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 *);
118 #define ETHER_IS_BROADCAST(addr) \
119 (bcmp(etherbroadcastaddr, (addr), ETHER_ADDR_LEN) == 0)
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,
204 struct ether_header *eh;
205 uint32_t lleflags = 0;
207 #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);
219 if (m->m_flags & M_BCAST)
220 memcpy(eh->ether_dhost, ifp->if_broadcastaddr,
223 const struct in_addr *a;
224 a = &(((const struct sockaddr_in *)dst)->sin_addr);
225 ETHER_MAP_IP_MULTICAST(a, eh->ether_dhost);
227 etype = htons(ETHERTYPE_IP);
228 memcpy(&eh->ether_type, &etype, sizeof(etype));
229 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
235 if ((m->m_flags & M_MCAST) == 0)
236 error = nd6_resolve(ifp, 0, m, dst, phdr, &lleflags);
238 const struct in6_addr *a6;
239 a6 = &(((const struct sockaddr_in6 *)dst)->sin6_addr);
240 ETHER_MAP_IPV6_MULTICAST(a6, eh->ether_dhost);
241 etype = htons(ETHERTYPE_IPV6);
242 memcpy(&eh->ether_type, &etype, sizeof(etype));
243 memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
248 if_printf(ifp, "can't handle af%d\n", dst->sa_family);
251 return (EAFNOSUPPORT);
254 if (error == EHOSTDOWN) {
255 if (ro != NULL && (ro->ro_flags & RT_HAS_GW) != 0)
256 error = EHOSTUNREACH;
262 *pflags = RT_MAY_LOOP;
263 if (lleflags & LLE_IFADDR)
270 * Ethernet output routine.
271 * Encapsulate a packet of type family for the local net.
272 * Use trailer local net encapsulation if enough data in first
273 * packet leaves a multiple of 512 bytes of data in remainder.
276 ether_output(struct ifnet *ifp, struct mbuf *m,
277 const struct sockaddr *dst, struct route *ro)
280 char linkhdr[ETHER_HDR_LEN], *phdr;
281 struct ether_header *eh;
284 int hlen; /* link layer header length */
290 phdr = ro->ro_prepend;
292 pflags = ro->ro_flags;
295 error = mac_ifnet_check_transmit(ifp, m);
301 if (ifp->if_flags & IFF_MONITOR)
303 if (!((ifp->if_flags & IFF_UP) &&
304 (ifp->if_drv_flags & IFF_DRV_RUNNING)))
308 /* No prepend data supplied. Try to calculate ourselves. */
310 hlen = ETHER_HDR_LEN;
311 error = ether_resolve_addr(ifp, m, dst, ro, phdr, &pflags);
313 return (error == EWOULDBLOCK ? 0 : error);
316 if ((pflags & RT_L2_ME) != 0) {
317 update_mbuf_csumflags(m, m);
318 return (if_simloop(ifp, m, dst->sa_family, 0));
320 loop_copy = pflags & RT_MAY_LOOP;
323 * Add local net header. If no space in first mbuf,
326 * Note that we do prepend regardless of RT_HAS_HEADER flag.
327 * This is done because BPF code shifts m_data pointer
328 * to the end of ethernet header prior to calling if_output().
330 M_PREPEND(m, hlen, M_NOWAIT);
333 if ((pflags & RT_HAS_HEADER) == 0) {
334 eh = mtod(m, struct ether_header *);
335 memcpy(eh, phdr, hlen);
339 * If a simplex interface, and the packet is being sent to our
340 * Ethernet address or a broadcast address, loopback a copy.
341 * XXX To make a simplex device behave exactly like a duplex
342 * device, we should copy in the case of sending to our own
343 * ethernet address (thus letting the original actually appear
344 * on the wire). However, we don't do that here for security
345 * reasons and compatibility with the original behavior.
347 if ((m->m_flags & M_BCAST) && loop_copy && (ifp->if_flags & IFF_SIMPLEX) &&
348 ((t = pf_find_mtag(m)) == NULL || !t->routed)) {
352 * Because if_simloop() modifies the packet, we need a
353 * writable copy through m_dup() instead of a readonly
354 * one as m_copy[m] would give us. The alternative would
355 * be to modify if_simloop() to handle the readonly mbuf,
356 * but performancewise it is mostly equivalent (trading
357 * extra data copying vs. extra locking).
