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>
106 #include <net/if_clone.h>
107 #include <net/if_dl.h>
108 #include <net/if_types.h>
109 #include <net/if_var.h>
110 #include <net/pfil.h>
112 #include <netinet/in.h> /* for struct arpcom */
113 #include <netinet/in_systm.h>
114 #include <netinet/in_var.h>
115 #include <netinet/ip.h>
116 #include <netinet/ip_var.h>
118 #include <netinet/ip6.h>
119 #include <netinet6/ip6_var.h>
122 #include <netinet/ip_carp.h>
124 #include <machine/in_cksum.h>
125 #include <netinet/if_ether.h> /* for struct arpcom */
126 #include <net/bridgestp.h>
127 #include <net/if_bridgevar.h>
128 #include <net/if_llc.h>
129 #include <net/if_vlan_var.h>
131 #include <net/route.h>
132 #include <netinet/ip_fw.h>
133 #include <netinet/ip_dummynet.h>
136 * Size of the route hash table. Must be a power of two.
138 #ifndef BRIDGE_RTHASH_SIZE
139 #define BRIDGE_RTHASH_SIZE 1024
142 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1)
145 * Maximum number of addresses to cache.
147 #ifndef BRIDGE_RTABLE_MAX
148 #define BRIDGE_RTABLE_MAX 100
152 * Timeout (in seconds) for entries learned dynamically.
154 #ifndef BRIDGE_RTABLE_TIMEOUT
155 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */
159 * Number of seconds between walks of the route list.
161 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
162 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60)
166 * List of capabilities to possibly mask on the member interface.
168 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
170 * List of capabilities to disable on the member interface.
172 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO
175 * Bridge interface list entry.
177 struct bridge_iflist {
178 LIST_ENTRY(bridge_iflist) bif_next;
179 struct ifnet *bif_ifp; /* member if */
180 struct bstp_port bif_stp; /* STP state */
181 uint32_t bif_flags; /* member if flags */
182 int bif_savedcaps; /* saved capabilities */
183 uint32_t bif_addrmax; /* max # of addresses */
184 uint32_t bif_addrcnt; /* cur. # of addresses */
185 uint32_t bif_addrexceeded;/* # of address violations */
191 struct bridge_rtnode {
192 LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
193 LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
194 struct bridge_iflist *brt_dst; /* destination if */
195 unsigned long brt_expire; /* expiration time */
196 uint8_t brt_flags; /* address flags */
197 uint8_t brt_addr[ETHER_ADDR_LEN];
198 uint16_t brt_vlan; /* vlan id */
200 #define brt_ifp brt_dst->bif_ifp
203 * Software state for each bridge.
205 struct bridge_softc {
206 struct ifnet *sc_ifp; /* make this an interface */
207 LIST_ENTRY(bridge_softc) sc_list;
210 uint32_t sc_brtmax; /* max # of addresses */
211 uint32_t sc_brtcnt; /* cur. # of addresses */
212 uint32_t sc_brttimeout; /* rt timeout in seconds */
213 struct callout sc_brcallout; /* bridge callout */
214 uint32_t sc_iflist_ref; /* refcount for sc_iflist */
215 uint32_t sc_iflist_xcnt; /* refcount for sc_iflist */
216 LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
217 LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
218 LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
219 uint32_t sc_rthash_key; /* key for hash */
220 LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
221 struct bstp_state sc_stp; /* STP state */
222 uint32_t sc_brtexceeded; /* # of cache drops */
225 static struct mtx bridge_list_mtx;
226 eventhandler_tag bridge_detach_cookie = NULL;
228 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
230 uma_zone_t bridge_rtnode_zone;
232 static int bridge_clone_create(struct if_clone *, int, caddr_t);
233 static void bridge_clone_destroy(struct ifnet *);
235 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
236 static void bridge_mutecaps(struct bridge_softc *);
237 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
239 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
240 static void bridge_init(void *);
241 static void bridge_dummynet(struct mbuf *, struct ifnet *);
242 static void bridge_stop(struct ifnet *, int);
243 static void bridge_start(struct ifnet *);
244 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
245 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
247 static void bridge_enqueue(struct bridge_softc *, struct ifnet *,
249 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
251 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
254 static void bridge_timer(void *);
256 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
258 static void bridge_span(struct bridge_softc *, struct mbuf *);
260 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
261 uint16_t, struct bridge_iflist *, int, uint8_t);
262 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
264 static void bridge_rttrim(struct bridge_softc *);
265 static void bridge_rtage(struct bridge_softc *);
266 static void bridge_rtflush(struct bridge_softc *, int);
267 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
270 static int bridge_rtable_init(struct bridge_softc *);
271 static void bridge_rtable_fini(struct bridge_softc *);
273 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
274 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
275 const uint8_t *, uint16_t);
276 static int bridge_rtnode_insert(struct bridge_softc *,
277 struct bridge_rtnode *);
278 static void bridge_rtnode_destroy(struct bridge_softc *,
279 struct bridge_rtnode *);
280 static void bridge_rtable_expire(struct ifnet *, int);
281 static void bridge_state_change(struct ifnet *, int);
283 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
285 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
287 static void bridge_delete_member(struct bridge_softc *,
288 struct bridge_iflist *, int);
289 static void bridge_delete_span(struct bridge_softc *,
290 struct bridge_iflist *);
292 static int bridge_ioctl_add(struct bridge_softc *, void *);
293 static int bridge_ioctl_del(struct bridge_softc *, void *);
294 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
295 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
296 static int bridge_ioctl_scache(struct bridge_softc *, void *);
297 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
298 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
299 static int bridge_ioctl_rts(struct bridge_softc *, void *);
300 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
301 static int bridge_ioctl_sto(struct bridge_softc *, void *);
302 static int bridge_ioctl_gto(struct bridge_softc *, void *);
303 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
304 static int bridge_ioctl_flush(struct bridge_softc *, void *);
305 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
306 static int bridge_ioctl_spri(struct bridge_softc *, void *);
307 static int bridge_ioctl_ght(struct bridge_softc *, void *);
308 static int bridge_ioctl_sht(struct bridge_softc *, void *);
309 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
310 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
311 static int bridge_ioctl_gma(struct bridge_softc *, void *);
312 static int bridge_ioctl_sma(struct bridge_softc *, void *);
313 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
314 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
315 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
316 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
317 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
318 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
319 static int bridge_ioctl_grte(struct bridge_softc *, void *);
320 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
321 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
322 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
323 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
325 static int bridge_ip_checkbasic(struct mbuf **mp);
327 static int bridge_ip6_checkbasic(struct mbuf **mp);
329 static int bridge_fragment(struct ifnet *, struct mbuf *,
330 struct ether_header *, int, struct llc *);
332 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
333 #define VLANTAGOF(_m) \
334 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
336 static struct bstp_cb_ops bridge_ops = {
337 .bcb_state = bridge_state_change,
338 .bcb_rtage = bridge_rtable_expire
341 SYSCTL_DECL(_net_link);
