1 /* $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $ */
4 * SPDX-License-Identifier: BSD-4-Clause
6 * Copyright 2001 Wasabi Systems, Inc.
9 * Written by Jason R. Thorpe for Wasabi Systems, Inc.
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. All advertising materials mentioning features or use of this software
20 * must display the following acknowledgement:
21 * This product includes software developed for the NetBSD Project by
22 * Wasabi Systems, Inc.
23 * 4. The name of Wasabi Systems, Inc. may not be used to endorse
24 * or promote products derived from this software without specific prior
27 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37 * POSSIBILITY OF SUCH DAMAGE.
41 * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
42 * All rights reserved.
44 * Redistribution and use in source and binary forms, with or without
45 * modification, are permitted provided that the following conditions
47 * 1. Redistributions of source code must retain the above copyright
48 * notice, this list of conditions and the following disclaimer.
49 * 2. Redistributions in binary form must reproduce the above copyright
50 * notice, this list of conditions and the following disclaimer in the
51 * documentation and/or other materials provided with the distribution.
53 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
54 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
55 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
56 * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
57 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
58 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
59 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
61 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
62 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
63 * POSSIBILITY OF SUCH DAMAGE.
65 * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
69 * Network interface bridge support.
73 * - Currently only supports Ethernet-like interfaces (Ethernet,
74 * 802.11, VLANs on Ethernet, etc.) Figure out a nice way
75 * to bridge other types of interfaces (maybe consider
76 * heterogeneous bridges).
79 #include <sys/cdefs.h>
80 __FBSDID("$FreeBSD$");
83 #include "opt_inet6.h"
85 #include <sys/param.h>
86 #include <sys/eventhandler.h>
88 #include <sys/malloc.h>
89 #include <sys/protosw.h>
90 #include <sys/systm.h>
93 #include <sys/socket.h> /* for net/if.h */
94 #include <sys/sockio.h>
95 #include <sys/ctype.h> /* string functions */
96 #include <sys/kernel.h>
97 #include <sys/random.h>
98 #include <sys/syslog.h>
99 #include <sys/sysctl.h>
101 #include <sys/module.h>
102 #include <sys/priv.h>
103 #include <sys/proc.h>
104 #include <sys/lock.h>
105 #include <sys/mutex.h>
109 #include <net/if_clone.h>
110 #include <net/if_dl.h>
111 #include <net/if_types.h>
112 #include <net/if_var.h>
113 #include <net/if_private.h>
114 #include <net/pfil.h>
115 #include <net/vnet.h>
117 #include <netinet/in.h>
118 #include <netinet/in_systm.h>
119 #include <netinet/in_var.h>
120 #include <netinet/ip.h>
121 #include <netinet/ip_var.h>
123 #include <netinet/ip6.h>
124 #include <netinet6/ip6_var.h>
125 #include <netinet6/in6_ifattach.h>
127 #if defined(INET) || defined(INET6)
128 #include <netinet/ip_carp.h>
130 #include <machine/in_cksum.h>
131 #include <netinet/if_ether.h>
132 #include <net/bridgestp.h>
133 #include <net/if_bridgevar.h>
134 #include <net/if_llc.h>
135 #include <net/if_vlan_var.h>
137 #include <net/route.h>
141 * XXX: declare here to avoid to include many inet6 related files..
142 * should be more generalized?
144 extern void nd6_setmtu(struct ifnet *);
148 * Size of the route hash table. Must be a power of two.
150 #ifndef BRIDGE_RTHASH_SIZE
151 #define BRIDGE_RTHASH_SIZE 1024
154 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1)
157 * Default maximum number of addresses to cache.
159 #ifndef BRIDGE_RTABLE_MAX
160 #define BRIDGE_RTABLE_MAX 2000
164 * Timeout (in seconds) for entries learned dynamically.
166 #ifndef BRIDGE_RTABLE_TIMEOUT
167 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */
171 * Number of seconds between walks of the route list.
173 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
174 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60)
178 * List of capabilities to possibly mask on the member interface.
180 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
184 * List of capabilities to strip
186 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO
191 * The bridge relies heavily on the epoch(9) system to protect its data
192 * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we
193 * must ensure there is only one writer at a time.
195 * That is: for read accesses we only need to be in NET_EPOCH, but for write
196 * accesses we must hold:
198 * - BRIDGE_RT_LOCK, for any change to bridge_rtnodes
199 * - BRIDGE_LOCK, for any other change
201 * The BRIDGE_LOCK is a sleepable lock, because it is held across ioctl()
202 * calls to bridge member interfaces and these ioctl()s can sleep.
203 * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes
204 * required while we're in NET_EPOCH and then we're not allowed to sleep.
206 #define BRIDGE_LOCK_INIT(_sc) do { \
207 sx_init(&(_sc)->sc_sx, "if_bridge"); \
208 mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF); \
210 #define BRIDGE_LOCK_DESTROY(_sc) do { \
211 sx_destroy(&(_sc)->sc_sx); \
212 mtx_destroy(&(_sc)->sc_rt_mtx); \
214 #define BRIDGE_LOCK(_sc) sx_xlock(&(_sc)->sc_sx)
215 #define BRIDGE_UNLOCK(_sc) sx_xunlock(&(_sc)->sc_sx)
216 #define BRIDGE_LOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_XLOCKED)
217 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
218 MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx))
219 #define BRIDGE_UNLOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_UNLOCKED)
220 #define BRIDGE_RT_LOCK(_sc) mtx_lock(&(_sc)->sc_rt_mtx)
221 #define BRIDGE_RT_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_rt_mtx)
222 #define BRIDGE_RT_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED)
223 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
224 MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx))
227 * Bridge interface list entry.
229 struct bridge_iflist {
230 CK_LIST_ENTRY(bridge_iflist) bif_next;
231 struct ifnet *bif_ifp; /* member if */
232 struct bstp_port bif_stp; /* STP state */
233 uint32_t bif_flags; /* member if flags */
234 int bif_savedcaps; /* saved capabilities */
235 uint32_t bif_addrmax; /* max # of addresses */
236 uint32_t bif_addrcnt; /* cur. # of addresses */
237 uint32_t bif_addrexceeded;/* # of address violations */
238 struct epoch_context bif_epoch_ctx;
244 struct bridge_rtnode {
245 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
246 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
247 struct bridge_iflist *brt_dst; /* destination if */
248 unsigned long brt_expire; /* expiration time */
249 uint8_t brt_flags; /* address flags */
250 uint8_t brt_addr[ETHER_ADDR_LEN];
251 uint16_t brt_vlan; /* vlan id */
252 struct vnet *brt_vnet;
253 struct epoch_context brt_epoch_ctx;
255 #define brt_ifp brt_dst->bif_ifp
258 * Software state for each bridge.
