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. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * @(#)if_ethersubr.c 8.1 (Berkeley) 6/10/93
34 * $FreeBSD: src/sys/net/if_ethersubr.c,v 1.70.2.33 2003/04/28 15:45:53 archie Exp $
35 * $DragonFly: src/sys/net/if_ethersubr.c,v 1.83 2008/08/22 09:14:16 sephe Exp $
38 #include "opt_atalk.h"
40 #include "opt_inet6.h"
43 #include "opt_netgraph.h"
45 #include "opt_ethernet.h"
47 #include <sys/param.h>
48 #include <sys/systm.h>
49 #include <sys/globaldata.h>
50 #include <sys/kernel.h>
51 #include <sys/malloc.h>
53 #include <sys/msgport.h>
54 #include <sys/socket.h>
55 #include <sys/sockio.h>
56 #include <sys/sysctl.h>
57 #include <sys/thread.h>
58 #include <sys/thread2.h>
61 #include <net/netisr.h>
62 #include <net/route.h>
63 #include <net/if_llc.h>
64 #include <net/if_dl.h>
65 #include <net/if_types.h>
66 #include <net/ifq_var.h>
68 #include <net/ethernet.h>
69 #include <net/vlan/if_vlan_ether.h>
70 #include <net/netmsg2.h>
72 #if defined(INET) || defined(INET6)
73 #include <netinet/in.h>
74 #include <netinet/in_var.h>
75 #include <netinet/if_ether.h>
76 #include <net/ipfw/ip_fw.h>
77 #include <net/dummynet/ip_dummynet.h>
80 #include <netinet6/nd6.h>
84 #include <netinet/ip_carp.h>
88 #include <netproto/ipx/ipx.h>
89 #include <netproto/ipx/ipx_if.h>
90 int (*ef_inputp)(struct ifnet*, const struct ether_header *eh, struct mbuf *m);
91 int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, struct sockaddr *dst,
92 short *tp, int *hlen);
97 #include <netns/ns_if.h>
99 int ether_outputdebug = 0;
100 int ether_inputdebug = 0;
104 #include <netproto/atalk/at.h>
105 #include <netproto/atalk/at_var.h>
106 #include <netproto/atalk/at_extern.h>
108 #define llc_snap_org_code llc_un.type_snap.org_code
109 #define llc_snap_ether_type llc_un.type_snap.ether_type
111 extern u_char at_org_code[3];
112 extern u_char aarp_org_code[3];
113 #endif /* NETATALK */
116 #include <netproto/mpls/mpls.h>
119 /* netgraph node hooks for ng_ether(4) */
120 void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp);
121 void (*ng_ether_input_orphan_p)(struct ifnet *ifp,
122 struct mbuf *m, const struct ether_header *eh);
123 int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
124 void (*ng_ether_attach_p)(struct ifnet *ifp);
125 void (*ng_ether_detach_p)(struct ifnet *ifp);
127 void (*vlan_input2_p)(struct mbuf *);
129 static int ether_output(struct ifnet *, struct mbuf *, struct sockaddr *,
131 static void ether_restore_header(struct mbuf **, const struct ether_header *,
132 const struct ether_header *);
137 struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *);
138 int (*bridge_output_p)(struct ifnet *, struct mbuf *);
139 void (*bridge_dn_p)(struct mbuf *, struct ifnet *);
141 static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
144 const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = {
145 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
148 #define gotoerr(e) do { error = (e); goto bad; } while (0)
149 #define IFP2AC(ifp) ((struct arpcom *)(ifp))
151 static boolean_t ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst,
153 const struct ether_header *eh);
155 static int ether_ipfw;
156 static u_int ether_restore_hdr;
157 static u_int ether_prepend_hdr;
159 SYSCTL_DECL(_net_link);
160 SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
161 SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW,
162 ðer_ipfw, 0, "Pass ether pkts through firewall");
163 SYSCTL_UINT(_net_link_ether, OID_AUTO, restore_hdr, CTLFLAG_RW,
164 ðer_restore_hdr, 0, "# of ether header restoration");
165 SYSCTL_UINT(_net_link_ether, OID_AUTO, prepend_hdr, CTLFLAG_RW,
166 ðer_prepend_hdr, 0,
167 "# of ether header restoration which prepends mbuf");
170 * Ethernet output routine.
171 * Encapsulate a packet of type family for the local net.
172 * Use trailer local net encapsulation if enough data in first
173 * packet leaves a multiple of 512 bytes of data in remainder.
174 * Assumes that ifp is actually pointer to arpcom structure.
