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/if_private.h>
114 #include <net/pfil.h>
115 #include <net/vnet.h>
117 #include <netinet/in.h>
118 #include <netinet/in_systm.h>
119 #include <netinet/in_var.h>
120 #include <netinet/ip.h>
121 #include <netinet/ip_var.h>
123 #include <netinet/ip6.h>
124 #include <netinet6/ip6_var.h>
125 #include <netinet6/in6_ifattach.h>
127 #if defined(INET) || defined(INET6)
128 #include <netinet/ip_carp.h>
130 #include <machine/in_cksum.h>
131 #include <netinet/if_ether.h>
132 #include <net/bridgestp.h>
133 #include <net/if_bridgevar.h>
134 #include <net/if_llc.h>
135 #include <net/if_vlan_var.h>
137 #include <net/route.h>
140 * Size of the route hash table. Must be a power of two.
142 #ifndef BRIDGE_RTHASH_SIZE
143 #define BRIDGE_RTHASH_SIZE 1024
146 #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1)
149 * Default maximum number of addresses to cache.
151 #ifndef BRIDGE_RTABLE_MAX
152 #define BRIDGE_RTABLE_MAX 2000
156 * Timeout (in seconds) for entries learned dynamically.
158 #ifndef BRIDGE_RTABLE_TIMEOUT
159 #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */
163 * Number of seconds between walks of the route list.
165 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
166 #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60)
170 * List of capabilities to possibly mask on the member interface.
172 #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
176 * List of capabilities to strip
178 #define BRIDGE_IFCAPS_STRIP IFCAP_LRO
183 * The bridge relies heavily on the epoch(9) system to protect its data
184 * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we
185 * must ensure there is only one writer at a time.
187 * That is: for read accesses we only need to be in NET_EPOCH, but for write
188 * accesses we must hold:
190 * - BRIDGE_RT_LOCK, for any change to bridge_rtnodes
191 * - BRIDGE_LOCK, for any other change
193 * The BRIDGE_LOCK is a sleepable lock, because it is held across ioctl()
194 * calls to bridge member interfaces and these ioctl()s can sleep.
195 * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes
196 * required while we're in NET_EPOCH and then we're not allowed to sleep.
198 #define BRIDGE_LOCK_INIT(_sc) do { \
199 sx_init(&(_sc)->sc_sx, "if_bridge"); \
200 mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF); \
202 #define BRIDGE_LOCK_DESTROY(_sc) do { \
203 sx_destroy(&(_sc)->sc_sx); \
204 mtx_destroy(&(_sc)->sc_rt_mtx); \
206 #define BRIDGE_LOCK(_sc) sx_xlock(&(_sc)->sc_sx)
207 #define BRIDGE_UNLOCK(_sc) sx_xunlock(&(_sc)->sc_sx)
208 #define BRIDGE_LOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_XLOCKED)
209 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
210 MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx))
211 #define BRIDGE_UNLOCK_ASSERT(_sc) sx_assert(&(_sc)->sc_sx, SX_UNLOCKED)
212 #define BRIDGE_RT_LOCK(_sc) mtx_lock(&(_sc)->sc_rt_mtx)
213 #define BRIDGE_RT_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_rt_mtx)
214 #define BRIDGE_RT_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED)
215 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc) \
216 MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx))
219 * Bridge interface list entry.
221 struct bridge_iflist {
222 CK_LIST_ENTRY(bridge_iflist) bif_next;
223 struct ifnet *bif_ifp; /* member if */
224 struct bstp_port bif_stp; /* STP state */
225 uint32_t bif_flags; /* member if flags */
226 int bif_savedcaps; /* saved capabilities */
227 uint32_t bif_addrmax; /* max # of addresses */
228 uint32_t bif_addrcnt; /* cur. # of addresses */
229 uint32_t bif_addrexceeded;/* # of address violations */
230 struct epoch_context bif_epoch_ctx;
236 struct bridge_rtnode {
237 CK_LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */
238 CK_LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */
239 struct bridge_iflist *brt_dst; /* destination if */
240 unsigned long brt_expire; /* expiration time */
241 uint8_t brt_flags; /* address flags */
242 uint8_t brt_addr[ETHER_ADDR_LEN];
243 uint16_t brt_vlan; /* vlan id */
244 struct vnet *brt_vnet;
245 struct epoch_context brt_epoch_ctx;
247 #define brt_ifp brt_dst->bif_ifp
250 * Software state for each bridge.
252 struct bridge_softc {
253 struct ifnet *sc_ifp; /* make this an interface */
254 LIST_ENTRY(bridge_softc) sc_list;
256 struct mtx sc_rt_mtx;
257 uint32_t sc_brtmax; /* max # of addresses */
258 uint32_t sc_brtcnt; /* cur. # of addresses */
259 uint32_t sc_brttimeout; /* rt timeout in seconds */
260 struct callout sc_brcallout; /* bridge callout */
261 CK_LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */
262 CK_LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */
263 CK_LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */
264 uint32_t sc_rthash_key; /* key for hash */
265 CK_LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */
266 struct bstp_state sc_stp; /* STP state */
267 uint32_t sc_brtexceeded; /* # of cache drops */
268 struct ifnet *sc_ifaddr; /* member mac copied from */
269 struct ether_addr sc_defaddr; /* Default MAC address */
270 struct epoch_context sc_epoch_ctx;
273 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
274 #define V_bridge_list_sx VNET(bridge_list_sx)
275 static eventhandler_tag bridge_detach_cookie;
277 int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
279 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
280 #define V_bridge_rtnode_zone VNET(bridge_rtnode_zone)
282 static int bridge_clone_create(struct if_clone *, char *, size_t,
283 struct ifc_data *, struct ifnet **);
284 static int bridge_clone_destroy(struct if_clone *, struct ifnet *, uint32_t);
286 static int bridge_ioctl(struct ifnet *, u_long, caddr_t);
287 static void bridge_mutecaps(struct bridge_softc *);
288 static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
290 static void bridge_ifdetach(void *arg __unused, struct ifnet *);
291 static void bridge_init(void *);
292 static void bridge_dummynet(struct mbuf *, struct ifnet *);
293 static void bridge_stop(struct ifnet *, int);
294 static int bridge_transmit(struct ifnet *, struct mbuf *);
296 static void bridge_altq_start(if_t);
297 static int bridge_altq_transmit(if_t, struct mbuf *);
299 static void bridge_qflush(struct ifnet *);
300 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
301 static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
303 static int bridge_enqueue(struct bridge_softc *, struct ifnet *,
305 static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
307 static void bridge_forward(struct bridge_softc *, struct bridge_iflist *,
310 static void bridge_timer(void *);
312 static void bridge_broadcast(struct bridge_softc *, struct ifnet *,
314 static void bridge_span(struct bridge_softc *, struct mbuf *);
316 static int bridge_rtupdate(struct bridge_softc *, const uint8_t *,
317 uint16_t, struct bridge_iflist *, int, uint8_t);
318 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
320 static void bridge_rttrim(struct bridge_softc *);
321 static void bridge_rtage(struct bridge_softc *);
322 static void bridge_rtflush(struct bridge_softc *, int);
323 static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
326 static void bridge_rtable_init(struct bridge_softc *);
327 static void bridge_rtable_fini(struct bridge_softc *);
329 static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
330 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
331 const uint8_t *, uint16_t);
332 static int bridge_rtnode_insert(struct bridge_softc *,
333 struct bridge_rtnode *);
334 static void bridge_rtnode_destroy(struct bridge_softc *,
335 struct bridge_rtnode *);
336 static void bridge_rtable_expire(struct ifnet *, int);
337 static void bridge_state_change(struct ifnet *, int);
339 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
341 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
343 static void bridge_delete_member(struct bridge_softc *,
344 struct bridge_iflist *, int);
345 static void bridge_delete_span(struct bridge_softc *,
346 struct bridge_iflist *);
348 static int bridge_ioctl_add(struct bridge_softc *, void *);
349 static int bridge_ioctl_del(struct bridge_softc *, void *);
350 static int bridge_ioctl_gifflags(struct bridge_softc *, void *);
351 static int bridge_ioctl_sifflags(struct bridge_softc *, void *);
352 static int bridge_ioctl_scache(struct bridge_softc *, void *);
353 static int bridge_ioctl_gcache(struct bridge_softc *, void *);
354 static int bridge_ioctl_gifs(struct bridge_softc *, void *);
355 static int bridge_ioctl_rts(struct bridge_softc *, void *);
356 static int bridge_ioctl_saddr(struct bridge_softc *, void *);
357 static int bridge_ioctl_sto(struct bridge_softc *, void *);
358 static int bridge_ioctl_gto(struct bridge_softc *, void *);
359 static int bridge_ioctl_daddr(struct bridge_softc *, void *);
360 static int bridge_ioctl_flush(struct bridge_softc *, void *);
361 static int bridge_ioctl_gpri(struct bridge_softc *, void *);
362 static int bridge_ioctl_spri(struct bridge_softc *, void *);
363 static int bridge_ioctl_ght(struct bridge_softc *, void *);
364 static int bridge_ioctl_sht(struct bridge_softc *, void *);
365 static int bridge_ioctl_gfd(struct bridge_softc *, void *);
366 static int bridge_ioctl_sfd(struct bridge_softc *, void *);
367 static int bridge_ioctl_gma(struct bridge_softc *, void *);
368 static int bridge_ioctl_sma(struct bridge_softc *, void *);
369 static int bridge_ioctl_sifprio(struct bridge_softc *, void *);
370 static int bridge_ioctl_sifcost(struct bridge_softc *, void *);