342 SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
344 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
345 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
346 static int pfil_member = 1; /* run pfil hooks on the member interface */
347 static int pfil_ipfw = 0; /* layer2 filter with ipfw */
348 static int pfil_ipfw_arp = 0; /* layer2 filter with ipfw */
349 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
350 locally destined packets */
351 static int log_stp = 0; /* log STP state changes */
352 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
353 &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
354 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
355 &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
356 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
357 &pfil_bridge, 0, "Packet filter on the bridge interface");
358 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
359 &pfil_member, 0, "Packet filter on the member interface");
360 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
362 "Packet filter on the physical interface for locally destined packets");
363 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
364 &log_stp, 0, "Log STP state changes");
366 struct bridge_control {
367 int (*bc_func)(struct bridge_softc *, void *);
372 #define BC_F_COPYIN 0x01 /* copy arguments in */
373 #define BC_F_COPYOUT 0x02 /* copy arguments out */
374 #define BC_F_SUSER 0x04 /* do super-user check */
376 const struct bridge_control bridge_control_table[] = {
377 { bridge_ioctl_add, sizeof(struct ifbreq),
378 BC_F_COPYIN|BC_F_SUSER },
379 { bridge_ioctl_del, sizeof(struct ifbreq),
380 BC_F_COPYIN|BC_F_SUSER },
382 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
383 BC_F_COPYIN|BC_F_COPYOUT },
384 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
385 BC_F_COPYIN|BC_F_SUSER },
387 { bridge_ioctl_scache, sizeof(struct ifbrparam),
388 BC_F_COPYIN|BC_F_SUSER },
389 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
392 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
393 BC_F_COPYIN|BC_F_COPYOUT },
394 { bridge_ioctl_rts, sizeof(struct ifbaconf),
395 BC_F_COPYIN|BC_F_COPYOUT },
397 { bridge_ioctl_saddr, sizeof(struct ifbareq),
398 BC_F_COPYIN|BC_F_SUSER },
400 { bridge_ioctl_sto, sizeof(struct ifbrparam),
401 BC_F_COPYIN|BC_F_SUSER },
402 { bridge_ioctl_gto, sizeof(struct ifbrparam),
405 { bridge_ioctl_daddr, sizeof(struct ifbareq),
406 BC_F_COPYIN|BC_F_SUSER },
408 { bridge_ioctl_flush, sizeof(struct ifbreq),
409 BC_F_COPYIN|BC_F_SUSER },
411 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
413 { bridge_ioctl_spri, sizeof(struct ifbrparam),
414 BC_F_COPYIN|BC_F_SUSER },
416 { bridge_ioctl_ght, sizeof(struct ifbrparam),
418 { bridge_ioctl_sht, sizeof(struct ifbrparam),
419 BC_F_COPYIN|BC_F_SUSER },
421 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
423 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
424 BC_F_COPYIN|BC_F_SUSER },
426 { bridge_ioctl_gma, sizeof(struct ifbrparam),
428 { bridge_ioctl_sma, sizeof(struct ifbrparam),
429 BC_F_COPYIN|BC_F_SUSER },
431 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
432 BC_F_COPYIN|BC_F_SUSER },
434 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
435 BC_F_COPYIN|BC_F_SUSER },
437 { bridge_ioctl_addspan, sizeof(struct ifbreq),
438 BC_F_COPYIN|BC_F_SUSER },
439 { bridge_ioctl_delspan, sizeof(struct ifbreq),
440 BC_F_COPYIN|BC_F_SUSER },
442 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
445 { bridge_ioctl_grte, sizeof(struct ifbrparam),
448 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
449 BC_F_COPYIN|BC_F_COPYOUT },
451 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
452 BC_F_COPYIN|BC_F_SUSER },
454 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
455 BC_F_COPYIN|BC_F_SUSER },
457 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
458 BC_F_COPYIN|BC_F_SUSER },
461 const int bridge_control_table_size =
462 sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
464 LIST_HEAD(, bridge_softc) bridge_list;
466 IFC_SIMPLE_DECLARE(bridge, 0);
469 bridge_modevent(module_t mod, int type, void *data)
474 mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
475 if_clone_attach(&bridge_cloner);
476 bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
477 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
479 LIST_INIT(&bridge_list);
480 bridge_input_p = bridge_input;
481 bridge_output_p = bridge_output;
482 bridge_dn_p = bridge_dummynet;
483 bridge_detach_cookie = EVENTHANDLER_REGISTER(
484 ifnet_departure_event, bridge_ifdetach, NULL,
485 EVENTHANDLER_PRI_ANY);
488 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
489 bridge_detach_cookie);
490 if_clone_detach(&bridge_cloner);
491 uma_zdestroy(bridge_rtnode_zone);
492 bridge_input_p = NULL;
493 bridge_output_p = NULL;
495 mtx_destroy(&bridge_list_mtx);
503 static moduledata_t bridge_mod = {
509 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
510 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
513 * handler for net.link.bridge.pfil_ipfw
516 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
518 int enable = pfil_ipfw;
521 error = sysctl_handle_int(oidp, &enable, 0, req);
522 enable = (enable) ? 1 : 0;
524 if (enable != pfil_ipfw) {
528 * Disable pfil so that ipfw doesnt run twice, if the user
529 * really wants both then they can re-enable pfil_bridge and/or
530 * pfil_member. Also allow non-ip packets as ipfw can filter by
542 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
543 &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
546 * bridge_clone_create:
548 * Create a new bridge instance.
551 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
553 struct bridge_softc *sc, *sc2;
554 struct ifnet *bifp, *ifp;
558 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
559 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
565 BRIDGE_LOCK_INIT(sc);
566 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
567 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
569 /* Initialize our routing table. */
570 bridge_rtable_init(sc);
572 callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
574 LIST_INIT(&sc->sc_iflist);
575 LIST_INIT(&sc->sc_spanlist);
578 if_initname(ifp, ifc->ifc_name, unit);
579 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
580 ifp->if_ioctl = bridge_ioctl;
581 ifp->if_start = bridge_start;
582 ifp->if_init = bridge_init;
583 ifp->if_type = IFT_BRIDGE;
584 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
585 ifp->if_snd.ifq_drv_maxlen = ifqmaxlen;
586 IFQ_SET_READY(&ifp->if_snd);
589 * Generate a random ethernet address with a locally administered
592 * Since we are using random ethernet addresses for the bridge, it is
593 * possible that we might have address collisions, so make sure that
594 * this hardware address isn't already in use on another bridge.
596 for (retry = 1; retry != 0;) {
597 arc4rand(eaddr, ETHER_ADDR_LEN, 1);
598 eaddr[0] &= ~1; /* clear multicast bit */
599 eaddr[0] |= 2; /* set the LAA bit */
601 mtx_lock(&bridge_list_mtx);
602 LIST_FOREACH(sc2, &bridge_list, sc_list) {
604 if (memcmp(eaddr, IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0)
607 mtx_unlock(&bridge_list_mtx);
610 bstp_attach(&sc->sc_stp, &bridge_ops);
611 ether_ifattach(ifp, eaddr);
612 /* Now undo some of the damage... */
613 ifp->if_baudrate = 0;
614 ifp->if_type = IFT_BRIDGE;
616 mtx_lock(&bridge_list_mtx);
617 LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
618 mtx_unlock(&bridge_list_mtx);
624 * bridge_clone_destroy:
626 * Destroy a bridge instance.
629 bridge_clone_destroy(struct ifnet *ifp)
631 struct bridge_softc *sc = ifp->if_softc;
632 struct bridge_iflist *bif;
637 ifp->if_flags &= ~IFF_UP;
639 while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
640 bridge_delete_member(sc, bif, 0);
642 while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
643 bridge_delete_span(sc, bif);
648 callout_drain(&sc->sc_brcallout);
650 mtx_lock(&bridge_list_mtx);
651 LIST_REMOVE(sc, sc_list);
652 mtx_unlock(&bridge_list_mtx);
654 bstp_detach(&sc->sc_stp);
656 if_free_type(ifp, IFT_ETHER);
658 /* Tear down the routing table. */
659 bridge_rtable_fini(sc);
661 BRIDGE_LOCK_DESTROY(sc);
668 * Handle a control request from the operator.
