1 /* $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $ */
4 * Copyright 2001 Wasabi Systems, Inc.
7 * Written by Jason R. Thorpe for Wasabi Systems, Inc.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed for the NetBSD Project by
20 * Wasabi Systems, Inc.
21 * 4. The name of Wasabi Systems, Inc. may not be used to endorse
22 * or promote products derived from this software without specific prior
25 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
27 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
28 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
29 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35 * POSSIBILITY OF SUCH DAMAGE.
39 * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
40 * All rights reserved.
42 * Redistribution and use in source and binary forms, with or without
43 * modification, are permitted provided that the following conditions
45 * 1. Redistributions of source code must retain the above copyright
46 * notice, this list of conditions and the following disclaimer.
47 * 2. Redistributions in binary form must reproduce the above copyright
48 * notice, this list of conditions and the following disclaimer in the
49 * documentation and/or other materials provided with the distribution.
51 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
52 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
53 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
54 * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
55 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
56 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
57 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
58 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
59 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
60 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
61 * POSSIBILITY OF SUCH DAMAGE.
63 * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
67 * Network interface bridge support.
71 * - Currently only supports Ethernet-like interfaces (Ethernet,
72 * 802.11, VLANs on Ethernet, etc.) Figure out a nice way
73 * to bridge other types of interfaces (FDDI-FDDI, and maybe
74 * consider heterogenous bridges).
77 #include <sys/cdefs.h>
78 __FBSDID("$FreeBSD$");
81 #include "opt_inet6.h"
84 #include <sys/param.h>
86 #include <sys/malloc.h>
87 #include <sys/protosw.h>
88 #include <sys/systm.h>
90 #include <sys/socket.h> /* for net/if.h */
91 #include <sys/sockio.h>
92 #include <sys/ctype.h> /* string functions */
93 #include <sys/kernel.h>
94 #include <sys/random.h>
95 #include <sys/syslog.h>
96 #include <sys/sysctl.h>
98 #include <sys/module.h>
100 #include <sys/proc.h>
101 #include <sys/lock.h>
102 #include <sys/mutex.h>
103 #include <sys/vimage.h>
107 #include <net/if_clone.h>
108 #include <net/if_dl.h>
109 #include <net/if_types.h>
110 #include <net/if_var.h>
111 #include <net/pfil.h>
113 #include <netinet/in.h> /* for struct arpcom */
114 #include <netinet/in_systm.h>
115 #include <netinet/in_var.h>
116 #include <netinet/ip.h>
117 #include <netinet/ip_var.h>
119 #include <netinet/ip6.h>
120 #include <netinet6/ip6_var.h>
123 #include <netinet/ip_carp.h>
125 #include <machine/in_cksum.h>
126 #include <netinet/if_ether.h> /* for struct arpcom */
127 #include <net/bridgestp.h>
128 #include <net/if_bridgevar.h>
129 #include <net/if_llc.h>
130 #include <net/if_vlan_var.h>
132 #include <net/route.h>
133 #include <netinet/ip_fw.h>
134 #include <netinet/ip_dummynet.h>
137 * Size of the route hash table. Must be a power of two.
139 #ifndef BRIDGE_RTHASH_SIZE
140 #define BRIDGE_RTHASH_SIZE 1024
143 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1)
146 * Maximum number of addresses to cache.
148 #ifndef BRIDGE_RTABLE_MAX
149 #define BRIDGE_RTABLE_MAX 100
153 * Timeout (in seconds) for entries learned dynamically.
155 #ifndef BRIDGE_RTABLE_TIMEOUT
156 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */
160 * Number of seconds between walks of the route list.
162 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
163 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60)
167 * List of capabilities to possibly mask on the member interface.
169 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
172 * Bridge interface list entry.
174 struct bridge_iflist {
175 LIST_ENTRY(bridge_iflist) bif_next;
176 struct ifnet *bif_ifp; /* member if */
177 struct bstp_port bif_stp; /* STP state */
178 uint32_t bif_flags; /* member if flags */
179 int bif_savedcaps; /* saved capabilities */
180 uint32_t bif_addrmax; /* max # of addresses */
181 uint32_t bif_addrcnt; /* cur. # of addresses */
182 uint32_t bif_addrexceeded;/* # of address violations */
188 struct bridge_rtnode {
189 LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
190 LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
191 struct bridge_iflist *brt_dst; /* destination if */
192 unsigned long brt_expire; /* expiration time */
193 uint8_t brt_flags; /* address flags */
194 uint8_t brt_addr[ETHER_ADDR_LEN];
195 uint16_t brt_vlan; /* vlan id */
197 #define brt_ifp brt_dst->bif_ifp
200 * Software state for each bridge.
202 struct bridge_softc {
203 struct ifnet *sc_ifp; /* make this an interface */
204 LIST_ENTRY(bridge_softc) sc_list;
207 uint32_t sc_brtmax; /* max # of addresses */
208 uint32_t sc_brtcnt; /* cur. # of addresses */
209 uint32_t sc_brttimeout; /* rt timeout in seconds */
210 struct callout sc_brcallout; /* bridge callout */
211 uint32_t sc_iflist_ref; /* refcount for sc_iflist */
212 uint32_t sc_iflist_xcnt; /* refcount for sc_iflist */
213 LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
214 LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
215 LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
216 uint32_t sc_rthash_key; /* key for hash */
217 LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
218 struct bstp_state sc_stp; /* STP state */
219 uint32_t sc_brtexceeded; /* # of cache drops */
220 u_char sc_defaddr[6]; /* Default MAC address */
223 static struct mtx bridge_list_mtx;
224 eventhandler_tag bridge_detach_cookie = NULL;
226 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
228 uma_zone_t bridge_rtnode_zone;
230 static int bridge_clone_create(struct if_clone *, int, caddr_t);
231 static void bridge_clone_destroy(struct ifnet *);
233 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
234 static void bridge_mutecaps(struct bridge_softc *);
235 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
237 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
238 static void bridge_init(void *);
239 static void bridge_dummynet(struct mbuf *, struct ifnet *);
240 static void bridge_stop(struct ifnet *, int);
241 static void bridge_start(struct ifnet *);
242 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
243 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
245 static void bridge_enqueue(struct bridge_softc *, struct ifnet *,
247 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
249 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
252 static void bridge_timer(void *);
254 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
256 static void bridge_span(struct bridge_softc *, struct mbuf *);
258 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
259 uint16_t, struct bridge_iflist *, int, uint8_t);
260 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
262 static void bridge_rttrim(struct bridge_softc *);
263 static void bridge_rtage(struct bridge_softc *);
264 static void bridge_rtflush(struct bridge_softc *, int);
265 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
268 static int bridge_rtable_init(struct bridge_softc *);
269 static void bridge_rtable_fini(struct bridge_softc *);
271 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
272 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
273 const uint8_t *, uint16_t);
274 static int bridge_rtnode_insert(struct bridge_softc *,
275 struct bridge_rtnode *);
276 static void bridge_rtnode_destroy(struct bridge_softc *,
277 struct bridge_rtnode *);
278 static void bridge_rtable_expire(struct ifnet *, int);
279 static void bridge_state_change(struct ifnet *, int);
281 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
283 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
285 static void bridge_delete_member(struct bridge_softc *,
286 struct bridge_iflist *, int);
287 static void bridge_delete_span(struct bridge_softc *,
288 struct bridge_iflist *);
290 static int bridge_ioctl_add(struct bridge_softc *, void *);
291 static int bridge_ioctl_del(struct bridge_softc *, void *);
292 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
293 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
294 static int bridge_ioctl_scache(struct bridge_softc *, void *);
295 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
296 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
297 static int bridge_ioctl_rts(struct bridge_softc *, void *);
298 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
299 static int bridge_ioctl_sto(struct bridge_softc *, void *);
300 static int bridge_ioctl_gto(struct bridge_softc *, void *);
301 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
302 static int bridge_ioctl_flush(struct bridge_softc *, void *);
