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 across 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 */
237 struct epoch_context bif_epoch_ctx;
243 struct bridge_rtnode {
244 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
245 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
246 struct bridge_iflist *brt_dst; /* destination if */
247 unsigned long brt_expire; /* expiration time */
248 uint8_t brt_flags; /* address flags */
249 uint8_t brt_addr[ETHER_ADDR_LEN];
250 uint16_t brt_vlan; /* vlan id */
251 struct vnet *brt_vnet;
252 struct epoch_context brt_epoch_ctx;
254 #define brt_ifp brt_dst->bif_ifp
257 * Software state for each bridge.
259 struct bridge_softc {
260 struct ifnet *sc_ifp; /* make this an interface */
261 LIST_ENTRY(bridge_softc) sc_list;
263 struct mtx sc_rt_mtx;
264 uint32_t sc_brtmax; /* max # of addresses */
265 uint32_t sc_brtcnt; /* cur. # of addresses */
266 uint32_t sc_brttimeout; /* rt timeout in seconds */
267 struct callout sc_brcallout; /* bridge callout */
268 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
269 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
270 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
271 uint32_t sc_rthash_key; /* key for hash */
272 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
273 struct bstp_state sc_stp; /* STP state */
274 uint32_t sc_brtexceeded; /* # of cache drops */
275 struct ifnet *sc_ifaddr; /* member mac copied from */
276 struct ether_addr sc_defaddr; /* Default MAC address */
277 struct epoch_context sc_epoch_ctx;
280 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
281 #define V_bridge_list_sx VNET(bridge_list_sx)
282 static eventhandler_tag bridge_detach_cookie;
284 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
286 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
287 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
289 static int bridge_clone_create(struct if_clone *, int, caddr_t);
290 static void bridge_clone_destroy(struct ifnet *);
292 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
293 static void bridge_mutecaps(struct bridge_softc *);
294 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
296 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
297 static void bridge_init(void *);
298 static void bridge_dummynet(struct mbuf *, struct ifnet *);
299 static void bridge_stop(struct ifnet *, int);
300 static int bridge_transmit(struct ifnet *, struct mbuf *);
302 static void bridge_altq_start(if_t);
303 static int bridge_altq_transmit(if_t, struct mbuf *);
305 static void bridge_qflush(struct ifnet *);
306 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
307 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
309 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
311 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
313 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
316 static void bridge_timer(void *);
318 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
320 static void bridge_span(struct bridge_softc *, struct mbuf *);
322 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
323 uint16_t, struct bridge_iflist *, int, uint8_t);
324 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
326 static void bridge_rttrim(struct bridge_softc *);
327 static void bridge_rtage(struct bridge_softc *);
328 static void bridge_rtflush(struct bridge_softc *, int);
329 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
332 static void bridge_rtable_init(struct bridge_softc *);
333 static void bridge_rtable_fini(struct bridge_softc *);
335 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
336 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
337 const uint8_t *, uint16_t);
338 static int bridge_rtnode_insert(struct bridge_softc *,
339 struct bridge_rtnode *);
340 static void bridge_rtnode_destroy(struct bridge_softc *,
341 struct bridge_rtnode *);
342 static void bridge_rtable_expire(struct ifnet *, int);
343 static void bridge_state_change(struct ifnet *, int);
345 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
347 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
349 static void bridge_delete_member(struct bridge_softc *,
350 struct bridge_iflist *, int);
351 static void bridge_delete_span(struct bridge_softc *,
352 struct bridge_iflist *);
354 static int bridge_ioctl_add(struct bridge_softc *, void *);
355 static int bridge_ioctl_del(struct bridge_softc *, void *);
356 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
357 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
358 static int bridge_ioctl_scache(struct bridge_softc *, void *);
359 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
360 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
361 static int bridge_ioctl_rts(struct bridge_softc *, void *);
362 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
363 static int bridge_ioctl_sto(struct bridge_softc *, void *);
364 static int bridge_ioctl_gto(struct bridge_softc *, void *);
365 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
366 static int bridge_ioctl_flush(struct bridge_softc *, void *);
367 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
368 static int bridge_ioctl_spri(struct bridge_softc *, void *);
369 static int bridge_ioctl_ght(struct bridge_softc *, void *);
370 static int bridge_ioctl_sht(struct bridge_softc *, void *);
371 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
372 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
373 static int bridge_ioctl_gma(struct bridge_softc *, void *);
374 static int bridge_ioctl_sma(struct bridge_softc *, void *);
375 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
376 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
377 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
378 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
379 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
380 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
381 static int bridge_ioctl_grte(struct bridge_softc *, void *);
382 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
383 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
384 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
385 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
387 static int bridge_ip_checkbasic(struct mbuf **mp);
389 static int bridge_ip6_checkbasic(struct mbuf **mp);
391 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
392 struct ether_header *, int, struct llc *);
393 static void bridge_linkstate(struct ifnet *ifp);
394 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 | CTLFLAG_MPSAFE, 0,
409 /* only pass IP[46] packets when pfil is enabled */
410 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
411 #define V_pfil_onlyip VNET(pfil_onlyip)
412 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
413 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
414 "Only pass IP packets when pfil is enabled");
416 /* run pfil hooks on the bridge interface */
417 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
418 #define V_pfil_bridge VNET(pfil_bridge)
419 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
420 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
421 "Packet filter on the bridge interface");
423 /* layer2 filter with ipfw */
424 VNET_DEFINE_STATIC(int, pfil_ipfw);
425 #define V_pfil_ipfw VNET(pfil_ipfw)
427 /* layer2 ARP filter with ipfw */
428 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
429 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
430 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
431 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
432 "Filter ARP packets through IPFW layer2");
434 /* run pfil hooks on the member interface */
435 VNET_DEFINE_STATIC(int, pfil_member) = 1;
436 #define V_pfil_member VNET(pfil_member)
437 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
438 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
439 "Packet filter on the member interface");
441 /* run pfil hooks on the physical interface for locally destined packets */
442 VNET_DEFINE_STATIC(int, pfil_local_phys);
443 #define V_pfil_local_phys VNET(pfil_local_phys)
444 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
445 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
446 "Packet filter on the physical interface for locally destined packets");
448 /* log STP state changes */
449 VNET_DEFINE_STATIC(int, log_stp);
450 #define V_log_stp VNET(log_stp)
451 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
452 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
453 "Log STP state changes");
455 /* share MAC with first bridge member */
456 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
457 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
458 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
459 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
460 "Inherit MAC address from the first bridge member");
462 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
463 #define V_allow_llz_overlap VNET(allow_llz_overlap)
464 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
465 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
466 "Allow overlap of link-local scope "
467 "zones of a bridge interface and the member interfaces");
469 struct bridge_control {
