1 /* $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $ */
4 * SPDX-License-Identifier: BSD-4-Clause
6 * Copyright 2001 Wasabi Systems, Inc.
9 * Written by Jason R. Thorpe for Wasabi Systems, Inc.
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. All advertising materials mentioning features or use of this software
20 * must display the following acknowledgement:
21 * This product includes software developed for the NetBSD Project by
22 * Wasabi Systems, Inc.
23 * 4. The name of Wasabi Systems, Inc. may not be used to endorse
24 * or promote products derived from this software without specific prior
27 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37 * POSSIBILITY OF SUCH DAMAGE.
41 * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
42 * All rights reserved.
44 * Redistribution and use in source and binary forms, with or without
45 * modification, are permitted provided that the following conditions
47 * 1. Redistributions of source code must retain the above copyright
48 * notice, this list of conditions and the following disclaimer.
49 * 2. Redistributions in binary form must reproduce the above copyright
50 * notice, this list of conditions and the following disclaimer in the
51 * documentation and/or other materials provided with the distribution.
53 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
54 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
55 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
56 * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
57 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
58 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
59 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
61 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
62 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
63 * POSSIBILITY OF SUCH DAMAGE.
65 * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
69 * Network interface bridge support.
73 * - Currently only supports Ethernet-like interfaces (Ethernet,
74 * 802.11, VLANs on Ethernet, etc.) Figure out a nice way
75 * to bridge other types of interfaces (maybe consider
76 * heterogeneous bridges).
79 #include <sys/cdefs.h>
80 __FBSDID("$FreeBSD$");
83 #include "opt_inet6.h"
85 #include <sys/param.h>
86 #include <sys/eventhandler.h>
88 #include <sys/malloc.h>
89 #include <sys/protosw.h>
90 #include <sys/systm.h>
93 #include <sys/socket.h> /* for net/if.h */
94 #include <sys/sockio.h>
95 #include <sys/ctype.h> /* string functions */
96 #include <sys/kernel.h>
97 #include <sys/random.h>
98 #include <sys/syslog.h>
99 #include <sys/sysctl.h>
101 #include <sys/module.h>
102 #include <sys/priv.h>
103 #include <sys/proc.h>
104 #include <sys/lock.h>
105 #include <sys/mutex.h>
109 #include <net/if_clone.h>
110 #include <net/if_dl.h>
111 #include <net/if_types.h>
112 #include <net/if_var.h>
113 #include <net/pfil.h>
114 #include <net/vnet.h>
116 #include <netinet/in.h>
117 #include <netinet/in_systm.h>
118 #include <netinet/in_var.h>
119 #include <netinet/ip.h>
120 #include <netinet/ip_var.h>
122 #include <netinet/ip6.h>
123 #include <netinet6/ip6_var.h>
124 #include <netinet6/in6_ifattach.h>
126 #if defined(INET) || defined(INET6)
127 #include <netinet/ip_carp.h>
129 #include <machine/in_cksum.h>
130 #include <netinet/if_ether.h>
131 #include <net/bridgestp.h>
132 #include <net/if_bridgevar.h>
133 #include <net/if_llc.h>
134 #include <net/if_vlan_var.h>
136 #include <net/route.h>
139 * Size of the route hash table. Must be a power of two.
141 #ifndef BRIDGE_RTHASH_SIZE
142 #define BRIDGE_RTHASH_SIZE 1024
145 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1)
148 * Default maximum number of addresses to cache.
150 #ifndef BRIDGE_RTABLE_MAX
151 #define BRIDGE_RTABLE_MAX 2000
155 * Timeout (in seconds) for entries learned dynamically.
157 #ifndef BRIDGE_RTABLE_TIMEOUT
158 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */
162 * Number of seconds between walks of the route list.
164 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
165 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60)
169 * List of capabilities to possibly mask on the member interface.
171 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
175 * List of capabilities to strip
177 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO
180 * Bridge interface list entry.
182 struct bridge_iflist {
183 LIST_ENTRY(bridge_iflist) bif_next;
184 struct ifnet *bif_ifp; /* member if */
185 struct bstp_port bif_stp; /* STP state */
186 uint32_t bif_flags; /* member if flags */
187 int bif_savedcaps; /* saved capabilities */
188 uint32_t bif_addrmax; /* max # of addresses */
189 uint32_t bif_addrcnt; /* cur. # of addresses */
190 uint32_t bif_addrexceeded;/* # of address violations */
196 struct bridge_rtnode {
197 LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
198 LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
199 struct bridge_iflist *brt_dst; /* destination if */
200 unsigned long brt_expire; /* expiration time */
201 uint8_t brt_flags; /* address flags */
202 uint8_t brt_addr[ETHER_ADDR_LEN];
203 uint16_t brt_vlan; /* vlan id */
205 #define brt_ifp brt_dst->bif_ifp
208 * Software state for each bridge.
210 struct bridge_softc {
211 struct ifnet *sc_ifp; /* make this an interface */
212 LIST_ENTRY(bridge_softc) sc_list;
215 uint32_t sc_brtmax; /* max # of addresses */
216 uint32_t sc_brtcnt; /* cur. # of addresses */
217 uint32_t sc_brttimeout; /* rt timeout in seconds */
218 struct callout sc_brcallout; /* bridge callout */
219 uint32_t sc_iflist_ref; /* refcount for sc_iflist */
220 uint32_t sc_iflist_xcnt; /* refcount for sc_iflist */
221 LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
222 LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
223 LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
224 uint32_t sc_rthash_key; /* key for hash */
225 LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
226 struct bstp_state sc_stp; /* STP state */
227 uint32_t sc_brtexceeded; /* # of cache drops */
228 struct ifnet *sc_ifaddr; /* member mac copied from */
229 struct ether_addr sc_defaddr; /* Default MAC address */
232 VNET_DEFINE_STATIC(struct mtx, bridge_list_mtx);
233 #define V_bridge_list_mtx VNET(bridge_list_mtx)
234 static eventhandler_tag bridge_detach_cookie;
236 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
238 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
239 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
241 static int bridge_clone_create(struct if_clone *, int, caddr_t);
242 static void bridge_clone_destroy(struct ifnet *);
244 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
245 static void bridge_mutecaps(struct bridge_softc *);
246 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
248 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
249 static void bridge_init(void *);
250 static void bridge_dummynet(struct mbuf *, struct ifnet *);
251 static void bridge_stop(struct ifnet *, int);
252 static int bridge_transmit(struct ifnet *, struct mbuf *);
253 static void bridge_qflush(struct ifnet *);
254 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
255 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
257 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
259 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
261 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
264 static void bridge_timer(void *);
266 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
268 static void bridge_span(struct bridge_softc *, struct mbuf *);
270 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
271 uint16_t, struct bridge_iflist *, int, uint8_t);
272 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
274 static void bridge_rttrim(struct bridge_softc *);
275 static void bridge_rtage(struct bridge_softc *);
276 static void bridge_rtflush(struct bridge_softc *, int);
277 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
280 static void bridge_rtable_init(struct bridge_softc *);
281 static void bridge_rtable_fini(struct bridge_softc *);
283 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
284 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
285 const uint8_t *, uint16_t);
286 static int bridge_rtnode_insert(struct bridge_softc *,
287 struct bridge_rtnode *);
288 static void bridge_rtnode_destroy(struct bridge_softc *,
289 struct bridge_rtnode *);
290 static void bridge_rtable_expire(struct ifnet *, int);
291 static void bridge_state_change(struct ifnet *, int);
293 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
295 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
297 static void bridge_delete_member(struct bridge_softc *,
298 struct bridge_iflist *, int);
299 static void bridge_delete_span(struct bridge_softc *,
300 struct bridge_iflist *);
302 static int bridge_ioctl_add(struct bridge_softc *, void *);
303 static int bridge_ioctl_del(struct bridge_softc *, void *);
304 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
305 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
306 static int bridge_ioctl_scache(struct bridge_softc *, void *);
307 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
308 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
309 static int bridge_ioctl_rts(struct bridge_softc *, void *);
310 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
311 static int bridge_ioctl_sto(struct bridge_softc *, void *);
312 static int bridge_ioctl_gto(struct bridge_softc *, void *);
313 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
314 static int bridge_ioctl_flush(struct bridge_softc *, void *);
315 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
316 static int bridge_ioctl_spri(struct bridge_softc *, void *);
317 static int bridge_ioctl_ght(struct bridge_softc *, void *);
318 static int bridge_ioctl_sht(struct bridge_softc *, void *);
319 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
320 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
321 static int bridge_ioctl_gma(struct bridge_softc *, void *);
322 static int bridge_ioctl_sma(struct bridge_softc *, void *);