303 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
304 static int bridge_ioctl_spri(struct bridge_softc *, void *);
305 static int bridge_ioctl_ght(struct bridge_softc *, void *);
306 static int bridge_ioctl_sht(struct bridge_softc *, void *);
307 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
308 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
309 static int bridge_ioctl_gma(struct bridge_softc *, void *);
310 static int bridge_ioctl_sma(struct bridge_softc *, void *);
311 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
312 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
313 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
314 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
315 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
316 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
317 static int bridge_ioctl_grte(struct bridge_softc *, void *);
318 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
319 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
320 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
321 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
323 static int bridge_ip_checkbasic(struct mbuf **mp);
325 static int bridge_ip6_checkbasic(struct mbuf **mp);
327 static int bridge_fragment(struct ifnet *, struct mbuf *,
328 struct ether_header *, int, struct llc *);
330 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
331 #define VLANTAGOF(_m) \
332 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
334 static struct bstp_cb_ops bridge_ops = {
335 .bcb_state = bridge_state_change,
336 .bcb_rtage = bridge_rtable_expire
339 SYSCTL_DECL(_net_link);
340 SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
342 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
343 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
344 static int pfil_member = 1; /* run pfil hooks on the member interface */
345 static int pfil_ipfw = 0; /* layer2 filter with ipfw */
346 static int pfil_ipfw_arp = 0; /* layer2 filter with ipfw */
347 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
348 locally destined packets */
349 static int log_stp = 0; /* log STP state changes */
350 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
351 &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
352 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
353 &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
354 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
355 &pfil_bridge, 0, "Packet filter on the bridge interface");
356 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
357 &pfil_member, 0, "Packet filter on the member interface");
358 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
360 "Packet filter on the physical interface for locally destined packets");
361 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
362 &log_stp, 0, "Log STP state changes");
364 struct bridge_control {
365 int (*bc_func)(struct bridge_softc *, void *);
370 #define BC_F_COPYIN 0x01 /* copy arguments in */
371 #define BC_F_COPYOUT 0x02 /* copy arguments out */
372 #define BC_F_SUSER 0x04 /* do super-user check */
374 const struct bridge_control bridge_control_table[] = {
375 { bridge_ioctl_add, sizeof(struct ifbreq),
376 BC_F_COPYIN|BC_F_SUSER },
377 { bridge_ioctl_del, sizeof(struct ifbreq),
378 BC_F_COPYIN|BC_F_SUSER },
380 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
381 BC_F_COPYIN|BC_F_COPYOUT },
382 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
383 BC_F_COPYIN|BC_F_SUSER },
385 { bridge_ioctl_scache, sizeof(struct ifbrparam),
386 BC_F_COPYIN|BC_F_SUSER },
387 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
390 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
391 BC_F_COPYIN|BC_F_COPYOUT },
392 { bridge_ioctl_rts, sizeof(struct ifbaconf),
393 BC_F_COPYIN|BC_F_COPYOUT },
395 { bridge_ioctl_saddr, sizeof(struct ifbareq),
396 BC_F_COPYIN|BC_F_SUSER },
398 { bridge_ioctl_sto, sizeof(struct ifbrparam),
399 BC_F_COPYIN|BC_F_SUSER },
400 { bridge_ioctl_gto, sizeof(struct ifbrparam),
403 { bridge_ioctl_daddr, sizeof(struct ifbareq),
404 BC_F_COPYIN|BC_F_SUSER },
406 { bridge_ioctl_flush, sizeof(struct ifbreq),
407 BC_F_COPYIN|BC_F_SUSER },
409 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
411 { bridge_ioctl_spri, sizeof(struct ifbrparam),
412 BC_F_COPYIN|BC_F_SUSER },
414 { bridge_ioctl_ght, sizeof(struct ifbrparam),
416 { bridge_ioctl_sht, sizeof(struct ifbrparam),
417 BC_F_COPYIN|BC_F_SUSER },
419 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
421 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
422 BC_F_COPYIN|BC_F_SUSER },
424 { bridge_ioctl_gma, sizeof(struct ifbrparam),
426 { bridge_ioctl_sma, sizeof(struct ifbrparam),
427 BC_F_COPYIN|BC_F_SUSER },
429 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
430 BC_F_COPYIN|BC_F_SUSER },
432 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
433 BC_F_COPYIN|BC_F_SUSER },
435 { bridge_ioctl_addspan, sizeof(struct ifbreq),
436 BC_F_COPYIN|BC_F_SUSER },
437 { bridge_ioctl_delspan, sizeof(struct ifbreq),
438 BC_F_COPYIN|BC_F_SUSER },
440 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
443 { bridge_ioctl_grte, sizeof(struct ifbrparam),
446 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
447 BC_F_COPYIN|BC_F_COPYOUT },
449 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
450 BC_F_COPYIN|BC_F_SUSER },
452 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
453 BC_F_COPYIN|BC_F_SUSER },
455 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
456 BC_F_COPYIN|BC_F_SUSER },
459 const int bridge_control_table_size =
460 sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
462 LIST_HEAD(, bridge_softc) bridge_list;
464 IFC_SIMPLE_DECLARE(bridge, 0);
467 bridge_modevent(module_t mod, int type, void *data)
472 mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
473 if_clone_attach(&bridge_cloner);
474 bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
475 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
477 LIST_INIT(&bridge_list);
478 bridge_input_p = bridge_input;
479 bridge_output_p = bridge_output;
480 bridge_dn_p = bridge_dummynet;
481 bridge_detach_cookie = EVENTHANDLER_REGISTER(
482 ifnet_departure_event, bridge_ifdetach, NULL,
483 EVENTHANDLER_PRI_ANY);
486 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
487 bridge_detach_cookie);
488 if_clone_detach(&bridge_cloner);
489 uma_zdestroy(bridge_rtnode_zone);
490 bridge_input_p = NULL;
491 bridge_output_p = NULL;
493 mtx_destroy(&bridge_list_mtx);
501 static moduledata_t bridge_mod = {
507 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
508 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
511 * handler for net.link.bridge.pfil_ipfw
514 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
516 int enable = pfil_ipfw;
519 error = sysctl_handle_int(oidp, &enable, 0, req);
520 enable = (enable) ? 1 : 0;
522 if (enable != pfil_ipfw) {
526 * Disable pfil so that ipfw doesnt run twice, if the user
527 * really wants both then they can re-enable pfil_bridge and/or
528 * pfil_member. Also allow non-ip packets as ipfw can filter by
540 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
541 &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
544 * bridge_clone_create:
546 * Create a new bridge instance.
549 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
551 struct bridge_softc *sc, *sc2;
552 struct ifnet *bifp, *ifp;
555 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
556 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
562 BRIDGE_LOCK_INIT(sc);
563 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
564 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
566 /* Initialize our routing table. */
567 bridge_rtable_init(sc);
569 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
571 LIST_INIT(&sc->sc_iflist);
572 LIST_INIT(&sc->sc_spanlist);
575 if_initname(ifp, ifc->ifc_name, unit);
576 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
577 ifp->if_ioctl = bridge_ioctl;
578 ifp->if_start = bridge_start;
579 ifp->if_init = bridge_init;
580 ifp->if_type = IFT_BRIDGE;
581 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
582 ifp->if_snd.ifq_drv_maxlen = ifqmaxlen;
583 IFQ_SET_READY(&ifp->if_snd);
586 * Generate a random ethernet address with a locally administered
589 * Since we are using random ethernet addresses for the bridge, it is
590 * possible that we might have address collisions, so make sure that
591 * this hardware address isn't already in use on another bridge.
593 for (retry = 1; retry != 0;) {
594 arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
595 sc->sc_defaddr[0] &= ~1; /* clear multicast bit */
596 sc->sc_defaddr[0] |= 2; /* set the LAA bit */
598 mtx_lock(&bridge_list_mtx);
599 LIST_FOREACH(sc2, &bridge_list, sc_list) {
601 if (memcmp(sc->sc_defaddr,
602 IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0)
605 mtx_unlock(&bridge_list_mtx);
608 bstp_attach(&sc->sc_stp, &bridge_ops);
609 ether_ifattach(ifp, sc->sc_defaddr);
610 /* Now undo some of the damage... */
611 ifp->if_baudrate = 0;
612 ifp->if_type = IFT_BRIDGE;
614 mtx_lock(&bridge_list_mtx);
615 LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
616 mtx_unlock(&bridge_list_mtx);
622 * bridge_clone_destroy:
624 * Destroy a bridge instance.