260 struct bridge_softc {
261 struct ifnet *sc_ifp; /* make this an interface */
262 LIST_ENTRY(bridge_softc) sc_list;
264 struct mtx sc_rt_mtx;
265 uint32_t sc_brtmax; /* max # of addresses */
266 uint32_t sc_brtcnt; /* cur. # of addresses */
267 uint32_t sc_brttimeout; /* rt timeout in seconds */
268 struct callout sc_brcallout; /* bridge callout */
269 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
270 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
271 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
272 uint32_t sc_rthash_key; /* key for hash */
273 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
274 struct bstp_state sc_stp; /* STP state */
275 uint32_t sc_brtexceeded; /* # of cache drops */
276 struct ifnet *sc_ifaddr; /* member mac copied from */
277 struct ether_addr sc_defaddr; /* Default MAC address */
278 struct epoch_context sc_epoch_ctx;
281 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
282 #define V_bridge_list_sx VNET(bridge_list_sx)
283 static eventhandler_tag bridge_detach_cookie;
285 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
287 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
288 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
290 static int bridge_clone_create(struct if_clone *, char *, size_t,
291 struct ifc_data *, struct ifnet **);
292 static int bridge_clone_destroy(struct if_clone *, struct ifnet *, uint32_t);
294 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
295 static void bridge_mutecaps(struct bridge_softc *);
296 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
298 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
299 static void bridge_init(void *);
300 static void bridge_dummynet(struct mbuf *, struct ifnet *);
301 static void bridge_stop(struct ifnet *, int);
302 static int bridge_transmit(struct ifnet *, struct mbuf *);
304 static void bridge_altq_start(if_t);
305 static int bridge_altq_transmit(if_t, struct mbuf *);
307 static void bridge_qflush(struct ifnet *);
308 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
309 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
311 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
313 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
315 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
318 static void bridge_timer(void *);
320 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
322 static void bridge_span(struct bridge_softc *, struct mbuf *);
324 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
325 uint16_t, struct bridge_iflist *, int, uint8_t);
326 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
328 static void bridge_rttrim(struct bridge_softc *);
329 static void bridge_rtage(struct bridge_softc *);
330 static void bridge_rtflush(struct bridge_softc *, int);
331 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
334 static void bridge_rtable_init(struct bridge_softc *);
335 static void bridge_rtable_fini(struct bridge_softc *);
337 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
338 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
339 const uint8_t *, uint16_t);
340 static int bridge_rtnode_insert(struct bridge_softc *,
341 struct bridge_rtnode *);
342 static void bridge_rtnode_destroy(struct bridge_softc *,
343 struct bridge_rtnode *);
344 static void bridge_rtable_expire(struct ifnet *, int);
345 static void bridge_state_change(struct ifnet *, int);
347 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
349 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
351 static void bridge_delete_member(struct bridge_softc *,
352 struct bridge_iflist *, int);
353 static void bridge_delete_span(struct bridge_softc *,
354 struct bridge_iflist *);
356 static int bridge_ioctl_add(struct bridge_softc *, void *);
357 static int bridge_ioctl_del(struct bridge_softc *, void *);
358 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
359 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
360 static int bridge_ioctl_scache(struct bridge_softc *, void *);
361 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
362 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
363 static int bridge_ioctl_rts(struct bridge_softc *, void *);
364 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
365 static int bridge_ioctl_sto(struct bridge_softc *, void *);
366 static int bridge_ioctl_gto(struct bridge_softc *, void *);
367 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
368 static int bridge_ioctl_flush(struct bridge_softc *, void *);
369 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
370 static int bridge_ioctl_spri(struct bridge_softc *, void *);
371 static int bridge_ioctl_ght(struct bridge_softc *, void *);
372 static int bridge_ioctl_sht(struct bridge_softc *, void *);
373 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
374 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
375 static int bridge_ioctl_gma(struct bridge_softc *, void *);
376 static int bridge_ioctl_sma(struct bridge_softc *, void *);
377 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
378 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
379 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
380 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
381 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
382 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
383 static int bridge_ioctl_grte(struct bridge_softc *, void *);
384 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
385 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
386 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
387 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
389 static int bridge_ip_checkbasic(struct mbuf **mp);
391 static int bridge_ip6_checkbasic(struct mbuf **mp);
393 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
394 struct ether_header *, int, struct llc *);
395 static void bridge_linkstate(struct ifnet *ifp);
396 static void bridge_linkcheck(struct bridge_softc *sc);
398 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
399 #define VLANTAGOF(_m) \
400 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
402 static struct bstp_cb_ops bridge_ops = {
403 .bcb_state = bridge_state_change,
404 .bcb_rtage = bridge_rtable_expire
407 SYSCTL_DECL(_net_link);
408 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
411 /* only pass IP[46] packets when pfil is enabled */
412 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
413 #define V_pfil_onlyip VNET(pfil_onlyip)
414 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
415 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
416 "Only pass IP packets when pfil is enabled");
418 /* run pfil hooks on the bridge interface */
419 VNET_DEFINE_STATIC(int, pfil_bridge) = 0;
420 #define V_pfil_bridge VNET(pfil_bridge)
421 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
422 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
423 "Packet filter on the bridge interface");
425 /* layer2 filter with ipfw */
426 VNET_DEFINE_STATIC(int, pfil_ipfw);
427 #define V_pfil_ipfw VNET(pfil_ipfw)
429 /* layer2 ARP filter with ipfw */
430 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
431 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
432 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
433 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
434 "Filter ARP packets through IPFW layer2");
436 /* run pfil hooks on the member interface */
437 VNET_DEFINE_STATIC(int, pfil_member) = 0;
438 #define V_pfil_member VNET(pfil_member)
439 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
440 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
441 "Packet filter on the member interface");
443 /* run pfil hooks on the physical interface for locally destined packets */
444 VNET_DEFINE_STATIC(int, pfil_local_phys);
445 #define V_pfil_local_phys VNET(pfil_local_phys)
446 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
447 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
448 "Packet filter on the physical interface for locally destined packets");
450 /* log STP state changes */
451 VNET_DEFINE_STATIC(int, log_stp);
452 #define V_log_stp VNET(log_stp)
453 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
454 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
455 "Log STP state changes");
457 /* share MAC with first bridge member */
458 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
459 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
460 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
461 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
462 "Inherit MAC address from the first bridge member");
464 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
465 #define V_allow_llz_overlap VNET(allow_llz_overlap)
466 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
467 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
468 "Allow overlap of link-local scope "
469 "zones of a bridge interface and the member interfaces");
471 struct bridge_control {
472 int (*bc_func)(struct bridge_softc *, void *);
477 #define BC_F_COPYIN 0x01 /* copy arguments in */
478 #define BC_F_COPYOUT 0x02 /* copy arguments out */
479 #define BC_F_SUSER 0x04 /* do super-user check */
481 static const struct bridge_control bridge_control_table[] = {
482 { bridge_ioctl_add, sizeof(struct ifbreq),
483 BC_F_COPYIN|BC_F_SUSER },
484 { bridge_ioctl_del, sizeof(struct ifbreq),
485 BC_F_COPYIN|BC_F_SUSER },
487 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
488 BC_F_COPYIN|BC_F_COPYOUT },
489 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
490 BC_F_COPYIN|BC_F_SUSER },
492 { bridge_ioctl_scache, sizeof(struct ifbrparam),
493 BC_F_COPYIN|BC_F_SUSER },
494 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
497 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
498 BC_F_COPYIN|BC_F_COPYOUT },
499 { bridge_ioctl_rts, sizeof(struct ifbaconf),
500 BC_F_COPYIN|BC_F_COPYOUT },
502 { bridge_ioctl_saddr, sizeof(struct ifbareq),
503 BC_F_COPYIN|BC_F_SUSER },
505 { bridge_ioctl_sto, sizeof(struct ifbrparam),
506 BC_F_COPYIN|BC_F_SUSER },
507 { bridge_ioctl_gto, sizeof(struct ifbrparam),
510 { bridge_ioctl_daddr, sizeof(struct ifbareq),
511 BC_F_COPYIN|BC_F_SUSER },
513 { bridge_ioctl_flush, sizeof(struct ifbreq),
514 BC_F_COPYIN|BC_F_SUSER },
516 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
518 { bridge_ioctl_spri, sizeof(struct ifbrparam),
519 BC_F_COPYIN|BC_F_SUSER },
521 { bridge_ioctl_ght, sizeof(struct ifbrparam),
523 { bridge_ioctl_sht, sizeof(struct ifbrparam),
524 BC_F_COPYIN|BC_F_SUSER },
526 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
528 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
529 BC_F_COPYIN|BC_F_SUSER },
531 { bridge_ioctl_gma, sizeof(struct ifbrparam),
533 { bridge_ioctl_sma, sizeof(struct ifbrparam),
534 BC_F_COPYIN|BC_F_SUSER },
536 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
537 BC_F_COPYIN|BC_F_SUSER },
539 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
540 BC_F_COPYIN|BC_F_SUSER },
542 { bridge_ioctl_addspan, sizeof(struct ifbreq),
543 BC_F_COPYIN|BC_F_SUSER },
544 { bridge_ioctl_delspan, sizeof(struct ifbreq),
545 BC_F_COPYIN|BC_F_SUSER },
547 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
550 { bridge_ioctl_grte, sizeof(struct ifbrparam),
553 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
554 BC_F_COPYIN|BC_F_COPYOUT },
556 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
557 BC_F_COPYIN|BC_F_SUSER },
559 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
560 BC_F_COPYIN|BC_F_SUSER },
562 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
563 BC_F_COPYIN|BC_F_SUSER },
566 static const int bridge_control_table_size = nitems(bridge_control_table);
568 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
569 #define V_bridge_list VNET(bridge_list)
570 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \
572 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx)
573 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx)
574 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx)
576 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
577 #define V_bridge_cloner VNET(bridge_cloner)
579 static const char bridge_name[] = "bridge";
582 vnet_bridge_init(const void *unused __unused)
585 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
586 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
588 BRIDGE_LIST_LOCK_INIT();
589 LIST_INIT(&V_bridge_list);
591 struct if_clone_addreq req = {
592 .create_f = bridge_clone_create,
593 .destroy_f = bridge_clone_destroy,
594 .flags = IFC_F_AUTOUNIT,
596 V_bridge_cloner = ifc_attach_cloner(bridge_name, &req);
598 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
599 vnet_bridge_init, NULL);
602 vnet_bridge_uninit(const void *unused __unused)
605 ifc_detach_cloner(V_bridge_cloner);
606 V_bridge_cloner = NULL;
607 BRIDGE_LIST_LOCK_DESTROY();
609 /* Callbacks may use the UMA zone. */
610 NET_EPOCH_DRAIN_CALLBACKS();
612 uma_zdestroy(V_bridge_rtnode_zone);
614 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
615 vnet_bridge_uninit, NULL);
618 bridge_modevent(module_t mod, int type, void *data)
623 bridge_dn_p = bridge_dummynet;
624 bridge_detach_cookie = EVENTHANDLER_REGISTER(
625 ifnet_departure_event, bridge_ifdetach, NULL,
626 EVENTHANDLER_PRI_ANY);
629 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
630 bridge_detach_cookie);
639 static moduledata_t bridge_mod = {
645 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
646 MODULE_VERSION(if_bridge, 1);
647 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
650 * handler for net.link.bridge.ipfw
653 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
655 int enable = V_pfil_ipfw;
658 error = sysctl_handle_int(oidp, &enable, 0, req);
661 if (enable != V_pfil_ipfw) {
662 V_pfil_ipfw = enable;
665 * Disable pfil so that ipfw doesnt run twice, if the user
666 * really wants both then they can re-enable pfil_bridge and/or
667 * pfil_member. Also allow non-ip packets as ipfw can filter by
679 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
680 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
681 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
682 "Layer2 filter with IPFW");
686 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
688 struct bridge_softc *sc = ifp->if_softc;
689 struct bridge_iflist *bif;
693 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
694 bridge_delete_member(sc, bif, 0);
696 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
697 bridge_delete_span(sc, bif);
702 ether_reassign(ifp, newvnet, arg);
707 * bridge_clone_create:
709 * Create a new bridge instance.