671 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
673 struct bridge_softc *sc = ifp->if_softc;
674 struct thread *td = curthread;
676 struct ifbreq ifbreq;
677 struct ifbifconf ifbifconf;
678 struct ifbareq ifbareq;
679 struct ifbaconf ifbaconf;
680 struct ifbrparam ifbrparam;
681 struct ifbropreq ifbropreq;
683 struct ifdrv *ifd = (struct ifdrv *) data;
684 const struct bridge_control *bc;
695 if (ifd->ifd_cmd >= bridge_control_table_size) {
699 bc = &bridge_control_table[ifd->ifd_cmd];
701 if (cmd == SIOCGDRVSPEC &&
702 (bc->bc_flags & BC_F_COPYOUT) == 0) {
706 else if (cmd == SIOCSDRVSPEC &&
707 (bc->bc_flags & BC_F_COPYOUT) != 0) {
712 if (bc->bc_flags & BC_F_SUSER) {
713 error = priv_check(td, PRIV_NET_BRIDGE);
718 if (ifd->ifd_len != bc->bc_argsize ||
719 ifd->ifd_len > sizeof(args)) {
724 bzero(&args, sizeof(args));
725 if (bc->bc_flags & BC_F_COPYIN) {
726 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
732 error = (*bc->bc_func)(sc, &args);
737 if (bc->bc_flags & BC_F_COPYOUT)
738 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
743 if (!(ifp->if_flags & IFF_UP) &&
744 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
746 * If interface is marked down and it is running,
747 * then stop and disable it.
752 } else if ((ifp->if_flags & IFF_UP) &&
753 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
755 * If interface is marked up and it is stopped, then
763 /* Do not allow the MTU to be changed on the bridge */
769 * drop the lock as ether_ioctl() will call bridge_start() and
770 * cause the lock to be recursed.
772 error = ether_ioctl(ifp, cmd, data);
782 * Clear or restore unwanted capabilities on the member interface
785 bridge_mutecaps(struct bridge_softc *sc)
787 struct bridge_iflist *bif;
790 /* Initial bitmask of capabilities to test */
791 mask = BRIDGE_IFCAPS_MASK;
793 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
794 /* Every member must support it or its disabled */
795 mask &= bif->bif_savedcaps;
798 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
799 enabled = bif->bif_ifp->if_capenable;
800 enabled &= ~BRIDGE_IFCAPS_STRIP;
801 /* strip off mask bits and enable them again if allowed */
802 enabled &= ~BRIDGE_IFCAPS_MASK;
805 bridge_set_ifcap(sc, bif, enabled);
811 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
813 struct ifnet *ifp = bif->bif_ifp;
817 bzero(&ifr, sizeof(ifr));
818 ifr.ifr_reqcap = set;
820 if (ifp->if_capenable != set) {
822 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
823 IFF_UNLOCKGIANT(ifp);
825 if_printf(sc->sc_ifp,
826 "error setting interface capabilities on %s\n",
832 * bridge_lookup_member:
834 * Lookup a bridge member interface.
836 static struct bridge_iflist *
837 bridge_lookup_member(struct bridge_softc *sc, const char *name)
839 struct bridge_iflist *bif;
842 BRIDGE_LOCK_ASSERT(sc);
844 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
846 if (strcmp(ifp->if_xname, name) == 0)
854 * bridge_lookup_member_if:
856 * Lookup a bridge member interface by ifnet*.
858 static struct bridge_iflist *
859 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
861 struct bridge_iflist *bif;
863 BRIDGE_LOCK_ASSERT(sc);
865 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
866 if (bif->bif_ifp == member_ifp)
874 * bridge_delete_member:
876 * Delete the specified member interface.
879 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
882 struct ifnet *ifs = bif->bif_ifp;
884 BRIDGE_LOCK_ASSERT(sc);
887 switch (ifs->if_type) {
891 * Take the interface out of promiscuous mode.
893 (void) ifpromisc(ifs, 0);
901 panic("bridge_delete_member: impossible");
905 /* reneable any interface capabilities */
906 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
909 if (bif->bif_flags & IFBIF_STP)
910 bstp_disable(&bif->bif_stp);
912 ifs->if_bridge = NULL;
914 LIST_REMOVE(bif, bif_next);
917 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
918 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
919 KASSERT(bif->bif_addrcnt == 0,
920 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
923 bstp_destroy(&bif->bif_stp); /* prepare to free */
929 * bridge_delete_span:
931 * Delete the specified span interface.
934 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
936 BRIDGE_LOCK_ASSERT(sc);
938 KASSERT(bif->bif_ifp->if_bridge == NULL,
939 ("%s: not a span interface", __func__));
941 LIST_REMOVE(bif, bif_next);
946 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
948 struct ifbreq *req = arg;
949 struct bridge_iflist *bif = NULL;
953 ifs = ifunit(req->ifbr_ifsname);
956 if (ifs->if_ioctl == NULL) /* must be supported */
959 /* If it's in the span list, it can't be a member. */
960 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
961 if (ifs == bif->bif_ifp)
964 /* Allow the first Ethernet member to define the MTU */
965 if (ifs->if_type != IFT_GIF) {
966 if (LIST_EMPTY(&sc->sc_iflist))
967 sc->sc_ifp->if_mtu = ifs->if_mtu;
968 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
969 if_printf(sc->sc_ifp, "invalid MTU for %s\n",
975 if (ifs->if_bridge == sc)
978 if (ifs->if_bridge != NULL)
981 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
986 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
987 bif->bif_savedcaps = ifs->if_capenable;
989 switch (ifs->if_type) {
993 * Place the interface into promiscuous mode.
995 error = ifpromisc(ifs, 1);
1008 ifs->if_bridge = sc;
1009 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1011 * XXX: XLOCK HERE!?!
1013 * NOTE: insert_***HEAD*** should be safe for the traversals.
1015 LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1017 /* Set interface capabilities to the intersection set of all members */
1018 bridge_mutecaps(sc);
1022 free(bif, M_DEVBUF);
1028 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1030 struct ifbreq *req = arg;
1031 struct bridge_iflist *bif;
1033 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1037 bridge_delete_member(sc, bif, 0);
1043 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1045 struct ifbreq *req = arg;
1046 struct bridge_iflist *bif;
1047 struct bstp_port *bp;
1049 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1054 req->ifbr_ifsflags = bif->bif_flags;
1055 req->ifbr_state = bp->bp_state;
1056 req->ifbr_priority = bp->bp_priority;
1057 req->ifbr_path_cost = bp->bp_path_cost;
1058 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1059 req->ifbr_proto = bp->bp_protover;
1060 req->ifbr_role = bp->bp_role;
1061 req->ifbr_stpflags = bp->bp_flags;
1062 req->ifbr_addrcnt = bif->bif_addrcnt;
1063 req->ifbr_addrmax = bif->bif_addrmax;
1064 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1066 /* Copy STP state options as flags */
1067 if (bp->bp_operedge)
1068 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1069 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1070 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1071 if (bp->bp_ptp_link)
1072 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1073 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1074 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1075 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1076 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1077 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1078 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1083 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1085 struct ifbreq *req = arg;
1086 struct bridge_iflist *bif;
1087 struct bstp_port *bp;
1090 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1095 if (req->ifbr_ifsflags & IFBIF_SPAN)
1096 /* SPAN is readonly */
1099 if (req->ifbr_ifsflags & IFBIF_STP) {
1100 if ((bif->bif_flags & IFBIF_STP) == 0) {
1101 error = bstp_enable(&bif->bif_stp);
1106 if ((bif->bif_flags & IFBIF_STP) != 0)
1107 bstp_disable(&bif->bif_stp);
1110 /* Pass on STP flags */
1111 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1112 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1113 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1114 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1116 /* Save the bits relating to the bridge */
1117 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1123 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1125 struct ifbrparam *param = arg;
1127 sc->sc_brtmax = param->ifbrp_csize;
1134 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1136 struct ifbrparam *param = arg;
1138 param->ifbrp_csize = sc->sc_brtmax;
1144 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1146 struct ifbifconf *bifc = arg;
1147 struct bridge_iflist *bif;
1150 int count, buflen, len, error = 0;
1153 LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1155 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1158 buflen = sizeof(breq) * count;
1159 if (bifc->ifbic_len == 0) {
1160 bifc->ifbic_len = buflen;
1164 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1169 len = min(bifc->ifbic_len, buflen);
1170 bzero(&breq, sizeof(breq));
1171 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1172 if (len < sizeof(breq))
1175 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1176 sizeof(breq.ifbr_ifsname));
1177 /* Fill in the ifbreq structure */
1178 error = bridge_ioctl_gifflags(sc, &breq);
1181 memcpy(buf, &breq, sizeof(breq));
1183 buf += sizeof(breq);
1184 len -= sizeof(breq);
1186 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1187 if (len < sizeof(breq))
1190 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1191 sizeof(breq.ifbr_ifsname));
1192 breq.ifbr_ifsflags = bif->bif_flags;
