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>
81 #include "opt_inet6.h"
83 #include <sys/param.h>
84 #include <sys/eventhandler.h>
86 #include <sys/malloc.h>
87 #include <sys/protosw.h>
88 #include <sys/systm.h>
91 #include <sys/socket.h> /* for net/if.h */
92 #include <sys/sockio.h>
93 #include <sys/ctype.h> /* string functions */
94 #include <sys/kernel.h>
95 #include <sys/random.h>
96 #include <sys/syslog.h>
97 #include <sys/sysctl.h>
99 #include <sys/module.h>
100 #include <sys/priv.h>
101 #include <sys/proc.h>
102 #include <sys/lock.h>
103 #include <sys/mutex.h>
107 #include <net/if_clone.h>
108 #include <net/if_dl.h>
109 #include <net/if_types.h>
110 #include <net/if_var.h>
111 #include <net/pfil.h>
112 #include <net/vnet.h>
114 #include <netinet/in.h>
115 #include <netinet/in_systm.h>
116 #include <netinet/in_var.h>
117 #include <netinet/ip.h>
118 #include <netinet/ip_var.h>
120 #include <netinet/ip6.h>
121 #include <netinet6/ip6_var.h>
122 #include <netinet6/in6_ifattach.h>
124 #if defined(INET) || defined(INET6)
125 #include <netinet/ip_carp.h>
127 #include <machine/in_cksum.h>
128 #include <netinet/if_ether.h>
129 #include <net/bridgestp.h>
130 #include <net/if_bridgevar.h>
131 #include <net/if_llc.h>
132 #include <net/if_vlan_var.h>
134 #include <net/route.h>
138 * XXX: declare here to avoid to include many inet6 related files..
139 * should be more generalized?
141 extern void nd6_setmtu(struct ifnet *);
145 * Size of the route hash table. Must be a power of two.
147 #ifndef BRIDGE_RTHASH_SIZE
148 #define BRIDGE_RTHASH_SIZE 1024
151 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1)
154 * Default maximum number of addresses to cache.
156 #ifndef BRIDGE_RTABLE_MAX
157 #define BRIDGE_RTABLE_MAX 2000
161 * Timeout (in seconds) for entries learned dynamically.
163 #ifndef BRIDGE_RTABLE_TIMEOUT
164 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */
168 * Number of seconds between walks of the route list.
170 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
171 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60)
175 * List of capabilities to possibly mask on the member interface.
177 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
181 * List of capabilities to strip
183 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO
188 * The bridge relies heavily on the epoch(9) system to protect its data
189 * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we
190 * must ensure there is only one writer at a time.
192 * That is: for read accesses we only need to be in NET_EPOCH, but for write
193 * accesses we must hold:
195 * - BRIDGE_RT_LOCK, for any change to bridge_rtnodes
196 * - BRIDGE_LOCK, for any other change
198 * The BRIDGE_LOCK is a sleepable lock, because it is held across ioctl()
199 * calls to bridge member interfaces and these ioctl()s can sleep.
200 * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes
201 * required while we're in NET_EPOCH and then we're not allowed to sleep.
203 #define BRIDGE_LOCK_INIT(_sc) do { \
204 sx_init(&(_sc)->sc_sx, "if_bridge"); \
205 mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF); \
207 #define BRIDGE_LOCK_DESTROY(_sc) do { \
208 sx_destroy(&(_sc)->sc_sx); \
209 mtx_destroy(&(_sc)->sc_rt_mtx); \
211 #define BRIDGE_LOCK(_sc) sx_xlock(&(_sc)->sc_sx)
212 #define BRIDGE_UNLOCK(_sc) sx_xunlock(&(_sc)->sc_sx)
213 #define BRIDGE_LOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_XLOCKED)
214 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
215 MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx))
216 #define BRIDGE_UNLOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_UNLOCKED)
217 #define BRIDGE_RT_LOCK(_sc) mtx_lock(&(_sc)->sc_rt_mtx)
218 #define BRIDGE_RT_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_rt_mtx)
219 #define BRIDGE_RT_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED)
220 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
221 MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx))
224 * Bridge interface list entry.
226 struct bridge_iflist {
227 CK_LIST_ENTRY(bridge_iflist) bif_next;
228 struct ifnet *bif_ifp; /* member if */
229 struct bstp_port bif_stp; /* STP state */
230 uint32_t bif_flags; /* member if flags */
231 int bif_savedcaps; /* saved capabilities */
232 uint32_t bif_addrmax; /* max # of addresses */
233 uint32_t bif_addrcnt; /* cur. # of addresses */
234 uint32_t bif_addrexceeded;/* # of address violations */
235 struct epoch_context bif_epoch_ctx;
241 struct bridge_rtnode {
242 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
243 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
244 struct bridge_iflist *brt_dst; /* destination if */
245 unsigned long brt_expire; /* expiration time */
246 uint8_t brt_flags; /* address flags */
247 uint8_t brt_addr[ETHER_ADDR_LEN];
248 uint16_t brt_vlan; /* vlan id */
249 struct vnet *brt_vnet;
250 struct epoch_context brt_epoch_ctx;
252 #define brt_ifp brt_dst->bif_ifp
255 * Software state for each bridge.
257 struct bridge_softc {
258 struct ifnet *sc_ifp; /* make this an interface */
259 LIST_ENTRY(bridge_softc) sc_list;
261 struct mtx sc_rt_mtx;
262 uint32_t sc_brtmax; /* max # of addresses */
263 uint32_t sc_brtcnt; /* cur. # of addresses */
264 uint32_t sc_brttimeout; /* rt timeout in seconds */
265 struct callout sc_brcallout; /* bridge callout */
266 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
267 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
268 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
269 uint32_t sc_rthash_key; /* key for hash */
270 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
271 struct bstp_state sc_stp; /* STP state */
272 uint32_t sc_brtexceeded; /* # of cache drops */
273 struct ifnet *sc_ifaddr; /* member mac copied from */
274 struct ether_addr sc_defaddr; /* Default MAC address */
275 struct epoch_context sc_epoch_ctx;
278 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
279 #define V_bridge_list_sx VNET(bridge_list_sx)
280 static eventhandler_tag bridge_detach_cookie;
282 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
284 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
285 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
287 static int bridge_clone_create(struct if_clone *, int, caddr_t);
288 static void bridge_clone_destroy(struct ifnet *);
290 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
291 static void bridge_mutecaps(struct bridge_softc *);
292 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
294 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
295 static void bridge_init(void *);
296 static void bridge_dummynet(struct mbuf *, struct ifnet *);
297 static void bridge_stop(struct ifnet *, int);
298 static int bridge_transmit(struct ifnet *, struct mbuf *);
300 static void bridge_altq_start(if_t);
301 static int bridge_altq_transmit(if_t, struct mbuf *);
303 static void bridge_qflush(struct ifnet *);
304 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
305 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
307 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
309 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
311 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
314 static void bridge_timer(void *);
316 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
318 static void bridge_span(struct bridge_softc *, struct mbuf *);
320 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
321 uint16_t, struct bridge_iflist *, int, uint8_t);
322 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
324 static void bridge_rttrim(struct bridge_softc *);
325 static void bridge_rtage(struct bridge_softc *);
326 static void bridge_rtflush(struct bridge_softc *, int);
327 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
330 static void bridge_rtable_init(struct bridge_softc *);
331 static void bridge_rtable_fini(struct bridge_softc *);
333 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
334 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
335 const uint8_t *, uint16_t);
336 static int bridge_rtnode_insert(struct bridge_softc *,
337 struct bridge_rtnode *);
338 static void bridge_rtnode_destroy(struct bridge_softc *,
339 struct bridge_rtnode *);
340 static void bridge_rtable_expire(struct ifnet *, int);
341 static void bridge_state_change(struct ifnet *, int);
343 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
345 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
347 static void bridge_delete_member(struct bridge_softc *,
348 struct bridge_iflist *, int);
349 static void bridge_delete_span(struct bridge_softc *,
350 struct bridge_iflist *);
352 static int bridge_ioctl_add(struct bridge_softc *, void *);
353 static int bridge_ioctl_del(struct bridge_softc *, void *);
354 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
355 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
356 static int bridge_ioctl_scache(struct bridge_softc *, void *);
357 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
358 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
359 static int bridge_ioctl_rts(struct bridge_softc *, void *);
360 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
361 static int bridge_ioctl_sto(struct bridge_softc *, void *);
362 static int bridge_ioctl_gto(struct bridge_softc *, void *);
363 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
364 static int bridge_ioctl_flush(struct bridge_softc *, void *);
365 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
366 static int bridge_ioctl_spri(struct bridge_softc *, void *);
367 static int bridge_ioctl_ght(struct bridge_softc *, void *);
368 static int bridge_ioctl_sht(struct bridge_softc *, void *);
369 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
370 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
371 static int bridge_ioctl_gma(struct bridge_softc *, void *);
372 static int bridge_ioctl_sma(struct bridge_softc *, void *);
373 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
374 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
