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");
679 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
681 struct bridge_softc *sc = ifp->if_softc;
682 struct bridge_iflist *bif;
686 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
687 bridge_delete_member(sc, bif, 0);
689 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
690 bridge_delete_span(sc, bif);
695 ether_reassign(ifp, newvnet, arg);
700 * bridge_clone_create:
702 * Create a new bridge instance.
705 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
707 struct bridge_softc *sc, *sc2;
708 struct ifnet *bifp, *ifp;
710 unsigned long hostid;
712 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
713 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
719 BRIDGE_LOCK_INIT(sc);
720 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
721 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
723 /* Initialize our routing table. */
724 bridge_rtable_init(sc);
726 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
728 CK_LIST_INIT(&sc->sc_iflist);
729 CK_LIST_INIT(&sc->sc_spanlist);
732 if_initname(ifp, bridge_name, unit);
733 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
734 ifp->if_ioctl = bridge_ioctl;
735 ifp->if_transmit = bridge_transmit;
736 ifp->if_qflush = bridge_qflush;
737 ifp->if_init = bridge_init;
738 ifp->if_type = IFT_BRIDGE;
741 * Generate an ethernet address with a locally administered address.
743 * Since we are using random ethernet addresses for the bridge, it is
744 * possible that we might have address collisions, so make sure that
745 * this hardware address isn't already in use on another bridge.
746 * The first try uses the hostid and falls back to arc4rand().
749 getcredhostid(curthread->td_ucred, &hostid);
751 if (fb || hostid == 0) {
752 ether_gen_addr(ifp, &sc->sc_defaddr);
754 sc->sc_defaddr.octet[0] = 0x2;
755 sc->sc_defaddr.octet[1] = (hostid >> 24) & 0xff;
756 sc->sc_defaddr.octet[2] = (hostid >> 16) & 0xff;
757 sc->sc_defaddr.octet[3] = (hostid >> 8 ) & 0xff;
758 sc->sc_defaddr.octet[4] = hostid & 0xff;
759 sc->sc_defaddr.octet[5] = ifp->if_dunit & 0xff;
765 LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
767 if (memcmp(sc->sc_defaddr.octet,
768 IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
773 BRIDGE_LIST_UNLOCK();
774 } while (retry == 1);
776 bstp_attach(&sc->sc_stp, &bridge_ops);
777 ether_ifattach(ifp, sc->sc_defaddr.octet);
778 /* Now undo some of the damage... */
779 ifp->if_baudrate = 0;
780 ifp->if_type = IFT_BRIDGE;
782 ifp->if_reassign = bridge_reassign;
786 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
787 BRIDGE_LIST_UNLOCK();
793 bridge_clone_destroy_cb(struct epoch_context *ctx)
795 struct bridge_softc *sc;
797 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
799 BRIDGE_LOCK_DESTROY(sc);
804 * bridge_clone_destroy:
806 * Destroy a bridge instance.
809 bridge_clone_destroy(struct ifnet *ifp)
811 struct bridge_softc *sc = ifp->if_softc;
812 struct bridge_iflist *bif;
813 struct epoch_tracker et;
818 ifp->if_flags &= ~IFF_UP;
820 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
821 bridge_delete_member(sc, bif, 0);
823 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
824 bridge_delete_span(sc, bif);
827 /* Tear down the routing table. */
828 bridge_rtable_fini(sc);
832 NET_EPOCH_ENTER_ET(et);
834 callout_drain(&sc->sc_brcallout);
837 LIST_REMOVE(sc, sc_list);
838 BRIDGE_LIST_UNLOCK();
840 bstp_detach(&sc->sc_stp);
841 NET_EPOCH_EXIT_ET(et);
846 epoch_call(net_epoch_preempt, &sc->sc_epoch_ctx, bridge_clone_destroy_cb);
852 * Handle a control request from the operator.
855 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
857 struct bridge_softc *sc = ifp->if_softc;
858 struct ifreq *ifr = (struct ifreq *)data;
859 struct bridge_iflist *bif;
860 struct thread *td = curthread;
862 struct ifbreq ifbreq;
863 struct ifbifconf ifbifconf;
864 struct ifbareq ifbareq;
865 struct ifbaconf ifbaconf;
866 struct ifbrparam ifbrparam;
867 struct ifbropreq ifbropreq;
869 struct ifdrv *ifd = (struct ifdrv *) data;
870 const struct bridge_control *bc;
871 int error = 0, oldmtu;
883 if (ifd->ifd_cmd >= bridge_control_table_size) {
887 bc = &bridge_control_table[ifd->ifd_cmd];
889 if (cmd == SIOCGDRVSPEC &&
890 (bc->bc_flags & BC_F_COPYOUT) == 0) {
894 else if (cmd == SIOCSDRVSPEC &&
895 (bc->bc_flags & BC_F_COPYOUT) != 0) {
900 if (bc->bc_flags & BC_F_SUSER) {
901 error = priv_check(td, PRIV_NET_BRIDGE);
906 if (ifd->ifd_len != bc->bc_argsize ||
907 ifd->ifd_len > sizeof(args)) {
912 bzero(&args, sizeof(args));
913 if (bc->bc_flags & BC_F_COPYIN) {
914 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
919 oldmtu = ifp->if_mtu;
920 error = (*bc->bc_func)(sc, &args);
925 * Bridge MTU may change during addition of the first port.
926 * If it did, do network layer specific procedure.
928 if (ifp->if_mtu != oldmtu) {
935 if (bc->bc_flags & BC_F_COPYOUT)
936 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
941 if (!(ifp->if_flags & IFF_UP) &&
942 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
944 * If interface is marked down and it is running,
945 * then stop and disable it.
948 } else if ((ifp->if_flags & IFF_UP) &&
949 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
951 * If interface is marked up and it is stopped, then
961 if (ifr->ifr_mtu < 576) {
965 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
966 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
969 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
970 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
971 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
972 " != %d\n", sc->sc_ifp->if_xname,
973 bif->bif_ifp->if_mtu,
974 bif->bif_ifp->if_xname, ifr->ifr_mtu);
980 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
984 * drop the lock as ether_ioctl() will call bridge_start() and
985 * cause the lock to be recursed.
