1 /* $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $ */
4 * SPDX-License-Identifier: BSD-4-Clause
6 * Copyright 2001 Wasabi Systems, Inc.
9 * Written by Jason R. Thorpe for Wasabi Systems, Inc.
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. All advertising materials mentioning features or use of this software
20 * must display the following acknowledgement:
21 * This product includes software developed for the NetBSD Project by
22 * Wasabi Systems, Inc.
23 * 4. The name of Wasabi Systems, Inc. may not be used to endorse
24 * or promote products derived from this software without specific prior
27 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37 * POSSIBILITY OF SUCH DAMAGE.
41 * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
42 * All rights reserved.
44 * Redistribution and use in source and binary forms, with or without
45 * modification, are permitted provided that the following conditions
47 * 1. Redistributions of source code must retain the above copyright
48 * notice, this list of conditions and the following disclaimer.
49 * 2. Redistributions in binary form must reproduce the above copyright
50 * notice, this list of conditions and the following disclaimer in the
51 * documentation and/or other materials provided with the distribution.
53 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
54 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
55 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
56 * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
57 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
58 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
59 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
61 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
62 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
63 * POSSIBILITY OF SUCH DAMAGE.
65 * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
69 * Network interface bridge support.
73 * - Currently only supports Ethernet-like interfaces (Ethernet,
74 * 802.11, VLANs on Ethernet, etc.) Figure out a nice way
75 * to bridge other types of interfaces (maybe consider
76 * heterogeneous bridges).
79 #include <sys/cdefs.h>
80 __FBSDID("$FreeBSD$");
83 #include "opt_inet6.h"
85 #include <sys/param.h>
86 #include <sys/eventhandler.h>
88 #include <sys/malloc.h>
89 #include <sys/protosw.h>
90 #include <sys/systm.h>
93 #include <sys/socket.h> /* for net/if.h */
94 #include <sys/sockio.h>
95 #include <sys/ctype.h> /* string functions */
96 #include <sys/kernel.h>
97 #include <sys/random.h>
98 #include <sys/syslog.h>
99 #include <sys/sysctl.h>
101 #include <sys/module.h>
102 #include <sys/priv.h>
103 #include <sys/proc.h>
104 #include <sys/lock.h>
105 #include <sys/mutex.h>
109 #include <net/if_clone.h>
110 #include <net/if_dl.h>
111 #include <net/if_types.h>
112 #include <net/if_var.h>
113 #include <net/pfil.h>
114 #include <net/vnet.h>
116 #include <netinet/in.h>
117 #include <netinet/in_systm.h>
118 #include <netinet/in_var.h>
119 #include <netinet/ip.h>
120 #include <netinet/ip_var.h>
122 #include <netinet/ip6.h>
123 #include <netinet6/ip6_var.h>
124 #include <netinet6/in6_ifattach.h>
126 #if defined(INET) || defined(INET6)
127 #include <netinet/ip_carp.h>
129 #include <machine/in_cksum.h>
130 #include <netinet/if_ether.h>
131 #include <net/bridgestp.h>
132 #include <net/if_bridgevar.h>
133 #include <net/if_llc.h>
134 #include <net/if_vlan_var.h>
136 #include <net/route.h>
140 * XXX: declare here to avoid to include many inet6 related files..
141 * should be more generalized?
143 extern void nd6_setmtu(struct ifnet *);
147 * Size of the route hash table. Must be a power of two.
149 #ifndef BRIDGE_RTHASH_SIZE
150 #define BRIDGE_RTHASH_SIZE 1024
153 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1)
156 * Default maximum number of addresses to cache.
158 #ifndef BRIDGE_RTABLE_MAX
159 #define BRIDGE_RTABLE_MAX 2000
163 * Timeout (in seconds) for entries learned dynamically.
165 #ifndef BRIDGE_RTABLE_TIMEOUT
166 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */
170 * Number of seconds between walks of the route list.
172 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
173 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60)
177 * List of capabilities to possibly mask on the member interface.
179 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
183 * List of capabilities to strip
185 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO
190 * The bridge relies heavily on the epoch(9) system to protect its data
191 * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we
192 * must ensure there is only one writer at a time.
194 * That is: for read accesses we only need to be in NET_EPOCH, but for write
195 * accesses we must hold:
197 * - BRIDGE_RT_LOCK, for any change to bridge_rtnodes
198 * - BRIDGE_LOCK, for any other change
200 * The BRIDGE_LOCK is a sleepable lock, because it is held accross ioctl()
201 * calls to bridge member interfaces and these ioctl()s can sleep.
202 * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes
203 * required while we're in NET_EPOCH and then we're not allowed to sleep.
205 #define BRIDGE_LOCK_INIT(_sc) do { \
206 sx_init(&(_sc)->sc_sx, "if_bridge"); \
207 mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF); \
209 #define BRIDGE_LOCK_DESTROY(_sc) do { \
210 sx_destroy(&(_sc)->sc_sx); \
211 mtx_destroy(&(_sc)->sc_rt_mtx); \
213 #define BRIDGE_LOCK(_sc) sx_xlock(&(_sc)->sc_sx)
214 #define BRIDGE_UNLOCK(_sc) sx_xunlock(&(_sc)->sc_sx)
215 #define BRIDGE_LOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_XLOCKED)
216 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
217 MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx))
218 #define BRIDGE_UNLOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_UNLOCKED)
219 #define BRIDGE_RT_LOCK(_sc) mtx_lock(&(_sc)->sc_rt_mtx)
220 #define BRIDGE_RT_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_rt_mtx)
221 #define BRIDGE_RT_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED)
222 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
223 MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx))
226 * Bridge interface list entry.
228 struct bridge_iflist {
229 CK_LIST_ENTRY(bridge_iflist) bif_next;
230 struct ifnet *bif_ifp; /* member if */
231 struct bstp_port bif_stp; /* STP state */
232 uint32_t bif_flags; /* member if flags */
233 int bif_savedcaps; /* saved capabilities */
234 uint32_t bif_addrmax; /* max # of addresses */
235 uint32_t bif_addrcnt; /* cur. # of addresses */
236 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 */
253 struct vnet *brt_vnet;
254 struct epoch_context brt_epoch_ctx;
256 #define brt_ifp brt_dst->bif_ifp
259 * Software state for each bridge.
261 struct bridge_softc {
262 struct ifnet *sc_ifp; /* make this an interface */
263 LIST_ENTRY(bridge_softc) sc_list;
265 struct mtx sc_rt_mtx;
266 uint32_t sc_brtmax; /* max # of addresses */
267 uint32_t sc_brtcnt; /* cur. # of addresses */
268 uint32_t sc_brttimeout; /* rt timeout in seconds */
269 struct callout sc_brcallout; /* bridge callout */
270 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
271 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
272 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
273 uint32_t sc_rthash_key; /* key for hash */
274 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
275 struct bstp_state sc_stp; /* STP state */
276 uint32_t sc_brtexceeded; /* # of cache drops */
277 struct ifnet *sc_ifaddr; /* member mac copied from */
278 struct ether_addr sc_defaddr; /* Default MAC address */
280 struct epoch_context sc_epoch_ctx;
283 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
284 #define V_bridge_list_sx VNET(bridge_list_sx)
285 static eventhandler_tag bridge_detach_cookie;
287 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
289 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
290 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
292 static int bridge_clone_create(struct if_clone *, int, caddr_t);
293 static void bridge_clone_destroy(struct ifnet *);
295 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
296 static void bridge_mutecaps(struct bridge_softc *);
297 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
299 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
300 static void bridge_init(void *);
301 static void bridge_dummynet(struct mbuf *, struct ifnet *);
302 static void bridge_stop(struct ifnet *, int);
303 static int bridge_transmit(struct ifnet *, struct mbuf *);
304 static void bridge_qflush(struct ifnet *);
305 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
306 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
308 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
310 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
312 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
315 static void bridge_timer(void *);
317 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
319 static void bridge_span(struct bridge_softc *, struct mbuf *);
321 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
322 uint16_t, struct bridge_iflist *, int, uint8_t);
323 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
325 static void bridge_rttrim(struct bridge_softc *);
326 static void bridge_rtage(struct bridge_softc *);
327 static void bridge_rtflush(struct bridge_softc *, int);
328 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
331 static void bridge_rtable_init(struct bridge_softc *);
332 static void bridge_rtable_fini(struct bridge_softc *);
334 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
335 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
336 const uint8_t *, uint16_t);
337 static int bridge_rtnode_insert(struct bridge_softc *,
338 struct bridge_rtnode *);
339 static void bridge_rtnode_destroy(struct bridge_softc *,
340 struct bridge_rtnode *);
341 static void bridge_rtable_expire(struct ifnet *, int);
342 static void bridge_state_change(struct ifnet *, int);
344 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
346 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
348 static void bridge_delete_member(struct bridge_softc *,
349 struct bridge_iflist *, int);
350 static void bridge_delete_span(struct bridge_softc *,
351 struct bridge_iflist *);
353 static int bridge_ioctl_add(struct bridge_softc *, void *);
354 static int bridge_ioctl_del(struct bridge_softc *, void *);
355 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
356 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
357 static int bridge_ioctl_scache(struct bridge_softc *, void *);
358 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
359 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
360 static int bridge_ioctl_rts(struct bridge_softc *, void *);
361 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
362 static int bridge_ioctl_sto(struct bridge_softc *, void *);