323 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
324 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
325 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
326 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
327 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
328 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
329 static int bridge_ioctl_grte(struct bridge_softc *, void *);
330 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
331 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
332 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
333 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
335 static int bridge_ip_checkbasic(struct mbuf **mp);
337 static int bridge_ip6_checkbasic(struct mbuf **mp);
339 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
340 struct ether_header *, int, struct llc *);
341 static void bridge_linkstate(struct ifnet *ifp);
342 static void bridge_linkcheck(struct bridge_softc *sc);
345 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
346 #define VLANTAGOF(_m) \
347 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
349 static struct bstp_cb_ops bridge_ops = {
350 .bcb_state = bridge_state_change,
351 .bcb_rtage = bridge_rtable_expire
354 SYSCTL_DECL(_net_link);
355 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
357 /* only pass IP[46] packets when pfil is enabled */
358 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
359 #define V_pfil_onlyip VNET(pfil_onlyip)
360 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
361 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
362 "Only pass IP packets when pfil is enabled");
364 /* run pfil hooks on the bridge interface */
365 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
366 #define V_pfil_bridge VNET(pfil_bridge)
367 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
368 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
369 "Packet filter on the bridge interface");
371 /* layer2 filter with ipfw */
372 VNET_DEFINE_STATIC(int, pfil_ipfw);
373 #define V_pfil_ipfw VNET(pfil_ipfw)
375 /* layer2 ARP filter with ipfw */
376 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
377 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
378 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
379 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
380 "Filter ARP packets through IPFW layer2");
382 /* run pfil hooks on the member interface */
383 VNET_DEFINE_STATIC(int, pfil_member) = 1;
384 #define V_pfil_member VNET(pfil_member)
385 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
386 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
387 "Packet filter on the member interface");
389 /* run pfil hooks on the physical interface for locally destined packets */
390 VNET_DEFINE_STATIC(int, pfil_local_phys);
391 #define V_pfil_local_phys VNET(pfil_local_phys)
392 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
393 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
394 "Packet filter on the physical interface for locally destined packets");
396 /* log STP state changes */
397 VNET_DEFINE_STATIC(int, log_stp);
398 #define V_log_stp VNET(log_stp)
399 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
400 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
401 "Log STP state changes");
403 /* share MAC with first bridge member */
404 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
405 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
406 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
407 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
408 "Inherit MAC address from the first bridge member");
410 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
411 #define V_allow_llz_overlap VNET(allow_llz_overlap)
412 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
413 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
414 "Allow overlap of link-local scope "
415 "zones of a bridge interface and the member interfaces");
417 struct bridge_control {
418 int (*bc_func)(struct bridge_softc *, void *);
423 #define BC_F_COPYIN 0x01 /* copy arguments in */
424 #define BC_F_COPYOUT 0x02 /* copy arguments out */
425 #define BC_F_SUSER 0x04 /* do super-user check */
427 const struct bridge_control bridge_control_table[] = {
428 { bridge_ioctl_add, sizeof(struct ifbreq),
429 BC_F_COPYIN|BC_F_SUSER },
430 { bridge_ioctl_del, sizeof(struct ifbreq),
431 BC_F_COPYIN|BC_F_SUSER },
433 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
434 BC_F_COPYIN|BC_F_COPYOUT },
435 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
436 BC_F_COPYIN|BC_F_SUSER },
438 { bridge_ioctl_scache, sizeof(struct ifbrparam),
439 BC_F_COPYIN|BC_F_SUSER },
440 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
443 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
444 BC_F_COPYIN|BC_F_COPYOUT },
445 { bridge_ioctl_rts, sizeof(struct ifbaconf),
446 BC_F_COPYIN|BC_F_COPYOUT },
448 { bridge_ioctl_saddr, sizeof(struct ifbareq),
449 BC_F_COPYIN|BC_F_SUSER },
451 { bridge_ioctl_sto, sizeof(struct ifbrparam),
452 BC_F_COPYIN|BC_F_SUSER },
453 { bridge_ioctl_gto, sizeof(struct ifbrparam),
456 { bridge_ioctl_daddr, sizeof(struct ifbareq),
457 BC_F_COPYIN|BC_F_SUSER },
459 { bridge_ioctl_flush, sizeof(struct ifbreq),
460 BC_F_COPYIN|BC_F_SUSER },
462 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
464 { bridge_ioctl_spri, sizeof(struct ifbrparam),
465 BC_F_COPYIN|BC_F_SUSER },
467 { bridge_ioctl_ght, sizeof(struct ifbrparam),
469 { bridge_ioctl_sht, sizeof(struct ifbrparam),
470 BC_F_COPYIN|BC_F_SUSER },
472 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
474 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
475 BC_F_COPYIN|BC_F_SUSER },
477 { bridge_ioctl_gma, sizeof(struct ifbrparam),
479 { bridge_ioctl_sma, sizeof(struct ifbrparam),
480 BC_F_COPYIN|BC_F_SUSER },
482 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
483 BC_F_COPYIN|BC_F_SUSER },
485 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
486 BC_F_COPYIN|BC_F_SUSER },
488 { bridge_ioctl_addspan, sizeof(struct ifbreq),
489 BC_F_COPYIN|BC_F_SUSER },
490 { bridge_ioctl_delspan, sizeof(struct ifbreq),
491 BC_F_COPYIN|BC_F_SUSER },
493 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
496 { bridge_ioctl_grte, sizeof(struct ifbrparam),
499 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
500 BC_F_COPYIN|BC_F_COPYOUT },
502 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
503 BC_F_COPYIN|BC_F_SUSER },
505 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
506 BC_F_COPYIN|BC_F_SUSER },
508 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
509 BC_F_COPYIN|BC_F_SUSER },
512 const int bridge_control_table_size = nitems(bridge_control_table);
514 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
515 #define V_bridge_list VNET(bridge_list)
516 #define BRIDGE_LIST_LOCK_INIT(x) mtx_init(&V_bridge_list_mtx, \
517 "if_bridge list", NULL, MTX_DEF)
518 #define BRIDGE_LIST_LOCK_DESTROY(x) mtx_destroy(&V_bridge_list_mtx)
519 #define BRIDGE_LIST_LOCK(x) mtx_lock(&V_bridge_list_mtx)
520 #define BRIDGE_LIST_UNLOCK(x) mtx_unlock(&V_bridge_list_mtx)
522 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
523 #define V_bridge_cloner VNET(bridge_cloner)
525 static const char bridge_name[] = "bridge";
528 vnet_bridge_init(const void *unused __unused)
531 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
532 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
534 BRIDGE_LIST_LOCK_INIT();
535 LIST_INIT(&V_bridge_list);
536 V_bridge_cloner = if_clone_simple(bridge_name,
537 bridge_clone_create, bridge_clone_destroy, 0);
539 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
540 vnet_bridge_init, NULL);
543 vnet_bridge_uninit(const void *unused __unused)
546 if_clone_detach(V_bridge_cloner);
547 V_bridge_cloner = NULL;
548 BRIDGE_LIST_LOCK_DESTROY();
549 uma_zdestroy(V_bridge_rtnode_zone);
551 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
552 vnet_bridge_uninit, NULL);
555 bridge_modevent(module_t mod, int type, void *data)
560 bridge_dn_p = bridge_dummynet;
561 bridge_detach_cookie = EVENTHANDLER_REGISTER(
562 ifnet_departure_event, bridge_ifdetach, NULL,
563 EVENTHANDLER_PRI_ANY);
566 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
567 bridge_detach_cookie);
576 static moduledata_t bridge_mod = {
582 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
583 MODULE_VERSION(if_bridge, 1);
584 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
587 * handler for net.link.bridge.ipfw
590 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
592 int enable = V_pfil_ipfw;
595 error = sysctl_handle_int(oidp, &enable, 0, req);
598 if (enable != V_pfil_ipfw) {
599 V_pfil_ipfw = enable;
602 * Disable pfil so that ipfw doesnt run twice, if the user
603 * really wants both then they can re-enable pfil_bridge and/or
604 * pfil_member. Also allow non-ip packets as ipfw can filter by
616 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
617 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
618 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
619 "Layer2 filter with IPFW");
622 * bridge_clone_create:
624 * Create a new bridge instance.
627 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
629 struct bridge_softc *sc, *sc2;
630 struct ifnet *bifp, *ifp;
632 unsigned long hostid;
634 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
635 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
641 BRIDGE_LOCK_INIT(sc);
642 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
643 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
645 /* Initialize our routing table. */
646 bridge_rtable_init(sc);
648 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
650 LIST_INIT(&sc->sc_iflist);
651 LIST_INIT(&sc->sc_spanlist);
654 if_initname(ifp, bridge_name, unit);
655 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
656 ifp->if_ioctl = bridge_ioctl;
657 ifp->if_transmit = bridge_transmit;
658 ifp->if_qflush = bridge_qflush;
659 ifp->if_init = bridge_init;
660 ifp->if_type = IFT_BRIDGE;
663 * Generate an ethernet address with a locally administered address.