359 * XXX This is a local workaround. A number of less
360 * often used kernel parts suffer from the same bug.
361 * See PR kern/105943 for a proposed general solution.
363 if ((n = m_dup(m, M_NOWAIT)) != NULL) {
364 update_mbuf_csumflags(m, n);
365 (void)if_simloop(ifp, n, dst->sa_family, hlen);
367 if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1);
371 * Bridges require special output handling.
373 if (ifp->if_bridge) {
374 BRIDGE_OUTPUT(ifp, m, error);
378 #if defined(INET) || defined(INET6)
380 (error = (*carp_output_p)(ifp, m, dst)))
384 /* Handle ng_ether(4) processing, if any */
385 if (ifp->if_l2com != NULL) {
386 KASSERT(ng_ether_output_p != NULL,
387 ("ng_ether_output_p is NULL"));
388 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
397 /* Continue with link-layer output */
398 return ether_output_frame(ifp, m);
402 * Ethernet link layer output routine to send a raw frame to the device.
404 * This assumes that the 14 byte Ethernet header is present and contiguous
405 * in the first mbuf (if BRIDGE'ing).
408 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
412 if (PFIL_HOOKED(&V_link_pfil_hook)) {
413 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_OUT, NULL);
423 * Queue message on interface, update output statistics if
424 * successful, and start output if interface not yet active.
426 return ((ifp->if_transmit)(ifp, m));
430 * Process a received Ethernet packet; the packet is in the
431 * mbuf chain m with the ethernet header at the front.
434 ether_input_internal(struct ifnet *ifp, struct mbuf *m)
436 struct ether_header *eh;
439 if ((ifp->if_flags & IFF_UP) == 0) {
444 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
445 if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n");
450 if (m->m_len < ETHER_HDR_LEN) {
451 /* XXX maybe should pullup? */
452 if_printf(ifp, "discard frame w/o leading ethernet "
453 "header (len %u pkt len %u)\n",
454 m->m_len, m->m_pkthdr.len);
455 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
459 eh = mtod(m, struct ether_header *);
460 etype = ntohs(eh->ether_type);
461 random_harvest_queue(m, sizeof(*m), 2, RANDOM_NET_ETHER);
463 CURVNET_SET_QUIET(ifp->if_vnet);
465 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
466 if (ETHER_IS_BROADCAST(eh->ether_dhost))
467 m->m_flags |= M_BCAST;
469 m->m_flags |= M_MCAST;
470 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
475 * Tag the mbuf with an appropriate MAC label before any other
476 * consumers can get to it.
478 mac_ifnet_create_mbuf(ifp, m);
482 * Give bpf a chance at the packet.
484 ETHER_BPF_MTAP(ifp, m);
487 * If the CRC is still on the packet, trim it off. We do this once
488 * and once only in case we are re-entered. Nothing else on the
489 * Ethernet receive path expects to see the FCS.
491 if (m->m_flags & M_HASFCS) {
492 m_adj(m, -ETHER_CRC_LEN);
493 m->m_flags &= ~M_HASFCS;
496 if (!(ifp->if_capenable & IFCAP_HWSTATS))
497 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
499 /* Allow monitor mode to claim this frame, after stats are updated. */
500 if (ifp->if_flags & IFF_MONITOR) {
506 /* Handle input from a lagg(4) port */
507 if (ifp->if_type == IFT_IEEE8023ADLAG) {
508 KASSERT(lagg_input_p != NULL,
509 ("%s: if_lagg not loaded!", __func__));
510 m = (*lagg_input_p)(ifp, m);
512 ifp = m->m_pkthdr.rcvif;
520 * If the hardware did not process an 802.1Q tag, do this now,
521 * to allow 802.1P priority frames to be passed to the main input
523 * TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels.