371 static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
372 static int bridge_ioctl_addspan(struct bridge_softc *, void *);
373 static int bridge_ioctl_delspan(struct bridge_softc *, void *);
374 static int bridge_ioctl_gbparam(struct bridge_softc *, void *);
375 static int bridge_ioctl_grte(struct bridge_softc *, void *);
376 static int bridge_ioctl_gifsstp(struct bridge_softc *, void *);
377 static int bridge_ioctl_sproto(struct bridge_softc *, void *);
378 static int bridge_ioctl_stxhc(struct bridge_softc *, void *);
379 static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
381 static int bridge_ip_checkbasic(struct mbuf **mp);
383 static int bridge_ip6_checkbasic(struct mbuf **mp);
385 static int bridge_fragment(struct ifnet *, struct mbuf **mp,
386 struct ether_header *, int, struct llc *);
387 static void bridge_linkstate(struct ifnet *ifp);
388 static void bridge_linkcheck(struct bridge_softc *sc);
390 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
391 #define VLANTAGOF(_m) \
392 (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
394 static struct bstp_cb_ops bridge_ops = {
395 .bcb_state = bridge_state_change,
396 .bcb_rtage = bridge_rtable_expire
399 SYSCTL_DECL(_net_link);
400 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
403 /* only pass IP[46] packets when pfil is enabled */
404 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
405 #define V_pfil_onlyip VNET(pfil_onlyip)
406 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
407 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
408 "Only pass IP packets when pfil is enabled");
410 /* run pfil hooks on the bridge interface */
411 VNET_DEFINE_STATIC(int, pfil_bridge) = 0;
412 #define V_pfil_bridge VNET(pfil_bridge)
413 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
414 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
415 "Packet filter on the bridge interface");
417 /* layer2 filter with ipfw */
418 VNET_DEFINE_STATIC(int, pfil_ipfw);
419 #define V_pfil_ipfw VNET(pfil_ipfw)
421 /* layer2 ARP filter with ipfw */
422 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
423 #define V_pfil_ipfw_arp VNET(pfil_ipfw_arp)
424 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
425 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
426 "Filter ARP packets through IPFW layer2");
428 /* run pfil hooks on the member interface */
429 VNET_DEFINE_STATIC(int, pfil_member) = 0;
430 #define V_pfil_member VNET(pfil_member)
431 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
432 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
433 "Packet filter on the member interface");
435 /* run pfil hooks on the physical interface for locally destined packets */
436 VNET_DEFINE_STATIC(int, pfil_local_phys);
437 #define V_pfil_local_phys VNET(pfil_local_phys)
438 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
439 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
440 "Packet filter on the physical interface for locally destined packets");
442 /* log STP state changes */
443 VNET_DEFINE_STATIC(int, log_stp);
444 #define V_log_stp VNET(log_stp)
445 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
446 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
447 "Log STP state changes");
449 /* share MAC with first bridge member */
450 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
451 #define V_bridge_inherit_mac VNET(bridge_inherit_mac)
452 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
453 CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
454 "Inherit MAC address from the first bridge member");
456 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
457 #define V_allow_llz_overlap VNET(allow_llz_overlap)
458 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
459 CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
460 "Allow overlap of link-local scope "
461 "zones of a bridge interface and the member interfaces");
463 struct bridge_control {
464 int (*bc_func)(struct bridge_softc *, void *);
469 #define BC_F_COPYIN 0x01 /* copy arguments in */
470 #define BC_F_COPYOUT 0x02 /* copy arguments out */
471 #define BC_F_SUSER 0x04 /* do super-user check */
473 static const struct bridge_control bridge_control_table[] = {
474 { bridge_ioctl_add, sizeof(struct ifbreq),
475 BC_F_COPYIN|BC_F_SUSER },
476 { bridge_ioctl_del, sizeof(struct ifbreq),
477 BC_F_COPYIN|BC_F_SUSER },
479 { bridge_ioctl_gifflags, sizeof(struct ifbreq),
480 BC_F_COPYIN|BC_F_COPYOUT },
481 { bridge_ioctl_sifflags, sizeof(struct ifbreq),
482 BC_F_COPYIN|BC_F_SUSER },
484 { bridge_ioctl_scache, sizeof(struct ifbrparam),
485 BC_F_COPYIN|BC_F_SUSER },
486 { bridge_ioctl_gcache, sizeof(struct ifbrparam),
489 { bridge_ioctl_gifs, sizeof(struct ifbifconf),
490 BC_F_COPYIN|BC_F_COPYOUT },
491 { bridge_ioctl_rts, sizeof(struct ifbaconf),
492 BC_F_COPYIN|BC_F_COPYOUT },
494 { bridge_ioctl_saddr, sizeof(struct ifbareq),
495 BC_F_COPYIN|BC_F_SUSER },
497 { bridge_ioctl_sto, sizeof(struct ifbrparam),
498 BC_F_COPYIN|BC_F_SUSER },
499 { bridge_ioctl_gto, sizeof(struct ifbrparam),
502 { bridge_ioctl_daddr, sizeof(struct ifbareq),
503 BC_F_COPYIN|BC_F_SUSER },
505 { bridge_ioctl_flush, sizeof(struct ifbreq),
506 BC_F_COPYIN|BC_F_SUSER },
508 { bridge_ioctl_gpri, sizeof(struct ifbrparam),
510 { bridge_ioctl_spri, sizeof(struct ifbrparam),
511 BC_F_COPYIN|BC_F_SUSER },
513 { bridge_ioctl_ght, sizeof(struct ifbrparam),
515 { bridge_ioctl_sht, sizeof(struct ifbrparam),
516 BC_F_COPYIN|BC_F_SUSER },
518 { bridge_ioctl_gfd, sizeof(struct ifbrparam),
520 { bridge_ioctl_sfd, sizeof(struct ifbrparam),
521 BC_F_COPYIN|BC_F_SUSER },
523 { bridge_ioctl_gma, sizeof(struct ifbrparam),
525 { bridge_ioctl_sma, sizeof(struct ifbrparam),
526 BC_F_COPYIN|BC_F_SUSER },
528 { bridge_ioctl_sifprio, sizeof(struct ifbreq),
529 BC_F_COPYIN|BC_F_SUSER },
531 { bridge_ioctl_sifcost, sizeof(struct ifbreq),
532 BC_F_COPYIN|BC_F_SUSER },
534 { bridge_ioctl_addspan, sizeof(struct ifbreq),
535 BC_F_COPYIN|BC_F_SUSER },
536 { bridge_ioctl_delspan, sizeof(struct ifbreq),
537 BC_F_COPYIN|BC_F_SUSER },
539 { bridge_ioctl_gbparam, sizeof(struct ifbropreq),
542 { bridge_ioctl_grte, sizeof(struct ifbrparam),
545 { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf),
546 BC_F_COPYIN|BC_F_COPYOUT },
548 { bridge_ioctl_sproto, sizeof(struct ifbrparam),
549 BC_F_COPYIN|BC_F_SUSER },
551 { bridge_ioctl_stxhc, sizeof(struct ifbrparam),
552 BC_F_COPYIN|BC_F_SUSER },
554 { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq),
555 BC_F_COPYIN|BC_F_SUSER },
558 static const int bridge_control_table_size = nitems(bridge_control_table);
560 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
561 #define V_bridge_list VNET(bridge_list)
562 #define BRIDGE_LIST_LOCK_INIT(x) sx_init(&V_bridge_list_sx, \
564 #define BRIDGE_LIST_LOCK_DESTROY(x) sx_destroy(&V_bridge_list_sx)
565 #define BRIDGE_LIST_LOCK(x) sx_xlock(&V_bridge_list_sx)
566 #define BRIDGE_LIST_UNLOCK(x) sx_xunlock(&V_bridge_list_sx)
568 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
569 #define V_bridge_cloner VNET(bridge_cloner)
571 static const char bridge_name[] = "bridge";
574 vnet_bridge_init(const void *unused __unused)
577 V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
578 sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
580 BRIDGE_LIST_LOCK_INIT();
581 LIST_INIT(&V_bridge_list);
583 struct if_clone_addreq req = {
584 .create_f = bridge_clone_create,
585 .destroy_f = bridge_clone_destroy,
586 .flags = IFC_F_AUTOUNIT,
588 V_bridge_cloner = ifc_attach_cloner(bridge_name, &req);
590 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
591 vnet_bridge_init, NULL);
594 vnet_bridge_uninit(const void *unused __unused)
597 ifc_detach_cloner(V_bridge_cloner);
598 V_bridge_cloner = NULL;
599 BRIDGE_LIST_LOCK_DESTROY();
601 /* Callbacks may use the UMA zone. */
602 NET_EPOCH_DRAIN_CALLBACKS();
604 uma_zdestroy(V_bridge_rtnode_zone);
606 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
607 vnet_bridge_uninit, NULL);
610 bridge_modevent(module_t mod, int type, void *data)
615 bridge_dn_p = bridge_dummynet;
616 bridge_detach_cookie = EVENTHANDLER_REGISTER(
617 ifnet_departure_event, bridge_ifdetach, NULL,
618 EVENTHANDLER_PRI_ANY);
621 EVENTHANDLER_DEREGISTER(ifnet_departure_event,
622 bridge_detach_cookie);
631 static moduledata_t bridge_mod = {
637 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
638 MODULE_VERSION(if_bridge, 1);
639 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
642 * handler for net.link.bridge.ipfw
645 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
647 int enable = V_pfil_ipfw;
650 error = sysctl_handle_int(oidp, &enable, 0, req);
653 if (enable != V_pfil_ipfw) {
654 V_pfil_ipfw = enable;
657 * Disable pfil so that ipfw doesnt run twice, if the user
658 * really wants both then they can re-enable pfil_bridge and/or
659 * pfil_member. Also allow non-ip packets as ipfw can filter by
671 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
672 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
673 &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
674 "Layer2 filter with IPFW");
678 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
680 struct bridge_softc *sc = ifp->if_softc;
681 struct bridge_iflist *bif;
685 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
686 bridge_delete_member(sc, bif, 0);
688 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
689 bridge_delete_span(sc, bif);
694 ether_reassign(ifp, newvnet, arg);
699 * bridge_clone_create:
701 * Create a new bridge instance.