1193 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1194 memcpy(buf, &breq, sizeof(breq));
1196 buf += sizeof(breq);
1197 len -= sizeof(breq);
1201 bifc->ifbic_len = sizeof(breq) * count;
1202 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1204 free(outbuf, M_TEMP);
1209 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1211 struct ifbaconf *bac = arg;
1212 struct bridge_rtnode *brt;
1213 struct ifbareq bareq;
1215 int count, buflen, len, error = 0;
1217 if (bac->ifbac_len == 0)
1221 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1223 buflen = sizeof(bareq) * count;
1226 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1231 len = min(bac->ifbac_len, buflen);
1232 bzero(&bareq, sizeof(bareq));
1233 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1234 if (len < sizeof(bareq))
1236 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1237 sizeof(bareq.ifba_ifsname));
1238 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1239 bareq.ifba_vlan = brt->brt_vlan;
1240 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1241 time_uptime < brt->brt_expire)
1242 bareq.ifba_expire = brt->brt_expire - time_uptime;
1244 bareq.ifba_expire = 0;
1245 bareq.ifba_flags = brt->brt_flags;
1247 memcpy(buf, &bareq, sizeof(bareq));
1249 buf += sizeof(bareq);
1250 len -= sizeof(bareq);
1254 bac->ifbac_len = sizeof(bareq) * count;
1255 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1257 free(outbuf, M_TEMP);
1262 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1264 struct ifbareq *req = arg;
1265 struct bridge_iflist *bif;
1268 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1272 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1279 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1281 struct ifbrparam *param = arg;
1283 sc->sc_brttimeout = param->ifbrp_ctime;
1288 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1290 struct ifbrparam *param = arg;
1292 param->ifbrp_ctime = sc->sc_brttimeout;
1297 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1299 struct ifbareq *req = arg;
1301 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1305 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1307 struct ifbreq *req = arg;
1309 bridge_rtflush(sc, req->ifbr_ifsflags);
1314 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1316 struct ifbrparam *param = arg;
1317 struct bstp_state *bs = &sc->sc_stp;
1319 param->ifbrp_prio = bs->bs_bridge_priority;
1324 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1326 struct ifbrparam *param = arg;
1328 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1332 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1334 struct ifbrparam *param = arg;
1335 struct bstp_state *bs = &sc->sc_stp;
1337 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1342 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1344 struct ifbrparam *param = arg;
1346 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1350 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1352 struct ifbrparam *param = arg;
1353 struct bstp_state *bs = &sc->sc_stp;
1355 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1360 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1362 struct ifbrparam *param = arg;
1364 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1368 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1370 struct ifbrparam *param = arg;
1371 struct bstp_state *bs = &sc->sc_stp;
1373 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1378 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1380 struct ifbrparam *param = arg;
1382 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1386 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1388 struct ifbreq *req = arg;
1389 struct bridge_iflist *bif;
1391 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1395 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1399 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1401 struct ifbreq *req = arg;
1402 struct bridge_iflist *bif;
1404 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1408 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1412 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1414 struct ifbreq *req = arg;
1415 struct bridge_iflist *bif;
1417 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1421 bif->bif_addrmax = req->ifbr_addrmax;
1426 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1428 struct ifbreq *req = arg;
1429 struct bridge_iflist *bif = NULL;
1432 ifs = ifunit(req->ifbr_ifsname);
1436 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1437 if (ifs == bif->bif_ifp)
1440 if (ifs->if_bridge != NULL)
1443 switch (ifs->if_type) {
1452 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1457 bif->bif_flags = IFBIF_SPAN;
1459 LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1465 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1467 struct ifbreq *req = arg;
1468 struct bridge_iflist *bif;
1471 ifs = ifunit(req->ifbr_ifsname);
1475 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1476 if (ifs == bif->bif_ifp)
1482 bridge_delete_span(sc, bif);
1488 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1490 struct ifbropreq *req = arg;
1491 struct bstp_state *bs = &sc->sc_stp;
1492 struct bstp_port *root_port;
1494 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1495 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1496 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1498 root_port = bs->bs_root_port;
1499 if (root_port == NULL)
1500 req->ifbop_root_port = 0;
1502 req->ifbop_root_port = root_port->bp_ifp->if_index;
1504 req->ifbop_holdcount = bs->bs_txholdcount;
1505 req->ifbop_priority = bs->bs_bridge_priority;
1506 req->ifbop_protocol = bs->bs_protover;
1507 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1508 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1509 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1510 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1511 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1512 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1518 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1520 struct ifbrparam *param = arg;
1522 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1527 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1529 struct ifbpstpconf *bifstp = arg;
1530 struct bridge_iflist *bif;
1531 struct bstp_port *bp;
1532 struct ifbpstpreq bpreq;
1534 int count, buflen, len, error = 0;
1537 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1538 if ((bif->bif_flags & IFBIF_STP) != 0)
1542 buflen = sizeof(bpreq) * count;
1543 if (bifstp->ifbpstp_len == 0) {
1544 bifstp->ifbpstp_len = buflen;
1549 outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1554 len = min(bifstp->ifbpstp_len, buflen);
1555 bzero(&bpreq, sizeof(bpreq));
1556 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1557 if (len < sizeof(bpreq))
1560 if ((bif->bif_flags & IFBIF_STP) == 0)
1564 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1565 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1566 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1567 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1568 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1569 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1571 memcpy(buf, &bpreq, sizeof(bpreq));
1573 buf += sizeof(bpreq);
1574 len -= sizeof(bpreq);
1578 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1579 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1581 free(outbuf, M_TEMP);
1586 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1588 struct ifbrparam *param = arg;
1590 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1594 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1596 struct ifbrparam *param = arg;
1598 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1604 * Detach an interface from a bridge. Called when a member
1605 * interface is detaching.
1608 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1610 struct bridge_softc *sc = ifp->if_bridge;
1611 struct bridge_iflist *bif;
1613 /* Check if the interface is a bridge member */
1617 bif = bridge_lookup_member_if(sc, ifp);
1619 bridge_delete_member(sc, bif, 1);
1625 /* Check if the interface is a span port */
1626 mtx_lock(&bridge_list_mtx);
1627 LIST_FOREACH(sc, &bridge_list, sc_list) {
1629 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1630 if (ifp == bif->bif_ifp) {
1631 bridge_delete_span(sc, bif);
1637 mtx_unlock(&bridge_list_mtx);
1643 * Initialize a bridge interface.
1646 bridge_init(void *xsc)
1648 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1649 struct ifnet *ifp = sc->sc_ifp;
1651 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1655 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1658 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1659 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1667 * Stop the bridge interface.
1670 bridge_stop(struct ifnet *ifp, int disable)
1672 struct bridge_softc *sc = ifp->if_softc;
1674 BRIDGE_LOCK_ASSERT(sc);
1676 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1679 callout_stop(&sc->sc_brcallout);
1680 bstp_stop(&sc->sc_stp);
1682 bridge_rtflush(sc, IFBF_FLUSHDYN);
1684 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1690 * Enqueue a packet on a bridge member interface.
1694 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1700 len = m->m_pkthdr.len;
1701 mflags = m->m_flags;
1703 /* We may be sending a fragment so traverse the mbuf */
1706 m->m_nextpkt = NULL;
1709 * If underlying interface can not do VLAN tag insertion itself
1710 * then attach a packet tag that holds it.