375 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
376 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
377 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
378 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
379 static int bridge_ioctl_grte(struct bridge_softc *, void *);
380 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
381 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
382 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
383 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
385 static int bridge_ip_checkbasic(struct mbuf **mp);
387 static int bridge_ip6_checkbasic(struct mbuf **mp);
389 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
390 struct ether_header *, int, struct llc *);
391 static void bridge_linkstate(struct ifnet *ifp);
392 static void bridge_linkcheck(struct bridge_softc *sc);
394 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
395 #define VLANTAGOF(_m) \
396 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
398 static struct bstp_cb_ops bridge_ops = {
399 .bcb_state = bridge_state_change,
400 .bcb_rtage = bridge_rtable_expire
403 SYSCTL_DECL(_net_link);
404 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
407 /* only pass IP[46] packets when pfil is enabled */
408 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
409 #define V_pfil_onlyip VNET(pfil_onlyip)
410 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
411 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
412 "Only pass IP packets when pfil is enabled");
414 /* run pfil hooks on the bridge interface */
415 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
416 #define V_pfil_bridge VNET(pfil_bridge)
417 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
418 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
419 "Packet filter on the bridge interface");
421 /* layer2 filter with ipfw */
422 VNET_DEFINE_STATIC(int, pfil_ipfw);
423 #define V_pfil_ipfw VNET(pfil_ipfw)
425 /* layer2 ARP filter with ipfw */
426 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
427 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
428 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
429 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
430 "Filter ARP packets through IPFW layer2");
432 /* run pfil hooks on the member interface */
433 VNET_DEFINE_STATIC(int, pfil_member) = 1;
434 #define V_pfil_member VNET(pfil_member)
435 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
436 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
437 "Packet filter on the member interface");
439 /* run pfil hooks on the physical interface for locally destined packets */
440 VNET_DEFINE_STATIC(int, pfil_local_phys);
441 #define V_pfil_local_phys VNET(pfil_local_phys)
442 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
443 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
444 "Packet filter on the physical interface for locally destined packets");
446 /* log STP state changes */
447 VNET_DEFINE_STATIC(int, log_stp);
448 #define V_log_stp VNET(log_stp)
449 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
450 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
451 "Log STP state changes");
453 /* share MAC with first bridge member */
454 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
455 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
456 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
457 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
458 "Inherit MAC address from the first bridge member");
460 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
461 #define V_allow_llz_overlap VNET(allow_llz_overlap)
462 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
463 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
464 "Allow overlap of link-local scope "
465 "zones of a bridge interface and the member interfaces");
467 /* log MAC address port flapping */
468 VNET_DEFINE_STATIC(bool, log_mac_flap) = true;
469 #define V_log_mac_flap VNET(log_mac_flap)
470 SYSCTL_BOOL(_net_link_bridge, OID_AUTO, log_mac_flap,
471 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(log_mac_flap), true,
472 "Log MAC address port flapping");
474 VNET_DEFINE_STATIC(int, log_interval) = 5;
475 VNET_DEFINE_STATIC(int, log_count) = 0;
476 VNET_DEFINE_STATIC(struct timeval, log_last) = { 0 };
478 #define V_log_interval VNET(log_interval)
479 #define V_log_count VNET(log_count)
480 #define V_log_last VNET(log_last)
482 struct bridge_control {
483 int (*bc_func)(struct bridge_softc *, void *);
488 #define BC_F_COPYIN 0x01 /* copy arguments in */
489 #define BC_F_COPYOUT 0x02 /* copy arguments out */
490 #define BC_F_SUSER 0x04 /* do super-user check */
492 static const struct bridge_control bridge_control_table[] = {
493 { bridge_ioctl_add, sizeof(struct ifbreq),
494 BC_F_COPYIN|BC_F_SUSER },
495 { bridge_ioctl_del, sizeof(struct ifbreq),
496 BC_F_COPYIN|BC_F_SUSER },
498 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
499 BC_F_COPYIN|BC_F_COPYOUT },
500 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
501 BC_F_COPYIN|BC_F_SUSER },
503 { bridge_ioctl_scache, sizeof(struct ifbrparam),
504 BC_F_COPYIN|BC_F_SUSER },
505 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
508 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
509 BC_F_COPYIN|BC_F_COPYOUT },
510 { bridge_ioctl_rts, sizeof(struct ifbaconf),
511 BC_F_COPYIN|BC_F_COPYOUT },
513 { bridge_ioctl_saddr, sizeof(struct ifbareq),
514 BC_F_COPYIN|BC_F_SUSER },
516 { bridge_ioctl_sto, sizeof(struct ifbrparam),
517 BC_F_COPYIN|BC_F_SUSER },
518 { bridge_ioctl_gto, sizeof(struct ifbrparam),
521 { bridge_ioctl_daddr, sizeof(struct ifbareq),
522 BC_F_COPYIN|BC_F_SUSER },
524 { bridge_ioctl_flush, sizeof(struct ifbreq),
525 BC_F_COPYIN|BC_F_SUSER },
527 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
529 { bridge_ioctl_spri, sizeof(struct ifbrparam),
530 BC_F_COPYIN|BC_F_SUSER },
532 { bridge_ioctl_ght, sizeof(struct ifbrparam),
534 { bridge_ioctl_sht, sizeof(struct ifbrparam),
535 BC_F_COPYIN|BC_F_SUSER },
537 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
539 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
540 BC_F_COPYIN|BC_F_SUSER },
542 { bridge_ioctl_gma, sizeof(struct ifbrparam),
544 { bridge_ioctl_sma, sizeof(struct ifbrparam),
545 BC_F_COPYIN|BC_F_SUSER },
547 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
548 BC_F_COPYIN|BC_F_SUSER },
550 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
551 BC_F_COPYIN|BC_F_SUSER },
553 { bridge_ioctl_addspan, sizeof(struct ifbreq),
554 BC_F_COPYIN|BC_F_SUSER },
555 { bridge_ioctl_delspan, sizeof(struct ifbreq),
556 BC_F_COPYIN|BC_F_SUSER },
558 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
561 { bridge_ioctl_grte, sizeof(struct ifbrparam),
564 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
565 BC_F_COPYIN|BC_F_COPYOUT },
567 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
568 BC_F_COPYIN|BC_F_SUSER },
570 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
571 BC_F_COPYIN|BC_F_SUSER },
573 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
574 BC_F_COPYIN|BC_F_SUSER },
577 static const int bridge_control_table_size = nitems(bridge_control_table);
579 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
580 #define V_bridge_list VNET(bridge_list)
581 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \
583 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx)
584 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx)
585 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx)
587 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
588 #define V_bridge_cloner VNET(bridge_cloner)
590 static const char bridge_name[] = "bridge";
593 vnet_bridge_init(const void *unused __unused)
596 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
597 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
599 BRIDGE_LIST_LOCK_INIT();
600 LIST_INIT(&V_bridge_list);
601 V_bridge_cloner = if_clone_simple(bridge_name,
602 bridge_clone_create, bridge_clone_destroy, 0);
604 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
605 vnet_bridge_init, NULL);
608 vnet_bridge_uninit(const void *unused __unused)
611 if_clone_detach(V_bridge_cloner);
612 V_bridge_cloner = NULL;
613 BRIDGE_LIST_LOCK_DESTROY();
615 /* Callbacks may use the UMA zone. */
616 NET_EPOCH_DRAIN_CALLBACKS();
618 uma_zdestroy(V_bridge_rtnode_zone);
620 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
621 vnet_bridge_uninit, NULL);
624 bridge_modevent(module_t mod, int type, void *data)
629 bridge_dn_p = bridge_dummynet;
630 bridge_detach_cookie = EVENTHANDLER_REGISTER(
631 ifnet_departure_event, bridge_ifdetach, NULL,
632 EVENTHANDLER_PRI_ANY);
635 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
636 bridge_detach_cookie);
645 static moduledata_t bridge_mod = {
651 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
652 MODULE_VERSION(if_bridge, 1);
653 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
656 * handler for net.link.bridge.ipfw
659 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
661 int enable = V_pfil_ipfw;
664 error = sysctl_handle_int(oidp, &enable, 0, req);
667 if (enable != V_pfil_ipfw) {
668 V_pfil_ipfw = enable;
671 * Disable pfil so that ipfw doesnt run twice, if the user
672 * really wants both then they can re-enable pfil_bridge and/or
673 * pfil_member. Also allow non-ip packets as ipfw can filter by
685 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
686 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
687 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
688 "Layer2 filter with IPFW");
692 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
694 struct bridge_softc *sc = ifp->if_softc;
695 struct bridge_iflist *bif;
699 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
700 bridge_delete_member(sc, bif, 0);
702 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
703 bridge_delete_span(sc, bif);
708 ether_reassign(ifp, newvnet, arg);
713 * bridge_clone_create:
715 * Create a new bridge instance.