177 ether_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst,
180 struct ether_header *eh, *deh;
183 int hlen = ETHER_HDR_LEN; /* link layer header length */
184 struct arpcom *ac = IFP2AC(ifp);
187 ASSERT_NOT_SERIALIZED(ifp->if_serializer);
189 if (ifp->if_flags & IFF_MONITOR)
191 if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING))
194 M_PREPEND(m, sizeof(struct ether_header), MB_DONTWAIT);
197 eh = mtod(m, struct ether_header *);
198 edst = eh->ether_dhost;
201 * Fill in the destination ethernet address and frame type.
203 switch (dst->sa_family) {
206 if (!arpresolve(ifp, rt, m, dst, edst))
207 return (0); /* if not yet resolved */
209 if (m->m_flags & M_MPLSLABELED)
210 eh->ether_type = htons(ETHERTYPE_MPLS);
213 eh->ether_type = htons(ETHERTYPE_IP);
218 if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, edst))
219 return (0); /* Something bad happenned. */
220 eh->ether_type = htons(ETHERTYPE_IPV6);
225 if (ef_outputp != NULL) {
226 error = ef_outputp(ifp, &m, dst, &eh->ether_type,
231 eh->ether_type = htons(ETHERTYPE_IPX);
232 bcopy(&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host),
233 edst, ETHER_ADDR_LEN);
239 struct at_ifaddr *aa;
241 if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) {
246 * In the phase 2 case, need to prepend an mbuf for
247 * the llc header. Since we must preserve the value
248 * of m, which is passed to us by value, we m_copy()
249 * the first mbuf, and use it for our llc header.
251 if (aa->aa_flags & AFA_PHASE2) {
254 M_PREPEND(m, sizeof(struct llc), MB_DONTWAIT);
255 eh = mtod(m, struct ether_header *);
256 edst = eh->ether_dhost;
257 llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP;
258 llc.llc_control = LLC_UI;
259 bcopy(at_org_code, llc.llc_snap_org_code,
261 llc.llc_snap_ether_type = htons(ETHERTYPE_AT);
263 mtod(m, caddr_t) + sizeof(struct ether_header),
265 eh->ether_type = htons(m->m_pkthdr.len);
266 hlen = sizeof(struct llc) + ETHER_HDR_LEN;
268 eh->ether_type = htons(ETHERTYPE_AT);
270 if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst))
279 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
280 eh->ether_type = 0x8137;
282 case 0x0: /* Novell 802.3 */
283 eh->ether_type = htons(m->m_pkthdr.len);
285 case 0xe0e0: /* Novell 802.2 and Token-Ring */
286 M_PREPEND(m, 3, MB_DONTWAIT);
287 eh = mtod(m, struct ether_header *);
288 edst = eh->ether_dhost;
289 eh->ether_type = htons(m->m_pkthdr.len);
290 cp = mtod(m, u_char *) + sizeof(struct ether_header);
296 bcopy(&(((struct sockaddr_ns *)dst)->sns_addr.x_host), edst,
299 * XXX if ns_thishost is the same as the node's ethernet
300 * address then just the default code will catch this anyhow.
301 * So I'm not sure if this next clause should be here at all?
304 if (bcmp(edst, &ns_thishost, ETHER_ADDR_LEN) == 0) {
305 m->m_pkthdr.rcvif = ifp;
306 netisr_dispatch(NETISR_NS, m);
309 if (bcmp(edst, &ns_broadhost, ETHER_ADDR_LEN) == 0)
310 m->m_flags |= M_BCAST;
313 case pseudo_AF_HDRCMPLT:
315 loop_copy = -1; /* if this is for us, don't do it */
316 deh = (struct ether_header *)dst->sa_data;
317 memcpy(edst, deh->ether_dhost, ETHER_ADDR_LEN);
318 eh->ether_type = deh->ether_type;
322 if_printf(ifp, "can't handle af%d\n", dst->sa_family);
323 gotoerr(EAFNOSUPPORT);
326 if (dst->sa_family == pseudo_AF_HDRCMPLT) /* unlikely */
327 memcpy(eh->ether_shost,
328 ((struct ether_header *)dst->sa_data)->ether_shost,
331 memcpy(eh->ether_shost, ac->ac_enaddr, ETHER_ADDR_LEN);
334 * Bridges require special output handling.
336 if (ifp->if_bridge) {
337 KASSERT(bridge_output_p != NULL,
338 ("%s: if_bridge not loaded!", __func__));
339 return bridge_output_p(ifp, m);
343 * If a simplex interface, and the packet is being sent to our
344 * Ethernet address or a broadcast address, loopback a copy.
