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 *, char *, size_t,
290 struct ifc_data *, struct ifnet **);
291 static int bridge_clone_destroy(struct if_clone *, struct ifnet *, uint32_t);
293 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
294 static void bridge_mutecaps(struct bridge_softc *);
295 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
297 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
298 static void bridge_init(void *);
299 static void bridge_dummynet(struct mbuf *, struct ifnet *);
300 static void bridge_stop(struct ifnet *, int);
301 static int bridge_transmit(struct ifnet *, struct mbuf *);
303 static void bridge_altq_start(if_t);
304 static int bridge_altq_transmit(if_t, struct mbuf *);
306 static void bridge_qflush(struct ifnet *);
307 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
308 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
310 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
312 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
314 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
317 static void bridge_timer(void *);
319 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
321 static void bridge_span(struct bridge_softc *, struct mbuf *);
323 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
324 uint16_t, struct bridge_iflist *, int, uint8_t);
325 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
327 static void bridge_rttrim(struct bridge_softc *);
328 static void bridge_rtage(struct bridge_softc *);
329 static void bridge_rtflush(struct bridge_softc *, int);
330 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
333 static void bridge_rtable_init(struct bridge_softc *);
334 static void bridge_rtable_fini(struct bridge_softc *);
336 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
337 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
338 const uint8_t *, uint16_t);
339 static int bridge_rtnode_insert(struct bridge_softc *,
340 struct bridge_rtnode *);
341 static void bridge_rtnode_destroy(struct bridge_softc *,
342 struct bridge_rtnode *);
343 static void bridge_rtable_expire(struct ifnet *, int);
344 static void bridge_state_change(struct ifnet *, int);
346 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
348 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
350 static void bridge_delete_member(struct bridge_softc *,
351 struct bridge_iflist *, int);
352 static void bridge_delete_span(struct bridge_softc *,
353 struct bridge_iflist *);
355 static int bridge_ioctl_add(struct bridge_softc *, void *);
356 static int bridge_ioctl_del(struct bridge_softc *, void *);
357 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
358 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
359 static int bridge_ioctl_scache(struct bridge_softc *, void *);
360 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
361 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
362 static int bridge_ioctl_rts(struct bridge_softc *, void *);
363 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
364 static int bridge_ioctl_sto(struct bridge_softc *, void *);
365 static int bridge_ioctl_gto(struct bridge_softc *, void *);
366 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
367 static int bridge_ioctl_flush(struct bridge_softc *, void *);
368 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
369 static int bridge_ioctl_spri(struct bridge_softc *, void *);
370 static int bridge_ioctl_ght(struct bridge_softc *, void *);
371 static int bridge_ioctl_sht(struct bridge_softc *, void *);
372 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
373 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
374 static int bridge_ioctl_gma(struct bridge_softc *, void *);
375 static int bridge_ioctl_sma(struct bridge_softc *, void *);
376 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
377 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
378 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
379 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
380 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
381 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
382 static int bridge_ioctl_grte(struct bridge_softc *, void *);
383 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
384 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
385 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
386 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
388 static int bridge_ip_checkbasic(struct mbuf **mp);
390 static int bridge_ip6_checkbasic(struct mbuf **mp);
392 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
393 struct ether_header *, int, struct llc *);
394 static void bridge_linkstate(struct ifnet *ifp);
395 static void bridge_linkcheck(struct bridge_softc *sc);
397 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
398 #define VLANTAGOF(_m) \
399 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
401 static struct bstp_cb_ops bridge_ops = {
402 .bcb_state = bridge_state_change,
403 .bcb_rtage = bridge_rtable_expire
406 SYSCTL_DECL(_net_link);
407 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
410 /* only pass IP[46] packets when pfil is enabled */
411 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
412 #define V_pfil_onlyip VNET(pfil_onlyip)
413 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
414 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
415 "Only pass IP packets when pfil is enabled");
417 /* run pfil hooks on the bridge interface */
418 VNET_DEFINE_STATIC(int, pfil_bridge) = 0;
419 #define V_pfil_bridge VNET(pfil_bridge)
420 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
421 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
422 "Packet filter on the bridge interface");
424 /* layer2 filter with ipfw */
425 VNET_DEFINE_STATIC(int, pfil_ipfw);
426 #define V_pfil_ipfw VNET(pfil_ipfw)
428 /* layer2 ARP filter with ipfw */
429 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
430 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
431 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
432 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
433 "Filter ARP packets through IPFW layer2");
435 /* run pfil hooks on the member interface */
436 VNET_DEFINE_STATIC(int, pfil_member) = 0;
437 #define V_pfil_member VNET(pfil_member)
438 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
439 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
440 "Packet filter on the member interface");
442 /* run pfil hooks on the physical interface for locally destined packets */
443 VNET_DEFINE_STATIC(int, pfil_local_phys);
444 #define V_pfil_local_phys VNET(pfil_local_phys)
445 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
446 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
447 "Packet filter on the physical interface for locally destined packets");
449 /* log STP state changes */
450 VNET_DEFINE_STATIC(int, log_stp);
451 #define V_log_stp VNET(log_stp)
452 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
453 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
454 "Log STP state changes");
456 /* share MAC with first bridge member */
457 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
458 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
459 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
460 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
461 "Inherit MAC address from the first bridge member");
463 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
464 #define V_allow_llz_overlap VNET(allow_llz_overlap)
465 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
466 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
467 "Allow overlap of link-local scope "
468 "zones of a bridge interface and the member interfaces");
470 struct bridge_control {
471 int (*bc_func)(struct bridge_softc *, void *);
476 #define BC_F_COPYIN 0x01 /* copy arguments in */
477 #define BC_F_COPYOUT 0x02 /* copy arguments out */
478 #define BC_F_SUSER 0x04 /* do super-user check */
480 const struct bridge_control bridge_control_table[] = {
481 { bridge_ioctl_add, sizeof(struct ifbreq),
482 BC_F_COPYIN|BC_F_SUSER },
483 { bridge_ioctl_del, sizeof(struct ifbreq),
484 BC_F_COPYIN|BC_F_SUSER },
486 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
487 BC_F_COPYIN|BC_F_COPYOUT },
488 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
489 BC_F_COPYIN|BC_F_SUSER },
491 { bridge_ioctl_scache, sizeof(struct ifbrparam),
492 BC_F_COPYIN|BC_F_SUSER },
493 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
496 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
497 BC_F_COPYIN|BC_F_COPYOUT },
498 { bridge_ioctl_rts, sizeof(struct ifbaconf),
499 BC_F_COPYIN|BC_F_COPYOUT },
501 { bridge_ioctl_saddr, sizeof(struct ifbareq),
502 BC_F_COPYIN|BC_F_SUSER },
504 { bridge_ioctl_sto, sizeof(struct ifbrparam),
505 BC_F_COPYIN|BC_F_SUSER },
506 { bridge_ioctl_gto, sizeof(struct ifbrparam),
509 { bridge_ioctl_daddr, sizeof(struct ifbareq),
510 BC_F_COPYIN|BC_F_SUSER },
512 { bridge_ioctl_flush, sizeof(struct ifbreq),
513 BC_F_COPYIN|BC_F_SUSER },
515 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
517 { bridge_ioctl_spri, sizeof(struct ifbrparam),