363 static int bridge_ioctl_gto(struct bridge_softc *, void *);
364 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
365 static int bridge_ioctl_flush(struct bridge_softc *, void *);
366 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
367 static int bridge_ioctl_spri(struct bridge_softc *, void *);
368 static int bridge_ioctl_ght(struct bridge_softc *, void *);
369 static int bridge_ioctl_sht(struct bridge_softc *, void *);
370 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
371 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
372 static int bridge_ioctl_gma(struct bridge_softc *, void *);
373 static int bridge_ioctl_sma(struct bridge_softc *, void *);
374 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
375 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
376 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
377 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
378 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
379 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
380 static int bridge_ioctl_grte(struct bridge_softc *, void *);
381 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
382 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
383 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
384 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
386 static int bridge_ip_checkbasic(struct mbuf **mp);
388 static int bridge_ip6_checkbasic(struct mbuf **mp);
390 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
391 struct ether_header *, int, struct llc *);
392 static void bridge_linkstate(struct ifnet *ifp);
393 static void bridge_linkcheck(struct bridge_softc *sc);
396 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
397 #define VLANTAGOF(_m) \
398 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
400 static struct bstp_cb_ops bridge_ops = {
401 .bcb_state = bridge_state_change,
402 .bcb_rtage = bridge_rtable_expire
405 SYSCTL_DECL(_net_link);
406 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
408 /* only pass IP[46] packets when pfil is enabled */
409 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
410 #define V_pfil_onlyip VNET(pfil_onlyip)
411 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
412 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
413 "Only pass IP packets when pfil is enabled");
415 /* run pfil hooks on the bridge interface */
416 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
417 #define V_pfil_bridge VNET(pfil_bridge)
418 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
419 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
420 "Packet filter on the bridge interface");
422 /* layer2 filter with ipfw */
423 VNET_DEFINE_STATIC(int, pfil_ipfw);
424 #define V_pfil_ipfw VNET(pfil_ipfw)
426 /* layer2 ARP filter with ipfw */
427 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
428 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
429 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
430 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
431 "Filter ARP packets through IPFW layer2");
433 /* run pfil hooks on the member interface */
434 VNET_DEFINE_STATIC(int, pfil_member) = 1;
435 #define V_pfil_member VNET(pfil_member)
436 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
437 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
438 "Packet filter on the member interface");
440 /* run pfil hooks on the physical interface for locally destined packets */
441 VNET_DEFINE_STATIC(int, pfil_local_phys);
442 #define V_pfil_local_phys VNET(pfil_local_phys)
443 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
444 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
445 "Packet filter on the physical interface for locally destined packets");
447 /* log STP state changes */
448 VNET_DEFINE_STATIC(int, log_stp);
449 #define V_log_stp VNET(log_stp)
450 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
451 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
452 "Log STP state changes");
454 /* share MAC with first bridge member */
455 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
456 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
457 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
458 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
459 "Inherit MAC address from the first bridge member");
461 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
462 #define V_allow_llz_overlap VNET(allow_llz_overlap)
463 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
464 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
465 "Allow overlap of link-local scope "
466 "zones of a bridge interface and the member interfaces");
468 struct bridge_control {
469 int (*bc_func)(struct bridge_softc *, void *);
474 #define BC_F_COPYIN 0x01 /* copy arguments in */
475 #define BC_F_COPYOUT 0x02 /* copy arguments out */
476 #define BC_F_SUSER 0x04 /* do super-user check */
478 const struct bridge_control bridge_control_table[] = {
479 { bridge_ioctl_add, sizeof(struct ifbreq),
480 BC_F_COPYIN|BC_F_SUSER },
481 { bridge_ioctl_del, sizeof(struct ifbreq),
482 BC_F_COPYIN|BC_F_SUSER },
484 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
485 BC_F_COPYIN|BC_F_COPYOUT },
486 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
487 BC_F_COPYIN|BC_F_SUSER },
489 { bridge_ioctl_scache, sizeof(struct ifbrparam),
490 BC_F_COPYIN|BC_F_SUSER },
491 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
494 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
495 BC_F_COPYIN|BC_F_COPYOUT },
496 { bridge_ioctl_rts, sizeof(struct ifbaconf),
497 BC_F_COPYIN|BC_F_COPYOUT },
499 { bridge_ioctl_saddr, sizeof(struct ifbareq),
500 BC_F_COPYIN|BC_F_SUSER },
502 { bridge_ioctl_sto, sizeof(struct ifbrparam),
503 BC_F_COPYIN|BC_F_SUSER },
504 { bridge_ioctl_gto, sizeof(struct ifbrparam),
507 { bridge_ioctl_daddr, sizeof(struct ifbareq),
508 BC_F_COPYIN|BC_F_SUSER },
510 { bridge_ioctl_flush, sizeof(struct ifbreq),
511 BC_F_COPYIN|BC_F_SUSER },
513 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
515 { bridge_ioctl_spri, sizeof(struct ifbrparam),
516 BC_F_COPYIN|BC_F_SUSER },
518 { bridge_ioctl_ght, sizeof(struct ifbrparam),
520 { bridge_ioctl_sht, sizeof(struct ifbrparam),
521 BC_F_COPYIN|BC_F_SUSER },
523 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
525 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
526 BC_F_COPYIN|BC_F_SUSER },
528 { bridge_ioctl_gma, sizeof(struct ifbrparam),
530 { bridge_ioctl_sma, sizeof(struct ifbrparam),
531 BC_F_COPYIN|BC_F_SUSER },
533 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
534 BC_F_COPYIN|BC_F_SUSER },
536 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
537 BC_F_COPYIN|BC_F_SUSER },
539 { bridge_ioctl_addspan, sizeof(struct ifbreq),
540 BC_F_COPYIN|BC_F_SUSER },
541 { bridge_ioctl_delspan, sizeof(struct ifbreq),
542 BC_F_COPYIN|BC_F_SUSER },
544 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
547 { bridge_ioctl_grte, sizeof(struct ifbrparam),
550 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
551 BC_F_COPYIN|BC_F_COPYOUT },
553 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
554 BC_F_COPYIN|BC_F_SUSER },
556 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
557 BC_F_COPYIN|BC_F_SUSER },
559 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
560 BC_F_COPYIN|BC_F_SUSER },
563 const int bridge_control_table_size = nitems(bridge_control_table);
565 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
566 #define V_bridge_list VNET(bridge_list)
567 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \
569 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx)
570 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx)
571 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx)
573 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
574 #define V_bridge_cloner VNET(bridge_cloner)
576 static const char bridge_name[] = "bridge";
579 vnet_bridge_init(const void *unused __unused)
582 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
583 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
585 BRIDGE_LIST_LOCK_INIT();
586 LIST_INIT(&V_bridge_list);
587 V_bridge_cloner = if_clone_simple(bridge_name,
588 bridge_clone_create, bridge_clone_destroy, 0);
590 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
591 vnet_bridge_init, NULL);
594 vnet_bridge_uninit(const void *unused __unused)
597 if_clone_detach(V_bridge_cloner);
598 V_bridge_cloner = NULL;
599 BRIDGE_LIST_LOCK_DESTROY();
601 /* Before we can destroy the uma zone, because there are callbacks that
603 epoch_drain_callbacks(net_epoch_preempt);
605 uma_zdestroy(V_bridge_rtnode_zone);
607 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
608 vnet_bridge_uninit, NULL);
611 bridge_modevent(module_t mod, int type, void *data)
616 bridge_dn_p = bridge_dummynet;
617 bridge_detach_cookie = EVENTHANDLER_REGISTER(
618 ifnet_departure_event, bridge_ifdetach, NULL,
619 EVENTHANDLER_PRI_ANY);
622 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
623 bridge_detach_cookie);
632 static moduledata_t bridge_mod = {
638 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
639 MODULE_VERSION(if_bridge, 1);
640 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
643 * handler for net.link.bridge.ipfw
646 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
648 int enable = V_pfil_ipfw;
651 error = sysctl_handle_int(oidp, &enable, 0, req);
654 if (enable != V_pfil_ipfw) {
655 V_pfil_ipfw = enable;
658 * Disable pfil so that ipfw doesnt run twice, if the user
659 * really wants both then they can re-enable pfil_bridge and/or
660 * pfil_member. Also allow non-ip packets as ipfw can filter by
672 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
673 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
674 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
675 "Layer2 filter with IPFW");
678 * bridge_clone_create:
680 * Create a new bridge instance.
683 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
685 struct bridge_softc *sc, *sc2;
686 struct ifnet *bifp, *ifp;
688 unsigned long hostid;
690 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
691 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
697 BRIDGE_LOCK_INIT(sc);
698 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
699 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
701 /* Initialize our routing table. */
702 bridge_rtable_init(sc);
704 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
706 CK_LIST_INIT(&sc->sc_iflist);
707 CK_LIST_INIT(&sc->sc_spanlist);
710 if_initname(ifp, bridge_name, unit);
711 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
712 ifp->if_ioctl = bridge_ioctl;
713 ifp->if_transmit = bridge_transmit;
714 ifp->if_qflush = bridge_qflush;
715 ifp->if_init = bridge_init;
716 ifp->if_type = IFT_BRIDGE;
719 * Generate an ethernet address with a locally administered address.