988 error = ether_ioctl(ifp, cmd, data);
1001 * Clear or restore unwanted capabilities on the member interface
1004 bridge_mutecaps(struct bridge_softc *sc)
1006 struct bridge_iflist *bif;
1009 BRIDGE_LOCK_ASSERT(sc);
1011 /* Initial bitmask of capabilities to test */
1012 mask = BRIDGE_IFCAPS_MASK;
1014 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1015 /* Every member must support it or its disabled */
1016 mask &= bif->bif_savedcaps;
1019 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1020 enabled = bif->bif_ifp->if_capenable;
1021 enabled &= ~BRIDGE_IFCAPS_STRIP;
1022 /* strip off mask bits and enable them again if allowed */
1023 enabled &= ~BRIDGE_IFCAPS_MASK;
1025 bridge_set_ifcap(sc, bif, enabled);
1030 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1032 struct ifnet *ifp = bif->bif_ifp;
1034 int error, mask, stuck;
1036 bzero(&ifr, sizeof(ifr));
1037 ifr.ifr_reqcap = set;
1039 if (ifp->if_capenable != set) {
1040 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1042 if_printf(sc->sc_ifp,
1043 "error setting capabilities on %s: %d\n",
1044 ifp->if_xname, error);
1045 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1046 stuck = ifp->if_capenable & mask & ~set;
1048 if_printf(sc->sc_ifp,
1049 "can't disable some capabilities on %s: 0x%x\n",
1050 ifp->if_xname, stuck);
1055 * bridge_lookup_member:
1057 * Lookup a bridge member interface.
1059 static struct bridge_iflist *
1060 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1062 struct bridge_iflist *bif;
1065 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1067 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1069 if (strcmp(ifp->if_xname, name) == 0)
1077 * bridge_lookup_member_if:
1079 * Lookup a bridge member interface by ifnet*.
1081 static struct bridge_iflist *
1082 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1084 struct bridge_iflist *bif;
1086 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1088 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1089 if (bif->bif_ifp == member_ifp)
1097 bridge_delete_member_cb(struct epoch_context *ctx)
1099 struct bridge_iflist *bif;
1101 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1103 free(bif, M_DEVBUF);
1107 * bridge_delete_member:
1109 * Delete the specified member interface.
1112 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1115 struct ifnet *ifs = bif->bif_ifp;
1116 struct ifnet *fif = NULL;
1117 struct bridge_iflist *bifl;
1119 BRIDGE_LOCK_ASSERT(sc);
1121 if (bif->bif_flags & IFBIF_STP)
1122 bstp_disable(&bif->bif_stp);
1124 ifs->if_bridge = NULL;
1125 CK_LIST_REMOVE(bif, bif_next);
1128 * If removing the interface that gave the bridge its mac address, set
1129 * the mac address of the bridge to the address of the next member, or
1130 * to its default address if no members are left.
1132 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1133 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1134 bcopy(&sc->sc_defaddr,
1135 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1136 sc->sc_ifaddr = NULL;
1138 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1139 fif = bifl->bif_ifp;
1140 bcopy(IF_LLADDR(fif),
1141 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1142 sc->sc_ifaddr = fif;
1144 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1147 bridge_linkcheck(sc);
1148 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1150 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1151 BRIDGE_RT_UNLOCK(sc);
1152 KASSERT(bif->bif_addrcnt == 0,
1153 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1155 ifs->if_bridge_output = NULL;
1156 ifs->if_bridge_input = NULL;
1157 ifs->if_bridge_linkstate = NULL;
1159 switch (ifs->if_type) {
1163 * Take the interface out of promiscuous mode, but only
1164 * if it was promiscuous in the first place. It might
1165 * not be if we're in the bridge_ioctl_add() error path.
1167 if (ifs->if_flags & IFF_PROMISC)
1168 (void) ifpromisc(ifs, 0);
1176 panic("bridge_delete_member: impossible");
1180 /* reneable any interface capabilities */
1181 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1183 bstp_destroy(&bif->bif_stp); /* prepare to free */
1185 epoch_call(net_epoch_preempt, &bif->bif_epoch_ctx,
1186 bridge_delete_member_cb);
1190 * bridge_delete_span:
1192 * Delete the specified span interface.
1195 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1197 BRIDGE_LOCK_ASSERT(sc);
1199 KASSERT(bif->bif_ifp->if_bridge == NULL,
1200 ("%s: not a span interface", __func__));
1202 CK_LIST_REMOVE(bif, bif_next);
1204 epoch_call(net_epoch_preempt, &bif->bif_epoch_ctx,
1205 bridge_delete_member_cb);
1209 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1211 struct ifbreq *req = arg;
1212 struct bridge_iflist *bif = NULL;
1216 ifs = ifunit(req->ifbr_ifsname);
1219 if (ifs->if_ioctl == NULL) /* must be supported */
1222 /* If it's in the span list, it can't be a member. */
1223 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1224 if (ifs == bif->bif_ifp)
1227 if (ifs->if_bridge == sc)
1230 if (ifs->if_bridge != NULL)
1233 switch (ifs->if_type) {
1237 /* permitted interface types */
1245 * Two valid inet6 addresses with link-local scope must not be
1246 * on the parent interface and the member interfaces at the
1247 * same time. This restriction is needed to prevent violation
1248 * of link-local scope zone. Attempts to add a member
1249 * interface which has inet6 addresses when the parent has
1250 * inet6 triggers removal of all inet6 addresses on the member
1254 /* Check if the parent interface has a link-local scope addr. */
1255 if (V_allow_llz_overlap == 0 &&
1256 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1258 * If any, remove all inet6 addresses from the member
1261 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1262 if (in6ifa_llaonifp(bif->bif_ifp)) {
1263 in6_ifdetach(bif->bif_ifp);
1264 if_printf(sc->sc_ifp,
1265 "IPv6 addresses on %s have been removed "
1266 "before adding it as a member to prevent "
1267 "IPv6 address scope violation.\n",
1268 bif->bif_ifp->if_xname);
1271 if (in6ifa_llaonifp(ifs)) {
1273 if_printf(sc->sc_ifp,
1274 "IPv6 addresses on %s have been removed "
1275 "before adding it as a member to prevent "
1276 "IPv6 address scope violation.\n",
1281 /* Allow the first Ethernet member to define the MTU */
1282 if (CK_LIST_EMPTY(&sc->sc_iflist))
1283 sc->sc_ifp->if_mtu = ifs->if_mtu;
1284 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1285 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1286 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1290 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1295 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1296 bif->bif_savedcaps = ifs->if_capenable;
1299 * Assign the interface's MAC address to the bridge if it's the first
1300 * member and the MAC address of the bridge has not been changed from
1301 * the default randomly generated one.
1303 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1304 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1305 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1306 sc->sc_ifaddr = ifs;
1307 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1310 ifs->if_bridge = sc;
1311 ifs->if_bridge_output = bridge_output;
1312 ifs->if_bridge_input = bridge_input;
1313 ifs->if_bridge_linkstate = bridge_linkstate;
1314 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1316 * XXX: XLOCK HERE!?!
1318 * NOTE: insert_***HEAD*** should be safe for the traversals.