518 BC_F_COPYIN|BC_F_SUSER },
520 { bridge_ioctl_ght, sizeof(struct ifbrparam),
522 { bridge_ioctl_sht, sizeof(struct ifbrparam),
523 BC_F_COPYIN|BC_F_SUSER },
525 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
527 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
528 BC_F_COPYIN|BC_F_SUSER },
530 { bridge_ioctl_gma, sizeof(struct ifbrparam),
532 { bridge_ioctl_sma, sizeof(struct ifbrparam),
533 BC_F_COPYIN|BC_F_SUSER },
535 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
536 BC_F_COPYIN|BC_F_SUSER },
538 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
539 BC_F_COPYIN|BC_F_SUSER },
541 { bridge_ioctl_addspan, sizeof(struct ifbreq),
542 BC_F_COPYIN|BC_F_SUSER },
543 { bridge_ioctl_delspan, sizeof(struct ifbreq),
544 BC_F_COPYIN|BC_F_SUSER },
546 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
549 { bridge_ioctl_grte, sizeof(struct ifbrparam),
552 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
553 BC_F_COPYIN|BC_F_COPYOUT },
555 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
556 BC_F_COPYIN|BC_F_SUSER },
558 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
559 BC_F_COPYIN|BC_F_SUSER },
561 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
562 BC_F_COPYIN|BC_F_SUSER },
565 const int bridge_control_table_size = nitems(bridge_control_table);
567 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
568 #define V_bridge_list VNET(bridge_list)
569 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \
571 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx)
572 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx)
573 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx)
575 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
576 #define V_bridge_cloner VNET(bridge_cloner)
578 static const char bridge_name[] = "bridge";
581 vnet_bridge_init(const void *unused __unused)
584 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
585 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
587 BRIDGE_LIST_LOCK_INIT();
588 LIST_INIT(&V_bridge_list);
590 struct if_clone_addreq req = {
591 .create_f = bridge_clone_create,
592 .destroy_f = bridge_clone_destroy,
593 .flags = IFC_F_AUTOUNIT,
595 V_bridge_cloner = ifc_attach_cloner(bridge_name, &req);
597 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
598 vnet_bridge_init, NULL);
601 vnet_bridge_uninit(const void *unused __unused)
604 ifc_detach_cloner(V_bridge_cloner);
605 V_bridge_cloner = NULL;
606 BRIDGE_LIST_LOCK_DESTROY();
608 /* Callbacks may use the UMA zone. */
609 NET_EPOCH_DRAIN_CALLBACKS();
611 uma_zdestroy(V_bridge_rtnode_zone);
613 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
614 vnet_bridge_uninit, NULL);
617 bridge_modevent(module_t mod, int type, void *data)
622 bridge_dn_p = bridge_dummynet;
623 bridge_detach_cookie = EVENTHANDLER_REGISTER(
624 ifnet_departure_event, bridge_ifdetach, NULL,
625 EVENTHANDLER_PRI_ANY);
628 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
629 bridge_detach_cookie);
638 static moduledata_t bridge_mod = {
644 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
645 MODULE_VERSION(if_bridge, 1);
646 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
649 * handler for net.link.bridge.ipfw
652 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
654 int enable = V_pfil_ipfw;
657 error = sysctl_handle_int(oidp, &enable, 0, req);
660 if (enable != V_pfil_ipfw) {
661 V_pfil_ipfw = enable;
664 * Disable pfil so that ipfw doesnt run twice, if the user
665 * really wants both then they can re-enable pfil_bridge and/or
666 * pfil_member. Also allow non-ip packets as ipfw can filter by
678 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
679 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
680 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
681 "Layer2 filter with IPFW");
685 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
687 struct bridge_softc *sc = ifp->if_softc;
688 struct bridge_iflist *bif;
692 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
693 bridge_delete_member(sc, bif, 0);
695 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
696 bridge_delete_span(sc, bif);
701 ether_reassign(ifp, newvnet, arg);
706 * bridge_clone_create:
708 * Create a new bridge instance.
711 bridge_clone_create(struct if_clone *ifc, char *name, size_t len,
712 struct ifc_data *ifd, struct ifnet **ifpp)
714 struct bridge_softc *sc;
717 sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
718 ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
724 BRIDGE_LOCK_INIT(sc);
725 sc->sc_brtmax = BRIDGE_RTABLE_MAX;
726 sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
728 /* Initialize our routing table. */
729 bridge_rtable_init(sc);
731 callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
733 CK_LIST_INIT(&sc->sc_iflist);
734 CK_LIST_INIT(&sc->sc_spanlist);
737 if_initname(ifp, bridge_name, ifd->unit);
738 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
739 ifp->if_ioctl = bridge_ioctl;
741 ifp->if_start = bridge_altq_start;
742 ifp->if_transmit = bridge_altq_transmit;
743 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
744 ifp->if_snd.ifq_drv_maxlen = 0;
745 IFQ_SET_READY(&ifp->if_snd);
747 ifp->if_transmit = bridge_transmit;
749 ifp->if_qflush = bridge_qflush;
750 ifp->if_init = bridge_init;
751 ifp->if_type = IFT_BRIDGE;
753 ether_gen_addr(ifp, &sc->sc_defaddr);
755 bstp_attach(&sc->sc_stp, &bridge_ops);
756 ether_ifattach(ifp, sc->sc_defaddr.octet);
757 /* Now undo some of the damage... */
758 ifp->if_baudrate = 0;
759 ifp->if_type = IFT_BRIDGE;
761 ifp->if_reassign = bridge_reassign;
765 LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
766 BRIDGE_LIST_UNLOCK();
773 bridge_clone_destroy_cb(struct epoch_context *ctx)
775 struct bridge_softc *sc;
777 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
779 BRIDGE_LOCK_DESTROY(sc);
784 * bridge_clone_destroy:
786 * Destroy a bridge instance.
789 bridge_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags)
791 struct bridge_softc *sc = ifp->if_softc;
792 struct bridge_iflist *bif;
793 struct epoch_tracker et;
798 ifp->if_flags &= ~IFF_UP;
800 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
801 bridge_delete_member(sc, bif, 0);
803 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
804 bridge_delete_span(sc, bif);
807 /* Tear down the routing table. */
808 bridge_rtable_fini(sc);
814 callout_drain(&sc->sc_brcallout);
817 LIST_REMOVE(sc, sc_list);
818 BRIDGE_LIST_UNLOCK();
820 bstp_detach(&sc->sc_stp);
822 IFQ_PURGE(&ifp->if_snd);
829 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
837 * Handle a control request from the operator.
840 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
842 struct bridge_softc *sc = ifp->if_softc;
843 struct ifreq *ifr = (struct ifreq *)data;
844 struct bridge_iflist *bif;
845 struct thread *td = curthread;
847 struct ifbreq ifbreq;
848 struct ifbifconf ifbifconf;
849 struct ifbareq ifbareq;
850 struct ifbaconf ifbaconf;
851 struct ifbrparam ifbrparam;
852 struct ifbropreq ifbropreq;
854 struct ifdrv *ifd = (struct ifdrv *) data;
855 const struct bridge_control *bc;
856 int error = 0, oldmtu;
867 if (ifd->ifd_cmd >= bridge_control_table_size) {
871 bc = &bridge_control_table[ifd->ifd_cmd];
873 if (cmd == SIOCGDRVSPEC &&
874 (bc->bc_flags & BC_F_COPYOUT) == 0) {
878 else if (cmd == SIOCSDRVSPEC &&
879 (bc->bc_flags & BC_F_COPYOUT) != 0) {
884 if (bc->bc_flags & BC_F_SUSER) {
885 error = priv_check(td, PRIV_NET_BRIDGE);
890 if (ifd->ifd_len != bc->bc_argsize ||
891 ifd->ifd_len > sizeof(args)) {
896 bzero(&args, sizeof(args));
897 if (bc->bc_flags & BC_F_COPYIN) {
898 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
903 oldmtu = ifp->if_mtu;
904 error = (*bc->bc_func)(sc, &args);
909 * Bridge MTU may change during addition of the first port.
910 * If it did, do network layer specific procedure.
912 if (ifp->if_mtu != oldmtu) {
919 if (bc->bc_flags & BC_F_COPYOUT)
920 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
925 if (!(ifp->if_flags & IFF_UP) &&
926 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
928 * If interface is marked down and it is running,
929 * then stop and disable it.
932 } else if ((ifp->if_flags & IFF_UP) &&
933 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
935 * If interface is marked up and it is stopped, then
945 oldmtu = sc->sc_ifp->if_mtu;
947 if (ifr->ifr_mtu < 576) {
951 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
952 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
955 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
956 error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
957 SIOCSIFMTU, (caddr_t)ifr);
959 log(LOG_NOTICE, "%s: invalid MTU: %u for"
960 " member %s\n", sc->sc_ifp->if_xname,
962 bif->bif_ifp->if_xname);
968 /* Restore the previous MTU on all member interfaces. */
969 ifr->ifr_mtu = oldmtu;
970 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
971 (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
972 SIOCSIFMTU, (caddr_t)ifr);
975 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
980 * drop the lock as ether_ioctl() will call bridge_start() and
981 * cause the lock to be recursed.