721 * Since we are using random ethernet addresses for the bridge, it is
722 * possible that we might have address collisions, so make sure that
723 * this hardware address isn't already in use on another bridge.
724 * The first try uses the hostid and falls back to arc4rand().
727 getcredhostid(curthread->td_ucred, &hostid);
729 if (fb || hostid == 0) {
730 ether_gen_addr(ifp, &sc->sc_defaddr);
732 sc->sc_defaddr.octet[0] = 0x2;
733 sc->sc_defaddr.octet[1] = (hostid >> 24) & 0xff;
734 sc->sc_defaddr.octet[2] = (hostid >> 16) & 0xff;
735 sc->sc_defaddr.octet[3] = (hostid >> 8 ) & 0xff;
736 sc->sc_defaddr.octet[4] = hostid & 0xff;
737 sc->sc_defaddr.octet[5] = ifp->if_dunit & 0xff;
743 LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
745 if (memcmp(sc->sc_defaddr.octet,
746 IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
751 BRIDGE_LIST_UNLOCK();
752 } while (retry == 1);
754 bstp_attach(&sc->sc_stp, &bridge_ops);
755 ether_ifattach(ifp, sc->sc_defaddr.octet);
756 /* Now undo some of the damage... */
757 ifp->if_baudrate = 0;
758 ifp->if_type = IFT_BRIDGE;
761 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
762 BRIDGE_LIST_UNLOCK();
768 bridge_clone_destroy_cb(struct epoch_context *ctx)
770 struct bridge_softc *sc;
772 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
774 BRIDGE_LOCK_DESTROY(sc);
779 * bridge_clone_destroy:
781 * Destroy a bridge instance.
784 bridge_clone_destroy(struct ifnet *ifp)
786 struct bridge_softc *sc = ifp->if_softc;
787 struct bridge_iflist *bif;
788 struct epoch_tracker et;
793 ifp->if_flags &= ~IFF_UP;
795 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
796 bridge_delete_member(sc, bif, 0);
798 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
799 bridge_delete_span(sc, bif);
802 /* Tear down the routing table. */
803 bridge_rtable_fini(sc);
807 NET_EPOCH_ENTER_ET(et);
809 callout_drain(&sc->sc_brcallout);
812 LIST_REMOVE(sc, sc_list);
813 BRIDGE_LIST_UNLOCK();
815 bstp_detach(&sc->sc_stp);
816 NET_EPOCH_EXIT_ET(et);
821 epoch_call(net_epoch_preempt, &sc->sc_epoch_ctx, bridge_clone_destroy_cb);
827 * Handle a control request from the operator.
830 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
832 struct bridge_softc *sc = ifp->if_softc;
833 struct ifreq *ifr = (struct ifreq *)data;
834 struct bridge_iflist *bif;
835 struct thread *td = curthread;
837 struct ifbreq ifbreq;
838 struct ifbifconf ifbifconf;
839 struct ifbareq ifbareq;
840 struct ifbaconf ifbaconf;
841 struct ifbrparam ifbrparam;
842 struct ifbropreq ifbropreq;
844 struct ifdrv *ifd = (struct ifdrv *) data;
845 const struct bridge_control *bc;
846 int error = 0, oldmtu;
858 if (ifd->ifd_cmd >= bridge_control_table_size) {
862 bc = &bridge_control_table[ifd->ifd_cmd];
864 if (cmd == SIOCGDRVSPEC &&
865 (bc->bc_flags & BC_F_COPYOUT) == 0) {
869 else if (cmd == SIOCSDRVSPEC &&
870 (bc->bc_flags & BC_F_COPYOUT) != 0) {
875 if (bc->bc_flags & BC_F_SUSER) {
876 error = priv_check(td, PRIV_NET_BRIDGE);
881 if (ifd->ifd_len != bc->bc_argsize ||
882 ifd->ifd_len > sizeof(args)) {
887 bzero(&args, sizeof(args));
888 if (bc->bc_flags & BC_F_COPYIN) {
889 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
894 oldmtu = ifp->if_mtu;
895 error = (*bc->bc_func)(sc, &args);
900 * Bridge MTU may change during addition of the first port.
901 * If it did, do network layer specific procedure.
903 if (ifp->if_mtu != oldmtu) {
910 if (bc->bc_flags & BC_F_COPYOUT)
911 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
916 if (!(ifp->if_flags & IFF_UP) &&
917 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
919 * If interface is marked down and it is running,
920 * then stop and disable it.
923 } else if ((ifp->if_flags & IFF_UP) &&
924 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
926 * If interface is marked up and it is stopped, then
936 if (ifr->ifr_mtu < 576) {
940 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
941 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
944 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
945 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
946 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
947 " != %d\n", sc->sc_ifp->if_xname,
948 bif->bif_ifp->if_mtu,
949 bif->bif_ifp->if_xname, ifr->ifr_mtu);
955 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
959 * drop the lock as ether_ioctl() will call bridge_start() and
960 * cause the lock to be recursed.
963 error = ether_ioctl(ifp, cmd, data);
976 * Clear or restore unwanted capabilities on the member interface
979 bridge_mutecaps(struct bridge_softc *sc)
981 struct bridge_iflist *bif;
984 BRIDGE_LOCK_ASSERT(sc);
986 /* Initial bitmask of capabilities to test */
987 mask = BRIDGE_IFCAPS_MASK;
989 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
990 /* Every member must support it or its disabled */
991 mask &= bif->bif_savedcaps;
994 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
995 enabled = bif->bif_ifp->if_capenable;
996 enabled &= ~BRIDGE_IFCAPS_STRIP;
997 /* strip off mask bits and enable them again if allowed */
998 enabled &= ~BRIDGE_IFCAPS_MASK;
1000 bridge_set_ifcap(sc, bif, enabled);
1005 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1007 struct ifnet *ifp = bif->bif_ifp;
1009 int error, mask, stuck;
1011 bzero(&ifr, sizeof(ifr));
1012 ifr.ifr_reqcap = set;
1014 if (ifp->if_capenable != set) {
1015 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1017 if_printf(sc->sc_ifp,
1018 "error setting capabilities on %s: %d\n",
1019 ifp->if_xname, error);
1020 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1021 stuck = ifp->if_capenable & mask & ~set;
1023 if_printf(sc->sc_ifp,
1024 "can't disable some capabilities on %s: 0x%x\n",
1025 ifp->if_xname, stuck);
1030 * bridge_lookup_member:
1032 * Lookup a bridge member interface.
1034 static struct bridge_iflist *
1035 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1037 struct bridge_iflist *bif;
1040 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1042 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1044 if (strcmp(ifp->if_xname, name) == 0)
1052 * bridge_lookup_member_if:
1054 * Lookup a bridge member interface by ifnet*.
1056 static struct bridge_iflist *
1057 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1059 struct bridge_iflist *bif;
1061 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1063 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1064 if (bif->bif_ifp == member_ifp)
1072 bridge_delete_member_cb(struct epoch_context *ctx)
1074 struct bridge_iflist *bif;
1076 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1078 free(bif, M_DEVBUF);
1082 * bridge_delete_member:
1084 * Delete the specified member interface.
1087 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1090 struct ifnet *ifs = bif->bif_ifp;
1091 struct ifnet *fif = NULL;
1092 struct bridge_iflist *bifl;
1094 BRIDGE_LOCK_ASSERT(sc);
1096 if (bif->bif_flags & IFBIF_STP)
1097 bstp_disable(&bif->bif_stp);
1099 ifs->if_bridge = NULL;
1100 CK_LIST_REMOVE(bif, bif_next);
1103 * If removing the interface that gave the bridge its mac address, set
1104 * the mac address of the bridge to the address of the next member, or
1105 * to its default address if no members are left.
1107 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1108 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1109 bcopy(&sc->sc_defaddr,
1110 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1111 sc->sc_ifaddr = NULL;
1113 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1114 fif = bifl->bif_ifp;
1115 bcopy(IF_LLADDR(fif),
1116 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1117 sc->sc_ifaddr = fif;
1119 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1122 bridge_linkcheck(sc);
1123 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1125 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1126 BRIDGE_RT_UNLOCK(sc);
1127 KASSERT(bif->bif_addrcnt == 0,
1128 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1130 ifs->if_bridge_output = NULL;
1131 ifs->if_bridge_input = NULL;
1132 ifs->if_bridge_linkstate = NULL;
1134 switch (ifs->if_type) {
1138 * Take the interface out of promiscuous mode, but only
1139 * if it was promiscuous in the first place. It might
1140 * not be if we're in the bridge_ioctl_add() error path.
1142 if (ifs->if_flags & IFF_PROMISC)
1143 (void) ifpromisc(ifs, 0);
1151 panic("bridge_delete_member: impossible");
1155 /* reneable any interface capabilities */
1156 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1158 bstp_destroy(&bif->bif_stp); /* prepare to free */
1160 epoch_call(net_epoch_preempt, &bif->bif_epoch_ctx,
1161 bridge_delete_member_cb);
1165 * bridge_delete_span:
1167 * Delete the specified span interface.