345 * XXX To make a simplex device behave exactly like a duplex
346 * device, we should copy in the case of sending to our own
347 * ethernet address (thus letting the original actually appear
348 * on the wire). However, we don't do that here for security
349 * reasons and compatibility with the original behavior.
351 if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) {
354 if (m->m_pkthdr.csum_flags & CSUM_IP)
355 csum_flags |= (CSUM_IP_CHECKED | CSUM_IP_VALID);
356 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
357 csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR);
358 if ((m->m_flags & M_BCAST) || (loop_copy > 0)) {
361 if ((n = m_copypacket(m, MB_DONTWAIT)) != NULL) {
362 n->m_pkthdr.csum_flags |= csum_flags;
363 if (csum_flags & CSUM_DATA_VALID)
364 n->m_pkthdr.csum_data = 0xffff;
365 if_simloop(ifp, n, dst->sa_family, hlen);
368 } else if (bcmp(eh->ether_dhost, eh->ether_shost,
369 ETHER_ADDR_LEN) == 0) {
370 m->m_pkthdr.csum_flags |= csum_flags;
371 if (csum_flags & CSUM_DATA_VALID)
372 m->m_pkthdr.csum_data = 0xffff;
373 if_simloop(ifp, m, dst->sa_family, hlen);
374 return (0); /* XXX */
379 if (ifp->if_carp && (error = carp_output(ifp, m, dst, NULL)))
384 /* Handle ng_ether(4) processing, if any */
385 if (ng_ether_output_p != NULL) {
386 if ((error = (*ng_ether_output_p)(ifp, &m)) != 0)
392 /* Continue with link-layer output */
393 return ether_output_frame(ifp, m);
401 * Ethernet link layer output routine to send a raw frame to the device.
403 * This assumes that the 14 byte Ethernet header is present and contiguous
407 ether_output_frame(struct ifnet *ifp, struct mbuf *m)
409 struct ip_fw *rule = NULL;
411 struct altq_pktattr pktattr;
414 ASSERT_NOT_SERIALIZED(ifp->if_serializer);
416 /* Extract info from dummynet tag */
417 mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL);
419 rule = ((struct dn_pkt *)m_tag_data(mtag))->dn_priv;
421 m_tag_delete(m, mtag);
425 if (ifq_is_enabled(&ifp->if_snd))
426 altq_etherclassify(&ifp->if_snd, m, &pktattr);
428 if (IPFW_LOADED && ether_ipfw != 0) {
429 struct ether_header save_eh, *eh;
431 eh = mtod(m, struct ether_header *);
433 m_adj(m, ETHER_HDR_LEN);
434 if (!ether_ipfw_chk(&m, ifp, &rule, eh)) {
438 return ENOBUFS; /* pkt dropped */
440 return 0; /* consumed e.g. in a pipe */
443 /* packet was ok, restore the ethernet header */
444 ether_restore_header(&m, eh, &save_eh);
453 * Queue message on interface, update output statistics if
454 * successful, and start output if interface not yet active.
456 error = ifq_dispatch(ifp, m, &pktattr);
461 * ipfw processing for ethernet packets (in and out).
462 * The second parameter is NULL from ether_demux(), and ifp from
463 * ether_output_frame().
466 ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, struct ip_fw **rule,
467 const struct ether_header *eh)
469 struct ether_header save_eh = *eh; /* might be a ptr in m */
470 struct ip_fw_args args;
474 if (*rule != NULL && fw_one_pass)
475 return TRUE; /* dummynet packet, already partially processed */
478 * I need some amount of data to be contiguous.
480 i = min((*m0)->m_pkthdr.len, max_protohdr);
481 if ((*m0)->m_len < i) {
482 *m0 = m_pullup(*m0, i);
490 if ((mtag = m_tag_find(*m0, PACKET_TAG_IPFW_DIVERT, NULL)) != NULL)
491 m_tag_delete(*m0, mtag);
492 if ((*m0)->m_pkthdr.fw_flags & IPFORWARD_MBUF_TAGGED) {
493 mtag = m_tag_find(*m0, PACKET_TAG_IPFORWARD, NULL);
494 KKASSERT(mtag != NULL);
495 m_tag_delete(*m0, mtag);
496 (*m0)->m_pkthdr.fw_flags &= ~IPFORWARD_MBUF_TAGGED;
499 args.m = *m0; /* the packet we are looking at */
500 args.oif = dst; /* destination, if any */
501 args.rule = *rule; /* matching rule to restart */
502 args.eh = &save_eh; /* MAC header for bridged/MAC packets */
503 i = ip_fw_chk_ptr(&args);
507 if ((i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL) /* drop */
510 if (i == 0) /* a PASS rule. */
513 if (i & IP_FW_PORT_DYNT_FLAG) {
515 * Pass the pkt to dummynet, which consumes it.