470 int (*bc_func)(struct bridge_softc *, void *);
475 #define BC_F_COPYIN 0x01 /* copy arguments in */
476 #define BC_F_COPYOUT 0x02 /* copy arguments out */
477 #define BC_F_SUSER 0x04 /* do super-user check */
479 const struct bridge_control bridge_control_table[] = {
480 { bridge_ioctl_add, sizeof(struct ifbreq),
481 BC_F_COPYIN|BC_F_SUSER },
482 { bridge_ioctl_del, sizeof(struct ifbreq),
483 BC_F_COPYIN|BC_F_SUSER },
485 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
486 BC_F_COPYIN|BC_F_COPYOUT },
487 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
488 BC_F_COPYIN|BC_F_SUSER },
490 { bridge_ioctl_scache, sizeof(struct ifbrparam),
491 BC_F_COPYIN|BC_F_SUSER },
492 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
495 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
496 BC_F_COPYIN|BC_F_COPYOUT },
497 { bridge_ioctl_rts, sizeof(struct ifbaconf),
498 BC_F_COPYIN|BC_F_COPYOUT },
500 { bridge_ioctl_saddr, sizeof(struct ifbareq),
501 BC_F_COPYIN|BC_F_SUSER },
503 { bridge_ioctl_sto, sizeof(struct ifbrparam),
504 BC_F_COPYIN|BC_F_SUSER },
505 { bridge_ioctl_gto, sizeof(struct ifbrparam),
508 { bridge_ioctl_daddr, sizeof(struct ifbareq),
509 BC_F_COPYIN|BC_F_SUSER },
511 { bridge_ioctl_flush, sizeof(struct ifbreq),
512 BC_F_COPYIN|BC_F_SUSER },
514 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
516 { bridge_ioctl_spri, sizeof(struct ifbrparam),
517 BC_F_COPYIN|BC_F_SUSER },
519 { bridge_ioctl_ght, sizeof(struct ifbrparam),
521 { bridge_ioctl_sht, sizeof(struct ifbrparam),
522 BC_F_COPYIN|BC_F_SUSER },
524 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
526 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
527 BC_F_COPYIN|BC_F_SUSER },
529 { bridge_ioctl_gma, sizeof(struct ifbrparam),
531 { bridge_ioctl_sma, sizeof(struct ifbrparam),
532 BC_F_COPYIN|BC_F_SUSER },
534 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
535 BC_F_COPYIN|BC_F_SUSER },
537 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
538 BC_F_COPYIN|BC_F_SUSER },
540 { bridge_ioctl_addspan, sizeof(struct ifbreq),
541 BC_F_COPYIN|BC_F_SUSER },
542 { bridge_ioctl_delspan, sizeof(struct ifbreq),
543 BC_F_COPYIN|BC_F_SUSER },
545 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
548 { bridge_ioctl_grte, sizeof(struct ifbrparam),
551 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
552 BC_F_COPYIN|BC_F_COPYOUT },
554 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
555 BC_F_COPYIN|BC_F_SUSER },
557 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
558 BC_F_COPYIN|BC_F_SUSER },
560 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
561 BC_F_COPYIN|BC_F_SUSER },
564 const int bridge_control_table_size = nitems(bridge_control_table);
566 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
567 #define V_bridge_list VNET(bridge_list)
568 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \
570 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx)
571 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx)
572 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx)
574 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
575 #define V_bridge_cloner VNET(bridge_cloner)
577 static const char bridge_name[] = "bridge";
580 vnet_bridge_init(const void *unused __unused)
583 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
584 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
586 BRIDGE_LIST_LOCK_INIT();
587 LIST_INIT(&V_bridge_list);
588 V_bridge_cloner = if_clone_simple(bridge_name,
589 bridge_clone_create, bridge_clone_destroy, 0);
591 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
592 vnet_bridge_init, NULL);
595 vnet_bridge_uninit(const void *unused __unused)
598 if_clone_detach(V_bridge_cloner);
599 V_bridge_cloner = NULL;
600 BRIDGE_LIST_LOCK_DESTROY();
602 /* Callbacks may use the UMA zone. */
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 | CTLFLAG_NEEDGIANT,
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;
710 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
711 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
717 BRIDGE_LOCK_INIT(sc);
718 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
719 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
721 /* Initialize our routing table. */
722 bridge_rtable_init(sc);
724 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
726 CK_LIST_INIT(&sc->sc_iflist);
727 CK_LIST_INIT(&sc->sc_spanlist);
730 if_initname(ifp, bridge_name, unit);
731 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
732 ifp->if_ioctl = bridge_ioctl;
734 ifp->if_start = bridge_altq_start;
735 ifp->if_transmit = bridge_altq_transmit;
736 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
737 ifp->if_snd.ifq_drv_maxlen = 0;
738 IFQ_SET_READY(&ifp->if_snd);
740 ifp->if_transmit = bridge_transmit;
742 ifp->if_qflush = bridge_qflush;
743 ifp->if_init = bridge_init;
744 ifp->if_type = IFT_BRIDGE;
746 ether_gen_addr(ifp, &sc->sc_defaddr);
748 bstp_attach(&sc->sc_stp, &bridge_ops);
749 ether_ifattach(ifp, sc->sc_defaddr.octet);
750 /* Now undo some of the damage... */
751 ifp->if_baudrate = 0;
752 ifp->if_type = IFT_BRIDGE;
754 ifp->if_reassign = bridge_reassign;
758 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
759 BRIDGE_LIST_UNLOCK();
765 bridge_clone_destroy_cb(struct epoch_context *ctx)
767 struct bridge_softc *sc;
769 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
771 BRIDGE_LOCK_DESTROY(sc);
776 * bridge_clone_destroy:
778 * Destroy a bridge instance.
781 bridge_clone_destroy(struct ifnet *ifp)
783 struct bridge_softc *sc = ifp->if_softc;
784 struct bridge_iflist *bif;
785 struct epoch_tracker et;
790 ifp->if_flags &= ~IFF_UP;
792 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
793 bridge_delete_member(sc, bif, 0);
795 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
796 bridge_delete_span(sc, bif);
799 /* Tear down the routing table. */
800 bridge_rtable_fini(sc);
806 callout_drain(&sc->sc_brcallout);
809 LIST_REMOVE(sc, sc_list);
810 BRIDGE_LIST_UNLOCK();
812 bstp_detach(&sc->sc_stp);
814 IFQ_PURGE(&ifp->if_snd);
821 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
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;
857 if (ifd->ifd_cmd >= bridge_control_table_size) {
861 bc = &bridge_control_table[ifd->ifd_cmd];
863 if (cmd == SIOCGDRVSPEC &&
864 (bc->bc_flags & BC_F_COPYOUT) == 0) {
868 else if (cmd == SIOCSDRVSPEC &&
869 (bc->bc_flags & BC_F_COPYOUT) != 0) {
874 if (bc->bc_flags & BC_F_SUSER) {
875 error = priv_check(td, PRIV_NET_BRIDGE);
880 if (ifd->ifd_len != bc->bc_argsize ||
881 ifd->ifd_len > sizeof(args)) {
886 bzero(&args, sizeof(args));
887 if (bc->bc_flags & BC_F_COPYIN) {
888 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
893 oldmtu = ifp->if_mtu;
894 error = (*bc->bc_func)(sc, &args);
899 * Bridge MTU may change during addition of the first port.
900 * If it did, do network layer specific procedure.
902 if (ifp->if_mtu != oldmtu) {
909 if (bc->bc_flags & BC_F_COPYOUT)
910 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
915 if (!(ifp->if_flags & IFF_UP) &&
916 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
918 * If interface is marked down and it is running,
919 * then stop and disable it.
922 } else if ((ifp->if_flags & IFF_UP) &&
923 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
925 * If interface is marked up and it is stopped, then
935 if (ifr->ifr_mtu < 576) {
939 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
940 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
943 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
944 if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
945 log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
946 " != %d\n", sc->sc_ifp->if_xname,
947 bif->bif_ifp->if_mtu,
948 bif->bif_ifp->if_xname, ifr->ifr_mtu);
954 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
958 * drop the lock as ether_ioctl() will call bridge_start() and
959 * cause the lock to be recursed.
962 error = ether_ioctl(ifp, cmd, data);
975 * Clear or restore unwanted capabilities on the member interface
978 bridge_mutecaps(struct bridge_softc *sc)
980 struct bridge_iflist *bif;
983 BRIDGE_LOCK_ASSERT(sc);
985 /* Initial bitmask of capabilities to test */
986 mask = BRIDGE_IFCAPS_MASK;
988 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
989 /* Every member must support it or its disabled */
990 mask &= bif->bif_savedcaps;
993 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
994 enabled = bif->bif_ifp->if_capenable;
995 enabled &= ~BRIDGE_IFCAPS_STRIP;
996 /* strip off mask bits and enable them again if allowed */
997 enabled &= ~BRIDGE_IFCAPS_MASK;
999 bridge_set_ifcap(sc, bif, enabled);
1004 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1006 struct ifnet *ifp = bif->bif_ifp;
1008 int error, mask, stuck;
1010 bzero(&ifr, sizeof(ifr));
1011 ifr.ifr_reqcap = set;
1013 if (ifp->if_capenable != set) {
1014 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1016 if_printf(sc->sc_ifp,
1017 "error setting capabilities on %s: %d\n",
1018 ifp->if_xname, error);
1019 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1020 stuck = ifp->if_capenable & mask & ~set;
1022 if_printf(sc->sc_ifp,
1023 "can't disable some capabilities on %s: 0x%x\n",
1024 ifp->if_xname, stuck);
1029 * bridge_lookup_member:
1031 * Lookup a bridge member interface.