1170 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1172 BRIDGE_LOCK_ASSERT(sc);
1174 KASSERT(bif->bif_ifp->if_bridge == NULL,
1175 ("%s: not a span interface", __func__));
1177 CK_LIST_REMOVE(bif, bif_next);
1179 epoch_call(net_epoch_preempt, &bif->bif_epoch_ctx,
1180 bridge_delete_member_cb);
1184 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1186 struct ifbreq *req = arg;
1187 struct bridge_iflist *bif = NULL;
1191 ifs = ifunit(req->ifbr_ifsname);
1194 if (ifs->if_ioctl == NULL) /* must be supported */
1197 /* If it's in the span list, it can't be a member. */
1198 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1199 if (ifs == bif->bif_ifp)
1202 if (ifs->if_bridge == sc)
1205 if (ifs->if_bridge != NULL)
1208 switch (ifs->if_type) {
1212 /* permitted interface types */
1220 * Two valid inet6 addresses with link-local scope must not be
1221 * on the parent interface and the member interfaces at the
1222 * same time. This restriction is needed to prevent violation
1223 * of link-local scope zone. Attempts to add a member
1224 * interface which has inet6 addresses when the parent has
1225 * inet6 triggers removal of all inet6 addresses on the member
1229 /* Check if the parent interface has a link-local scope addr. */
1230 if (V_allow_llz_overlap == 0 &&
1231 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1233 * If any, remove all inet6 addresses from the member
1236 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1237 if (in6ifa_llaonifp(bif->bif_ifp)) {
1238 in6_ifdetach(bif->bif_ifp);
1239 if_printf(sc->sc_ifp,
1240 "IPv6 addresses on %s have been removed "
1241 "before adding it as a member to prevent "
1242 "IPv6 address scope violation.\n",
1243 bif->bif_ifp->if_xname);
1246 if (in6ifa_llaonifp(ifs)) {
1248 if_printf(sc->sc_ifp,
1249 "IPv6 addresses on %s have been removed "
1250 "before adding it as a member to prevent "
1251 "IPv6 address scope violation.\n",
1256 /* Allow the first Ethernet member to define the MTU */
1257 if (CK_LIST_EMPTY(&sc->sc_iflist))
1258 sc->sc_ifp->if_mtu = ifs->if_mtu;
1259 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1260 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1261 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1265 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1270 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1271 bif->bif_savedcaps = ifs->if_capenable;
1274 * Assign the interface's MAC address to the bridge if it's the first
1275 * member and the MAC address of the bridge has not been changed from
1276 * the default randomly generated one.
1278 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1279 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1280 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1281 sc->sc_ifaddr = ifs;
1282 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1285 ifs->if_bridge = sc;
1286 ifs->if_bridge_output = bridge_output;
1287 ifs->if_bridge_input = bridge_input;
1288 ifs->if_bridge_linkstate = bridge_linkstate;
1289 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1291 * XXX: XLOCK HERE!?!
1293 * NOTE: insert_***HEAD*** should be safe for the traversals.
1295 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1297 /* Set interface capabilities to the intersection set of all members */
1298 bridge_mutecaps(sc);
1299 bridge_linkcheck(sc);
1301 /* Place the interface into promiscuous mode */
1302 switch (ifs->if_type) {
1305 error = ifpromisc(ifs, 1);
1310 bridge_delete_member(sc, bif, 0);
1315 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1317 struct ifbreq *req = arg;
1318 struct bridge_iflist *bif;
1320 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1324 bridge_delete_member(sc, bif, 0);
1330 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1332 struct ifbreq *req = arg;
1333 struct bridge_iflist *bif;
1334 struct bstp_port *bp;
1336 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1341 req->ifbr_ifsflags = bif->bif_flags;
1342 req->ifbr_state = bp->bp_state;
1343 req->ifbr_priority = bp->bp_priority;
1344 req->ifbr_path_cost = bp->bp_path_cost;
1345 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1346 req->ifbr_proto = bp->bp_protover;
1347 req->ifbr_role = bp->bp_role;
1348 req->ifbr_stpflags = bp->bp_flags;
1349 req->ifbr_addrcnt = bif->bif_addrcnt;
1350 req->ifbr_addrmax = bif->bif_addrmax;
1351 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1353 /* Copy STP state options as flags */
1354 if (bp->bp_operedge)
1355 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1356 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1357 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1358 if (bp->bp_ptp_link)
1359 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1360 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1361 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1362 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1363 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1364 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1365 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1370 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1372 struct ifbreq *req = arg;
1373 struct bridge_iflist *bif;
1374 struct bstp_port *bp;
1377 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1382 if (req->ifbr_ifsflags & IFBIF_SPAN)
1383 /* SPAN is readonly */
1386 if (req->ifbr_ifsflags & IFBIF_STP) {
1387 if ((bif->bif_flags & IFBIF_STP) == 0) {
1388 error = bstp_enable(&bif->bif_stp);
1393 if ((bif->bif_flags & IFBIF_STP) != 0)
1394 bstp_disable(&bif->bif_stp);
1397 /* Pass on STP flags */
1398 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1399 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1400 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1401 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1403 /* Save the bits relating to the bridge */
1404 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1410 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1412 struct ifbrparam *param = arg;
1414 sc->sc_brtmax = param->ifbrp_csize;
1421 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1423 struct ifbrparam *param = arg;
1425 param->ifbrp_csize = sc->sc_brtmax;
1431 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1433 struct ifbifconf *bifc = arg;
1434 struct bridge_iflist *bif;
1437 int count, buflen, len, error = 0;
1440 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1442 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1445 buflen = sizeof(breq) * count;
1446 if (bifc->ifbic_len == 0) {
1447 bifc->ifbic_len = buflen;
1450 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1456 len = min(bifc->ifbic_len, buflen);
1457 bzero(&breq, sizeof(breq));
1458 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1459 if (len < sizeof(breq))
1462 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1463 sizeof(breq.ifbr_ifsname));
1464 /* Fill in the ifbreq structure */
1465 error = bridge_ioctl_gifflags(sc, &breq);
1468 memcpy(buf, &breq, sizeof(breq));
1470 buf += sizeof(breq);
1471 len -= sizeof(breq);
1473 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1474 if (len < sizeof(breq))
1477 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1478 sizeof(breq.ifbr_ifsname));
1479 breq.ifbr_ifsflags = bif->bif_flags;
1480 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1481 memcpy(buf, &breq, sizeof(breq));
1483 buf += sizeof(breq);
1484 len -= sizeof(breq);
1487 bifc->ifbic_len = sizeof(breq) * count;
1488 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1489 free(outbuf, M_TEMP);
1494 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1496 struct ifbaconf *bac = arg;
1497 struct bridge_rtnode *brt;
1498 struct ifbareq bareq;
1500 int count, buflen, len, error = 0;
1502 if (bac->ifbac_len == 0)
1506 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1508 buflen = sizeof(bareq) * count;
1510 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1516 len = min(bac->ifbac_len, buflen);
1517 bzero(&bareq, sizeof(bareq));
1518 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1519 if (len < sizeof(bareq))
1521 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1522 sizeof(bareq.ifba_ifsname));
1523 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1524 bareq.ifba_vlan = brt->brt_vlan;
1525 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1526 time_uptime < brt->brt_expire)
1527 bareq.ifba_expire = brt->brt_expire - time_uptime;
1529 bareq.ifba_expire = 0;
1530 bareq.ifba_flags = brt->brt_flags;
1532 memcpy(buf, &bareq, sizeof(bareq));
1534 buf += sizeof(bareq);
1535 len -= sizeof(bareq);
1538 bac->ifbac_len = sizeof(bareq) * count;
1539 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1540 free(outbuf, M_TEMP);
1545 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1547 struct ifbareq *req = arg;
1548 struct bridge_iflist *bif;
1549 struct epoch_tracker et;
1552 NET_EPOCH_ENTER_ET(et);
1554 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1556 NET_EPOCH_EXIT_ET(et);
1560 /* bridge_rtupdate() may acquire the lock. */
1561 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1563 NET_EPOCH_EXIT_ET(et);
1569 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1571 struct ifbrparam *param = arg;
1573 sc->sc_brttimeout = param->ifbrp_ctime;
1578 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1580 struct ifbrparam *param = arg;
1582 param->ifbrp_ctime = sc->sc_brttimeout;
1587 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1589 struct ifbareq *req = arg;
1591 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1595 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1597 struct ifbreq *req = arg;
1600 bridge_rtflush(sc, req->ifbr_ifsflags);
1601 BRIDGE_RT_UNLOCK(sc);
1607 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1609 struct ifbrparam *param = arg;
1610 struct bstp_state *bs = &sc->sc_stp;
1612 param->ifbrp_prio = bs->bs_bridge_priority;
1617 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1619 struct ifbrparam *param = arg;
1621 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1625 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1627 struct ifbrparam *param = arg;
1628 struct bstp_state *bs = &sc->sc_stp;
1630 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1635 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1637 struct ifbrparam *param = arg;
1639 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1643 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1645 struct ifbrparam *param = arg;
1646 struct bstp_state *bs = &sc->sc_stp;
1648 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1653 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1655 struct ifbrparam *param = arg;
1657 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1661 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1663 struct ifbrparam *param = arg;
1664 struct bstp_state *bs = &sc->sc_stp;
1666 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1671 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1673 struct ifbrparam *param = arg;
1675 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1679 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1681 struct ifbreq *req = arg;
