
553 * If the hardware did not process an 802.1Q tag, do this now,
554 * to allow 802.1P priority frames to be passed to the main input
556 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels.
558 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) {
559 struct ether_vlan_header *evl;
561 if (m->m_len < sizeof(*evl) &&
562 (m = m_pullup(m, sizeof(*evl))) == NULL) {
564 if_printf(ifp, "cannot pullup VLAN header\n");
566 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
571 evl = mtod(m, struct ether_vlan_header *);
572 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag);
573 m->m_flags |= M_VLANTAG;
575 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN,
576 ETHER_HDR_LEN - ETHER_TYPE_LEN);
577 m_adj(m, ETHER_VLAN_ENCAP_LEN);
578 eh = mtod(m, struct ether_header *);
581 M_SETFIB(m, ifp->if_fib);
583 /* Allow ng_ether(4) to claim this frame. */
584 if (ifp->if_l2com != NULL) {
585 KASSERT(ng_ether_input_p != NULL,
586 ("%s: ng_ether_input_p is NULL", __func__));
587 m->m_flags &= ~M_PROMISC;
588 (*ng_ether_input_p)(ifp, &m);
593 eh = mtod(m, struct ether_header *);
597 * Allow if_bridge(4) to claim this frame.
598 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it
599 * and the frame should be delivered locally.
601 if (ifp->if_bridge != NULL) {
602 m->m_flags &= ~M_PROMISC;
603 BRIDGE_INPUT(ifp, m);
608 eh = mtod(m, struct ether_header *);
611 #if defined(INET) || defined(INET6)
613 * Clear M_PROMISC on frame so that carp(4) will see it when the
614 * mbuf flows up to Layer 3.
615 * FreeBSD's implementation of carp(4) uses the inprotosw
616 * to dispatch IPPROTO_CARP. carp(4) also allocates its own
617 * Ethernet addresses of the form 00:00:5e:00:01:xx, which
618 * is outside the scope of the M_PROMISC test below.
619 * TODO: Maintain a hash table of ethernet addresses other than
620 * ether_dhost which may be active on this ifp.
622 if (ifp->if_carp && (*carp_forus_p)(ifp, eh->ether_dhost)) {
623 m->m_flags &= ~M_PROMISC;
628 * If the frame received was not for our MAC address, set the
629 * M_PROMISC flag on the mbuf chain. The frame may need to
630 * be seen by the rest of the Ethernet input path in case of
631 * re-entry (e.g. bridge, vlan, netgraph) but should not be
632 * seen by upper protocol layers.
634 if (!ETHER_IS_MULTICAST(eh->ether_dhost) &&
635 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0)
636 m->m_flags |= M_PROMISC;
644 * Ethernet input dispatch; by default, direct dispatch here regardless of
645 * global configuration. However, if RSS is enabled, hook up RSS affinity
646 * so that when deferred or hybrid dispatch is enabled, we can redistribute
649 * XXXRW: Would be nice if the ifnet passed up a flag indicating whether or
650 * not it had already done work distribution via multi-queue. Then we could
651 * direct dispatch in the event load balancing was already complete and
652 * handle the case of interfaces with different capabilities better.
654 * XXXRW: Sort of want an M_DISTRIBUTED flag to avoid multiple distributions
655 * at multiple layers?