1033 static struct bridge_iflist *
1034 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1036 struct bridge_iflist *bif;
1039 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1041 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1043 if (strcmp(ifp->if_xname, name) == 0)
1051 * bridge_lookup_member_if:
1053 * Lookup a bridge member interface by ifnet*.
1055 static struct bridge_iflist *
1056 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1058 struct bridge_iflist *bif;
1060 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1062 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1063 if (bif->bif_ifp == member_ifp)
1071 bridge_delete_member_cb(struct epoch_context *ctx)
1073 struct bridge_iflist *bif;
1075 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1077 free(bif, M_DEVBUF);
1081 * bridge_delete_member:
1083 * Delete the specified member interface.
1086 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1089 struct ifnet *ifs = bif->bif_ifp;
1090 struct ifnet *fif = NULL;
1091 struct bridge_iflist *bifl;
1093 BRIDGE_LOCK_ASSERT(sc);
1095 if (bif->bif_flags & IFBIF_STP)
1096 bstp_disable(&bif->bif_stp);
1098 ifs->if_bridge = NULL;
1099 CK_LIST_REMOVE(bif, bif_next);
1102 * If removing the interface that gave the bridge its mac address, set
1103 * the mac address of the bridge to the address of the next member, or
1104 * to its default address if no members are left.
1106 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1107 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1108 bcopy(&sc->sc_defaddr,
1109 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1110 sc->sc_ifaddr = NULL;
1112 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1113 fif = bifl->bif_ifp;
1114 bcopy(IF_LLADDR(fif),
1115 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1116 sc->sc_ifaddr = fif;
1118 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1121 bridge_linkcheck(sc);
1122 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1124 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1125 BRIDGE_RT_UNLOCK(sc);
1126 KASSERT(bif->bif_addrcnt == 0,
1127 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1129 ifs->if_bridge_output = NULL;
1130 ifs->if_bridge_input = NULL;
1131 ifs->if_bridge_linkstate = NULL;
1133 switch (ifs->if_type) {
1137 * Take the interface out of promiscuous mode, but only
1138 * if it was promiscuous in the first place. It might
1139 * not be if we're in the bridge_ioctl_add() error path.
1141 if (ifs->if_flags & IFF_PROMISC)
1142 (void) ifpromisc(ifs, 0);
1150 panic("bridge_delete_member: impossible");
1154 /* reneable any interface capabilities */
1155 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1157 bstp_destroy(&bif->bif_stp); /* prepare to free */
1159 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1163 * bridge_delete_span:
1165 * Delete the specified span interface.
1168 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1170 BRIDGE_LOCK_ASSERT(sc);
1172 KASSERT(bif->bif_ifp->if_bridge == NULL,
1173 ("%s: not a span interface", __func__));
1175 CK_LIST_REMOVE(bif, bif_next);
1177 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1181 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1183 struct ifbreq *req = arg;
1184 struct bridge_iflist *bif = NULL;
1188 ifs = ifunit(req->ifbr_ifsname);
1191 if (ifs->if_ioctl == NULL) /* must be supported */
1194 /* If it's in the span list, it can't be a member. */
1195 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1196 if (ifs == bif->bif_ifp)
1199 if (ifs->if_bridge == sc)
1202 if (ifs->if_bridge != NULL)
1205 switch (ifs->if_type) {
1209 /* permitted interface types */
1217 * Two valid inet6 addresses with link-local scope must not be
1218 * on the parent interface and the member interfaces at the
1219 * same time. This restriction is needed to prevent violation
1220 * of link-local scope zone. Attempts to add a member
1221 * interface which has inet6 addresses when the parent has
1222 * inet6 triggers removal of all inet6 addresses on the member
1226 /* Check if the parent interface has a link-local scope addr. */
1227 if (V_allow_llz_overlap == 0 &&
1228 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1230 * If any, remove all inet6 addresses from the member
1233 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1234 if (in6ifa_llaonifp(bif->bif_ifp)) {
1235 in6_ifdetach(bif->bif_ifp);
1236 if_printf(sc->sc_ifp,
1237 "IPv6 addresses on %s have been removed "
1238 "before adding it as a member to prevent "
1239 "IPv6 address scope violation.\n",
1240 bif->bif_ifp->if_xname);
1243 if (in6ifa_llaonifp(ifs)) {
1245 if_printf(sc->sc_ifp,
1246 "IPv6 addresses on %s have been removed "
1247 "before adding it as a member to prevent "
1248 "IPv6 address scope violation.\n",
1253 /* Allow the first Ethernet member to define the MTU */
1254 if (CK_LIST_EMPTY(&sc->sc_iflist))
1255 sc->sc_ifp->if_mtu = ifs->if_mtu;
1256 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1257 if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1258 ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1262 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1267 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1268 bif->bif_savedcaps = ifs->if_capenable;
1271 * Assign the interface's MAC address to the bridge if it's the first
1272 * member and the MAC address of the bridge has not been changed from
1273 * the default randomly generated one.
1275 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1276 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1277 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1278 sc->sc_ifaddr = ifs;
1279 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1282 ifs->if_bridge = sc;
1283 ifs->if_bridge_output = bridge_output;
1284 ifs->if_bridge_input = bridge_input;
1285 ifs->if_bridge_linkstate = bridge_linkstate;
1286 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1288 * XXX: XLOCK HERE!?!
1290 * NOTE: insert_***HEAD*** should be safe for the traversals.