1682 struct bridge_iflist *bif;
1684 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1688 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1692 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1694 struct ifbreq *req = arg;
1695 struct bridge_iflist *bif;
1697 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1701 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1705 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1707 struct ifbreq *req = arg;
1708 struct bridge_iflist *bif;
1710 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1714 bif->bif_addrmax = req->ifbr_addrmax;
1719 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1721 struct ifbreq *req = arg;
1722 struct bridge_iflist *bif = NULL;
1725 ifs = ifunit(req->ifbr_ifsname);
1729 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1730 if (ifs == bif->bif_ifp)
1733 if (ifs->if_bridge != NULL)
1736 switch (ifs->if_type) {
1745 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1750 bif->bif_flags = IFBIF_SPAN;
1752 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1758 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1760 struct ifbreq *req = arg;
1761 struct bridge_iflist *bif;
1764 ifs = ifunit(req->ifbr_ifsname);
1768 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1769 if (ifs == bif->bif_ifp)
1775 bridge_delete_span(sc, bif);
1781 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1783 struct ifbropreq *req = arg;
1784 struct bstp_state *bs = &sc->sc_stp;
1785 struct bstp_port *root_port;
1787 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1788 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1789 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1791 root_port = bs->bs_root_port;
1792 if (root_port == NULL)
1793 req->ifbop_root_port = 0;
1795 req->ifbop_root_port = root_port->bp_ifp->if_index;
1797 req->ifbop_holdcount = bs->bs_txholdcount;
1798 req->ifbop_priority = bs->bs_bridge_priority;
1799 req->ifbop_protocol = bs->bs_protover;
1800 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1801 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1802 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1803 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1804 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1805 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1811 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1813 struct ifbrparam *param = arg;
1815 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1820 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1822 struct ifbpstpconf *bifstp = arg;
1823 struct bridge_iflist *bif;
1824 struct bstp_port *bp;
1825 struct ifbpstpreq bpreq;
1827 int count, buflen, len, error = 0;
1830 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1831 if ((bif->bif_flags & IFBIF_STP) != 0)
1835 buflen = sizeof(bpreq) * count;
1836 if (bifstp->ifbpstp_len == 0) {
1837 bifstp->ifbpstp_len = buflen;
1841 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1847 len = min(bifstp->ifbpstp_len, buflen);
1848 bzero(&bpreq, sizeof(bpreq));
1849 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1850 if (len < sizeof(bpreq))
1853 if ((bif->bif_flags & IFBIF_STP) == 0)
1857 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1858 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1859 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1860 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1861 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1862 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1864 memcpy(buf, &bpreq, sizeof(bpreq));
1866 buf += sizeof(bpreq);
1867 len -= sizeof(bpreq);
1870 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1871 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1872 free(outbuf, M_TEMP);
1877 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1879 struct ifbrparam *param = arg;
1881 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1885 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1887 struct ifbrparam *param = arg;
1889 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1895 * Detach an interface from a bridge. Called when a member
1896 * interface is detaching.
1899 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1901 struct bridge_softc *sc = ifp->if_bridge;
1902 struct bridge_iflist *bif;
1904 if (ifp->if_flags & IFF_RENAMING)
1906 if (V_bridge_cloner == NULL) {
1908 * This detach handler can be called after
1909 * vnet_bridge_uninit(). Just return in that case.
1913 /* Check if the interface is a bridge member */
1917 bif = bridge_lookup_member_if(sc, ifp);
1919 bridge_delete_member(sc, bif, 1);
1925 /* Check if the interface is a span port */
1927 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1929 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1930 if (ifp == bif->bif_ifp) {
1931 bridge_delete_span(sc, bif);
1937 BRIDGE_LIST_UNLOCK();
1943 * Initialize a bridge interface.
1946 bridge_init(void *xsc)
1948 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1949 struct ifnet *ifp = sc->sc_ifp;
1951 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1955 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1958 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1959 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1967 * Stop the bridge interface.
1970 bridge_stop(struct ifnet *ifp, int disable)
1972 struct bridge_softc *sc = ifp->if_softc;
1974 BRIDGE_LOCK_ASSERT(sc);
1976 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1980 callout_stop(&sc->sc_brcallout);
1982 bstp_stop(&sc->sc_stp);
1984 bridge_rtflush(sc, IFBF_FLUSHDYN);
1985 BRIDGE_RT_UNLOCK(sc);
1987 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1993 * Enqueue a packet on a bridge member interface.
1997 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2003 /* We may be sending a fragment so traverse the mbuf */
2006 m->m_nextpkt = NULL;
2007 len = m->m_pkthdr.len;
2008 mflags = m->m_flags;
2011 * If underlying interface can not do VLAN tag insertion itself
2012 * then attach a packet tag that holds it.
2014 if ((m->m_flags & M_VLANTAG) &&
2015 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2016 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2019 "unable to prepend VLAN header\n");
2020 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2023 m->m_flags &= ~M_VLANTAG;
2026 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2027 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2029 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2033 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2034 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2035 if (mflags & M_MCAST)
2036 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2045 * Receive a queued packet from dummynet and pass it on to the output
2048 * The mbuf has the Ethernet header already attached.
2051 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2053 struct bridge_softc *sc;
2055 sc = ifp->if_bridge;
2058 * The packet didnt originate from a member interface. This should only
2059 * ever happen if a member interface is removed while packets are
2067 if (PFIL_HOOKED(&V_inet_pfil_hook)
2069 || PFIL_HOOKED(&V_inet6_pfil_hook)
2072 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2078 bridge_enqueue(sc, ifp, m);
2084 * Send output from a bridge member interface. This
2085 * performs the bridging function for locally originated
2088 * The mbuf has the Ethernet header already attached. We must
2089 * enqueue or free the mbuf before returning.
2092 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2095 struct epoch_tracker et;
2096 struct ether_header *eh;
2097 struct ifnet *dst_if;
2098 struct bridge_softc *sc;
2101 NET_EPOCH_ENTER_ET(et);
2103 if (m->m_len < ETHER_HDR_LEN) {
2104 m = m_pullup(m, ETHER_HDR_LEN);
2106 NET_EPOCH_EXIT_ET(et);
2111 eh = mtod(m, struct ether_header *);
2112 sc = ifp->if_bridge;
2113 vlan = VLANTAGOF(m);
2116 * If bridge is down, but the original output interface is up,
2117 * go ahead and send out that interface. Otherwise, the packet
2120 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2126 * If the packet is a multicast, or we don't know a better way to
2127 * get there, send to all interfaces.
2129 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2132 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2133 if (dst_if == NULL) {
2134 struct bridge_iflist *bif;
2140 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2141 dst_if = bif->bif_ifp;
2143 if (dst_if->if_type == IFT_GIF)
2145 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2149 * If this is not the original output interface,
2150 * and the interface is participating in spanning
2151 * tree, make sure the port is in a state that
2152 * allows forwarding.
2154 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2155 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2158 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2162 mc = m_copypacket(m, M_NOWAIT);
2164 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2169 bridge_enqueue(sc, dst_if, mc);
2173 NET_EPOCH_EXIT_ET(et);
2179 * XXX Spanning tree consideration here?
2183 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2185 NET_EPOCH_EXIT_ET(et);
2189 bridge_enqueue(sc, dst_if, m);
2190 NET_EPOCH_EXIT_ET(et);
2197 * Do output on a bridge.
2201 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2203 struct epoch_tracker et;
2204 struct bridge_softc *sc;
2205 struct ether_header *eh;
2206 struct ifnet *dst_if;
2209 NET_EPOCH_ENTER_ET(et);
2213 ETHER_BPF_MTAP(ifp, m);
2215 eh = mtod(m, struct ether_header *);
2217 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2218 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2219 error = bridge_enqueue(sc, dst_if, m);
2221 bridge_broadcast(sc, ifp, m, 0);
2223 NET_EPOCH_EXIT_ET(et);
2229 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2232 bridge_qflush(struct ifnet *ifp __unused)
2239 * The forwarding function of the bridge.
2241 * NOTE: Releases the lock on return.
2244 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2247 struct bridge_iflist *dbif;
2248 struct ifnet *src_if, *dst_if, *ifp;
2249 struct ether_header *eh;
2254 MPASS(in_epoch(net_epoch_preempt));
2256 src_if = m->m_pkthdr.rcvif;
2259 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2260 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2261 vlan = VLANTAGOF(m);
2263 if ((sbif->bif_flags & IFBIF_STP) &&
2264 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2267 eh = mtod(m, struct ether_header *);
2268 dst = eh->ether_dhost;
2270 /* If the interface is learning, record the address. */
2271 if (sbif->bif_flags & IFBIF_LEARNING) {
2272 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2273 sbif, 0, IFBAF_DYNAMIC);
2275 * If the interface has addresses limits then deny any source
2276 * that is not in the cache.
2278 if (error && sbif->bif_addrmax)
2282 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2283 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2287 * At this point, the port either doesn't participate
2288 * in spanning tree or it is in the forwarding state.
2292 * If the packet is unicast, destined for someone on
2293 * "this" side of the bridge, drop it.