718 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
720 struct bridge_softc *sc;
723 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
724 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
730 BRIDGE_LOCK_INIT(sc);
731 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
732 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
734 /* Initialize our routing table. */
735 bridge_rtable_init(sc);
737 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
739 CK_LIST_INIT(&sc->sc_iflist);
740 CK_LIST_INIT(&sc->sc_spanlist);
743 if_initname(ifp, bridge_name, unit);
744 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
745 ifp->if_ioctl = bridge_ioctl;
747 ifp->if_start = bridge_altq_start;
748 ifp->if_transmit = bridge_altq_transmit;
749 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
750 ifp->if_snd.ifq_drv_maxlen = 0;
751 IFQ_SET_READY(&ifp->if_snd);
753 ifp->if_transmit = bridge_transmit;
755 ifp->if_qflush = bridge_qflush;
756 ifp->if_init = bridge_init;
757 ifp->if_type = IFT_BRIDGE;
759 ether_gen_addr(ifp, &sc->sc_defaddr);
761 bstp_attach(&sc->sc_stp, &bridge_ops);
762 ether_ifattach(ifp, sc->sc_defaddr.octet);
763 /* Now undo some of the damage... */
764 ifp->if_baudrate = 0;
765 ifp->if_type = IFT_BRIDGE;
767 ifp->if_reassign = bridge_reassign;
771 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
772 BRIDGE_LIST_UNLOCK();
778 bridge_clone_destroy_cb(struct epoch_context *ctx)
780 struct bridge_softc *sc;
782 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
784 BRIDGE_LOCK_DESTROY(sc);
789 * bridge_clone_destroy:
791 * Destroy a bridge instance.
794 bridge_clone_destroy(struct ifnet *ifp)
796 struct bridge_softc *sc = ifp->if_softc;
797 struct bridge_iflist *bif;
798 struct epoch_tracker et;
803 ifp->if_flags &= ~IFF_UP;
805 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
806 bridge_delete_member(sc, bif, 0);
808 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
809 bridge_delete_span(sc, bif);
812 /* Tear down the routing table. */
813 bridge_rtable_fini(sc);
819 callout_drain(&sc->sc_brcallout);
822 LIST_REMOVE(sc, sc_list);
823 BRIDGE_LIST_UNLOCK();
825 bstp_detach(&sc->sc_stp);
827 IFQ_PURGE(&ifp->if_snd);
834 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
840 * Handle a control request from the operator.
843 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
845 struct bridge_softc *sc = ifp->if_softc;
846 struct ifreq *ifr = (struct ifreq *)data;
847 struct bridge_iflist *bif;
848 struct thread *td = curthread;
850 struct ifbreq ifbreq;
851 struct ifbifconf ifbifconf;
852 struct ifbareq ifbareq;
853 struct ifbaconf ifbaconf;
854 struct ifbrparam ifbrparam;
855 struct ifbropreq ifbropreq;
857 struct ifdrv *ifd = (struct ifdrv *) data;
858 const struct bridge_control *bc;
859 int error = 0, oldmtu;
870 if (ifd->ifd_cmd >= bridge_control_table_size) {
874 bc = &bridge_control_table[ifd->ifd_cmd];
876 if (cmd == SIOCGDRVSPEC &&
877 (bc->bc_flags & BC_F_COPYOUT) == 0) {
881 else if (cmd == SIOCSDRVSPEC &&
882 (bc->bc_flags & BC_F_COPYOUT) != 0) {
887 if (bc->bc_flags & BC_F_SUSER) {
888 error = priv_check(td, PRIV_NET_BRIDGE);
893 if (ifd->ifd_len != bc->bc_argsize ||
894 ifd->ifd_len > sizeof(args)) {
899 bzero(&args, sizeof(args));
900 if (bc->bc_flags & BC_F_COPYIN) {
901 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
906 oldmtu = ifp->if_mtu;
907 error = (*bc->bc_func)(sc, &args);
912 * Bridge MTU may change during addition of the first port.
913 * If it did, do network layer specific procedure.
915 if (ifp->if_mtu != oldmtu) {
922 if (bc->bc_flags & BC_F_COPYOUT)
923 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
928 if (!(ifp->if_flags & IFF_UP) &&
929 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
931 * If interface is marked down and it is running,
932 * then stop and disable it.
935 } else if ((ifp->if_flags & IFF_UP) &&
936 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
938 * If interface is marked up and it is stopped, then
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 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
958 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
959 " != %d\n", sc->sc_ifp->if_xname,
960 bif->bif_ifp->if_mtu,
961 bif->bif_ifp->if_xname, ifr->ifr_mtu);
967 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
971 * drop the lock as ether_ioctl() will call bridge_start() and
972 * cause the lock to be recursed.
975 error = ether_ioctl(ifp, cmd, data);
988 * Clear or restore unwanted capabilities on the member interface
991 bridge_mutecaps(struct bridge_softc *sc)
993 struct bridge_iflist *bif;
996 BRIDGE_LOCK_ASSERT(sc);
998 /* Initial bitmask of capabilities to test */
999 mask = BRIDGE_IFCAPS_MASK;
1001 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1002 /* Every member must support it or its disabled */
1003 mask &= bif->bif_savedcaps;
1006 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1007 enabled = bif->bif_ifp->if_capenable;
1008 enabled &= ~BRIDGE_IFCAPS_STRIP;
1009 /* strip off mask bits and enable them again if allowed */
1010 enabled &= ~BRIDGE_IFCAPS_MASK;
1012 bridge_set_ifcap(sc, bif, enabled);
1017 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1019 struct ifnet *ifp = bif->bif_ifp;
1021 int error, mask, stuck;
1023 bzero(&ifr, sizeof(ifr));
1024 ifr.ifr_reqcap = set;
1026 if (ifp->if_capenable != set) {
1027 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1029 if_printf(sc->sc_ifp,
1030 "error setting capabilities on %s: %d\n",
1031 ifp->if_xname, error);
1032 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1033 stuck = ifp->if_capenable & mask & ~set;
1035 if_printf(sc->sc_ifp,
1036 "can't disable some capabilities on %s: 0x%x\n",
1037 ifp->if_xname, stuck);
1042 * bridge_lookup_member:
1044 * Lookup a bridge member interface.
1046 static struct bridge_iflist *
1047 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1049 struct bridge_iflist *bif;
1052 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1054 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1056 if (strcmp(ifp->if_xname, name) == 0)
1064 * bridge_lookup_member_if:
1066 * Lookup a bridge member interface by ifnet*.
1068 static struct bridge_iflist *
1069 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1071 struct bridge_iflist *bif;
1073 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1075 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1076 if (bif->bif_ifp == member_ifp)
1084 bridge_delete_member_cb(struct epoch_context *ctx)
1086 struct bridge_iflist *bif;
1088 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1090 free(bif, M_DEVBUF);
1094 * bridge_delete_member:
1096 * Delete the specified member interface.
1099 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1102 struct ifnet *ifs = bif->bif_ifp;
1103 struct ifnet *fif = NULL;
1104 struct bridge_iflist *bifl;
1106 BRIDGE_LOCK_ASSERT(sc);
1108 if (bif->bif_flags & IFBIF_STP)
1109 bstp_disable(&bif->bif_stp);
1111 ifs->if_bridge = NULL;
1112 CK_LIST_REMOVE(bif, bif_next);
1115 * If removing the interface that gave the bridge its mac address, set
1116 * the mac address of the bridge to the address of the next member, or
1117 * to its default address if no members are left.
1119 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1120 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1121 bcopy(&sc->sc_defaddr,
1122 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1123 sc->sc_ifaddr = NULL;
1125 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1126 fif = bifl->bif_ifp;
1127 bcopy(IF_LLADDR(fif),
1128 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1129 sc->sc_ifaddr = fif;
1131 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1134 bridge_linkcheck(sc);
1135 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1137 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1138 BRIDGE_RT_UNLOCK(sc);
1139 KASSERT(bif->bif_addrcnt == 0,
1140 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1142 ifs->if_bridge_output = NULL;
1143 ifs->if_bridge_input = NULL;
1144 ifs->if_bridge_linkstate = NULL;
1146 switch (ifs->if_type) {
1150 * Take the interface out of promiscuous mode, but only
1151 * if it was promiscuous in the first place. It might
1152 * not be if we're in the bridge_ioctl_add() error path.
1154 if (ifs->if_flags & IFF_PROMISC)
1155 (void) ifpromisc(ifs, 0);
1163 panic("bridge_delete_member: impossible");
1167 /* reneable any interface capabilities */
1168 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1170 bstp_destroy(&bif->bif_stp); /* prepare to free */
1172 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1176 * bridge_delete_span:
1178 * Delete the specified span interface.