627 bridge_clone_destroy(struct ifnet *ifp)
629 struct bridge_softc *sc = ifp->if_softc;
630 struct bridge_iflist *bif;
635 ifp->if_flags &= ~IFF_UP;
637 while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
638 bridge_delete_member(sc, bif, 0);
640 while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
641 bridge_delete_span(sc, bif);
646 callout_drain(&sc->sc_brcallout);
648 mtx_lock(&bridge_list_mtx);
649 LIST_REMOVE(sc, sc_list);
650 mtx_unlock(&bridge_list_mtx);
652 bstp_detach(&sc->sc_stp);
654 if_free_type(ifp, IFT_ETHER);
656 /* Tear down the routing table. */
657 bridge_rtable_fini(sc);
659 BRIDGE_LOCK_DESTROY(sc);
666 * Handle a control request from the operator.
669 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
671 struct bridge_softc *sc = ifp->if_softc;
672 struct thread *td = curthread;
674 struct ifbreq ifbreq;
675 struct ifbifconf ifbifconf;
676 struct ifbareq ifbareq;
677 struct ifbaconf ifbaconf;
678 struct ifbrparam ifbrparam;
679 struct ifbropreq ifbropreq;
681 struct ifdrv *ifd = (struct ifdrv *) data;
682 const struct bridge_control *bc;
693 if (ifd->ifd_cmd >= bridge_control_table_size) {
697 bc = &bridge_control_table[ifd->ifd_cmd];
699 if (cmd == SIOCGDRVSPEC &&
700 (bc->bc_flags & BC_F_COPYOUT) == 0) {
704 else if (cmd == SIOCSDRVSPEC &&
705 (bc->bc_flags & BC_F_COPYOUT) != 0) {
710 if (bc->bc_flags & BC_F_SUSER) {
711 error = priv_check(td, PRIV_NET_BRIDGE);
716 if (ifd->ifd_len != bc->bc_argsize ||
717 ifd->ifd_len > sizeof(args)) {
722 bzero(&args, sizeof(args));
723 if (bc->bc_flags & BC_F_COPYIN) {
724 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
730 error = (*bc->bc_func)(sc, &args);
735 if (bc->bc_flags & BC_F_COPYOUT)
736 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
741 if (!(ifp->if_flags & IFF_UP) &&
742 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
744 * If interface is marked down and it is running,
745 * then stop and disable it.
750 } else if ((ifp->if_flags & IFF_UP) &&
751 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
753 * If interface is marked up and it is stopped, then
761 /* Do not allow the MTU to be changed on the bridge */
767 * drop the lock as ether_ioctl() will call bridge_start() and
768 * cause the lock to be recursed.
770 error = ether_ioctl(ifp, cmd, data);
780 * Clear or restore unwanted capabilities on the member interface
783 bridge_mutecaps(struct bridge_softc *sc)
785 struct bridge_iflist *bif;
788 /* Initial bitmask of capabilities to test */
789 mask = BRIDGE_IFCAPS_MASK;
791 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
792 /* Every member must support it or its disabled */
793 mask &= bif->bif_savedcaps;
796 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
797 enabled = bif->bif_ifp->if_capenable;
798 /* strip off mask bits and enable them again if allowed */
799 enabled &= ~BRIDGE_IFCAPS_MASK;
802 * Receive offload can only be enabled if all members also
803 * support send offload.
805 if ((enabled & IFCAP_TSO) == 0)
806 enabled &= ~IFCAP_LRO;
808 bridge_set_ifcap(sc, bif, enabled);
814 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
816 struct ifnet *ifp = bif->bif_ifp;
820 bzero(&ifr, sizeof(ifr));
821 ifr.ifr_reqcap = set;
823 if (ifp->if_capenable != set) {
825 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
826 IFF_UNLOCKGIANT(ifp);
828 if_printf(sc->sc_ifp,
829 "error setting interface capabilities on %s\n",
835 * bridge_lookup_member:
837 * Lookup a bridge member interface.
839 static struct bridge_iflist *
840 bridge_lookup_member(struct bridge_softc *sc, const char *name)
842 struct bridge_iflist *bif;
845 BRIDGE_LOCK_ASSERT(sc);
847 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
849 if (strcmp(ifp->if_xname, name) == 0)
857 * bridge_lookup_member_if:
859 * Lookup a bridge member interface by ifnet*.
861 static struct bridge_iflist *
862 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
864 struct bridge_iflist *bif;
866 BRIDGE_LOCK_ASSERT(sc);
868 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
869 if (bif->bif_ifp == member_ifp)
877 * bridge_delete_member:
879 * Delete the specified member interface.
882 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
885 struct ifnet *ifs = bif->bif_ifp;
886 struct ifnet *fif = NULL;
888 BRIDGE_LOCK_ASSERT(sc);
891 switch (ifs->if_type) {
895 * Take the interface out of promiscuous mode.
897 (void) ifpromisc(ifs, 0);
905 panic("bridge_delete_member: impossible");
909 /* reneable any interface capabilities */
910 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
913 if (bif->bif_flags & IFBIF_STP)
914 bstp_disable(&bif->bif_stp);
916 ifs->if_bridge = NULL;
918 LIST_REMOVE(bif, bif_next);
922 * If removing the interface that gave the bridge its mac address, set
923 * the mac address of the bridge to the address of the next member, or
924 * to its default address if no members are left.
926 if (!memcmp(IF_LLADDR(sc->sc_ifp), IF_LLADDR(ifs), ETHER_ADDR_LEN)) {
927 if (LIST_EMPTY(&sc->sc_iflist))
928 bcopy(sc->sc_defaddr,
929 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
931 fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
932 bcopy(IF_LLADDR(fif),
933 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
937 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
938 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
939 KASSERT(bif->bif_addrcnt == 0,
940 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
943 bstp_destroy(&bif->bif_stp); /* prepare to free */
949 * bridge_delete_span:
951 * Delete the specified span interface.
954 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
956 BRIDGE_LOCK_ASSERT(sc);
958 KASSERT(bif->bif_ifp->if_bridge == NULL,
959 ("%s: not a span interface", __func__));
961 LIST_REMOVE(bif, bif_next);
966 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
968 struct ifbreq *req = arg;
969 struct bridge_iflist *bif = NULL;
973 ifs = ifunit(req->ifbr_ifsname);
976 if (ifs->if_ioctl == NULL) /* must be supported */
979 /* If it's in the span list, it can't be a member. */
980 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
981 if (ifs == bif->bif_ifp)
984 /* Allow the first Ethernet member to define the MTU */
985 if (ifs->if_type != IFT_GIF) {
986 if (LIST_EMPTY(&sc->sc_iflist))
987 sc->sc_ifp->if_mtu = ifs->if_mtu;
988 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
989 if_printf(sc->sc_ifp, "invalid MTU for %s\n",
995 if (ifs->if_bridge == sc)
998 if (ifs->if_bridge != NULL)
1001 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1006 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1007 bif->bif_savedcaps = ifs->if_capenable;
1009 switch (ifs->if_type) {
1013 * Place the interface into promiscuous mode.
1015 error = ifpromisc(ifs, 1);
1029 * Assign the interface's MAC address to the bridge if it's the first
1030 * member and the MAC address of the bridge has not been changed from
1031 * the default randomly generated one.
1033 if (LIST_EMPTY(&sc->sc_iflist) &&
1034 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN))
1035 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1037 ifs->if_bridge = sc;
1038 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1040 * XXX: XLOCK HERE!?!
1042 * NOTE: insert_***HEAD*** should be safe for the traversals.