519 m = *m0; /* pass the original to dummynet */
520 *m0 = NULL; /* and nothing back to the caller */
522 ether_restore_header(&m, eh, &save_eh);
526 ip_fw_dn_io_ptr(m, (i & 0xffff),
527 dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args);
531 * XXX at some point add support for divert/forward actions.
532 * If none of the above matches, we have to drop the pkt.
538 ether_input(struct ifnet *ifp, struct mbuf *m)
540 ether_input_chain2(ifp, m, NULL);
544 * Perform common duties while attaching to interface list
547 ether_ifattach(struct ifnet *ifp, uint8_t *lla, lwkt_serialize_t serializer)
549 ether_ifattach_bpf(ifp, lla, DLT_EN10MB, sizeof(struct ether_header),
554 ether_ifattach_bpf(struct ifnet *ifp, uint8_t *lla, u_int dlt, u_int hdrlen,
555 lwkt_serialize_t serializer)
557 struct sockaddr_dl *sdl;
559 ifp->if_type = IFT_ETHER;
560 ifp->if_addrlen = ETHER_ADDR_LEN;
561 ifp->if_hdrlen = ETHER_HDR_LEN;
562 if_attach(ifp, serializer);
563 ifp->if_mtu = ETHERMTU;
564 if (ifp->if_baudrate == 0)
565 ifp->if_baudrate = 10000000;
566 ifp->if_output = ether_output;
567 ifp->if_input = ether_input;
568 ifp->if_resolvemulti = ether_resolvemulti;
569 ifp->if_broadcastaddr = etherbroadcastaddr;
570 sdl = IF_LLSOCKADDR(ifp);
571 sdl->sdl_type = IFT_ETHER;
572 sdl->sdl_alen = ifp->if_addrlen;
573 bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
575 * XXX Keep the current drivers happy.
576 * XXX Remove once all drivers have been cleaned up
578 if (lla != IFP2AC(ifp)->ac_enaddr)
579 bcopy(lla, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen);
580 bpfattach(ifp, dlt, hdrlen);
581 if (ng_ether_attach_p != NULL)
582 (*ng_ether_attach_p)(ifp);
584 if_printf(ifp, "MAC address: %6D\n", lla, ":");
588 * Perform common duties while detaching an Ethernet interface
591 ether_ifdetach(struct ifnet *ifp)
595 if (ng_ether_detach_p != NULL)
596 (*ng_ether_detach_p)(ifp);
602 ether_ioctl(struct ifnet *ifp, int command, caddr_t data)
604 struct ifaddr *ifa = (struct ifaddr *) data;
605 struct ifreq *ifr = (struct ifreq *) data;
608 #define IF_INIT(ifp) \
610 if (((ifp)->if_flags & IFF_UP) == 0) { \
611 (ifp)->if_flags |= IFF_UP; \
612 (ifp)->if_init((ifp)->if_softc); \
616 ASSERT_SERIALIZED(ifp->if_serializer);
620 switch (ifa->ifa_addr->sa_family) {
623 IF_INIT(ifp); /* before arpwhohas */
624 arp_ifinit(ifp, ifa);
629 * XXX - This code is probably wrong
633 struct ipx_addr *ina = &IA_SIPX(ifa)->sipx_addr;
634 struct arpcom *ac = IFP2AC(ifp);
636 if (ipx_nullhost(*ina))
637 ina->x_host = *(union ipx_host *) ac->ac_enaddr;
639 bcopy(ina->x_host.c_host, ac->ac_enaddr,
640 sizeof ac->ac_enaddr);
642 IF_INIT(ifp); /* Set new address. */
648 * XXX - This code is probably wrong
652 struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr);
653 struct arpcom *ac = IFP2AC(ifp);
655 if (ns_nullhost(*ina))
656 ina->x_host = *(union ns_host *)(ac->ac_enaddr);
658 bcopy(ina->x_host.c_host, ac->ac_enaddr,
659 sizeof ac->ac_enaddr);
675 bcopy(IFP2AC(ifp)->ac_enaddr,
676 ((struct sockaddr *)ifr->ifr_data)->sa_data,
682 * Set the interface MTU.