2295 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2296 dst_if = bridge_rtlookup(sc, dst, vlan);
2297 if (src_if == dst_if)
2301 * Check if its a reserved multicast address, any address
2302 * listed in 802.1D section 7.12.6 may not be forwarded by the
2304 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2306 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2307 dst[2] == 0xc2 && dst[3] == 0x00 &&
2308 dst[4] == 0x00 && dst[5] <= 0x0f)
2311 /* ...forward it to all interfaces. */
2312 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2317 * If we have a destination interface which is a member of our bridge,
2318 * OR this is a unicast packet, push it through the bpf(4) machinery.
2319 * For broadcast or multicast packets, don't bother because it will
2320 * be reinjected into ether_input. We do this before we pass the packets
2321 * through the pfil(9) framework, as it is possible that pfil(9) will
2322 * drop the packet, or possibly modify it, making it difficult to debug
2323 * firewall issues on the bridge.
2325 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2326 ETHER_BPF_MTAP(ifp, m);
2328 /* run the packet filter */
2329 if (PFIL_HOOKED(&V_inet_pfil_hook)
2331 || PFIL_HOOKED(&V_inet6_pfil_hook)
2334 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2340 if (dst_if == NULL) {
2341 bridge_broadcast(sc, src_if, m, 1);
2346 * At this point, we're dealing with a unicast frame
2347 * going to a different interface.
2349 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2352 dbif = bridge_lookup_member_if(sc, dst_if);
2354 /* Not a member of the bridge (anymore?) */
2357 /* Private segments can not talk to each other */
2358 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2361 if ((dbif->bif_flags & IFBIF_STP) &&
2362 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2365 if (PFIL_HOOKED(&V_inet_pfil_hook)
2367 || PFIL_HOOKED(&V_inet6_pfil_hook)
2370 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2376 bridge_enqueue(sc, dst_if, m);
2386 * Receive input from a member interface. Queue the packet for
2387 * bridging if it is not for us.
2389 static struct mbuf *
2390 bridge_input(struct ifnet *ifp, struct mbuf *m)
2392 struct epoch_tracker et;
2393 struct bridge_softc *sc = ifp->if_bridge;
2394 struct bridge_iflist *bif, *bif2;
2396 struct ether_header *eh;
2397 struct mbuf *mc, *mc2;
2401 NET_EPOCH_ENTER_ET(et);
2403 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2404 NET_EPOCH_EXIT_ET(et);
2409 vlan = VLANTAGOF(m);
2412 * Implement support for bridge monitoring. If this flag has been
2413 * set on this interface, discard the packet once we push it through
2414 * the bpf(4) machinery, but before we do, increment the byte and
2415 * packet counters associated with this interface.
2417 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2418 m->m_pkthdr.rcvif = bifp;
2419 ETHER_BPF_MTAP(bifp, m);
2420 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2421 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2423 NET_EPOCH_EXIT_ET(et);
2426 bif = bridge_lookup_member_if(sc, ifp);
2428 NET_EPOCH_EXIT_ET(et);
2432 eh = mtod(m, struct ether_header *);
2436 if (m->m_flags & (M_BCAST|M_MCAST)) {
2437 /* Tap off 802.1D packets; they do not get forwarded. */
2438 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2439 ETHER_ADDR_LEN) == 0) {
2440 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2441 NET_EPOCH_EXIT_ET(et);
2445 if ((bif->bif_flags & IFBIF_STP) &&
2446 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2447 NET_EPOCH_EXIT_ET(et);
2452 * Make a deep copy of the packet and enqueue the copy
2453 * for bridge processing; return the original packet for
2456 mc = m_dup(m, M_NOWAIT);
2458 NET_EPOCH_EXIT_ET(et);
2462 /* Perform the bridge forwarding function with the copy. */
2463 bridge_forward(sc, bif, mc);
2466 * Reinject the mbuf as arriving on the bridge so we have a
2467 * chance at claiming multicast packets. We can not loop back
2468 * here from ether_input as a bridge is never a member of a
2471 KASSERT(bifp->if_bridge == NULL,
2472 ("loop created in bridge_input"));
2473 mc2 = m_dup(m, M_NOWAIT);
2475 /* Keep the layer3 header aligned */
2476 int i = min(mc2->m_pkthdr.len, max_protohdr);
2477 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2480 mc2->m_pkthdr.rcvif = bifp;
2481 (*bifp->if_input)(bifp, mc2);
2484 /* Return the original packet for local processing. */
2485 NET_EPOCH_EXIT_ET(et);
2489 if ((bif->bif_flags & IFBIF_STP) &&
2490 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2491 NET_EPOCH_EXIT_ET(et);
2495 #if (defined(INET) || defined(INET6))
2496 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2497 || ((iface)->if_carp \
2498 && (*carp_forus_p)((iface), eh->ether_dhost))
2499 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2500 || ((iface)->if_carp \
2501 && (*carp_forus_p)((iface), eh->ether_shost))
2503 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2504 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2508 # define OR_PFIL_HOOKED_INET6 \
2509 || PFIL_HOOKED(&V_inet6_pfil_hook)
2511 # define OR_PFIL_HOOKED_INET6
2514 #define GRAB_OUR_PACKETS(iface) \
2515 if ((iface)->if_type == IFT_GIF) \
2517 /* It is destined for us. */ \
2518 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2519 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2521 if ((iface)->if_type == IFT_BRIDGE) { \
2522 ETHER_BPF_MTAP(iface, m); \
2523 if_inc_counter(iface, IFCOUNTER_IPACKETS, 1); \
2524 if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2525 /* Filter on the physical interface. */ \
2526 if (V_pfil_local_phys && \
2527 (PFIL_HOOKED(&V_inet_pfil_hook) \
2528 OR_PFIL_HOOKED_INET6)) { \
2529 if (bridge_pfil(&m, NULL, ifp, \
2530 PFIL_IN) != 0 || m == NULL) { \
2531 NET_EPOCH_EXIT_ET(et); \
2534 eh = mtod(m, struct ether_header *); \
2537 if (bif->bif_flags & IFBIF_LEARNING) { \
2538 error = bridge_rtupdate(sc, eh->ether_shost, \
2539 vlan, bif, 0, IFBAF_DYNAMIC); \
2540 if (error && bif->bif_addrmax) { \
2542 NET_EPOCH_EXIT_ET(et); \
2546 m->m_pkthdr.rcvif = iface; \
2547 NET_EPOCH_EXIT_ET(et); \
2551 /* We just received a packet that we sent out. */ \
2552 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2553 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2556 NET_EPOCH_EXIT_ET(et); \
2561 * Unicast. Make sure it's not for the bridge.
2563 do { GRAB_OUR_PACKETS(bifp) } while (0);
2566 * Give a chance for ifp at first priority. This will help when the
2567 * packet comes through the interface like VLAN's with the same MACs
2568 * on several interfaces from the same bridge. This also will save
2569 * some CPU cycles in case the destination interface and the input
2570 * interface (eq ifp) are the same.
2572 do { GRAB_OUR_PACKETS(ifp) } while (0);
2574 /* Now check the all bridge members. */
2575 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2576 GRAB_OUR_PACKETS(bif2->bif_ifp)
2579 #undef OR_CARP_CHECK_WE_ARE_DST
2580 #undef OR_CARP_CHECK_WE_ARE_SRC
2581 #undef OR_PFIL_HOOKED_INET6
2582 #undef GRAB_OUR_PACKETS
2584 /* Perform the bridge forwarding function. */
2585 bridge_forward(sc, bif, m);
2587 NET_EPOCH_EXIT_ET(et);
2594 * Send a frame to all interfaces that are members of
2595 * the bridge, except for the one on which the packet
2598 * NOTE: Releases the lock on return.
2601 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2602 struct mbuf *m, int runfilt)
2604 struct bridge_iflist *dbif, *sbif;
2606 struct ifnet *dst_if;
2609 MPASS(in_epoch(net_epoch_preempt));
2611 sbif = bridge_lookup_member_if(sc, src_if);
2613 /* Filter on the bridge interface before broadcasting */
2614 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2616 || PFIL_HOOKED(&V_inet6_pfil_hook)
2619 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2625 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2626 dst_if = dbif->bif_ifp;
2627 if (dst_if == src_if)
2630 /* Private segments can not talk to each other */
2631 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2634 if ((dbif->bif_flags & IFBIF_STP) &&
2635 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2638 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2639 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2642 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2645 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2649 mc = m_dup(m, M_NOWAIT);
2651 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2657 * Filter on the output interface. Pass a NULL bridge interface
2658 * pointer so we do not redundantly filter on the bridge for
2659 * each interface we broadcast on.
2661 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2663 || PFIL_HOOKED(&V_inet6_pfil_hook)
2667 /* Keep the layer3 header aligned */
2668 i = min(mc->m_pkthdr.len, max_protohdr);
2669 mc = m_copyup(mc, i, ETHER_ALIGN);
2671 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2675 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2681 bridge_enqueue(sc, dst_if, mc);
2690 * Duplicate a packet out one or more interfaces that are in span mode,
2691 * the original mbuf is unmodified.
2694 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2696 struct bridge_iflist *bif;
2697 struct ifnet *dst_if;
2700 MPASS(in_epoch(net_epoch_preempt));
2702 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2705 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2706 dst_if = bif->bif_ifp;
2708 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2711 mc = m_copypacket(m, M_NOWAIT);
2713 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2717 bridge_enqueue(sc, dst_if, mc);
2724 * Add a bridge routing entry.