525 if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) {
526 struct ether_vlan_header *evl;
528 if (m->m_len < sizeof(*evl) &&
529 (m = m_pullup(m, sizeof(*evl))) == NULL) {
531 if_printf(ifp, "cannot pullup VLAN header\n");
533 if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
538 evl = mtod(m, struct ether_vlan_header *);
539 m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag);
540 m->m_flags |= M_VLANTAG;
542 bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN,
543 ETHER_HDR_LEN - ETHER_TYPE_LEN);
544 m_adj(m, ETHER_VLAN_ENCAP_LEN);
545 eh = mtod(m, struct ether_header *);
548 M_SETFIB(m, ifp->if_fib);
550 /* Allow ng_ether(4) to claim this frame. */
551 if (ifp->if_l2com != NULL) {
552 KASSERT(ng_ether_input_p != NULL,
553 ("%s: ng_ether_input_p is NULL", __func__));
554 m->m_flags &= ~M_PROMISC;
555 (*ng_ether_input_p)(ifp, &m);
560 eh = mtod(m, struct ether_header *);
564 * Allow if_bridge(4) to claim this frame.
565 * The BRIDGE_INPUT() macro will update ifp if the bridge changed it
566 * and the frame should be delivered locally.
568 if (ifp->if_bridge != NULL) {
569 m->m_flags &= ~M_PROMISC;
570 BRIDGE_INPUT(ifp, m);
575 eh = mtod(m, struct ether_header *);
578 #if defined(INET) || defined(INET6)
580 * Clear M_PROMISC on frame so that carp(4) will see it when the
581 * mbuf flows up to Layer 3.
582 * FreeBSD's implementation of carp(4) uses the inprotosw
583 * to dispatch IPPROTO_CARP. carp(4) also allocates its own
584 * Ethernet addresses of the form 00:00:5e:00:01:xx, which
585 * is outside the scope of the M_PROMISC test below.
586 * TODO: Maintain a hash table of ethernet addresses other than
587 * ether_dhost which may be active on this ifp.
589 if (ifp->if_carp && (*carp_forus_p)(ifp, eh->ether_dhost)) {
590 m->m_flags &= ~M_PROMISC;
595 * If the frame received was not for our MAC address, set the
596 * M_PROMISC flag on the mbuf chain. The frame may need to
597 * be seen by the rest of the Ethernet input path in case of
598 * re-entry (e.g. bridge, vlan, netgraph) but should not be
599 * seen by upper protocol layers.
601 if (!ETHER_IS_MULTICAST(eh->ether_dhost) &&
602 bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0)
603 m->m_flags |= M_PROMISC;
611 * Ethernet input dispatch; by default, direct dispatch here regardless of
612 * global configuration. However, if RSS is enabled, hook up RSS affinity
613 * so that when deferred or hybrid dispatch is enabled, we can redistribute
616 * XXXRW: Would be nice if the ifnet passed up a flag indicating whether or
617 * not it had already done work distribution via multi-queue. Then we could
618 * direct dispatch in the event load balancing was already complete and
619 * handle the case of interfaces with different capabilities better.
621 * XXXRW: Sort of want an M_DISTRIBUTED flag to avoid multiple distributions
622 * at multiple layers?