1044 LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1046 /* Set interface capabilities to the intersection set of all members */
1047 bridge_mutecaps(sc);
1051 free(bif, M_DEVBUF);
1057 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1059 struct ifbreq *req = arg;
1060 struct bridge_iflist *bif;
1062 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1066 bridge_delete_member(sc, bif, 0);
1072 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1074 struct ifbreq *req = arg;
1075 struct bridge_iflist *bif;
1076 struct bstp_port *bp;
1078 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1083 req->ifbr_ifsflags = bif->bif_flags;
1084 req->ifbr_state = bp->bp_state;
1085 req->ifbr_priority = bp->bp_priority;
1086 req->ifbr_path_cost = bp->bp_path_cost;
1087 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1088 req->ifbr_proto = bp->bp_protover;
1089 req->ifbr_role = bp->bp_role;
1090 req->ifbr_stpflags = bp->bp_flags;
1091 req->ifbr_addrcnt = bif->bif_addrcnt;
1092 req->ifbr_addrmax = bif->bif_addrmax;
1093 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1095 /* Copy STP state options as flags */
1096 if (bp->bp_operedge)
1097 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1098 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1099 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1100 if (bp->bp_ptp_link)
1101 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1102 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1103 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1104 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1105 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1106 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1107 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1112 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1114 struct ifbreq *req = arg;
1115 struct bridge_iflist *bif;
1116 struct bstp_port *bp;
1119 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1124 if (req->ifbr_ifsflags & IFBIF_SPAN)
1125 /* SPAN is readonly */
1128 if (req->ifbr_ifsflags & IFBIF_STP) {
1129 if ((bif->bif_flags & IFBIF_STP) == 0) {
1130 error = bstp_enable(&bif->bif_stp);
1135 if ((bif->bif_flags & IFBIF_STP) != 0)
1136 bstp_disable(&bif->bif_stp);
1139 /* Pass on STP flags */
1140 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1141 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1142 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1143 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1145 /* Save the bits relating to the bridge */
1146 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1152 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1154 struct ifbrparam *param = arg;
1156 sc->sc_brtmax = param->ifbrp_csize;
1163 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1165 struct ifbrparam *param = arg;
1167 param->ifbrp_csize = sc->sc_brtmax;
1173 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1175 struct ifbifconf *bifc = arg;
1176 struct bridge_iflist *bif;
1179 int count, buflen, len, error = 0;
1182 LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1184 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1187 buflen = sizeof(breq) * count;
1188 if (bifc->ifbic_len == 0) {
1189 bifc->ifbic_len = buflen;
1193 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1198 len = min(bifc->ifbic_len, buflen);
1199 bzero(&breq, sizeof(breq));
1200 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1201 if (len < sizeof(breq))
1204 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1205 sizeof(breq.ifbr_ifsname));
1206 /* Fill in the ifbreq structure */
1207 error = bridge_ioctl_gifflags(sc, &breq);
1210 memcpy(buf, &breq, sizeof(breq));
1212 buf += sizeof(breq);
1213 len -= sizeof(breq);
1215 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1216 if (len < sizeof(breq))
1219 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1220 sizeof(breq.ifbr_ifsname));
1221 breq.ifbr_ifsflags = bif->bif_flags;
1222 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1223 memcpy(buf, &breq, sizeof(breq));
1225 buf += sizeof(breq);
1226 len -= sizeof(breq);
1230 bifc->ifbic_len = sizeof(breq) * count;
1231 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1233 free(outbuf, M_TEMP);
1238 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1240 struct ifbaconf *bac = arg;
1241 struct bridge_rtnode *brt;
1242 struct ifbareq bareq;
1244 int count, buflen, len, error = 0;
1246 if (bac->ifbac_len == 0)
1250 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1252 buflen = sizeof(bareq) * count;
1255 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1260 len = min(bac->ifbac_len, buflen);
1261 bzero(&bareq, sizeof(bareq));
1262 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1263 if (len < sizeof(bareq))
1265 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1266 sizeof(bareq.ifba_ifsname));
1267 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1268 bareq.ifba_vlan = brt->brt_vlan;
1269 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1270 time_uptime < brt->brt_expire)
1271 bareq.ifba_expire = brt->brt_expire - time_uptime;
1273 bareq.ifba_expire = 0;
1274 bareq.ifba_flags = brt->brt_flags;
1276 memcpy(buf, &bareq, sizeof(bareq));
1278 buf += sizeof(bareq);
1279 len -= sizeof(bareq);
1283 bac->ifbac_len = sizeof(bareq) * count;
1284 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1286 free(outbuf, M_TEMP);
1291 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1293 struct ifbareq *req = arg;
1294 struct bridge_iflist *bif;
1297 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1301 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1308 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1310 struct ifbrparam *param = arg;
1312 sc->sc_brttimeout = param->ifbrp_ctime;
1317 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1319 struct ifbrparam *param = arg;
1321 param->ifbrp_ctime = sc->sc_brttimeout;
1326 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1328 struct ifbareq *req = arg;
1330 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1334 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1336 struct ifbreq *req = arg;
1338 bridge_rtflush(sc, req->ifbr_ifsflags);
1343 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1345 struct ifbrparam *param = arg;
1346 struct bstp_state *bs = &sc->sc_stp;
1348 param->ifbrp_prio = bs->bs_bridge_priority;
1353 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1355 struct ifbrparam *param = arg;
1357 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1361 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1363 struct ifbrparam *param = arg;
1364 struct bstp_state *bs = &sc->sc_stp;
1366 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1371 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1373 struct ifbrparam *param = arg;
1375 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1379 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1381 struct ifbrparam *param = arg;
1382 struct bstp_state *bs = &sc->sc_stp;
1384 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1389 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1391 struct ifbrparam *param = arg;
1393 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1397 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1399 struct ifbrparam *param = arg;
1400 struct bstp_state *bs = &sc->sc_stp;
1402 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1407 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1409 struct ifbrparam *param = arg;
1411 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1415 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1417 struct ifbreq *req = arg;
1418 struct bridge_iflist *bif;
1420 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1424 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1428 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1430 struct ifbreq *req = arg;
1431 struct bridge_iflist *bif;
1433 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1437 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1441 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1443 struct ifbreq *req = arg;
1444 struct bridge_iflist *bif;
1446 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1450 bif->bif_addrmax = req->ifbr_addrmax;
1455 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1457 struct ifbreq *req = arg;
1458 struct bridge_iflist *bif = NULL;
1461 ifs = ifunit(req->ifbr_ifsname);
1465 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1466 if (ifs == bif->bif_ifp)
1469 if (ifs->if_bridge != NULL)
1472 switch (ifs->if_type) {
1481 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1486 bif->bif_flags = IFBIF_SPAN;
1488 LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1494 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1496 struct ifbreq *req = arg;
1497 struct bridge_iflist *bif;
1500 ifs = ifunit(req->ifbr_ifsname);
1504 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1505 if (ifs == bif->bif_ifp)
1511 bridge_delete_span(sc, bif);
1517 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1519 struct ifbropreq *req = arg;
1520 struct bstp_state *bs = &sc->sc_stp;
1521 struct bstp_port *root_port;
1523 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1524 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1525 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1527 root_port = bs->bs_root_port;
1528 if (root_port == NULL)
1529 req->ifbop_root_port = 0;
1531 req->ifbop_root_port = root_port->bp_ifp->if_index;
1533 req->ifbop_holdcount = bs->bs_txholdcount;
1534 req->ifbop_priority = bs->bs_bridge_priority;
1535 req->ifbop_protocol = bs->bs_protover;
1536 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1537 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1538 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1539 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1540 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1541 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1547 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1549 struct ifbrparam *param = arg;
1551 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1556 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1558 struct ifbpstpconf *bifstp = arg;
1559 struct bridge_iflist *bif;
1560 struct bstp_port *bp;
1561 struct ifbpstpreq bpreq;
1563 int count, buflen, len, error = 0;
1566 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1567 if ((bif->bif_flags & IFBIF_STP) != 0)
1571 buflen = sizeof(bpreq) * count;
1572 if (bifstp->ifbpstp_len == 0) {
1573 bifstp->ifbpstp_len = buflen;
1578 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1583 len = min(bifstp->ifbpstp_len, buflen);
1584 bzero(&bpreq, sizeof(bpreq));
1585 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1586 if (len < sizeof(bpreq))
1589 if ((bif->bif_flags & IFBIF_STP) == 0)
1593 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1594 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1595 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1596 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1597 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1598 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1600 memcpy(buf, &bpreq, sizeof(bpreq));
1602 buf += sizeof(bpreq);
1603 len -= sizeof(bpreq);
1607 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1608 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1610 free(outbuf, M_TEMP);
1615 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1617 struct ifbrparam *param = arg;
1619 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1623 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1625 struct ifbrparam *param = arg;
1627 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1633 * Detach an interface from a bridge. Called when a member
1634 * interface is detaching.