1181 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1183 BRIDGE_LOCK_ASSERT(sc);
1185 KASSERT(bif->bif_ifp->if_bridge == NULL,
1186 ("%s: not a span interface", __func__));
1188 CK_LIST_REMOVE(bif, bif_next);
1190 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1194 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1196 struct ifbreq *req = arg;
1197 struct bridge_iflist *bif = NULL;
1201 ifs = ifunit(req->ifbr_ifsname);
1204 if (ifs->if_ioctl == NULL) /* must be supported */
1207 /* If it's in the span list, it can't be a member. */
1208 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1209 if (ifs == bif->bif_ifp)
1212 if (ifs->if_bridge == sc)
1215 if (ifs->if_bridge != NULL)
1218 switch (ifs->if_type) {
1222 /* permitted interface types */
1230 * Two valid inet6 addresses with link-local scope must not be
1231 * on the parent interface and the member interfaces at the
1232 * same time. This restriction is needed to prevent violation
1233 * of link-local scope zone. Attempts to add a member
1234 * interface which has inet6 addresses when the parent has
1235 * inet6 triggers removal of all inet6 addresses on the member
1239 /* Check if the parent interface has a link-local scope addr. */
1240 if (V_allow_llz_overlap == 0 &&
1241 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1243 * If any, remove all inet6 addresses from the member
1246 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1247 if (in6ifa_llaonifp(bif->bif_ifp)) {
1248 in6_ifdetach(bif->bif_ifp);
1249 if_printf(sc->sc_ifp,
1250 "IPv6 addresses on %s have been removed "
1251 "before adding it as a member to prevent "
1252 "IPv6 address scope violation.\n",
1253 bif->bif_ifp->if_xname);
1256 if (in6ifa_llaonifp(ifs)) {
1258 if_printf(sc->sc_ifp,
1259 "IPv6 addresses on %s have been removed "
1260 "before adding it as a member to prevent "
1261 "IPv6 address scope violation.\n",
1266 /* Allow the first Ethernet member to define the MTU */
1267 if (CK_LIST_EMPTY(&sc->sc_iflist))
1268 sc->sc_ifp->if_mtu = ifs->if_mtu;
1269 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1270 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1271 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1275 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1280 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1281 bif->bif_savedcaps = ifs->if_capenable;
1284 * Assign the interface's MAC address to the bridge if it's the first
1285 * member and the MAC address of the bridge has not been changed from
1286 * the default randomly generated one.
1288 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1289 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1290 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1291 sc->sc_ifaddr = ifs;
1292 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1295 ifs->if_bridge = sc;
1296 ifs->if_bridge_output = bridge_output;
1297 ifs->if_bridge_input = bridge_input;
1298 ifs->if_bridge_linkstate = bridge_linkstate;
1299 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1301 * XXX: XLOCK HERE!?!
1303 * NOTE: insert_***HEAD*** should be safe for the traversals.
1305 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1307 /* Set interface capabilities to the intersection set of all members */
1308 bridge_mutecaps(sc);
1309 bridge_linkcheck(sc);
1311 /* Place the interface into promiscuous mode */
1312 switch (ifs->if_type) {
1315 error = ifpromisc(ifs, 1);
1320 bridge_delete_member(sc, bif, 0);
1325 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1327 struct ifbreq *req = arg;
1328 struct bridge_iflist *bif;
1330 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1334 bridge_delete_member(sc, bif, 0);
1340 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1342 struct ifbreq *req = arg;
1343 struct bridge_iflist *bif;
1344 struct bstp_port *bp;
1346 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1351 req->ifbr_ifsflags = bif->bif_flags;
1352 req->ifbr_state = bp->bp_state;
1353 req->ifbr_priority = bp->bp_priority;
1354 req->ifbr_path_cost = bp->bp_path_cost;
1355 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1356 req->ifbr_proto = bp->bp_protover;
1357 req->ifbr_role = bp->bp_role;
1358 req->ifbr_stpflags = bp->bp_flags;
1359 req->ifbr_addrcnt = bif->bif_addrcnt;
1360 req->ifbr_addrmax = bif->bif_addrmax;
1361 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1363 /* Copy STP state options as flags */
1364 if (bp->bp_operedge)
1365 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1366 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1367 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1368 if (bp->bp_ptp_link)
1369 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1370 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1371 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1372 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1373 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1374 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1375 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1380 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1382 struct epoch_tracker et;
1383 struct ifbreq *req = arg;
1384 struct bridge_iflist *bif;
1385 struct bstp_port *bp;
1388 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1393 if (req->ifbr_ifsflags & IFBIF_SPAN)
1394 /* SPAN is readonly */
1397 NET_EPOCH_ENTER(et);
1399 if (req->ifbr_ifsflags & IFBIF_STP) {
1400 if ((bif->bif_flags & IFBIF_STP) == 0) {
1401 error = bstp_enable(&bif->bif_stp);
1408 if ((bif->bif_flags & IFBIF_STP) != 0)
1409 bstp_disable(&bif->bif_stp);
1412 /* Pass on STP flags */
1413 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1414 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1415 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1416 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1418 /* Save the bits relating to the bridge */
1419 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1427 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1429 struct ifbrparam *param = arg;
1431 sc->sc_brtmax = param->ifbrp_csize;
1438 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1440 struct ifbrparam *param = arg;
1442 param->ifbrp_csize = sc->sc_brtmax;
1448 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1450 struct ifbifconf *bifc = arg;
1451 struct bridge_iflist *bif;
1454 int count, buflen, len, error = 0;
1457 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1459 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1462 buflen = sizeof(breq) * count;
1463 if (bifc->ifbic_len == 0) {
1464 bifc->ifbic_len = buflen;
1467 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1473 len = min(bifc->ifbic_len, buflen);
1474 bzero(&breq, sizeof(breq));
1475 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1476 if (len < sizeof(breq))
1479 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1480 sizeof(breq.ifbr_ifsname));
1481 /* Fill in the ifbreq structure */
1482 error = bridge_ioctl_gifflags(sc, &breq);
1485 memcpy(buf, &breq, sizeof(breq));
1487 buf += sizeof(breq);
1488 len -= sizeof(breq);
1490 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1491 if (len < sizeof(breq))
1494 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1495 sizeof(breq.ifbr_ifsname));
1496 breq.ifbr_ifsflags = bif->bif_flags;
1497 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1498 memcpy(buf, &breq, sizeof(breq));
1500 buf += sizeof(breq);
1501 len -= sizeof(breq);
1504 bifc->ifbic_len = sizeof(breq) * count;
1505 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1506 free(outbuf, M_TEMP);
1511 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1513 struct ifbaconf *bac = arg;
1514 struct bridge_rtnode *brt;
1515 struct ifbareq bareq;
1517 int count, buflen, len, error = 0;
1519 if (bac->ifbac_len == 0)
1523 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1525 buflen = sizeof(bareq) * count;
1527 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1533 len = min(bac->ifbac_len, buflen);
1534 bzero(&bareq, sizeof(bareq));
1535 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1536 if (len < sizeof(bareq))
1538 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1539 sizeof(bareq.ifba_ifsname));
1540 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1541 bareq.ifba_vlan = brt->brt_vlan;
1542 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1543 time_uptime < brt->brt_expire)
1544 bareq.ifba_expire = brt->brt_expire - time_uptime;
1546 bareq.ifba_expire = 0;
1547 bareq.ifba_flags = brt->brt_flags;
1549 memcpy(buf, &bareq, sizeof(bareq));
1551 buf += sizeof(bareq);
1552 len -= sizeof(bareq);
1555 bac->ifbac_len = sizeof(bareq) * count;
1556 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1557 free(outbuf, M_TEMP);
1562 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1564 struct ifbareq *req = arg;
1565 struct bridge_iflist *bif;
1566 struct epoch_tracker et;
1569 NET_EPOCH_ENTER(et);
1570 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1576 /* bridge_rtupdate() may acquire the lock. */
1577 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1585 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1587 struct ifbrparam *param = arg;
1589 sc->sc_brttimeout = param->ifbrp_ctime;
1594 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1596 struct ifbrparam *param = arg;
1598 param->ifbrp_ctime = sc->sc_brttimeout;
1603 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1605 struct ifbareq *req = arg;
1607 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1611 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1613 struct ifbreq *req = arg;
1616 bridge_rtflush(sc, req->ifbr_ifsflags);
1617 BRIDGE_RT_UNLOCK(sc);
1623 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1625 struct ifbrparam *param = arg;
1626 struct bstp_state *bs = &sc->sc_stp;
1628 param->ifbrp_prio = bs->bs_bridge_priority;
1633 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1635 struct ifbrparam *param = arg;
1637 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1641 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1643 struct ifbrparam *param = arg;
1644 struct bstp_state *bs = &sc->sc_stp;
1646 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1651 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1653 struct ifbrparam *param = arg;
1655 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1659 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1661 struct ifbrparam *param = arg;
1662 struct bstp_state *bs = &sc->sc_stp;
1664 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1669 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1671 struct ifbrparam *param = arg;
1673 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1677 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1679 struct ifbrparam *param = arg;
1680 struct bstp_state *bs = &sc->sc_stp;
1682 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1687 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1689 struct ifbrparam *param = arg;
1691 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1695 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1697 struct ifbreq *req = arg;
1698 struct bridge_iflist *bif;