984 error = ether_ioctl(ifp, cmd, data);
997 * Clear or restore unwanted capabilities on the member interface
1000 bridge_mutecaps(struct bridge_softc *sc)
1002 struct bridge_iflist *bif;
1005 BRIDGE_LOCK_ASSERT(sc);
1007 /* Initial bitmask of capabilities to test */
1008 mask = BRIDGE_IFCAPS_MASK;
1010 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1011 /* Every member must support it or its disabled */
1012 mask &= bif->bif_savedcaps;
1015 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1016 enabled = bif->bif_ifp->if_capenable;
1017 enabled &= ~BRIDGE_IFCAPS_STRIP;
1018 /* strip off mask bits and enable them again if allowed */
1019 enabled &= ~BRIDGE_IFCAPS_MASK;
1021 bridge_set_ifcap(sc, bif, enabled);
1026 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1028 struct ifnet *ifp = bif->bif_ifp;
1030 int error, mask, stuck;
1032 bzero(&ifr, sizeof(ifr));
1033 ifr.ifr_reqcap = set;
1035 if (ifp->if_capenable != set) {
1036 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1038 if_printf(sc->sc_ifp,
1039 "error setting capabilities on %s: %d\n",
1040 ifp->if_xname, error);
1041 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1042 stuck = ifp->if_capenable & mask & ~set;
1044 if_printf(sc->sc_ifp,
1045 "can't disable some capabilities on %s: 0x%x\n",
1046 ifp->if_xname, stuck);
1051 * bridge_lookup_member:
1053 * Lookup a bridge member interface.
1055 static struct bridge_iflist *
1056 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1058 struct bridge_iflist *bif;
1061 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1063 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1065 if (strcmp(ifp->if_xname, name) == 0)
1073 * bridge_lookup_member_if:
1075 * Lookup a bridge member interface by ifnet*.
1077 static struct bridge_iflist *
1078 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1080 struct bridge_iflist *bif;
1082 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1084 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1085 if (bif->bif_ifp == member_ifp)
1093 bridge_delete_member_cb(struct epoch_context *ctx)
1095 struct bridge_iflist *bif;
1097 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1099 free(bif, M_DEVBUF);
1103 * bridge_delete_member:
1105 * Delete the specified member interface.
1108 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1111 struct ifnet *ifs = bif->bif_ifp;
1112 struct ifnet *fif = NULL;
1113 struct bridge_iflist *bifl;
1115 BRIDGE_LOCK_ASSERT(sc);
1117 if (bif->bif_flags & IFBIF_STP)
1118 bstp_disable(&bif->bif_stp);
1120 ifs->if_bridge = NULL;
1121 CK_LIST_REMOVE(bif, bif_next);
1124 * If removing the interface that gave the bridge its mac address, set
1125 * the mac address of the bridge to the address of the next member, or
1126 * to its default address if no members are left.
1128 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1129 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1130 bcopy(&sc->sc_defaddr,
1131 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1132 sc->sc_ifaddr = NULL;
1134 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1135 fif = bifl->bif_ifp;
1136 bcopy(IF_LLADDR(fif),
1137 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1138 sc->sc_ifaddr = fif;
1140 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1143 bridge_linkcheck(sc);
1144 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1146 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1147 BRIDGE_RT_UNLOCK(sc);
1148 KASSERT(bif->bif_addrcnt == 0,
1149 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1151 ifs->if_bridge_output = NULL;
1152 ifs->if_bridge_input = NULL;
1153 ifs->if_bridge_linkstate = NULL;
1155 switch (ifs->if_type) {
1159 * Take the interface out of promiscuous mode, but only
1160 * if it was promiscuous in the first place. It might
1161 * not be if we're in the bridge_ioctl_add() error path.
1163 if (ifs->if_flags & IFF_PROMISC)
1164 (void) ifpromisc(ifs, 0);
1172 panic("bridge_delete_member: impossible");
1176 /* reneable any interface capabilities */
1177 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1179 bstp_destroy(&bif->bif_stp); /* prepare to free */
1181 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1185 * bridge_delete_span:
1187 * Delete the specified span interface.
1190 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1192 BRIDGE_LOCK_ASSERT(sc);
1194 KASSERT(bif->bif_ifp->if_bridge == NULL,
1195 ("%s: not a span interface", __func__));
1197 CK_LIST_REMOVE(bif, bif_next);
1199 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1203 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1205 struct ifbreq *req = arg;
1206 struct bridge_iflist *bif = NULL;
1210 ifs = ifunit(req->ifbr_ifsname);
1213 if (ifs->if_ioctl == NULL) /* must be supported */
1216 /* If it's in the span list, it can't be a member. */
1217 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1218 if (ifs == bif->bif_ifp)
1221 if (ifs->if_bridge == sc)
1224 if (ifs->if_bridge != NULL)
1227 switch (ifs->if_type) {
1231 /* permitted interface types */
1239 * Two valid inet6 addresses with link-local scope must not be
1240 * on the parent interface and the member interfaces at the
1241 * same time. This restriction is needed to prevent violation
1242 * of link-local scope zone. Attempts to add a member
1243 * interface which has inet6 addresses when the parent has
1244 * inet6 triggers removal of all inet6 addresses on the member
1248 /* Check if the parent interface has a link-local scope addr. */
1249 if (V_allow_llz_overlap == 0 &&
1250 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1252 * If any, remove all inet6 addresses from the member
1255 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1256 if (in6ifa_llaonifp(bif->bif_ifp)) {
1257 in6_ifdetach(bif->bif_ifp);
1258 if_printf(sc->sc_ifp,
1259 "IPv6 addresses on %s have been removed "
1260 "before adding it as a member to prevent "
1261 "IPv6 address scope violation.\n",
1262 bif->bif_ifp->if_xname);
1265 if (in6ifa_llaonifp(ifs)) {
1267 if_printf(sc->sc_ifp,
1268 "IPv6 addresses on %s have been removed "
1269 "before adding it as a member to prevent "
1270 "IPv6 address scope violation.\n",
1275 /* Allow the first Ethernet member to define the MTU */
1276 if (CK_LIST_EMPTY(&sc->sc_iflist))
1277 sc->sc_ifp->if_mtu = ifs->if_mtu;
1278 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1281 snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s",
1283 ifr.ifr_mtu = sc->sc_ifp->if_mtu;
1285 error = (*ifs->if_ioctl)(ifs,
1286 SIOCSIFMTU, (caddr_t)&ifr);
1288 log(LOG_NOTICE, "%s: invalid MTU: %u for"
1289 " new member %s\n", sc->sc_ifp->if_xname,
1296 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1301 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1302 bif->bif_savedcaps = ifs->if_capenable;
1305 * Assign the interface's MAC address to the bridge if it's the first
1306 * member and the MAC address of the bridge has not been changed from
1307 * the default randomly generated one.
1309 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1310 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1311 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1312 sc->sc_ifaddr = ifs;
1313 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1316 ifs->if_bridge = sc;
1317 ifs->if_bridge_output = bridge_output;
1318 ifs->if_bridge_input = bridge_input;
1319 ifs->if_bridge_linkstate = bridge_linkstate;
1320 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1322 * XXX: XLOCK HERE!?!
1324 * NOTE: insert_***HEAD*** should be safe for the traversals.