704 bridge_clone_create(struct if_clone *ifc, char *name, size_t len,
705 struct ifc_data *ifd, struct ifnet **ifpp)
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, ifd->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();
766 bridge_clone_destroy_cb(struct epoch_context *ctx)
768 struct bridge_softc *sc;
770 sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
772 BRIDGE_LOCK_DESTROY(sc);
777 * bridge_clone_destroy:
779 * Destroy a bridge instance.
782 bridge_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags)
784 struct bridge_softc *sc = ifp->if_softc;
785 struct bridge_iflist *bif;
786 struct epoch_tracker et;
791 ifp->if_flags &= ~IFF_UP;
793 while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
794 bridge_delete_member(sc, bif, 0);
796 while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
797 bridge_delete_span(sc, bif);
800 /* Tear down the routing table. */
801 bridge_rtable_fini(sc);
807 callout_drain(&sc->sc_brcallout);
810 LIST_REMOVE(sc, sc_list);
811 BRIDGE_LIST_UNLOCK();
813 bstp_detach(&sc->sc_stp);
815 IFQ_PURGE(&ifp->if_snd);
822 NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
830 * Handle a control request from the operator.
833 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
835 struct bridge_softc *sc = ifp->if_softc;
836 struct ifreq *ifr = (struct ifreq *)data;
837 struct bridge_iflist *bif;
838 struct thread *td = curthread;
840 struct ifbreq ifbreq;
841 struct ifbifconf ifbifconf;
842 struct ifbareq ifbareq;
843 struct ifbaconf ifbaconf;
844 struct ifbrparam ifbrparam;
845 struct ifbropreq ifbropreq;
847 struct ifdrv *ifd = (struct ifdrv *) data;
848 const struct bridge_control *bc;
849 int error = 0, oldmtu;
860 if (ifd->ifd_cmd >= bridge_control_table_size) {
864 bc = &bridge_control_table[ifd->ifd_cmd];
866 if (cmd == SIOCGDRVSPEC &&
867 (bc->bc_flags & BC_F_COPYOUT) == 0) {
871 else if (cmd == SIOCSDRVSPEC &&
872 (bc->bc_flags & BC_F_COPYOUT) != 0) {
877 if (bc->bc_flags & BC_F_SUSER) {
878 error = priv_check(td, PRIV_NET_BRIDGE);
883 if (ifd->ifd_len != bc->bc_argsize ||
884 ifd->ifd_len > sizeof(args)) {
889 bzero(&args, sizeof(args));
890 if (bc->bc_flags & BC_F_COPYIN) {
891 error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
896 oldmtu = ifp->if_mtu;
897 error = (*bc->bc_func)(sc, &args);
902 * Bridge MTU may change during addition of the first port.
903 * If it did, do network layer specific procedure.
905 if (ifp->if_mtu != oldmtu)
908 if (bc->bc_flags & BC_F_COPYOUT)
909 error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
914 if (!(ifp->if_flags & IFF_UP) &&
915 (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
917 * If interface is marked down and it is running,
918 * then stop and disable it.
921 } else if ((ifp->if_flags & IFF_UP) &&
922 !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
924 * If interface is marked up and it is stopped, then
934 oldmtu = sc->sc_ifp->if_mtu;
936 if (ifr->ifr_mtu < 576) {
940 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
941 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
944 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
945 error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
946 SIOCSIFMTU, (caddr_t)ifr);
948 log(LOG_NOTICE, "%s: invalid MTU: %u for"
949 " member %s\n", sc->sc_ifp->if_xname,
951 bif->bif_ifp->if_xname);
957 /* Restore the previous MTU on all member interfaces. */
958 ifr->ifr_mtu = oldmtu;
959 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
960 (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
961 SIOCSIFMTU, (caddr_t)ifr);
964 sc->sc_ifp->if_mtu = ifr->ifr_mtu;
969 * drop the lock as ether_ioctl() will call bridge_start() and
970 * cause the lock to be recursed.
973 error = ether_ioctl(ifp, cmd, data);
986 * Clear or restore unwanted capabilities on the member interface
989 bridge_mutecaps(struct bridge_softc *sc)
991 struct bridge_iflist *bif;
994 BRIDGE_LOCK_ASSERT(sc);
996 /* Initial bitmask of capabilities to test */
997 mask = BRIDGE_IFCAPS_MASK;
999 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1000 /* Every member must support it or its disabled */
1001 mask &= bif->bif_savedcaps;
1004 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1005 enabled = bif->bif_ifp->if_capenable;
1006 enabled &= ~BRIDGE_IFCAPS_STRIP;
1007 /* strip off mask bits and enable them again if allowed */
1008 enabled &= ~BRIDGE_IFCAPS_MASK;
1010 bridge_set_ifcap(sc, bif, enabled);
1015 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1017 struct ifnet *ifp = bif->bif_ifp;
1019 int error, mask, stuck;
1021 bzero(&ifr, sizeof(ifr));
1022 ifr.ifr_reqcap = set;
1024 if (ifp->if_capenable != set) {
1025 error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1027 if_printf(sc->sc_ifp,
1028 "error setting capabilities on %s: %d\n",
1029 ifp->if_xname, error);
1030 mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1031 stuck = ifp->if_capenable & mask & ~set;
1033 if_printf(sc->sc_ifp,
1034 "can't disable some capabilities on %s: 0x%x\n",
1035 ifp->if_xname, stuck);
1040 * bridge_lookup_member:
1042 * Lookup a bridge member interface.
1044 static struct bridge_iflist *
1045 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1047 struct bridge_iflist *bif;
1050 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1052 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1054 if (strcmp(ifp->if_xname, name) == 0)
1062 * bridge_lookup_member_if:
1064 * Lookup a bridge member interface by ifnet*.
1066 static struct bridge_iflist *
1067 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1069 struct bridge_iflist *bif;
1071 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1073 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1074 if (bif->bif_ifp == member_ifp)
1082 bridge_delete_member_cb(struct epoch_context *ctx)
1084 struct bridge_iflist *bif;
1086 bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1088 free(bif, M_DEVBUF);
1092 * bridge_delete_member:
1094 * Delete the specified member interface.
1097 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1100 struct ifnet *ifs = bif->bif_ifp;
1101 struct ifnet *fif = NULL;
1102 struct bridge_iflist *bifl;
1104 BRIDGE_LOCK_ASSERT(sc);
1106 if (bif->bif_flags & IFBIF_STP)
1107 bstp_disable(&bif->bif_stp);
1109 ifs->if_bridge = NULL;
1110 CK_LIST_REMOVE(bif, bif_next);
1113 * If removing the interface that gave the bridge its mac address, set
1114 * the mac address of the bridge to the address of the next member, or
1115 * to its default address if no members are left.
1117 if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1118 if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1119 bcopy(&sc->sc_defaddr,
1120 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1121 sc->sc_ifaddr = NULL;
1123 bifl = CK_LIST_FIRST(&sc->sc_iflist);
1124 fif = bifl->bif_ifp;
1125 bcopy(IF_LLADDR(fif),
1126 IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1127 sc->sc_ifaddr = fif;
1129 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1132 bridge_linkcheck(sc);
1133 bridge_mutecaps(sc); /* recalcuate now this interface is removed */
1135 bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1136 BRIDGE_RT_UNLOCK(sc);
1137 KASSERT(bif->bif_addrcnt == 0,
1138 ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1140 ifs->if_bridge_output = NULL;
1141 ifs->if_bridge_input = NULL;
1142 ifs->if_bridge_linkstate = NULL;
1144 switch (ifs->if_type) {
1148 * Take the interface out of promiscuous mode, but only
1149 * if it was promiscuous in the first place. It might
1150 * not be if we're in the bridge_ioctl_add() error path.
1152 if (ifs->if_flags & IFF_PROMISC)
1153 (void) ifpromisc(ifs, 0);
1161 panic("bridge_delete_member: impossible");
1165 /* reneable any interface capabilities */
1166 bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1168 bstp_destroy(&bif->bif_stp); /* prepare to free */
1170 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1174 * bridge_delete_span:
1176 * Delete the specified span interface.