1292 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1294 /* Set interface capabilities to the intersection set of all members */
1295 bridge_mutecaps(sc);
1296 bridge_linkcheck(sc);
1298 /* Place the interface into promiscuous mode */
1299 switch (ifs->if_type) {
1302 error = ifpromisc(ifs, 1);
1307 bridge_delete_member(sc, bif, 0);
1312 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1314 struct ifbreq *req = arg;
1315 struct bridge_iflist *bif;
1317 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1321 bridge_delete_member(sc, bif, 0);
1327 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1329 struct ifbreq *req = arg;
1330 struct bridge_iflist *bif;
1331 struct bstp_port *bp;
1333 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1338 req->ifbr_ifsflags = bif->bif_flags;
1339 req->ifbr_state = bp->bp_state;
1340 req->ifbr_priority = bp->bp_priority;
1341 req->ifbr_path_cost = bp->bp_path_cost;
1342 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1343 req->ifbr_proto = bp->bp_protover;
1344 req->ifbr_role = bp->bp_role;
1345 req->ifbr_stpflags = bp->bp_flags;
1346 req->ifbr_addrcnt = bif->bif_addrcnt;
1347 req->ifbr_addrmax = bif->bif_addrmax;
1348 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1350 /* Copy STP state options as flags */
1351 if (bp->bp_operedge)
1352 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1353 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1354 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1355 if (bp->bp_ptp_link)
1356 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1357 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1358 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1359 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1360 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1361 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1362 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1367 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1369 struct epoch_tracker et;
1370 struct ifbreq *req = arg;
1371 struct bridge_iflist *bif;
1372 struct bstp_port *bp;
1375 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1380 if (req->ifbr_ifsflags & IFBIF_SPAN)
1381 /* SPAN is readonly */
1384 NET_EPOCH_ENTER(et);
1386 if (req->ifbr_ifsflags & IFBIF_STP) {
1387 if ((bif->bif_flags & IFBIF_STP) == 0) {
1388 error = bstp_enable(&bif->bif_stp);
1395 if ((bif->bif_flags & IFBIF_STP) != 0)
1396 bstp_disable(&bif->bif_stp);
1399 /* Pass on STP flags */
1400 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1401 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1402 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1403 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1405 /* Save the bits relating to the bridge */
1406 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1414 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1416 struct ifbrparam *param = arg;
1418 sc->sc_brtmax = param->ifbrp_csize;
1425 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1427 struct ifbrparam *param = arg;
1429 param->ifbrp_csize = sc->sc_brtmax;
1435 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1437 struct ifbifconf *bifc = arg;
1438 struct bridge_iflist *bif;
1441 int count, buflen, len, error = 0;
1444 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1446 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1449 buflen = sizeof(breq) * count;
1450 if (bifc->ifbic_len == 0) {
1451 bifc->ifbic_len = buflen;
1454 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1460 len = min(bifc->ifbic_len, buflen);
1461 bzero(&breq, sizeof(breq));
1462 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1463 if (len < sizeof(breq))
1466 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1467 sizeof(breq.ifbr_ifsname));
1468 /* Fill in the ifbreq structure */
1469 error = bridge_ioctl_gifflags(sc, &breq);
1472 memcpy(buf, &breq, sizeof(breq));
1474 buf += sizeof(breq);
1475 len -= sizeof(breq);
1477 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1478 if (len < sizeof(breq))
1481 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1482 sizeof(breq.ifbr_ifsname));
1483 breq.ifbr_ifsflags = bif->bif_flags;
1484 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1485 memcpy(buf, &breq, sizeof(breq));
1487 buf += sizeof(breq);
1488 len -= sizeof(breq);
1491 bifc->ifbic_len = sizeof(breq) * count;
1492 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1493 free(outbuf, M_TEMP);
1498 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1500 struct ifbaconf *bac = arg;
1501 struct bridge_rtnode *brt;
1502 struct ifbareq bareq;
1504 int count, buflen, len, error = 0;
1506 if (bac->ifbac_len == 0)
1510 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1512 buflen = sizeof(bareq) * count;
1514 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1520 len = min(bac->ifbac_len, buflen);
1521 bzero(&bareq, sizeof(bareq));
1522 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1523 if (len < sizeof(bareq))
1525 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1526 sizeof(bareq.ifba_ifsname));
1527 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1528 bareq.ifba_vlan = brt->brt_vlan;
1529 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1530 time_uptime < brt->brt_expire)
1531 bareq.ifba_expire = brt->brt_expire - time_uptime;
1533 bareq.ifba_expire = 0;
1534 bareq.ifba_flags = brt->brt_flags;
1536 memcpy(buf, &bareq, sizeof(bareq));
1538 buf += sizeof(bareq);
1539 len -= sizeof(bareq);
1542 bac->ifbac_len = sizeof(bareq) * count;
1543 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1544 free(outbuf, M_TEMP);
1549 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1551 struct ifbareq *req = arg;
1552 struct bridge_iflist *bif;
1553 struct epoch_tracker et;
1556 NET_EPOCH_ENTER(et);
1557 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1563 /* bridge_rtupdate() may acquire the lock. */
1564 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1572 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1574 struct ifbrparam *param = arg;
1576 sc->sc_brttimeout = param->ifbrp_ctime;
1581 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1583 struct ifbrparam *param = arg;
1585 param->ifbrp_ctime = sc->sc_brttimeout;
1590 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1592 struct ifbareq *req = arg;
1594 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1598 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1600 struct ifbreq *req = arg;
1603 bridge_rtflush(sc, req->ifbr_ifsflags);
1604 BRIDGE_RT_UNLOCK(sc);
1610 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1612 struct ifbrparam *param = arg;
1613 struct bstp_state *bs = &sc->sc_stp;
1615 param->ifbrp_prio = bs->bs_bridge_priority;
1620 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1622 struct ifbrparam *param = arg;
1624 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1628 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1630 struct ifbrparam *param = arg;
1631 struct bstp_state *bs = &sc->sc_stp;
1633 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1638 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1640 struct ifbrparam *param = arg;
1642 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1646 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1648 struct ifbrparam *param = arg;
1649 struct bstp_state *bs = &sc->sc_stp;
1651 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1656 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1658 struct ifbrparam *param = arg;
1660 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1664 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1666 struct ifbrparam *param = arg;
1667 struct bstp_state *bs = &sc->sc_stp;
1669 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1674 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1676 struct ifbrparam *param = arg;
1678 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1682 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1684 struct ifbreq *req = arg;
1685 struct bridge_iflist *bif;
1687 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1691 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1695 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1697 struct ifbreq *req = arg;
1698 struct bridge_iflist *bif;
1700 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1704 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1708 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1710 struct ifbreq *req = arg;
1711 struct bridge_iflist *bif;
1713 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1717 bif->bif_addrmax = req->ifbr_addrmax;
1722 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1724 struct ifbreq *req = arg;
1725 struct bridge_iflist *bif = NULL;
1728 ifs = ifunit(req->ifbr_ifsname);
1732 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1733 if (ifs == bif->bif_ifp)
1736 if (ifs->if_bridge != NULL)
1739 switch (ifs->if_type) {
1748 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1753 bif->bif_flags = IFBIF_SPAN;
1755 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1761 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1763 struct ifbreq *req = arg;
1764 struct bridge_iflist *bif;
1767 ifs = ifunit(req->ifbr_ifsname);
1771 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1772 if (ifs == bif->bif_ifp)
1778 bridge_delete_span(sc, bif);
1784 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1786 struct ifbropreq *req = arg;
1787 struct bstp_state *bs = &sc->sc_stp;
1788 struct bstp_port *root_port;
1790 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1791 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1792 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1794 root_port = bs->bs_root_port;
1795 if (root_port == NULL)
1796 req->ifbop_root_port = 0;
1798 req->ifbop_root_port = root_port->bp_ifp->if_index;
1800 req->ifbop_holdcount = bs->bs_txholdcount;
1801 req->ifbop_priority = bs->bs_bridge_priority;
1802 req->ifbop_protocol = bs->bs_protover;
1803 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1804 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1805 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1806 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1807 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1808 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1814 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1816 struct ifbrparam *param = arg;
1818 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1823 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1825 struct ifbpstpconf *bifstp = arg;
1826 struct bridge_iflist *bif;
1827 struct bstp_port *bp;
1828 struct ifbpstpreq bpreq;
1830 int count, buflen, len, error = 0;
1833 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1834 if ((bif->bif_flags & IFBIF_STP) != 0)
1838 buflen = sizeof(bpreq) * count;
1839 if (bifstp->ifbpstp_len == 0) {
1840 bifstp->ifbpstp_len = buflen;
1844 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1850 len = min(bifstp->ifbpstp_len, buflen);
1851 bzero(&bpreq, sizeof(bpreq));
1852 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1853 if (len < sizeof(bpreq))
1856 if ((bif->bif_flags & IFBIF_STP) == 0)
1860 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1861 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1862 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1863 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1864 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1865 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1867 memcpy(buf, &bpreq, sizeof(bpreq));
1869 buf += sizeof(bpreq);
1870 len -= sizeof(bpreq);
1873 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1874 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1875 free(outbuf, M_TEMP);
1880 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1882 struct ifbrparam *param = arg;
1884 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1888 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1890 struct ifbrparam *param = arg;
1892 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1898 * Detach an interface from a bridge. Called when a member
1899 * interface is detaching.
1902 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1904 struct bridge_softc *sc = ifp->if_bridge;
1905 struct bridge_iflist *bif;
1907 if (ifp->if_flags & IFF_RENAMING)
1909 if (V_bridge_cloner == NULL) {
1911 * This detach handler can be called after
1912 * vnet_bridge_uninit(). Just return in that case.