1712 if ((m->m_flags & M_VLANTAG) &&
1713 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1714 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1717 "unable to prepend VLAN header\n");
1718 dst_ifp->if_oerrors++;
1721 m->m_flags &= ~M_VLANTAG;
1725 IFQ_ENQUEUE(&dst_ifp->if_snd, m, err);
1730 sc->sc_ifp->if_opackets++;
1731 sc->sc_ifp->if_obytes += len;
1733 dst_ifp->if_obytes += len;
1735 if (mflags & M_MCAST) {
1736 sc->sc_ifp->if_omcasts++;
1737 dst_ifp->if_omcasts++;
1741 if ((dst_ifp->if_drv_flags & IFF_DRV_OACTIVE) == 0)
1742 (*dst_ifp->if_start)(dst_ifp);
1748 * Receive a queued packet from dummynet and pass it on to the output
1751 * The mbuf has the Ethernet header already attached.
1754 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1756 struct bridge_softc *sc;
1758 sc = ifp->if_bridge;
1761 * The packet didnt originate from a member interface. This should only
1762 * ever happen if a member interface is removed while packets are
1770 if (PFIL_HOOKED(&inet_pfil_hook)
1772 || PFIL_HOOKED(&inet6_pfil_hook)
1775 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1781 bridge_enqueue(sc, ifp, m);
1787 * Send output from a bridge member interface. This
1788 * performs the bridging function for locally originated
1791 * The mbuf has the Ethernet header already attached. We must
1792 * enqueue or free the mbuf before returning.
1795 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1798 struct ether_header *eh;
1799 struct ifnet *dst_if;
1800 struct bridge_softc *sc;
1803 if (m->m_len < ETHER_HDR_LEN) {
1804 m = m_pullup(m, ETHER_HDR_LEN);
1809 eh = mtod(m, struct ether_header *);
1810 sc = ifp->if_bridge;
1811 vlan = VLANTAGOF(m);
1816 * If bridge is down, but the original output interface is up,
1817 * go ahead and send out that interface. Otherwise, the packet
1820 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1826 * If the packet is a multicast, or we don't know a better way to
1827 * get there, send to all interfaces.
1829 if (ETHER_IS_MULTICAST(eh->ether_dhost))
1832 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1833 if (dst_if == NULL) {
1834 struct bridge_iflist *bif;
1836 int error = 0, used = 0;
1840 BRIDGE_LOCK2REF(sc, error);
1846 LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1847 dst_if = bif->bif_ifp;
1849 if (dst_if->if_type == IFT_GIF)
1851 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1855 * If this is not the original output interface,
1856 * and the interface is participating in spanning
1857 * tree, make sure the port is in a state that
1858 * allows forwarding.
1860 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1861 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1864 if (LIST_NEXT(bif, bif_next) == NULL) {
1868 mc = m_copypacket(m, M_DONTWAIT);
1870 sc->sc_ifp->if_oerrors++;
1875 bridge_enqueue(sc, dst_if, mc);
1885 * XXX Spanning tree consideration here?
1889 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1896 bridge_enqueue(sc, dst_if, m);
1903 * Start output on a bridge.
1907 bridge_start(struct ifnet *ifp)
1909 struct bridge_softc *sc;
1911 struct ether_header *eh;
1912 struct ifnet *dst_if;
1916 ifp->if_drv_flags |= IFF_DRV_OACTIVE;
1918 IFQ_DEQUEUE(&ifp->if_snd, m);
1921 ETHER_BPF_MTAP(ifp, m);
1923 eh = mtod(m, struct ether_header *);
1927 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
1928 dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1);
1932 bridge_broadcast(sc, ifp, m, 0);
1935 bridge_enqueue(sc, dst_if, m);
1938 ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
1944 * The forwarding function of the bridge.
1946 * NOTE: Releases the lock on return.
1949 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
1952 struct bridge_iflist *dbif;
1953 struct ifnet *src_if, *dst_if, *ifp;
1954 struct ether_header *eh;
1959 src_if = m->m_pkthdr.rcvif;
1963 ifp->if_ibytes += m->m_pkthdr.len;
1964 vlan = VLANTAGOF(m);
1966 if ((sbif->bif_flags & IFBIF_STP) &&
1967 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1970 eh = mtod(m, struct ether_header *);
1971 dst = eh->ether_dhost;
1973 /* If the interface is learning, record the address. */
1974 if (sbif->bif_flags & IFBIF_LEARNING) {
1975 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
1976 sbif, 0, IFBAF_DYNAMIC);
1978 * If the interface has addresses limits then deny any source
1979 * that is not in the cache.
1981 if (error && sbif->bif_addrmax)
1985 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
1986 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
1990 * At this point, the port either doesn't participate
1991 * in spanning tree or it is in the forwarding state.
1995 * If the packet is unicast, destined for someone on
1996 * "this" side of the bridge, drop it.
1998 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
1999 dst_if = bridge_rtlookup(sc, dst, vlan);
2000 if (src_if == dst_if)
2004 * Check if its a reserved multicast address, any address
2005 * listed in 802.1D section 7.12.6 may not be forwarded by the
2007 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2009 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2010 dst[2] == 0xc2 && dst[3] == 0x00 &&
2011 dst[4] == 0x00 && dst[5] <= 0x0f)
2014 /* ...forward it to all interfaces. */
2020 * If we have a destination interface which is a member of our bridge,
2021 * OR this is a unicast packet, push it through the bpf(4) machinery.
2022 * For broadcast or multicast packets, don't bother because it will
2023 * be reinjected into ether_input. We do this before we pass the packets
2024 * through the pfil(9) framework, as it is possible that pfil(9) will
2025 * drop the packet, or possibly modify it, making it difficult to debug
2026 * firewall issues on the bridge.
2028 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2029 ETHER_BPF_MTAP(ifp, m);
2031 /* run the packet filter */
2032 if (PFIL_HOOKED(&inet_pfil_hook)
2034 || PFIL_HOOKED(&inet6_pfil_hook)
2038 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2045 if (dst_if == NULL) {
2046 bridge_broadcast(sc, src_if, m, 1);
2051 * At this point, we're dealing with a unicast frame
2052 * going to a different interface.
2054 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2057 dbif = bridge_lookup_member_if(sc, dst_if);
2059 /* Not a member of the bridge (anymore?) */
2062 /* Private segments can not talk to each other */
2063 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2066 if ((dbif->bif_flags & IFBIF_STP) &&
2067 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2072 if (PFIL_HOOKED(&inet_pfil_hook)
2074 || PFIL_HOOKED(&inet6_pfil_hook)
2077 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2083 bridge_enqueue(sc, dst_if, m);
2094 * Receive input from a member interface. Queue the packet for
2095 * bridging if it is not for us.
2097 static struct mbuf *
2098 bridge_input(struct ifnet *ifp, struct mbuf *m)
2100 struct bridge_softc *sc = ifp->if_bridge;
2101 struct bridge_iflist *bif, *bif2;
2103 struct ether_header *eh;
2104 struct mbuf *mc, *mc2;
2108 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2112 vlan = VLANTAGOF(m);
2115 * Implement support for bridge monitoring. If this flag has been
2116 * set on this interface, discard the packet once we push it through
2117 * the bpf(4) machinery, but before we do, increment the byte and
2118 * packet counters associated with this interface.