1179 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1181 BRIDGE_LOCK_ASSERT(sc);
1183 KASSERT(bif->bif_ifp->if_bridge == NULL,
1184 ("%s: not a span interface", __func__));
1186 CK_LIST_REMOVE(bif, bif_next);
1188 NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1192 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1194 struct ifbreq *req = arg;
1195 struct bridge_iflist *bif = NULL;
1199 ifs = ifunit(req->ifbr_ifsname);
1202 if (ifs->if_ioctl == NULL) /* must be supported */
1205 /* If it's in the span list, it can't be a member. */
1206 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1207 if (ifs == bif->bif_ifp)
1210 if (ifs->if_bridge == sc)
1213 if (ifs->if_bridge != NULL)
1216 switch (ifs->if_type) {
1220 /* permitted interface types */
1228 * Two valid inet6 addresses with link-local scope must not be
1229 * on the parent interface and the member interfaces at the
1230 * same time. This restriction is needed to prevent violation
1231 * of link-local scope zone. Attempts to add a member
1232 * interface which has inet6 addresses when the parent has
1233 * inet6 triggers removal of all inet6 addresses on the member
1237 /* Check if the parent interface has a link-local scope addr. */
1238 if (V_allow_llz_overlap == 0 &&
1239 in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1241 * If any, remove all inet6 addresses from the member
1244 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1245 if (in6ifa_llaonifp(bif->bif_ifp)) {
1246 in6_ifdetach(bif->bif_ifp);
1247 if_printf(sc->sc_ifp,
1248 "IPv6 addresses on %s have been removed "
1249 "before adding it as a member to prevent "
1250 "IPv6 address scope violation.\n",
1251 bif->bif_ifp->if_xname);
1254 if (in6ifa_llaonifp(ifs)) {
1256 if_printf(sc->sc_ifp,
1257 "IPv6 addresses on %s have been removed "
1258 "before adding it as a member to prevent "
1259 "IPv6 address scope violation.\n",
1264 /* Allow the first Ethernet member to define the MTU */
1265 if (CK_LIST_EMPTY(&sc->sc_iflist))
1266 sc->sc_ifp->if_mtu = ifs->if_mtu;
1267 else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1270 snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s",
1272 ifr.ifr_mtu = sc->sc_ifp->if_mtu;
1274 error = (*ifs->if_ioctl)(ifs,
1275 SIOCSIFMTU, (caddr_t)&ifr);
1277 log(LOG_NOTICE, "%s: invalid MTU: %u for"
1278 " new member %s\n", sc->sc_ifp->if_xname,
1285 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1290 bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1291 bif->bif_savedcaps = ifs->if_capenable;
1294 * Assign the interface's MAC address to the bridge if it's the first
1295 * member and the MAC address of the bridge has not been changed from
1296 * the default randomly generated one.
1298 if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1299 !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1300 bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1301 sc->sc_ifaddr = ifs;
1302 EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1305 ifs->if_bridge = sc;
1306 ifs->if_bridge_output = bridge_output;
1307 ifs->if_bridge_input = bridge_input;
1308 ifs->if_bridge_linkstate = bridge_linkstate;
1309 bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1311 * XXX: XLOCK HERE!?!
1313 * NOTE: insert_***HEAD*** should be safe for the traversals.
1315 CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1317 /* Set interface capabilities to the intersection set of all members */
1318 bridge_mutecaps(sc);
1319 bridge_linkcheck(sc);
1321 /* Place the interface into promiscuous mode */
1322 switch (ifs->if_type) {
1325 error = ifpromisc(ifs, 1);
1330 bridge_delete_member(sc, bif, 0);
1335 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1337 struct ifbreq *req = arg;
1338 struct bridge_iflist *bif;
1340 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1344 bridge_delete_member(sc, bif, 0);
1350 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1352 struct ifbreq *req = arg;
1353 struct bridge_iflist *bif;
1354 struct bstp_port *bp;
1356 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1361 req->ifbr_ifsflags = bif->bif_flags;
1362 req->ifbr_state = bp->bp_state;
1363 req->ifbr_priority = bp->bp_priority;
1364 req->ifbr_path_cost = bp->bp_path_cost;
1365 req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1366 req->ifbr_proto = bp->bp_protover;
1367 req->ifbr_role = bp->bp_role;
1368 req->ifbr_stpflags = bp->bp_flags;
1369 req->ifbr_addrcnt = bif->bif_addrcnt;
1370 req->ifbr_addrmax = bif->bif_addrmax;
1371 req->ifbr_addrexceeded = bif->bif_addrexceeded;
1373 /* Copy STP state options as flags */
1374 if (bp->bp_operedge)
1375 req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1376 if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1377 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1378 if (bp->bp_ptp_link)
1379 req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1380 if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1381 req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1382 if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1383 req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1384 if (bp->bp_flags & BSTP_PORT_ADMCOST)
1385 req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1390 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1392 struct epoch_tracker et;
1393 struct ifbreq *req = arg;
1394 struct bridge_iflist *bif;
1395 struct bstp_port *bp;
1398 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1403 if (req->ifbr_ifsflags & IFBIF_SPAN)
1404 /* SPAN is readonly */
1407 NET_EPOCH_ENTER(et);
1409 if (req->ifbr_ifsflags & IFBIF_STP) {
1410 if ((bif->bif_flags & IFBIF_STP) == 0) {
1411 error = bstp_enable(&bif->bif_stp);
1418 if ((bif->bif_flags & IFBIF_STP) != 0)
1419 bstp_disable(&bif->bif_stp);
1422 /* Pass on STP flags */
1423 bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1424 bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1425 bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1426 bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1428 /* Save the bits relating to the bridge */
1429 bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1437 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1439 struct ifbrparam *param = arg;
1441 sc->sc_brtmax = param->ifbrp_csize;
1448 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1450 struct ifbrparam *param = arg;
1452 param->ifbrp_csize = sc->sc_brtmax;
1458 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1460 struct ifbifconf *bifc = arg;
1461 struct bridge_iflist *bif;
1464 int count, buflen, len, error = 0;
1467 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1469 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1472 buflen = sizeof(breq) * count;
1473 if (bifc->ifbic_len == 0) {
1474 bifc->ifbic_len = buflen;
1477 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1483 len = min(bifc->ifbic_len, buflen);
1484 bzero(&breq, sizeof(breq));
1485 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1486 if (len < sizeof(breq))
1489 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1490 sizeof(breq.ifbr_ifsname));
1491 /* Fill in the ifbreq structure */
1492 error = bridge_ioctl_gifflags(sc, &breq);
1495 memcpy(buf, &breq, sizeof(breq));
1497 buf += sizeof(breq);
1498 len -= sizeof(breq);
1500 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1501 if (len < sizeof(breq))
1504 strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1505 sizeof(breq.ifbr_ifsname));
1506 breq.ifbr_ifsflags = bif->bif_flags;
1507 breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1508 memcpy(buf, &breq, sizeof(breq));
1510 buf += sizeof(breq);
1511 len -= sizeof(breq);
1514 bifc->ifbic_len = sizeof(breq) * count;
1515 error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1516 free(outbuf, M_TEMP);
1521 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1523 struct ifbaconf *bac = arg;
1524 struct bridge_rtnode *brt;
1525 struct ifbareq bareq;
1527 int count, buflen, len, error = 0;
1529 if (bac->ifbac_len == 0)
1533 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1535 buflen = sizeof(bareq) * count;
1537 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1543 len = min(bac->ifbac_len, buflen);
1544 bzero(&bareq, sizeof(bareq));
1545 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1546 if (len < sizeof(bareq))
1548 strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1549 sizeof(bareq.ifba_ifsname));
1550 memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1551 bareq.ifba_vlan = brt->brt_vlan;
1552 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1553 time_uptime < brt->brt_expire)
1554 bareq.ifba_expire = brt->brt_expire - time_uptime;
1556 bareq.ifba_expire = 0;
1557 bareq.ifba_flags = brt->brt_flags;
1559 memcpy(buf, &bareq, sizeof(bareq));
1561 buf += sizeof(bareq);
1562 len -= sizeof(bareq);
1565 bac->ifbac_len = sizeof(bareq) * count;
1566 error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1567 free(outbuf, M_TEMP);
1572 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1574 struct ifbareq *req = arg;
1575 struct bridge_iflist *bif;
1576 struct epoch_tracker et;
1579 NET_EPOCH_ENTER(et);
1580 bif = bridge_lookup_member(sc, req->ifba_ifsname);
1586 /* bridge_rtupdate() may acquire the lock. */
1587 error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1595 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1597 struct ifbrparam *param = arg;
1599 sc->sc_brttimeout = param->ifbrp_ctime;
1604 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1606 struct ifbrparam *param = arg;
1608 param->ifbrp_ctime = sc->sc_brttimeout;
1613 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1615 struct ifbareq *req = arg;
1617 return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1621 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1623 struct ifbreq *req = arg;
1626 bridge_rtflush(sc, req->ifbr_ifsflags);
1627 BRIDGE_RT_UNLOCK(sc);
1633 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1635 struct ifbrparam *param = arg;
1636 struct bstp_state *bs = &sc->sc_stp;
1638 param->ifbrp_prio = bs->bs_bridge_priority;
1643 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1645 struct ifbrparam *param = arg;
1647 return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1651 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1653 struct ifbrparam *param = arg;
1654 struct bstp_state *bs = &sc->sc_stp;
1656 param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1661 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1663 struct ifbrparam *param = arg;
1665 return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1669 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1671 struct ifbrparam *param = arg;
1672 struct bstp_state *bs = &sc->sc_stp;
1674 param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1679 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1681 struct ifbrparam *param = arg;
1683 return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1687 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1689 struct ifbrparam *param = arg;
1690 struct bstp_state *bs = &sc->sc_stp;
1692 param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1697 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1699 struct ifbrparam *param = arg;
1701 return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1705 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1707 struct ifbreq *req = arg;
1708 struct bridge_iflist *bif;