1916 /* Check if the interface is a bridge member */
1920 bif = bridge_lookup_member_if(sc, ifp);
1922 bridge_delete_member(sc, bif, 1);
1928 /* Check if the interface is a span port */
1930 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1932 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1933 if (ifp == bif->bif_ifp) {
1934 bridge_delete_span(sc, bif);
1940 BRIDGE_LIST_UNLOCK();
1946 * Initialize a bridge interface.
1949 bridge_init(void *xsc)
1951 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1952 struct ifnet *ifp = sc->sc_ifp;
1954 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1958 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1961 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1962 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1970 * Stop the bridge interface.
1973 bridge_stop(struct ifnet *ifp, int disable)
1975 struct bridge_softc *sc = ifp->if_softc;
1977 BRIDGE_LOCK_ASSERT(sc);
1979 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1983 callout_stop(&sc->sc_brcallout);
1985 bstp_stop(&sc->sc_stp);
1987 bridge_rtflush(sc, IFBF_FLUSHDYN);
1988 BRIDGE_RT_UNLOCK(sc);
1990 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1996 * Enqueue a packet on a bridge member interface.
2000 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2006 /* We may be sending a fragment so traverse the mbuf */
2009 m->m_nextpkt = NULL;
2010 len = m->m_pkthdr.len;
2011 mflags = m->m_flags;
2014 * If underlying interface can not do VLAN tag insertion itself
2015 * then attach a packet tag that holds it.
2017 if ((m->m_flags & M_VLANTAG) &&
2018 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2019 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2022 "unable to prepend VLAN header\n");
2023 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2026 m->m_flags &= ~M_VLANTAG;
2029 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2030 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2032 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2036 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2037 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2038 if (mflags & M_MCAST)
2039 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2048 * Receive a queued packet from dummynet and pass it on to the output
2051 * The mbuf has the Ethernet header already attached.
2054 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2056 struct bridge_softc *sc;
2058 sc = ifp->if_bridge;
2061 * The packet didnt originate from a member interface. This should only
2062 * ever happen if a member interface is removed while packets are
2070 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2072 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2075 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2081 bridge_enqueue(sc, ifp, m);
2087 * Send output from a bridge member interface. This
2088 * performs the bridging function for locally originated
2091 * The mbuf has the Ethernet header already attached. We must
2092 * enqueue or free the mbuf before returning.
2095 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2098 struct ether_header *eh;
2099 struct ifnet *bifp, *dst_if;
2100 struct bridge_softc *sc;
2105 if (m->m_len < ETHER_HDR_LEN) {
2106 m = m_pullup(m, ETHER_HDR_LEN);
2111 eh = mtod(m, struct ether_header *);
2112 sc = ifp->if_bridge;
2113 vlan = VLANTAGOF(m);
2118 * If bridge is down, but the original output interface is up,
2119 * go ahead and send out that interface. Otherwise, the packet
2122 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2128 * If the packet is a multicast, or we don't know a better way to
2129 * get there, send to all interfaces.
2131 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2134 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2135 /* Tap any traffic not passing back out the originating interface */
2137 ETHER_BPF_MTAP(bifp, m);
2138 if (dst_if == NULL) {
2139 struct bridge_iflist *bif;
2145 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2146 dst_if = bif->bif_ifp;
2148 if (dst_if->if_type == IFT_GIF)
2150 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2154 * If this is not the original output interface,
2155 * and the interface is participating in spanning
2156 * tree, make sure the port is in a state that
2157 * allows forwarding.
2159 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2160 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2163 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2167 mc = m_dup(m, M_NOWAIT);
2169 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2174 bridge_enqueue(sc, dst_if, mc);
2183 * XXX Spanning tree consideration here?
2187 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2192 bridge_enqueue(sc, dst_if, m);
2199 * Do output on a bridge.
2203 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2205 struct bridge_softc *sc;
2206 struct ether_header *eh;
2207 struct ifnet *dst_if;
2212 ETHER_BPF_MTAP(ifp, m);
2214 eh = mtod(m, struct ether_header *);
2216 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2217 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2218 error = bridge_enqueue(sc, dst_if, m);
2220 bridge_broadcast(sc, ifp, m, 0);
2227 bridge_altq_start(if_t ifp)
2229 struct ifaltq *ifq = &ifp->if_snd;
2233 IFQ_DEQUEUE_NOLOCK(ifq, m);
2235 bridge_transmit(ifp, m);
2236 IFQ_DEQUEUE_NOLOCK(ifq, m);
2242 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2246 if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2247 IFQ_ENQUEUE(&ifp->if_snd, m, err);
2249 bridge_altq_start(ifp);
2251 err = bridge_transmit(ifp, m);
2258 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2261 bridge_qflush(struct ifnet *ifp __unused)
2268 * The forwarding function of the bridge.
2270 * NOTE: Releases the lock on return.
2273 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2276 struct bridge_iflist *dbif;
2277 struct ifnet *src_if, *dst_if, *ifp;
2278 struct ether_header *eh;
2285 src_if = m->m_pkthdr.rcvif;
2288 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2289 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2290 vlan = VLANTAGOF(m);
2292 if ((sbif->bif_flags & IFBIF_STP) &&
2293 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2296 eh = mtod(m, struct ether_header *);
2297 dst = eh->ether_dhost;
2299 /* If the interface is learning, record the address. */
2300 if (sbif->bif_flags & IFBIF_LEARNING) {
2301 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2302 sbif, 0, IFBAF_DYNAMIC);
2304 * If the interface has addresses limits then deny any source
2305 * that is not in the cache.
2307 if (error && sbif->bif_addrmax)
2311 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2312 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2316 * At this point, the port either doesn't participate
2317 * in spanning tree or it is in the forwarding state.
2321 * If the packet is unicast, destined for someone on
2322 * "this" side of the bridge, drop it.
2324 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2325 dst_if = bridge_rtlookup(sc, dst, vlan);
2326 if (src_if == dst_if)
2330 * Check if its a reserved multicast address, any address
2331 * listed in 802.1D section 7.12.6 may not be forwarded by the
2333 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2335 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2336 dst[2] == 0xc2 && dst[3] == 0x00 &&
2337 dst[4] == 0x00 && dst[5] <= 0x0f)
2340 /* ...forward it to all interfaces. */
2341 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2346 * If we have a destination interface which is a member of our bridge,
2347 * OR this is a unicast packet, push it through the bpf(4) machinery.
2348 * For broadcast or multicast packets, don't bother because it will
2349 * be reinjected into ether_input. We do this before we pass the packets
2350 * through the pfil(9) framework, as it is possible that pfil(9) will
2351 * drop the packet, or possibly modify it, making it difficult to debug
2352 * firewall issues on the bridge.
2354 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2355 ETHER_BPF_MTAP(ifp, m);
2357 /* run the packet filter */
2358 if (PFIL_HOOKED_IN(V_inet_pfil_head)
2360 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2363 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2369 if (dst_if == NULL) {
2370 bridge_broadcast(sc, src_if, m, 1);
2375 * At this point, we're dealing with a unicast frame
2376 * going to a different interface.
2378 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2381 dbif = bridge_lookup_member_if(sc, dst_if);
2383 /* Not a member of the bridge (anymore?) */
2386 /* Private segments can not talk to each other */
2387 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2390 if ((dbif->bif_flags & IFBIF_STP) &&
2391 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2394 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2396 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2399 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2405 bridge_enqueue(sc, dst_if, m);
2415 * Receive input from a member interface. Queue the packet for
2416 * bridging if it is not for us.
2418 static struct mbuf *
2419 bridge_input(struct ifnet *ifp, struct mbuf *m)
2421 struct bridge_softc *sc = ifp->if_bridge;
2422 struct bridge_iflist *bif, *bif2;
2424 struct ether_header *eh;
2425 struct mbuf *mc, *mc2;
2431 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2435 vlan = VLANTAGOF(m);
2438 * Implement support for bridge monitoring. If this flag has been
2439 * set on this interface, discard the packet once we push it through
2440 * the bpf(4) machinery, but before we do, increment the byte and
2441 * packet counters associated with this interface.