1326 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1328 /* Set interface capabilities to the intersection set of all members */
1329 bridge_mutecaps(sc);
1330 bridge_linkcheck(sc);
1332 /* Place the interface into promiscuous mode */
1333 switch (ifs->if_type) {
1336 error = ifpromisc(ifs, 1);
1341 bridge_delete_member(sc, bif, 0);
1346 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1348 struct ifbreq *req = arg;
1349 struct bridge_iflist *bif;
1351 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1355 bridge_delete_member(sc, bif, 0);
1361 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1363 struct ifbreq *req = arg;
1364 struct bridge_iflist *bif;
1365 struct bstp_port *bp;
1367 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1372 req->ifbr_ifsflags = bif->bif_flags;
1373 req->ifbr_state = bp->bp_state;
1374 req->ifbr_priority = bp->bp_priority;
1375 req->ifbr_path_cost = bp->bp_path_cost;
1376 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1377 req->ifbr_proto = bp->bp_protover;
1378 req->ifbr_role = bp->bp_role;
1379 req->ifbr_stpflags = bp->bp_flags;
1380 req->ifbr_addrcnt = bif->bif_addrcnt;
1381 req->ifbr_addrmax = bif->bif_addrmax;
1382 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1384 /* Copy STP state options as flags */
1385 if (bp->bp_operedge)
1386 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1387 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1388 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1389 if (bp->bp_ptp_link)
1390 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1391 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1392 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1393 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1394 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1395 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1396 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1401 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1403 struct epoch_tracker et;
1404 struct ifbreq *req = arg;
1405 struct bridge_iflist *bif;
1406 struct bstp_port *bp;
1409 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1414 if (req->ifbr_ifsflags & IFBIF_SPAN)
1415 /* SPAN is readonly */
1418 NET_EPOCH_ENTER(et);
1420 if (req->ifbr_ifsflags & IFBIF_STP) {
1421 if ((bif->bif_flags & IFBIF_STP) == 0) {
1422 error = bstp_enable(&bif->bif_stp);
1429 if ((bif->bif_flags & IFBIF_STP) != 0)
1430 bstp_disable(&bif->bif_stp);
1433 /* Pass on STP flags */
1434 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1435 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1436 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1437 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1439 /* Save the bits relating to the bridge */
1440 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1448 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1450 struct ifbrparam *param = arg;
1452 sc->sc_brtmax = param->ifbrp_csize;
1459 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1461 struct ifbrparam *param = arg;
1463 param->ifbrp_csize = sc->sc_brtmax;
1469 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1471 struct ifbifconf *bifc = arg;
1472 struct bridge_iflist *bif;
1475 int count, buflen, len, error = 0;
1478 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1480 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1483 buflen = sizeof(breq) * count;
1484 if (bifc->ifbic_len == 0) {
1485 bifc->ifbic_len = buflen;
1488 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1494 len = min(bifc->ifbic_len, buflen);
1495 bzero(&breq, sizeof(breq));
1496 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1497 if (len < sizeof(breq))
1500 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1501 sizeof(breq.ifbr_ifsname));
1502 /* Fill in the ifbreq structure */
1503 error = bridge_ioctl_gifflags(sc, &breq);
1506 memcpy(buf, &breq, sizeof(breq));
1508 buf += sizeof(breq);
1509 len -= sizeof(breq);
1511 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1512 if (len < sizeof(breq))
1515 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1516 sizeof(breq.ifbr_ifsname));
1517 breq.ifbr_ifsflags = bif->bif_flags;
1518 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1519 memcpy(buf, &breq, sizeof(breq));
1521 buf += sizeof(breq);
1522 len -= sizeof(breq);
1525 bifc->ifbic_len = sizeof(breq) * count;
1526 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1527 free(outbuf, M_TEMP);
1532 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1534 struct ifbaconf *bac = arg;
1535 struct bridge_rtnode *brt;
1536 struct ifbareq bareq;
1538 int count, buflen, len, error = 0;
1540 if (bac->ifbac_len == 0)
1544 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1546 buflen = sizeof(bareq) * count;
1548 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1554 len = min(bac->ifbac_len, buflen);
1555 bzero(&bareq, sizeof(bareq));
1556 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1557 if (len < sizeof(bareq))
1559 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1560 sizeof(bareq.ifba_ifsname));
1561 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1562 bareq.ifba_vlan = brt->brt_vlan;
1563 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1564 time_uptime < brt->brt_expire)
1565 bareq.ifba_expire = brt->brt_expire - time_uptime;
1567 bareq.ifba_expire = 0;
1568 bareq.ifba_flags = brt->brt_flags;
1570 memcpy(buf, &bareq, sizeof(bareq));
1572 buf += sizeof(bareq);
1573 len -= sizeof(bareq);
1576 bac->ifbac_len = sizeof(bareq) * count;
1577 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1578 free(outbuf, M_TEMP);
1583 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1585 struct ifbareq *req = arg;
1586 struct bridge_iflist *bif;
1587 struct epoch_tracker et;
1590 NET_EPOCH_ENTER(et);
1591 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1597 /* bridge_rtupdate() may acquire the lock. */
1598 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1606 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1608 struct ifbrparam *param = arg;
1610 sc->sc_brttimeout = param->ifbrp_ctime;
1615 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1617 struct ifbrparam *param = arg;
1619 param->ifbrp_ctime = sc->sc_brttimeout;
1624 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1626 struct ifbareq *req = arg;
1628 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1632 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1634 struct ifbreq *req = arg;
1637 bridge_rtflush(sc, req->ifbr_ifsflags);
1638 BRIDGE_RT_UNLOCK(sc);
1644 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1646 struct ifbrparam *param = arg;
1647 struct bstp_state *bs = &sc->sc_stp;
1649 param->ifbrp_prio = bs->bs_bridge_priority;
1654 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1656 struct ifbrparam *param = arg;
1658 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1662 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1664 struct ifbrparam *param = arg;
1665 struct bstp_state *bs = &sc->sc_stp;
1667 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1672 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1674 struct ifbrparam *param = arg;
1676 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1680 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1682 struct ifbrparam *param = arg;
1683 struct bstp_state *bs = &sc->sc_stp;
1685 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1690 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1692 struct ifbrparam *param = arg;
1694 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1698 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1700 struct ifbrparam *param = arg;
1701 struct bstp_state *bs = &sc->sc_stp;
1703 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1708 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1710 struct ifbrparam *param = arg;
1712 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1716 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1718 struct ifbreq *req = arg;
1719 struct bridge_iflist *bif;
1721 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1725 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1729 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1731 struct ifbreq *req = arg;
1732 struct bridge_iflist *bif;
1734 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1738 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1742 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1744 struct ifbreq *req = arg;
1745 struct bridge_iflist *bif;
1747 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1751 bif->bif_addrmax = req->ifbr_addrmax;
1756 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1758 struct ifbreq *req = arg;
1759 struct bridge_iflist *bif = NULL;
1762 ifs = ifunit(req->ifbr_ifsname);
1766 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1767 if (ifs == bif->bif_ifp)
1770 if (ifs->if_bridge != NULL)
1773 switch (ifs->if_type) {
1782 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1787 bif->bif_flags = IFBIF_SPAN;
1789 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1795 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1797 struct ifbreq *req = arg;
1798 struct bridge_iflist *bif;
1801 ifs = ifunit(req->ifbr_ifsname);
1805 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1806 if (ifs == bif->bif_ifp)
1812 bridge_delete_span(sc, bif);
1818 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1820 struct ifbropreq *req = arg;
1821 struct bstp_state *bs = &sc->sc_stp;
1822 struct bstp_port *root_port;
1824 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1825 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1826 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1828 root_port = bs->bs_root_port;
1829 if (root_port == NULL)
1830 req->ifbop_root_port = 0;
1832 req->ifbop_root_port = root_port->bp_ifp->if_index;
1834 req->ifbop_holdcount = bs->bs_txholdcount;
1835 req->ifbop_priority = bs->bs_bridge_priority;
1836 req->ifbop_protocol = bs->bs_protover;
1837 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1838 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1839 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1840 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1841 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1842 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1848 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1850 struct ifbrparam *param = arg;
1852 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1857 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1859 struct ifbpstpconf *bifstp = arg;
1860 struct bridge_iflist *bif;
1861 struct bstp_port *bp;
1862 struct ifbpstpreq bpreq;
1864 int count, buflen, len, error = 0;
1867 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1868 if ((bif->bif_flags & IFBIF_STP) != 0)
1872 buflen = sizeof(bpreq) * count;
1873 if (bifstp->ifbpstp_len == 0) {
1874 bifstp->ifbpstp_len = buflen;
1878 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1884 len = min(bifstp->ifbpstp_len, buflen);
1885 bzero(&bpreq, sizeof(bpreq));
1886 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1887 if (len < sizeof(bpreq))
1890 if ((bif->bif_flags & IFBIF_STP) == 0)
1894 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1895 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1896 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1897 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1898 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1899 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1901 memcpy(buf, &bpreq, sizeof(bpreq));
1903 buf += sizeof(bpreq);
1904 len -= sizeof(bpreq);
1907 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1908 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1909 free(outbuf, M_TEMP);
1914 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1916 struct ifbrparam *param = arg;
1918 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1922 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1924 struct ifbrparam *param = arg;
1926 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1932 * Detach an interface from a bridge. Called when a member
1933 * interface is detaching.