624 * XXXRW: For now, enable all this only if RSS is compiled in, although it
625 * works fine without RSS. Need to characterise the performance overhead
626 * of the detour through the netisr code in the event the result is always
630 ether_nh_input(struct mbuf *m)
634 KASSERT(m->m_pkthdr.rcvif != NULL,
635 ("%s: NULL interface pointer", __func__));
636 ether_input_internal(m->m_pkthdr.rcvif, m);
639 static struct netisr_handler ether_nh = {
641 .nh_handler = ether_nh_input,
642 .nh_proto = NETISR_ETHER,
644 .nh_policy = NETISR_POLICY_CPU,
645 .nh_dispatch = NETISR_DISPATCH_DIRECT,
646 .nh_m2cpuid = rss_m2cpuid,
648 .nh_policy = NETISR_POLICY_SOURCE,
649 .nh_dispatch = NETISR_DISPATCH_DIRECT,
654 ether_init(__unused void *arg)
657 netisr_register(ðer_nh);
659 SYSINIT(ether, SI_SUB_INIT_IF, SI_ORDER_ANY, ether_init, NULL);
662 vnet_ether_init(__unused void *arg)
666 /* Initialize packet filter hooks. */
667 V_link_pfil_hook.ph_type = PFIL_TYPE_AF;
668 V_link_pfil_hook.ph_af = AF_LINK;
669 if ((i = pfil_head_register(&V_link_pfil_hook)) != 0)
670 printf("%s: WARNING: unable to register pfil link hook, "
671 "error %d\n", __func__, i);
673 VNET_SYSINIT(vnet_ether_init, SI_SUB_PROTO_IF, SI_ORDER_ANY,
674 vnet_ether_init, NULL);
677 vnet_ether_destroy(__unused void *arg)
681 if ((i = pfil_head_unregister(&V_link_pfil_hook)) != 0)
682 printf("%s: WARNING: unable to unregister pfil link hook, "
683 "error %d\n", __func__, i);
685 VNET_SYSUNINIT(vnet_ether_uninit, SI_SUB_PROTO_IF, SI_ORDER_ANY,
686 vnet_ether_destroy, NULL);
691 ether_input(struct ifnet *ifp, struct mbuf *m)
697 * The drivers are allowed to pass in a chain of packets linked with
698 * m_nextpkt. We split them up into separate packets here and pass
699 * them up. This allows the drivers to amortize the receive lock.
706 * We will rely on rcvif being set properly in the deferred context,
707 * so assert it is correct here.
709 KASSERT(m->m_pkthdr.rcvif == ifp, ("%s: ifnet mismatch", __func__));
710 netisr_dispatch(NETISR_ETHER, m);
716 * Upper layer processing for a received Ethernet packet.
719 ether_demux(struct ifnet *ifp, struct mbuf *m)
721 struct ether_header *eh;
725 KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__));
727 /* Do not grab PROMISC frames in case we are re-entered. */
728 if (PFIL_HOOKED(&V_link_pfil_hook) && !(m->m_flags & M_PROMISC)) {
729 i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_IN, NULL);
731 if (i != 0 || m == NULL)
735 eh = mtod(m, struct ether_header *);
736 ether_type = ntohs(eh->ether_type);
739 * If this frame has a VLAN tag other than 0, call vlan_input()
740 * if its module is loaded. Otherwise, drop.
742 if ((m->m_flags & M_VLANTAG) &&
743 EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) {
744 if (ifp->if_vlantrunk == NULL) {
745 if_inc_counter(ifp, IFCOUNTER_NOPROTO, 1);
749 KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!",
751 /* Clear before possibly re-entering ether_input(). */
752 m->m_flags &= ~M_PROMISC;
753 (*vlan_input_p)(ifp, m);
758 * Pass promiscuously received frames to the upper layer if the user
759 * requested this by setting IFF_PPROMISC. Otherwise, drop them.
761 if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) {
767 * Reset layer specific mbuf flags to avoid confusing upper layers.
768 * Strip off Ethernet header.
770 m->m_flags &= ~M_VLANTAG;
772 m_adj(m, ETHER_HDR_LEN);
775 * Dispatch frame to upper layer.
777 switch (ether_type) {
784 if (ifp->if_flags & IFF_NOARP) {
785 /* Discard packet if ARP is disabled on interface */
800 netisr_dispatch(isr, m);
805 * Packet is to be discarded. If netgraph is present,
806 * hand the packet to it for last chance processing;
807 * otherwise dispose of it.
809 if (ifp->if_l2com != NULL) {
810 KASSERT(ng_ether_input_orphan_p != NULL,
811 ("ng_ether_input_orphan_p is NULL"));
813 * Put back the ethernet header so netgraph has a
814 * consistent view of inbound packets.
816 M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
817 (*ng_ether_input_orphan_p)(ifp, m);
824 * Convert Ethernet address to printable (loggable) representation.
825 * This routine is for compatibility; it's better to just use
827 * printf("%6D", <pointer to address>, ":");
829 * since there's no static buffer involved.