665 * Since we are using random ethernet addresses for the bridge, it is
666 * possible that we might have address collisions, so make sure that
667 * this hardware address isn't already in use on another bridge.
668 * The first try uses the hostid and falls back to arc4rand().
671 getcredhostid(curthread->td_ucred, &hostid);
673 if (fb || hostid == 0) {
674 ether_gen_addr(ifp, &sc->sc_defaddr);
676 sc->sc_defaddr.octet[0] = 0x2;
677 sc->sc_defaddr.octet[1] = (hostid >> 24) & 0xff;
678 sc->sc_defaddr.octet[2] = (hostid >> 16) & 0xff;
679 sc->sc_defaddr.octet[3] = (hostid >> 8 ) & 0xff;
680 sc->sc_defaddr.octet[4] = hostid & 0xff;
681 sc->sc_defaddr.octet[5] = ifp->if_dunit & 0xff;
687 LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
689 if (memcmp(sc->sc_defaddr.octet,
690 IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
695 BRIDGE_LIST_UNLOCK();
696 } while (retry == 1);
698 bstp_attach(&sc->sc_stp, &bridge_ops);
699 ether_ifattach(ifp, sc->sc_defaddr.octet);
700 /* Now undo some of the damage... */
701 ifp->if_baudrate = 0;
702 ifp->if_type = IFT_BRIDGE;
705 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
706 BRIDGE_LIST_UNLOCK();
712 * bridge_clone_destroy:
714 * Destroy a bridge instance.
717 bridge_clone_destroy(struct ifnet *ifp)
719 struct bridge_softc *sc = ifp->if_softc;
720 struct bridge_iflist *bif;
725 ifp->if_flags &= ~IFF_UP;
727 while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
728 bridge_delete_member(sc, bif, 0);
730 while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
731 bridge_delete_span(sc, bif);
734 /* Tear down the routing table. */
735 bridge_rtable_fini(sc);
739 callout_drain(&sc->sc_brcallout);
742 LIST_REMOVE(sc, sc_list);
743 BRIDGE_LIST_UNLOCK();
745 bstp_detach(&sc->sc_stp);
749 BRIDGE_LOCK_DESTROY(sc);
756 * Handle a control request from the operator.
759 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
761 struct bridge_softc *sc = ifp->if_softc;
762 struct ifreq *ifr = (struct ifreq *)data;
763 struct bridge_iflist *bif;
764 struct thread *td = curthread;
766 struct ifbreq ifbreq;
767 struct ifbifconf ifbifconf;
768 struct ifbareq ifbareq;
769 struct ifbaconf ifbaconf;
770 struct ifbrparam ifbrparam;
771 struct ifbropreq ifbropreq;
773 struct ifdrv *ifd = (struct ifdrv *) data;
774 const struct bridge_control *bc;
785 if (ifd->ifd_cmd >= bridge_control_table_size) {
789 bc = &bridge_control_table[ifd->ifd_cmd];
791 if (cmd == SIOCGDRVSPEC &&
792 (bc->bc_flags & BC_F_COPYOUT) == 0) {
796 else if (cmd == SIOCSDRVSPEC &&
797 (bc->bc_flags & BC_F_COPYOUT) != 0) {
802 if (bc->bc_flags & BC_F_SUSER) {
803 error = priv_check(td, PRIV_NET_BRIDGE);
808 if (ifd->ifd_len != bc->bc_argsize ||
809 ifd->ifd_len > sizeof(args)) {
814 bzero(&args, sizeof(args));
815 if (bc->bc_flags & BC_F_COPYIN) {
816 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
822 error = (*bc->bc_func)(sc, &args);
827 if (bc->bc_flags & BC_F_COPYOUT)
828 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
833 if (!(ifp->if_flags & IFF_UP) &&
834 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
836 * If interface is marked down and it is running,
837 * then stop and disable it.
842 } else if ((ifp->if_flags & IFF_UP) &&
843 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
845 * If interface is marked up and it is stopped, then
853 if (ifr->ifr_mtu < 576) {
857 if (LIST_EMPTY(&sc->sc_iflist)) {
858 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
862 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
863 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
864 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
865 " != %d\n", sc->sc_ifp->if_xname,
866 bif->bif_ifp->if_mtu,
867 bif->bif_ifp->if_xname, ifr->ifr_mtu);
873 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
878 * drop the lock as ether_ioctl() will call bridge_start() and
879 * cause the lock to be recursed.
881 error = ether_ioctl(ifp, cmd, data);
891 * Clear or restore unwanted capabilities on the member interface
894 bridge_mutecaps(struct bridge_softc *sc)
896 struct bridge_iflist *bif;
899 /* Initial bitmask of capabilities to test */
900 mask = BRIDGE_IFCAPS_MASK;
902 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
903 /* Every member must support it or its disabled */
904 mask &= bif->bif_savedcaps;
908 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
909 enabled = bif->bif_ifp->if_capenable;
910 enabled &= ~BRIDGE_IFCAPS_STRIP;
911 /* strip off mask bits and enable them again if allowed */
912 enabled &= ~BRIDGE_IFCAPS_MASK;
915 bridge_set_ifcap(sc, bif, enabled);
923 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
925 struct ifnet *ifp = bif->bif_ifp;
927 int error, mask, stuck;
929 BRIDGE_UNLOCK_ASSERT(sc);
931 bzero(&ifr, sizeof(ifr));
932 ifr.ifr_reqcap = set;
934 if (ifp->if_capenable != set) {
935 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
937 if_printf(sc->sc_ifp,
938 "error setting capabilities on %s: %d\n",
939 ifp->if_xname, error);
940 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
941 stuck = ifp->if_capenable & mask & ~set;
943 if_printf(sc->sc_ifp,
944 "can't disable some capabilities on %s: 0x%x\n",
945 ifp->if_xname, stuck);
950 * bridge_lookup_member:
952 * Lookup a bridge member interface.
954 static struct bridge_iflist *
955 bridge_lookup_member(struct bridge_softc *sc, const char *name)
957 struct bridge_iflist *bif;
960 BRIDGE_LOCK_ASSERT(sc);
962 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
964 if (strcmp(ifp->if_xname, name) == 0)
972 * bridge_lookup_member_if:
974 * Lookup a bridge member interface by ifnet*.
976 static struct bridge_iflist *
977 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
979 struct bridge_iflist *bif;
981 BRIDGE_LOCK_ASSERT(sc);
983 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
984 if (bif->bif_ifp == member_ifp)
992 * bridge_delete_member:
994 * Delete the specified member interface.
997 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1000 struct ifnet *ifs = bif->bif_ifp;
1001 struct ifnet *fif = NULL;
1003 BRIDGE_LOCK_ASSERT(sc);
1005 if (bif->bif_flags & IFBIF_STP)
1006 bstp_disable(&bif->bif_stp);
1008 ifs->if_bridge = NULL;
1010 LIST_REMOVE(bif, bif_next);
1014 * If removing the interface that gave the bridge its mac address, set
1015 * the mac address of the bridge to the address of the next member, or
1016 * to its default address if no members are left.
1018 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1019 if (LIST_EMPTY(&sc->sc_iflist)) {
1020 bcopy(&sc->sc_defaddr,
1021 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1022 sc->sc_ifaddr = NULL;
1024 fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
1025 bcopy(IF_LLADDR(fif),
1026 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1027 sc->sc_ifaddr = fif;
1029 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1032 bridge_linkcheck(sc);
1033 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1034 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1035 KASSERT(bif->bif_addrcnt == 0,
1036 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1038 ifs->if_bridge_output = NULL;
1039 ifs->if_bridge_input = NULL;
1040 ifs->if_bridge_linkstate = NULL;
1043 switch (ifs->if_type) {
1047 * Take the interface out of promiscuous mode, but only
1048 * if it was promiscuous in the first place. It might
1049 * not be if we're in the bridge_ioctl_add() error path.
1051 if (ifs->if_flags & IFF_PROMISC)
1052 (void) ifpromisc(ifs, 0);
1060 panic("bridge_delete_member: impossible");
1064 /* reneable any interface capabilities */
1065 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1067 bstp_destroy(&bif->bif_stp); /* prepare to free */
1069 free(bif, M_DEVBUF);
1073 * bridge_delete_span:
1075 * Delete the specified span interface.