1320 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1322 /* Set interface capabilities to the intersection set of all members */
1323 bridge_mutecaps(sc);
1324 bridge_linkcheck(sc);
1326 /* Place the interface into promiscuous mode */
1327 switch (ifs->if_type) {
1330 error = ifpromisc(ifs, 1);
1335 bridge_delete_member(sc, bif, 0);
1340 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1342 struct ifbreq *req = arg;
1343 struct bridge_iflist *bif;
1345 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1349 bridge_delete_member(sc, bif, 0);
1355 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1357 struct ifbreq *req = arg;
1358 struct bridge_iflist *bif;
1359 struct bstp_port *bp;
1361 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1366 req->ifbr_ifsflags = bif->bif_flags;
1367 req->ifbr_state = bp->bp_state;
1368 req->ifbr_priority = bp->bp_priority;
1369 req->ifbr_path_cost = bp->bp_path_cost;
1370 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1371 req->ifbr_proto = bp->bp_protover;
1372 req->ifbr_role = bp->bp_role;
1373 req->ifbr_stpflags = bp->bp_flags;
1374 req->ifbr_addrcnt = bif->bif_addrcnt;
1375 req->ifbr_addrmax = bif->bif_addrmax;
1376 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1378 /* Copy STP state options as flags */
1379 if (bp->bp_operedge)
1380 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1381 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1382 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1383 if (bp->bp_ptp_link)
1384 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1385 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1386 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1387 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1388 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1389 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1390 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1395 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1397 struct ifbreq *req = arg;
1398 struct bridge_iflist *bif;
1399 struct bstp_port *bp;
1402 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1407 if (req->ifbr_ifsflags & IFBIF_SPAN)
1408 /* SPAN is readonly */
1413 if (req->ifbr_ifsflags & IFBIF_STP) {
1414 if ((bif->bif_flags & IFBIF_STP) == 0) {
1415 error = bstp_enable(&bif->bif_stp);
1422 if ((bif->bif_flags & IFBIF_STP) != 0)
1423 bstp_disable(&bif->bif_stp);
1426 /* Pass on STP flags */
1427 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1428 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1429 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1430 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1432 /* Save the bits relating to the bridge */
1433 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1441 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1443 struct ifbrparam *param = arg;
1445 sc->sc_brtmax = param->ifbrp_csize;
1452 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1454 struct ifbrparam *param = arg;
1456 param->ifbrp_csize = sc->sc_brtmax;
1462 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1464 struct ifbifconf *bifc = arg;
1465 struct bridge_iflist *bif;
1468 int count, buflen, len, error = 0;
1471 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1473 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1476 buflen = sizeof(breq) * count;
1477 if (bifc->ifbic_len == 0) {
1478 bifc->ifbic_len = buflen;
1481 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1487 len = min(bifc->ifbic_len, buflen);
1488 bzero(&breq, sizeof(breq));
1489 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1490 if (len < sizeof(breq))
1493 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1494 sizeof(breq.ifbr_ifsname));
1495 /* Fill in the ifbreq structure */
1496 error = bridge_ioctl_gifflags(sc, &breq);
1499 memcpy(buf, &breq, sizeof(breq));
1501 buf += sizeof(breq);
1502 len -= sizeof(breq);
1504 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1505 if (len < sizeof(breq))
1508 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1509 sizeof(breq.ifbr_ifsname));
1510 breq.ifbr_ifsflags = bif->bif_flags;
1511 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1512 memcpy(buf, &breq, sizeof(breq));
1514 buf += sizeof(breq);
1515 len -= sizeof(breq);
1518 bifc->ifbic_len = sizeof(breq) * count;
1519 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1520 free(outbuf, M_TEMP);
1525 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1527 struct ifbaconf *bac = arg;
1528 struct bridge_rtnode *brt;
1529 struct ifbareq bareq;
1531 int count, buflen, len, error = 0;
1533 if (bac->ifbac_len == 0)
1537 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1539 buflen = sizeof(bareq) * count;
1541 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1547 len = min(bac->ifbac_len, buflen);
1548 bzero(&bareq, sizeof(bareq));
1549 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1550 if (len < sizeof(bareq))
1552 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1553 sizeof(bareq.ifba_ifsname));
1554 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1555 bareq.ifba_vlan = brt->brt_vlan;
1556 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1557 time_uptime < brt->brt_expire)
1558 bareq.ifba_expire = brt->brt_expire - time_uptime;
1560 bareq.ifba_expire = 0;
1561 bareq.ifba_flags = brt->brt_flags;
1563 memcpy(buf, &bareq, sizeof(bareq));
1565 buf += sizeof(bareq);
1566 len -= sizeof(bareq);
1569 bac->ifbac_len = sizeof(bareq) * count;
1570 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1571 free(outbuf, M_TEMP);
1576 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1578 struct ifbareq *req = arg;
1579 struct bridge_iflist *bif;
1580 struct epoch_tracker et;
1583 NET_EPOCH_ENTER_ET(et);
1585 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1587 NET_EPOCH_EXIT_ET(et);
1591 /* bridge_rtupdate() may acquire the lock. */
1592 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1594 NET_EPOCH_EXIT_ET(et);
1600 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1602 struct ifbrparam *param = arg;
1604 sc->sc_brttimeout = param->ifbrp_ctime;
1609 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1611 struct ifbrparam *param = arg;
1613 param->ifbrp_ctime = sc->sc_brttimeout;
1618 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1620 struct ifbareq *req = arg;
1622 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1626 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1628 struct ifbreq *req = arg;
1631 bridge_rtflush(sc, req->ifbr_ifsflags);
1632 BRIDGE_RT_UNLOCK(sc);
1638 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1640 struct ifbrparam *param = arg;
1641 struct bstp_state *bs = &sc->sc_stp;
1643 param->ifbrp_prio = bs->bs_bridge_priority;
1648 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1650 struct ifbrparam *param = arg;
1652 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1656 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1658 struct ifbrparam *param = arg;
1659 struct bstp_state *bs = &sc->sc_stp;
1661 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1666 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1668 struct ifbrparam *param = arg;
1670 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1674 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1676 struct ifbrparam *param = arg;
1677 struct bstp_state *bs = &sc->sc_stp;
1679 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1684 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1686 struct ifbrparam *param = arg;
1688 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1692 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1694 struct ifbrparam *param = arg;
1695 struct bstp_state *bs = &sc->sc_stp;
1697 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1702 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1704 struct ifbrparam *param = arg;
1706 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1710 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1712 struct ifbreq *req = arg;
1713 struct bridge_iflist *bif;
1715 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1719 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1723 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1725 struct ifbreq *req = arg;
1726 struct bridge_iflist *bif;
1728 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1732 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1736 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1738 struct ifbreq *req = arg;
1739 struct bridge_iflist *bif;
1741 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1745 bif->bif_addrmax = req->ifbr_addrmax;
1750 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1752 struct ifbreq *req = arg;
1753 struct bridge_iflist *bif = NULL;
1756 ifs = ifunit(req->ifbr_ifsname);
1760 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1761 if (ifs == bif->bif_ifp)
1764 if (ifs->if_bridge != NULL)
1767 switch (ifs->if_type) {
1776 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1781 bif->bif_flags = IFBIF_SPAN;
1783 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1789 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1791 struct ifbreq *req = arg;
1792 struct bridge_iflist *bif;
1795 ifs = ifunit(req->ifbr_ifsname);
1799 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1800 if (ifs == bif->bif_ifp)
1806 bridge_delete_span(sc, bif);
1812 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1814 struct ifbropreq *req = arg;
1815 struct bstp_state *bs = &sc->sc_stp;
1816 struct bstp_port *root_port;
1818 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1819 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1820 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1822 root_port = bs->bs_root_port;
1823 if (root_port == NULL)
1824 req->ifbop_root_port = 0;
1826 req->ifbop_root_port = root_port->bp_ifp->if_index;
1828 req->ifbop_holdcount = bs->bs_txholdcount;
1829 req->ifbop_priority = bs->bs_bridge_priority;
1830 req->ifbop_protocol = bs->bs_protover;
1831 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1832 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1833 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1834 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1835 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1836 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1842 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1844 struct ifbrparam *param = arg;
1846 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1851 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1853 struct ifbpstpconf *bifstp = arg;
1854 struct bridge_iflist *bif;
1855 struct bstp_port *bp;
1856 struct ifbpstpreq bpreq;
1858 int count, buflen, len, error = 0;
1861 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1862 if ((bif->bif_flags & IFBIF_STP) != 0)
1866 buflen = sizeof(bpreq) * count;
1867 if (bifstp->ifbpstp_len == 0) {
1868 bifstp->ifbpstp_len = buflen;
1872 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1878 len = min(bifstp->ifbpstp_len, buflen);
1879 bzero(&bpreq, sizeof(bpreq));
1880 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1881 if (len < sizeof(bpreq))
1884 if ((bif->bif_flags & IFBIF_STP) == 0)
1888 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1889 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1890 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1891 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1892 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1893 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1895 memcpy(buf, &bpreq, sizeof(bpreq));
1897 buf += sizeof(bpreq);
1898 len -= sizeof(bpreq);
1901 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1902 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1903 free(outbuf, M_TEMP);
1908 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1910 struct ifbrparam *param = arg;
1912 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1916 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1918 struct ifbrparam *param = arg;
1920 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1926 * Detach an interface from a bridge. Called when a member
1927 * interface is detaching.