1710 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1714 return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1718 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1720 struct ifbreq *req = arg;
1721 struct bridge_iflist *bif;
1723 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1727 return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1731 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1733 struct ifbreq *req = arg;
1734 struct bridge_iflist *bif;
1736 bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1740 bif->bif_addrmax = req->ifbr_addrmax;
1745 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1747 struct ifbreq *req = arg;
1748 struct bridge_iflist *bif = NULL;
1751 ifs = ifunit(req->ifbr_ifsname);
1755 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1756 if (ifs == bif->bif_ifp)
1759 if (ifs->if_bridge != NULL)
1762 switch (ifs->if_type) {
1771 bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1776 bif->bif_flags = IFBIF_SPAN;
1778 CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1784 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1786 struct ifbreq *req = arg;
1787 struct bridge_iflist *bif;
1790 ifs = ifunit(req->ifbr_ifsname);
1794 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1795 if (ifs == bif->bif_ifp)
1801 bridge_delete_span(sc, bif);
1807 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1809 struct ifbropreq *req = arg;
1810 struct bstp_state *bs = &sc->sc_stp;
1811 struct bstp_port *root_port;
1813 req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1814 req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1815 req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1817 root_port = bs->bs_root_port;
1818 if (root_port == NULL)
1819 req->ifbop_root_port = 0;
1821 req->ifbop_root_port = root_port->bp_ifp->if_index;
1823 req->ifbop_holdcount = bs->bs_txholdcount;
1824 req->ifbop_priority = bs->bs_bridge_priority;
1825 req->ifbop_protocol = bs->bs_protover;
1826 req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1827 req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1828 req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1829 req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1830 req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1831 req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1837 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1839 struct ifbrparam *param = arg;
1841 param->ifbrp_cexceeded = sc->sc_brtexceeded;
1846 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1848 struct ifbpstpconf *bifstp = arg;
1849 struct bridge_iflist *bif;
1850 struct bstp_port *bp;
1851 struct ifbpstpreq bpreq;
1853 int count, buflen, len, error = 0;
1856 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1857 if ((bif->bif_flags & IFBIF_STP) != 0)
1861 buflen = sizeof(bpreq) * count;
1862 if (bifstp->ifbpstp_len == 0) {
1863 bifstp->ifbpstp_len = buflen;
1867 outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1873 len = min(bifstp->ifbpstp_len, buflen);
1874 bzero(&bpreq, sizeof(bpreq));
1875 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1876 if (len < sizeof(bpreq))
1879 if ((bif->bif_flags & IFBIF_STP) == 0)
1883 bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1884 bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1885 bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1886 bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1887 bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1888 bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1890 memcpy(buf, &bpreq, sizeof(bpreq));
1892 buf += sizeof(bpreq);
1893 len -= sizeof(bpreq);
1896 bifstp->ifbpstp_len = sizeof(bpreq) * count;
1897 error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1898 free(outbuf, M_TEMP);
1903 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1905 struct ifbrparam *param = arg;
1907 return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1911 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1913 struct ifbrparam *param = arg;
1915 return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1921 * Detach an interface from a bridge. Called when a member
1922 * interface is detaching.
1925 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1927 struct bridge_softc *sc = ifp->if_bridge;
1928 struct bridge_iflist *bif;
1930 if (ifp->if_flags & IFF_RENAMING)
1932 if (V_bridge_cloner == NULL) {
1934 * This detach handler can be called after
1935 * vnet_bridge_uninit(). Just return in that case.
1939 /* Check if the interface is a bridge member */
1943 bif = bridge_lookup_member_if(sc, ifp);
1945 bridge_delete_member(sc, bif, 1);
1951 /* Check if the interface is a span port */
1953 LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1955 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1956 if (ifp == bif->bif_ifp) {
1957 bridge_delete_span(sc, bif);
1963 BRIDGE_LIST_UNLOCK();
1969 * Initialize a bridge interface.
1972 bridge_init(void *xsc)
1974 struct bridge_softc *sc = (struct bridge_softc *)xsc;
1975 struct ifnet *ifp = sc->sc_ifp;
1977 if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1981 callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1984 ifp->if_drv_flags |= IFF_DRV_RUNNING;
1985 bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */
1993 * Stop the bridge interface.
1996 bridge_stop(struct ifnet *ifp, int disable)
1998 struct bridge_softc *sc = ifp->if_softc;
2000 BRIDGE_LOCK_ASSERT(sc);
2002 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2006 callout_stop(&sc->sc_brcallout);
2008 bstp_stop(&sc->sc_stp);
2010 bridge_rtflush(sc, IFBF_FLUSHDYN);
2011 BRIDGE_RT_UNLOCK(sc);
2013 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2019 * Enqueue a packet on a bridge member interface.
2023 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2029 /* We may be sending a fragment so traverse the mbuf */
2032 m->m_nextpkt = NULL;
2033 len = m->m_pkthdr.len;
2034 mflags = m->m_flags;
2037 * If underlying interface can not do VLAN tag insertion itself
2038 * then attach a packet tag that holds it.
2040 if ((m->m_flags & M_VLANTAG) &&
2041 (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2042 m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2045 "unable to prepend VLAN header\n");
2046 if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2049 m->m_flags &= ~M_VLANTAG;
2052 M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2053 if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2056 for (m = m0, n = 1; m != NULL; m = m0, n++) {
2060 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, n);
2064 if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2065 if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2066 if (mflags & M_MCAST)
2067 if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2076 * Receive a queued packet from dummynet and pass it on to the output
2079 * The mbuf has the Ethernet header already attached.
2082 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2084 struct bridge_softc *sc;
2086 sc = ifp->if_bridge;
2089 * The packet didnt originate from a member interface. This should only
2090 * ever happen if a member interface is removed while packets are
2098 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2100 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2103 if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2109 bridge_enqueue(sc, ifp, m);
2115 * Send output from a bridge member interface. This
2116 * performs the bridging function for locally originated
2119 * The mbuf has the Ethernet header already attached. We must
2120 * enqueue or free the mbuf before returning.
2123 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2126 struct ether_header *eh;
2127 struct ifnet *bifp, *dst_if;
2128 struct bridge_softc *sc;
2133 if (m->m_len < ETHER_HDR_LEN) {
2134 m = m_pullup(m, ETHER_HDR_LEN);
2139 eh = mtod(m, struct ether_header *);
2140 sc = ifp->if_bridge;
2141 vlan = VLANTAGOF(m);
2146 * If bridge is down, but the original output interface is up,
2147 * go ahead and send out that interface. Otherwise, the packet
2150 if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2156 * If the packet is a multicast, or we don't know a better way to
2157 * get there, send to all interfaces.
2159 if (ETHER_IS_MULTICAST(eh->ether_dhost))
2162 dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2163 /* Tap any traffic not passing back out the originating interface */
2165 ETHER_BPF_MTAP(bifp, m);
2166 if (dst_if == NULL) {
2167 struct bridge_iflist *bif;
2173 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2174 dst_if = bif->bif_ifp;
2176 if (dst_if->if_type == IFT_GIF)
2178 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2182 * If this is not the original output interface,
2183 * and the interface is participating in spanning
2184 * tree, make sure the port is in a state that
2185 * allows forwarding.
2187 if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2188 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2191 if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2195 mc = m_dup(m, M_NOWAIT);
2197 if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2202 bridge_enqueue(sc, dst_if, mc);
2211 * XXX Spanning tree consideration here?
2215 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2220 bridge_enqueue(sc, dst_if, m);
2227 * Do output on a bridge.
2231 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2233 struct bridge_softc *sc;
2234 struct ether_header *eh;
2235 struct ifnet *dst_if;
2240 ETHER_BPF_MTAP(ifp, m);
2242 eh = mtod(m, struct ether_header *);
2244 if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2245 (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2246 error = bridge_enqueue(sc, dst_if, m);
2248 bridge_broadcast(sc, ifp, m, 0);
2255 bridge_altq_start(if_t ifp)
2257 struct ifaltq *ifq = &ifp->if_snd;
2261 IFQ_DEQUEUE_NOLOCK(ifq, m);
2263 bridge_transmit(ifp, m);
2264 IFQ_DEQUEUE_NOLOCK(ifq, m);
2270 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2274 if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2275 IFQ_ENQUEUE(&ifp->if_snd, m, err);
2277 bridge_altq_start(ifp);
2279 err = bridge_transmit(ifp, m);
2286 * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2289 bridge_qflush(struct ifnet *ifp __unused)
2296 * The forwarding function of the bridge.
2298 * NOTE: Releases the lock on return.
2301 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2304 struct bridge_iflist *dbif;
2305 struct ifnet *src_if, *dst_if, *ifp;
2306 struct ether_header *eh;
2313 src_if = m->m_pkthdr.rcvif;
2316 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2317 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2318 vlan = VLANTAGOF(m);
2320 if ((sbif->bif_flags & IFBIF_STP) &&
2321 sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2324 eh = mtod(m, struct ether_header *);
2325 dst = eh->ether_dhost;
2327 /* If the interface is learning, record the address. */
2328 if (sbif->bif_flags & IFBIF_LEARNING) {
2329 error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2330 sbif, 0, IFBAF_DYNAMIC);
2332 * If the interface has addresses limits then deny any source
2333 * that is not in the cache.
2335 if (error && sbif->bif_addrmax)
2339 if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2340 sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2344 * At this point, the port either doesn't participate
2345 * in spanning tree or it is in the forwarding state.