712 bridge_clone_create(struct if_clone *ifc, char *name, size_t len,
713 struct ifc_data *ifd, struct ifnet **ifpp)
715 struct bridge_softc *sc;
718 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
719 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
725 BRIDGE_LOCK_INIT(sc);
726 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
727 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
729 /* Initialize our routing table. */
730 bridge_rtable_init(sc);
732 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
734 CK_LIST_INIT(&sc->sc_iflist);
735 CK_LIST_INIT(&sc->sc_spanlist);
738 if_initname(ifp, bridge_name, ifd->unit);
739 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
740 ifp->if_ioctl = bridge_ioctl;
742 ifp->if_start = bridge_altq_start;
743 ifp->if_transmit = bridge_altq_transmit;
744 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
745 ifp->if_snd.ifq_drv_maxlen = 0;
746 IFQ_SET_READY(&ifp->if_snd);
748 ifp->if_transmit = bridge_transmit;
750 ifp->if_qflush = bridge_qflush;
751 ifp->if_init = bridge_init;
752 ifp->if_type = IFT_BRIDGE;
754 ether_gen_addr(ifp, &sc->sc_defaddr);
756 bstp_attach(&sc->sc_stp, &bridge_ops);
757 ether_ifattach(ifp, sc->sc_defaddr.octet);
758 /* Now undo some of the damage... */
759 ifp->if_baudrate = 0;
760 ifp->if_type = IFT_BRIDGE;
762 ifp->if_reassign = bridge_reassign;
766 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
767 BRIDGE_LIST_UNLOCK();
774 bridge_clone_destroy_cb(struct epoch_context *ctx)
776 struct bridge_softc *sc;
778 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
780 BRIDGE_LOCK_DESTROY(sc);
785 * bridge_clone_destroy:
787 * Destroy a bridge instance.
790 bridge_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags)
792 struct bridge_softc *sc = ifp->if_softc;
793 struct bridge_iflist *bif;
794 struct epoch_tracker et;
799 ifp->if_flags &= ~IFF_UP;
801 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
802 bridge_delete_member(sc, bif, 0);
804 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
805 bridge_delete_span(sc, bif);
808 /* Tear down the routing table. */
809 bridge_rtable_fini(sc);
815 callout_drain(&sc->sc_brcallout);
818 LIST_REMOVE(sc, sc_list);
819 BRIDGE_LIST_UNLOCK();
821 bstp_detach(&sc->sc_stp);
823 IFQ_PURGE(&ifp->if_snd);
830 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
838 * Handle a control request from the operator.
841 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
843 struct bridge_softc *sc = ifp->if_softc;
844 struct ifreq *ifr = (struct ifreq *)data;
845 struct bridge_iflist *bif;
846 struct thread *td = curthread;
848 struct ifbreq ifbreq;
849 struct ifbifconf ifbifconf;
850 struct ifbareq ifbareq;
851 struct ifbaconf ifbaconf;
852 struct ifbrparam ifbrparam;
853 struct ifbropreq ifbropreq;
855 struct ifdrv *ifd = (struct ifdrv *) data;
856 const struct bridge_control *bc;
857 int error = 0, oldmtu;
868 if (ifd->ifd_cmd >= bridge_control_table_size) {
872 bc = &bridge_control_table[ifd->ifd_cmd];
874 if (cmd == SIOCGDRVSPEC &&
875 (bc->bc_flags & BC_F_COPYOUT) == 0) {
879 else if (cmd == SIOCSDRVSPEC &&
880 (bc->bc_flags & BC_F_COPYOUT) != 0) {
885 if (bc->bc_flags & BC_F_SUSER) {
886 error = priv_check(td, PRIV_NET_BRIDGE);
891 if (ifd->ifd_len != bc->bc_argsize ||
892 ifd->ifd_len > sizeof(args)) {
897 bzero(&args, sizeof(args));
898 if (bc->bc_flags & BC_F_COPYIN) {
899 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
904 oldmtu = ifp->if_mtu;
905 error = (*bc->bc_func)(sc, &args);
910 * Bridge MTU may change during addition of the first port.
911 * If it did, do network layer specific procedure.
913 if (ifp->if_mtu != oldmtu) {
920 if (bc->bc_flags & BC_F_COPYOUT)
921 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
926 if (!(ifp->if_flags & IFF_UP) &&
927 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
929 * If interface is marked down and it is running,
930 * then stop and disable it.
933 } else if ((ifp->if_flags & IFF_UP) &&
934 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
936 * If interface is marked up and it is stopped, then
946 oldmtu = sc->sc_ifp->if_mtu;
948 if (ifr->ifr_mtu < 576) {
952 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
953 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
956 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
957 error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
958 SIOCSIFMTU, (caddr_t)ifr);
960 log(LOG_NOTICE, "%s: invalid MTU: %u for"
961 " member %s\n", sc->sc_ifp->if_xname,
963 bif->bif_ifp->if_xname);
969 /* Restore the previous MTU on all member interfaces. */
970 ifr->ifr_mtu = oldmtu;
971 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
972 (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
973 SIOCSIFMTU, (caddr_t)ifr);
976 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
981 * drop the lock as ether_ioctl() will call bridge_start() and
982 * cause the lock to be recursed.
985 error = ether_ioctl(ifp, cmd, data);
998 * Clear or restore unwanted capabilities on the member interface
1001 bridge_mutecaps(struct bridge_softc *sc)
1003 struct bridge_iflist *bif;
1006 BRIDGE_LOCK_ASSERT(sc);
1008 /* Initial bitmask of capabilities to test */
1009 mask = BRIDGE_IFCAPS_MASK;
1011 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1012 /* Every member must support it or its disabled */
1013 mask &= bif->bif_savedcaps;
1016 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1017 enabled = bif->bif_ifp->if_capenable;
1018 enabled &= ~BRIDGE_IFCAPS_STRIP;
1019 /* strip off mask bits and enable them again if allowed */
1020 enabled &= ~BRIDGE_IFCAPS_MASK;
1022 bridge_set_ifcap(sc, bif, enabled);
1027 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1029 struct ifnet *ifp = bif->bif_ifp;
1031 int error, mask, stuck;
1033 bzero(&ifr, sizeof(ifr));
1034 ifr.ifr_reqcap = set;
1036 if (ifp->if_capenable != set) {
1037 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1039 if_printf(sc->sc_ifp,
1040 "error setting capabilities on %s: %d\n",
1041 ifp->if_xname, error);
1042 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1043 stuck = ifp->if_capenable & mask & ~set;
1045 if_printf(sc->sc_ifp,
1046 "can't disable some capabilities on %s: 0x%x\n",
1047 ifp->if_xname, stuck);
1052 * bridge_lookup_member:
1054 * Lookup a bridge member interface.
1056 static struct bridge_iflist *
1057 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1059 struct bridge_iflist *bif;
1062 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1064 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1066 if (strcmp(ifp->if_xname, name) == 0)
1074 * bridge_lookup_member_if:
1076 * Lookup a bridge member interface by ifnet*.
1078 static struct bridge_iflist *
1079 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1081 struct bridge_iflist *bif;
1083 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1085 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1086 if (bif->bif_ifp == member_ifp)
1094 bridge_delete_member_cb(struct epoch_context *ctx)
1096 struct bridge_iflist *bif;
1098 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1100 free(bif, M_DEVBUF);
1104 * bridge_delete_member:
1106 * Delete the specified member interface.
1109 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1112 struct ifnet *ifs = bif->bif_ifp;
1113 struct ifnet *fif = NULL;
1114 struct bridge_iflist *bifl;
1116 BRIDGE_LOCK_ASSERT(sc);
1118 if (bif->bif_flags & IFBIF_STP)
1119 bstp_disable(&bif->bif_stp);
1121 ifs->if_bridge = NULL;
1122 CK_LIST_REMOVE(bif, bif_next);
1125 * If removing the interface that gave the bridge its mac address, set
1126 * the mac address of the bridge to the address of the next member, or
1127 * to its default address if no members are left.