684 if (ifr->ifr_mtu > ETHERMTU) {
687 ifp->if_mtu = ifr->ifr_mtu;
702 struct sockaddr **llsa,
705 struct sockaddr_dl *sdl;
706 struct sockaddr_in *sin;
708 struct sockaddr_in6 *sin6;
712 switch(sa->sa_family) {
715 * No mapping needed. Just check that it's a valid MC address.
717 sdl = (struct sockaddr_dl *)sa;
718 e_addr = LLADDR(sdl);
719 if ((e_addr[0] & 1) != 1)
720 return EADDRNOTAVAIL;
726 sin = (struct sockaddr_in *)sa;
727 if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
728 return EADDRNOTAVAIL;
729 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
731 sdl->sdl_len = sizeof *sdl;
732 sdl->sdl_family = AF_LINK;
733 sdl->sdl_index = ifp->if_index;
734 sdl->sdl_type = IFT_ETHER;
735 sdl->sdl_alen = ETHER_ADDR_LEN;
736 e_addr = LLADDR(sdl);
737 ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
738 *llsa = (struct sockaddr *)sdl;
743 sin6 = (struct sockaddr_in6 *)sa;
744 if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
746 * An IP6 address of 0 means listen to all
747 * of the Ethernet multicast address used for IP6.
748 * (This is used for multicast routers.)
750 ifp->if_flags |= IFF_ALLMULTI;
754 if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
755 return EADDRNOTAVAIL;
756 MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR,
758 sdl->sdl_len = sizeof *sdl;
759 sdl->sdl_family = AF_LINK;
760 sdl->sdl_index = ifp->if_index;
761 sdl->sdl_type = IFT_ETHER;
762 sdl->sdl_alen = ETHER_ADDR_LEN;
763 e_addr = LLADDR(sdl);
764 ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
765 *llsa = (struct sockaddr *)sdl;
771 * Well, the text isn't quite right, but it's the name
780 * This is for reference. We have a table-driven version
781 * of the little-endian crc32 generator, which is faster
782 * than the double-loop.
785 ether_crc32_le(const uint8_t *buf, size_t len)
787 uint32_t c, crc, carry;
790 crc = 0xffffffffU; /* initial value */
792 for (i = 0; i < len; i++) {
794 for (j = 0; j < 8; j++) {
795 carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01);
799 crc = (crc ^ ETHER_CRC_POLY_LE);
807 ether_crc32_le(const uint8_t *buf, size_t len)
809 static const uint32_t crctab[] = {
810 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
811 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
812 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
813 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
818 crc = 0xffffffffU; /* initial value */
820 for (i = 0; i < len; i++) {
822 crc = (crc >> 4) ^ crctab[crc & 0xf];
823 crc = (crc >> 4) ^ crctab[crc & 0xf];
831 ether_crc32_be(const uint8_t *buf, size_t len)
833 uint32_t c, crc, carry;
836 crc = 0xffffffffU; /* initial value */
838 for (i = 0; i < len; i++) {
840 for (j = 0; j < 8; j++) {
841 carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01);
845 crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
853 * find the size of ethernet header, and call classifier
856 altq_etherclassify(struct ifaltq *ifq, struct mbuf *m,
857 struct altq_pktattr *pktattr)
859 struct ether_header *eh;
861 int hlen, af, hdrsize;
864 hlen = sizeof(struct ether_header);
865 eh = mtod(m, struct ether_header *);
867 ether_type = ntohs(eh->ether_type);
868 if (ether_type < ETHERMTU) {
870 struct llc *llc = (struct llc *)(eh + 1);
873 if (m->m_len < hlen ||
874 llc->llc_dsap != LLC_SNAP_LSAP ||
875 llc->llc_ssap != LLC_SNAP_LSAP ||
876 llc->llc_control != LLC_UI)
877 goto bad; /* not snap! */
879 ether_type = ntohs(llc->llc_un.type_snap.ether_type);
882 if (ether_type == ETHERTYPE_IP) {
884 hdrsize = 20; /* sizeof(struct ip) */
886 } else if (ether_type == ETHERTYPE_IPV6) {
888 hdrsize = 40; /* sizeof(struct ip6_hdr) */
893 while (m->m_len <= hlen) {
897 hdr = m->m_data + hlen;
898 if (m->m_len < hlen + hdrsize) {
900 * ip header is not in a single mbuf. this should not
901 * happen in the current code.
902 * (todo: use m_pulldown in the future)
908 ifq_classify(ifq, m, af, pktattr);
915 pktattr->pattr_class = NULL;
916 pktattr->pattr_hdr = NULL;
917 pktattr->pattr_af = AF_UNSPEC;
921 ether_restore_header(struct mbuf **m0, const struct ether_header *eh,
922 const struct ether_header *save_eh)
924 struct mbuf *m = *m0;
929 * Prepend the header, optimize for the common case of
930 * eh pointing into the mbuf.