1930 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1932 struct bridge_softc *sc = ifp->if_bridge;
1933 struct bridge_iflist *bif;
1935 if (ifp->if_flags & IFF_RENAMING)
1937 if (V_bridge_cloner == NULL) {
1939 * This detach handler can be called after
1940 * vnet_bridge_uninit(). Just return in that case.
1944 /* Check if the interface is a bridge member */
1948 bif = bridge_lookup_member_if(sc, ifp);
1950 bridge_delete_member(sc, bif, 1);
1956 /* Check if the interface is a span port */
1958 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1960 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1961 if (ifp == bif->bif_ifp) {
1962 bridge_delete_span(sc, bif);
1968 BRIDGE_LIST_UNLOCK();
1974 * Initialize a bridge interface.
1977 bridge_init(void *xsc)
1979 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1980 struct ifnet *ifp = sc->sc_ifp;
1982 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1986 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1989 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1990 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1998 * Stop the bridge interface.
2001 bridge_stop(struct ifnet *ifp, int disable)
2003 struct bridge_softc *sc = ifp->if_softc;
2005 BRIDGE_LOCK_ASSERT(sc);
2007 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2011 callout_stop(&sc->sc_brcallout);
2013 bstp_stop(&sc->sc_stp);
2015 bridge_rtflush(sc, IFBF_FLUSHDYN);
2016 BRIDGE_RT_UNLOCK(sc);
2018 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2024 * Enqueue a packet on a bridge member interface.
2028 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2034 /* We may be sending a fragment so traverse the mbuf */
2037 m->m_nextpkt = NULL;
2038 len = m->m_pkthdr.len;
2039 mflags = m->m_flags;
2042 * If underlying interface can not do VLAN tag insertion itself
2043 * then attach a packet tag that holds it.
2045 if ((m->m_flags & M_VLANTAG) &&
2046 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2047 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2050 "unable to prepend VLAN header\n");
2051 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2054 m->m_flags &= ~M_VLANTAG;
2057 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2058 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2060 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2064 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2065 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2066 if (mflags & M_MCAST)
2067 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2076 * Receive a queued packet from dummynet and pass it on to the output
2079 * The mbuf has the Ethernet header already attached.
2082 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2084 struct bridge_softc *sc;
2086 sc = ifp->if_bridge;
2089 * The packet didnt originate from a member interface. This should only
2090 * ever happen if a member interface is removed while packets are
2098 if (PFIL_HOOKED(&V_inet_pfil_hook)
2100 || PFIL_HOOKED(&V_inet6_pfil_hook)
2103 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2109 bridge_enqueue(sc, ifp, m);
2115 * Send output from a bridge member interface. This
2116 * performs the bridging function for locally originated
2119 * The mbuf has the Ethernet header already attached. We must
2120 * enqueue or free the mbuf before returning.
2123 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2126 struct epoch_tracker et;
2127 struct ether_header *eh;
2128 struct ifnet *dst_if;
2129 struct bridge_softc *sc;
2132 NET_EPOCH_ENTER_ET(et);
2134 if (m->m_len < ETHER_HDR_LEN) {
2135 m = m_pullup(m, ETHER_HDR_LEN);
2137 NET_EPOCH_EXIT_ET(et);
2142 eh = mtod(m, struct ether_header *);
2143 sc = ifp->if_bridge;
2144 vlan = VLANTAGOF(m);
2147 * If bridge is down, but the original output interface is up,
2148 * go ahead and send out that interface. Otherwise, the packet
2151 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2157 * If the packet is a multicast, or we don't know a better way to
2158 * get there, send to all interfaces.
2160 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2163 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2164 if (dst_if == NULL) {
2165 struct bridge_iflist *bif;
2171 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2172 dst_if = bif->bif_ifp;
2174 if (dst_if->if_type == IFT_GIF)
2176 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2180 * If this is not the original output interface,
2181 * and the interface is participating in spanning
2182 * tree, make sure the port is in a state that
2183 * allows forwarding.
2185 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2186 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2189 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2193 mc = m_copypacket(m, M_NOWAIT);
2195 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2200 bridge_enqueue(sc, dst_if, mc);
2204 NET_EPOCH_EXIT_ET(et);
2210 * XXX Spanning tree consideration here?
2214 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2216 NET_EPOCH_EXIT_ET(et);
2220 bridge_enqueue(sc, dst_if, m);
2221 NET_EPOCH_EXIT_ET(et);
2228 * Do output on a bridge.
2232 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2234 struct epoch_tracker et;
2235 struct bridge_softc *sc;
2236 struct ether_header *eh;
2237 struct ifnet *dst_if;
2240 NET_EPOCH_ENTER_ET(et);
2244 ETHER_BPF_MTAP(ifp, m);
2246 eh = mtod(m, struct ether_header *);
2248 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2249 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2250 error = bridge_enqueue(sc, dst_if, m);
2252 bridge_broadcast(sc, ifp, m, 0);
2254 NET_EPOCH_EXIT_ET(et);
2260 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2263 bridge_qflush(struct ifnet *ifp __unused)
2270 * The forwarding function of the bridge.