1129 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1130 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1131 bcopy(&sc->sc_defaddr,
1132 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1133 sc->sc_ifaddr = NULL;
1135 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1136 fif = bifl->bif_ifp;
1137 bcopy(IF_LLADDR(fif),
1138 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1139 sc->sc_ifaddr = fif;
1141 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1144 bridge_linkcheck(sc);
1145 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1147 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1148 BRIDGE_RT_UNLOCK(sc);
1149 KASSERT(bif->bif_addrcnt == 0,
1150 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1152 ifs->if_bridge_output = NULL;
1153 ifs->if_bridge_input = NULL;
1154 ifs->if_bridge_linkstate = NULL;
1156 switch (ifs->if_type) {
1160 * Take the interface out of promiscuous mode, but only
1161 * if it was promiscuous in the first place. It might
1162 * not be if we're in the bridge_ioctl_add() error path.
1164 if (ifs->if_flags & IFF_PROMISC)
1165 (void) ifpromisc(ifs, 0);
1173 panic("bridge_delete_member: impossible");
1177 /* reneable any interface capabilities */
1178 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1180 bstp_destroy(&bif->bif_stp); /* prepare to free */
1182 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1186 * bridge_delete_span:
1188 * Delete the specified span interface.
1191 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1193 BRIDGE_LOCK_ASSERT(sc);
1195 KASSERT(bif->bif_ifp->if_bridge == NULL,
1196 ("%s: not a span interface", __func__));
1198 CK_LIST_REMOVE(bif, bif_next);
1200 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1204 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1206 struct ifbreq *req = arg;
1207 struct bridge_iflist *bif = NULL;
1211 ifs = ifunit(req->ifbr_ifsname);
1214 if (ifs->if_ioctl == NULL) /* must be supported */
1217 /* If it's in the span list, it can't be a member. */
1218 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1219 if (ifs == bif->bif_ifp)
1222 if (ifs->if_bridge == sc)
1225 if (ifs->if_bridge != NULL)
1228 switch (ifs->if_type) {
1232 /* permitted interface types */
1240 * Two valid inet6 addresses with link-local scope must not be
1241 * on the parent interface and the member interfaces at the
1242 * same time. This restriction is needed to prevent violation
1243 * of link-local scope zone. Attempts to add a member
1244 * interface which has inet6 addresses when the parent has
1245 * inet6 triggers removal of all inet6 addresses on the member
1249 /* Check if the parent interface has a link-local scope addr. */
1250 if (V_allow_llz_overlap == 0 &&
1251 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1253 * If any, remove all inet6 addresses from the member
1256 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1257 if (in6ifa_llaonifp(bif->bif_ifp)) {
1258 in6_ifdetach(bif->bif_ifp);
1259 if_printf(sc->sc_ifp,
1260 "IPv6 addresses on %s have been removed "
1261 "before adding it as a member to prevent "
1262 "IPv6 address scope violation.\n",
1263 bif->bif_ifp->if_xname);
1266 if (in6ifa_llaonifp(ifs)) {
1268 if_printf(sc->sc_ifp,
1269 "IPv6 addresses on %s have been removed "
1270 "before adding it as a member to prevent "
1271 "IPv6 address scope violation.\n",
1276 /* Allow the first Ethernet member to define the MTU */
1277 if (CK_LIST_EMPTY(&sc->sc_iflist))
1278 sc->sc_ifp->if_mtu = ifs->if_mtu;
1279 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1282 snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s",
1284 ifr.ifr_mtu = sc->sc_ifp->if_mtu;
1286 error = (*ifs->if_ioctl)(ifs,
1287 SIOCSIFMTU, (caddr_t)&ifr);
1289 log(LOG_NOTICE, "%s: invalid MTU: %u for"
1290 " new member %s\n", sc->sc_ifp->if_xname,
1297 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1302 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1303 bif->bif_savedcaps = ifs->if_capenable;
1306 * Assign the interface's MAC address to the bridge if it's the first
1307 * member and the MAC address of the bridge has not been changed from
1308 * the default randomly generated one.
1310 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1311 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1312 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1313 sc->sc_ifaddr = ifs;
1314 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1317 ifs->if_bridge = sc;
1318 ifs->if_bridge_output = bridge_output;
1319 ifs->if_bridge_input = bridge_input;
1320 ifs->if_bridge_linkstate = bridge_linkstate;
1321 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1323 * XXX: XLOCK HERE!?!
1325 * NOTE: insert_***HEAD*** should be safe for the traversals.
1327 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1329 /* Set interface capabilities to the intersection set of all members */
1330 bridge_mutecaps(sc);
1331 bridge_linkcheck(sc);
1333 /* Place the interface into promiscuous mode */
1334 switch (ifs->if_type) {
1337 error = ifpromisc(ifs, 1);
1342 bridge_delete_member(sc, bif, 0);
1347 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1349 struct ifbreq *req = arg;
1350 struct bridge_iflist *bif;
1352 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1356 bridge_delete_member(sc, bif, 0);
1362 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1364 struct ifbreq *req = arg;
1365 struct bridge_iflist *bif;
1366 struct bstp_port *bp;
1368 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1373 req->ifbr_ifsflags = bif->bif_flags;
1374 req->ifbr_state = bp->bp_state;
1375 req->ifbr_priority = bp->bp_priority;
1376 req->ifbr_path_cost = bp->bp_path_cost;
1377 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1378 req->ifbr_proto = bp->bp_protover;
1379 req->ifbr_role = bp->bp_role;
1380 req->ifbr_stpflags = bp->bp_flags;
1381 req->ifbr_addrcnt = bif->bif_addrcnt;
1382 req->ifbr_addrmax = bif->bif_addrmax;
1383 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1385 /* Copy STP state options as flags */
1386 if (bp->bp_operedge)
1387 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1388 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1389 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1390 if (bp->bp_ptp_link)
1391 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1392 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1393 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1394 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1395 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1396 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1397 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1402 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1404 struct epoch_tracker et;
1405 struct ifbreq *req = arg;
1406 struct bridge_iflist *bif;
1407 struct bstp_port *bp;
1410 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1415 if (req->ifbr_ifsflags & IFBIF_SPAN)
1416 /* SPAN is readonly */
1419 NET_EPOCH_ENTER(et);
1421 if (req->ifbr_ifsflags & IFBIF_STP) {
1422 if ((bif->bif_flags & IFBIF_STP) == 0) {
1423 error = bstp_enable(&bif->bif_stp);
1430 if ((bif->bif_flags & IFBIF_STP) != 0)
1431 bstp_disable(&bif->bif_stp);
1434 /* Pass on STP flags */
1435 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1436 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1437 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1438 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1440 /* Save the bits relating to the bridge */
1441 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1449 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1451 struct ifbrparam *param = arg;
1453 sc->sc_brtmax = param->ifbrp_csize;
1460 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1462 struct ifbrparam *param = arg;
1464 param->ifbrp_csize = sc->sc_brtmax;
1470 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1472 struct ifbifconf *bifc = arg;
1473 struct bridge_iflist *bif;
1476 int count, buflen, len, error = 0;
1479 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1481 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1484 buflen = sizeof(breq) * count;
1485 if (bifc->ifbic_len == 0) {
1486 bifc->ifbic_len = buflen;
1489 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1495 len = min(bifc->ifbic_len, buflen);
1496 bzero(&breq, sizeof(breq));
1497 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1498 if (len < sizeof(breq))
1501 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1502 sizeof(breq.ifbr_ifsname));
1503 /* Fill in the ifbreq structure */
1504 error = bridge_ioctl_gifflags(sc, &breq);
1507 memcpy(buf, &breq, sizeof(breq));
1509 buf += sizeof(breq);
1510 len -= sizeof(breq);
1512 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1513 if (len < sizeof(breq))
1516 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1517 sizeof(breq.ifbr_ifsname));
1518 breq.ifbr_ifsflags = bif->bif_flags;
1519 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1520 memcpy(buf, &breq, sizeof(breq));
1522 buf += sizeof(breq);
1523 len -= sizeof(breq);
1526 bifc->ifbic_len = sizeof(breq) * count;
1527 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1528 free(outbuf, M_TEMP);
1533 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1535 struct ifbaconf *bac = arg;
1536 struct bridge_rtnode *brt;
1537 struct ifbareq bareq;
1539 int count, buflen, len, error = 0;
1541 if (bac->ifbac_len == 0)
1545 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1547 buflen = sizeof(bareq) * count;
1549 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1555 len = min(bac->ifbac_len, buflen);
1556 bzero(&bareq, sizeof(bareq));
1557 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1558 if (len < sizeof(bareq))
1560 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1561 sizeof(bareq.ifba_ifsname));
1562 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1563 bareq.ifba_vlan = brt->brt_vlan;
1564 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1565 time_uptime < brt->brt_expire)
1566 bareq.ifba_expire = brt->brt_expire - time_uptime;
1568 bareq.ifba_expire = 0;
1569 bareq.ifba_flags = brt->brt_flags;
1571 memcpy(buf, &bareq, sizeof(bareq));
1573 buf += sizeof(bareq);
1574 len -= sizeof(bareq);
1577 bac->ifbac_len = sizeof(bareq) * count;
1578 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1579 free(outbuf, M_TEMP);
1584 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1586 struct ifbareq *req = arg;
1587 struct bridge_iflist *bif;
1588 struct epoch_tracker et;
1591 NET_EPOCH_ENTER(et);
1592 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1598 /* bridge_rtupdate() may acquire the lock. */
1599 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1607 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1609 struct ifbrparam *param = arg;
1611 sc->sc_brttimeout = param->ifbrp_ctime;
1616 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1618 struct ifbrparam *param = arg;
1620 param->ifbrp_ctime = sc->sc_brttimeout;
1625 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1627 struct ifbareq *req = arg;
1629 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1633 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1635 struct ifbreq *req = arg;
1638 bridge_rtflush(sc, req->ifbr_ifsflags);
1639 BRIDGE_RT_UNLOCK(sc);
1645 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1647 struct ifbrparam *param = arg;
1648 struct bstp_state *bs = &sc->sc_stp;
1650 param->ifbrp_prio = bs->bs_bridge_priority;