932 if ((const void *)(eh + 1) == (void *)m->m_data) {
933 m->m_data -= ETHER_HDR_LEN;
934 m->m_len += ETHER_HDR_LEN;
935 m->m_pkthdr.len += ETHER_HDR_LEN;
939 M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT);
941 bcopy(save_eh, mtod(m, struct ether_header *),
948 #ifdef ETHER_INPUT_CHAIN
951 ether_input_ipifunc(void *arg)
953 struct mbuf *m, *next;
961 port = m->m_pkthdr.header;
962 m->m_pkthdr.header = NULL;
965 &m->m_hdr.mh_netmsg.nm_netmsg.nm_lmsg);
972 ether_input_dispatch(struct mbuf_chain *chain)
977 for (i = 0; i < ncpus; ++i) {
978 if (chain[i].mc_head != NULL) {
979 lwkt_send_ipiq(globaldata_find(i),
980 ether_input_ipifunc, chain[i].mc_head);
984 if (chain->mc_head != NULL)
985 ether_input_ipifunc(chain->mc_head);
990 ether_input_chain_init(struct mbuf_chain *chain)
995 for (i = 0; i < ncpus; ++i)
996 chain[i].mc_head = chain[i].mc_tail = NULL;
998 chain->mc_head = chain->mc_tail = NULL;
1002 #endif /* ETHER_INPUT_CHAIN */
1005 * Upper layer processing for a received Ethernet packet.
1008 ether_demux_oncpu(struct ifnet *ifp, struct mbuf *m)
1010 struct ether_header *eh;
1011 int isr, redispatch;
1013 struct ip_fw *rule = NULL;
1020 KASSERT(m->m_len >= ETHER_HDR_LEN,
1021 ("ether header is no contiguous!\n"));
1023 eh = mtod(m, struct ether_header *);
1025 /* Extract info from dummynet tag */
1026 mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL);
1028 rule = ((struct dn_pkt *)m_tag_data(mtag))->dn_priv;
1029 KKASSERT(ifp == NULL);
1030 ifp = m->m_pkthdr.rcvif;
1032 m_tag_delete(m, mtag);
1035 if (rule) /* packet is passing the second time */
1040 * XXX: Okay, we need to call carp_forus() and - if it is for
1041 * us jump over code that does the normal check
1042 * "ac_enaddr == ether_dhost". The check sequence is a bit
1043 * different from OpenBSD, so we jump over as few code as
1044 * possible, to catch _all_ sanity checks. This needs
1045 * evaluation, to see if the carp ether_dhost values break any
1048 if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost))
1053 * Discard packet if upper layers shouldn't see it because
1054 * it was unicast to a different Ethernet address. If the
1055 * driver is working properly, then this situation can only
1056 * happen when the interface is in promiscuous mode.
1058 if (((ifp->if_flags & (IFF_PROMISC | IFF_PPROMISC)) == IFF_PROMISC) &&
1059 (eh->ether_dhost[0] & 1) == 0 &&
1060 bcmp(eh->ether_dhost, IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN)) {
1066 if (IPFW_LOADED && ether_ipfw != 0) {
1067 struct ether_header save_eh = *eh;
1069 /* XXX old crufty stuff, needs to be removed */
1070 m_adj(m, sizeof(struct ether_header));
1072 if (!ether_ipfw_chk(&m, NULL, &rule, eh)) {
1077 ether_restore_header(&m, eh, &save_eh);
1080 eh = mtod(m, struct ether_header *);
1083 ether_type = ntohs(eh->ether_type);
1084 KKASSERT(ether_type != ETHERTYPE_VLAN);
1086 if (m->m_flags & M_VLANTAG) {
1087 if (vlan_input2_p != NULL) {
1090 m->m_pkthdr.rcvif->if_noproto++;
1096 m_adj(m, sizeof(struct ether_header));
1099 switch (ether_type) {
1102 if (ipflow_fastforward(m))
1108 if (ifp->if_flags & IFF_NOARP) {
1109 /* Discard packet if ARP is disabled on interface */
1118 case ETHERTYPE_IPV6:
1125 if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
1132 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
1140 isr = NETISR_ATALK1;
1142 case ETHERTYPE_AARP:
1148 case ETHERTYPE_MPLS:
1149 case ETHERTYPE_MPLS_MCAST:
1150 /* Should have been set by ether_input_chain2(). */
1151 KKASSERT(m->m_flags & M_MPLSLABELED);
1158 * The accurate msgport is not determined before