1936 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1938 struct bridge_softc *sc = ifp->if_bridge;
1939 struct bridge_iflist *bif;
1941 if (ifp->if_flags & IFF_RENAMING)
1943 if (V_bridge_cloner == NULL) {
1945 * This detach handler can be called after
1946 * vnet_bridge_uninit(). Just return in that case.
1950 /* Check if the interface is a bridge member */
1954 bif = bridge_lookup_member_if(sc, ifp);
1956 bridge_delete_member(sc, bif, 1);
1962 /* Check if the interface is a span port */
1964 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1966 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1967 if (ifp == bif->bif_ifp) {
1968 bridge_delete_span(sc, bif);
1974 BRIDGE_LIST_UNLOCK();
1980 * Initialize a bridge interface.
1983 bridge_init(void *xsc)
1985 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1986 struct ifnet *ifp = sc->sc_ifp;
1988 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1992 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1995 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1996 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
2004 * Stop the bridge interface.
2007 bridge_stop(struct ifnet *ifp, int disable)
2009 struct bridge_softc *sc = ifp->if_softc;
2011 BRIDGE_LOCK_ASSERT(sc);
2013 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2017 callout_stop(&sc->sc_brcallout);
2019 bstp_stop(&sc->sc_stp);
2021 bridge_rtflush(sc, IFBF_FLUSHDYN);
2022 BRIDGE_RT_UNLOCK(sc);
2024 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2030 * Enqueue a packet on a bridge member interface.
2034 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2040 /* We may be sending a fragment so traverse the mbuf */
2043 m->m_nextpkt = NULL;
2044 len = m->m_pkthdr.len;
2045 mflags = m->m_flags;
2048 * If underlying interface can not do VLAN tag insertion itself
2049 * then attach a packet tag that holds it.
2051 if ((m->m_flags & M_VLANTAG) &&
2052 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2053 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2056 "unable to prepend VLAN header\n");
2057 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2060 m->m_flags &= ~M_VLANTAG;
2063 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2064 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2067 for (m = m0, n = 1; m != NULL; m = m0, n++) {
2071 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, n);
2075 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2076 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2077 if (mflags & M_MCAST)
2078 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2087 * Receive a queued packet from dummynet and pass it on to the output
2090 * The mbuf has the Ethernet header already attached.
2093 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2095 struct bridge_softc *sc;
2097 sc = ifp->if_bridge;
2100 * The packet didnt originate from a member interface. This should only
2101 * ever happen if a member interface is removed while packets are
2109 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2111 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2114 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2120 bridge_enqueue(sc, ifp, m);
2126 * Send output from a bridge member interface. This
2127 * performs the bridging function for locally originated
2130 * The mbuf has the Ethernet header already attached. We must
2131 * enqueue or free the mbuf before returning.
2134 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2137 struct ether_header *eh;
2138 struct ifnet *bifp, *dst_if;
2139 struct bridge_softc *sc;
2144 if (m->m_len < ETHER_HDR_LEN) {
2145 m = m_pullup(m, ETHER_HDR_LEN);
2150 eh = mtod(m, struct ether_header *);
2151 sc = ifp->if_bridge;
2152 vlan = VLANTAGOF(m);
2157 * If bridge is down, but the original output interface is up,
2158 * go ahead and send out that interface. Otherwise, the packet
2161 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2167 * If the packet is a multicast, or we don't know a better way to
2168 * get there, send to all interfaces.
2170 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2173 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2174 /* Tap any traffic not passing back out the originating interface */
2176 ETHER_BPF_MTAP(bifp, m);
2177 if (dst_if == NULL) {
2178 struct bridge_iflist *bif;
2184 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2185 dst_if = bif->bif_ifp;
2187 if (dst_if->if_type == IFT_GIF)
2189 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2193 * If this is not the original output interface,
2194 * and the interface is participating in spanning
2195 * tree, make sure the port is in a state that
2196 * allows forwarding.
2198 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2199 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2202 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2206 mc = m_dup(m, M_NOWAIT);
2208 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2213 bridge_enqueue(sc, dst_if, mc);
2222 * XXX Spanning tree consideration here?
2226 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2231 bridge_enqueue(sc, dst_if, m);
2238 * Do output on a bridge.
2242 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2244 struct bridge_softc *sc;
2245 struct ether_header *eh;
2246 struct ifnet *dst_if;
2251 ETHER_BPF_MTAP(ifp, m);
2253 eh = mtod(m, struct ether_header *);
2255 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2256 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2257 error = bridge_enqueue(sc, dst_if, m);
2259 bridge_broadcast(sc, ifp, m, 0);
2266 bridge_altq_start(if_t ifp)
2268 struct ifaltq *ifq = &ifp->if_snd;
2272 IFQ_DEQUEUE_NOLOCK(ifq, m);
2274 bridge_transmit(ifp, m);
2275 IFQ_DEQUEUE_NOLOCK(ifq, m);
2281 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2285 if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2286 IFQ_ENQUEUE(&ifp->if_snd, m, err);
2288 bridge_altq_start(ifp);
2290 err = bridge_transmit(ifp, m);
2297 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2300 bridge_qflush(struct ifnet *ifp __unused)
2307 * The forwarding function of the bridge.
2309 * NOTE: Releases the lock on return.
2312 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2315 struct bridge_iflist *dbif;
2316 struct ifnet *src_if, *dst_if, *ifp;
2317 struct ether_header *eh;
2324 src_if = m->m_pkthdr.rcvif;
2327 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2328 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2329 vlan = VLANTAGOF(m);
2331 if ((sbif->bif_flags & IFBIF_STP) &&
2332 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2335 eh = mtod(m, struct ether_header *);
2336 dst = eh->ether_dhost;
2338 /* If the interface is learning, record the address. */
2339 if (sbif->bif_flags & IFBIF_LEARNING) {
2340 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2341 sbif, 0, IFBAF_DYNAMIC);
2343 * If the interface has addresses limits then deny any source
2344 * that is not in the cache.
2346 if (error && sbif->bif_addrmax)
2350 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2351 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2355 * At this point, the port either doesn't participate
2356 * in spanning tree or it is in the forwarding state.
2360 * If the packet is unicast, destined for someone on
2361 * "this" side of the bridge, drop it.
2363 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2364 dst_if = bridge_rtlookup(sc, dst, vlan);
2365 if (src_if == dst_if)
2369 * Check if its a reserved multicast address, any address
2370 * listed in 802.1D section 7.12.6 may not be forwarded by the
2372 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2374 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2375 dst[2] == 0xc2 && dst[3] == 0x00 &&
2376 dst[4] == 0x00 && dst[5] <= 0x0f)
2379 /* ...forward it to all interfaces. */
2380 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2385 * If we have a destination interface which is a member of our bridge,
2386 * OR this is a unicast packet, push it through the bpf(4) machinery.
2387 * For broadcast or multicast packets, don't bother because it will
2388 * be reinjected into ether_input. We do this before we pass the packets
2389 * through the pfil(9) framework, as it is possible that pfil(9) will
2390 * drop the packet, or possibly modify it, making it difficult to debug
2391 * firewall issues on the bridge.
2393 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2394 ETHER_BPF_MTAP(ifp, m);
2396 /* run the packet filter */
2397 if (PFIL_HOOKED_IN(V_inet_pfil_head)
2399 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2402 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2408 if (dst_if == NULL) {
2409 bridge_broadcast(sc, src_if, m, 1);
2414 * At this point, we're dealing with a unicast frame
2415 * going to a different interface.