2443 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2444 m->m_pkthdr.rcvif = bifp;
2445 ETHER_BPF_MTAP(bifp, m);
2446 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2447 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2451 bif = bridge_lookup_member_if(sc, ifp);
2456 eh = mtod(m, struct ether_header *);
2460 if (m->m_flags & (M_BCAST|M_MCAST)) {
2461 /* Tap off 802.1D packets; they do not get forwarded. */
2462 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2463 ETHER_ADDR_LEN) == 0) {
2464 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2468 if ((bif->bif_flags & IFBIF_STP) &&
2469 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2474 * Make a deep copy of the packet and enqueue the copy
2475 * for bridge processing; return the original packet for
2478 mc = m_dup(m, M_NOWAIT);
2483 /* Perform the bridge forwarding function with the copy. */
2484 bridge_forward(sc, bif, mc);
2487 * Reinject the mbuf as arriving on the bridge so we have a
2488 * chance at claiming multicast packets. We can not loop back
2489 * here from ether_input as a bridge is never a member of a
2492 KASSERT(bifp->if_bridge == NULL,
2493 ("loop created in bridge_input"));
2494 mc2 = m_dup(m, M_NOWAIT);
2496 /* Keep the layer3 header aligned */
2497 int i = min(mc2->m_pkthdr.len, max_protohdr);
2498 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2501 mc2->m_pkthdr.rcvif = bifp;
2502 (*bifp->if_input)(bifp, mc2);
2505 /* Return the original packet for local processing. */
2509 if ((bif->bif_flags & IFBIF_STP) &&
2510 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2514 #if (defined(INET) || defined(INET6))
2515 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2516 || ((iface)->if_carp \
2517 && (*carp_forus_p)((iface), eh->ether_dhost))
2518 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2519 || ((iface)->if_carp \
2520 && (*carp_forus_p)((iface), eh->ether_shost))
2522 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2523 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2527 # define OR_PFIL_HOOKED_INET6 \
2528 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2530 # define OR_PFIL_HOOKED_INET6
2533 #define GRAB_OUR_PACKETS(iface) \
2534 if ((iface)->if_type == IFT_GIF) \
2536 /* It is destined for us. */ \
2537 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2538 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2540 if (bif->bif_flags & IFBIF_LEARNING) { \
2541 error = bridge_rtupdate(sc, eh->ether_shost, \
2542 vlan, bif, 0, IFBAF_DYNAMIC); \
2543 if (error && bif->bif_addrmax) { \
2548 m->m_pkthdr.rcvif = iface; \
2549 if ((iface) == ifp) { \
2550 /* Skip bridge processing... src == dest */ \
2553 /* It's passing over or to the bridge, locally. */ \
2554 ETHER_BPF_MTAP(bifp, m); \
2555 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \
2556 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2557 /* Filter on the physical interface. */ \
2558 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2559 OR_PFIL_HOOKED_INET6)) { \
2560 if (bridge_pfil(&m, NULL, ifp, \
2561 PFIL_IN) != 0 || m == NULL) { \
2565 if ((iface) != bifp) \
2566 ETHER_BPF_MTAP(iface, m); \
2570 /* We just received a packet that we sent out. */ \
2571 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2572 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2579 * Unicast. Make sure it's not for the bridge.
2581 do { GRAB_OUR_PACKETS(bifp) } while (0);
2584 * Give a chance for ifp at first priority. This will help when the
2585 * packet comes through the interface like VLAN's with the same MACs
2586 * on several interfaces from the same bridge. This also will save
2587 * some CPU cycles in case the destination interface and the input
2588 * interface (eq ifp) are the same.
2590 do { GRAB_OUR_PACKETS(ifp) } while (0);
2592 /* Now check the all bridge members. */
2593 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2594 GRAB_OUR_PACKETS(bif2->bif_ifp)
2597 #undef OR_CARP_CHECK_WE_ARE_DST
2598 #undef OR_CARP_CHECK_WE_ARE_SRC
2599 #undef OR_PFIL_HOOKED_INET6
2600 #undef GRAB_OUR_PACKETS
2602 /* Perform the bridge forwarding function. */
2603 bridge_forward(sc, bif, m);
2611 * Send a frame to all interfaces that are members of
2612 * the bridge, except for the one on which the packet
2615 * NOTE: Releases the lock on return.
2618 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2619 struct mbuf *m, int runfilt)
2621 struct bridge_iflist *dbif, *sbif;
2623 struct ifnet *dst_if;
2628 sbif = bridge_lookup_member_if(sc, src_if);
2630 /* Filter on the bridge interface before broadcasting */
2631 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2633 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2636 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2642 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2643 dst_if = dbif->bif_ifp;
2644 if (dst_if == src_if)
2647 /* Private segments can not talk to each other */
2648 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2651 if ((dbif->bif_flags & IFBIF_STP) &&
2652 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2655 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2656 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2659 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2662 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2666 mc = m_dup(m, M_NOWAIT);
2668 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2674 * Filter on the output interface. Pass a NULL bridge interface
2675 * pointer so we do not redundantly filter on the bridge for
2676 * each interface we broadcast on.
2678 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2680 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2684 /* Keep the layer3 header aligned */
2685 i = min(mc->m_pkthdr.len, max_protohdr);
2686 mc = m_copyup(mc, i, ETHER_ALIGN);
2688 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2692 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2698 bridge_enqueue(sc, dst_if, mc);
2707 * Duplicate a packet out one or more interfaces that are in span mode,
2708 * the original mbuf is unmodified.
2711 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2713 struct bridge_iflist *bif;
2714 struct ifnet *dst_if;
2719 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2722 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2723 dst_if = bif->bif_ifp;
2725 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2728 mc = m_dup(m, M_NOWAIT);
2730 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2734 bridge_enqueue(sc, dst_if, mc);
2741 * Add a bridge routing entry.
2744 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2745 struct bridge_iflist *bif, int setflags, uint8_t flags)
2747 struct bridge_rtnode *brt;
2750 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2752 /* Check the source address is valid and not multicast. */
2753 if (ETHER_IS_MULTICAST(dst) ||
2754 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2755 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2758 /* 802.1p frames map to vlan 1 */
2763 * A route for this destination might already exist. If so,
2764 * update it, otherwise create a new one.
2766 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2769 /* Check again, now that we have the lock. There could have
2770 * been a race and we only want to insert this once. */
2771 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2772 BRIDGE_RT_UNLOCK(sc);
2776 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2777 sc->sc_brtexceeded++;
2778 BRIDGE_RT_UNLOCK(sc);
2781 /* Check per interface address limits (if enabled) */
2782 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2783 bif->bif_addrexceeded++;
2784 BRIDGE_RT_UNLOCK(sc);
2789 * Allocate a new bridge forwarding node, and
2790 * initialize the expiration time and Ethernet
2793 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2795 BRIDGE_RT_UNLOCK(sc);
2798 brt->brt_vnet = curvnet;
2800 if (bif->bif_flags & IFBIF_STICKY)
2801 brt->brt_flags = IFBAF_STICKY;
2803 brt->brt_flags = IFBAF_DYNAMIC;
2805 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2806 brt->brt_vlan = vlan;
2808 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2809 uma_zfree(V_bridge_rtnode_zone, brt);
2810 BRIDGE_RT_UNLOCK(sc);
2816 BRIDGE_RT_UNLOCK(sc);
2819 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2820 brt->brt_dst != bif) {
2822 brt->brt_dst->bif_addrcnt--;
2824 brt->brt_dst->bif_addrcnt++;
2825 BRIDGE_RT_UNLOCK(sc);
2828 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2829 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2831 brt->brt_flags = flags;
2839 * Lookup the destination interface for an address.