1159 * we reach here, so redo the dispatching
1163 if (ef_inputp && ef_inputp(ifp, eh, m) == 0)
1167 checksum = mtod(m, ushort *);
1169 if ((ether_type <= ETHERMTU) &&
1170 ((*checksum == 0xffff) || (*checksum == 0xE0E0))) {
1171 if (*checksum == 0xE0E0) {
1172 m->m_pkthdr.len -= 3;
1181 if (ether_type > ETHERMTU)
1183 l = mtod(m, struct llc *);
1184 if (l->llc_dsap == LLC_SNAP_LSAP &&
1185 l->llc_ssap == LLC_SNAP_LSAP &&
1186 l->llc_control == LLC_UI) {
1187 if (bcmp(&(l->llc_snap_org_code)[0], at_org_code,
1188 sizeof at_org_code) == 0 &&
1189 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) {
1190 m_adj(m, sizeof(struct llc));
1191 isr = NETISR_ATALK2;
1194 if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code,
1195 sizeof aarp_org_code) == 0 &&
1196 ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) {
1197 m_adj(m, sizeof(struct llc));
1204 if (ng_ether_input_orphan_p != NULL)
1205 ng_ether_input_orphan_p(ifp, m, eh);
1214 netisr_dispatch(isr, m);
1218 * First we perform any link layer operations, then continue to the
1219 * upper layers with ether_demux_oncpu().
1222 ether_input_oncpu(struct ifnet *ifp, struct mbuf *m)
1224 if ((ifp->if_flags & (IFF_UP | IFF_MONITOR)) != IFF_UP) {
1226 * Receiving interface's flags are changed, when this
1227 * packet is waiting for processing; discard it.
1234 * Tap the packet off here for a bridge. bridge_input()
1235 * will return NULL if it has consumed the packet, otherwise
1236 * it gets processed as normal. Note that bridge_input()
1237 * will always return the original packet if we need to
1238 * process it locally.
1240 if (ifp->if_bridge) {
1241 KASSERT(bridge_input_p != NULL,
1242 ("%s: if_bridge not loaded!", __func__));
1244 if(m->m_flags & M_PROTO1) {
1245 m->m_flags &= ~M_PROTO1;
1247 /* clear M_PROMISC, in case the packets comes from a vlan */
1248 /* m->m_flags &= ~M_PROMISC; */
1249 m = bridge_input_p(ifp, m);
1253 KASSERT(ifp == m->m_pkthdr.rcvif,
1254 ("bridge_input_p changed rcvif\n"));
1258 /* Handle ng_ether(4) processing, if any */
1259 if (ng_ether_input_p != NULL) {
1260 ng_ether_input_p(ifp, &m);
1265 /* Continue with upper layer processing */
1266 ether_demux_oncpu(ifp, m);
1270 ether_input_handler(struct netmsg *nmsg)
1272 struct netmsg_packet *nmp = (struct netmsg_packet *)nmsg;
1278 ifp = m->m_pkthdr.rcvif;
1280 ether_input_oncpu(ifp, m);
1283 static __inline void
1284 ether_init_netpacket(int num, struct mbuf *m)
1286 struct netmsg_packet *pmsg;
1288 pmsg = &m->m_hdr.mh_netmsg;
1289 netmsg_init(&pmsg->nm_netmsg, &netisr_apanic_rport, 0,
1290 ether_input_handler);
1291 pmsg->nm_packet = m;
1292 pmsg->nm_netmsg.nm_lmsg.u.ms_result = num;
1295 static __inline struct lwkt_port *
1296 ether_mport(int num, struct mbuf **m)
1298 if (num == NETISR_MAX) {
1300 * All packets whose target msgports can't be
1301 * determined here are dispatched to netisr0,
1302 * where further dispatching may happen.
1304 return cpu_portfn(0);
1306 return netisr_find_port(num, m);
1310 * Process a received Ethernet packet.
1312 * The ethernet header is assumed to be in the mbuf so the caller
1313 * MUST MAKE SURE that there are at least sizeof(struct ether_header)
1314 * bytes in the first mbuf.
1316 * We first try to find the target msgport for this ether frame, if
1317 * there is no target msgport for it, this ether frame is discarded,
1318 * else we do following processing according to whether 'chain' is
1320 * - If 'chain' is NULL, this ether frame is sent to the target msgport
1321 * immediately. This situation happens when ether_input_chain2 is
1322 * accessed through ifnet.if_input.