2727 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2728 struct bridge_iflist *bif, int setflags, uint8_t flags)
2730 struct bridge_rtnode *brt;
2733 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2735 /* Check the source address is valid and not multicast. */
2736 if (ETHER_IS_MULTICAST(dst) ||
2737 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2738 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2741 /* 802.1p frames map to vlan 1 */
2746 * A route for this destination might already exist. If so,
2747 * update it, otherwise create a new one.
2749 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2752 /* Check again, now that we have the lock. There could have
2753 * been a race and we only want to insert this once. */
2754 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2755 BRIDGE_RT_UNLOCK(sc);
2759 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2760 sc->sc_brtexceeded++;
2761 BRIDGE_RT_UNLOCK(sc);
2764 /* Check per interface address limits (if enabled) */
2765 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2766 bif->bif_addrexceeded++;
2767 BRIDGE_RT_UNLOCK(sc);
2772 * Allocate a new bridge forwarding node, and
2773 * initialize the expiration time and Ethernet
2776 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2778 BRIDGE_RT_UNLOCK(sc);
2781 brt->brt_vnet = curvnet;
2783 if (bif->bif_flags & IFBIF_STICKY)
2784 brt->brt_flags = IFBAF_STICKY;
2786 brt->brt_flags = IFBAF_DYNAMIC;
2788 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2789 brt->brt_vlan = vlan;
2791 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2792 uma_zfree(V_bridge_rtnode_zone, brt);
2793 BRIDGE_RT_UNLOCK(sc);
2799 BRIDGE_RT_UNLOCK(sc);
2802 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2803 brt->brt_dst != bif) {
2805 brt->brt_dst->bif_addrcnt--;
2807 brt->brt_dst->bif_addrcnt++;
2808 BRIDGE_RT_UNLOCK(sc);
2811 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2812 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2814 brt->brt_flags = flags;
2822 * Lookup the destination interface for an address.
2824 static struct ifnet *
2825 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2827 struct bridge_rtnode *brt;
2829 MPASS(in_epoch(net_epoch_preempt));
2831 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2834 return (brt->brt_ifp);
2840 * Trim the routine table so that we have a number
2841 * of routing entries less than or equal to the
2845 bridge_rttrim(struct bridge_softc *sc)
2847 struct bridge_rtnode *brt, *nbrt;
2849 MPASS(in_epoch(net_epoch_preempt));
2850 BRIDGE_RT_LOCK_ASSERT(sc);
2852 /* Make sure we actually need to do this. */
2853 if (sc->sc_brtcnt <= sc->sc_brtmax)
2856 /* Force an aging cycle; this might trim enough addresses. */
2858 if (sc->sc_brtcnt <= sc->sc_brtmax)
2861 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2862 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2863 bridge_rtnode_destroy(sc, brt);
2864 if (sc->sc_brtcnt <= sc->sc_brtmax)
2873 * Aging timer for the bridge.
2876 bridge_timer(void *arg)
2878 struct bridge_softc *sc = arg;
2880 BRIDGE_RT_LOCK_ASSERT(sc);
2882 /* Destruction of rtnodes requires a proper vnet context */
2883 CURVNET_SET(sc->sc_ifp->if_vnet);
2886 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2887 callout_reset(&sc->sc_brcallout,
2888 bridge_rtable_prune_period * hz, bridge_timer, sc);
2895 * Perform an aging cycle.
2898 bridge_rtage(struct bridge_softc *sc)
2900 struct bridge_rtnode *brt, *nbrt;
2902 BRIDGE_RT_LOCK_ASSERT(sc);
2904 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2905 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2906 if (time_uptime >= brt->brt_expire)
2907 bridge_rtnode_destroy(sc, brt);
2915 * Remove all dynamic addresses from the bridge.
2918 bridge_rtflush(struct bridge_softc *sc, int full)
2920 struct bridge_rtnode *brt, *nbrt;
2922 BRIDGE_RT_LOCK_ASSERT(sc);
2924 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2925 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2926 bridge_rtnode_destroy(sc, brt);
2933 * Remove an address from the table.
2936 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2938 struct bridge_rtnode *brt;
2944 * If vlan is zero then we want to delete for all vlans so the lookup
2945 * may return more than one.
2947 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2948 bridge_rtnode_destroy(sc, brt);
2952 BRIDGE_RT_UNLOCK(sc);
2954 return (found ? 0 : ENOENT);
2960 * Delete routes to a speicifc member interface.
2963 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2965 struct bridge_rtnode *brt, *nbrt;
2967 BRIDGE_RT_LOCK_ASSERT(sc);
2969 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2970 if (brt->brt_ifp == ifp && (full ||
2971 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2972 bridge_rtnode_destroy(sc, brt);
2977 * bridge_rtable_init:
2979 * Initialize the route table for this bridge.
2982 bridge_rtable_init(struct bridge_softc *sc)
2986 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2987 M_DEVBUF, M_WAITOK);
2989 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2990 CK_LIST_INIT(&sc->sc_rthash[i]);
2992 sc->sc_rthash_key = arc4random();
2993 CK_LIST_INIT(&sc->sc_rtlist);
2997 * bridge_rtable_fini:
2999 * Deconstruct the route table for this bridge.
3002 bridge_rtable_fini(struct bridge_softc *sc)
3005 KASSERT(sc->sc_brtcnt == 0,
3006 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3007 free(sc->sc_rthash, M_DEVBUF);
3011 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3012 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3014 #define mix(a, b, c) \
3016 a -= b; a -= c; a ^= (c >> 13); \
3017 b -= c; b -= a; b ^= (a << 8); \
3018 c -= a; c -= b; c ^= (b >> 13); \
3019 a -= b; a -= c; a ^= (c >> 12); \
3020 b -= c; b -= a; b ^= (a << 16); \
3021 c -= a; c -= b; c ^= (b >> 5); \
3022 a -= b; a -= c; a ^= (c >> 3); \
3023 b -= c; b -= a; b ^= (a << 10); \
3024 c -= a; c -= b; c ^= (b >> 15); \
3025 } while (/*CONSTCOND*/0)
3027 static __inline uint32_t
3028 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3030 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3041 return (c & BRIDGE_RTHASH_MASK);
3047 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3051 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3052 d = ((int)a[i]) - ((int)b[i]);
3059 * bridge_rtnode_lookup:
3061 * Look up a bridge route node for the specified destination. Compare the
3062 * vlan id or if zero then just return the first match.
3064 static struct bridge_rtnode *
3065 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3067 struct bridge_rtnode *brt;
3071 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3073 hash = bridge_rthash(sc, addr);
3074 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3075 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3076 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3086 * bridge_rtnode_insert:
3088 * Insert the specified bridge node into the route table. We
3089 * assume the entry is not already in the table.
3092 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3094 struct bridge_rtnode *lbrt;
3098 BRIDGE_RT_LOCK_ASSERT(sc);
3100 hash = bridge_rthash(sc, brt->brt_addr);
3102 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3104 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3109 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3110 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3113 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3116 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3117 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3120 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3121 } while (lbrt != NULL);
3124 panic("bridge_rtnode_insert: impossible");
3128 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3135 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3137 struct bridge_rtnode *brt;
3139 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3141 CURVNET_SET(brt->brt_vnet);
3142 uma_zfree(V_bridge_rtnode_zone, brt);
3147 * bridge_rtnode_destroy:
3149 * Destroy a bridge rtnode.
3152 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3155 BRIDGE_RT_LOCK_ASSERT(sc);
3157 CK_LIST_REMOVE(brt, brt_hash);
3159 CK_LIST_REMOVE(brt, brt_list);
3161 brt->brt_dst->bif_addrcnt--;
3163 epoch_call(net_epoch_preempt, &brt->brt_epoch_ctx,
3164 bridge_rtnode_destroy_cb);
3168 * bridge_rtable_expire:
3170 * Set the expiry time for all routes on an interface.
3173 bridge_rtable_expire(struct ifnet *ifp, int age)
3175 struct bridge_softc *sc = ifp->if_bridge;
3176 struct bridge_rtnode *brt;
3178 CURVNET_SET(ifp->if_vnet);
3182 * If the age is zero then flush, otherwise set all the expiry times to
3183 * age for the interface
3186 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3188 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3189 /* Cap the expiry time to 'age' */
3190 if (brt->brt_ifp == ifp &&
3191 brt->brt_expire > time_uptime + age &&
3192 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3193 brt->brt_expire = time_uptime + age;
3196 BRIDGE_RT_UNLOCK(sc);
3201 * bridge_state_change:
3203 * Callback from the bridgestp code when a port changes states.
3206 bridge_state_change(struct ifnet *ifp, int state)
3208 struct bridge_softc *sc = ifp->if_bridge;
3209 static const char *stpstates[] = {
3218 CURVNET_SET(ifp->if_vnet);
3220 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3221 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3226 * Send bridge packets through pfil if they are one of the types pfil can deal
3227 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3228 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3232 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3234 int snap, error, i, hlen;
3235 struct ether_header *eh1, eh2;
3238 u_int16_t ether_type;
3241 error = -1; /* Default error if not error == 0 */
3244 /* we may return with the IP fields swapped, ensure its not shared */
3245 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3248 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3249 return (0); /* filtering is disabled */
3251 i = min((*mp)->m_pkthdr.len, max_protohdr);
3252 if ((*mp)->m_len < i) {
3253 *mp = m_pullup(*mp, i);
3255 printf("%s: m_pullup failed\n", __func__);
3260 eh1 = mtod(*mp, struct ether_header *);
3261 ether_type = ntohs(eh1->ether_type);
3264 * Check for SNAP/LLC.