1078 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1080 BRIDGE_LOCK_ASSERT(sc);
1082 KASSERT(bif->bif_ifp->if_bridge == NULL,
1083 ("%s: not a span interface", __func__));
1085 LIST_REMOVE(bif, bif_next);
1086 free(bif, M_DEVBUF);
1090 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1092 struct ifbreq *req = arg;
1093 struct bridge_iflist *bif = NULL;
1097 ifs = ifunit(req->ifbr_ifsname);
1100 if (ifs->if_ioctl == NULL) /* must be supported */
1103 /* If it's in the span list, it can't be a member. */
1104 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1105 if (ifs == bif->bif_ifp)
1108 if (ifs->if_bridge == sc)
1111 if (ifs->if_bridge != NULL)
1114 switch (ifs->if_type) {
1118 /* permitted interface types */
1126 * Two valid inet6 addresses with link-local scope must not be
1127 * on the parent interface and the member interfaces at the
1128 * same time. This restriction is needed to prevent violation
1129 * of link-local scope zone. Attempts to add a member
1130 * interface which has inet6 addresses when the parent has
1131 * inet6 triggers removal of all inet6 addresses on the member
1135 /* Check if the parent interface has a link-local scope addr. */
1136 if (V_allow_llz_overlap == 0 &&
1137 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1139 * If any, remove all inet6 addresses from the member
1143 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1144 if (in6ifa_llaonifp(bif->bif_ifp)) {
1146 in6_ifdetach(bif->bif_ifp);
1148 if_printf(sc->sc_ifp,
1149 "IPv6 addresses on %s have been removed "
1150 "before adding it as a member to prevent "
1151 "IPv6 address scope violation.\n",
1152 bif->bif_ifp->if_xname);
1156 if (in6ifa_llaonifp(ifs)) {
1160 if_printf(sc->sc_ifp,
1161 "IPv6 addresses on %s have been removed "
1162 "before adding it as a member to prevent "
1163 "IPv6 address scope violation.\n",
1168 /* Allow the first Ethernet member to define the MTU */
1169 if (LIST_EMPTY(&sc->sc_iflist))
1170 sc->sc_ifp->if_mtu = ifs->if_mtu;
1171 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1172 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1173 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1177 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1182 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1183 bif->bif_savedcaps = ifs->if_capenable;
1186 * Assign the interface's MAC address to the bridge if it's the first
1187 * member and the MAC address of the bridge has not been changed from
1188 * the default randomly generated one.
1190 if (V_bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1191 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1192 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1193 sc->sc_ifaddr = ifs;
1194 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1197 ifs->if_bridge = sc;
1198 ifs->if_bridge_output = bridge_output;
1199 ifs->if_bridge_input = bridge_input;
1200 ifs->if_bridge_linkstate = bridge_linkstate;
1201 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1203 * XXX: XLOCK HERE!?!
1205 * NOTE: insert_***HEAD*** should be safe for the traversals.
1207 LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1209 /* Set interface capabilities to the intersection set of all members */
1210 bridge_mutecaps(sc);
1211 bridge_linkcheck(sc);
1213 /* Place the interface into promiscuous mode */
1214 switch (ifs->if_type) {
1218 error = ifpromisc(ifs, 1);
1224 bridge_delete_member(sc, bif, 0);
1229 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1231 struct ifbreq *req = arg;
1232 struct bridge_iflist *bif;
1234 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1238 bridge_delete_member(sc, bif, 0);
1244 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1246 struct ifbreq *req = arg;
1247 struct bridge_iflist *bif;
1248 struct bstp_port *bp;
1250 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1255 req->ifbr_ifsflags = bif->bif_flags;
1256 req->ifbr_state = bp->bp_state;
1257 req->ifbr_priority = bp->bp_priority;
1258 req->ifbr_path_cost = bp->bp_path_cost;
1259 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1260 req->ifbr_proto = bp->bp_protover;
1261 req->ifbr_role = bp->bp_role;
1262 req->ifbr_stpflags = bp->bp_flags;
1263 req->ifbr_addrcnt = bif->bif_addrcnt;
1264 req->ifbr_addrmax = bif->bif_addrmax;
1265 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1267 /* Copy STP state options as flags */
1268 if (bp->bp_operedge)
1269 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1270 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1271 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1272 if (bp->bp_ptp_link)
1273 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1274 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1275 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1276 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1277 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1278 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1279 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1284 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1286 struct ifbreq *req = arg;
1287 struct bridge_iflist *bif;
1288 struct bstp_port *bp;
1291 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1296 if (req->ifbr_ifsflags & IFBIF_SPAN)
1297 /* SPAN is readonly */
1300 if (req->ifbr_ifsflags & IFBIF_STP) {
1301 if ((bif->bif_flags & IFBIF_STP) == 0) {
1302 error = bstp_enable(&bif->bif_stp);
1307 if ((bif->bif_flags & IFBIF_STP) != 0)
1308 bstp_disable(&bif->bif_stp);
1311 /* Pass on STP flags */
1312 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1313 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1314 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1315 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1317 /* Save the bits relating to the bridge */
1318 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1324 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1326 struct ifbrparam *param = arg;
1328 sc->sc_brtmax = param->ifbrp_csize;
1335 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1337 struct ifbrparam *param = arg;
1339 param->ifbrp_csize = sc->sc_brtmax;
1345 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1347 struct ifbifconf *bifc = arg;
1348 struct bridge_iflist *bif;
1351 int count, buflen, len, error = 0;
1354 LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1356 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1359 buflen = sizeof(breq) * count;
1360 if (bifc->ifbic_len == 0) {
1361 bifc->ifbic_len = buflen;
1365 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1370 len = min(bifc->ifbic_len, buflen);
1371 bzero(&breq, sizeof(breq));
1372 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1373 if (len < sizeof(breq))
1376 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1377 sizeof(breq.ifbr_ifsname));
1378 /* Fill in the ifbreq structure */
1379 error = bridge_ioctl_gifflags(sc, &breq);
1382 memcpy(buf, &breq, sizeof(breq));
1384 buf += sizeof(breq);
1385 len -= sizeof(breq);
1387 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1388 if (len < sizeof(breq))
1391 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1392 sizeof(breq.ifbr_ifsname));
1393 breq.ifbr_ifsflags = bif->bif_flags;
1394 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1395 memcpy(buf, &breq, sizeof(breq));
1397 buf += sizeof(breq);
1398 len -= sizeof(breq);
1402 bifc->ifbic_len = sizeof(breq) * count;
1403 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1405 free(outbuf, M_TEMP);
1410 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1412 struct ifbaconf *bac = arg;
1413 struct bridge_rtnode *brt;
1414 struct ifbareq bareq;
1416 int count, buflen, len, error = 0;
1418 if (bac->ifbac_len == 0)
1422 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1424 buflen = sizeof(bareq) * count;
1427 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1432 len = min(bac->ifbac_len, buflen);
1433 bzero(&bareq, sizeof(bareq));
1434 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1435 if (len < sizeof(bareq))
1437 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1438 sizeof(bareq.ifba_ifsname));
1439 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1440 bareq.ifba_vlan = brt->brt_vlan;
1441 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1442 time_uptime < brt->brt_expire)
1443 bareq.ifba_expire = brt->brt_expire - time_uptime;
1445 bareq.ifba_expire = 0;
1446 bareq.ifba_flags = brt->brt_flags;
1448 memcpy(buf, &bareq, sizeof(bareq));
1450 buf += sizeof(bareq);
1451 len -= sizeof(bareq);
1455 bac->ifbac_len = sizeof(bareq) * count;
1456 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1458 free(outbuf, M_TEMP);
1463 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1465 struct ifbareq *req = arg;
1466 struct bridge_iflist *bif;
1469 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1473 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1480 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1482 struct ifbrparam *param = arg;
1484 sc->sc_brttimeout = param->ifbrp_ctime;
1489 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1491 struct ifbrparam *param = arg;
1493 param->ifbrp_ctime = sc->sc_brttimeout;
1498 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1500 struct ifbareq *req = arg;
1502 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1506 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1508 struct ifbreq *req = arg;
1510 bridge_rtflush(sc, req->ifbr_ifsflags);
1515 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1517 struct ifbrparam *param = arg;
1518 struct bstp_state *bs = &sc->sc_stp;
1520 param->ifbrp_prio = bs->bs_bridge_priority;
1525 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1527 struct ifbrparam *param = arg;
1529 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1533 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1535 struct ifbrparam *param = arg;
1536 struct bstp_state *bs = &sc->sc_stp;
1538 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1543 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1545 struct ifbrparam *param = arg;
1547 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1551 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1553 struct ifbrparam *param = arg;
1554 struct bstp_state *bs = &sc->sc_stp;
1556 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1561 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1563 struct ifbrparam *param = arg;
1565 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1569 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1571 struct ifbrparam *param = arg;
1572 struct bstp_state *bs = &sc->sc_stp;
1574 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1579 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1581 struct ifbrparam *param = arg;