1637 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1639 struct bridge_softc *sc = ifp->if_bridge;
1640 struct bridge_iflist *bif;
1642 /* Check if the interface is a bridge member */
1646 bif = bridge_lookup_member_if(sc, ifp);
1648 bridge_delete_member(sc, bif, 1);
1654 /* Check if the interface is a span port */
1655 mtx_lock(&bridge_list_mtx);
1656 LIST_FOREACH(sc, &bridge_list, sc_list) {
1658 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1659 if (ifp == bif->bif_ifp) {
1660 bridge_delete_span(sc, bif);
1666 mtx_unlock(&bridge_list_mtx);
1672 * Initialize a bridge interface.
1675 bridge_init(void *xsc)
1677 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1678 struct ifnet *ifp = sc->sc_ifp;
1680 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1684 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1687 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1688 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1696 * Stop the bridge interface.
1699 bridge_stop(struct ifnet *ifp, int disable)
1701 struct bridge_softc *sc = ifp->if_softc;
1703 BRIDGE_LOCK_ASSERT(sc);
1705 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1708 callout_stop(&sc->sc_brcallout);
1709 bstp_stop(&sc->sc_stp);
1711 bridge_rtflush(sc, IFBF_FLUSHDYN);
1713 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1719 * Enqueue a packet on a bridge member interface.
1723 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1729 len = m->m_pkthdr.len;
1730 mflags = m->m_flags;
1732 /* We may be sending a fragment so traverse the mbuf */
1735 m->m_nextpkt = NULL;
1738 * If underlying interface can not do VLAN tag insertion itself
1739 * then attach a packet tag that holds it.
1741 if ((m->m_flags & M_VLANTAG) &&
1742 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1743 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1746 "unable to prepend VLAN header\n");
1747 dst_ifp->if_oerrors++;
1750 m->m_flags &= ~M_VLANTAG;
1754 IFQ_ENQUEUE(&dst_ifp->if_snd, m, err);
1759 sc->sc_ifp->if_opackets++;
1760 sc->sc_ifp->if_obytes += len;
1762 dst_ifp->if_obytes += len;
1764 if (mflags & M_MCAST) {
1765 sc->sc_ifp->if_omcasts++;
1766 dst_ifp->if_omcasts++;
1770 if ((dst_ifp->if_drv_flags & IFF_DRV_OACTIVE) == 0)
1771 (*dst_ifp->if_start)(dst_ifp);
1777 * Receive a queued packet from dummynet and pass it on to the output
1780 * The mbuf has the Ethernet header already attached.
1783 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1785 struct bridge_softc *sc;
1787 sc = ifp->if_bridge;
1790 * The packet didnt originate from a member interface. This should only
1791 * ever happen if a member interface is removed while packets are
1799 if (PFIL_HOOKED(&inet_pfil_hook)
1801 || PFIL_HOOKED(&inet6_pfil_hook)
1804 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1810 bridge_enqueue(sc, ifp, m);
1816 * Send output from a bridge member interface. This
1817 * performs the bridging function for locally originated
1820 * The mbuf has the Ethernet header already attached. We must
1821 * enqueue or free the mbuf before returning.
1824 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1827 struct ether_header *eh;
1828 struct ifnet *dst_if;
1829 struct bridge_softc *sc;
1832 if (m->m_len < ETHER_HDR_LEN) {
1833 m = m_pullup(m, ETHER_HDR_LEN);
1838 eh = mtod(m, struct ether_header *);
1839 sc = ifp->if_bridge;
1840 vlan = VLANTAGOF(m);
1845 * If bridge is down, but the original output interface is up,
1846 * go ahead and send out that interface. Otherwise, the packet
1849 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1855 * If the packet is a multicast, or we don't know a better way to
1856 * get there, send to all interfaces.
1858 if (ETHER_IS_MULTICAST(eh->ether_dhost))
1861 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1862 if (dst_if == NULL) {
1863 struct bridge_iflist *bif;
1865 int error = 0, used = 0;
1869 BRIDGE_LOCK2REF(sc, error);
1875 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1876 dst_if = bif->bif_ifp;
1878 if (dst_if->if_type == IFT_GIF)
1880 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1884 * If this is not the original output interface,
1885 * and the interface is participating in spanning
1886 * tree, make sure the port is in a state that
1887 * allows forwarding.
1889 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1890 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1893 if (LIST_NEXT(bif, bif_next) == NULL) {
1897 mc = m_copypacket(m, M_DONTWAIT);
1899 sc->sc_ifp->if_oerrors++;
1904 bridge_enqueue(sc, dst_if, mc);
1914 * XXX Spanning tree consideration here?
1918 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1925 bridge_enqueue(sc, dst_if, m);
1932 * Start output on a bridge.
1936 bridge_start(struct ifnet *ifp)
1938 struct bridge_softc *sc;
1940 struct ether_header *eh;
1941 struct ifnet *dst_if;
1945 ifp->if_drv_flags |= IFF_DRV_OACTIVE;
1947 IFQ_DEQUEUE(&ifp->if_snd, m);
1950 ETHER_BPF_MTAP(ifp, m);
1952 eh = mtod(m, struct ether_header *);
1956 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
1957 dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1);
1961 bridge_broadcast(sc, ifp, m, 0);
1964 bridge_enqueue(sc, dst_if, m);
1967 ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
1973 * The forwarding function of the bridge.
1975 * NOTE: Releases the lock on return.
1978 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
1981 struct bridge_iflist *dbif;
1982 struct ifnet *src_if, *dst_if, *ifp;
1983 struct ether_header *eh;
1988 src_if = m->m_pkthdr.rcvif;
1992 ifp->if_ibytes += m->m_pkthdr.len;
1993 vlan = VLANTAGOF(m);
1995 if ((sbif->bif_flags & IFBIF_STP) &&
1996 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1999 eh = mtod(m, struct ether_header *);
2000 dst = eh->ether_dhost;
2002 /* If the interface is learning, record the address. */
2003 if (sbif->bif_flags & IFBIF_LEARNING) {
2004 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2005 sbif, 0, IFBAF_DYNAMIC);
2007 * If the interface has addresses limits then deny any source
2008 * that is not in the cache.
2010 if (error && sbif->bif_addrmax)
2014 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2015 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2019 * At this point, the port either doesn't participate
2020 * in spanning tree or it is in the forwarding state.
2024 * If the packet is unicast, destined for someone on
2025 * "this" side of the bridge, drop it.
2027 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2028 dst_if = bridge_rtlookup(sc, dst, vlan);
2029 if (src_if == dst_if)
2033 * Check if its a reserved multicast address, any address
2034 * listed in 802.1D section 7.12.6 may not be forwarded by the
2036 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2038 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2039 dst[2] == 0xc2 && dst[3] == 0x00 &&
2040 dst[4] == 0x00 && dst[5] <= 0x0f)
2043 /* ...forward it to all interfaces. */
2049 * If we have a destination interface which is a member of our bridge,
2050 * OR this is a unicast packet, push it through the bpf(4) machinery.
2051 * For broadcast or multicast packets, don't bother because it will
2052 * be reinjected into ether_input. We do this before we pass the packets
2053 * through the pfil(9) framework, as it is possible that pfil(9) will
2054 * drop the packet, or possibly modify it, making it difficult to debug
2055 * firewall issues on the bridge.
2057 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2058 ETHER_BPF_MTAP(ifp, m);
2060 /* run the packet filter */
2061 if (PFIL_HOOKED(&inet_pfil_hook)
2063 || PFIL_HOOKED(&inet6_pfil_hook)
2067 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2074 if (dst_if == NULL) {
2075 bridge_broadcast(sc, src_if, m, 1);
2080 * At this point, we're dealing with a unicast frame
2081 * going to a different interface.
2083 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2086 dbif = bridge_lookup_member_if(sc, dst_if);
2088 /* Not a member of the bridge (anymore?) */
2091 /* Private segments can not talk to each other */
2092 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2095 if ((dbif->bif_flags & IFBIF_STP) &&
2096 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2101 if (PFIL_HOOKED(&inet_pfil_hook)
2103 || PFIL_HOOKED(&inet6_pfil_hook)
2106 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2112 bridge_enqueue(sc, dst_if, m);
2123 * Receive input from a member interface. Queue the packet for
2124 * bridging if it is not for us.