2120 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2121 m->m_pkthdr.rcvif = bifp;
2122 ETHER_BPF_MTAP(bifp, m);
2123 bifp->if_ipackets++;
2124 bifp->if_ibytes += m->m_pkthdr.len;
2129 bif = bridge_lookup_member_if(sc, ifp);
2135 eh = mtod(m, struct ether_header *);
2139 if (m->m_flags & (M_BCAST|M_MCAST)) {
2140 /* Tap off 802.1D packets; they do not get forwarded. */
2141 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2142 ETHER_ADDR_LEN) == 0) {
2143 m = bstp_input(&bif->bif_stp, ifp, m);
2150 if ((bif->bif_flags & IFBIF_STP) &&
2151 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2157 * Make a deep copy of the packet and enqueue the copy
2158 * for bridge processing; return the original packet for
2161 mc = m_dup(m, M_DONTWAIT);
2167 /* Perform the bridge forwarding function with the copy. */
2168 bridge_forward(sc, bif, mc);
2171 * Reinject the mbuf as arriving on the bridge so we have a
2172 * chance at claiming multicast packets. We can not loop back
2173 * here from ether_input as a bridge is never a member of a
2176 KASSERT(bifp->if_bridge == NULL,
2177 ("loop created in bridge_input"));
2178 mc2 = m_dup(m, M_DONTWAIT);
2180 /* Keep the layer3 header aligned */
2181 int i = min(mc2->m_pkthdr.len, max_protohdr);
2182 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2185 mc2->m_pkthdr.rcvif = bifp;
2186 (*bifp->if_input)(bifp, mc2);
2189 /* Return the original packet for local processing. */
2193 if ((bif->bif_flags & IFBIF_STP) &&
2194 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2200 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2201 || ((iface)->if_carp \
2202 && carp_forus((iface)->if_carp, eh->ether_dhost))
2203 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2204 || ((iface)->if_carp \
2205 && carp_forus((iface)->if_carp, eh->ether_shost))
2207 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2208 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2212 # define OR_PFIL_HOOKED_INET6 \
2213 || PFIL_HOOKED(&inet6_pfil_hook)
2215 # define OR_PFIL_HOOKED_INET6
2218 #define GRAB_OUR_PACKETS(iface) \
2219 if ((iface)->if_type == IFT_GIF) \
2221 /* It is destined for us. */ \
2222 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2223 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2225 if ((iface)->if_type == IFT_BRIDGE) { \
2226 ETHER_BPF_MTAP(iface, m); \
2227 iface->if_ipackets++; \
2228 /* Filter on the physical interface. */ \
2229 if (pfil_local_phys && \
2230 (PFIL_HOOKED(&inet_pfil_hook) \
2231 OR_PFIL_HOOKED_INET6)) { \
2232 if (bridge_pfil(&m, NULL, ifp, \
2233 PFIL_IN) != 0 || m == NULL) { \
2234 BRIDGE_UNLOCK(sc); \
2239 if (bif->bif_flags & IFBIF_LEARNING) { \
2240 error = bridge_rtupdate(sc, eh->ether_shost, \
2241 vlan, bif, 0, IFBAF_DYNAMIC); \
2242 if (error && bif->bif_addrmax) { \
2243 BRIDGE_UNLOCK(sc); \
2248 m->m_pkthdr.rcvif = iface; \
2249 BRIDGE_UNLOCK(sc); \
2253 /* We just received a packet that we sent out. */ \
2254 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2255 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2257 BRIDGE_UNLOCK(sc); \
2263 * Unicast. Make sure it's not for the bridge.
2265 do { GRAB_OUR_PACKETS(bifp) } while (0);
2268 * Give a chance for ifp at first priority. This will help when the
2269 * packet comes through the interface like VLAN's with the same MACs
2270 * on several interfaces from the same bridge. This also will save
2271 * some CPU cycles in case the destination interface and the input
2272 * interface (eq ifp) are the same.
2274 do { GRAB_OUR_PACKETS(ifp) } while (0);
2276 /* Now check the all bridge members. */
2277 LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2278 GRAB_OUR_PACKETS(bif2->bif_ifp)
2281 #undef OR_CARP_CHECK_WE_ARE_DST
2282 #undef OR_CARP_CHECK_WE_ARE_SRC
2283 #undef OR_PFIL_HOOKED_INET6
2284 #undef GRAB_OUR_PACKETS
2286 /* Perform the bridge forwarding function. */
2287 bridge_forward(sc, bif, m);
2295 * Send a frame to all interfaces that are members of
2296 * the bridge, except for the one on which the packet
2299 * NOTE: Releases the lock on return.
2302 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2303 struct mbuf *m, int runfilt)
2305 struct bridge_iflist *dbif, *sbif;
2307 struct ifnet *dst_if;
2308 int error = 0, used = 0, i;
2310 sbif = bridge_lookup_member_if(sc, src_if);
2312 BRIDGE_LOCK2REF(sc, error);
2318 /* Filter on the bridge interface before broadcasting */
2319 if (runfilt && (PFIL_HOOKED(&inet_pfil_hook)
2321 || PFIL_HOOKED(&inet6_pfil_hook)
2324 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2330 LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2331 dst_if = dbif->bif_ifp;
2332 if (dst_if == src_if)
2335 /* Private segments can not talk to each other */
2336 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2339 if ((dbif->bif_flags & IFBIF_STP) &&
2340 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2343 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2344 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2347 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2350 if (LIST_NEXT(dbif, bif_next) == NULL) {
2354 mc = m_dup(m, M_DONTWAIT);
2356 sc->sc_ifp->if_oerrors++;
2362 * Filter on the output interface. Pass a NULL bridge interface
2363 * pointer so we do not redundantly filter on the bridge for
2364 * each interface we broadcast on.
2366 if (runfilt && (PFIL_HOOKED(&inet_pfil_hook)
2368 || PFIL_HOOKED(&inet6_pfil_hook)
2372 /* Keep the layer3 header aligned */
2373 i = min(mc->m_pkthdr.len, max_protohdr);
2374 mc = m_copyup(mc, i, ETHER_ALIGN);
2376 sc->sc_ifp->if_oerrors++;
2380 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2386 bridge_enqueue(sc, dst_if, mc);
2398 * Duplicate a packet out one or more interfaces that are in span mode,
2399 * the original mbuf is unmodified.
2402 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2404 struct bridge_iflist *bif;
2405 struct ifnet *dst_if;
2408 if (LIST_EMPTY(&sc->sc_spanlist))
2411 LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2412 dst_if = bif->bif_ifp;
2414 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2417 mc = m_copypacket(m, M_DONTWAIT);
2419 sc->sc_ifp->if_oerrors++;
2423 bridge_enqueue(sc, dst_if, mc);
2430 * Add a bridge routing entry.
2433 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2434 struct bridge_iflist *bif, int setflags, uint8_t flags)
2436 struct bridge_rtnode *brt;
2439 BRIDGE_LOCK_ASSERT(sc);
2441 /* Check the source address is valid and not multicast. */
2442 if (ETHER_IS_MULTICAST(dst) ||
2443 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2444 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2447 /* 802.1p frames map to vlan 1 */
2452 * A route for this destination might already exist. If so,
2453 * update it, otherwise create a new one.
2455 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2456 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2457 sc->sc_brtexceeded++;
2460 /* Check per interface address limits (if enabled) */
2461 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2462 bif->bif_addrexceeded++;
2467 * Allocate a new bridge forwarding node, and
2468 * initialize the expiration time and Ethernet
2471 brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2475 if (bif->bif_flags & IFBIF_STICKY)
2476 brt->brt_flags = IFBAF_STICKY;
2478 brt->brt_flags = IFBAF_DYNAMIC;
2480 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2481 brt->brt_vlan = vlan;
2483 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2484 uma_zfree(bridge_rtnode_zone, brt);
2491 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2492 brt->brt_dst != bif) {
2493 brt->brt_dst->bif_addrcnt--;
2495 brt->brt_dst->bif_addrcnt++;
2498 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2499 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2501 brt->brt_flags = flags;
2509 * Lookup the destination interface for an address.