1583 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1587 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1589 struct ifbreq *req = arg;
1590 struct bridge_iflist *bif;
1592 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1596 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1600 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1602 struct ifbreq *req = arg;
1603 struct bridge_iflist *bif;
1605 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1609 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1613 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1615 struct ifbreq *req = arg;
1616 struct bridge_iflist *bif;
1618 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1622 bif->bif_addrmax = req->ifbr_addrmax;
1627 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1629 struct ifbreq *req = arg;
1630 struct bridge_iflist *bif = NULL;
1633 ifs = ifunit(req->ifbr_ifsname);
1637 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1638 if (ifs == bif->bif_ifp)
1641 if (ifs->if_bridge != NULL)
1644 switch (ifs->if_type) {
1653 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1658 bif->bif_flags = IFBIF_SPAN;
1660 LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1666 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1668 struct ifbreq *req = arg;
1669 struct bridge_iflist *bif;
1672 ifs = ifunit(req->ifbr_ifsname);
1676 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1677 if (ifs == bif->bif_ifp)
1683 bridge_delete_span(sc, bif);
1689 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1691 struct ifbropreq *req = arg;
1692 struct bstp_state *bs = &sc->sc_stp;
1693 struct bstp_port *root_port;
1695 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1696 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1697 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1699 root_port = bs->bs_root_port;
1700 if (root_port == NULL)
1701 req->ifbop_root_port = 0;
1703 req->ifbop_root_port = root_port->bp_ifp->if_index;
1705 req->ifbop_holdcount = bs->bs_txholdcount;
1706 req->ifbop_priority = bs->bs_bridge_priority;
1707 req->ifbop_protocol = bs->bs_protover;
1708 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1709 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1710 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1711 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1712 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1713 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1719 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1721 struct ifbrparam *param = arg;
1723 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1728 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1730 struct ifbpstpconf *bifstp = arg;
1731 struct bridge_iflist *bif;
1732 struct bstp_port *bp;
1733 struct ifbpstpreq bpreq;
1735 int count, buflen, len, error = 0;
1738 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1739 if ((bif->bif_flags & IFBIF_STP) != 0)
1743 buflen = sizeof(bpreq) * count;
1744 if (bifstp->ifbpstp_len == 0) {
1745 bifstp->ifbpstp_len = buflen;
1750 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1755 len = min(bifstp->ifbpstp_len, buflen);
1756 bzero(&bpreq, sizeof(bpreq));
1757 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1758 if (len < sizeof(bpreq))
1761 if ((bif->bif_flags & IFBIF_STP) == 0)
1765 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1766 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1767 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1768 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1769 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1770 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1772 memcpy(buf, &bpreq, sizeof(bpreq));
1774 buf += sizeof(bpreq);
1775 len -= sizeof(bpreq);
1779 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1780 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1782 free(outbuf, M_TEMP);
1787 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1789 struct ifbrparam *param = arg;
1791 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1795 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1797 struct ifbrparam *param = arg;
1799 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1805 * Detach an interface from a bridge. Called when a member
1806 * interface is detaching.
1809 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1811 struct bridge_softc *sc = ifp->if_bridge;
1812 struct bridge_iflist *bif;
1814 if (ifp->if_flags & IFF_RENAMING)
1816 if (V_bridge_cloner == NULL) {
1818 * This detach handler can be called after
1819 * vnet_bridge_uninit(). Just return in that case.
1823 /* Check if the interface is a bridge member */
1827 bif = bridge_lookup_member_if(sc, ifp);
1829 bridge_delete_member(sc, bif, 1);
1835 /* Check if the interface is a span port */
1837 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1839 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1840 if (ifp == bif->bif_ifp) {
1841 bridge_delete_span(sc, bif);
1847 BRIDGE_LIST_UNLOCK();
1853 * Initialize a bridge interface.
1856 bridge_init(void *xsc)
1858 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1859 struct ifnet *ifp = sc->sc_ifp;
1861 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1865 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1868 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1869 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1877 * Stop the bridge interface.
1880 bridge_stop(struct ifnet *ifp, int disable)
1882 struct bridge_softc *sc = ifp->if_softc;
1884 BRIDGE_LOCK_ASSERT(sc);
1886 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1889 callout_stop(&sc->sc_brcallout);
1890 bstp_stop(&sc->sc_stp);
1892 bridge_rtflush(sc, IFBF_FLUSHDYN);
1894 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1900 * Enqueue a packet on a bridge member interface.
1904 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1910 /* We may be sending a fragment so traverse the mbuf */
1913 m->m_nextpkt = NULL;
1914 len = m->m_pkthdr.len;
1915 mflags = m->m_flags;
1918 * If underlying interface can not do VLAN tag insertion itself
1919 * then attach a packet tag that holds it.
1921 if ((m->m_flags & M_VLANTAG) &&
1922 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1923 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1926 "unable to prepend VLAN header\n");
1927 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1930 m->m_flags &= ~M_VLANTAG;
1933 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1934 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1936 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1940 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1941 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1942 if (mflags & M_MCAST)
1943 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
1952 * Receive a queued packet from dummynet and pass it on to the output
1955 * The mbuf has the Ethernet header already attached.
1958 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1960 struct bridge_softc *sc;
1962 sc = ifp->if_bridge;
1965 * The packet didnt originate from a member interface. This should only
1966 * ever happen if a member interface is removed while packets are
1974 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
1976 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
1979 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1985 bridge_enqueue(sc, ifp, m);
1991 * Send output from a bridge member interface. This
1992 * performs the bridging function for locally originated
1995 * The mbuf has the Ethernet header already attached. We must
1996 * enqueue or free the mbuf before returning.
1999 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2002 struct ether_header *eh;
2003 struct ifnet *dst_if;
2004 struct bridge_softc *sc;
2007 if (m->m_len < ETHER_HDR_LEN) {
2008 m = m_pullup(m, ETHER_HDR_LEN);
2013 eh = mtod(m, struct ether_header *);
2014 sc = ifp->if_bridge;
2015 vlan = VLANTAGOF(m);
2020 * If bridge is down, but the original output interface is up,
2021 * go ahead and send out that interface. Otherwise, the packet
2024 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2030 * If the packet is a multicast, or we don't know a better way to
2031 * get there, send to all interfaces.
2033 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2036 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2037 if (dst_if == NULL) {
2038 struct bridge_iflist *bif;
2040 int error = 0, used = 0;
2044 BRIDGE_LOCK2REF(sc, error);
2050 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2051 dst_if = bif->bif_ifp;
2053 if (dst_if->if_type == IFT_GIF)
2055 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2059 * If this is not the original output interface,
2060 * and the interface is participating in spanning
2061 * tree, make sure the port is in a state that
2062 * allows forwarding.
2064 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2065 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2068 if (LIST_NEXT(bif, bif_next) == NULL) {
2072 mc = m_copypacket(m, M_NOWAIT);
2074 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2079 bridge_enqueue(sc, dst_if, mc);
2089 * XXX Spanning tree consideration here?
2093 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2100 bridge_enqueue(sc, dst_if, m);
2107 * Do output on a bridge.
2111 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2113 struct bridge_softc *sc;
2114 struct ether_header *eh;
2115 struct ifnet *dst_if;
2120 ETHER_BPF_MTAP(ifp, m);
2122 eh = mtod(m, struct ether_header *);
2125 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2126 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2128 error = bridge_enqueue(sc, dst_if, m);
2130 bridge_broadcast(sc, ifp, m, 0);
2136 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2139 bridge_qflush(struct ifnet *ifp __unused)
2146 * The forwarding function of the bridge.
2148 * NOTE: Releases the lock on return.
2151 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2154 struct bridge_iflist *dbif;
2155 struct ifnet *src_if, *dst_if, *ifp;
2156 struct ether_header *eh;
2161 src_if = m->m_pkthdr.rcvif;
2164 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2165 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2166 vlan = VLANTAGOF(m);
2168 if ((sbif->bif_flags & IFBIF_STP) &&
2169 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2172 eh = mtod(m, struct ether_header *);
2173 dst = eh->ether_dhost;
2175 /* If the interface is learning, record the address. */
2176 if (sbif->bif_flags & IFBIF_LEARNING) {
2177 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2178 sbif, 0, IFBAF_DYNAMIC);
2180 * If the interface has addresses limits then deny any source
2181 * that is not in the cache.