2349 * If the packet is unicast, destined for someone on
2350 * "this" side of the bridge, drop it.
2352 if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2353 dst_if = bridge_rtlookup(sc, dst, vlan);
2354 if (src_if == dst_if)
2358 * Check if its a reserved multicast address, any address
2359 * listed in 802.1D section 7.12.6 may not be forwarded by the
2361 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2363 if (dst[0] == 0x01 && dst[1] == 0x80 &&
2364 dst[2] == 0xc2 && dst[3] == 0x00 &&
2365 dst[4] == 0x00 && dst[5] <= 0x0f)
2368 /* ...forward it to all interfaces. */
2369 if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2374 * If we have a destination interface which is a member of our bridge,
2375 * OR this is a unicast packet, push it through the bpf(4) machinery.
2376 * For broadcast or multicast packets, don't bother because it will
2377 * be reinjected into ether_input. We do this before we pass the packets
2378 * through the pfil(9) framework, as it is possible that pfil(9) will
2379 * drop the packet, or possibly modify it, making it difficult to debug
2380 * firewall issues on the bridge.
2382 if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2383 ETHER_BPF_MTAP(ifp, m);
2385 /* run the packet filter */
2386 if (PFIL_HOOKED_IN(V_inet_pfil_head)
2388 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2391 if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2397 if (dst_if == NULL) {
2398 bridge_broadcast(sc, src_if, m, 1);
2403 * At this point, we're dealing with a unicast frame
2404 * going to a different interface.
2406 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2409 dbif = bridge_lookup_member_if(sc, dst_if);
2411 /* Not a member of the bridge (anymore?) */
2414 /* Private segments can not talk to each other */
2415 if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2418 if ((dbif->bif_flags & IFBIF_STP) &&
2419 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2422 if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2424 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2427 if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2433 bridge_enqueue(sc, dst_if, m);
2443 * Receive input from a member interface. Queue the packet for
2444 * bridging if it is not for us.
2446 static struct mbuf *
2447 bridge_input(struct ifnet *ifp, struct mbuf *m)
2449 struct bridge_softc *sc = ifp->if_bridge;
2450 struct bridge_iflist *bif, *bif2;
2452 struct ether_header *eh;
2453 struct mbuf *mc, *mc2;
2459 if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2463 vlan = VLANTAGOF(m);
2466 * Implement support for bridge monitoring. If this flag has been
2467 * set on this interface, discard the packet once we push it through
2468 * the bpf(4) machinery, but before we do, increment the byte and
2469 * packet counters associated with this interface.
2471 if ((bifp->if_flags & IFF_MONITOR) != 0) {
2472 m->m_pkthdr.rcvif = bifp;
2473 ETHER_BPF_MTAP(bifp, m);
2474 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2475 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2479 bif = bridge_lookup_member_if(sc, ifp);
2484 eh = mtod(m, struct ether_header *);
2488 if (m->m_flags & (M_BCAST|M_MCAST)) {
2489 /* Tap off 802.1D packets; they do not get forwarded. */
2490 if (memcmp(eh->ether_dhost, bstp_etheraddr,
2491 ETHER_ADDR_LEN) == 0) {
2492 bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2496 if ((bif->bif_flags & IFBIF_STP) &&
2497 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2502 * Make a deep copy of the packet and enqueue the copy
2503 * for bridge processing; return the original packet for
2506 mc = m_dup(m, M_NOWAIT);
2511 /* Perform the bridge forwarding function with the copy. */
2512 bridge_forward(sc, bif, mc);
2515 * Reinject the mbuf as arriving on the bridge so we have a
2516 * chance at claiming multicast packets. We can not loop back
2517 * here from ether_input as a bridge is never a member of a
2520 KASSERT(bifp->if_bridge == NULL,
2521 ("loop created in bridge_input"));
2522 mc2 = m_dup(m, M_NOWAIT);
2524 /* Keep the layer3 header aligned */
2525 int i = min(mc2->m_pkthdr.len, max_protohdr);
2526 mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2529 mc2->m_pkthdr.rcvif = bifp;
2530 (*bifp->if_input)(bifp, mc2);
2533 /* Return the original packet for local processing. */
2537 if ((bif->bif_flags & IFBIF_STP) &&
2538 bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2542 #if (defined(INET) || defined(INET6))
2543 # define OR_CARP_CHECK_WE_ARE_DST(iface) \
2544 || ((iface)->if_carp \
2545 && (*carp_forus_p)((iface), eh->ether_dhost))
2546 # define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2547 || ((iface)->if_carp \
2548 && (*carp_forus_p)((iface), eh->ether_shost))
2550 # define OR_CARP_CHECK_WE_ARE_DST(iface)
2551 # define OR_CARP_CHECK_WE_ARE_SRC(iface)
2555 # define OR_PFIL_HOOKED_INET6 \
2556 || PFIL_HOOKED_IN(V_inet6_pfil_head)
2558 # define OR_PFIL_HOOKED_INET6
2561 #define GRAB_OUR_PACKETS(iface) \
2562 if ((iface)->if_type == IFT_GIF) \
2564 /* It is destined for us. */ \
2565 if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \
2566 OR_CARP_CHECK_WE_ARE_DST((iface)) \
2568 if (bif->bif_flags & IFBIF_LEARNING) { \
2569 error = bridge_rtupdate(sc, eh->ether_shost, \
2570 vlan, bif, 0, IFBAF_DYNAMIC); \
2571 if (error && bif->bif_addrmax) { \
2576 m->m_pkthdr.rcvif = iface; \
2577 if ((iface) == ifp) { \
2578 /* Skip bridge processing... src == dest */ \
2581 /* It's passing over or to the bridge, locally. */ \
2582 ETHER_BPF_MTAP(bifp, m); \
2583 if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1); \
2584 if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); \
2585 /* Filter on the physical interface. */ \
2586 if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) \
2587 OR_PFIL_HOOKED_INET6)) { \
2588 if (bridge_pfil(&m, NULL, ifp, \
2589 PFIL_IN) != 0 || m == NULL) { \
2593 if ((iface) != bifp) \
2594 ETHER_BPF_MTAP(iface, m); \
2598 /* We just received a packet that we sent out. */ \
2599 if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2600 OR_CARP_CHECK_WE_ARE_SRC((iface)) \
2607 * Unicast. Make sure it's not for the bridge.
2609 do { GRAB_OUR_PACKETS(bifp) } while (0);
2612 * Give a chance for ifp at first priority. This will help when the
2613 * packet comes through the interface like VLAN's with the same MACs
2614 * on several interfaces from the same bridge. This also will save
2615 * some CPU cycles in case the destination interface and the input
2616 * interface (eq ifp) are the same.
2618 do { GRAB_OUR_PACKETS(ifp) } while (0);
2620 /* Now check the all bridge members. */
2621 CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2622 GRAB_OUR_PACKETS(bif2->bif_ifp)
2625 #undef OR_CARP_CHECK_WE_ARE_DST
2626 #undef OR_CARP_CHECK_WE_ARE_SRC
2627 #undef OR_PFIL_HOOKED_INET6
2628 #undef GRAB_OUR_PACKETS
2630 /* Perform the bridge forwarding function. */
2631 bridge_forward(sc, bif, m);
2639 * Send a frame to all interfaces that are members of
2640 * the bridge, except for the one on which the packet
2643 * NOTE: Releases the lock on return.
2646 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2647 struct mbuf *m, int runfilt)
2649 struct bridge_iflist *dbif, *sbif;
2651 struct ifnet *dst_if;
2656 sbif = bridge_lookup_member_if(sc, src_if);
2658 /* Filter on the bridge interface before broadcasting */
2659 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2661 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2664 if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2670 CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2671 dst_if = dbif->bif_ifp;
2672 if (dst_if == src_if)
2675 /* Private segments can not talk to each other */
2676 if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2679 if ((dbif->bif_flags & IFBIF_STP) &&
2680 dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2683 if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2684 (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2687 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2690 if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2694 mc = m_dup(m, M_NOWAIT);
2696 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2702 * Filter on the output interface. Pass a NULL bridge interface
2703 * pointer so we do not redundantly filter on the bridge for
2704 * each interface we broadcast on.
2706 if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2708 || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2712 /* Keep the layer3 header aligned */
2713 i = min(mc->m_pkthdr.len, max_protohdr);
2714 mc = m_copyup(mc, i, ETHER_ALIGN);
2716 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2720 if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2726 bridge_enqueue(sc, dst_if, mc);
2735 * Duplicate a packet out one or more interfaces that are in span mode,
2736 * the original mbuf is unmodified.
2739 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2741 struct bridge_iflist *bif;
2742 struct ifnet *dst_if;
2747 if (CK_LIST_EMPTY(&sc->sc_spanlist))
2750 CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2751 dst_if = bif->bif_ifp;
2753 if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2756 mc = m_dup(m, M_NOWAIT);
2758 if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2762 bridge_enqueue(sc, dst_if, mc);
2769 * Add a bridge routing entry.
2772 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2773 struct bridge_iflist *bif, int setflags, uint8_t flags)
2775 struct bridge_rtnode *brt;
2778 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2780 /* Check the source address is valid and not multicast. */
2781 if (ETHER_IS_MULTICAST(dst) ||
2782 (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2783 dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2786 /* 802.1p frames map to vlan 1 */
2791 * A route for this destination might already exist. If so,
2792 * update it, otherwise create a new one.