657 * XXXRW: For now, enable all this only if RSS is compiled in, although it
658 * works fine without RSS. Need to characterise the performance overhead
659 * of the detour through the netisr code in the event the result is always
663 ether_nh_input(struct mbuf *m)
667 KASSERT(m->m_pkthdr.rcvif != NULL,
668 ("%s: NULL interface pointer", __func__));
669 ether_input_internal(m->m_pkthdr.rcvif, m);
672 static struct netisr_handler ether_nh = {
674 .nh_handler = ether_nh_input,
675 .nh_proto = NETISR_ETHER,
677 .nh_policy = NETISR_POLICY_CPU,
678 .nh_dispatch = NETISR_DISPATCH_DIRECT,
679 .nh_m2cpuid = rss_m2cpuid,
681 .nh_policy = NETISR_POLICY_SOURCE,
682 .nh_dispatch = NETISR_DISPATCH_DIRECT,
687 ether_init(__unused void *arg)
690 netisr_register(ðer_nh);
692 SYSINIT(ether, SI_SUB_INIT_IF, SI_ORDER_ANY, ether_init, NULL);
695 vnet_ether_init(__unused void *arg)
699 /* Initialize packet filter hooks. */
700 V_link_pfil_hook.ph_type = PFIL_TYPE_AF;
701 V_link_pfil_hook.ph_af = AF_LINK;
702 if ((i = pfil_head_register(&V_link_pfil_hook)) != 0)
703 printf("%s: WARNING: unable to register pfil link hook, "
704 "error %d\n", __func__, i);
706 netisr_register_vnet(ðer_nh);
709 VNET_SYSINIT(vnet_ether_init, SI_SUB_PROTO_IF, SI_ORDER_ANY,
710 vnet_ether_init, NULL);
714 vnet_ether_pfil_destroy(__unused void *arg)
718 if ((i = pfil_head_unregister(&V_link_pfil_hook)) != 0)
719 printf("%s: WARNING: unable to unregister pfil link hook, "
720 "error %d\n", __func__, i);
722 VNET_SYSUNINIT(vnet_ether_pfil_uninit, SI_SUB_PROTO_PFIL, SI_ORDER_ANY,
723 vnet_ether_pfil_destroy, NULL);
726 vnet_ether_destroy(__unused void *arg)
729 netisr_unregister_vnet(ðer_nh);
731 VNET_SYSUNINIT(vnet_ether_uninit, SI_SUB_PROTO_IF, SI_ORDER_ANY,
732 vnet_ether_destroy, NULL);
738 ether_input(struct ifnet *ifp, struct mbuf *m)
744 * The drivers are allowed to pass in a chain of packets linked with
745 * m_nextpkt. We split them up into separate packets here and pass
746 * them up. This allows the drivers to amortize the receive lock.
753 * We will rely on rcvif being set properly in the deferred context,
754 * so assert it is correct here.
756 KASSERT(m->m_pkthdr.rcvif == ifp, ("%s: ifnet mismatch m %p "
757 "rcvif %p ifp %p", __func__, m, m->m_pkthdr.rcvif, ifp));
758 CURVNET_SET_QUIET(ifp->if_vnet);
759 netisr_dispatch(NETISR_ETHER, m);
766 * Upper layer processing for a received Ethernet packet.
769 ether_demux(struct ifnet *ifp, struct mbuf *m)
771 struct ether_header *eh;
775 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__));
777 /* Do not grab PROMISC frames in case we are re-entered. */
778 if (PFIL_HOOKED(&V_link_pfil_hook) && !(m->m_flags & M_PROMISC)) {
779 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_IN, NULL);
781 if (i != 0 || m == NULL)
785 eh = mtod(m, struct ether_header *);
786 ether_type = ntohs(eh->ether_type);
789 * If this frame has a VLAN tag other than 0, call vlan_input()
790 * if its module is loaded. Otherwise, drop.
792 if ((m->m_flags & M_VLANTAG) &&
793 EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) {
794 if (ifp->if_vlantrunk == NULL) {
795 if_inc_counter(ifp, IFCOUNTER_NOPROTO, 1);
799 KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!",
801 /* Clear before possibly re-entering ether_input(). */
802 m->m_flags &= ~M_PROMISC;
803 (*vlan_input_p)(ifp, m);
808 * Pass promiscuously received frames to the upper layer if the user
809 * requested this by setting IFF_PPROMISC. Otherwise, drop them.
811 if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) {
817 * Reset layer specific mbuf flags to avoid confusing upper layers.
818 * Strip off Ethernet header.
820 m->m_flags &= ~M_VLANTAG;
822 m_adj(m, ETHER_HDR_LEN);
825 * Dispatch frame to upper layer.
827 switch (ether_type) {
834 if (ifp->if_flags & IFF_NOARP) {
835 /* Discard packet if ARP is disabled on interface */
850 netisr_dispatch(isr, m);
855 * Packet is to be discarded. If netgraph is present,
856 * hand the packet to it for last chance processing;
857 * otherwise dispose of it.
859 if (ifp->if_l2com != NULL) {
860 KASSERT(ng_ether_input_orphan_p != NULL,
861 ("ng_ether_input_orphan_p is NULL"));
863 * Put back the ethernet header so netgraph has a
864 * consistent view of inbound packets.
866 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
867 (*ng_ether_input_orphan_p)(ifp, m);
874 * Convert Ethernet address to printable (loggable) representation.