1700 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1704 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1708 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1710 struct ifbreq *req = arg;
1711 struct bridge_iflist *bif;
1713 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1717 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1721 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1723 struct ifbreq *req = arg;
1724 struct bridge_iflist *bif;
1726 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1730 bif->bif_addrmax = req->ifbr_addrmax;
1735 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1737 struct ifbreq *req = arg;
1738 struct bridge_iflist *bif = NULL;
1741 ifs = ifunit(req->ifbr_ifsname);
1745 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1746 if (ifs == bif->bif_ifp)
1749 if (ifs->if_bridge != NULL)
1752 switch (ifs->if_type) {
1761 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1766 bif->bif_flags = IFBIF_SPAN;
1768 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1774 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1776 struct ifbreq *req = arg;
1777 struct bridge_iflist *bif;
1780 ifs = ifunit(req->ifbr_ifsname);
1784 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1785 if (ifs == bif->bif_ifp)
1791 bridge_delete_span(sc, bif);
1797 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1799 struct ifbropreq *req = arg;
1800 struct bstp_state *bs = &sc->sc_stp;
1801 struct bstp_port *root_port;
1803 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1804 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1805 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1807 root_port = bs->bs_root_port;
1808 if (root_port == NULL)
1809 req->ifbop_root_port = 0;
1811 req->ifbop_root_port = root_port->bp_ifp->if_index;
1813 req->ifbop_holdcount = bs->bs_txholdcount;
1814 req->ifbop_priority = bs->bs_bridge_priority;
1815 req->ifbop_protocol = bs->bs_protover;
1816 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1817 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1818 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1819 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1820 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1821 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1827 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1829 struct ifbrparam *param = arg;
1831 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1836 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1838 struct ifbpstpconf *bifstp = arg;
1839 struct bridge_iflist *bif;
1840 struct bstp_port *bp;
1841 struct ifbpstpreq bpreq;
1843 int count, buflen, len, error = 0;
1846 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1847 if ((bif->bif_flags & IFBIF_STP) != 0)
1851 buflen = sizeof(bpreq) * count;
1852 if (bifstp->ifbpstp_len == 0) {
1853 bifstp->ifbpstp_len = buflen;
1857 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1863 len = min(bifstp->ifbpstp_len, buflen);
1864 bzero(&bpreq, sizeof(bpreq));
1865 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1866 if (len < sizeof(bpreq))
1869 if ((bif->bif_flags & IFBIF_STP) == 0)
1873 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1874 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1875 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1876 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1877 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1878 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1880 memcpy(buf, &bpreq, sizeof(bpreq));
1882 buf += sizeof(bpreq);
1883 len -= sizeof(bpreq);
1886 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1887 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1888 free(outbuf, M_TEMP);
1893 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1895 struct ifbrparam *param = arg;
1897 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1901 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1903 struct ifbrparam *param = arg;
1905 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1911 * Detach an interface from a bridge. Called when a member
1912 * interface is detaching.
1915 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1917 struct bridge_softc *sc = ifp->if_bridge;
1918 struct bridge_iflist *bif;
1920 if (ifp->if_flags & IFF_RENAMING)
1922 if (V_bridge_cloner == NULL) {
1924 * This detach handler can be called after
1925 * vnet_bridge_uninit(). Just return in that case.
1929 /* Check if the interface is a bridge member */
1933 bif = bridge_lookup_member_if(sc, ifp);
1935 bridge_delete_member(sc, bif, 1);
1941 /* Check if the interface is a span port */
1943 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1945 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1946 if (ifp == bif->bif_ifp) {
1947 bridge_delete_span(sc, bif);
1953 BRIDGE_LIST_UNLOCK();
1959 * Initialize a bridge interface.
1962 bridge_init(void *xsc)
1964 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1965 struct ifnet *ifp = sc->sc_ifp;
1967 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1971 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1974 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1975 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1983 * Stop the bridge interface.
1986 bridge_stop(struct ifnet *ifp, int disable)
1988 struct bridge_softc *sc = ifp->if_softc;
1990 BRIDGE_LOCK_ASSERT(sc);
1992 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1996 callout_stop(&sc->sc_brcallout);
1998 bstp_stop(&sc->sc_stp);
2000 bridge_rtflush(sc, IFBF_FLUSHDYN);
2001 BRIDGE_RT_UNLOCK(sc);
2003 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2009 * Enqueue a packet on a bridge member interface.
2013 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2019 /* We may be sending a fragment so traverse the mbuf */
2022 m->m_nextpkt = NULL;
2023 len = m->m_pkthdr.len;
2024 mflags = m->m_flags;
2027 * If underlying interface can not do VLAN tag insertion itself
2028 * then attach a packet tag that holds it.
2030 if ((m->m_flags & M_VLANTAG) &&
2031 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2032 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2035 "unable to prepend VLAN header\n");
2036 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2039 m->m_flags &= ~M_VLANTAG;
2042 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2043 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2046 for (m = m0, n = 1; m != NULL; m = m0, n++) {
2050 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, n);
2054 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2055 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2056 if (mflags & M_MCAST)
2057 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2066 * Receive a queued packet from dummynet and pass it on to the output
2069 * The mbuf has the Ethernet header already attached.
2072 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2074 struct bridge_softc *sc;
2076 sc = ifp->if_bridge;
2079 * The packet didnt originate from a member interface. This should only
2080 * ever happen if a member interface is removed while packets are
2088 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2090 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2093 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2099 bridge_enqueue(sc, ifp, m);
2105 * Send output from a bridge member interface. This
2106 * performs the bridging function for locally originated
2109 * The mbuf has the Ethernet header already attached. We must
2110 * enqueue or free the mbuf before returning.
2113 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2116 struct ether_header *eh;
2117 struct ifnet *bifp, *dst_if;
2118 struct bridge_softc *sc;
2123 if (m->m_len < ETHER_HDR_LEN) {
2124 m = m_pullup(m, ETHER_HDR_LEN);
2129 eh = mtod(m, struct ether_header *);
2130 sc = ifp->if_bridge;
2131 vlan = VLANTAGOF(m);
2136 * If bridge is down, but the original output interface is up,
2137 * go ahead and send out that interface. Otherwise, the packet
2140 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2146 * If the packet is a multicast, or we don't know a better way to
2147 * get there, send to all interfaces.
2149 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2152 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2153 /* Tap any traffic not passing back out the originating interface */
2155 ETHER_BPF_MTAP(bifp, m);
2156 if (dst_if == NULL) {
2157 struct bridge_iflist *bif;
2163 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2164 dst_if = bif->bif_ifp;
2166 if (dst_if->if_type == IFT_GIF)
2168 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2172 * If this is not the original output interface,
2173 * and the interface is participating in spanning
2174 * tree, make sure the port is in a state that
2175 * allows forwarding.
2177 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2178 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2181 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2185 mc = m_dup(m, M_NOWAIT);
2187 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2192 bridge_enqueue(sc, dst_if, mc);
2201 * XXX Spanning tree consideration here?
2205 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2210 bridge_enqueue(sc, dst_if, m);
2217 * Do output on a bridge.
2221 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2223 struct bridge_softc *sc;
2224 struct ether_header *eh;
2225 struct ifnet *dst_if;
2230 ETHER_BPF_MTAP(ifp, m);
2232 eh = mtod(m, struct ether_header *);
2234 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2235 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2236 error = bridge_enqueue(sc, dst_if, m);
2238 bridge_broadcast(sc, ifp, m, 0);
2245 bridge_altq_start(if_t ifp)
2247 struct ifaltq *ifq = &ifp->if_snd;
2251 IFQ_DEQUEUE_NOLOCK(ifq, m);
2253 bridge_transmit(ifp, m);
2254 IFQ_DEQUEUE_NOLOCK(ifq, m);
2260 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2264 if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2265 IFQ_ENQUEUE(&ifp->if_snd, m, err);
2267 bridge_altq_start(ifp);
2269 err = bridge_transmit(ifp, m);
2276 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2279 bridge_qflush(struct ifnet *ifp __unused)
2286 * The forwarding function of the bridge.
2288 * NOTE: Releases the lock on return.
2291 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2294 struct bridge_iflist *dbif;
2295 struct ifnet *src_if, *dst_if, *ifp;
2296 struct ether_header *eh;
2303 src_if = m->m_pkthdr.rcvif;
2306 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2307 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2308 vlan = VLANTAGOF(m);
2310 if ((sbif->bif_flags & IFBIF_STP) &&
2311 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2314 eh = mtod(m, struct ether_header *);
2315 dst = eh->ether_dhost;
2317 /* If the interface is learning, record the address. */
2318 if (sbif->bif_flags & IFBIF_LEARNING) {
2319 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2320 sbif, 0, IFBAF_DYNAMIC);
2322 * If the interface has addresses limits then deny any source
2323 * that is not in the cache.
2325 if (error && sbif->bif_addrmax)
2329 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2330 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2334 * At this point, the port either doesn't participate
2335 * in spanning tree or it is in the forwarding state.