2126 static struct mbuf *
2127 bridge_input(struct ifnet *ifp, struct mbuf *m)
2129 struct bridge_softc *sc = ifp->if_bridge;
2130 struct bridge_iflist *bif, *bif2;
2132 struct ether_header *eh;
2133 struct mbuf *mc, *mc2;
2137 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2141 vlan = VLANTAGOF(m);
2144 * Implement support for bridge monitoring. If this flag has been
2145 * set on this interface, discard the packet once we push it through
2146 * the bpf(4) machinery, but before we do, increment the byte and
2147 * packet counters associated with this interface.
2149 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2150 m->m_pkthdr.rcvif = bifp;
2151 ETHER_BPF_MTAP(bifp, m);
2152 bifp->if_ipackets++;
2153 bifp->if_ibytes += m->m_pkthdr.len;
2158 bif = bridge_lookup_member_if(sc, ifp);
2164 eh = mtod(m, struct ether_header *);
2168 if (m->m_flags & (M_BCAST|M_MCAST)) {
2169 /* Tap off 802.1D packets; they do not get forwarded. */
2170 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2171 ETHER_ADDR_LEN) == 0) {
2172 m = bstp_input(&bif->bif_stp, ifp, m);
2179 if ((bif->bif_flags & IFBIF_STP) &&
2180 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2186 * Make a deep copy of the packet and enqueue the copy
2187 * for bridge processing; return the original packet for
2190 mc = m_dup(m, M_DONTWAIT);
2196 /* Perform the bridge forwarding function with the copy. */
2197 bridge_forward(sc, bif, mc);
2200 * Reinject the mbuf as arriving on the bridge so we have a
2201 * chance at claiming multicast packets. We can not loop back
2202 * here from ether_input as a bridge is never a member of a
2205 KASSERT(bifp->if_bridge == NULL,
2206 ("loop created in bridge_input"));
2207 mc2 = m_dup(m, M_DONTWAIT);
2209 /* Keep the layer3 header aligned */
2210 int i = min(mc2->m_pkthdr.len, max_protohdr);
2211 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2214 mc2->m_pkthdr.rcvif = bifp;
2215 (*bifp->if_input)(bifp, mc2);
2218 /* Return the original packet for local processing. */
2222 if ((bif->bif_flags & IFBIF_STP) &&
2223 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2229 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2230 || ((iface)->if_carp \
2231 && carp_forus((iface)->if_carp, eh->ether_dhost))
2232 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2233 || ((iface)->if_carp \
2234 && carp_forus((iface)->if_carp, eh->ether_shost))
2236 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2237 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2241 # define OR_PFIL_HOOKED_INET6 \
2242 || PFIL_HOOKED(&inet6_pfil_hook)
2244 # define OR_PFIL_HOOKED_INET6
2247 #define GRAB_OUR_PACKETS(iface) \
2248 if ((iface)->if_type == IFT_GIF) \
2250 /* It is destined for us. */ \
2251 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2252 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2254 if ((iface)->if_type == IFT_BRIDGE) { \
2255 ETHER_BPF_MTAP(iface, m); \
2256 iface->if_ipackets++; \
2257 /* Filter on the physical interface. */ \
2258 if (pfil_local_phys && \
2259 (PFIL_HOOKED(&inet_pfil_hook) \
2260 OR_PFIL_HOOKED_INET6)) { \
2261 if (bridge_pfil(&m, NULL, ifp, \
2262 PFIL_IN) != 0 || m == NULL) { \
2263 BRIDGE_UNLOCK(sc); \
2268 if (bif->bif_flags & IFBIF_LEARNING) { \
2269 error = bridge_rtupdate(sc, eh->ether_shost, \
2270 vlan, bif, 0, IFBAF_DYNAMIC); \
2271 if (error && bif->bif_addrmax) { \
2272 BRIDGE_UNLOCK(sc); \
2277 m->m_pkthdr.rcvif = iface; \
2278 BRIDGE_UNLOCK(sc); \
2282 /* We just received a packet that we sent out. */ \
2283 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2284 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2286 BRIDGE_UNLOCK(sc); \
2292 * Unicast. Make sure it's not for the bridge.
2294 do { GRAB_OUR_PACKETS(bifp) } while (0);
2297 * Give a chance for ifp at first priority. This will help when the
2298 * packet comes through the interface like VLAN's with the same MACs
2299 * on several interfaces from the same bridge. This also will save
2300 * some CPU cycles in case the destination interface and the input
2301 * interface (eq ifp) are the same.
2303 do { GRAB_OUR_PACKETS(ifp) } while (0);
2305 /* Now check the all bridge members. */
2306 LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2307 GRAB_OUR_PACKETS(bif2->bif_ifp)
2310 #undef OR_CARP_CHECK_WE_ARE_DST
2311 #undef OR_CARP_CHECK_WE_ARE_SRC
2312 #undef OR_PFIL_HOOKED_INET6
2313 #undef GRAB_OUR_PACKETS
2315 /* Perform the bridge forwarding function. */
2316 bridge_forward(sc, bif, m);
2324 * Send a frame to all interfaces that are members of
2325 * the bridge, except for the one on which the packet
2328 * NOTE: Releases the lock on return.
2331 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2332 struct mbuf *m, int runfilt)
2334 struct bridge_iflist *dbif, *sbif;
2336 struct ifnet *dst_if;
2337 int error = 0, used = 0, i;
2339 sbif = bridge_lookup_member_if(sc, src_if);
2341 BRIDGE_LOCK2REF(sc, error);
2347 /* Filter on the bridge interface before broadcasting */
2348 if (runfilt && (PFIL_HOOKED(&inet_pfil_hook)
2350 || PFIL_HOOKED(&inet6_pfil_hook)
2353 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2359 LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2360 dst_if = dbif->bif_ifp;
2361 if (dst_if == src_if)
2364 /* Private segments can not talk to each other */
2365 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2368 if ((dbif->bif_flags & IFBIF_STP) &&
2369 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2372 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2373 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2376 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2379 if (LIST_NEXT(dbif, bif_next) == NULL) {
2383 mc = m_dup(m, M_DONTWAIT);
2385 sc->sc_ifp->if_oerrors++;
2391 * Filter on the output interface. Pass a NULL bridge interface
2392 * pointer so we do not redundantly filter on the bridge for
2393 * each interface we broadcast on.
2395 if (runfilt && (PFIL_HOOKED(&inet_pfil_hook)
2397 || PFIL_HOOKED(&inet6_pfil_hook)
2401 /* Keep the layer3 header aligned */
2402 i = min(mc->m_pkthdr.len, max_protohdr);
2403 mc = m_copyup(mc, i, ETHER_ALIGN);
2405 sc->sc_ifp->if_oerrors++;
2409 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2415 bridge_enqueue(sc, dst_if, mc);
2427 * Duplicate a packet out one or more interfaces that are in span mode,
2428 * the original mbuf is unmodified.
2431 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2433 struct bridge_iflist *bif;
2434 struct ifnet *dst_if;
2437 if (LIST_EMPTY(&sc->sc_spanlist))
2440 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2441 dst_if = bif->bif_ifp;
2443 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2446 mc = m_copypacket(m, M_DONTWAIT);
2448 sc->sc_ifp->if_oerrors++;
2452 bridge_enqueue(sc, dst_if, mc);
2459 * Add a bridge routing entry.
2462 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2463 struct bridge_iflist *bif, int setflags, uint8_t flags)
2465 struct bridge_rtnode *brt;
2468 BRIDGE_LOCK_ASSERT(sc);
2470 /* Check the source address is valid and not multicast. */
2471 if (ETHER_IS_MULTICAST(dst) ||
2472 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2473 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2476 /* 802.1p frames map to vlan 1 */
2481 * A route for this destination might already exist. If so,
2482 * update it, otherwise create a new one.
2484 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2485 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2486 sc->sc_brtexceeded++;
2489 /* Check per interface address limits (if enabled) */
2490 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2491 bif->bif_addrexceeded++;
2496 * Allocate a new bridge forwarding node, and
2497 * initialize the expiration time and Ethernet
2500 brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2504 if (bif->bif_flags & IFBIF_STICKY)
2505 brt->brt_flags = IFBAF_STICKY;
2507 brt->brt_flags = IFBAF_DYNAMIC;
2509 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2510 brt->brt_vlan = vlan;
2512 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2513 uma_zfree(bridge_rtnode_zone, brt);
2520 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2521 brt->brt_dst != bif) {
2522 brt->brt_dst->bif_addrcnt--;
2524 brt->brt_dst->bif_addrcnt++;
2527 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2528 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2530 brt->brt_flags = flags;
2538 * Lookup the destination interface for an address.