2272 * NOTE: Releases the lock on return.
2275 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2278 struct bridge_iflist *dbif;
2279 struct ifnet *src_if, *dst_if, *ifp;
2280 struct ether_header *eh;
2285 MPASS(in_epoch(net_epoch_preempt));
2287 src_if = m->m_pkthdr.rcvif;
2290 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2291 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2292 vlan = VLANTAGOF(m);
2294 if ((sbif->bif_flags & IFBIF_STP) &&
2295 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2298 eh = mtod(m, struct ether_header *);
2299 dst = eh->ether_dhost;
2301 /* If the interface is learning, record the address. */
2302 if (sbif->bif_flags & IFBIF_LEARNING) {
2303 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2304 sbif, 0, IFBAF_DYNAMIC);
2306 * If the interface has addresses limits then deny any source
2307 * that is not in the cache.
2309 if (error && sbif->bif_addrmax)
2313 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2314 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2318 * At this point, the port either doesn't participate
2319 * in spanning tree or it is in the forwarding state.
2323 * If the packet is unicast, destined for someone on
2324 * "this" side of the bridge, drop it.
2326 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2327 dst_if = bridge_rtlookup(sc, dst, vlan);
2328 if (src_if == dst_if)
2332 * Check if its a reserved multicast address, any address
2333 * listed in 802.1D section 7.12.6 may not be forwarded by the
2335 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2337 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2338 dst[2] == 0xc2 && dst[3] == 0x00 &&
2339 dst[4] == 0x00 && dst[5] <= 0x0f)
2342 /* ...forward it to all interfaces. */
2343 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2348 * If we have a destination interface which is a member of our bridge,
2349 * OR this is a unicast packet, push it through the bpf(4) machinery.
2350 * For broadcast or multicast packets, don't bother because it will
2351 * be reinjected into ether_input. We do this before we pass the packets
2352 * through the pfil(9) framework, as it is possible that pfil(9) will
2353 * drop the packet, or possibly modify it, making it difficult to debug
2354 * firewall issues on the bridge.
2356 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2357 ETHER_BPF_MTAP(ifp, m);
2359 /* run the packet filter */
2360 if (PFIL_HOOKED(&V_inet_pfil_hook)
2362 || PFIL_HOOKED(&V_inet6_pfil_hook)
2365 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2371 if (dst_if == NULL) {
2372 bridge_broadcast(sc, src_if, m, 1);
2377 * At this point, we're dealing with a unicast frame
2378 * going to a different interface.
2380 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2383 dbif = bridge_lookup_member_if(sc, dst_if);
2385 /* Not a member of the bridge (anymore?) */
2388 /* Private segments can not talk to each other */
2389 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2392 if ((dbif->bif_flags & IFBIF_STP) &&
2393 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2396 if (PFIL_HOOKED(&V_inet_pfil_hook)
2398 || PFIL_HOOKED(&V_inet6_pfil_hook)
2401 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2407 bridge_enqueue(sc, dst_if, m);
2417 * Receive input from a member interface. Queue the packet for
2418 * bridging if it is not for us.
2420 static struct mbuf *
2421 bridge_input(struct ifnet *ifp, struct mbuf *m)
2423 struct epoch_tracker et;
2424 struct bridge_softc *sc = ifp->if_bridge;
2425 struct bridge_iflist *bif, *bif2;
2427 struct ether_header *eh;
2428 struct mbuf *mc, *mc2;
2432 NET_EPOCH_ENTER_ET(et);
2434 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2435 NET_EPOCH_EXIT_ET(et);
2440 vlan = VLANTAGOF(m);
2443 * Implement support for bridge monitoring. If this flag has been
2444 * set on this interface, discard the packet once we push it through
2445 * the bpf(4) machinery, but before we do, increment the byte and
2446 * packet counters associated with this interface.
2448 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2449 m->m_pkthdr.rcvif = bifp;
2450 ETHER_BPF_MTAP(bifp, m);
2451 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2452 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2454 NET_EPOCH_EXIT_ET(et);
2457 bif = bridge_lookup_member_if(sc, ifp);
2459 NET_EPOCH_EXIT_ET(et);
2463 eh = mtod(m, struct ether_header *);
2467 if (m->m_flags & (M_BCAST|M_MCAST)) {
2468 /* Tap off 802.1D packets; they do not get forwarded. */
2469 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2470 ETHER_ADDR_LEN) == 0) {
2471 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2472 NET_EPOCH_EXIT_ET(et);
2476 if ((bif->bif_flags & IFBIF_STP) &&
2477 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2478 NET_EPOCH_EXIT_ET(et);
2483 * Make a deep copy of the packet and enqueue the copy
2484 * for bridge processing; return the original packet for
2487 mc = m_dup(m, M_NOWAIT);
2489 NET_EPOCH_EXIT_ET(et);
2493 /* Perform the bridge forwarding function with the copy. */
2494 bridge_forward(sc, bif, mc);
2497 * Reinject the mbuf as arriving on the bridge so we have a
2498 * chance at claiming multicast packets. We can not loop back
2499 * here from ether_input as a bridge is never a member of a
2502 KASSERT(bifp->if_bridge == NULL,
2503 ("loop created in bridge_input"));
2504 mc2 = m_dup(m, M_NOWAIT);
2506 /* Keep the layer3 header aligned */
2507 int i = min(mc2->m_pkthdr.len, max_protohdr);
2508 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2511 mc2->m_pkthdr.rcvif = bifp;
2512 (*bifp->if_input)(bifp, mc2);
2515 /* Return the original packet for local processing. */
2516 NET_EPOCH_EXIT_ET(et);
2520 if ((bif->bif_flags & IFBIF_STP) &&
2521 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2522 NET_EPOCH_EXIT_ET(et);
2526 #if (defined(INET) || defined(INET6))
2527 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2528 || ((iface)->if_carp \
2529 && (*carp_forus_p)((iface), eh->ether_dhost))
2530 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2531 || ((iface)->if_carp \
2532 && (*carp_forus_p)((iface), eh->ether_shost))
2534 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2535 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2539 # define OR_PFIL_HOOKED_INET6 \
2540 || PFIL_HOOKED(&V_inet6_pfil_hook)
2542 # define OR_PFIL_HOOKED_INET6
2545 #define GRAB_OUR_PACKETS(iface) \
2546 if ((iface)->if_type == IFT_GIF) \
2548 /* It is destined for us. */ \
2549 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2550 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2552 if ((iface)->if_type == IFT_BRIDGE) { \
2553 ETHER_BPF_MTAP(iface, m); \
2554 if_inc_counter(iface, IFCOUNTER_IPACKETS, 1); \
2555 if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2556 /* Filter on the physical interface. */ \
2557 if (V_pfil_local_phys && \
2558 (PFIL_HOOKED(&V_inet_pfil_hook) \
2559 OR_PFIL_HOOKED_INET6)) { \
2560 if (bridge_pfil(&m, NULL, ifp, \
2561 PFIL_IN) != 0 || m == NULL) { \
2562 NET_EPOCH_EXIT_ET(et); \
2565 eh = mtod(m, struct ether_header *); \
2568 if (bif->bif_flags & IFBIF_LEARNING) { \
2569 error = bridge_rtupdate(sc, eh->ether_shost, \
2570 vlan, bif, 0, IFBAF_DYNAMIC); \
2571 if (error && bif->bif_addrmax) { \
2573 NET_EPOCH_EXIT_ET(et); \
2577 m->m_pkthdr.rcvif = iface; \
2578 NET_EPOCH_EXIT_ET(et); \
2582 /* We just received a packet that we sent out. */ \
2583 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2584 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2587 NET_EPOCH_EXIT_ET(et); \
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 OR_CARP_CHECK_WE_ARE_DST
2611 #undef OR_CARP_CHECK_WE_ARE_SRC
2612 #undef OR_PFIL_HOOKED_INET6
2613 #undef GRAB_OUR_PACKETS
2615 /* Perform the bridge forwarding function. */
2616 bridge_forward(sc, bif, m);
2618 NET_EPOCH_EXIT_ET(et);
2625 * Send a frame to all interfaces that are members of
2626 * the bridge, except for the one on which the packet
2629 * NOTE: Releases the lock on return.