2339 * If the packet is unicast, destined for someone on
2340 * "this" side of the bridge, drop it.
2342 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2343 dst_if = bridge_rtlookup(sc, dst, vlan);
2344 if (src_if == dst_if)
2348 * Check if its a reserved multicast address, any address
2349 * listed in 802.1D section 7.12.6 may not be forwarded by the
2351 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2353 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2354 dst[2] == 0xc2 && dst[3] == 0x00 &&
2355 dst[4] == 0x00 && dst[5] <= 0x0f)
2358 /* ...forward it to all interfaces. */
2359 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2364 * If we have a destination interface which is a member of our bridge,
2365 * OR this is a unicast packet, push it through the bpf(4) machinery.
2366 * For broadcast or multicast packets, don't bother because it will
2367 * be reinjected into ether_input. We do this before we pass the packets
2368 * through the pfil(9) framework, as it is possible that pfil(9) will
2369 * drop the packet, or possibly modify it, making it difficult to debug
2370 * firewall issues on the bridge.
2372 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2373 ETHER_BPF_MTAP(ifp, m);
2375 /* run the packet filter */
2376 if (PFIL_HOOKED_IN(V_inet_pfil_head)
2378 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2381 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2387 if (dst_if == NULL) {
2388 bridge_broadcast(sc, src_if, m, 1);
2393 * At this point, we're dealing with a unicast frame
2394 * going to a different interface.
2396 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2399 dbif = bridge_lookup_member_if(sc, dst_if);
2401 /* Not a member of the bridge (anymore?) */
2404 /* Private segments can not talk to each other */
2405 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2408 if ((dbif->bif_flags & IFBIF_STP) &&
2409 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2412 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2414 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2417 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2423 bridge_enqueue(sc, dst_if, m);
2433 * Receive input from a member interface. Queue the packet for
2434 * bridging if it is not for us.
2436 static struct mbuf *
2437 bridge_input(struct ifnet *ifp, struct mbuf *m)
2439 struct bridge_softc *sc = ifp->if_bridge;
2440 struct bridge_iflist *bif, *bif2;
2442 struct ether_header *eh;
2443 struct mbuf *mc, *mc2;
2449 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2453 vlan = VLANTAGOF(m);
2456 * Implement support for bridge monitoring. If this flag has been
2457 * set on this interface, discard the packet once we push it through
2458 * the bpf(4) machinery, but before we do, increment the byte and
2459 * packet counters associated with this interface.
2461 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2462 m->m_pkthdr.rcvif = bifp;
2463 ETHER_BPF_MTAP(bifp, m);
2464 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2465 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2469 bif = bridge_lookup_member_if(sc, ifp);
2474 eh = mtod(m, struct ether_header *);
2478 if (m->m_flags & (M_BCAST|M_MCAST)) {
2479 /* Tap off 802.1D packets; they do not get forwarded. */
2480 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2481 ETHER_ADDR_LEN) == 0) {
2482 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2486 if ((bif->bif_flags & IFBIF_STP) &&
2487 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2492 * Make a deep copy of the packet and enqueue the copy
2493 * for bridge processing; return the original packet for
2496 mc = m_dup(m, M_NOWAIT);
2501 /* Perform the bridge forwarding function with the copy. */
2502 bridge_forward(sc, bif, mc);
2505 * Reinject the mbuf as arriving on the bridge so we have a
2506 * chance at claiming multicast packets. We can not loop back
2507 * here from ether_input as a bridge is never a member of a
2510 KASSERT(bifp->if_bridge == NULL,
2511 ("loop created in bridge_input"));
2512 mc2 = m_dup(m, M_NOWAIT);
2514 /* Keep the layer3 header aligned */
2515 int i = min(mc2->m_pkthdr.len, max_protohdr);
2516 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2519 mc2->m_pkthdr.rcvif = bifp;
2520 (*bifp->if_input)(bifp, mc2);
2523 /* Return the original packet for local processing. */
2527 if ((bif->bif_flags & IFBIF_STP) &&
2528 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2532 #if defined(INET) || defined(INET6)
2533 #define CARP_CHECK_WE_ARE_DST(iface) \
2534 ((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_dhost))
2535 #define CARP_CHECK_WE_ARE_SRC(iface) \
2536 ((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_shost))
2538 #define CARP_CHECK_WE_ARE_DST(iface) false
2539 #define CARP_CHECK_WE_ARE_SRC(iface) false
2543 #define PFIL_HOOKED_INET6 PFIL_HOOKED_IN(V_inet6_pfil_head)
2545 #define PFIL_HOOKED_INET6 false
2548 #define GRAB_OUR_PACKETS(iface) \
2549 if ((iface)->if_type == IFT_GIF) \
2551 /* It is destined for us. */ \
2552 if (memcmp(IF_LLADDR(iface), eh->ether_dhost, ETHER_ADDR_LEN) == 0 || \
2553 CARP_CHECK_WE_ARE_DST(iface)) { \
2554 if (bif->bif_flags & IFBIF_LEARNING) { \
2555 error = bridge_rtupdate(sc, eh->ether_shost, \
2556 vlan, bif, 0, IFBAF_DYNAMIC); \
2557 if (error && bif->bif_addrmax) { \
2562 m->m_pkthdr.rcvif = iface; \
2563 if ((iface) == ifp) { \
2564 /* Skip bridge processing... src == dest */ \
2567 /* It's passing over or to the bridge, locally. */ \
2568 ETHER_BPF_MTAP(bifp, m); \
2569 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \
2570 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2571 /* Filter on the physical interface. */ \
2572 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) || \
2573 PFIL_HOOKED_INET6)) { \
2574 if (bridge_pfil(&m, NULL, ifp, \
2575 PFIL_IN) != 0 || m == NULL) { \
2579 if ((iface) != bifp) \
2580 ETHER_BPF_MTAP(iface, m); \
2584 /* We just received a packet that we sent out. */ \
2585 if (memcmp(IF_LLADDR(iface), eh->ether_shost, ETHER_ADDR_LEN) == 0 || \
2586 CARP_CHECK_WE_ARE_SRC(iface)) { \
2592 * Unicast. Make sure it's not for the bridge.
2594 do { GRAB_OUR_PACKETS(bifp) } while (0);
2597 * Give a chance for ifp at first priority. This will help when the
2598 * packet comes through the interface like VLAN's with the same MACs
2599 * on several interfaces from the same bridge. This also will save
2600 * some CPU cycles in case the destination interface and the input
2601 * interface (eq ifp) are the same.
2603 do { GRAB_OUR_PACKETS(ifp) } while (0);
2605 /* Now check the all bridge members. */
2606 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2607 GRAB_OUR_PACKETS(bif2->bif_ifp)
2610 #undef CARP_CHECK_WE_ARE_DST
2611 #undef CARP_CHECK_WE_ARE_SRC
2612 #undef PFIL_HOOKED_INET6
2613 #undef GRAB_OUR_PACKETS
2615 /* Perform the bridge forwarding function. */
2616 bridge_forward(sc, bif, m);
2624 * Send a frame to all interfaces that are members of
2625 * the bridge, except for the one on which the packet
2628 * NOTE: Releases the lock on return.
2631 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2632 struct mbuf *m, int runfilt)
2634 struct bridge_iflist *dbif, *sbif;
2636 struct ifnet *dst_if;
2641 sbif = bridge_lookup_member_if(sc, src_if);
2643 /* Filter on the bridge interface before broadcasting */
2644 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2646 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2649 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2655 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2656 dst_if = dbif->bif_ifp;
2657 if (dst_if == src_if)
2660 /* Private segments can not talk to each other */
2661 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2664 if ((dbif->bif_flags & IFBIF_STP) &&
2665 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2668 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2669 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2672 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2675 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2679 mc = m_dup(m, M_NOWAIT);
2681 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2687 * Filter on the output interface. Pass a NULL bridge interface
2688 * pointer so we do not redundantly filter on the bridge for
2689 * each interface we broadcast on.
2691 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2693 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2697 /* Keep the layer3 header aligned */
2698 i = min(mc->m_pkthdr.len, max_protohdr);
2699 mc = m_copyup(mc, i, ETHER_ALIGN);
2701 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2705 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2711 bridge_enqueue(sc, dst_if, mc);
2720 * Duplicate a packet out one or more interfaces that are in span mode,
2721 * the original mbuf is unmodified.
2724 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2726 struct bridge_iflist *bif;
2727 struct ifnet *dst_if;
2732 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2735 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2736 dst_if = bif->bif_ifp;
2738 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2741 mc = m_dup(m, M_NOWAIT);
2743 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2747 bridge_enqueue(sc, dst_if, mc);
2754 * Add a bridge routing entry.
2757 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2758 struct bridge_iflist *bif, int setflags, uint8_t flags)
2760 struct bridge_rtnode *brt;
2761 struct bridge_iflist *obif;
2764 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2766 /* Check the source address is valid and not multicast. */
2767 if (ETHER_IS_MULTICAST(dst) ||
2768 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2769 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2772 /* 802.1p frames map to vlan 1 */
2777 * A route for this destination might already exist. If so,
2778 * update it, otherwise create a new one.