1655 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1657 struct ifbrparam *param = arg;
1659 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1663 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1665 struct ifbrparam *param = arg;
1666 struct bstp_state *bs = &sc->sc_stp;
1668 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1673 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1675 struct ifbrparam *param = arg;
1677 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1681 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1683 struct ifbrparam *param = arg;
1684 struct bstp_state *bs = &sc->sc_stp;
1686 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1691 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1693 struct ifbrparam *param = arg;
1695 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1699 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1701 struct ifbrparam *param = arg;
1702 struct bstp_state *bs = &sc->sc_stp;
1704 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1709 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1711 struct ifbrparam *param = arg;
1713 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1717 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1719 struct ifbreq *req = arg;
1720 struct bridge_iflist *bif;
1722 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1726 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1730 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1732 struct ifbreq *req = arg;
1733 struct bridge_iflist *bif;
1735 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1739 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1743 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1745 struct ifbreq *req = arg;
1746 struct bridge_iflist *bif;
1748 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1752 bif->bif_addrmax = req->ifbr_addrmax;
1757 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1759 struct ifbreq *req = arg;
1760 struct bridge_iflist *bif = NULL;
1763 ifs = ifunit(req->ifbr_ifsname);
1767 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1768 if (ifs == bif->bif_ifp)
1771 if (ifs->if_bridge != NULL)
1774 switch (ifs->if_type) {
1783 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1788 bif->bif_flags = IFBIF_SPAN;
1790 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1796 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1798 struct ifbreq *req = arg;
1799 struct bridge_iflist *bif;
1802 ifs = ifunit(req->ifbr_ifsname);
1806 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1807 if (ifs == bif->bif_ifp)
1813 bridge_delete_span(sc, bif);
1819 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1821 struct ifbropreq *req = arg;
1822 struct bstp_state *bs = &sc->sc_stp;
1823 struct bstp_port *root_port;
1825 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1826 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1827 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1829 root_port = bs->bs_root_port;
1830 if (root_port == NULL)
1831 req->ifbop_root_port = 0;
1833 req->ifbop_root_port = root_port->bp_ifp->if_index;
1835 req->ifbop_holdcount = bs->bs_txholdcount;
1836 req->ifbop_priority = bs->bs_bridge_priority;
1837 req->ifbop_protocol = bs->bs_protover;
1838 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1839 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1840 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1841 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1842 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1843 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1849 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1851 struct ifbrparam *param = arg;
1853 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1858 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1860 struct ifbpstpconf *bifstp = arg;
1861 struct bridge_iflist *bif;
1862 struct bstp_port *bp;
1863 struct ifbpstpreq bpreq;
1865 int count, buflen, len, error = 0;
1868 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1869 if ((bif->bif_flags & IFBIF_STP) != 0)
1873 buflen = sizeof(bpreq) * count;
1874 if (bifstp->ifbpstp_len == 0) {
1875 bifstp->ifbpstp_len = buflen;
1879 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1885 len = min(bifstp->ifbpstp_len, buflen);
1886 bzero(&bpreq, sizeof(bpreq));
1887 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1888 if (len < sizeof(bpreq))
1891 if ((bif->bif_flags & IFBIF_STP) == 0)
1895 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1896 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1897 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1898 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1899 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1900 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1902 memcpy(buf, &bpreq, sizeof(bpreq));
1904 buf += sizeof(bpreq);
1905 len -= sizeof(bpreq);
1908 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1909 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1910 free(outbuf, M_TEMP);
1915 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1917 struct ifbrparam *param = arg;
1919 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1923 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1925 struct ifbrparam *param = arg;
1927 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1933 * Detach an interface from a bridge. Called when a member
1934 * interface is detaching.
1937 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1939 struct bridge_softc *sc = ifp->if_bridge;
1940 struct bridge_iflist *bif;
1942 if (ifp->if_flags & IFF_RENAMING)
1944 if (V_bridge_cloner == NULL) {
1946 * This detach handler can be called after
1947 * vnet_bridge_uninit(). Just return in that case.
1951 /* Check if the interface is a bridge member */
1955 bif = bridge_lookup_member_if(sc, ifp);
1957 bridge_delete_member(sc, bif, 1);
1963 /* Check if the interface is a span port */
1965 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1967 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1968 if (ifp == bif->bif_ifp) {
1969 bridge_delete_span(sc, bif);
1975 BRIDGE_LIST_UNLOCK();
1981 * Initialize a bridge interface.
1984 bridge_init(void *xsc)
1986 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1987 struct ifnet *ifp = sc->sc_ifp;
1989 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1993 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1996 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1997 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
2005 * Stop the bridge interface.
2008 bridge_stop(struct ifnet *ifp, int disable)
2010 struct bridge_softc *sc = ifp->if_softc;
2012 BRIDGE_LOCK_ASSERT(sc);
2014 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2018 callout_stop(&sc->sc_brcallout);
2020 bstp_stop(&sc->sc_stp);
2022 bridge_rtflush(sc, IFBF_FLUSHDYN);
2023 BRIDGE_RT_UNLOCK(sc);
2025 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2031 * Enqueue a packet on a bridge member interface.
2035 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2041 /* We may be sending a fragment so traverse the mbuf */
2044 m->m_nextpkt = NULL;
2045 len = m->m_pkthdr.len;
2046 mflags = m->m_flags;
2049 * If underlying interface can not do VLAN tag insertion itself
2050 * then attach a packet tag that holds it.
2052 if ((m->m_flags & M_VLANTAG) &&
2053 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2054 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2057 "unable to prepend VLAN header\n");
2058 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2061 m->m_flags &= ~M_VLANTAG;
2064 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2065 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2068 for (m = m0, n = 1; m != NULL; m = m0, n++) {
2072 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, n);
2076 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2077 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2078 if (mflags & M_MCAST)
2079 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2088 * Receive a queued packet from dummynet and pass it on to the output
2091 * The mbuf has the Ethernet header already attached.
2094 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2096 struct bridge_softc *sc;
2098 sc = ifp->if_bridge;
2101 * The packet didnt originate from a member interface. This should only
2102 * ever happen if a member interface is removed while packets are
2110 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2112 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2115 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2121 bridge_enqueue(sc, ifp, m);
2127 * Send output from a bridge member interface. This
2128 * performs the bridging function for locally originated
2131 * The mbuf has the Ethernet header already attached. We must
2132 * enqueue or free the mbuf before returning.
2135 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2138 struct ether_header *eh;
2139 struct ifnet *bifp, *dst_if;
2140 struct bridge_softc *sc;
2145 if (m->m_len < ETHER_HDR_LEN) {
2146 m = m_pullup(m, ETHER_HDR_LEN);
2151 eh = mtod(m, struct ether_header *);
2152 sc = ifp->if_bridge;
2153 vlan = VLANTAGOF(m);
2158 * If bridge is down, but the original output interface is up,
2159 * go ahead and send out that interface. Otherwise, the packet
2162 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2168 * If the packet is a multicast, or we don't know a better way to
2169 * get there, send to all interfaces.
2171 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2174 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2175 /* Tap any traffic not passing back out the originating interface */
2177 ETHER_BPF_MTAP(bifp, m);
2178 if (dst_if == NULL) {
2179 struct bridge_iflist *bif;
2185 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2186 dst_if = bif->bif_ifp;
2188 if (dst_if->if_type == IFT_GIF)
2190 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2194 * If this is not the original output interface,
2195 * and the interface is participating in spanning
2196 * tree, make sure the port is in a state that
2197 * allows forwarding.
2199 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2200 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2203 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2207 mc = m_dup(m, M_NOWAIT);
2209 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2214 bridge_enqueue(sc, dst_if, mc);
2223 * XXX Spanning tree consideration here?
2227 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2232 bridge_enqueue(sc, dst_if, m);
2239 * Do output on a bridge.
2243 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2245 struct bridge_softc *sc;
2246 struct ether_header *eh;
2247 struct ifnet *dst_if;
2252 ETHER_BPF_MTAP(ifp, m);
2254 eh = mtod(m, struct ether_header *);
2256 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2257 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2258 error = bridge_enqueue(sc, dst_if, m);
2260 bridge_broadcast(sc, ifp, m, 0);
2267 bridge_altq_start(if_t ifp)
2269 struct ifaltq *ifq = &ifp->if_snd;
2273 IFQ_DEQUEUE_NOLOCK(ifq, m);
2275 bridge_transmit(ifp, m);
2276 IFQ_DEQUEUE_NOLOCK(ifq, m);
2282 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2286 if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2287 IFQ_ENQUEUE(&ifp->if_snd, m, err);
2289 bridge_altq_start(ifp);
2291 err = bridge_transmit(ifp, m);
2298 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2301 bridge_qflush(struct ifnet *ifp __unused)
2308 * The forwarding function of the bridge.