2511 static struct ifnet *
2512 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2514 struct bridge_rtnode *brt;
2516 BRIDGE_LOCK_ASSERT(sc);
2518 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2521 return (brt->brt_ifp);
2527 * Trim the routine table so that we have a number
2528 * of routing entries less than or equal to the
2532 bridge_rttrim(struct bridge_softc *sc)
2534 struct bridge_rtnode *brt, *nbrt;
2536 BRIDGE_LOCK_ASSERT(sc);
2538 /* Make sure we actually need to do this. */
2539 if (sc->sc_brtcnt <= sc->sc_brtmax)
2542 /* Force an aging cycle; this might trim enough addresses. */
2544 if (sc->sc_brtcnt <= sc->sc_brtmax)
2547 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2548 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2549 bridge_rtnode_destroy(sc, brt);
2550 if (sc->sc_brtcnt <= sc->sc_brtmax)
2559 * Aging timer for the bridge.
2562 bridge_timer(void *arg)
2564 struct bridge_softc *sc = arg;
2566 BRIDGE_LOCK_ASSERT(sc);
2570 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2571 callout_reset(&sc->sc_brcallout,
2572 bridge_rtable_prune_period * hz, bridge_timer, sc);
2578 * Perform an aging cycle.
2581 bridge_rtage(struct bridge_softc *sc)
2583 struct bridge_rtnode *brt, *nbrt;
2585 BRIDGE_LOCK_ASSERT(sc);
2587 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2588 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2589 if (time_uptime >= brt->brt_expire)
2590 bridge_rtnode_destroy(sc, brt);
2598 * Remove all dynamic addresses from the bridge.
2601 bridge_rtflush(struct bridge_softc *sc, int full)
2603 struct bridge_rtnode *brt, *nbrt;
2605 BRIDGE_LOCK_ASSERT(sc);
2607 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2608 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2609 bridge_rtnode_destroy(sc, brt);
2616 * Remove an address from the table.
2619 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2621 struct bridge_rtnode *brt;
2624 BRIDGE_LOCK_ASSERT(sc);
2627 * If vlan is zero then we want to delete for all vlans so the lookup
2628 * may return more than one.
2630 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2631 bridge_rtnode_destroy(sc, brt);
2635 return (found ? 0 : ENOENT);
2641 * Delete routes to a speicifc member interface.
2644 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2646 struct bridge_rtnode *brt, *nbrt;
2648 BRIDGE_LOCK_ASSERT(sc);
2650 LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2651 if (brt->brt_ifp == ifp && (full ||
2652 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2653 bridge_rtnode_destroy(sc, brt);
2658 * bridge_rtable_init:
2660 * Initialize the route table for this bridge.
2663 bridge_rtable_init(struct bridge_softc *sc)
2667 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2668 M_DEVBUF, M_NOWAIT);
2669 if (sc->sc_rthash == NULL)
2672 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2673 LIST_INIT(&sc->sc_rthash[i]);
2675 sc->sc_rthash_key = arc4random();
2677 LIST_INIT(&sc->sc_rtlist);
2683 * bridge_rtable_fini:
2685 * Deconstruct the route table for this bridge.
2688 bridge_rtable_fini(struct bridge_softc *sc)
2691 KASSERT(sc->sc_brtcnt == 0,
2692 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2693 free(sc->sc_rthash, M_DEVBUF);
2697 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2698 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2700 #define mix(a, b, c) \
2702 a -= b; a -= c; a ^= (c >> 13); \
2703 b -= c; b -= a; b ^= (a << 8); \
2704 c -= a; c -= b; c ^= (b >> 13); \
2705 a -= b; a -= c; a ^= (c >> 12); \
2706 b -= c; b -= a; b ^= (a << 16); \
2707 c -= a; c -= b; c ^= (b >> 5); \
2708 a -= b; a -= c; a ^= (c >> 3); \
2709 b -= c; b -= a; b ^= (a << 10); \
2710 c -= a; c -= b; c ^= (b >> 15); \
2711 } while (/*CONSTCOND*/0)
2713 static __inline uint32_t
2714 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2716 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2727 return (c & BRIDGE_RTHASH_MASK);
2733 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2737 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2738 d = ((int)a[i]) - ((int)b[i]);
2745 * bridge_rtnode_lookup:
2747 * Look up a bridge route node for the specified destination. Compare the
2748 * vlan id or if zero then just return the first match.
2750 static struct bridge_rtnode *
2751 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2753 struct bridge_rtnode *brt;
2757 BRIDGE_LOCK_ASSERT(sc);
2759 hash = bridge_rthash(sc, addr);
2760 LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2761 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2762 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2772 * bridge_rtnode_insert:
2774 * Insert the specified bridge node into the route table. We
2775 * assume the entry is not already in the table.
2778 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2780 struct bridge_rtnode *lbrt;
2784 BRIDGE_LOCK_ASSERT(sc);
2786 hash = bridge_rthash(sc, brt->brt_addr);
2788 lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2790 LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2795 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2796 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2799 LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2802 if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2803 LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2806 lbrt = LIST_NEXT(lbrt, brt_hash);
2807 } while (lbrt != NULL);
2810 panic("bridge_rtnode_insert: impossible");
2814 LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2821 * bridge_rtnode_destroy:
2823 * Destroy a bridge rtnode.
2826 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2828 BRIDGE_LOCK_ASSERT(sc);
2830 LIST_REMOVE(brt, brt_hash);
2832 LIST_REMOVE(brt, brt_list);
2834 brt->brt_dst->bif_addrcnt--;
2835 uma_zfree(bridge_rtnode_zone, brt);
2839 * bridge_rtable_expire:
2841 * Set the expiry time for all routes on an interface.
2844 bridge_rtable_expire(struct ifnet *ifp, int age)
2846 struct bridge_softc *sc = ifp->if_bridge;
2847 struct bridge_rtnode *brt;
2852 * If the age is zero then flush, otherwise set all the expiry times to
2853 * age for the interface
2856 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2858 LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2859 /* Cap the expiry time to 'age' */
2860 if (brt->brt_ifp == ifp &&
2861 brt->brt_expire > time_uptime + age &&
2862 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2863 brt->brt_expire = time_uptime + age;
2870 * bridge_state_change:
2872 * Callback from the bridgestp code when a port changes states.
2875 bridge_state_change(struct ifnet *ifp, int state)
2877 struct bridge_softc *sc = ifp->if_bridge;
2878 static const char *stpstates[] = {
2888 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2889 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2893 * Send bridge packets through pfil if they are one of the types pfil can deal
2894 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
2895 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2899 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2901 int snap, error, i, hlen;
2902 struct ether_header *eh1, eh2;
2903 struct ip_fw_args args;
2906 u_int16_t ether_type;
2909 error = -1; /* Default error if not error == 0 */
2912 /* we may return with the IP fields swapped, ensure its not shared */
2913 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2916 if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2917 return (0); /* filtering is disabled */
2919 i = min((*mp)->m_pkthdr.len, max_protohdr);
2920 if ((*mp)->m_len < i) {
2921 *mp = m_pullup(*mp, i);
2923 printf("%s: m_pullup failed\n", __func__);
2928 eh1 = mtod(*mp, struct ether_header *);
2929 ether_type = ntohs(eh1->ether_type);
2932 * Check for SNAP/LLC.
2934 if (ether_type < ETHERMTU) {
2935 struct llc *llc2 = (struct llc *)(eh1 + 1);
2937 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
2938 llc2->llc_dsap == LLC_SNAP_LSAP &&
2939 llc2->llc_ssap == LLC_SNAP_LSAP &&
2940 llc2->llc_control == LLC_UI) {
2941 ether_type = htons(llc2->llc_un.type_snap.ether_type);
2947 * If we're trying to filter bridge traffic, don't look at anything
2948 * other than IP and ARP traffic. If the filter doesn't understand
2949 * IPv6, don't allow IPv6 through the bridge either. This is lame
2950 * since if we really wanted, say, an AppleTalk filter, we are hosed,
2951 * but of course we don't have an AppleTalk filter to begin with.