875 * This routine is for compatibility; it's better to just use
877 * printf("%6D", <pointer to address>, ":");
879 * since there's no static buffer involved.
882 ether_sprintf(const u_char *ap)
884 static char etherbuf[18];
885 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
890 * Perform common duties while attaching to interface list
893 ether_ifattach(struct ifnet *ifp, const u_int8_t *lla)
897 struct sockaddr_dl *sdl;
899 ifp->if_addrlen = ETHER_ADDR_LEN;
900 ifp->if_hdrlen = ETHER_HDR_LEN;
902 ifp->if_mtu = ETHERMTU;
903 ifp->if_output = ether_output;
904 ifp->if_input = ether_input;
905 ifp->if_resolvemulti = ether_resolvemulti;
906 ifp->if_requestencap = ether_requestencap;
908 ifp->if_reassign = ether_reassign;
910 if (ifp->if_baudrate == 0)
911 ifp->if_baudrate = IF_Mbps(10); /* just a default */
912 ifp->if_broadcastaddr = etherbroadcastaddr;
915 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
916 sdl = (struct sockaddr_dl *)ifa->ifa_addr;
917 sdl->sdl_type = IFT_ETHER;
918 sdl->sdl_alen = ifp->if_addrlen;
919 bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
921 if (ifp->if_hw_addr != NULL)
922 bcopy(lla, ifp->if_hw_addr, ifp->if_addrlen);
924 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
925 if (ng_ether_attach_p != NULL)
926 (*ng_ether_attach_p)(ifp);
928 /* Announce Ethernet MAC address if non-zero. */
929 for (i = 0; i < ifp->if_addrlen; i++)
932 if (i != ifp->if_addrlen)
933 if_printf(ifp, "Ethernet address: %6D\n", lla, ":");
935 uuid_ether_add(LLADDR(sdl));
937 /* Add necessary bits are setup; announce it now. */
938 EVENTHANDLER_INVOKE(ether_ifattach_event, ifp);
939 if (IS_DEFAULT_VNET(curvnet))
940 devctl_notify("ETHERNET", ifp->if_xname, "IFATTACH", NULL);
944 * Perform common duties while detaching an Ethernet interface
947 ether_ifdetach(struct ifnet *ifp)
949 struct sockaddr_dl *sdl;
951 sdl = (struct sockaddr_dl *)(ifp->if_addr->ifa_addr);
952 uuid_ether_del(LLADDR(sdl));
954 if (ifp->if_l2com != NULL) {
955 KASSERT(ng_ether_detach_p != NULL,
956 ("ng_ether_detach_p is NULL"));
957 (*ng_ether_detach_p)(ifp);
966 ether_reassign(struct ifnet *ifp, struct vnet *new_vnet, char *unused __unused)
969 if (ifp->if_l2com != NULL) {
970 KASSERT(ng_ether_detach_p != NULL,
971 ("ng_ether_detach_p is NULL"));
972 (*ng_ether_detach_p)(ifp);
975 if (ng_ether_attach_p != NULL) {
976 CURVNET_SET_QUIET(new_vnet);
977 (*ng_ether_attach_p)(ifp);
983 SYSCTL_DECL(_net_link);
984 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
988 * This is for reference. We have a table-driven version
989 * of the little-endian crc32 generator, which is faster
990 * than the double-loop.
993 ether_crc32_le(const uint8_t *buf, size_t len)
1000 crc = 0xffffffff; /* initial value */
1002 for (i = 0; i < len; i++) {
1003 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
1004 carry = (crc ^ data) & 1;
1007 crc = (crc ^ ETHER_CRC_POLY_LE);
1015 ether_crc32_le(const uint8_t *buf, size_t len)
1017 static const uint32_t crctab[] = {
1018 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
1019 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
1020 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
1021 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
1026 crc = 0xffffffff; /* initial value */
1028 for (i = 0; i < len; i++) {
1030 crc = (crc >> 4) ^ crctab[crc & 0xf];
1031 crc = (crc >> 4) ^ crctab[crc & 0xf];
1039 ether_crc32_be(const uint8_t *buf, size_t len)
1042 uint32_t crc, carry;
1046 crc = 0xffffffff; /* initial value */
1048 for (i = 0; i < len; i++) {
1049 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
1050 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
1053 crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1061 ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
1063 struct ifaddr *ifa = (struct ifaddr *) data;
1064 struct ifreq *ifr = (struct ifreq *) data;
1069 ifp->if_flags |= IFF_UP;
1071 switch (ifa->ifa_addr->sa_family) {
1074 ifp->if_init(ifp->if_softc); /* before arpwhohas */
1075 arp_ifinit(ifp, ifa);
1079 ifp->if_init(ifp->if_softc);
1086 struct sockaddr *sa;
1088 sa = (struct sockaddr *) & ifr->ifr_data;
1089 bcopy(IF_LLADDR(ifp),
1090 (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
1096 * Set the interface MTU.