2540 static struct ifnet *
2541 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2543 struct bridge_rtnode *brt;
2545 BRIDGE_LOCK_ASSERT(sc);
2547 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2550 return (brt->brt_ifp);
2556 * Trim the routine table so that we have a number
2557 * of routing entries less than or equal to the
2561 bridge_rttrim(struct bridge_softc *sc)
2563 struct bridge_rtnode *brt, *nbrt;
2565 BRIDGE_LOCK_ASSERT(sc);
2567 /* Make sure we actually need to do this. */
2568 if (sc->sc_brtcnt <= sc->sc_brtmax)
2571 /* Force an aging cycle; this might trim enough addresses. */
2573 if (sc->sc_brtcnt <= sc->sc_brtmax)
2576 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2577 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2578 bridge_rtnode_destroy(sc, brt);
2579 if (sc->sc_brtcnt <= sc->sc_brtmax)
2588 * Aging timer for the bridge.
2591 bridge_timer(void *arg)
2593 struct bridge_softc *sc = arg;
2595 BRIDGE_LOCK_ASSERT(sc);
2599 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2600 callout_reset(&sc->sc_brcallout,
2601 bridge_rtable_prune_period * hz, bridge_timer, sc);
2607 * Perform an aging cycle.
2610 bridge_rtage(struct bridge_softc *sc)
2612 struct bridge_rtnode *brt, *nbrt;
2614 BRIDGE_LOCK_ASSERT(sc);
2616 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2617 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2618 if (time_uptime >= brt->brt_expire)
2619 bridge_rtnode_destroy(sc, brt);
2627 * Remove all dynamic addresses from the bridge.
2630 bridge_rtflush(struct bridge_softc *sc, int full)
2632 struct bridge_rtnode *brt, *nbrt;
2634 BRIDGE_LOCK_ASSERT(sc);
2636 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2637 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2638 bridge_rtnode_destroy(sc, brt);
2645 * Remove an address from the table.
2648 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2650 struct bridge_rtnode *brt;
2653 BRIDGE_LOCK_ASSERT(sc);
2656 * If vlan is zero then we want to delete for all vlans so the lookup
2657 * may return more than one.
2659 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2660 bridge_rtnode_destroy(sc, brt);
2664 return (found ? 0 : ENOENT);
2670 * Delete routes to a speicifc member interface.
2673 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2675 struct bridge_rtnode *brt, *nbrt;
2677 BRIDGE_LOCK_ASSERT(sc);
2679 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2680 if (brt->brt_ifp == ifp && (full ||
2681 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2682 bridge_rtnode_destroy(sc, brt);
2687 * bridge_rtable_init:
2689 * Initialize the route table for this bridge.
2692 bridge_rtable_init(struct bridge_softc *sc)
2696 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2697 M_DEVBUF, M_NOWAIT);
2698 if (sc->sc_rthash == NULL)
2701 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2702 LIST_INIT(&sc->sc_rthash[i]);
2704 sc->sc_rthash_key = arc4random();
2706 LIST_INIT(&sc->sc_rtlist);
2712 * bridge_rtable_fini:
2714 * Deconstruct the route table for this bridge.
2717 bridge_rtable_fini(struct bridge_softc *sc)
2720 KASSERT(sc->sc_brtcnt == 0,
2721 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2722 free(sc->sc_rthash, M_DEVBUF);
2726 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2727 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2729 #define mix(a, b, c) \
2731 a -= b; a -= c; a ^= (c >> 13); \
2732 b -= c; b -= a; b ^= (a << 8); \
2733 c -= a; c -= b; c ^= (b >> 13); \
2734 a -= b; a -= c; a ^= (c >> 12); \
2735 b -= c; b -= a; b ^= (a << 16); \
2736 c -= a; c -= b; c ^= (b >> 5); \
2737 a -= b; a -= c; a ^= (c >> 3); \
2738 b -= c; b -= a; b ^= (a << 10); \
2739 c -= a; c -= b; c ^= (b >> 15); \
2740 } while (/*CONSTCOND*/0)
2742 static __inline uint32_t
2743 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2745 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2756 return (c & BRIDGE_RTHASH_MASK);
2762 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2766 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2767 d = ((int)a[i]) - ((int)b[i]);
2774 * bridge_rtnode_lookup:
2776 * Look up a bridge route node for the specified destination. Compare the
2777 * vlan id or if zero then just return the first match.
2779 static struct bridge_rtnode *
2780 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2782 struct bridge_rtnode *brt;
2786 BRIDGE_LOCK_ASSERT(sc);
2788 hash = bridge_rthash(sc, addr);
2789 LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2790 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2791 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2801 * bridge_rtnode_insert:
2803 * Insert the specified bridge node into the route table. We
2804 * assume the entry is not already in the table.
2807 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2809 struct bridge_rtnode *lbrt;
2813 BRIDGE_LOCK_ASSERT(sc);
2815 hash = bridge_rthash(sc, brt->brt_addr);
2817 lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2819 LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2824 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2825 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2828 LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2831 if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2832 LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2835 lbrt = LIST_NEXT(lbrt, brt_hash);
2836 } while (lbrt != NULL);
2839 panic("bridge_rtnode_insert: impossible");
2843 LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2850 * bridge_rtnode_destroy:
2852 * Destroy a bridge rtnode.
2855 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2857 BRIDGE_LOCK_ASSERT(sc);
2859 LIST_REMOVE(brt, brt_hash);
2861 LIST_REMOVE(brt, brt_list);
2863 brt->brt_dst->bif_addrcnt--;
2864 uma_zfree(bridge_rtnode_zone, brt);
2868 * bridge_rtable_expire:
2870 * Set the expiry time for all routes on an interface.
2873 bridge_rtable_expire(struct ifnet *ifp, int age)
2875 struct bridge_softc *sc = ifp->if_bridge;
2876 struct bridge_rtnode *brt;
2881 * If the age is zero then flush, otherwise set all the expiry times to
2882 * age for the interface
2885 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2887 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2888 /* Cap the expiry time to 'age' */
2889 if (brt->brt_ifp == ifp &&
2890 brt->brt_expire > time_uptime + age &&
2891 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2892 brt->brt_expire = time_uptime + age;
2899 * bridge_state_change:
2901 * Callback from the bridgestp code when a port changes states.
2904 bridge_state_change(struct ifnet *ifp, int state)
2906 struct bridge_softc *sc = ifp->if_bridge;
2907 static const char *stpstates[] = {
2917 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2918 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2922 * Send bridge packets through pfil if they are one of the types pfil can deal
2923 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
2924 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2928 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2930 int snap, error, i, hlen;
2931 struct ether_header *eh1, eh2;
2932 struct ip_fw_args args;
2935 u_int16_t ether_type;
2938 error = -1; /* Default error if not error == 0 */
2941 /* we may return with the IP fields swapped, ensure its not shared */
2942 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2945 if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2946 return (0); /* filtering is disabled */
2948 i = min((*mp)->m_pkthdr.len, max_protohdr);
2949 if ((*mp)->m_len < i) {
2950 *mp = m_pullup(*mp, i);
2952 printf("%s: m_pullup failed\n", __func__);
2957 eh1 = mtod(*mp, struct ether_header *);
2958 ether_type = ntohs(eh1->ether_type);
2961 * Check for SNAP/LLC.
2963 if (ether_type < ETHERMTU) {
2964 struct llc *llc2 = (struct llc *)(eh1 + 1);
2966 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
2967 llc2->llc_dsap == LLC_SNAP_LSAP &&
2968 llc2->llc_ssap == LLC_SNAP_LSAP &&
2969 llc2->llc_control == LLC_UI) {
2970 ether_type = htons(llc2->llc_un.type_snap.ether_type);
2976 * If we're trying to filter bridge traffic, don't look at anything
2977 * other than IP and ARP traffic. If the filter doesn't understand
2978 * IPv6, don't allow IPv6 through the bridge either. This is lame
2979 * since if we really wanted, say, an AppleTalk filter, we are hosed,
2980 * but of course we don't have an AppleTalk filter to begin with.