1323 * - If 'chain' is not NULL, this ether frame is queued to the 'chain'
1324 * bucket indexed by the target msgport's cpuid and the target msgport
1325 * is saved in mbuf's m_pkthdr.m_head. Caller of ether_input_chain2
1326 * must initialize 'chain' by calling ether_input_chain_init().
1327 * ether_input_dispatch must be called later to send ether frames
1328 * queued on 'chain' to their target msgport.
1331 ether_input_chain2(struct ifnet *ifp, struct mbuf *m, struct mbuf_chain *chain)
1333 struct ether_header *eh, *save_eh, save_eh0;
1334 struct lwkt_port *port;
1335 uint16_t ether_type;
1338 ASSERT_SERIALIZED(ifp->if_serializer);
1341 /* Discard packet if interface is not up */
1342 if (!(ifp->if_flags & IFF_UP)) {
1347 if (m->m_len < sizeof(struct ether_header)) {
1348 /* XXX error in the caller. */
1352 eh = mtod(m, struct ether_header *);
1354 m->m_pkthdr.rcvif = ifp;
1356 if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
1357 if (bcmp(ifp->if_broadcastaddr, eh->ether_dhost,
1358 ifp->if_addrlen) == 0)
1359 m->m_flags |= M_BCAST;
1361 m->m_flags |= M_MCAST;
1365 ETHER_BPF_MTAP(ifp, m);
1367 ifp->if_ibytes += m->m_pkthdr.len;
1369 if (ifp->if_flags & IFF_MONITOR) {
1371 * Interface marked for monitoring; discard packet.
1377 if (ntohs(eh->ether_type) == ETHERTYPE_VLAN &&
1378 (m->m_flags & M_VLANTAG) == 0) {
1380 * Extract vlan tag if hardware does not do it for us
1382 vlan_ether_decap(&m);
1385 eh = mtod(m, struct ether_header *);
1387 ether_type = ntohs(eh->ether_type);
1389 if ((m->m_flags & M_VLANTAG) && ether_type == ETHERTYPE_VLAN) {
1391 * To prevent possible dangerous recursion,
1392 * we don't do vlan-in-vlan
1398 KKASSERT(ether_type != ETHERTYPE_VLAN);
1401 * Map ether type to netisr id.
1403 switch (ether_type) {
1415 case ETHERTYPE_IPV6:
1427 case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */
1434 isr = NETISR_ATALK1;
1436 case ETHERTYPE_AARP:
1442 case ETHERTYPE_MPLS:
1443 case ETHERTYPE_MPLS_MCAST:
1444 m->m_flags |= M_MPLSLABELED;
1451 * NETISR_MAX is an invalid value; it is chosen to let
1452 * ether_mport() know that we are not able to decide
1453 * this packet's msgport here.
1460 * If the packet is in contiguous memory, following
1461 * m_adj() could ensure that the hidden ether header
1462 * will not be destroyed, else we will have to save
1463 * the ether header for the later restoration.
1465 if (m->m_pkthdr.len != m->m_len) {
1467 save_eh = &save_eh0;
1473 * Temporarily remove ether header; ether_mport()
1474 * expects a packet without ether header.
1476 m_adj(m, sizeof(struct ether_header));
1479 * Find the packet's target msgport.
1481 port = ether_mport(isr, &m);
1483 KKASSERT(m == NULL);
1488 * Restore ether header.
1490 if (save_eh != NULL) {
1491 ether_restore_header(&m, eh, save_eh);
1495 m->m_data -= ETHER_HDR_LEN;
1496 m->m_len += ETHER_HDR_LEN;
1497 m->m_pkthdr.len += ETHER_HDR_LEN;
1501 * Initialize mbuf's netmsg packet _after_ possible
1502 * ether header restoration, else the initialized
1503 * netmsg packet may be lost during ether header
1506 ether_init_netpacket(isr, m);
1508 #ifdef ETHER_INPUT_CHAIN
1509 if (chain != NULL) {
1510 struct mbuf_chain *c;
1513 m->m_pkthdr.header = port; /* XXX */
1514 cpuid = port->mpu_td->td_gd->gd_cpuid;
1517 if (c->mc_head == NULL) {
1518 c->mc_head = c->mc_tail = m;
1520 c->mc_tail->m_nextpkt = m;
1523 m->m_nextpkt = NULL;
1525 #endif /* ETHER_INPUT_CHAIN */
1526 lwkt_sendmsg(port, &m->m_hdr.mh_netmsg.nm_netmsg.nm_lmsg);