2183 if (error && sbif->bif_addrmax)
2187 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2188 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2192 * At this point, the port either doesn't participate
2193 * in spanning tree or it is in the forwarding state.
2197 * If the packet is unicast, destined for someone on
2198 * "this" side of the bridge, drop it.
2200 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2201 dst_if = bridge_rtlookup(sc, dst, vlan);
2202 if (src_if == dst_if)
2206 * Check if its a reserved multicast address, any address
2207 * listed in 802.1D section 7.12.6 may not be forwarded by the
2209 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2211 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2212 dst[2] == 0xc2 && dst[3] == 0x00 &&
2213 dst[4] == 0x00 && dst[5] <= 0x0f)
2216 /* ...forward it to all interfaces. */
2217 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2222 * If we have a destination interface which is a member of our bridge,
2223 * OR this is a unicast packet, push it through the bpf(4) machinery.
2224 * For broadcast or multicast packets, don't bother because it will
2225 * be reinjected into ether_input. We do this before we pass the packets
2226 * through the pfil(9) framework, as it is possible that pfil(9) will
2227 * drop the packet, or possibly modify it, making it difficult to debug
2228 * firewall issues on the bridge.
2230 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2231 ETHER_BPF_MTAP(ifp, m);
2233 /* run the packet filter */
2234 if (PFIL_HOOKED_IN(V_inet_pfil_head)
2236 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2240 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2247 if (dst_if == NULL) {
2248 bridge_broadcast(sc, src_if, m, 1);
2253 * At this point, we're dealing with a unicast frame
2254 * going to a different interface.
2256 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2259 dbif = bridge_lookup_member_if(sc, dst_if);
2261 /* Not a member of the bridge (anymore?) */
2264 /* Private segments can not talk to each other */
2265 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2268 if ((dbif->bif_flags & IFBIF_STP) &&
2269 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2274 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2276 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2279 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2285 bridge_enqueue(sc, dst_if, m);
2296 * Receive input from a member interface. Queue the packet for
2297 * bridging if it is not for us.
2299 static struct mbuf *
2300 bridge_input(struct ifnet *ifp, struct mbuf *m)
2302 struct bridge_softc *sc = ifp->if_bridge;
2303 struct bridge_iflist *bif, *bif2;
2305 struct ether_header *eh;
2306 struct mbuf *mc, *mc2;
2310 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2314 vlan = VLANTAGOF(m);
2317 * Implement support for bridge monitoring. If this flag has been
2318 * set on this interface, discard the packet once we push it through
2319 * the bpf(4) machinery, but before we do, increment the byte and
2320 * packet counters associated with this interface.
2322 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2323 m->m_pkthdr.rcvif = bifp;
2324 ETHER_BPF_MTAP(bifp, m);
2325 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2326 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2331 bif = bridge_lookup_member_if(sc, ifp);
2337 eh = mtod(m, struct ether_header *);
2341 if (m->m_flags & (M_BCAST|M_MCAST)) {
2342 /* Tap off 802.1D packets; they do not get forwarded. */
2343 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2344 ETHER_ADDR_LEN) == 0) {
2345 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2350 if ((bif->bif_flags & IFBIF_STP) &&
2351 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2357 * Make a deep copy of the packet and enqueue the copy
2358 * for bridge processing; return the original packet for
2361 mc = m_dup(m, M_NOWAIT);
2367 /* Perform the bridge forwarding function with the copy. */
2368 bridge_forward(sc, bif, mc);
2371 * Reinject the mbuf as arriving on the bridge so we have a
2372 * chance at claiming multicast packets. We can not loop back
2373 * here from ether_input as a bridge is never a member of a
2376 KASSERT(bifp->if_bridge == NULL,
2377 ("loop created in bridge_input"));
2378 mc2 = m_dup(m, M_NOWAIT);
2380 /* Keep the layer3 header aligned */
2381 int i = min(mc2->m_pkthdr.len, max_protohdr);
2382 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2385 mc2->m_pkthdr.rcvif = bifp;
2386 (*bifp->if_input)(bifp, mc2);
2389 /* Return the original packet for local processing. */
2393 if ((bif->bif_flags & IFBIF_STP) &&
2394 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2399 #if (defined(INET) || defined(INET6))
2400 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2401 || ((iface)->if_carp \
2402 && (*carp_forus_p)((iface), eh->ether_dhost))
2403 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2404 || ((iface)->if_carp \
2405 && (*carp_forus_p)((iface), eh->ether_shost))
2407 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2408 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2412 # define OR_PFIL_HOOKED_INET6 \
2413 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2415 # define OR_PFIL_HOOKED_INET6
2418 #define GRAB_OUR_PACKETS(iface) \
2419 if ((iface)->if_type == IFT_GIF) \
2421 /* It is destined for us. */ \
2422 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2423 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2425 if (bif->bif_flags & IFBIF_LEARNING) { \
2426 error = bridge_rtupdate(sc, eh->ether_shost, \
2427 vlan, bif, 0, IFBAF_DYNAMIC); \
2428 if (error && bif->bif_addrmax) { \
2429 BRIDGE_UNLOCK(sc); \
2434 m->m_pkthdr.rcvif = iface; \
2435 if ((iface) == ifp) { \
2436 /* Skip bridge processing... src == dest */ \
2437 BRIDGE_UNLOCK(sc); \
2440 /* It's passing over or to the bridge, locally. */ \
2441 ETHER_BPF_MTAP(bifp, m); \
2442 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \
2443 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2444 /* Filter on the physical interface. */ \
2445 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2446 OR_PFIL_HOOKED_INET6)) { \
2447 if (bridge_pfil(&m, NULL, ifp, \
2448 PFIL_IN) != 0 || m == NULL) { \
2449 BRIDGE_UNLOCK(sc); \
2453 BRIDGE_UNLOCK(sc); \
2457 /* We just received a packet that we sent out. */ \
2458 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2459 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2461 BRIDGE_UNLOCK(sc); \
2467 * Unicast. Make sure it's not for the bridge.
2469 do { GRAB_OUR_PACKETS(bifp) } while (0);
2472 * Give a chance for ifp at first priority. This will help when the
2473 * packet comes through the interface like VLAN's with the same MACs
2474 * on several interfaces from the same bridge. This also will save
2475 * some CPU cycles in case the destination interface and the input
2476 * interface (eq ifp) are the same.
2478 do { GRAB_OUR_PACKETS(ifp) } while (0);
2480 /* Now check the all bridge members. */
2481 LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2482 GRAB_OUR_PACKETS(bif2->bif_ifp)
2485 #undef OR_CARP_CHECK_WE_ARE_DST
2486 #undef OR_CARP_CHECK_WE_ARE_SRC
2487 #undef OR_PFIL_HOOKED_INET6
2488 #undef GRAB_OUR_PACKETS
2490 /* Perform the bridge forwarding function. */
2491 bridge_forward(sc, bif, m);
2499 * Send a frame to all interfaces that are members of
2500 * the bridge, except for the one on which the packet
2503 * NOTE: Releases the lock on return.
2506 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2507 struct mbuf *m, int runfilt)
2509 struct bridge_iflist *dbif, *sbif;
2511 struct ifnet *dst_if;
2512 int error = 0, used = 0, i;
2514 sbif = bridge_lookup_member_if(sc, src_if);
2516 BRIDGE_LOCK2REF(sc, error);
2522 /* Filter on the bridge interface before broadcasting */
2523 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2525 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2528 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2534 LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2535 dst_if = dbif->bif_ifp;
2536 if (dst_if == src_if)
2539 /* Private segments can not talk to each other */
2540 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2543 if ((dbif->bif_flags & IFBIF_STP) &&
2544 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2547 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2548 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2551 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2554 if (LIST_NEXT(dbif, bif_next) == NULL) {
2558 mc = m_dup(m, M_NOWAIT);
2560 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2566 * Filter on the output interface. Pass a NULL bridge interface
2567 * pointer so we do not redundantly filter on the bridge for
2568 * each interface we broadcast on.
2570 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2572 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2576 /* Keep the layer3 header aligned */
2577 i = min(mc->m_pkthdr.len, max_protohdr);
2578 mc = m_copyup(mc, i, ETHER_ALIGN);
2580 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2584 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2590 bridge_enqueue(sc, dst_if, mc);
2602 * Duplicate a packet out one or more interfaces that are in span mode,
2603 * the original mbuf is unmodified.
2606 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2608 struct bridge_iflist *bif;
2609 struct ifnet *dst_if;
2612 if (LIST_EMPTY(&sc->sc_spanlist))
2615 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2616 dst_if = bif->bif_ifp;
2618 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2621 mc = m_copypacket(m, M_NOWAIT);
2623 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2627 bridge_enqueue(sc, dst_if, mc);
2634 * Add a bridge routing entry.