2632 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2633 struct mbuf *m, int runfilt)
2635 struct bridge_iflist *dbif, *sbif;
2637 struct ifnet *dst_if;
2640 MPASS(in_epoch(net_epoch_preempt));
2642 sbif = bridge_lookup_member_if(sc, src_if);
2644 /* Filter on the bridge interface before broadcasting */
2645 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2647 || PFIL_HOOKED(&V_inet6_pfil_hook)
2650 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2656 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2657 dst_if = dbif->bif_ifp;
2658 if (dst_if == src_if)
2661 /* Private segments can not talk to each other */
2662 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2665 if ((dbif->bif_flags & IFBIF_STP) &&
2666 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2669 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2670 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2673 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2676 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2680 mc = m_dup(m, M_NOWAIT);
2682 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2688 * Filter on the output interface. Pass a NULL bridge interface
2689 * pointer so we do not redundantly filter on the bridge for
2690 * each interface we broadcast on.
2692 if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2694 || PFIL_HOOKED(&V_inet6_pfil_hook)
2698 /* Keep the layer3 header aligned */
2699 i = min(mc->m_pkthdr.len, max_protohdr);
2700 mc = m_copyup(mc, i, ETHER_ALIGN);
2702 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2706 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2712 bridge_enqueue(sc, dst_if, mc);
2721 * Duplicate a packet out one or more interfaces that are in span mode,
2722 * the original mbuf is unmodified.
2725 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2727 struct bridge_iflist *bif;
2728 struct ifnet *dst_if;
2731 MPASS(in_epoch(net_epoch_preempt));
2733 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2736 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2737 dst_if = bif->bif_ifp;
2739 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2742 mc = m_copypacket(m, M_NOWAIT);
2744 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2748 bridge_enqueue(sc, dst_if, mc);
2755 * Add a bridge routing entry.
2758 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2759 struct bridge_iflist *bif, int setflags, uint8_t flags)
2761 struct bridge_rtnode *brt;
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 ((brt = 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;
2822 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2823 uma_zfree(V_bridge_rtnode_zone, brt);
2824 BRIDGE_RT_UNLOCK(sc);
2830 BRIDGE_RT_UNLOCK(sc);
2833 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2834 brt->brt_dst != bif) {
2836 brt->brt_dst->bif_addrcnt--;
2838 brt->brt_dst->bif_addrcnt++;
2839 BRIDGE_RT_UNLOCK(sc);
2842 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2843 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2845 brt->brt_flags = flags;
2853 * Lookup the destination interface for an address.
2855 static struct ifnet *
2856 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2858 struct bridge_rtnode *brt;
2860 MPASS(in_epoch(net_epoch_preempt));
2862 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2865 return (brt->brt_ifp);
2871 * Trim the routine table so that we have a number
2872 * of routing entries less than or equal to the
2876 bridge_rttrim(struct bridge_softc *sc)
2878 struct bridge_rtnode *brt, *nbrt;
2880 MPASS(in_epoch(net_epoch_preempt));
2881 BRIDGE_RT_LOCK_ASSERT(sc);
2883 /* Make sure we actually need to do this. */
2884 if (sc->sc_brtcnt <= sc->sc_brtmax)
2887 /* Force an aging cycle; this might trim enough addresses. */
2889 if (sc->sc_brtcnt <= sc->sc_brtmax)
2892 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2893 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2894 bridge_rtnode_destroy(sc, brt);
2895 if (sc->sc_brtcnt <= sc->sc_brtmax)
2904 * Aging timer for the bridge.
2907 bridge_timer(void *arg)
2909 struct bridge_softc *sc = arg;
2911 BRIDGE_RT_LOCK_ASSERT(sc);
2913 /* Destruction of rtnodes requires a proper vnet context */
2914 CURVNET_SET(sc->sc_ifp->if_vnet);
2917 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2918 callout_reset(&sc->sc_brcallout,
2919 bridge_rtable_prune_period * hz, bridge_timer, sc);
2926 * Perform an aging cycle.
2929 bridge_rtage(struct bridge_softc *sc)
2931 struct bridge_rtnode *brt, *nbrt;
2933 BRIDGE_RT_LOCK_ASSERT(sc);
2935 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2936 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2937 if (time_uptime >= brt->brt_expire)
2938 bridge_rtnode_destroy(sc, brt);
2946 * Remove all dynamic addresses from the bridge.
2949 bridge_rtflush(struct bridge_softc *sc, int full)
2951 struct bridge_rtnode *brt, *nbrt;
2953 BRIDGE_RT_LOCK_ASSERT(sc);
2955 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2956 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2957 bridge_rtnode_destroy(sc, brt);
2964 * Remove an address from the table.
2967 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2969 struct bridge_rtnode *brt;
2975 * If vlan is zero then we want to delete for all vlans so the lookup
2976 * may return more than one.
2978 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2979 bridge_rtnode_destroy(sc, brt);
2983 BRIDGE_RT_UNLOCK(sc);
2985 return (found ? 0 : ENOENT);
2991 * Delete routes to a speicifc member interface.
2994 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2996 struct bridge_rtnode *brt, *nbrt;
2998 BRIDGE_RT_LOCK_ASSERT(sc);
3000 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3001 if (brt->brt_ifp == ifp && (full ||
3002 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3003 bridge_rtnode_destroy(sc, brt);
3008 * bridge_rtable_init:
3010 * Initialize the route table for this bridge.