2794 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2797 /* Check again, now that we have the lock. There could have
2798 * been a race and we only want to insert this once. */
2799 if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) != NULL) {
2800 BRIDGE_RT_UNLOCK(sc);
2804 if (sc->sc_brtcnt >= sc->sc_brtmax) {
2805 sc->sc_brtexceeded++;
2806 BRIDGE_RT_UNLOCK(sc);
2809 /* Check per interface address limits (if enabled) */
2810 if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2811 bif->bif_addrexceeded++;
2812 BRIDGE_RT_UNLOCK(sc);
2817 * Allocate a new bridge forwarding node, and
2818 * initialize the expiration time and Ethernet
2821 brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2823 BRIDGE_RT_UNLOCK(sc);
2826 brt->brt_vnet = curvnet;
2828 if (bif->bif_flags & IFBIF_STICKY)
2829 brt->brt_flags = IFBAF_STICKY;
2831 brt->brt_flags = IFBAF_DYNAMIC;
2833 memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2834 brt->brt_vlan = vlan;
2836 if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2837 uma_zfree(V_bridge_rtnode_zone, brt);
2838 BRIDGE_RT_UNLOCK(sc);
2844 BRIDGE_RT_UNLOCK(sc);
2847 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2848 brt->brt_dst != bif) {
2850 brt->brt_dst->bif_addrcnt--;
2852 brt->brt_dst->bif_addrcnt++;
2853 BRIDGE_RT_UNLOCK(sc);
2856 if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2857 brt->brt_expire = time_uptime + sc->sc_brttimeout;
2859 brt->brt_flags = flags;
2867 * Lookup the destination interface for an address.
2869 static struct ifnet *
2870 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2872 struct bridge_rtnode *brt;
2876 if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2879 return (brt->brt_ifp);
2885 * Trim the routine table so that we have a number
2886 * of routing entries less than or equal to the
2890 bridge_rttrim(struct bridge_softc *sc)
2892 struct bridge_rtnode *brt, *nbrt;
2895 BRIDGE_RT_LOCK_ASSERT(sc);
2897 /* Make sure we actually need to do this. */
2898 if (sc->sc_brtcnt <= sc->sc_brtmax)
2901 /* Force an aging cycle; this might trim enough addresses. */
2903 if (sc->sc_brtcnt <= sc->sc_brtmax)
2906 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2907 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2908 bridge_rtnode_destroy(sc, brt);
2909 if (sc->sc_brtcnt <= sc->sc_brtmax)
2918 * Aging timer for the bridge.
2921 bridge_timer(void *arg)
2923 struct bridge_softc *sc = arg;
2925 BRIDGE_RT_LOCK_ASSERT(sc);
2927 /* Destruction of rtnodes requires a proper vnet context */
2928 CURVNET_SET(sc->sc_ifp->if_vnet);
2931 if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2932 callout_reset(&sc->sc_brcallout,
2933 bridge_rtable_prune_period * hz, bridge_timer, sc);
2940 * Perform an aging cycle.
2943 bridge_rtage(struct bridge_softc *sc)
2945 struct bridge_rtnode *brt, *nbrt;
2947 BRIDGE_RT_LOCK_ASSERT(sc);
2949 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2950 if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2951 if (time_uptime >= brt->brt_expire)
2952 bridge_rtnode_destroy(sc, brt);
2960 * Remove all dynamic addresses from the bridge.
2963 bridge_rtflush(struct bridge_softc *sc, int full)
2965 struct bridge_rtnode *brt, *nbrt;
2967 BRIDGE_RT_LOCK_ASSERT(sc);
2969 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2970 if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2971 bridge_rtnode_destroy(sc, brt);
2978 * Remove an address from the table.
2981 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2983 struct bridge_rtnode *brt;
2989 * If vlan is zero then we want to delete for all vlans so the lookup
2990 * may return more than one.
2992 while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2993 bridge_rtnode_destroy(sc, brt);
2997 BRIDGE_RT_UNLOCK(sc);
2999 return (found ? 0 : ENOENT);
3005 * Delete routes to a speicifc member interface.
3008 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3010 struct bridge_rtnode *brt, *nbrt;
3012 BRIDGE_RT_LOCK_ASSERT(sc);
3014 CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3015 if (brt->brt_ifp == ifp && (full ||
3016 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3017 bridge_rtnode_destroy(sc, brt);
3022 * bridge_rtable_init:
3024 * Initialize the route table for this bridge.
3027 bridge_rtable_init(struct bridge_softc *sc)
3031 sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3032 M_DEVBUF, M_WAITOK);
3034 for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3035 CK_LIST_INIT(&sc->sc_rthash[i]);
3037 sc->sc_rthash_key = arc4random();
3038 CK_LIST_INIT(&sc->sc_rtlist);
3042 * bridge_rtable_fini:
3044 * Deconstruct the route table for this bridge.
3047 bridge_rtable_fini(struct bridge_softc *sc)
3050 KASSERT(sc->sc_brtcnt == 0,
3051 ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3052 free(sc->sc_rthash, M_DEVBUF);
3056 * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3057 * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3059 #define mix(a, b, c) \
3061 a -= b; a -= c; a ^= (c >> 13); \
3062 b -= c; b -= a; b ^= (a << 8); \
3063 c -= a; c -= b; c ^= (b >> 13); \
3064 a -= b; a -= c; a ^= (c >> 12); \
3065 b -= c; b -= a; b ^= (a << 16); \
3066 c -= a; c -= b; c ^= (b >> 5); \
3067 a -= b; a -= c; a ^= (c >> 3); \
3068 b -= c; b -= a; b ^= (a << 10); \
3069 c -= a; c -= b; c ^= (b >> 15); \
3070 } while (/*CONSTCOND*/0)
3072 static __inline uint32_t
3073 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3075 uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3086 return (c & BRIDGE_RTHASH_MASK);
3092 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3096 for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3097 d = ((int)a[i]) - ((int)b[i]);
3104 * bridge_rtnode_lookup:
3106 * Look up a bridge route node for the specified destination. Compare the
3107 * vlan id or if zero then just return the first match.
3109 static struct bridge_rtnode *
3110 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3112 struct bridge_rtnode *brt;
3116 BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3118 hash = bridge_rthash(sc, addr);
3119 CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3120 dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3121 if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
3131 * bridge_rtnode_insert:
3133 * Insert the specified bridge node into the route table. We
3134 * assume the entry is not already in the table.
3137 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3139 struct bridge_rtnode *lbrt;
3143 BRIDGE_RT_LOCK_ASSERT(sc);
3145 hash = bridge_rthash(sc, brt->brt_addr);
3147 lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3149 CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3154 dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3155 if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3158 CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3161 if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3162 CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3165 lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3166 } while (lbrt != NULL);
3169 panic("bridge_rtnode_insert: impossible");
3173 CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3180 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3182 struct bridge_rtnode *brt;
3184 brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3186 CURVNET_SET(brt->brt_vnet);
3187 uma_zfree(V_bridge_rtnode_zone, brt);
3192 * bridge_rtnode_destroy:
3194 * Destroy a bridge rtnode.
3197 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3199 BRIDGE_RT_LOCK_ASSERT(sc);
3201 CK_LIST_REMOVE(brt, brt_hash);
3203 CK_LIST_REMOVE(brt, brt_list);
3205 brt->brt_dst->bif_addrcnt--;
3207 NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3211 * bridge_rtable_expire:
3213 * Set the expiry time for all routes on an interface.
3216 bridge_rtable_expire(struct ifnet *ifp, int age)
3218 struct bridge_softc *sc = ifp->if_bridge;
3219 struct bridge_rtnode *brt;
3221 CURVNET_SET(ifp->if_vnet);
3225 * If the age is zero then flush, otherwise set all the expiry times to
3226 * age for the interface
3229 bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3231 CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3232 /* Cap the expiry time to 'age' */
3233 if (brt->brt_ifp == ifp &&
3234 brt->brt_expire > time_uptime + age &&
3235 (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3236 brt->brt_expire = time_uptime + age;
3239 BRIDGE_RT_UNLOCK(sc);
3244 * bridge_state_change:
3246 * Callback from the bridgestp code when a port changes states.
3249 bridge_state_change(struct ifnet *ifp, int state)
3251 struct bridge_softc *sc = ifp->if_bridge;
3252 static const char *stpstates[] = {
3261 CURVNET_SET(ifp->if_vnet);
3263 log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3264 sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3269 * Send bridge packets through pfil if they are one of the types pfil can deal
3270 * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without
3271 * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3275 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3277 int snap, error, i, hlen;
3278 struct ether_header *eh1, eh2;
3281 u_int16_t ether_type;
3285 error = -1; /* Default error if not error == 0 */
3288 /* we may return with the IP fields swapped, ensure its not shared */
3289 KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3292 if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3293 return (0); /* filtering is disabled */
3295 i = min((*mp)->m_pkthdr.len, max_protohdr);
3296 if ((*mp)->m_len < i) {
3297 *mp = m_pullup(*mp, i);
3299 printf("%s: m_pullup failed\n", __func__);
3304 eh1 = mtod(*mp, struct ether_header *);
3305 ether_type = ntohs(eh1->ether_type);
3308 * Check for SNAP/LLC.