2841 static struct ifnet *
2842 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2844 struct bridge_rtnode *brt;
2848 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2851 return (brt->brt_ifp);
2857 * Trim the routine table so that we have a number
2858 * of routing entries less than or equal to the
2862 bridge_rttrim(struct bridge_softc *sc)
2864 struct bridge_rtnode *brt, *nbrt;
2867 BRIDGE_RT_LOCK_ASSERT(sc);
2869 /* Make sure we actually need to do this. */
2870 if (sc->sc_brtcnt <= sc->sc_brtmax)
2873 /* Force an aging cycle; this might trim enough addresses. */
2875 if (sc->sc_brtcnt <= sc->sc_brtmax)
2878 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2879 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2880 bridge_rtnode_destroy(sc, brt);
2881 if (sc->sc_brtcnt <= sc->sc_brtmax)
2890 * Aging timer for the bridge.
2893 bridge_timer(void *arg)
2895 struct bridge_softc *sc = arg;
2897 BRIDGE_RT_LOCK_ASSERT(sc);
2899 /* Destruction of rtnodes requires a proper vnet context */
2900 CURVNET_SET(sc->sc_ifp->if_vnet);
2903 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2904 callout_reset(&sc->sc_brcallout,
2905 bridge_rtable_prune_period * hz, bridge_timer, sc);
2912 * Perform an aging cycle.
2915 bridge_rtage(struct bridge_softc *sc)
2917 struct bridge_rtnode *brt, *nbrt;
2919 BRIDGE_RT_LOCK_ASSERT(sc);
2921 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2922 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2923 if (time_uptime >= brt->brt_expire)
2924 bridge_rtnode_destroy(sc, brt);
2932 * Remove all dynamic addresses from the bridge.
2935 bridge_rtflush(struct bridge_softc *sc, int full)
2937 struct bridge_rtnode *brt, *nbrt;
2939 BRIDGE_RT_LOCK_ASSERT(sc);
2941 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2942 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2943 bridge_rtnode_destroy(sc, brt);
2950 * Remove an address from the table.
2953 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2955 struct bridge_rtnode *brt;
2961 * If vlan is zero then we want to delete for all vlans so the lookup
2962 * may return more than one.
2964 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2965 bridge_rtnode_destroy(sc, brt);
2969 BRIDGE_RT_UNLOCK(sc);
2971 return (found ? 0 : ENOENT);
2977 * Delete routes to a speicifc member interface.
2980 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2982 struct bridge_rtnode *brt, *nbrt;
2984 BRIDGE_RT_LOCK_ASSERT(sc);
2986 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2987 if (brt->brt_ifp == ifp && (full ||
2988 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2989 bridge_rtnode_destroy(sc, brt);
2994 * bridge_rtable_init:
2996 * Initialize the route table for this bridge.
2999 bridge_rtable_init(struct bridge_softc *sc)
3003 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3004 M_DEVBUF, M_WAITOK);
3006 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3007 CK_LIST_INIT(&sc->sc_rthash[i]);
3009 sc->sc_rthash_key = arc4random();
3010 CK_LIST_INIT(&sc->sc_rtlist);
3014 * bridge_rtable_fini:
3016 * Deconstruct the route table for this bridge.
3019 bridge_rtable_fini(struct bridge_softc *sc)
3022 KASSERT(sc->sc_brtcnt == 0,
3023 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3024 free(sc->sc_rthash, M_DEVBUF);
3028 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3029 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3031 #define mix(a, b, c) \
3033 a -= b; a -= c; a ^= (c >> 13); \
3034 b -= c; b -= a; b ^= (a << 8); \
3035 c -= a; c -= b; c ^= (b >> 13); \
3036 a -= b; a -= c; a ^= (c >> 12); \
3037 b -= c; b -= a; b ^= (a << 16); \
3038 c -= a; c -= b; c ^= (b >> 5); \
3039 a -= b; a -= c; a ^= (c >> 3); \
3040 b -= c; b -= a; b ^= (a << 10); \
3041 c -= a; c -= b; c ^= (b >> 15); \
3042 } while (/*CONSTCOND*/0)
3044 static __inline uint32_t
3045 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3047 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3058 return (c & BRIDGE_RTHASH_MASK);
3064 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3068 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3069 d = ((int)a[i]) - ((int)b[i]);
3076 * bridge_rtnode_lookup:
3078 * Look up a bridge route node for the specified destination. Compare the
3079 * vlan id or if zero then just return the first match.
3081 static struct bridge_rtnode *
3082 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3084 struct bridge_rtnode *brt;
3088 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3090 hash = bridge_rthash(sc, addr);
3091 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3092 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3093 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3103 * bridge_rtnode_insert:
3105 * Insert the specified bridge node into the route table. We
3106 * assume the entry is not already in the table.
3109 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3111 struct bridge_rtnode *lbrt;
3115 BRIDGE_RT_LOCK_ASSERT(sc);
3117 hash = bridge_rthash(sc, brt->brt_addr);
3119 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3121 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3126 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3127 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3130 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3133 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3134 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3137 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3138 } while (lbrt != NULL);
3141 panic("bridge_rtnode_insert: impossible");
3145 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3152 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3154 struct bridge_rtnode *brt;
3156 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3158 CURVNET_SET(brt->brt_vnet);
3159 uma_zfree(V_bridge_rtnode_zone, brt);
3164 * bridge_rtnode_destroy:
3166 * Destroy a bridge rtnode.
3169 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3171 BRIDGE_RT_LOCK_ASSERT(sc);
3173 CK_LIST_REMOVE(brt, brt_hash);
3175 CK_LIST_REMOVE(brt, brt_list);
3177 brt->brt_dst->bif_addrcnt--;
3179 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3183 * bridge_rtable_expire:
3185 * Set the expiry time for all routes on an interface.
3188 bridge_rtable_expire(struct ifnet *ifp, int age)
3190 struct bridge_softc *sc = ifp->if_bridge;
3191 struct bridge_rtnode *brt;
3193 CURVNET_SET(ifp->if_vnet);
3197 * If the age is zero then flush, otherwise set all the expiry times to
3198 * age for the interface
3201 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3203 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3204 /* Cap the expiry time to 'age' */
3205 if (brt->brt_ifp == ifp &&
3206 brt->brt_expire > time_uptime + age &&
3207 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3208 brt->brt_expire = time_uptime + age;
3211 BRIDGE_RT_UNLOCK(sc);
3216 * bridge_state_change:
3218 * Callback from the bridgestp code when a port changes states.
3221 bridge_state_change(struct ifnet *ifp, int state)
3223 struct bridge_softc *sc = ifp->if_bridge;
3224 static const char *stpstates[] = {
3233 CURVNET_SET(ifp->if_vnet);
3235 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3236 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3241 * Send bridge packets through pfil if they are one of the types pfil can deal
3242 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3243 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3247 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3249 int snap, error, i, hlen;
3250 struct ether_header *eh1, eh2;
3253 u_int16_t ether_type;
3257 error = -1; /* Default error if not error == 0 */
3260 /* we may return with the IP fields swapped, ensure its not shared */
3261 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3264 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3265 return (0); /* filtering is disabled */
3267 i = min((*mp)->m_pkthdr.len, max_protohdr);
3268 if ((*mp)->m_len < i) {
3269 *mp = m_pullup(*mp, i);
3271 printf("%s: m_pullup failed\n", __func__);
3276 eh1 = mtod(*mp, struct ether_header *);
3277 ether_type = ntohs(eh1->ether_type);
3280 * Check for SNAP/LLC.
3282 if (ether_type < ETHERMTU) {
3283 struct llc *llc2 = (struct llc *)(eh1 + 1);
3285 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3286 llc2->llc_dsap == LLC_SNAP_LSAP &&
3287 llc2->llc_ssap == LLC_SNAP_LSAP &&
3288 llc2->llc_control == LLC_UI) {
3289 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3295 * If we're trying to filter bridge traffic, don't look at anything
3296 * other than IP and ARP traffic. If the filter doesn't understand
3297 * IPv6, don't allow IPv6 through the bridge either. This is lame
3298 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3299 * but of course we don't have an AppleTalk filter to begin with.