3013 bridge_rtable_init(struct bridge_softc *sc)
3017 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3018 M_DEVBUF, M_WAITOK);
3020 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3021 CK_LIST_INIT(&sc->sc_rthash[i]);
3023 sc->sc_rthash_key = arc4random();
3024 CK_LIST_INIT(&sc->sc_rtlist);
3028 * bridge_rtable_fini:
3030 * Deconstruct the route table for this bridge.
3033 bridge_rtable_fini(struct bridge_softc *sc)
3036 KASSERT(sc->sc_brtcnt == 0,
3037 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3038 free(sc->sc_rthash, M_DEVBUF);
3042 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3043 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3045 #define mix(a, b, c) \
3047 a -= b; a -= c; a ^= (c >> 13); \
3048 b -= c; b -= a; b ^= (a << 8); \
3049 c -= a; c -= b; c ^= (b >> 13); \
3050 a -= b; a -= c; a ^= (c >> 12); \
3051 b -= c; b -= a; b ^= (a << 16); \
3052 c -= a; c -= b; c ^= (b >> 5); \
3053 a -= b; a -= c; a ^= (c >> 3); \
3054 b -= c; b -= a; b ^= (a << 10); \
3055 c -= a; c -= b; c ^= (b >> 15); \
3056 } while (/*CONSTCOND*/0)
3058 static __inline uint32_t
3059 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3061 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3072 return (c & BRIDGE_RTHASH_MASK);
3078 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3082 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3083 d = ((int)a[i]) - ((int)b[i]);
3090 * bridge_rtnode_lookup:
3092 * Look up a bridge route node for the specified destination. Compare the
3093 * vlan id or if zero then just return the first match.
3095 static struct bridge_rtnode *
3096 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3098 struct bridge_rtnode *brt;
3102 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3104 hash = bridge_rthash(sc, addr);
3105 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3106 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3107 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3117 * bridge_rtnode_insert:
3119 * Insert the specified bridge node into the route table. We
3120 * assume the entry is not already in the table.
3123 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3125 struct bridge_rtnode *lbrt;
3129 BRIDGE_RT_LOCK_ASSERT(sc);
3131 hash = bridge_rthash(sc, brt->brt_addr);
3133 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3135 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3140 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3141 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3144 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3147 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3148 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3151 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3152 } while (lbrt != NULL);
3155 panic("bridge_rtnode_insert: impossible");
3159 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3166 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3168 struct bridge_rtnode *brt;
3170 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3172 CURVNET_SET(brt->brt_vnet);
3173 uma_zfree(V_bridge_rtnode_zone, brt);
3178 * bridge_rtnode_destroy:
3180 * Destroy a bridge rtnode.
3183 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3186 BRIDGE_RT_LOCK_ASSERT(sc);
3188 CK_LIST_REMOVE(brt, brt_hash);
3190 CK_LIST_REMOVE(brt, brt_list);
3192 brt->brt_dst->bif_addrcnt--;
3194 epoch_call(net_epoch_preempt, &brt->brt_epoch_ctx,
3195 bridge_rtnode_destroy_cb);
3199 * bridge_rtable_expire:
3201 * Set the expiry time for all routes on an interface.
3204 bridge_rtable_expire(struct ifnet *ifp, int age)
3206 struct bridge_softc *sc = ifp->if_bridge;
3207 struct bridge_rtnode *brt;
3209 CURVNET_SET(ifp->if_vnet);
3213 * If the age is zero then flush, otherwise set all the expiry times to
3214 * age for the interface
3217 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3219 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3220 /* Cap the expiry time to 'age' */
3221 if (brt->brt_ifp == ifp &&
3222 brt->brt_expire > time_uptime + age &&
3223 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3224 brt->brt_expire = time_uptime + age;
3227 BRIDGE_RT_UNLOCK(sc);
3232 * bridge_state_change:
3234 * Callback from the bridgestp code when a port changes states.
3237 bridge_state_change(struct ifnet *ifp, int state)
3239 struct bridge_softc *sc = ifp->if_bridge;
3240 static const char *stpstates[] = {
3249 CURVNET_SET(ifp->if_vnet);
3251 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3252 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3257 * Send bridge packets through pfil if they are one of the types pfil can deal
3258 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3259 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3263 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3265 int snap, error, i, hlen;
3266 struct ether_header *eh1, eh2;
3269 u_int16_t ether_type;
3272 error = -1; /* Default error if not error == 0 */
3275 /* we may return with the IP fields swapped, ensure its not shared */
3276 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3279 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3280 return (0); /* filtering is disabled */
3282 i = min((*mp)->m_pkthdr.len, max_protohdr);
3283 if ((*mp)->m_len < i) {
3284 *mp = m_pullup(*mp, i);
3286 printf("%s: m_pullup failed\n", __func__);
3291 eh1 = mtod(*mp, struct ether_header *);
3292 ether_type = ntohs(eh1->ether_type);
3295 * Check for SNAP/LLC.
3297 if (ether_type < ETHERMTU) {
3298 struct llc *llc2 = (struct llc *)(eh1 + 1);
3300 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3301 llc2->llc_dsap == LLC_SNAP_LSAP &&
3302 llc2->llc_ssap == LLC_SNAP_LSAP &&
3303 llc2->llc_control == LLC_UI) {
3304 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3310 * If we're trying to filter bridge traffic, don't look at anything
3311 * other than IP and ARP traffic. If the filter doesn't understand
3312 * IPv6, don't allow IPv6 through the bridge either. This is lame
3313 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3314 * but of course we don't have an AppleTalk filter to begin with.
3315 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3318 switch (ether_type) {
3320 case ETHERTYPE_REVARP:
3321 if (V_pfil_ipfw_arp == 0)
3322 return (0); /* Automatically pass */
3327 case ETHERTYPE_IPV6:
3332 * Check to see if the user wants to pass non-ip
3333 * packets, these will not be checked by pfil(9) and
3334 * passed unconditionally so the default is to drop.
3340 /* Run the packet through pfil before stripping link headers */
3341 if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3342 dir == PFIL_OUT && ifp != NULL) {
3344 error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, 0,
3347 if (*mp == NULL || error != 0) /* packet consumed by filter */
3351 /* Strip off the Ethernet header and keep a copy. */
3352 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3353 m_adj(*mp, ETHER_HDR_LEN);
3355 /* Strip off snap header, if present */
3357 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3358 m_adj(*mp, sizeof(struct llc));
3362 * Check the IP header for alignment and errors
3364 if (dir == PFIL_IN) {
3365 switch (ether_type) {
3367 error = bridge_ip_checkbasic(mp);
3370 case ETHERTYPE_IPV6:
3371 error = bridge_ip6_checkbasic(mp);
3384 * Run the packet through pfil
3386 switch (ether_type) {
3389 * Run pfil on the member interface and the bridge, both can
3390 * be skipped by clearing pfil_member or pfil_bridge.