2637 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2638 struct bridge_iflist *bif, int setflags, uint8_t flags)
2640 struct bridge_rtnode *brt;
2643 BRIDGE_LOCK_ASSERT(sc);
2645 /* Check the source address is valid and not multicast. */
2646 if (ETHER_IS_MULTICAST(dst) ||
2647 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2648 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2651 /* 802.1p frames map to vlan 1 */
2656 * A route for this destination might already exist. If so,
2657 * update it, otherwise create a new one.
2659 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2660 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2661 sc->sc_brtexceeded++;
2664 /* Check per interface address limits (if enabled) */
2665 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2666 bif->bif_addrexceeded++;
2671 * Allocate a new bridge forwarding node, and
2672 * initialize the expiration time and Ethernet
2675 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2679 if (bif->bif_flags & IFBIF_STICKY)
2680 brt->brt_flags = IFBAF_STICKY;
2682 brt->brt_flags = IFBAF_DYNAMIC;
2684 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2685 brt->brt_vlan = vlan;
2687 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2688 uma_zfree(V_bridge_rtnode_zone, brt);
2695 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2696 brt->brt_dst != bif) {
2697 brt->brt_dst->bif_addrcnt--;
2699 brt->brt_dst->bif_addrcnt++;
2702 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2703 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2705 brt->brt_flags = flags;
2713 * Lookup the destination interface for an address.
2715 static struct ifnet *
2716 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2718 struct bridge_rtnode *brt;
2720 BRIDGE_LOCK_ASSERT(sc);
2722 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2725 return (brt->brt_ifp);
2731 * Trim the routine table so that we have a number
2732 * of routing entries less than or equal to the
2736 bridge_rttrim(struct bridge_softc *sc)
2738 struct bridge_rtnode *brt, *nbrt;
2740 BRIDGE_LOCK_ASSERT(sc);
2742 /* Make sure we actually need to do this. */
2743 if (sc->sc_brtcnt <= sc->sc_brtmax)
2746 /* Force an aging cycle; this might trim enough addresses. */
2748 if (sc->sc_brtcnt <= sc->sc_brtmax)
2751 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2752 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2753 bridge_rtnode_destroy(sc, brt);
2754 if (sc->sc_brtcnt <= sc->sc_brtmax)
2763 * Aging timer for the bridge.
2766 bridge_timer(void *arg)
2768 struct bridge_softc *sc = arg;
2770 BRIDGE_LOCK_ASSERT(sc);
2772 /* Destruction of rtnodes requires a proper vnet context */
2773 CURVNET_SET(sc->sc_ifp->if_vnet);
2776 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2777 callout_reset(&sc->sc_brcallout,
2778 bridge_rtable_prune_period * hz, bridge_timer, sc);
2785 * Perform an aging cycle.
2788 bridge_rtage(struct bridge_softc *sc)
2790 struct bridge_rtnode *brt, *nbrt;
2792 BRIDGE_LOCK_ASSERT(sc);
2794 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2795 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2796 if (time_uptime >= brt->brt_expire)
2797 bridge_rtnode_destroy(sc, brt);
2805 * Remove all dynamic addresses from the bridge.
2808 bridge_rtflush(struct bridge_softc *sc, int full)
2810 struct bridge_rtnode *brt, *nbrt;
2812 BRIDGE_LOCK_ASSERT(sc);
2814 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2815 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2816 bridge_rtnode_destroy(sc, brt);
2823 * Remove an address from the table.
2826 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2828 struct bridge_rtnode *brt;
2831 BRIDGE_LOCK_ASSERT(sc);
2834 * If vlan is zero then we want to delete for all vlans so the lookup
2835 * may return more than one.
2837 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2838 bridge_rtnode_destroy(sc, brt);
2842 return (found ? 0 : ENOENT);
2848 * Delete routes to a speicifc member interface.
2851 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2853 struct bridge_rtnode *brt, *nbrt;
2855 BRIDGE_LOCK_ASSERT(sc);
2857 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2858 if (brt->brt_ifp == ifp && (full ||
2859 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2860 bridge_rtnode_destroy(sc, brt);
2865 * bridge_rtable_init:
2867 * Initialize the route table for this bridge.
2870 bridge_rtable_init(struct bridge_softc *sc)
2874 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2875 M_DEVBUF, M_WAITOK);
2877 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2878 LIST_INIT(&sc->sc_rthash[i]);
2880 sc->sc_rthash_key = arc4random();
2881 LIST_INIT(&sc->sc_rtlist);
2885 * bridge_rtable_fini:
2887 * Deconstruct the route table for this bridge.
2890 bridge_rtable_fini(struct bridge_softc *sc)
2893 KASSERT(sc->sc_brtcnt == 0,
2894 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2895 free(sc->sc_rthash, M_DEVBUF);
2899 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2900 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2902 #define mix(a, b, c) \
2904 a -= b; a -= c; a ^= (c >> 13); \
2905 b -= c; b -= a; b ^= (a << 8); \
2906 c -= a; c -= b; c ^= (b >> 13); \
2907 a -= b; a -= c; a ^= (c >> 12); \
2908 b -= c; b -= a; b ^= (a << 16); \
2909 c -= a; c -= b; c ^= (b >> 5); \
2910 a -= b; a -= c; a ^= (c >> 3); \
2911 b -= c; b -= a; b ^= (a << 10); \
2912 c -= a; c -= b; c ^= (b >> 15); \
2913 } while (/*CONSTCOND*/0)
2915 static __inline uint32_t
2916 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2918 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2929 return (c & BRIDGE_RTHASH_MASK);
2935 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2939 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2940 d = ((int)a[i]) - ((int)b[i]);
2947 * bridge_rtnode_lookup:
2949 * Look up a bridge route node for the specified destination. Compare the
2950 * vlan id or if zero then just return the first match.
2952 static struct bridge_rtnode *
2953 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2955 struct bridge_rtnode *brt;
2959 BRIDGE_LOCK_ASSERT(sc);
2961 hash = bridge_rthash(sc, addr);
2962 LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2963 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2964 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2974 * bridge_rtnode_insert:
2976 * Insert the specified bridge node into the route table. We
2977 * assume the entry is not already in the table.
2980 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2982 struct bridge_rtnode *lbrt;
2986 BRIDGE_LOCK_ASSERT(sc);
2988 hash = bridge_rthash(sc, brt->brt_addr);
2990 lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2992 LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2997 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2998 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3001 LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3004 if (LIST_NEXT(lbrt, brt_hash) == NULL) {
3005 LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3008 lbrt = LIST_NEXT(lbrt, brt_hash);
3009 } while (lbrt != NULL);
3012 panic("bridge_rtnode_insert: impossible");
3016 LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3023 * bridge_rtnode_destroy:
3025 * Destroy a bridge rtnode.
3028 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3030 BRIDGE_LOCK_ASSERT(sc);
3032 LIST_REMOVE(brt, brt_hash);
3034 LIST_REMOVE(brt, brt_list);
3036 brt->brt_dst->bif_addrcnt--;
3037 uma_zfree(V_bridge_rtnode_zone, brt);
3041 * bridge_rtable_expire:
3043 * Set the expiry time for all routes on an interface.
3046 bridge_rtable_expire(struct ifnet *ifp, int age)
3048 struct bridge_softc *sc = ifp->if_bridge;
3049 struct bridge_rtnode *brt;
3051 CURVNET_SET(ifp->if_vnet);
3055 * If the age is zero then flush, otherwise set all the expiry times to
3056 * age for the interface
3059 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3061 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3062 /* Cap the expiry time to 'age' */
3063 if (brt->brt_ifp == ifp &&
3064 brt->brt_expire > time_uptime + age &&
3065 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3066 brt->brt_expire = time_uptime + age;
3074 * bridge_state_change:
3076 * Callback from the bridgestp code when a port changes states.
3079 bridge_state_change(struct ifnet *ifp, int state)
3081 struct bridge_softc *sc = ifp->if_bridge;
3082 static const char *stpstates[] = {
3091 CURVNET_SET(ifp->if_vnet);
3093 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3094 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3099 * Send bridge packets through pfil if they are one of the types pfil can deal
3100 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3101 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3105 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3107 int snap, error, i, hlen;
3108 struct ether_header *eh1, eh2;
3111 u_int16_t ether_type;
3115 error = -1; /* Default error if not error == 0 */
3118 /* we may return with the IP fields swapped, ensure its not shared */
3119 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3122 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3123 return (0); /* filtering is disabled */
3125 i = min((*mp)->m_pkthdr.len, max_protohdr);
3126 if ((*mp)->m_len < i) {
3127 *mp = m_pullup(*mp, i);
3129 printf("%s: m_pullup failed\n", __func__);
3134 eh1 = mtod(*mp, struct ether_header *);
3135 ether_type = ntohs(eh1->ether_type);
3138 * Check for SNAP/LLC.