2952 * (Note that since pfil doesn't understand ARP it will pass *ALL*
2955 switch (ether_type) {
2957 case ETHERTYPE_REVARP:
2958 if (pfil_ipfw_arp == 0)
2959 return (0); /* Automatically pass */
2964 case ETHERTYPE_IPV6:
2969 * Check to see if the user wants to pass non-ip
2970 * packets, these will not be checked by pfil(9) and
2971 * passed unconditionally so the default is to drop.
2977 /* Strip off the Ethernet header and keep a copy. */
2978 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
2979 m_adj(*mp, ETHER_HDR_LEN);
2981 /* Strip off snap header, if present */
2983 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
2984 m_adj(*mp, sizeof(struct llc));
2988 * Check the IP header for alignment and errors
2990 if (dir == PFIL_IN) {
2991 switch (ether_type) {
2993 error = bridge_ip_checkbasic(mp);
2996 case ETHERTYPE_IPV6:
2997 error = bridge_ip6_checkbasic(mp);
3007 if (IPFW_LOADED && pfil_ipfw != 0 && dir == PFIL_OUT && ifp != NULL) {
3009 args.rule = ip_dn_claim_rule(*mp);
3010 if (args.rule != NULL && fw_one_pass)
3011 goto ipfwpass; /* packet already partially processed */
3015 args.next_hop = NULL;
3017 args.inp = NULL; /* used by ipfw uid/gid/jail rules */
3018 i = ip_fw_chk_ptr(&args);
3024 if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) {
3026 /* put the Ethernet header back on */
3027 M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3030 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3033 * Pass the pkt to dummynet, which consumes it. The
3034 * packet will return to us via bridge_dummynet().
3037 ip_dn_io_ptr(mp, DN_TO_IFB_FWD, &args);
3041 if (i != IP_FW_PASS) /* drop */
3049 * Run the packet through pfil
3051 switch (ether_type) {
3054 * before calling the firewall, swap fields the same as
3055 * IP does. here we assume the header is contiguous
3057 ip = mtod(*mp, struct ip *);
3059 ip->ip_len = ntohs(ip->ip_len);
3060 ip->ip_off = ntohs(ip->ip_off);
3063 * Run pfil on the member interface and the bridge, both can
3064 * be skipped by clearing pfil_member or pfil_bridge.
3067 * in_if -> bridge_if -> out_if
3069 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3070 error = pfil_run_hooks(&inet_pfil_hook, mp, bifp,
3073 if (*mp == NULL || error != 0) /* filter may consume */
3076 if (pfil_member && ifp != NULL)
3077 error = pfil_run_hooks(&inet_pfil_hook, mp, ifp,
3080 if (*mp == NULL || error != 0) /* filter may consume */
3083 if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3084 error = pfil_run_hooks(&inet_pfil_hook, mp, bifp,
3087 if (*mp == NULL || error != 0) /* filter may consume */
3090 /* check if we need to fragment the packet */
3091 if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3092 i = (*mp)->m_pkthdr.len;
3093 if (i > ifp->if_mtu) {
3094 error = bridge_fragment(ifp, *mp, &eh2, snap,
3100 /* Recalculate the ip checksum and restore byte ordering */
3101 ip = mtod(*mp, struct ip *);
3102 hlen = ip->ip_hl << 2;
3103 if (hlen < sizeof(struct ip))
3105 if (hlen > (*mp)->m_len) {
3106 if ((*mp = m_pullup(*mp, hlen)) == 0)
3108 ip = mtod(*mp, struct ip *);
3112 ip->ip_len = htons(ip->ip_len);
3113 ip->ip_off = htons(ip->ip_off);
3115 if (hlen == sizeof(struct ip))
3116 ip->ip_sum = in_cksum_hdr(ip);
3118 ip->ip_sum = in_cksum(*mp, hlen);
3122 case ETHERTYPE_IPV6:
3123 if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3124 error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp,
3127 if (*mp == NULL || error != 0) /* filter may consume */
3130 if (pfil_member && ifp != NULL)
3131 error = pfil_run_hooks(&inet6_pfil_hook, mp, ifp,
3134 if (*mp == NULL || error != 0) /* filter may consume */
3137 if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3138 error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp,
3155 * Finally, put everything back the way it was and return
3158 M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT);
3161 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3164 M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3167 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3178 * Perform basic checks on header size since
3179 * pfil assumes ip_input has already processed
3180 * it for it. Cut-and-pasted from ip_input.c.
3181 * Given how simple the IPv6 version is,
3182 * does the IPv4 version really need to be
3185 * XXX Should we update ipstat here, or not?
3186 * XXX Right now we update ipstat but not
3190 bridge_ip_checkbasic(struct mbuf **mp)
3192 struct mbuf *m = *mp;
3200 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3201 if ((m = m_copyup(m, sizeof(struct ip),
3202 (max_linkhdr + 3) & ~3)) == NULL) {
3203 /* XXXJRT new stat, please */
3204 ipstat.ips_toosmall++;
3207 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3208 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3209 ipstat.ips_toosmall++;
3213 ip = mtod(m, struct ip *);
3214 if (ip == NULL) goto bad;
3216 if (ip->ip_v != IPVERSION) {
3217 ipstat.ips_badvers++;
3220 hlen = ip->ip_hl << 2;
3221 if (hlen < sizeof(struct ip)) { /* minimum header length */
3222 ipstat.ips_badhlen++;
3225 if (hlen > m->m_len) {
3226 if ((m = m_pullup(m, hlen)) == 0) {
3227 ipstat.ips_badhlen++;
3230 ip = mtod(m, struct ip *);
3231 if (ip == NULL) goto bad;
3234 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3235 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3237 if (hlen == sizeof(struct ip)) {
3238 sum = in_cksum_hdr(ip);
3240 sum = in_cksum(m, hlen);
3244 ipstat.ips_badsum++;
3248 /* Retrieve the packet length. */
3249 len = ntohs(ip->ip_len);
3252 * Check for additional length bogosity
3255 ipstat.ips_badlen++;
3260 * Check that the amount of data in the buffers
3261 * is as at least much as the IP header would have us expect.
3262 * Drop packet if shorter than we expect.
3264 if (m->m_pkthdr.len < len) {
3265 ipstat.ips_tooshort++;
3269 /* Checks out, proceed */
3280 * Same as above, but for IPv6.
3281 * Cut-and-pasted from ip6_input.c.
3282 * XXX Should we update ip6stat, or not?
3285 bridge_ip6_checkbasic(struct mbuf **mp)
3287 struct mbuf *m = *mp;
3288 struct ip6_hdr *ip6;
3291 * If the IPv6 header is not aligned, slurp it up into a new
3292 * mbuf with space for link headers, in the event we forward
3293 * it. Otherwise, if it is aligned, make sure the entire base
3294 * IPv6 header is in the first mbuf of the chain.
3296 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3297 struct ifnet *inifp = m->m_pkthdr.rcvif;
3298 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3299 (max_linkhdr + 3) & ~3)) == NULL) {
3300 /* XXXJRT new stat, please */
3301 ip6stat.ip6s_toosmall++;
3302 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3305 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3306 struct ifnet *inifp = m->m_pkthdr.rcvif;
3307 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3308 ip6stat.ip6s_toosmall++;
3309 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3314 ip6 = mtod(m, struct ip6_hdr *);
3316 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3317 ip6stat.ip6s_badvers++;
3318 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3322 /* Checks out, proceed */
3335 * Return a fragmented mbuf chain.
3338 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3339 int snap, struct llc *llc)
3345 if (m->m_len < sizeof(struct ip) &&
3346 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3348 ip = mtod(m, struct ip *);
3350 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist,
3355 /* walk the chain and re-add the Ethernet header */
3356 for (m0 = m; m0; m0 = m0->m_nextpkt) {
3359 M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT);
3364 bcopy(llc, mtod(m0, caddr_t),
3365 sizeof(struct llc));
3367 M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT);
3372 bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3378 ipstat.ips_fragmented++;