2310 * NOTE: Releases the lock on return.
2313 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2316 struct bridge_iflist *dbif;
2317 struct ifnet *src_if, *dst_if, *ifp;
2318 struct ether_header *eh;
2325 src_if = m->m_pkthdr.rcvif;
2328 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2329 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2330 vlan = VLANTAGOF(m);
2332 if ((sbif->bif_flags & IFBIF_STP) &&
2333 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2336 eh = mtod(m, struct ether_header *);
2337 dst = eh->ether_dhost;
2339 /* If the interface is learning, record the address. */
2340 if (sbif->bif_flags & IFBIF_LEARNING) {
2341 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2342 sbif, 0, IFBAF_DYNAMIC);
2344 * If the interface has addresses limits then deny any source
2345 * that is not in the cache.
2347 if (error && sbif->bif_addrmax)
2351 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2352 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2356 * At this point, the port either doesn't participate
2357 * in spanning tree or it is in the forwarding state.
2361 * If the packet is unicast, destined for someone on
2362 * "this" side of the bridge, drop it.
2364 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2365 dst_if = bridge_rtlookup(sc, dst, vlan);
2366 if (src_if == dst_if)
2370 * Check if its a reserved multicast address, any address
2371 * listed in 802.1D section 7.12.6 may not be forwarded by the
2373 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2375 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2376 dst[2] == 0xc2 && dst[3] == 0x00 &&
2377 dst[4] == 0x00 && dst[5] <= 0x0f)
2380 /* ...forward it to all interfaces. */
2381 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2386 * If we have a destination interface which is a member of our bridge,
2387 * OR this is a unicast packet, push it through the bpf(4) machinery.
2388 * For broadcast or multicast packets, don't bother because it will
2389 * be reinjected into ether_input. We do this before we pass the packets
2390 * through the pfil(9) framework, as it is possible that pfil(9) will
2391 * drop the packet, or possibly modify it, making it difficult to debug
2392 * firewall issues on the bridge.
2394 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2395 ETHER_BPF_MTAP(ifp, m);
2397 /* run the packet filter */
2398 if (PFIL_HOOKED_IN(V_inet_pfil_head)
2400 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2403 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2409 if (dst_if == NULL) {
2410 bridge_broadcast(sc, src_if, m, 1);
2415 * At this point, we're dealing with a unicast frame
2416 * going to a different interface.
2418 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2421 dbif = bridge_lookup_member_if(sc, dst_if);
2423 /* Not a member of the bridge (anymore?) */
2426 /* Private segments can not talk to each other */
2427 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2430 if ((dbif->bif_flags & IFBIF_STP) &&
2431 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2434 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2436 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2439 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2445 bridge_enqueue(sc, dst_if, m);
2455 * Receive input from a member interface. Queue the packet for
2456 * bridging if it is not for us.
2458 static struct mbuf *
2459 bridge_input(struct ifnet *ifp, struct mbuf *m)
2461 struct bridge_softc *sc = ifp->if_bridge;
2462 struct bridge_iflist *bif, *bif2;
2464 struct ether_header *eh;
2465 struct mbuf *mc, *mc2;
2471 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2475 vlan = VLANTAGOF(m);
2478 * Implement support for bridge monitoring. If this flag has been
2479 * set on this interface, discard the packet once we push it through
2480 * the bpf(4) machinery, but before we do, increment the byte and
2481 * packet counters associated with this interface.
2483 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2484 m->m_pkthdr.rcvif = bifp;
2485 ETHER_BPF_MTAP(bifp, m);
2486 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2487 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2491 bif = bridge_lookup_member_if(sc, ifp);
2496 eh = mtod(m, struct ether_header *);
2500 if (m->m_flags & (M_BCAST|M_MCAST)) {
2501 /* Tap off 802.1D packets; they do not get forwarded. */
2502 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2503 ETHER_ADDR_LEN) == 0) {
2504 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2508 if ((bif->bif_flags & IFBIF_STP) &&
2509 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2514 * Make a deep copy of the packet and enqueue the copy
2515 * for bridge processing; return the original packet for
2518 mc = m_dup(m, M_NOWAIT);
2523 /* Perform the bridge forwarding function with the copy. */
2524 bridge_forward(sc, bif, mc);
2527 * Reinject the mbuf as arriving on the bridge so we have a
2528 * chance at claiming multicast packets. We can not loop back
2529 * here from ether_input as a bridge is never a member of a
2532 KASSERT(bifp->if_bridge == NULL,
2533 ("loop created in bridge_input"));
2534 mc2 = m_dup(m, M_NOWAIT);
2536 /* Keep the layer3 header aligned */
2537 int i = min(mc2->m_pkthdr.len, max_protohdr);
2538 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2541 mc2->m_pkthdr.rcvif = bifp;
2542 (*bifp->if_input)(bifp, mc2);
2545 /* Return the original packet for local processing. */
2549 if ((bif->bif_flags & IFBIF_STP) &&
2550 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2554 #if (defined(INET) || defined(INET6))
2555 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2556 || ((iface)->if_carp \
2557 && (*carp_forus_p)((iface), eh->ether_dhost))
2558 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2559 || ((iface)->if_carp \
2560 && (*carp_forus_p)((iface), eh->ether_shost))
2562 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2563 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2567 # define OR_PFIL_HOOKED_INET6 \
2568 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2570 # define OR_PFIL_HOOKED_INET6
2573 #define GRAB_OUR_PACKETS(iface) \
2574 if ((iface)->if_type == IFT_GIF) \
2576 /* It is destined for us. */ \
2577 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2578 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2580 if (bif->bif_flags & IFBIF_LEARNING) { \
2581 error = bridge_rtupdate(sc, eh->ether_shost, \
2582 vlan, bif, 0, IFBAF_DYNAMIC); \
2583 if (error && bif->bif_addrmax) { \
2588 m->m_pkthdr.rcvif = iface; \
2589 if ((iface) == ifp) { \
2590 /* Skip bridge processing... src == dest */ \
2593 /* It's passing over or to the bridge, locally. */ \
2594 ETHER_BPF_MTAP(bifp, m); \
2595 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \
2596 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2597 /* Filter on the physical interface. */ \
2598 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2599 OR_PFIL_HOOKED_INET6)) { \
2600 if (bridge_pfil(&m, NULL, ifp, \
2601 PFIL_IN) != 0 || m == NULL) { \
2605 if ((iface) != bifp) \
2606 ETHER_BPF_MTAP(iface, m); \
2610 /* We just received a packet that we sent out. */ \
2611 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2612 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2619 * Unicast. Make sure it's not for the bridge.
2621 do { GRAB_OUR_PACKETS(bifp) } while (0);
2624 * Give a chance for ifp at first priority. This will help when the
2625 * packet comes through the interface like VLAN's with the same MACs
2626 * on several interfaces from the same bridge. This also will save
2627 * some CPU cycles in case the destination interface and the input
2628 * interface (eq ifp) are the same.
2630 do { GRAB_OUR_PACKETS(ifp) } while (0);
2632 /* Now check the all bridge members. */
2633 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2634 GRAB_OUR_PACKETS(bif2->bif_ifp)
2637 #undef OR_CARP_CHECK_WE_ARE_DST
2638 #undef OR_CARP_CHECK_WE_ARE_SRC
2639 #undef OR_PFIL_HOOKED_INET6
2640 #undef GRAB_OUR_PACKETS
2642 /* Perform the bridge forwarding function. */
2643 bridge_forward(sc, bif, m);
2651 * Send a frame to all interfaces that are members of
2652 * the bridge, except for the one on which the packet
2655 * NOTE: Releases the lock on return.
2658 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2659 struct mbuf *m, int runfilt)
2661 struct bridge_iflist *dbif, *sbif;
2663 struct ifnet *dst_if;
2668 sbif = bridge_lookup_member_if(sc, src_if);
2670 /* Filter on the bridge interface before broadcasting */
2671 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2673 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2676 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2682 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2683 dst_if = dbif->bif_ifp;
2684 if (dst_if == src_if)
2687 /* Private segments can not talk to each other */
2688 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2691 if ((dbif->bif_flags & IFBIF_STP) &&
2692 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2695 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2696 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2699 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2702 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2706 mc = m_dup(m, M_NOWAIT);
2708 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2714 * Filter on the output interface. Pass a NULL bridge interface
2715 * pointer so we do not redundantly filter on the bridge for
2716 * each interface we broadcast on.
2718 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2720 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2724 /* Keep the layer3 header aligned */
2725 i = min(mc->m_pkthdr.len, max_protohdr);
2726 mc = m_copyup(mc, i, ETHER_ALIGN);
2728 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2732 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2738 bridge_enqueue(sc, dst_if, mc);
2747 * Duplicate a packet out one or more interfaces that are in span mode,
2748 * the original mbuf is unmodified.