2417 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2420 dbif = bridge_lookup_member_if(sc, dst_if);
2422 /* Not a member of the bridge (anymore?) */
2425 /* Private segments can not talk to each other */
2426 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2429 if ((dbif->bif_flags & IFBIF_STP) &&
2430 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2433 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2435 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2438 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2444 bridge_enqueue(sc, dst_if, m);
2454 * Receive input from a member interface. Queue the packet for
2455 * bridging if it is not for us.
2457 static struct mbuf *
2458 bridge_input(struct ifnet *ifp, struct mbuf *m)
2460 struct bridge_softc *sc = ifp->if_bridge;
2461 struct bridge_iflist *bif, *bif2;
2463 struct ether_header *eh;
2464 struct mbuf *mc, *mc2;
2470 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2474 vlan = VLANTAGOF(m);
2477 * Implement support for bridge monitoring. If this flag has been
2478 * set on this interface, discard the packet once we push it through
2479 * the bpf(4) machinery, but before we do, increment the byte and
2480 * packet counters associated with this interface.
2482 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2483 m->m_pkthdr.rcvif = bifp;
2484 ETHER_BPF_MTAP(bifp, m);
2485 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2486 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2490 bif = bridge_lookup_member_if(sc, ifp);
2495 eh = mtod(m, struct ether_header *);
2499 if (m->m_flags & (M_BCAST|M_MCAST)) {
2500 /* Tap off 802.1D packets; they do not get forwarded. */
2501 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2502 ETHER_ADDR_LEN) == 0) {
2503 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2507 if ((bif->bif_flags & IFBIF_STP) &&
2508 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2513 * Make a deep copy of the packet and enqueue the copy
2514 * for bridge processing; return the original packet for
2517 mc = m_dup(m, M_NOWAIT);
2522 /* Perform the bridge forwarding function with the copy. */
2523 bridge_forward(sc, bif, mc);
2526 * Reinject the mbuf as arriving on the bridge so we have a
2527 * chance at claiming multicast packets. We can not loop back
2528 * here from ether_input as a bridge is never a member of a
2531 KASSERT(bifp->if_bridge == NULL,
2532 ("loop created in bridge_input"));
2533 mc2 = m_dup(m, M_NOWAIT);
2535 /* Keep the layer3 header aligned */
2536 int i = min(mc2->m_pkthdr.len, max_protohdr);
2537 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2540 mc2->m_pkthdr.rcvif = bifp;
2541 (*bifp->if_input)(bifp, mc2);
2544 /* Return the original packet for local processing. */
2548 if ((bif->bif_flags & IFBIF_STP) &&
2549 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2553 #if (defined(INET) || defined(INET6))
2554 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2555 || ((iface)->if_carp \
2556 && (*carp_forus_p)((iface), eh->ether_dhost))
2557 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2558 || ((iface)->if_carp \
2559 && (*carp_forus_p)((iface), eh->ether_shost))
2561 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2562 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2566 # define OR_PFIL_HOOKED_INET6 \
2567 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2569 # define OR_PFIL_HOOKED_INET6
2572 #define GRAB_OUR_PACKETS(iface) \
2573 if ((iface)->if_type == IFT_GIF) \
2575 /* It is destined for us. */ \
2576 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2577 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2579 if (bif->bif_flags & IFBIF_LEARNING) { \
2580 error = bridge_rtupdate(sc, eh->ether_shost, \
2581 vlan, bif, 0, IFBAF_DYNAMIC); \
2582 if (error && bif->bif_addrmax) { \
2587 m->m_pkthdr.rcvif = iface; \
2588 if ((iface) == ifp) { \
2589 /* Skip bridge processing... src == dest */ \
2592 /* It's passing over or to the bridge, locally. */ \
2593 ETHER_BPF_MTAP(bifp, m); \
2594 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \
2595 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2596 /* Filter on the physical interface. */ \
2597 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2598 OR_PFIL_HOOKED_INET6)) { \
2599 if (bridge_pfil(&m, NULL, ifp, \
2600 PFIL_IN) != 0 || m == NULL) { \
2604 if ((iface) != bifp) \
2605 ETHER_BPF_MTAP(iface, m); \
2609 /* We just received a packet that we sent out. */ \
2610 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2611 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2618 * Unicast. Make sure it's not for the bridge.
2620 do { GRAB_OUR_PACKETS(bifp) } while (0);
2623 * Give a chance for ifp at first priority. This will help when the
2624 * packet comes through the interface like VLAN's with the same MACs
2625 * on several interfaces from the same bridge. This also will save
2626 * some CPU cycles in case the destination interface and the input
2627 * interface (eq ifp) are the same.
2629 do { GRAB_OUR_PACKETS(ifp) } while (0);
2631 /* Now check the all bridge members. */
2632 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2633 GRAB_OUR_PACKETS(bif2->bif_ifp)
2636 #undef OR_CARP_CHECK_WE_ARE_DST
2637 #undef OR_CARP_CHECK_WE_ARE_SRC
2638 #undef OR_PFIL_HOOKED_INET6
2639 #undef GRAB_OUR_PACKETS
2641 /* Perform the bridge forwarding function. */
2642 bridge_forward(sc, bif, m);
2650 * Send a frame to all interfaces that are members of
2651 * the bridge, except for the one on which the packet
2654 * NOTE: Releases the lock on return.
2657 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2658 struct mbuf *m, int runfilt)
2660 struct bridge_iflist *dbif, *sbif;
2662 struct ifnet *dst_if;
2667 sbif = bridge_lookup_member_if(sc, src_if);
2669 /* Filter on the bridge interface before broadcasting */
2670 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2672 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2675 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2681 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2682 dst_if = dbif->bif_ifp;
2683 if (dst_if == src_if)
2686 /* Private segments can not talk to each other */
2687 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2690 if ((dbif->bif_flags & IFBIF_STP) &&
2691 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2694 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2695 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2698 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2701 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2705 mc = m_dup(m, M_NOWAIT);
2707 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2713 * Filter on the output interface. Pass a NULL bridge interface
2714 * pointer so we do not redundantly filter on the bridge for
2715 * each interface we broadcast on.
2717 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2719 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2723 /* Keep the layer3 header aligned */
2724 i = min(mc->m_pkthdr.len, max_protohdr);
2725 mc = m_copyup(mc, i, ETHER_ALIGN);
2727 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2731 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2737 bridge_enqueue(sc, dst_if, mc);
2746 * Duplicate a packet out one or more interfaces that are in span mode,
2747 * the original mbuf is unmodified.
2750 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2752 struct bridge_iflist *bif;
2753 struct ifnet *dst_if;
2758 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2761 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2762 dst_if = bif->bif_ifp;
2764 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2767 mc = m_dup(m, M_NOWAIT);
2769 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2773 bridge_enqueue(sc, dst_if, mc);
2780 * Add a bridge routing entry.
2783 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2784 struct bridge_iflist *bif, int setflags, uint8_t flags)
2786 struct bridge_rtnode *brt;
2789 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2791 /* Check the source address is valid and not multicast. */
2792 if (ETHER_IS_MULTICAST(dst) ||
2793 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2794 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2797 /* 802.1p frames map to vlan 1 */
2802 * A route for this destination might already exist. If so,
2803 * update it, otherwise create a new one.
2805 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2808 /* Check again, now that we have the lock. There could have
2809 * been a race and we only want to insert this once. */
2810 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2811 BRIDGE_RT_UNLOCK(sc);
2815 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2816 sc->sc_brtexceeded++;
2817 BRIDGE_RT_UNLOCK(sc);
2820 /* Check per interface address limits (if enabled) */
2821 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2822 bif->bif_addrexceeded++;
2823 BRIDGE_RT_UNLOCK(sc);
2828 * Allocate a new bridge forwarding node, and
2829 * initialize the expiration time and Ethernet
2832 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2834 BRIDGE_RT_UNLOCK(sc);
2837 brt->brt_vnet = curvnet;
2839 if (bif->bif_flags & IFBIF_STICKY)
2840 brt->brt_flags = IFBAF_STICKY;
2842 brt->brt_flags = IFBAF_DYNAMIC;
2844 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2845 brt->brt_vlan = vlan;
2847 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2848 uma_zfree(V_bridge_rtnode_zone, brt);
2849 BRIDGE_RT_UNLOCK(sc);
2855 BRIDGE_RT_UNLOCK(sc);
2858 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2859 brt->brt_dst != bif) {
2861 brt->brt_dst->bif_addrcnt--;
2863 brt->brt_dst->bif_addrcnt++;
2864 BRIDGE_RT_UNLOCK(sc);
2867 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2868 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2870 brt->brt_flags = flags;
2878 * Lookup the destination interface for an address.