3266 if (ether_type < ETHERMTU) {
3267 struct llc *llc2 = (struct llc *)(eh1 + 1);
3269 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3270 llc2->llc_dsap == LLC_SNAP_LSAP &&
3271 llc2->llc_ssap == LLC_SNAP_LSAP &&
3272 llc2->llc_control == LLC_UI) {
3273 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3279 * If we're trying to filter bridge traffic, don't look at anything
3280 * other than IP and ARP traffic. If the filter doesn't understand
3281 * IPv6, don't allow IPv6 through the bridge either. This is lame
3282 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3283 * but of course we don't have an AppleTalk filter to begin with.
3284 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3287 switch (ether_type) {
3289 case ETHERTYPE_REVARP:
3290 if (V_pfil_ipfw_arp == 0)
3291 return (0); /* Automatically pass */
3296 case ETHERTYPE_IPV6:
3301 * Check to see if the user wants to pass non-ip
3302 * packets, these will not be checked by pfil(9) and
3303 * passed unconditionally so the default is to drop.
3309 /* Run the packet through pfil before stripping link headers */
3310 if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3311 dir == PFIL_OUT && ifp != NULL) {
3313 error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, 0,
3316 if (*mp == NULL || error != 0) /* packet consumed by filter */
3320 /* Strip off the Ethernet header and keep a copy. */
3321 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3322 m_adj(*mp, ETHER_HDR_LEN);
3324 /* Strip off snap header, if present */
3326 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3327 m_adj(*mp, sizeof(struct llc));
3331 * Check the IP header for alignment and errors
3333 if (dir == PFIL_IN) {
3334 switch (ether_type) {
3336 error = bridge_ip_checkbasic(mp);
3339 case ETHERTYPE_IPV6:
3340 error = bridge_ip6_checkbasic(mp);
3353 * Run the packet through pfil
3355 switch (ether_type) {
3358 * Run pfil on the member interface and the bridge, both can
3359 * be skipped by clearing pfil_member or pfil_bridge.
3362 * in_if -> bridge_if -> out_if
3364 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3365 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3368 if (*mp == NULL || error != 0) /* filter may consume */
3371 if (V_pfil_member && ifp != NULL)
3372 error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3375 if (*mp == NULL || error != 0) /* filter may consume */
3378 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3379 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3382 if (*mp == NULL || error != 0) /* filter may consume */
3385 /* check if we need to fragment the packet */
3386 /* bridge_fragment generates a mbuf chain of packets */
3387 /* that already include eth headers */
3388 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3389 i = (*mp)->m_pkthdr.len;
3390 if (i > ifp->if_mtu) {
3391 error = bridge_fragment(ifp, mp, &eh2, snap,
3397 /* Recalculate the ip checksum. */
3398 ip = mtod(*mp, struct ip *);
3399 hlen = ip->ip_hl << 2;
3400 if (hlen < sizeof(struct ip))
3402 if (hlen > (*mp)->m_len) {
3403 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3405 ip = mtod(*mp, struct ip *);
3410 if (hlen == sizeof(struct ip))
3411 ip->ip_sum = in_cksum_hdr(ip);
3413 ip->ip_sum = in_cksum(*mp, hlen);
3417 case ETHERTYPE_IPV6:
3418 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3419 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3422 if (*mp == NULL || error != 0) /* filter may consume */
3425 if (V_pfil_member && ifp != NULL)
3426 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3429 if (*mp == NULL || error != 0) /* filter may consume */
3432 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3433 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3450 * Finally, put everything back the way it was and return
3453 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3456 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3459 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3462 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3473 * Perform basic checks on header size since
3474 * pfil assumes ip_input has already processed
3475 * it for it. Cut-and-pasted from ip_input.c.
3476 * Given how simple the IPv6 version is,
3477 * does the IPv4 version really need to be
3480 * XXX Should we update ipstat here, or not?
3481 * XXX Right now we update ipstat but not
3485 bridge_ip_checkbasic(struct mbuf **mp)
3487 struct mbuf *m = *mp;
3495 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3496 if ((m = m_copyup(m, sizeof(struct ip),
3497 (max_linkhdr + 3) & ~3)) == NULL) {
3498 /* XXXJRT new stat, please */
3499 KMOD_IPSTAT_INC(ips_toosmall);
3502 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3503 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3504 KMOD_IPSTAT_INC(ips_toosmall);
3508 ip = mtod(m, struct ip *);
3509 if (ip == NULL) goto bad;
3511 if (ip->ip_v != IPVERSION) {
3512 KMOD_IPSTAT_INC(ips_badvers);
3515 hlen = ip->ip_hl << 2;
3516 if (hlen < sizeof(struct ip)) { /* minimum header length */
3517 KMOD_IPSTAT_INC(ips_badhlen);
3520 if (hlen > m->m_len) {
3521 if ((m = m_pullup(m, hlen)) == NULL) {
3522 KMOD_IPSTAT_INC(ips_badhlen);
3525 ip = mtod(m, struct ip *);
3526 if (ip == NULL) goto bad;
3529 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3530 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3532 if (hlen == sizeof(struct ip)) {
3533 sum = in_cksum_hdr(ip);
3535 sum = in_cksum(m, hlen);
3539 KMOD_IPSTAT_INC(ips_badsum);
3543 /* Retrieve the packet length. */
3544 len = ntohs(ip->ip_len);
3547 * Check for additional length bogosity
3550 KMOD_IPSTAT_INC(ips_badlen);
3555 * Check that the amount of data in the buffers
3556 * is as at least much as the IP header would have us expect.
3557 * Drop packet if shorter than we expect.
3559 if (m->m_pkthdr.len < len) {
3560 KMOD_IPSTAT_INC(ips_tooshort);
3564 /* Checks out, proceed */
3575 * Same as above, but for IPv6.
3576 * Cut-and-pasted from ip6_input.c.
3577 * XXX Should we update ip6stat, or not?
3580 bridge_ip6_checkbasic(struct mbuf **mp)
3582 struct mbuf *m = *mp;
3583 struct ip6_hdr *ip6;
3586 * If the IPv6 header is not aligned, slurp it up into a new
3587 * mbuf with space for link headers, in the event we forward
3588 * it. Otherwise, if it is aligned, make sure the entire base
3589 * IPv6 header is in the first mbuf of the chain.
3591 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3592 struct ifnet *inifp = m->m_pkthdr.rcvif;
3593 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3594 (max_linkhdr + 3) & ~3)) == NULL) {
3595 /* XXXJRT new stat, please */
3596 IP6STAT_INC(ip6s_toosmall);
3597 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3600 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3601 struct ifnet *inifp = m->m_pkthdr.rcvif;
3602 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3603 IP6STAT_INC(ip6s_toosmall);
3604 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3609 ip6 = mtod(m, struct ip6_hdr *);
3611 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3612 IP6STAT_INC(ip6s_badvers);
3613 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3617 /* Checks out, proceed */
3630 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3633 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3634 int snap, struct llc *llc)
3636 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3640 if (m->m_len < sizeof(struct ip) &&
3641 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3643 ip = mtod(m, struct ip *);
3645 m->m_pkthdr.csum_flags |= CSUM_IP;
3646 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3651 * Walk the chain and re-add the Ethernet header for
3654 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3655 nextpkt = mcur->m_nextpkt;
3656 mcur->m_nextpkt = NULL;
3658 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3662 mprev->m_nextpkt = nextpkt;
3665 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3668 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3672 mprev->m_nextpkt = nextpkt;
3675 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3678 * The previous two M_PREPEND could have inserted one or two
3679 * mbufs in front so we have to update the previous packet's
3682 mcur->m_nextpkt = nextpkt;
3684 mprev->m_nextpkt = mcur;
3686 /* The first mbuf in the original chain needs to be
3693 KMOD_IPSTAT_INC(ips_fragmented);
3697 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3698 m = mcur->m_nextpkt;
3705 bridge_linkstate(struct ifnet *ifp)
3707 struct bridge_softc *sc = ifp->if_bridge;
3708 struct bridge_iflist *bif;
3709 struct epoch_tracker et;
3711 NET_EPOCH_ENTER_ET(et);
3713 bif = bridge_lookup_member_if(sc, ifp);
3715 NET_EPOCH_EXIT_ET(et);
3718 bridge_linkcheck(sc);
3720 bstp_linkstate(&bif->bif_stp);
3722 NET_EPOCH_EXIT_ET(et);
3726 bridge_linkcheck(struct bridge_softc *sc)
3728 struct bridge_iflist *bif;
3729 int new_link, hasls;
3731 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3733 new_link = LINK_STATE_DOWN;
3735 /* Our link is considered up if at least one of our ports is active */
3736 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3737 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3739 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3740 new_link = LINK_STATE_UP;
3744 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3745 /* If no interfaces support link-state then we default to up */
3746 new_link = LINK_STATE_UP;
3748 if_link_state_change(sc->sc_ifp, new_link);