2751 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2753 struct bridge_iflist *bif;
2754 struct ifnet *dst_if;
2759 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2762 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2763 dst_if = bif->bif_ifp;
2765 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2768 mc = m_dup(m, M_NOWAIT);
2770 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2774 bridge_enqueue(sc, dst_if, mc);
2781 * Add a bridge routing entry.
2784 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2785 struct bridge_iflist *bif, int setflags, uint8_t flags)
2787 struct bridge_rtnode *brt;
2790 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2792 /* Check the source address is valid and not multicast. */
2793 if (ETHER_IS_MULTICAST(dst) ||
2794 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2795 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2798 /* 802.1p frames map to vlan 1 */
2803 * A route for this destination might already exist. If so,
2804 * update it, otherwise create a new one.
2806 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2809 /* Check again, now that we have the lock. There could have
2810 * been a race and we only want to insert this once. */
2811 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2812 BRIDGE_RT_UNLOCK(sc);
2816 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2817 sc->sc_brtexceeded++;
2818 BRIDGE_RT_UNLOCK(sc);
2821 /* Check per interface address limits (if enabled) */
2822 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2823 bif->bif_addrexceeded++;
2824 BRIDGE_RT_UNLOCK(sc);
2829 * Allocate a new bridge forwarding node, and
2830 * initialize the expiration time and Ethernet
2833 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2835 BRIDGE_RT_UNLOCK(sc);
2838 brt->brt_vnet = curvnet;
2840 if (bif->bif_flags & IFBIF_STICKY)
2841 brt->brt_flags = IFBAF_STICKY;
2843 brt->brt_flags = IFBAF_DYNAMIC;
2845 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2846 brt->brt_vlan = vlan;
2848 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2849 uma_zfree(V_bridge_rtnode_zone, brt);
2850 BRIDGE_RT_UNLOCK(sc);
2856 BRIDGE_RT_UNLOCK(sc);
2859 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2860 brt->brt_dst != bif) {
2862 brt->brt_dst->bif_addrcnt--;
2864 brt->brt_dst->bif_addrcnt++;
2865 BRIDGE_RT_UNLOCK(sc);
2868 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2869 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2871 brt->brt_flags = flags;
2879 * Lookup the destination interface for an address.
2881 static struct ifnet *
2882 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2884 struct bridge_rtnode *brt;
2888 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2891 return (brt->brt_ifp);
2897 * Trim the routine table so that we have a number
2898 * of routing entries less than or equal to the
2902 bridge_rttrim(struct bridge_softc *sc)
2904 struct bridge_rtnode *brt, *nbrt;
2907 BRIDGE_RT_LOCK_ASSERT(sc);
2909 /* Make sure we actually need to do this. */
2910 if (sc->sc_brtcnt <= sc->sc_brtmax)
2913 /* Force an aging cycle; this might trim enough addresses. */
2915 if (sc->sc_brtcnt <= sc->sc_brtmax)
2918 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2919 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2920 bridge_rtnode_destroy(sc, brt);
2921 if (sc->sc_brtcnt <= sc->sc_brtmax)
2930 * Aging timer for the bridge.
2933 bridge_timer(void *arg)
2935 struct bridge_softc *sc = arg;
2937 BRIDGE_RT_LOCK_ASSERT(sc);
2939 /* Destruction of rtnodes requires a proper vnet context */
2940 CURVNET_SET(sc->sc_ifp->if_vnet);
2943 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2944 callout_reset(&sc->sc_brcallout,
2945 bridge_rtable_prune_period * hz, bridge_timer, sc);
2952 * Perform an aging cycle.
2955 bridge_rtage(struct bridge_softc *sc)
2957 struct bridge_rtnode *brt, *nbrt;
2959 BRIDGE_RT_LOCK_ASSERT(sc);
2961 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2962 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2963 if (time_uptime >= brt->brt_expire)
2964 bridge_rtnode_destroy(sc, brt);
2972 * Remove all dynamic addresses from the bridge.
2975 bridge_rtflush(struct bridge_softc *sc, int full)
2977 struct bridge_rtnode *brt, *nbrt;
2979 BRIDGE_RT_LOCK_ASSERT(sc);
2981 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2982 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2983 bridge_rtnode_destroy(sc, brt);
2990 * Remove an address from the table.
2993 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2995 struct bridge_rtnode *brt;
3001 * If vlan is zero then we want to delete for all vlans so the lookup
3002 * may return more than one.
3004 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
3005 bridge_rtnode_destroy(sc, brt);
3009 BRIDGE_RT_UNLOCK(sc);
3011 return (found ? 0 : ENOENT);
3017 * Delete routes to a speicifc member interface.
3020 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3022 struct bridge_rtnode *brt, *nbrt;
3024 BRIDGE_RT_LOCK_ASSERT(sc);
3026 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3027 if (brt->brt_ifp == ifp && (full ||
3028 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3029 bridge_rtnode_destroy(sc, brt);
3034 * bridge_rtable_init:
3036 * Initialize the route table for this bridge.
3039 bridge_rtable_init(struct bridge_softc *sc)
3043 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3044 M_DEVBUF, M_WAITOK);
3046 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3047 CK_LIST_INIT(&sc->sc_rthash[i]);
3049 sc->sc_rthash_key = arc4random();
3050 CK_LIST_INIT(&sc->sc_rtlist);
3054 * bridge_rtable_fini:
3056 * Deconstruct the route table for this bridge.
3059 bridge_rtable_fini(struct bridge_softc *sc)
3062 KASSERT(sc->sc_brtcnt == 0,
3063 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3064 free(sc->sc_rthash, M_DEVBUF);
3068 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3069 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3071 #define mix(a, b, c) \
3073 a -= b; a -= c; a ^= (c >> 13); \
3074 b -= c; b -= a; b ^= (a << 8); \
3075 c -= a; c -= b; c ^= (b >> 13); \
3076 a -= b; a -= c; a ^= (c >> 12); \
3077 b -= c; b -= a; b ^= (a << 16); \
3078 c -= a; c -= b; c ^= (b >> 5); \
3079 a -= b; a -= c; a ^= (c >> 3); \
3080 b -= c; b -= a; b ^= (a << 10); \
3081 c -= a; c -= b; c ^= (b >> 15); \
3082 } while (/*CONSTCOND*/0)
3084 static __inline uint32_t
3085 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3087 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3098 return (c & BRIDGE_RTHASH_MASK);
3104 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3108 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3109 d = ((int)a[i]) - ((int)b[i]);
3116 * bridge_rtnode_lookup:
3118 * Look up a bridge route node for the specified destination. Compare the
3119 * vlan id or if zero then just return the first match.
3121 static struct bridge_rtnode *
3122 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3124 struct bridge_rtnode *brt;
3128 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3130 hash = bridge_rthash(sc, addr);
3131 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3132 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3133 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3143 * bridge_rtnode_insert:
3145 * Insert the specified bridge node into the route table. We
3146 * assume the entry is not already in the table.
3149 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3151 struct bridge_rtnode *lbrt;
3155 BRIDGE_RT_LOCK_ASSERT(sc);
3157 hash = bridge_rthash(sc, brt->brt_addr);
3159 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3161 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3166 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3167 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3170 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3173 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3174 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3177 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3178 } while (lbrt != NULL);
3181 panic("bridge_rtnode_insert: impossible");
3185 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3192 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3194 struct bridge_rtnode *brt;
3196 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3198 CURVNET_SET(brt->brt_vnet);
3199 uma_zfree(V_bridge_rtnode_zone, brt);
3204 * bridge_rtnode_destroy:
3206 * Destroy a bridge rtnode.
3209 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3211 BRIDGE_RT_LOCK_ASSERT(sc);
3213 CK_LIST_REMOVE(brt, brt_hash);
3215 CK_LIST_REMOVE(brt, brt_list);
3217 brt->brt_dst->bif_addrcnt--;
3219 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3223 * bridge_rtable_expire:
3225 * Set the expiry time for all routes on an interface.
3228 bridge_rtable_expire(struct ifnet *ifp, int age)
3230 struct bridge_softc *sc = ifp->if_bridge;
3231 struct bridge_rtnode *brt;
3233 CURVNET_SET(ifp->if_vnet);
3237 * If the age is zero then flush, otherwise set all the expiry times to
3238 * age for the interface
3241 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3243 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3244 /* Cap the expiry time to 'age' */
3245 if (brt->brt_ifp == ifp &&
3246 brt->brt_expire > time_uptime + age &&
3247 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3248 brt->brt_expire = time_uptime + age;
3251 BRIDGE_RT_UNLOCK(sc);
3256 * bridge_state_change:
3258 * Callback from the bridgestp code when a port changes states.