832 ether_sprintf(const u_char *ap)
834 static char etherbuf[18];
835 snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
840 * Perform common duties while attaching to interface list
843 ether_ifattach(struct ifnet *ifp, const u_int8_t *lla)
847 struct sockaddr_dl *sdl;
849 ifp->if_addrlen = ETHER_ADDR_LEN;
850 ifp->if_hdrlen = ETHER_HDR_LEN;
852 ifp->if_mtu = ETHERMTU;
853 ifp->if_output = ether_output;
854 ifp->if_input = ether_input;
855 ifp->if_resolvemulti = ether_resolvemulti;
856 ifp->if_requestencap = ether_requestencap;
858 ifp->if_reassign = ether_reassign;
860 if (ifp->if_baudrate == 0)
861 ifp->if_baudrate = IF_Mbps(10); /* just a default */
862 ifp->if_broadcastaddr = etherbroadcastaddr;
865 KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
866 sdl = (struct sockaddr_dl *)ifa->ifa_addr;
867 sdl->sdl_type = IFT_ETHER;
868 sdl->sdl_alen = ifp->if_addrlen;
869 bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
871 bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
872 if (ng_ether_attach_p != NULL)
873 (*ng_ether_attach_p)(ifp);
875 /* Announce Ethernet MAC address if non-zero. */
876 for (i = 0; i < ifp->if_addrlen; i++)
879 if (i != ifp->if_addrlen)
880 if_printf(ifp, "Ethernet address: %6D\n", lla, ":");
882 uuid_ether_add(LLADDR(sdl));
886 * Perform common duties while detaching an Ethernet interface
889 ether_ifdetach(struct ifnet *ifp)
891 struct sockaddr_dl *sdl;
893 sdl = (struct sockaddr_dl *)(ifp->if_addr->ifa_addr);
894 uuid_ether_del(LLADDR(sdl));
896 if (ifp->if_l2com != NULL) {
897 KASSERT(ng_ether_detach_p != NULL,
898 ("ng_ether_detach_p is NULL"));
899 (*ng_ether_detach_p)(ifp);
908 ether_reassign(struct ifnet *ifp, struct vnet *new_vnet, char *unused __unused)
911 if (ifp->if_l2com != NULL) {
912 KASSERT(ng_ether_detach_p != NULL,
913 ("ng_ether_detach_p is NULL"));
914 (*ng_ether_detach_p)(ifp);
917 if (ng_ether_attach_p != NULL) {
918 CURVNET_SET_QUIET(new_vnet);
919 (*ng_ether_attach_p)(ifp);
925 SYSCTL_DECL(_net_link);
926 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
930 * This is for reference. We have a table-driven version
931 * of the little-endian crc32 generator, which is faster
932 * than the double-loop.
935 ether_crc32_le(const uint8_t *buf, size_t len)
942 crc = 0xffffffff; /* initial value */
944 for (i = 0; i < len; i++) {
945 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
946 carry = (crc ^ data) & 1;
949 crc = (crc ^ ETHER_CRC_POLY_LE);
957 ether_crc32_le(const uint8_t *buf, size_t len)
959 static const uint32_t crctab[] = {
960 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
961 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
962 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
963 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
968 crc = 0xffffffff; /* initial value */
970 for (i = 0; i < len; i++) {
972 crc = (crc >> 4) ^ crctab[crc & 0xf];
973 crc = (crc >> 4) ^ crctab[crc & 0xf];
981 ether_crc32_be(const uint8_t *buf, size_t len)
988 crc = 0xffffffff; /* initial value */
990 for (i = 0; i < len; i++) {
991 for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
992 carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
995 crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
1003 ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
1005 struct ifaddr *ifa = (struct ifaddr *) data;
1006 struct ifreq *ifr = (struct ifreq *) data;
1011 ifp->if_flags |= IFF_UP;
1013 switch (ifa->ifa_addr->sa_family) {
1016 ifp->if_init(ifp->if_softc); /* before arpwhohas */
1017 arp_ifinit(ifp, ifa);
1021 ifp->if_init(ifp->if_softc);
1028 struct sockaddr *sa;
1030 sa = (struct sockaddr *) & ifr->ifr_data;
1031 bcopy(IF_LLADDR(ifp),
1032 (caddr_t) sa->sa_data, ETHER_ADDR_LEN);
1038 * Set the interface MTU.