3300 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3303 switch (ether_type) {
3305 case ETHERTYPE_REVARP:
3306 if (V_pfil_ipfw_arp == 0)
3307 return (0); /* Automatically pass */
3312 case ETHERTYPE_IPV6:
3317 * Check to see if the user wants to pass non-ip
3318 * packets, these will not be checked by pfil(9) and
3319 * passed unconditionally so the default is to drop.
3325 /* Run the packet through pfil before stripping link headers */
3326 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3327 dir == PFIL_OUT && ifp != NULL) {
3328 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3336 /* Strip off the Ethernet header and keep a copy. */
3337 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3338 m_adj(*mp, ETHER_HDR_LEN);
3340 /* Strip off snap header, if present */
3342 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3343 m_adj(*mp, sizeof(struct llc));
3347 * Check the IP header for alignment and errors
3349 if (dir == PFIL_IN) {
3350 switch (ether_type) {
3352 error = bridge_ip_checkbasic(mp);
3355 case ETHERTYPE_IPV6:
3356 error = bridge_ip6_checkbasic(mp);
3369 * Run the packet through pfil
3372 switch (ether_type) {
3375 * Run pfil on the member interface and the bridge, both can
3376 * be skipped by clearing pfil_member or pfil_bridge.
3379 * in_if -> bridge_if -> out_if
3381 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3382 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3386 if (V_pfil_member && ifp != NULL && (rv =
3387 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3391 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3392 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3396 /* check if we need to fragment the packet */
3397 /* bridge_fragment generates a mbuf chain of packets */
3398 /* that already include eth headers */
3399 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3400 i = (*mp)->m_pkthdr.len;
3401 if (i > ifp->if_mtu) {
3402 error = bridge_fragment(ifp, mp, &eh2, snap,
3408 /* Recalculate the ip checksum. */
3409 ip = mtod(*mp, struct ip *);
3410 hlen = ip->ip_hl << 2;
3411 if (hlen < sizeof(struct ip))
3413 if (hlen > (*mp)->m_len) {
3414 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3416 ip = mtod(*mp, struct ip *);
3421 if (hlen == sizeof(struct ip))
3422 ip->ip_sum = in_cksum_hdr(ip);
3424 ip->ip_sum = in_cksum(*mp, hlen);
3428 case ETHERTYPE_IPV6:
3429 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3430 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3434 if (V_pfil_member && ifp != NULL && (rv =
3435 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3439 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3440 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3459 * Finally, put everything back the way it was and return
3462 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3465 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3468 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3471 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3482 * Perform basic checks on header size since
3483 * pfil assumes ip_input has already processed
3484 * it for it. Cut-and-pasted from ip_input.c.
3485 * Given how simple the IPv6 version is,
3486 * does the IPv4 version really need to be
3489 * XXX Should we update ipstat here, or not?
3490 * XXX Right now we update ipstat but not
3494 bridge_ip_checkbasic(struct mbuf **mp)
3496 struct mbuf *m = *mp;
3504 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3505 if ((m = m_copyup(m, sizeof(struct ip),
3506 (max_linkhdr + 3) & ~3)) == NULL) {
3507 /* XXXJRT new stat, please */
3508 KMOD_IPSTAT_INC(ips_toosmall);
3511 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3512 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3513 KMOD_IPSTAT_INC(ips_toosmall);
3517 ip = mtod(m, struct ip *);
3518 if (ip == NULL) goto bad;
3520 if (ip->ip_v != IPVERSION) {
3521 KMOD_IPSTAT_INC(ips_badvers);
3524 hlen = ip->ip_hl << 2;
3525 if (hlen < sizeof(struct ip)) { /* minimum header length */
3526 KMOD_IPSTAT_INC(ips_badhlen);
3529 if (hlen > m->m_len) {
3530 if ((m = m_pullup(m, hlen)) == NULL) {
3531 KMOD_IPSTAT_INC(ips_badhlen);
3534 ip = mtod(m, struct ip *);
3535 if (ip == NULL) goto bad;
3538 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3539 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3541 if (hlen == sizeof(struct ip)) {
3542 sum = in_cksum_hdr(ip);
3544 sum = in_cksum(m, hlen);
3548 KMOD_IPSTAT_INC(ips_badsum);
3552 /* Retrieve the packet length. */
3553 len = ntohs(ip->ip_len);
3556 * Check for additional length bogosity
3559 KMOD_IPSTAT_INC(ips_badlen);
3564 * Check that the amount of data in the buffers
3565 * is as at least much as the IP header would have us expect.
3566 * Drop packet if shorter than we expect.
3568 if (m->m_pkthdr.len < len) {
3569 KMOD_IPSTAT_INC(ips_tooshort);
3573 /* Checks out, proceed */
3584 * Same as above, but for IPv6.
3585 * Cut-and-pasted from ip6_input.c.
3586 * XXX Should we update ip6stat, or not?
3589 bridge_ip6_checkbasic(struct mbuf **mp)
3591 struct mbuf *m = *mp;
3592 struct ip6_hdr *ip6;
3595 * If the IPv6 header is not aligned, slurp it up into a new
3596 * mbuf with space for link headers, in the event we forward
3597 * it. Otherwise, if it is aligned, make sure the entire base
3598 * IPv6 header is in the first mbuf of the chain.
3600 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3601 struct ifnet *inifp = m->m_pkthdr.rcvif;
3602 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3603 (max_linkhdr + 3) & ~3)) == NULL) {
3604 /* XXXJRT new stat, please */
3605 IP6STAT_INC(ip6s_toosmall);
3606 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3609 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3610 struct ifnet *inifp = m->m_pkthdr.rcvif;
3611 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3612 IP6STAT_INC(ip6s_toosmall);
3613 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3618 ip6 = mtod(m, struct ip6_hdr *);
3620 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3621 IP6STAT_INC(ip6s_badvers);
3622 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3626 /* Checks out, proceed */
3639 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3642 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3643 int snap, struct llc *llc)
3645 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3649 if (m->m_len < sizeof(struct ip) &&
3650 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3652 ip = mtod(m, struct ip *);
3654 m->m_pkthdr.csum_flags |= CSUM_IP;
3655 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3660 * Walk the chain and re-add the Ethernet header for
3663 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3664 nextpkt = mcur->m_nextpkt;
3665 mcur->m_nextpkt = NULL;
3667 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3671 mprev->m_nextpkt = nextpkt;
3674 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3677 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3681 mprev->m_nextpkt = nextpkt;
3684 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3687 * The previous two M_PREPEND could have inserted one or two
3688 * mbufs in front so we have to update the previous packet's
3691 mcur->m_nextpkt = nextpkt;
3693 mprev->m_nextpkt = mcur;
3695 /* The first mbuf in the original chain needs to be
3702 KMOD_IPSTAT_INC(ips_fragmented);
3706 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3707 m = mcur->m_nextpkt;
3714 bridge_linkstate(struct ifnet *ifp)
3716 struct bridge_softc *sc = ifp->if_bridge;
3717 struct bridge_iflist *bif;
3718 struct epoch_tracker et;
3720 NET_EPOCH_ENTER(et);
3722 bif = bridge_lookup_member_if(sc, ifp);
3727 bridge_linkcheck(sc);
3729 bstp_linkstate(&bif->bif_stp);
3735 bridge_linkcheck(struct bridge_softc *sc)
3737 struct bridge_iflist *bif;
3738 int new_link, hasls;
3740 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3742 new_link = LINK_STATE_DOWN;
3744 /* Our link is considered up if at least one of our ports is active */
3745 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3746 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3748 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3749 new_link = LINK_STATE_UP;
3753 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3754 /* If no interfaces support link-state then we default to up */
3755 new_link = LINK_STATE_UP;
3757 if_link_state_change(sc->sc_ifp, new_link);