2880 static struct ifnet *
2881 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2883 struct bridge_rtnode *brt;
2887 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2890 return (brt->brt_ifp);
2896 * Trim the routine table so that we have a number
2897 * of routing entries less than or equal to the
2901 bridge_rttrim(struct bridge_softc *sc)
2903 struct bridge_rtnode *brt, *nbrt;
2906 BRIDGE_RT_LOCK_ASSERT(sc);
2908 /* Make sure we actually need to do this. */
2909 if (sc->sc_brtcnt <= sc->sc_brtmax)
2912 /* Force an aging cycle; this might trim enough addresses. */
2914 if (sc->sc_brtcnt <= sc->sc_brtmax)
2917 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2918 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2919 bridge_rtnode_destroy(sc, brt);
2920 if (sc->sc_brtcnt <= sc->sc_brtmax)
2929 * Aging timer for the bridge.
2932 bridge_timer(void *arg)
2934 struct bridge_softc *sc = arg;
2936 BRIDGE_RT_LOCK_ASSERT(sc);
2938 /* Destruction of rtnodes requires a proper vnet context */
2939 CURVNET_SET(sc->sc_ifp->if_vnet);
2942 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2943 callout_reset(&sc->sc_brcallout,
2944 bridge_rtable_prune_period * hz, bridge_timer, sc);
2951 * Perform an aging cycle.
2954 bridge_rtage(struct bridge_softc *sc)
2956 struct bridge_rtnode *brt, *nbrt;
2958 BRIDGE_RT_LOCK_ASSERT(sc);
2960 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2961 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2962 if (time_uptime >= brt->brt_expire)
2963 bridge_rtnode_destroy(sc, brt);
2971 * Remove all dynamic addresses from the bridge.
2974 bridge_rtflush(struct bridge_softc *sc, int full)
2976 struct bridge_rtnode *brt, *nbrt;
2978 BRIDGE_RT_LOCK_ASSERT(sc);
2980 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2981 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2982 bridge_rtnode_destroy(sc, brt);
2989 * Remove an address from the table.
2992 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2994 struct bridge_rtnode *brt;
3000 * If vlan is zero then we want to delete for all vlans so the lookup
3001 * may return more than one.
3003 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
3004 bridge_rtnode_destroy(sc, brt);
3008 BRIDGE_RT_UNLOCK(sc);
3010 return (found ? 0 : ENOENT);
3016 * Delete routes to a speicifc member interface.
3019 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3021 struct bridge_rtnode *brt, *nbrt;
3023 BRIDGE_RT_LOCK_ASSERT(sc);
3025 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3026 if (brt->brt_ifp == ifp && (full ||
3027 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3028 bridge_rtnode_destroy(sc, brt);
3033 * bridge_rtable_init:
3035 * Initialize the route table for this bridge.
3038 bridge_rtable_init(struct bridge_softc *sc)
3042 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3043 M_DEVBUF, M_WAITOK);
3045 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3046 CK_LIST_INIT(&sc->sc_rthash[i]);
3048 sc->sc_rthash_key = arc4random();
3049 CK_LIST_INIT(&sc->sc_rtlist);
3053 * bridge_rtable_fini:
3055 * Deconstruct the route table for this bridge.
3058 bridge_rtable_fini(struct bridge_softc *sc)
3061 KASSERT(sc->sc_brtcnt == 0,
3062 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3063 free(sc->sc_rthash, M_DEVBUF);
3067 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3068 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3070 #define mix(a, b, c) \
3072 a -= b; a -= c; a ^= (c >> 13); \
3073 b -= c; b -= a; b ^= (a << 8); \
3074 c -= a; c -= b; c ^= (b >> 13); \
3075 a -= b; a -= c; a ^= (c >> 12); \
3076 b -= c; b -= a; b ^= (a << 16); \
3077 c -= a; c -= b; c ^= (b >> 5); \
3078 a -= b; a -= c; a ^= (c >> 3); \
3079 b -= c; b -= a; b ^= (a << 10); \
3080 c -= a; c -= b; c ^= (b >> 15); \
3081 } while (/*CONSTCOND*/0)
3083 static __inline uint32_t
3084 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3086 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3097 return (c & BRIDGE_RTHASH_MASK);
3103 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3107 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3108 d = ((int)a[i]) - ((int)b[i]);
3115 * bridge_rtnode_lookup:
3117 * Look up a bridge route node for the specified destination. Compare the
3118 * vlan id or if zero then just return the first match.
3120 static struct bridge_rtnode *
3121 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3123 struct bridge_rtnode *brt;
3127 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3129 hash = bridge_rthash(sc, addr);
3130 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3131 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3132 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3142 * bridge_rtnode_insert:
3144 * Insert the specified bridge node into the route table. We
3145 * assume the entry is not already in the table.
3148 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3150 struct bridge_rtnode *lbrt;
3154 BRIDGE_RT_LOCK_ASSERT(sc);
3156 hash = bridge_rthash(sc, brt->brt_addr);
3158 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3160 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3165 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3166 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3169 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3172 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3173 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3176 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3177 } while (lbrt != NULL);
3180 panic("bridge_rtnode_insert: impossible");
3184 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3191 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3193 struct bridge_rtnode *brt;
3195 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3197 CURVNET_SET(brt->brt_vnet);
3198 uma_zfree(V_bridge_rtnode_zone, brt);
3203 * bridge_rtnode_destroy:
3205 * Destroy a bridge rtnode.
3208 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3210 BRIDGE_RT_LOCK_ASSERT(sc);
3212 CK_LIST_REMOVE(brt, brt_hash);
3214 CK_LIST_REMOVE(brt, brt_list);
3216 brt->brt_dst->bif_addrcnt--;
3218 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3222 * bridge_rtable_expire:
3224 * Set the expiry time for all routes on an interface.
3227 bridge_rtable_expire(struct ifnet *ifp, int age)
3229 struct bridge_softc *sc = ifp->if_bridge;
3230 struct bridge_rtnode *brt;
3232 CURVNET_SET(ifp->if_vnet);
3236 * If the age is zero then flush, otherwise set all the expiry times to
3237 * age for the interface
3240 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3242 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3243 /* Cap the expiry time to 'age' */
3244 if (brt->brt_ifp == ifp &&
3245 brt->brt_expire > time_uptime + age &&
3246 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3247 brt->brt_expire = time_uptime + age;
3250 BRIDGE_RT_UNLOCK(sc);
3255 * bridge_state_change:
3257 * Callback from the bridgestp code when a port changes states.
3260 bridge_state_change(struct ifnet *ifp, int state)
3262 struct bridge_softc *sc = ifp->if_bridge;
3263 static const char *stpstates[] = {
3272 CURVNET_SET(ifp->if_vnet);
3274 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3275 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3280 * Send bridge packets through pfil if they are one of the types pfil can deal
3281 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3282 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3286 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3288 int snap, error, i, hlen;
3289 struct ether_header *eh1, eh2;
3292 u_int16_t ether_type;
3296 error = -1; /* Default error if not error == 0 */
3299 /* we may return with the IP fields swapped, ensure its not shared */
3300 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3303 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3304 return (0); /* filtering is disabled */
3306 i = min((*mp)->m_pkthdr.len, max_protohdr);
3307 if ((*mp)->m_len < i) {
3308 *mp = m_pullup(*mp, i);
3310 printf("%s: m_pullup failed\n", __func__);
3315 eh1 = mtod(*mp, struct ether_header *);
3316 ether_type = ntohs(eh1->ether_type);
3319 * Check for SNAP/LLC.