3140 if (ether_type < ETHERMTU) {
3141 struct llc *llc2 = (struct llc *)(eh1 + 1);
3143 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3144 llc2->llc_dsap == LLC_SNAP_LSAP &&
3145 llc2->llc_ssap == LLC_SNAP_LSAP &&
3146 llc2->llc_control == LLC_UI) {
3147 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3153 * If we're trying to filter bridge traffic, don't look at anything
3154 * other than IP and ARP traffic. If the filter doesn't understand
3155 * IPv6, don't allow IPv6 through the bridge either. This is lame
3156 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3157 * but of course we don't have an AppleTalk filter to begin with.
3158 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3161 switch (ether_type) {
3163 case ETHERTYPE_REVARP:
3164 if (V_pfil_ipfw_arp == 0)
3165 return (0); /* Automatically pass */
3170 case ETHERTYPE_IPV6:
3175 * Check to see if the user wants to pass non-ip
3176 * packets, these will not be checked by pfil(9) and
3177 * passed unconditionally so the default is to drop.
3183 /* Run the packet through pfil before stripping link headers */
3184 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3185 dir == PFIL_OUT && ifp != NULL) {
3186 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3194 /* Strip off the Ethernet header and keep a copy. */
3195 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3196 m_adj(*mp, ETHER_HDR_LEN);
3198 /* Strip off snap header, if present */
3200 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3201 m_adj(*mp, sizeof(struct llc));
3205 * Check the IP header for alignment and errors
3207 if (dir == PFIL_IN) {
3208 switch (ether_type) {
3210 error = bridge_ip_checkbasic(mp);
3213 case ETHERTYPE_IPV6:
3214 error = bridge_ip6_checkbasic(mp);
3227 * Run the packet through pfil
3230 switch (ether_type) {
3233 * Run pfil on the member interface and the bridge, both can
3234 * be skipped by clearing pfil_member or pfil_bridge.
3237 * in_if -> bridge_if -> out_if
3239 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3240 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3244 if (V_pfil_member && ifp != NULL && (rv =
3245 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3249 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3250 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3254 /* check if we need to fragment the packet */
3255 /* bridge_fragment generates a mbuf chain of packets */
3256 /* that already include eth headers */
3257 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3258 i = (*mp)->m_pkthdr.len;
3259 if (i > ifp->if_mtu) {
3260 error = bridge_fragment(ifp, mp, &eh2, snap,
3266 /* Recalculate the ip checksum. */
3267 ip = mtod(*mp, struct ip *);
3268 hlen = ip->ip_hl << 2;
3269 if (hlen < sizeof(struct ip))
3271 if (hlen > (*mp)->m_len) {
3272 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3274 ip = mtod(*mp, struct ip *);
3279 if (hlen == sizeof(struct ip))
3280 ip->ip_sum = in_cksum_hdr(ip);
3282 ip->ip_sum = in_cksum(*mp, hlen);
3286 case ETHERTYPE_IPV6:
3287 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3288 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3292 if (V_pfil_member && ifp != NULL && (rv =
3293 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3297 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3298 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3317 * Finally, put everything back the way it was and return
3320 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3323 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3326 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3329 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3340 * Perform basic checks on header size since
3341 * pfil assumes ip_input has already processed
3342 * it for it. Cut-and-pasted from ip_input.c.
3343 * Given how simple the IPv6 version is,
3344 * does the IPv4 version really need to be
3347 * XXX Should we update ipstat here, or not?
3348 * XXX Right now we update ipstat but not
3352 bridge_ip_checkbasic(struct mbuf **mp)
3354 struct mbuf *m = *mp;
3362 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3363 if ((m = m_copyup(m, sizeof(struct ip),
3364 (max_linkhdr + 3) & ~3)) == NULL) {
3365 /* XXXJRT new stat, please */
3366 KMOD_IPSTAT_INC(ips_toosmall);
3369 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3370 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3371 KMOD_IPSTAT_INC(ips_toosmall);
3375 ip = mtod(m, struct ip *);
3376 if (ip == NULL) goto bad;
3378 if (ip->ip_v != IPVERSION) {
3379 KMOD_IPSTAT_INC(ips_badvers);
3382 hlen = ip->ip_hl << 2;
3383 if (hlen < sizeof(struct ip)) { /* minimum header length */
3384 KMOD_IPSTAT_INC(ips_badhlen);
3387 if (hlen > m->m_len) {
3388 if ((m = m_pullup(m, hlen)) == NULL) {
3389 KMOD_IPSTAT_INC(ips_badhlen);
3392 ip = mtod(m, struct ip *);
3393 if (ip == NULL) goto bad;
3396 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3397 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3399 if (hlen == sizeof(struct ip)) {
3400 sum = in_cksum_hdr(ip);
3402 sum = in_cksum(m, hlen);
3406 KMOD_IPSTAT_INC(ips_badsum);
3410 /* Retrieve the packet length. */
3411 len = ntohs(ip->ip_len);
3414 * Check for additional length bogosity
3417 KMOD_IPSTAT_INC(ips_badlen);
3422 * Check that the amount of data in the buffers
3423 * is as at least much as the IP header would have us expect.
3424 * Drop packet if shorter than we expect.
3426 if (m->m_pkthdr.len < len) {
3427 KMOD_IPSTAT_INC(ips_tooshort);
3431 /* Checks out, proceed */
3442 * Same as above, but for IPv6.
3443 * Cut-and-pasted from ip6_input.c.
3444 * XXX Should we update ip6stat, or not?
3447 bridge_ip6_checkbasic(struct mbuf **mp)
3449 struct mbuf *m = *mp;
3450 struct ip6_hdr *ip6;
3453 * If the IPv6 header is not aligned, slurp it up into a new
3454 * mbuf with space for link headers, in the event we forward
3455 * it. Otherwise, if it is aligned, make sure the entire base
3456 * IPv6 header is in the first mbuf of the chain.
3458 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3459 struct ifnet *inifp = m->m_pkthdr.rcvif;
3460 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3461 (max_linkhdr + 3) & ~3)) == NULL) {
3462 /* XXXJRT new stat, please */
3463 IP6STAT_INC(ip6s_toosmall);
3464 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3467 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3468 struct ifnet *inifp = m->m_pkthdr.rcvif;
3469 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3470 IP6STAT_INC(ip6s_toosmall);
3471 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3476 ip6 = mtod(m, struct ip6_hdr *);
3478 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3479 IP6STAT_INC(ip6s_badvers);
3480 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3484 /* Checks out, proceed */
3497 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3500 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3501 int snap, struct llc *llc)
3503 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3507 if (m->m_len < sizeof(struct ip) &&
3508 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3510 ip = mtod(m, struct ip *);
3512 m->m_pkthdr.csum_flags |= CSUM_IP;
3513 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3518 * Walk the chain and re-add the Ethernet header for
3521 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3522 nextpkt = mcur->m_nextpkt;
3523 mcur->m_nextpkt = NULL;
3525 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3529 mprev->m_nextpkt = nextpkt;
3532 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3535 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3539 mprev->m_nextpkt = nextpkt;
3542 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3545 * The previous two M_PREPEND could have inserted one or two
3546 * mbufs in front so we have to update the previous packet's
3549 mcur->m_nextpkt = nextpkt;
3551 mprev->m_nextpkt = mcur;
3553 /* The first mbuf in the original chain needs to be
3560 KMOD_IPSTAT_INC(ips_fragmented);
3564 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3565 m = mcur->m_nextpkt;
3572 bridge_linkstate(struct ifnet *ifp)
3574 struct bridge_softc *sc = ifp->if_bridge;
3575 struct bridge_iflist *bif;
3578 bif = bridge_lookup_member_if(sc, ifp);
3583 bridge_linkcheck(sc);
3586 bstp_linkstate(&bif->bif_stp);
3590 bridge_linkcheck(struct bridge_softc *sc)
3592 struct bridge_iflist *bif;
3593 int new_link, hasls;
3595 BRIDGE_LOCK_ASSERT(sc);
3596 new_link = LINK_STATE_DOWN;
3598 /* Our link is considered up if at least one of our ports is active */
3599 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3600 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3602 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3603 new_link = LINK_STATE_UP;
3607 if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3608 /* If no interfaces support link-state then we default to up */
3609 new_link = LINK_STATE_UP;
3611 if_link_state_change(sc->sc_ifp, new_link);