1098 if (ifr->ifr_mtu > ETHERMTU) {
1101 ifp->if_mtu = ifr->ifr_mtu;
1105 error = EINVAL; /* XXX netbsd has ENOTTY??? */
1112 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
1113 struct sockaddr *sa)
1115 struct sockaddr_dl *sdl;
1117 struct sockaddr_in *sin;
1120 struct sockaddr_in6 *sin6;
1124 switch(sa->sa_family) {
1127 * No mapping needed. Just check that it's a valid MC address.
1129 sdl = (struct sockaddr_dl *)sa;
1130 e_addr = LLADDR(sdl);
1131 if (!ETHER_IS_MULTICAST(e_addr))
1132 return EADDRNOTAVAIL;
1138 sin = (struct sockaddr_in *)sa;
1139 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1140 return EADDRNOTAVAIL;
1141 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1142 sdl->sdl_alen = ETHER_ADDR_LEN;
1143 e_addr = LLADDR(sdl);
1144 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1145 *llsa = (struct sockaddr *)sdl;
1150 sin6 = (struct sockaddr_in6 *)sa;
1151 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1153 * An IP6 address of 0 means listen to all
1154 * of the Ethernet multicast address used for IP6.
1155 * (This is used for multicast routers.)
1157 ifp->if_flags |= IFF_ALLMULTI;
1161 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1162 return EADDRNOTAVAIL;
1163 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1164 sdl->sdl_alen = ETHER_ADDR_LEN;
1165 e_addr = LLADDR(sdl);
1166 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1167 *llsa = (struct sockaddr *)sdl;
1173 * Well, the text isn't quite right, but it's the name
1176 return EAFNOSUPPORT;
1180 static moduledata_t ether_mod = {
1185 ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen)
1187 struct ether_vlan_header vlan;
1190 KASSERT((m->m_flags & M_VLANTAG) != 0,
1191 ("%s: vlan information not present", __func__));
1192 KASSERT(m->m_len >= sizeof(struct ether_header),
1193 ("%s: mbuf not large enough for header", __func__));
1194 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header));
1195 vlan.evl_proto = vlan.evl_encap_proto;
1196 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN);
1197 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag);
1198 m->m_len -= sizeof(struct ether_header);
1199 m->m_data += sizeof(struct ether_header);
1201 * If a data link has been supplied by the caller, then we will need to
1202 * re-create a stack allocated mbuf chain with the following structure:
1204 * (1) mbuf #1 will contain the supplied data link
1205 * (2) mbuf #2 will contain the vlan header
1206 * (3) mbuf #3 will contain the original mbuf's packet data
1208 * Otherwise, submit the packet and vlan header via bpf_mtap2().
1212 mv.m_data = (caddr_t)&vlan;
1213 mv.m_len = sizeof(vlan);
1219 bpf_mtap2(bp, &vlan, sizeof(vlan), m);
1220 m->m_len += sizeof(struct ether_header);
1221 m->m_data -= sizeof(struct ether_header);
1225 ether_vlanencap(struct mbuf *m, uint16_t tag)
1227 struct ether_vlan_header *evl;
1229 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_NOWAIT);
1232 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */
1234 if (m->m_len < sizeof(*evl)) {
1235 m = m_pullup(m, sizeof(*evl));
1241 * Transform the Ethernet header into an Ethernet header
1242 * with 802.1Q encapsulation.
1244 evl = mtod(m, struct ether_vlan_header *);
1245 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN,
1246 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN);
1247 evl->evl_encap_proto = htons(ETHERTYPE_VLAN);
1248 evl->evl_tag = htons(tag);
1252 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY);
1253 MODULE_VERSION(ether, 1);