3321 if (ether_type < ETHERMTU) {
3322 struct llc *llc2 = (struct llc *)(eh1 + 1);
3324 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3325 llc2->llc_dsap == LLC_SNAP_LSAP &&
3326 llc2->llc_ssap == LLC_SNAP_LSAP &&
3327 llc2->llc_control == LLC_UI) {
3328 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3334 * If we're trying to filter bridge traffic, don't look at anything
3335 * other than IP and ARP traffic. If the filter doesn't understand
3336 * IPv6, don't allow IPv6 through the bridge either. This is lame
3337 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3338 * but of course we don't have an AppleTalk filter to begin with.
3339 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3342 switch (ether_type) {
3344 case ETHERTYPE_REVARP:
3345 if (V_pfil_ipfw_arp == 0)
3346 return (0); /* Automatically pass */
3351 case ETHERTYPE_IPV6:
3356 * Check to see if the user wants to pass non-ip
3357 * packets, these will not be checked by pfil(9) and
3358 * passed unconditionally so the default is to drop.
3364 /* Run the packet through pfil before stripping link headers */
3365 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3366 dir == PFIL_OUT && ifp != NULL) {
3367 switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3375 /* Strip off the Ethernet header and keep a copy. */
3376 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3377 m_adj(*mp, ETHER_HDR_LEN);
3379 /* Strip off snap header, if present */
3381 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3382 m_adj(*mp, sizeof(struct llc));
3386 * Check the IP header for alignment and errors
3388 if (dir == PFIL_IN) {
3389 switch (ether_type) {
3391 error = bridge_ip_checkbasic(mp);
3394 case ETHERTYPE_IPV6:
3395 error = bridge_ip6_checkbasic(mp);
3408 * Run the packet through pfil
3411 switch (ether_type) {
3414 * Run pfil on the member interface and the bridge, both can
3415 * be skipped by clearing pfil_member or pfil_bridge.
3418 * in_if -> bridge_if -> out_if
3420 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3421 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3425 if (V_pfil_member && ifp != NULL && (rv =
3426 pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3430 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3431 pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3435 /* check if we need to fragment the packet */
3436 /* bridge_fragment generates a mbuf chain of packets */
3437 /* that already include eth headers */
3438 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3439 i = (*mp)->m_pkthdr.len;
3440 if (i > ifp->if_mtu) {
3441 error = bridge_fragment(ifp, mp, &eh2, snap,
3447 /* Recalculate the ip checksum. */
3448 ip = mtod(*mp, struct ip *);
3449 hlen = ip->ip_hl << 2;
3450 if (hlen < sizeof(struct ip))
3452 if (hlen > (*mp)->m_len) {
3453 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3455 ip = mtod(*mp, struct ip *);
3460 if (hlen == sizeof(struct ip))
3461 ip->ip_sum = in_cksum_hdr(ip);
3463 ip->ip_sum = in_cksum(*mp, hlen);
3467 case ETHERTYPE_IPV6:
3468 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3469 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3473 if (V_pfil_member && ifp != NULL && (rv =
3474 pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3478 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3479 pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3498 * Finally, put everything back the way it was and return
3501 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3504 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3507 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3510 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3521 * Perform basic checks on header size since
3522 * pfil assumes ip_input has already processed
3523 * it for it. Cut-and-pasted from ip_input.c.
3524 * Given how simple the IPv6 version is,
3525 * does the IPv4 version really need to be
3528 * XXX Should we update ipstat here, or not?
3529 * XXX Right now we update ipstat but not
3533 bridge_ip_checkbasic(struct mbuf **mp)
3535 struct mbuf *m = *mp;
3543 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3544 if ((m = m_copyup(m, sizeof(struct ip),
3545 (max_linkhdr + 3) & ~3)) == NULL) {
3546 /* XXXJRT new stat, please */
3547 KMOD_IPSTAT_INC(ips_toosmall);
3550 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3551 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3552 KMOD_IPSTAT_INC(ips_toosmall);
3556 ip = mtod(m, struct ip *);
3557 if (ip == NULL) goto bad;
3559 if (ip->ip_v != IPVERSION) {
3560 KMOD_IPSTAT_INC(ips_badvers);
3563 hlen = ip->ip_hl << 2;
3564 if (hlen < sizeof(struct ip)) { /* minimum header length */
3565 KMOD_IPSTAT_INC(ips_badhlen);
3568 if (hlen > m->m_len) {
3569 if ((m = m_pullup(m, hlen)) == NULL) {
3570 KMOD_IPSTAT_INC(ips_badhlen);
3573 ip = mtod(m, struct ip *);
3574 if (ip == NULL) goto bad;
3577 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3578 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3580 if (hlen == sizeof(struct ip)) {
3581 sum = in_cksum_hdr(ip);
3583 sum = in_cksum(m, hlen);
3587 KMOD_IPSTAT_INC(ips_badsum);
3591 /* Retrieve the packet length. */
3592 len = ntohs(ip->ip_len);
3595 * Check for additional length bogosity
3598 KMOD_IPSTAT_INC(ips_badlen);
3603 * Check that the amount of data in the buffers
3604 * is as at least much as the IP header would have us expect.
3605 * Drop packet if shorter than we expect.
3607 if (m->m_pkthdr.len < len) {
3608 KMOD_IPSTAT_INC(ips_tooshort);
3612 /* Checks out, proceed */
3623 * Same as above, but for IPv6.
3624 * Cut-and-pasted from ip6_input.c.
3625 * XXX Should we update ip6stat, or not?
3628 bridge_ip6_checkbasic(struct mbuf **mp)
3630 struct mbuf *m = *mp;
3631 struct ip6_hdr *ip6;
3634 * If the IPv6 header is not aligned, slurp it up into a new
3635 * mbuf with space for link headers, in the event we forward
3636 * it. Otherwise, if it is aligned, make sure the entire base
3637 * IPv6 header is in the first mbuf of the chain.
3639 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3640 struct ifnet *inifp = m->m_pkthdr.rcvif;
3641 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3642 (max_linkhdr + 3) & ~3)) == NULL) {
3643 /* XXXJRT new stat, please */
3644 IP6STAT_INC(ip6s_toosmall);
3645 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3648 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3649 struct ifnet *inifp = m->m_pkthdr.rcvif;
3650 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3651 IP6STAT_INC(ip6s_toosmall);
3652 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3657 ip6 = mtod(m, struct ip6_hdr *);
3659 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3660 IP6STAT_INC(ip6s_badvers);
3661 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3665 /* Checks out, proceed */
3678 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3681 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3682 int snap, struct llc *llc)
3684 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3688 if (m->m_len < sizeof(struct ip) &&
3689 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3691 ip = mtod(m, struct ip *);
3693 m->m_pkthdr.csum_flags |= CSUM_IP;
3694 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3699 * Walk the chain and re-add the Ethernet header for
3702 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3703 nextpkt = mcur->m_nextpkt;
3704 mcur->m_nextpkt = NULL;
3706 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3710 mprev->m_nextpkt = nextpkt;
3713 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3716 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3720 mprev->m_nextpkt = nextpkt;
3723 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3726 * The previous two M_PREPEND could have inserted one or two
3727 * mbufs in front so we have to update the previous packet's
3730 mcur->m_nextpkt = nextpkt;
3732 mprev->m_nextpkt = mcur;
3734 /* The first mbuf in the original chain needs to be
3741 KMOD_IPSTAT_INC(ips_fragmented);
3745 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3746 m = mcur->m_nextpkt;
3753 bridge_linkstate(struct ifnet *ifp)
3755 struct bridge_softc *sc = ifp->if_bridge;
3756 struct bridge_iflist *bif;
3757 struct epoch_tracker et;
3759 NET_EPOCH_ENTER(et);
3761 bif = bridge_lookup_member_if(sc, ifp);
3766 bridge_linkcheck(sc);
3768 bstp_linkstate(&bif->bif_stp);
3774 bridge_linkcheck(struct bridge_softc *sc)
3776 struct bridge_iflist *bif;
3777 int new_link, hasls;
3779 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3781 new_link = LINK_STATE_DOWN;
3783 /* Our link is considered up if at least one of our ports is active */
3784 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3785 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3787 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3788 new_link = LINK_STATE_UP;
3792 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3793 /* If no interfaces support link-state then we default to up */
3794 new_link = LINK_STATE_UP;
3796 if_link_state_change(sc->sc_ifp, new_link);