2780 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2783 /* Check again, now that we have the lock. There could have
2784 * been a race and we only want to insert this once. */
2785 if (bridge_rtnode_lookup(sc, dst, vlan) != NULL) {
2786 BRIDGE_RT_UNLOCK(sc);
2790 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2791 sc->sc_brtexceeded++;
2792 BRIDGE_RT_UNLOCK(sc);
2795 /* Check per interface address limits (if enabled) */
2796 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2797 bif->bif_addrexceeded++;
2798 BRIDGE_RT_UNLOCK(sc);
2803 * Allocate a new bridge forwarding node, and
2804 * initialize the expiration time and Ethernet
2807 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2809 BRIDGE_RT_UNLOCK(sc);
2812 brt->brt_vnet = curvnet;
2814 if (bif->bif_flags & IFBIF_STICKY)
2815 brt->brt_flags = IFBAF_STICKY;
2817 brt->brt_flags = IFBAF_DYNAMIC;
2819 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2820 brt->brt_vlan = vlan;
2823 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2824 uma_zfree(V_bridge_rtnode_zone, brt);
2825 BRIDGE_RT_UNLOCK(sc);
2830 BRIDGE_RT_UNLOCK(sc);
2833 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2834 (obif = brt->brt_dst) != bif) {
2835 MPASS(obif != NULL);
2838 brt->brt_dst->bif_addrcnt--;
2840 brt->brt_dst->bif_addrcnt++;
2841 BRIDGE_RT_UNLOCK(sc);
2843 if (V_log_mac_flap &&
2844 ppsratecheck(&V_log_last, &V_log_count, V_log_interval)) {
2846 "%s: mac address %6D vlan %d moved from %s to %s\n",
2847 sc->sc_ifp->if_xname,
2848 &brt->brt_addr[0], ":",
2850 obif->bif_ifp->if_xname,
2851 bif->bif_ifp->if_xname);
2855 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2856 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2858 brt->brt_flags = flags;
2866 * Lookup the destination interface for an address.
2868 static struct ifnet *
2869 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2871 struct bridge_rtnode *brt;
2875 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2878 return (brt->brt_ifp);
2884 * Trim the routine table so that we have a number
2885 * of routing entries less than or equal to the
2889 bridge_rttrim(struct bridge_softc *sc)
2891 struct bridge_rtnode *brt, *nbrt;
2894 BRIDGE_RT_LOCK_ASSERT(sc);
2896 /* Make sure we actually need to do this. */
2897 if (sc->sc_brtcnt <= sc->sc_brtmax)
2900 /* Force an aging cycle; this might trim enough addresses. */
2902 if (sc->sc_brtcnt <= sc->sc_brtmax)
2905 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2906 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2907 bridge_rtnode_destroy(sc, brt);
2908 if (sc->sc_brtcnt <= sc->sc_brtmax)
2917 * Aging timer for the bridge.
2920 bridge_timer(void *arg)
2922 struct bridge_softc *sc = arg;
2924 BRIDGE_RT_LOCK_ASSERT(sc);
2926 /* Destruction of rtnodes requires a proper vnet context */
2927 CURVNET_SET(sc->sc_ifp->if_vnet);
2930 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2931 callout_reset(&sc->sc_brcallout,
2932 bridge_rtable_prune_period * hz, bridge_timer, sc);
2939 * Perform an aging cycle.
2942 bridge_rtage(struct bridge_softc *sc)
2944 struct bridge_rtnode *brt, *nbrt;
2946 BRIDGE_RT_LOCK_ASSERT(sc);
2948 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2949 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2950 if (time_uptime >= brt->brt_expire)
2951 bridge_rtnode_destroy(sc, brt);
2959 * Remove all dynamic addresses from the bridge.
2962 bridge_rtflush(struct bridge_softc *sc, int full)
2964 struct bridge_rtnode *brt, *nbrt;
2966 BRIDGE_RT_LOCK_ASSERT(sc);
2968 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2969 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2970 bridge_rtnode_destroy(sc, brt);
2977 * Remove an address from the table.
2980 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2982 struct bridge_rtnode *brt;
2988 * If vlan is zero then we want to delete for all vlans so the lookup
2989 * may return more than one.
2991 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2992 bridge_rtnode_destroy(sc, brt);
2996 BRIDGE_RT_UNLOCK(sc);
2998 return (found ? 0 : ENOENT);
3004 * Delete routes to a speicifc member interface.
3007 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3009 struct bridge_rtnode *brt, *nbrt;
3011 BRIDGE_RT_LOCK_ASSERT(sc);
3013 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3014 if (brt->brt_ifp == ifp && (full ||
3015 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3016 bridge_rtnode_destroy(sc, brt);
3021 * bridge_rtable_init:
3023 * Initialize the route table for this bridge.
3026 bridge_rtable_init(struct bridge_softc *sc)
3030 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3031 M_DEVBUF, M_WAITOK);
3033 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3034 CK_LIST_INIT(&sc->sc_rthash[i]);
3036 sc->sc_rthash_key = arc4random();
3037 CK_LIST_INIT(&sc->sc_rtlist);
3041 * bridge_rtable_fini:
3043 * Deconstruct the route table for this bridge.
3046 bridge_rtable_fini(struct bridge_softc *sc)
3049 KASSERT(sc->sc_brtcnt == 0,
3050 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3051 free(sc->sc_rthash, M_DEVBUF);
3055 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3056 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3058 #define mix(a, b, c) \
3060 a -= b; a -= c; a ^= (c >> 13); \
3061 b -= c; b -= a; b ^= (a << 8); \
3062 c -= a; c -= b; c ^= (b >> 13); \
3063 a -= b; a -= c; a ^= (c >> 12); \
3064 b -= c; b -= a; b ^= (a << 16); \
3065 c -= a; c -= b; c ^= (b >> 5); \
3066 a -= b; a -= c; a ^= (c >> 3); \
3067 b -= c; b -= a; b ^= (a << 10); \
3068 c -= a; c -= b; c ^= (b >> 15); \
3069 } while (/*CONSTCOND*/0)
3071 static __inline uint32_t
3072 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3074 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3085 return (c & BRIDGE_RTHASH_MASK);
3091 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3095 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3096 d = ((int)a[i]) - ((int)b[i]);
3103 * bridge_rtnode_lookup:
3105 * Look up a bridge route node for the specified destination. Compare the
3106 * vlan id or if zero then just return the first match.
3108 static struct bridge_rtnode *
3109 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3111 struct bridge_rtnode *brt;
3115 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3117 hash = bridge_rthash(sc, addr);
3118 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3119 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3120 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3130 * bridge_rtnode_insert:
3132 * Insert the specified bridge node into the route table. We
3133 * assume the entry is not already in the table.
3136 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3138 struct bridge_rtnode *lbrt;
3142 BRIDGE_RT_LOCK_ASSERT(sc);
3144 hash = bridge_rthash(sc, brt->brt_addr);
3146 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3148 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3153 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3154 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3157 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3160 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3161 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3164 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3165 } while (lbrt != NULL);
3168 panic("bridge_rtnode_insert: impossible");
3172 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3179 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3181 struct bridge_rtnode *brt;
3183 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3185 CURVNET_SET(brt->brt_vnet);
3186 uma_zfree(V_bridge_rtnode_zone, brt);
3191 * bridge_rtnode_destroy:
3193 * Destroy a bridge rtnode.
3196 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3198 BRIDGE_RT_LOCK_ASSERT(sc);
3200 CK_LIST_REMOVE(brt, brt_hash);
3202 CK_LIST_REMOVE(brt, brt_list);
3204 brt->brt_dst->bif_addrcnt--;
3206 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3210 * bridge_rtable_expire:
3212 * Set the expiry time for all routes on an interface.
3215 bridge_rtable_expire(struct ifnet *ifp, int age)
3217 struct bridge_softc *sc = ifp->if_bridge;
3218 struct bridge_rtnode *brt;
3220 CURVNET_SET(ifp->if_vnet);
3224 * If the age is zero then flush, otherwise set all the expiry times to
3225 * age for the interface
3228 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3230 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3231 /* Cap the expiry time to 'age' */
3232 if (brt->brt_ifp == ifp &&
3233 brt->brt_expire > time_uptime + age &&
3234 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3235 brt->brt_expire = time_uptime + age;
3238 BRIDGE_RT_UNLOCK(sc);
3243 * bridge_state_change:
3245 * Callback from the bridgestp code when a port changes states.