3310 if (ether_type < ETHERMTU) {
3311 struct llc *llc2 = (struct llc *)(eh1 + 1);
3313 if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3314 llc2->llc_dsap == LLC_SNAP_LSAP &&
3315 llc2->llc_ssap == LLC_SNAP_LSAP &&
3316 llc2->llc_control == LLC_UI) {
3317 ether_type = htons(llc2->llc_un.type_snap.ether_type);
3323 * If we're trying to filter bridge traffic, don't look at anything
3324 * other than IP and ARP traffic. If the filter doesn't understand
3325 * IPv6, don't allow IPv6 through the bridge either. This is lame
3326 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3327 * but of course we don't have an AppleTalk filter to begin with.
3328 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3331 switch (ether_type) {
3333 case ETHERTYPE_REVARP:
3334 if (V_pfil_ipfw_arp == 0)
3335 return (0); /* Automatically pass */
3340 case ETHERTYPE_IPV6:
3345 * Check to see if the user wants to pass non-ip
3346 * packets, these will not be checked by pfil(9) and
3347 * passed unconditionally so the default is to drop.
3353 /* Run the packet through pfil before stripping link headers */
3354 if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3355 dir == PFIL_OUT && ifp != NULL) {
3356 switch (pfil_mbuf_out(V_link_pfil_head, mp, ifp, NULL)) {
3364 /* Strip off the Ethernet header and keep a copy. */
3365 m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3366 m_adj(*mp, ETHER_HDR_LEN);
3368 /* Strip off snap header, if present */
3370 m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3371 m_adj(*mp, sizeof(struct llc));
3375 * Check the IP header for alignment and errors
3377 if (dir == PFIL_IN) {
3378 switch (ether_type) {
3380 error = bridge_ip_checkbasic(mp);
3383 case ETHERTYPE_IPV6:
3384 error = bridge_ip6_checkbasic(mp);
3397 * Run the packet through pfil
3400 switch (ether_type) {
3403 * Run pfil on the member interface and the bridge, both can
3404 * be skipped by clearing pfil_member or pfil_bridge.
3407 * in_if -> bridge_if -> out_if
3409 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3410 pfil_mbuf_out(V_inet_pfil_head, mp, bifp, NULL)) !=
3414 if (V_pfil_member && ifp != NULL) {
3415 rv = (dir == PFIL_OUT) ?
3416 pfil_mbuf_out(V_inet_pfil_head, mp, ifp, NULL) :
3417 pfil_mbuf_in(V_inet_pfil_head, mp, ifp, NULL);
3418 if (rv != PFIL_PASS)
3422 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3423 pfil_mbuf_in(V_inet_pfil_head, mp, bifp, NULL)) !=
3427 /* check if we need to fragment the packet */
3428 /* bridge_fragment generates a mbuf chain of packets */
3429 /* that already include eth headers */
3430 if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3431 i = (*mp)->m_pkthdr.len;
3432 if (i > ifp->if_mtu) {
3433 error = bridge_fragment(ifp, mp, &eh2, snap,
3439 /* Recalculate the ip checksum. */
3440 ip = mtod(*mp, struct ip *);
3441 hlen = ip->ip_hl << 2;
3442 if (hlen < sizeof(struct ip))
3444 if (hlen > (*mp)->m_len) {
3445 if ((*mp = m_pullup(*mp, hlen)) == NULL)
3447 ip = mtod(*mp, struct ip *);
3452 if (hlen == sizeof(struct ip))
3453 ip->ip_sum = in_cksum_hdr(ip);
3455 ip->ip_sum = in_cksum(*mp, hlen);
3459 case ETHERTYPE_IPV6:
3460 if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3461 pfil_mbuf_out(V_inet6_pfil_head, mp, bifp, NULL)) !=
3465 if (V_pfil_member && ifp != NULL) {
3466 rv = (dir == PFIL_OUT) ?
3467 pfil_mbuf_out(V_inet6_pfil_head, mp, ifp, NULL) :
3468 pfil_mbuf_in(V_inet6_pfil_head, mp, ifp, NULL);
3469 if (rv != PFIL_PASS)
3473 if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3474 pfil_mbuf_in(V_inet6_pfil_head, mp, bifp, NULL)) !=
3493 * Finally, put everything back the way it was and return
3496 M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3499 bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3502 M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3505 bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3516 * Perform basic checks on header size since
3517 * pfil assumes ip_input has already processed
3518 * it for it. Cut-and-pasted from ip_input.c.
3519 * Given how simple the IPv6 version is,
3520 * does the IPv4 version really need to be
3523 * XXX Should we update ipstat here, or not?
3524 * XXX Right now we update ipstat but not
3528 bridge_ip_checkbasic(struct mbuf **mp)
3530 struct mbuf *m = *mp;
3538 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3539 if ((m = m_copyup(m, sizeof(struct ip),
3540 (max_linkhdr + 3) & ~3)) == NULL) {
3541 /* XXXJRT new stat, please */
3542 KMOD_IPSTAT_INC(ips_toosmall);
3545 } else if (__predict_false(m->m_len < sizeof (struct ip))) {
3546 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3547 KMOD_IPSTAT_INC(ips_toosmall);
3551 ip = mtod(m, struct ip *);
3552 if (ip == NULL) goto bad;
3554 if (ip->ip_v != IPVERSION) {
3555 KMOD_IPSTAT_INC(ips_badvers);
3558 hlen = ip->ip_hl << 2;
3559 if (hlen < sizeof(struct ip)) { /* minimum header length */
3560 KMOD_IPSTAT_INC(ips_badhlen);
3563 if (hlen > m->m_len) {
3564 if ((m = m_pullup(m, hlen)) == NULL) {
3565 KMOD_IPSTAT_INC(ips_badhlen);
3568 ip = mtod(m, struct ip *);
3569 if (ip == NULL) goto bad;
3572 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3573 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3575 if (hlen == sizeof(struct ip)) {
3576 sum = in_cksum_hdr(ip);
3578 sum = in_cksum(m, hlen);
3582 KMOD_IPSTAT_INC(ips_badsum);
3586 /* Retrieve the packet length. */
3587 len = ntohs(ip->ip_len);
3590 * Check for additional length bogosity
3593 KMOD_IPSTAT_INC(ips_badlen);
3598 * Check that the amount of data in the buffers
3599 * is as at least much as the IP header would have us expect.
3600 * Drop packet if shorter than we expect.
3602 if (m->m_pkthdr.len < len) {
3603 KMOD_IPSTAT_INC(ips_tooshort);
3607 /* Checks out, proceed */
3618 * Same as above, but for IPv6.
3619 * Cut-and-pasted from ip6_input.c.
3620 * XXX Should we update ip6stat, or not?
3623 bridge_ip6_checkbasic(struct mbuf **mp)
3625 struct mbuf *m = *mp;
3626 struct ip6_hdr *ip6;
3629 * If the IPv6 header is not aligned, slurp it up into a new
3630 * mbuf with space for link headers, in the event we forward
3631 * it. Otherwise, if it is aligned, make sure the entire base
3632 * IPv6 header is in the first mbuf of the chain.
3634 if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3635 struct ifnet *inifp = m->m_pkthdr.rcvif;
3636 if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3637 (max_linkhdr + 3) & ~3)) == NULL) {
3638 /* XXXJRT new stat, please */
3639 IP6STAT_INC(ip6s_toosmall);
3640 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3643 } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3644 struct ifnet *inifp = m->m_pkthdr.rcvif;
3645 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3646 IP6STAT_INC(ip6s_toosmall);
3647 in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3652 ip6 = mtod(m, struct ip6_hdr *);
3654 if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3655 IP6STAT_INC(ip6s_badvers);
3656 in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3660 /* Checks out, proceed */
3673 * Fragment mbuf chain in multiple packets and prepend ethernet header.
3676 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3677 int snap, struct llc *llc)
3679 struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3683 if (m->m_len < sizeof(struct ip) &&
3684 (m = m_pullup(m, sizeof(struct ip))) == NULL)
3686 ip = mtod(m, struct ip *);
3688 m->m_pkthdr.csum_flags |= CSUM_IP;
3689 error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3694 * Walk the chain and re-add the Ethernet header for
3697 for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3698 nextpkt = mcur->m_nextpkt;
3699 mcur->m_nextpkt = NULL;
3701 M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3705 mprev->m_nextpkt = nextpkt;
3708 bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3711 M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3715 mprev->m_nextpkt = nextpkt;
3718 bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3721 * The previous two M_PREPEND could have inserted one or two
3722 * mbufs in front so we have to update the previous packet's
3725 mcur->m_nextpkt = nextpkt;
3727 mprev->m_nextpkt = mcur;
3729 /* The first mbuf in the original chain needs to be
3736 KMOD_IPSTAT_INC(ips_fragmented);
3740 for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3741 m = mcur->m_nextpkt;
3748 bridge_linkstate(struct ifnet *ifp)
3750 struct bridge_softc *sc = ifp->if_bridge;
3751 struct bridge_iflist *bif;
3752 struct epoch_tracker et;
3754 NET_EPOCH_ENTER(et);
3756 bif = bridge_lookup_member_if(sc, ifp);
3761 bridge_linkcheck(sc);
3763 bstp_linkstate(&bif->bif_stp);
3769 bridge_linkcheck(struct bridge_softc *sc)
3771 struct bridge_iflist *bif;
3772 int new_link, hasls;
3774 BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3776 new_link = LINK_STATE_DOWN;
3778 /* Our link is considered up if at least one of our ports is active */
3779 CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3780 if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3782 if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3783 new_link = LINK_STATE_UP;
3787 if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3788 /* If no interfaces support link-state then we default to up */
3789 new_link = LINK_STATE_UP;
3791 if_link_state_change(sc->sc_ifp, new_link);