3261 bridge_state_change(struct ifnet *ifp, int state)
3263 struct bridge_softc *sc = ifp->if_bridge;
3264 static const char *stpstates[] = {
3273 CURVNET_SET(ifp->if_vnet);
3275 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3276 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3281 * Send bridge packets through pfil if they are one of the types pfil can deal
3282 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3283 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3287 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3289 int snap, error, i, hlen;
3290 struct ether_header *eh1, eh2;
3293 u_int16_t ether_type;
3297 error = -1; /* Default error if not error == 0 */
3300 /* we may return with the IP fields swapped, ensure its not shared */
3301 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3304 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3305 return (0); /* filtering is disabled */
3307 i = min((*mp)->m_pkthdr.len, max_protohdr);
3308 if ((*mp)->m_len < i) {
3309 *mp = m_pullup(*mp, i);
3311 printf("%s: m_pullup failed\n", __func__);
3316 eh1 = mtod(*mp, struct ether_header *);
3317 ether_type = ntohs(eh1->ether_type);
3320 * Check for SNAP/LLC.
3322 if (ether_type < ETHERMTU) {
3323 struct llc *llc2 = (struct llc *)(eh1 + 1);
3325 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3326 llc2->llc_dsap == LLC_SNAP_LSAP &&
3327 llc2->llc_ssap == LLC_SNAP_LSAP &&
3328 llc2->llc_control == LLC_UI) {
3329 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3335 * If we're trying to filter bridge traffic, don't look at anything
3336 * other than IP and ARP traffic. If the filter doesn't understand
3337 * IPv6, don't allow IPv6 through the bridge either. This is lame
3338 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3339 * but of course we don't have an AppleTalk filter to begin with.
3340 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3343 switch (ether_type) {
3345 case ETHERTYPE_REVARP:
3346 if (V_pfil_ipfw_arp == 0)
3347 return (0); /* Automatically pass */
3352 case ETHERTYPE_IPV6:
3357 * Check to see if the user wants to pass non-ip
3358 * packets, these will not be checked by pfil(9) and
3359 * passed unconditionally so the default is to drop.
3365 /* Run the packet through pfil before stripping link headers */
3366 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3367 dir == PFIL_OUT && ifp != NULL) {
3368 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3376 /* Strip off the Ethernet header and keep a copy. */
3377 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3378 m_adj(*mp, ETHER_HDR_LEN);
3380 /* Strip off snap header, if present */
3382 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3383 m_adj(*mp, sizeof(struct llc));
3387 * Check the IP header for alignment and errors
3389 if (dir == PFIL_IN) {
3390 switch (ether_type) {
3392 error = bridge_ip_checkbasic(mp);
3395 case ETHERTYPE_IPV6:
3396 error = bridge_ip6_checkbasic(mp);
3409 * Run the packet through pfil
3412 switch (ether_type) {
3415 * Run pfil on the member interface and the bridge, both can
3416 * be skipped by clearing pfil_member or pfil_bridge.
3419 * in_if -> bridge_if -> out_if
3421 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3422 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3426 if (V_pfil_member && ifp != NULL && (rv =
3427 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3431 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3432 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3436 /* check if we need to fragment the packet */
3437 /* bridge_fragment generates a mbuf chain of packets */
3438 /* that already include eth headers */
3439 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3440 i = (*mp)->m_pkthdr.len;
3441 if (i > ifp->if_mtu) {
3442 error = bridge_fragment(ifp, mp, &eh2, snap,
3448 /* Recalculate the ip checksum. */
3449 ip = mtod(*mp, struct ip *);
3450 hlen = ip->ip_hl << 2;
3451 if (hlen < sizeof(struct ip))
3453 if (hlen > (*mp)->m_len) {
3454 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3456 ip = mtod(*mp, struct ip *);
3461 if (hlen == sizeof(struct ip))
3462 ip->ip_sum = in_cksum_hdr(ip);
3464 ip->ip_sum = in_cksum(*mp, hlen);
3468 case ETHERTYPE_IPV6:
3469 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3470 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3474 if (V_pfil_member && ifp != NULL && (rv =
3475 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3479 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3480 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3499 * Finally, put everything back the way it was and return
3502 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3505 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3508 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3511 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3522 * Perform basic checks on header size since
3523 * pfil assumes ip_input has already processed
3524 * it for it. Cut-and-pasted from ip_input.c.
3525 * Given how simple the IPv6 version is,
3526 * does the IPv4 version really need to be
3529 * XXX Should we update ipstat here, or not?
3530 * XXX Right now we update ipstat but not
3534 bridge_ip_checkbasic(struct mbuf **mp)
3536 struct mbuf *m = *mp;
3544 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3545 if ((m = m_copyup(m, sizeof(struct ip),
3546 (max_linkhdr + 3) & ~3)) == NULL) {
3547 /* XXXJRT new stat, please */
3548 KMOD_IPSTAT_INC(ips_toosmall);
3551 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3552 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3553 KMOD_IPSTAT_INC(ips_toosmall);
3557 ip = mtod(m, struct ip *);
3558 if (ip == NULL) goto bad;
3560 if (ip->ip_v != IPVERSION) {
3561 KMOD_IPSTAT_INC(ips_badvers);
3564 hlen = ip->ip_hl << 2;
3565 if (hlen < sizeof(struct ip)) { /* minimum header length */
3566 KMOD_IPSTAT_INC(ips_badhlen);
3569 if (hlen > m->m_len) {
3570 if ((m = m_pullup(m, hlen)) == NULL) {
3571 KMOD_IPSTAT_INC(ips_badhlen);
3574 ip = mtod(m, struct ip *);
3575 if (ip == NULL) goto bad;
3578 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3579 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3581 if (hlen == sizeof(struct ip)) {
3582 sum = in_cksum_hdr(ip);
3584 sum = in_cksum(m, hlen);
3588 KMOD_IPSTAT_INC(ips_badsum);
3592 /* Retrieve the packet length. */
3593 len = ntohs(ip->ip_len);
3596 * Check for additional length bogosity
3599 KMOD_IPSTAT_INC(ips_badlen);
3604 * Check that the amount of data in the buffers
3605 * is as at least much as the IP header would have us expect.
3606 * Drop packet if shorter than we expect.
3608 if (m->m_pkthdr.len < len) {
3609 KMOD_IPSTAT_INC(ips_tooshort);
3613 /* Checks out, proceed */
3624 * Same as above, but for IPv6.
3625 * Cut-and-pasted from ip6_input.c.
3626 * XXX Should we update ip6stat, or not?
3629 bridge_ip6_checkbasic(struct mbuf **mp)
3631 struct mbuf *m = *mp;
3632 struct ip6_hdr *ip6;
3635 * If the IPv6 header is not aligned, slurp it up into a new
3636 * mbuf with space for link headers, in the event we forward
3637 * it. Otherwise, if it is aligned, make sure the entire base
3638 * IPv6 header is in the first mbuf of the chain.
3640 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3641 struct ifnet *inifp = m->m_pkthdr.rcvif;
3642 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3643 (max_linkhdr + 3) & ~3)) == NULL) {
3644 /* XXXJRT new stat, please */
3645 IP6STAT_INC(ip6s_toosmall);
3646 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3649 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3650 struct ifnet *inifp = m->m_pkthdr.rcvif;
3651 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3652 IP6STAT_INC(ip6s_toosmall);
3653 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3658 ip6 = mtod(m, struct ip6_hdr *);
3660 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3661 IP6STAT_INC(ip6s_badvers);
3662 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3666 /* Checks out, proceed */
3679 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3682 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3683 int snap, struct llc *llc)
3685 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3689 if (m->m_len < sizeof(struct ip) &&
3690 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3692 ip = mtod(m, struct ip *);
3694 m->m_pkthdr.csum_flags |= CSUM_IP;
3695 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3700 * Walk the chain and re-add the Ethernet header for
3703 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3704 nextpkt = mcur->m_nextpkt;
3705 mcur->m_nextpkt = NULL;
3707 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3711 mprev->m_nextpkt = nextpkt;
3714 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3717 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3721 mprev->m_nextpkt = nextpkt;
3724 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3727 * The previous two M_PREPEND could have inserted one or two
3728 * mbufs in front so we have to update the previous packet's
3731 mcur->m_nextpkt = nextpkt;
3733 mprev->m_nextpkt = mcur;
3735 /* The first mbuf in the original chain needs to be
3742 KMOD_IPSTAT_INC(ips_fragmented);
3746 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3747 m = mcur->m_nextpkt;
3754 bridge_linkstate(struct ifnet *ifp)
3756 struct bridge_softc *sc = ifp->if_bridge;
3757 struct bridge_iflist *bif;
3758 struct epoch_tracker et;
3760 NET_EPOCH_ENTER(et);
3762 bif = bridge_lookup_member_if(sc, ifp);
3767 bridge_linkcheck(sc);
3769 bstp_linkstate(&bif->bif_stp);
3775 bridge_linkcheck(struct bridge_softc *sc)
3777 struct bridge_iflist *bif;
3778 int new_link, hasls;
3780 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3782 new_link = LINK_STATE_DOWN;
3784 /* Our link is considered up if at least one of our ports is active */
3785 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3786 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3788 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3789 new_link = LINK_STATE_UP;
3793 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3794 /* If no interfaces support link-state then we default to up */
3795 new_link = LINK_STATE_UP;
3797 if_link_state_change(sc->sc_ifp, new_link);