3248 bridge_state_change(struct ifnet *ifp, int state)
3250 struct bridge_softc *sc = ifp->if_bridge;
3251 static const char *stpstates[] = {
3260 CURVNET_SET(ifp->if_vnet);
3262 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3263 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3268 * Send bridge packets through pfil if they are one of the types pfil can deal
3269 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3270 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3274 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3276 int snap, error, i, hlen;
3277 struct ether_header *eh1, eh2;
3280 u_int16_t ether_type;
3284 error = -1; /* Default error if not error == 0 */
3287 /* we may return with the IP fields swapped, ensure its not shared */
3288 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3291 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3292 return (0); /* filtering is disabled */
3294 i = min((*mp)->m_pkthdr.len, max_protohdr);
3295 if ((*mp)->m_len < i) {
3296 *mp = m_pullup(*mp, i);
3298 printf("%s: m_pullup failed\n", __func__);
3303 eh1 = mtod(*mp, struct ether_header *);
3304 ether_type = ntohs(eh1->ether_type);
3307 * Check for SNAP/LLC.
3309 if (ether_type < ETHERMTU) {
3310 struct llc *llc2 = (struct llc *)(eh1 + 1);
3312 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3313 llc2->llc_dsap == LLC_SNAP_LSAP &&
3314 llc2->llc_ssap == LLC_SNAP_LSAP &&
3315 llc2->llc_control == LLC_UI) {
3316 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3322 * If we're trying to filter bridge traffic, don't look at anything
3323 * other than IP and ARP traffic. If the filter doesn't understand
3324 * IPv6, don't allow IPv6 through the bridge either. This is lame
3325 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3326 * but of course we don't have an AppleTalk filter to begin with.
3327 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3330 switch (ether_type) {
3332 case ETHERTYPE_REVARP:
3333 if (V_pfil_ipfw_arp == 0)
3334 return (0); /* Automatically pass */
3339 case ETHERTYPE_IPV6:
3344 * Check to see if the user wants to pass non-ip
3345 * packets, these will not be checked by pfil(9) and
3346 * passed unconditionally so the default is to drop.
3352 /* Run the packet through pfil before stripping link headers */
3353 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3354 dir == PFIL_OUT && ifp != NULL) {
3355 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3363 /* Strip off the Ethernet header and keep a copy. */
3364 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3365 m_adj(*mp, ETHER_HDR_LEN);
3367 /* Strip off snap header, if present */
3369 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3370 m_adj(*mp, sizeof(struct llc));
3374 * Check the IP header for alignment and errors
3376 if (dir == PFIL_IN) {
3377 switch (ether_type) {
3379 error = bridge_ip_checkbasic(mp);
3382 case ETHERTYPE_IPV6:
3383 error = bridge_ip6_checkbasic(mp);
3396 * Run the packet through pfil
3399 switch (ether_type) {
3402 * Run pfil on the member interface and the bridge, both can
3403 * be skipped by clearing pfil_member or pfil_bridge.
3406 * in_if -> bridge_if -> out_if
3408 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3409 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3413 if (V_pfil_member && ifp != NULL && (rv =
3414 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3418 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3419 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3423 /* check if we need to fragment the packet */
3424 /* bridge_fragment generates a mbuf chain of packets */
3425 /* that already include eth headers */
3426 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3427 i = (*mp)->m_pkthdr.len;
3428 if (i > ifp->if_mtu) {
3429 error = bridge_fragment(ifp, mp, &eh2, snap,
3435 /* Recalculate the ip checksum. */
3436 ip = mtod(*mp, struct ip *);
3437 hlen = ip->ip_hl << 2;
3438 if (hlen < sizeof(struct ip))
3440 if (hlen > (*mp)->m_len) {
3441 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3443 ip = mtod(*mp, struct ip *);
3448 if (hlen == sizeof(struct ip))
3449 ip->ip_sum = in_cksum_hdr(ip);
3451 ip->ip_sum = in_cksum(*mp, hlen);
3455 case ETHERTYPE_IPV6:
3456 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3457 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3461 if (V_pfil_member && ifp != NULL && (rv =
3462 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3466 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3467 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3486 * Finally, put everything back the way it was and return
3489 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3492 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3495 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3498 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3509 * Perform basic checks on header size since
3510 * pfil assumes ip_input has already processed
3511 * it for it. Cut-and-pasted from ip_input.c.
3512 * Given how simple the IPv6 version is,
3513 * does the IPv4 version really need to be
3516 * XXX Should we update ipstat here, or not?
3517 * XXX Right now we update ipstat but not
3521 bridge_ip_checkbasic(struct mbuf **mp)
3523 struct mbuf *m = *mp;
3531 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3532 if ((m = m_copyup(m, sizeof(struct ip),
3533 (max_linkhdr + 3) & ~3)) == NULL) {
3534 /* XXXJRT new stat, please */
3535 KMOD_IPSTAT_INC(ips_toosmall);
3538 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3539 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3540 KMOD_IPSTAT_INC(ips_toosmall);
3544 ip = mtod(m, struct ip *);
3545 if (ip == NULL) goto bad;
3547 if (ip->ip_v != IPVERSION) {
3548 KMOD_IPSTAT_INC(ips_badvers);
3551 hlen = ip->ip_hl << 2;
3552 if (hlen < sizeof(struct ip)) { /* minimum header length */
3553 KMOD_IPSTAT_INC(ips_badhlen);
3556 if (hlen > m->m_len) {
3557 if ((m = m_pullup(m, hlen)) == NULL) {
3558 KMOD_IPSTAT_INC(ips_badhlen);
3561 ip = mtod(m, struct ip *);
3562 if (ip == NULL) goto bad;
3565 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3566 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3568 if (hlen == sizeof(struct ip)) {
3569 sum = in_cksum_hdr(ip);
3571 sum = in_cksum(m, hlen);
3575 KMOD_IPSTAT_INC(ips_badsum);
3579 /* Retrieve the packet length. */
3580 len = ntohs(ip->ip_len);
3583 * Check for additional length bogosity
3586 KMOD_IPSTAT_INC(ips_badlen);
3591 * Check that the amount of data in the buffers
3592 * is as at least much as the IP header would have us expect.
3593 * Drop packet if shorter than we expect.
3595 if (m->m_pkthdr.len < len) {
3596 KMOD_IPSTAT_INC(ips_tooshort);
3600 /* Checks out, proceed */
3611 * Same as above, but for IPv6.
3612 * Cut-and-pasted from ip6_input.c.
3613 * XXX Should we update ip6stat, or not?
3616 bridge_ip6_checkbasic(struct mbuf **mp)
3618 struct mbuf *m = *mp;
3619 struct ip6_hdr *ip6;
3622 * If the IPv6 header is not aligned, slurp it up into a new
3623 * mbuf with space for link headers, in the event we forward
3624 * it. Otherwise, if it is aligned, make sure the entire base
3625 * IPv6 header is in the first mbuf of the chain.
3627 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3628 struct ifnet *inifp = m->m_pkthdr.rcvif;
3629 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3630 (max_linkhdr + 3) & ~3)) == NULL) {
3631 /* XXXJRT new stat, please */
3632 IP6STAT_INC(ip6s_toosmall);
3633 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3636 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3637 struct ifnet *inifp = m->m_pkthdr.rcvif;
3638 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3639 IP6STAT_INC(ip6s_toosmall);
3640 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3645 ip6 = mtod(m, struct ip6_hdr *);
3647 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3648 IP6STAT_INC(ip6s_badvers);
3649 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3653 /* Checks out, proceed */
3666 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3669 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3670 int snap, struct llc *llc)
3672 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3676 if (m->m_len < sizeof(struct ip) &&
3677 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3679 ip = mtod(m, struct ip *);
3681 m->m_pkthdr.csum_flags |= CSUM_IP;
3682 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3687 * Walk the chain and re-add the Ethernet header for
3690 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3691 nextpkt = mcur->m_nextpkt;
3692 mcur->m_nextpkt = NULL;
3694 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3698 mprev->m_nextpkt = nextpkt;
3701 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3704 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3708 mprev->m_nextpkt = nextpkt;
3711 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3714 * The previous two M_PREPEND could have inserted one or two
3715 * mbufs in front so we have to update the previous packet's
3718 mcur->m_nextpkt = nextpkt;
3720 mprev->m_nextpkt = mcur;
3722 /* The first mbuf in the original chain needs to be
3729 KMOD_IPSTAT_INC(ips_fragmented);
3733 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3734 m = mcur->m_nextpkt;
3741 bridge_linkstate(struct ifnet *ifp)
3743 struct bridge_softc *sc = ifp->if_bridge;
3744 struct bridge_iflist *bif;
3745 struct epoch_tracker et;
3747 NET_EPOCH_ENTER(et);
3749 bif = bridge_lookup_member_if(sc, ifp);
3754 bridge_linkcheck(sc);
3756 bstp_linkstate(&bif->bif_stp);
3762 bridge_linkcheck(struct bridge_softc *sc)
3764 struct bridge_iflist *bif;
3765 int new_link, hasls;
3767 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3769 new_link = LINK_STATE_DOWN;
3771 /* Our link is considered up if at least one of our ports is active */
3772 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3773 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3775 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3776 new_link = LINK_STATE_UP;
3780 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3781 /* If no interfaces support link-state then we default to up */
3782 new_link = LINK_STATE_UP;
3784 if_link_state_change(sc->sc_ifp, new_link);