3393 * in_if -> bridge_if -> out_if
3395 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3396 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3399 if (*mp == NULL || error != 0) /* filter may consume */
3402 if (V_pfil_member && ifp != NULL)
3403 error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3406 if (*mp == NULL || error != 0) /* filter may consume */
3409 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3410 error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3413 if (*mp == NULL || error != 0) /* filter may consume */
3416 /* check if we need to fragment the packet */
3417 /* bridge_fragment generates a mbuf chain of packets */
3418 /* that already include eth headers */
3419 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3420 i = (*mp)->m_pkthdr.len;
3421 if (i > ifp->if_mtu) {
3422 error = bridge_fragment(ifp, mp, &eh2, snap,
3428 /* Recalculate the ip checksum. */
3429 ip = mtod(*mp, struct ip *);
3430 hlen = ip->ip_hl << 2;
3431 if (hlen < sizeof(struct ip))
3433 if (hlen > (*mp)->m_len) {
3434 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3436 ip = mtod(*mp, struct ip *);
3441 if (hlen == sizeof(struct ip))
3442 ip->ip_sum = in_cksum_hdr(ip);
3444 ip->ip_sum = in_cksum(*mp, hlen);
3448 case ETHERTYPE_IPV6:
3449 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3450 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3453 if (*mp == NULL || error != 0) /* filter may consume */
3456 if (V_pfil_member && ifp != NULL)
3457 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3460 if (*mp == NULL || error != 0) /* filter may consume */
3463 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3464 error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3481 * Finally, put everything back the way it was and return
3484 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3487 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3490 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3493 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3504 * Perform basic checks on header size since
3505 * pfil assumes ip_input has already processed
3506 * it for it. Cut-and-pasted from ip_input.c.
3507 * Given how simple the IPv6 version is,
3508 * does the IPv4 version really need to be
3511 * XXX Should we update ipstat here, or not?
3512 * XXX Right now we update ipstat but not
3516 bridge_ip_checkbasic(struct mbuf **mp)
3518 struct mbuf *m = *mp;
3526 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3527 if ((m = m_copyup(m, sizeof(struct ip),
3528 (max_linkhdr + 3) & ~3)) == NULL) {
3529 /* XXXJRT new stat, please */
3530 KMOD_IPSTAT_INC(ips_toosmall);
3533 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3534 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3535 KMOD_IPSTAT_INC(ips_toosmall);
3539 ip = mtod(m, struct ip *);
3540 if (ip == NULL) goto bad;
3542 if (ip->ip_v != IPVERSION) {
3543 KMOD_IPSTAT_INC(ips_badvers);
3546 hlen = ip->ip_hl << 2;
3547 if (hlen < sizeof(struct ip)) { /* minimum header length */
3548 KMOD_IPSTAT_INC(ips_badhlen);
3551 if (hlen > m->m_len) {
3552 if ((m = m_pullup(m, hlen)) == NULL) {
3553 KMOD_IPSTAT_INC(ips_badhlen);
3556 ip = mtod(m, struct ip *);
3557 if (ip == NULL) goto bad;
3560 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3561 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3563 if (hlen == sizeof(struct ip)) {
3564 sum = in_cksum_hdr(ip);
3566 sum = in_cksum(m, hlen);
3570 KMOD_IPSTAT_INC(ips_badsum);
3574 /* Retrieve the packet length. */
3575 len = ntohs(ip->ip_len);
3578 * Check for additional length bogosity
3581 KMOD_IPSTAT_INC(ips_badlen);
3586 * Check that the amount of data in the buffers
3587 * is as at least much as the IP header would have us expect.
3588 * Drop packet if shorter than we expect.
3590 if (m->m_pkthdr.len < len) {
3591 KMOD_IPSTAT_INC(ips_tooshort);
3595 /* Checks out, proceed */
3606 * Same as above, but for IPv6.
3607 * Cut-and-pasted from ip6_input.c.
3608 * XXX Should we update ip6stat, or not?
3611 bridge_ip6_checkbasic(struct mbuf **mp)
3613 struct mbuf *m = *mp;
3614 struct ip6_hdr *ip6;
3617 * If the IPv6 header is not aligned, slurp it up into a new
3618 * mbuf with space for link headers, in the event we forward
3619 * it. Otherwise, if it is aligned, make sure the entire base
3620 * IPv6 header is in the first mbuf of the chain.
3622 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3623 struct ifnet *inifp = m->m_pkthdr.rcvif;
3624 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3625 (max_linkhdr + 3) & ~3)) == NULL) {
3626 /* XXXJRT new stat, please */
3627 IP6STAT_INC(ip6s_toosmall);
3628 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3631 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3632 struct ifnet *inifp = m->m_pkthdr.rcvif;
3633 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3634 IP6STAT_INC(ip6s_toosmall);
3635 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3640 ip6 = mtod(m, struct ip6_hdr *);
3642 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3643 IP6STAT_INC(ip6s_badvers);
3644 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3648 /* Checks out, proceed */
3661 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3664 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3665 int snap, struct llc *llc)
3667 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3671 if (m->m_len < sizeof(struct ip) &&
3672 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3674 ip = mtod(m, struct ip *);
3676 m->m_pkthdr.csum_flags |= CSUM_IP;
3677 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3682 * Walk the chain and re-add the Ethernet header for
3685 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3686 nextpkt = mcur->m_nextpkt;
3687 mcur->m_nextpkt = NULL;
3689 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3693 mprev->m_nextpkt = nextpkt;
3696 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3699 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3703 mprev->m_nextpkt = nextpkt;
3706 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3709 * The previous two M_PREPEND could have inserted one or two
3710 * mbufs in front so we have to update the previous packet's
3713 mcur->m_nextpkt = nextpkt;
3715 mprev->m_nextpkt = mcur;
3717 /* The first mbuf in the original chain needs to be
3724 KMOD_IPSTAT_INC(ips_fragmented);
3728 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3729 m = mcur->m_nextpkt;
3736 bridge_linkstate(struct ifnet *ifp)
3738 struct bridge_softc *sc = ifp->if_bridge;
3739 struct bridge_iflist *bif;
3740 struct epoch_tracker et;
3742 NET_EPOCH_ENTER_ET(et);
3744 bif = bridge_lookup_member_if(sc, ifp);
3746 NET_EPOCH_EXIT_ET(et);
3749 bridge_linkcheck(sc);
3751 bstp_linkstate(&bif->bif_stp);
3753 NET_EPOCH_EXIT_ET(et);
3757 bridge_linkcheck(struct bridge_softc *sc)
3759 struct bridge_iflist *bif;
3760 int new_link, hasls;
3762 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3764 new_link = LINK_STATE_DOWN;
3766 /* Our link is considered up if at least one of our ports is active */
3767 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3768 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3770 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3771 new_link = LINK_STATE_UP;
3775 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3776 /* If no interfaces support link-state then we default to up */
3777 new_link = LINK_STATE_UP;
3779 if_link_state_change(sc->sc_ifp, new_link);