1040 if (ifr->ifr_mtu > ETHERMTU) {
1043 ifp->if_mtu = ifr->ifr_mtu;
1047 error = EINVAL; /* XXX netbsd has ENOTTY??? */
1054 ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
1055 struct sockaddr *sa)
1057 struct sockaddr_dl *sdl;
1059 struct sockaddr_in *sin;
1062 struct sockaddr_in6 *sin6;
1066 switch(sa->sa_family) {
1069 * No mapping needed. Just check that it's a valid MC address.
1071 sdl = (struct sockaddr_dl *)sa;
1072 e_addr = LLADDR(sdl);
1073 if (!ETHER_IS_MULTICAST(e_addr))
1074 return EADDRNOTAVAIL;
1080 sin = (struct sockaddr_in *)sa;
1081 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
1082 return EADDRNOTAVAIL;
1083 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1084 sdl->sdl_alen = ETHER_ADDR_LEN;
1085 e_addr = LLADDR(sdl);
1086 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
1087 *llsa = (struct sockaddr *)sdl;
1092 sin6 = (struct sockaddr_in6 *)sa;
1093 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
1095 * An IP6 address of 0 means listen to all
1096 * of the Ethernet multicast address used for IP6.
1097 * (This is used for multicast routers.)
1099 ifp->if_flags |= IFF_ALLMULTI;
1103 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
1104 return EADDRNOTAVAIL;
1105 sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
1106 sdl->sdl_alen = ETHER_ADDR_LEN;
1107 e_addr = LLADDR(sdl);
1108 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
1109 *llsa = (struct sockaddr *)sdl;
1115 * Well, the text isn't quite right, but it's the name
1118 return EAFNOSUPPORT;
1122 static moduledata_t ether_mod = {
1127 ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen)
1129 struct ether_vlan_header vlan;
1132 KASSERT((m->m_flags & M_VLANTAG) != 0,
1133 ("%s: vlan information not present", __func__));
1134 KASSERT(m->m_len >= sizeof(struct ether_header),
1135 ("%s: mbuf not large enough for header", __func__));
1136 bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header));
1137 vlan.evl_proto = vlan.evl_encap_proto;
1138 vlan.evl_encap_proto = htons(ETHERTYPE_VLAN);
1139 vlan.evl_tag = htons(m->m_pkthdr.ether_vtag);
1140 m->m_len -= sizeof(struct ether_header);
1141 m->m_data += sizeof(struct ether_header);
1143 * If a data link has been supplied by the caller, then we will need to
1144 * re-create a stack allocated mbuf chain with the following structure:
1146 * (1) mbuf #1 will contain the supplied data link
1147 * (2) mbuf #2 will contain the vlan header
1148 * (3) mbuf #3 will contain the original mbuf's packet data
1150 * Otherwise, submit the packet and vlan header via bpf_mtap2().
1154 mv.m_data = (caddr_t)&vlan;
1155 mv.m_len = sizeof(vlan);
1161 bpf_mtap2(bp, &vlan, sizeof(vlan), m);
1162 m->m_len += sizeof(struct ether_header);
1163 m->m_data -= sizeof(struct ether_header);
1167 ether_vlanencap(struct mbuf *m, uint16_t tag)
1169 struct ether_vlan_header *evl;
1171 M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_NOWAIT);
1174 /* M_PREPEND takes care of m_len, m_pkthdr.len for us */
1176 if (m->m_len < sizeof(*evl)) {
1177 m = m_pullup(m, sizeof(*evl));
1183 * Transform the Ethernet header into an Ethernet header
1184 * with 802.1Q encapsulation.
1186 evl = mtod(m, struct ether_vlan_header *);
1187 bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN,
1188 (char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN);
1189 evl->evl_encap_proto = htons(ETHERTYPE_VLAN);
1190 evl->evl_tag = htons(tag);
1194 DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY);
1195 MODULE_VERSION(ether, 1);