2981 * (Note that since pfil doesn't understand ARP it will pass *ALL*
2984 switch (ether_type) {
2986 case ETHERTYPE_REVARP:
2987 if (pfil_ipfw_arp == 0)
2988 return (0); /* Automatically pass */
2993 case ETHERTYPE_IPV6:
2998 * Check to see if the user wants to pass non-ip
2999 * packets, these will not be checked by pfil(9) and
3000 * passed unconditionally so the default is to drop.
3006 /* Strip off the Ethernet header and keep a copy. */
3007 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3008 m_adj(*mp, ETHER_HDR_LEN);
3010 /* Strip off snap header, if present */
3012 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3013 m_adj(*mp, sizeof(struct llc));
3017 * Check the IP header for alignment and errors
3019 if (dir == PFIL_IN) {
3020 switch (ether_type) {
3022 error = bridge_ip_checkbasic(mp);
3025 case ETHERTYPE_IPV6:
3026 error = bridge_ip6_checkbasic(mp);
3036 if (IPFW_LOADED && pfil_ipfw != 0 && dir == PFIL_OUT && ifp != NULL) {
3038 args.rule = ip_dn_claim_rule(*mp);
3039 if (args.rule != NULL && V_fw_one_pass)
3040 goto ipfwpass; /* packet already partially processed */
3044 args.next_hop = NULL;
3046 args.inp = NULL; /* used by ipfw uid/gid/jail rules */
3047 i = ip_fw_chk_ptr(&args);
3053 if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) {
3055 /* put the Ethernet header back on */
3056 M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3059 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3062 * Pass the pkt to dummynet, which consumes it. The
3063 * packet will return to us via bridge_dummynet().
3066 ip_dn_io_ptr(mp, DN_TO_IFB_FWD, &args);
3070 if (i != IP_FW_PASS) /* drop */
3078 * Run the packet through pfil
3080 switch (ether_type) {
3083 * before calling the firewall, swap fields the same as
3084 * IP does. here we assume the header is contiguous
3086 ip = mtod(*mp, struct ip *);
3088 ip->ip_len = ntohs(ip->ip_len);
3089 ip->ip_off = ntohs(ip->ip_off);
3092 * Run pfil on the member interface and the bridge, both can
3093 * be skipped by clearing pfil_member or pfil_bridge.
3096 * in_if -> bridge_if -> out_if
3098 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3099 error = pfil_run_hooks(&inet_pfil_hook, mp, bifp,
3102 if (*mp == NULL || error != 0) /* filter may consume */
3105 if (pfil_member && ifp != NULL)
3106 error = pfil_run_hooks(&inet_pfil_hook, mp, ifp,
3109 if (*mp == NULL || error != 0) /* filter may consume */
3112 if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3113 error = pfil_run_hooks(&inet_pfil_hook, mp, bifp,
3116 if (*mp == NULL || error != 0) /* filter may consume */
3119 /* check if we need to fragment the packet */
3120 if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3121 i = (*mp)->m_pkthdr.len;
3122 if (i > ifp->if_mtu) {
3123 error = bridge_fragment(ifp, *mp, &eh2, snap,
3129 /* Recalculate the ip checksum and restore byte ordering */
3130 ip = mtod(*mp, struct ip *);
3131 hlen = ip->ip_hl << 2;
3132 if (hlen < sizeof(struct ip))
3134 if (hlen > (*mp)->m_len) {
3135 if ((*mp = m_pullup(*mp, hlen)) == 0)
3137 ip = mtod(*mp, struct ip *);
3141 ip->ip_len = htons(ip->ip_len);
3142 ip->ip_off = htons(ip->ip_off);
3144 if (hlen == sizeof(struct ip))
3145 ip->ip_sum = in_cksum_hdr(ip);
3147 ip->ip_sum = in_cksum(*mp, hlen);
3151 case ETHERTYPE_IPV6:
3152 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3153 error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp,
3156 if (*mp == NULL || error != 0) /* filter may consume */
3159 if (pfil_member && ifp != NULL)
3160 error = pfil_run_hooks(&inet6_pfil_hook, mp, ifp,
3163 if (*mp == NULL || error != 0) /* filter may consume */
3166 if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3167 error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp,
3184 * Finally, put everything back the way it was and return
3187 M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT);
3190 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3193 M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3196 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3207 * Perform basic checks on header size since
3208 * pfil assumes ip_input has already processed
3209 * it for it. Cut-and-pasted from ip_input.c.
3210 * Given how simple the IPv6 version is,
3211 * does the IPv4 version really need to be
3214 * XXX Should we update ipstat here, or not?
3215 * XXX Right now we update ipstat but not
3219 bridge_ip_checkbasic(struct mbuf **mp)
3221 struct mbuf *m = *mp;
3229 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3230 if ((m = m_copyup(m, sizeof(struct ip),
3231 (max_linkhdr + 3) & ~3)) == NULL) {
3232 /* XXXJRT new stat, please */
3233 V_ipstat.ips_toosmall++;
3236 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3237 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3238 V_ipstat.ips_toosmall++;
3242 ip = mtod(m, struct ip *);
3243 if (ip == NULL) goto bad;
3245 if (ip->ip_v != IPVERSION) {
3246 V_ipstat.ips_badvers++;
3249 hlen = ip->ip_hl << 2;
3250 if (hlen < sizeof(struct ip)) { /* minimum header length */
3251 V_ipstat.ips_badhlen++;
3254 if (hlen > m->m_len) {
3255 if ((m = m_pullup(m, hlen)) == 0) {
3256 V_ipstat.ips_badhlen++;
3259 ip = mtod(m, struct ip *);
3260 if (ip == NULL) goto bad;
3263 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3264 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3266 if (hlen == sizeof(struct ip)) {
3267 sum = in_cksum_hdr(ip);
3269 sum = in_cksum(m, hlen);
3273 V_ipstat.ips_badsum++;
3277 /* Retrieve the packet length. */
3278 len = ntohs(ip->ip_len);
3281 * Check for additional length bogosity
3284 V_ipstat.ips_badlen++;
3289 * Check that the amount of data in the buffers
3290 * is as at least much as the IP header would have us expect.
3291 * Drop packet if shorter than we expect.
3293 if (m->m_pkthdr.len < len) {
3294 V_ipstat.ips_tooshort++;
3298 /* Checks out, proceed */
3309 * Same as above, but for IPv6.
3310 * Cut-and-pasted from ip6_input.c.
3311 * XXX Should we update ip6stat, or not?
3314 bridge_ip6_checkbasic(struct mbuf **mp)
3316 struct mbuf *m = *mp;
3317 struct ip6_hdr *ip6;
3320 * If the IPv6 header is not aligned, slurp it up into a new
3321 * mbuf with space for link headers, in the event we forward
3322 * it. Otherwise, if it is aligned, make sure the entire base
3323 * IPv6 header is in the first mbuf of the chain.
3325 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3326 struct ifnet *inifp = m->m_pkthdr.rcvif;
3327 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3328 (max_linkhdr + 3) & ~3)) == NULL) {
3329 /* XXXJRT new stat, please */
3330 V_ip6stat.ip6s_toosmall++;
3331 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3334 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3335 struct ifnet *inifp = m->m_pkthdr.rcvif;
3336 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3337 V_ip6stat.ip6s_toosmall++;
3338 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3343 ip6 = mtod(m, struct ip6_hdr *);
3345 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3346 V_ip6stat.ip6s_badvers++;
3347 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3351 /* Checks out, proceed */
3364 * Return a fragmented mbuf chain.
3367 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3368 int snap, struct llc *llc)
3374 if (m->m_len < sizeof(struct ip) &&
3375 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3377 ip = mtod(m, struct ip *);
3379 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist,
3384 /* walk the chain and re-add the Ethernet header */
3385 for (m0 = m; m0; m0 = m0->m_nextpkt) {
3388 M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT);
3393 bcopy(llc, mtod(m0, caddr_t),
3394 sizeof(struct llc));
3396 M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT);
3401 bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3407 V_ipstat.ips_fragmented++;