xref: /freebsd/sys/net/if_bridge.c (revision 5f757f3ff9144b609b3c433dfd370cc6bdc191ad)
1 /*	$NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $	*/
2 
3 /*-
4  * SPDX-License-Identifier: BSD-4-Clause
5  *
6  * Copyright 2001 Wasabi Systems, Inc.
7  * All rights reserved.
8  *
9  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
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
25  *    written permission.
26  *
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.
38  */
39 
40 /*
41  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
42  * All rights reserved.
43  *
44  * Redistribution and use in source and binary forms, with or without
45  * modification, are permitted provided that the following conditions
46  * are met:
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.
52  *
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.
64  *
65  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
66  */
67 
68 /*
69  * Network interface bridge support.
70  *
71  * TODO:
72  *
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).
77  */
78 
79 #include <sys/cdefs.h>
80 #include "opt_inet.h"
81 #include "opt_inet6.h"
82 
83 #include <sys/param.h>
84 #include <sys/eventhandler.h>
85 #include <sys/mbuf.h>
86 #include <sys/malloc.h>
87 #include <sys/protosw.h>
88 #include <sys/systm.h>
89 #include <sys/jail.h>
90 #include <sys/time.h>
91 #include <sys/socket.h> /* for net/if.h */
92 #include <sys/sockio.h>
93 #include <sys/ctype.h>  /* string functions */
94 #include <sys/kernel.h>
95 #include <sys/random.h>
96 #include <sys/syslog.h>
97 #include <sys/sysctl.h>
98 #include <vm/uma.h>
99 #include <sys/module.h>
100 #include <sys/priv.h>
101 #include <sys/proc.h>
102 #include <sys/lock.h>
103 #include <sys/mutex.h>
104 
105 #include <net/bpf.h>
106 #include <net/if.h>
107 #include <net/if_clone.h>
108 #include <net/if_dl.h>
109 #include <net/if_types.h>
110 #include <net/if_var.h>
111 #include <net/if_private.h>
112 #include <net/pfil.h>
113 #include <net/vnet.h>
114 
115 #include <netinet/in.h>
116 #include <netinet/in_systm.h>
117 #include <netinet/in_var.h>
118 #include <netinet/ip.h>
119 #include <netinet/ip_var.h>
120 #ifdef INET6
121 #include <netinet/ip6.h>
122 #include <netinet6/ip6_var.h>
123 #include <netinet6/in6_ifattach.h>
124 #endif
125 #if defined(INET) || defined(INET6)
126 #include <netinet/ip_carp.h>
127 #endif
128 #include <machine/in_cksum.h>
129 #include <netinet/if_ether.h>
130 #include <net/bridgestp.h>
131 #include <net/if_bridgevar.h>
132 #include <net/if_llc.h>
133 #include <net/if_vlan_var.h>
134 
135 #include <net/route.h>
136 
137 /*
138  * Size of the route hash table.  Must be a power of two.
139  */
140 #ifndef BRIDGE_RTHASH_SIZE
141 #define	BRIDGE_RTHASH_SIZE		1024
142 #endif
143 
144 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
145 
146 /*
147  * Default maximum number of addresses to cache.
148  */
149 #ifndef BRIDGE_RTABLE_MAX
150 #define	BRIDGE_RTABLE_MAX		2000
151 #endif
152 
153 /*
154  * Timeout (in seconds) for entries learned dynamically.
155  */
156 #ifndef BRIDGE_RTABLE_TIMEOUT
157 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
158 #endif
159 
160 /*
161  * Number of seconds between walks of the route list.
162  */
163 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
164 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
165 #endif
166 
167 /*
168  * List of capabilities to possibly mask on the member interface.
169  */
170 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
171 					 IFCAP_TXCSUM_IPV6)
172 
173 /*
174  * List of capabilities to strip
175  */
176 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
177 
178 /*
179  * Bridge locking
180  *
181  * The bridge relies heavily on the epoch(9) system to protect its data
182  * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we
183  * must ensure there is only one writer at a time.
184  *
185  * That is: for read accesses we only need to be in NET_EPOCH, but for write
186  * accesses we must hold:
187  *
188  *  - BRIDGE_RT_LOCK, for any change to bridge_rtnodes
189  *  - BRIDGE_LOCK, for any other change
190  *
191  * The BRIDGE_LOCK is a sleepable lock, because it is held across ioctl()
192  * calls to bridge member interfaces and these ioctl()s can sleep.
193  * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes
194  * required while we're in NET_EPOCH and then we're not allowed to sleep.
195  */
196 #define BRIDGE_LOCK_INIT(_sc)		do {			\
197 	sx_init(&(_sc)->sc_sx, "if_bridge");			\
198 	mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF);	\
199 } while (0)
200 #define BRIDGE_LOCK_DESTROY(_sc)	do {	\
201 	sx_destroy(&(_sc)->sc_sx);		\
202 	mtx_destroy(&(_sc)->sc_rt_mtx);		\
203 } while (0)
204 #define BRIDGE_LOCK(_sc)		sx_xlock(&(_sc)->sc_sx)
205 #define BRIDGE_UNLOCK(_sc)		sx_xunlock(&(_sc)->sc_sx)
206 #define BRIDGE_LOCK_ASSERT(_sc)		sx_assert(&(_sc)->sc_sx, SX_XLOCKED)
207 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc)	\
208 	    MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx))
209 #define BRIDGE_UNLOCK_ASSERT(_sc)	sx_assert(&(_sc)->sc_sx, SX_UNLOCKED)
210 #define BRIDGE_RT_LOCK(_sc)		mtx_lock(&(_sc)->sc_rt_mtx)
211 #define BRIDGE_RT_UNLOCK(_sc)		mtx_unlock(&(_sc)->sc_rt_mtx)
212 #define BRIDGE_RT_LOCK_ASSERT(_sc)	mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED)
213 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc)	\
214 	    MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx))
215 
216 /*
217  * Bridge interface list entry.
218  */
219 struct bridge_iflist {
220 	CK_LIST_ENTRY(bridge_iflist) bif_next;
221 	struct ifnet		*bif_ifp;	/* member if */
222 	struct bstp_port	bif_stp;	/* STP state */
223 	uint32_t		bif_flags;	/* member if flags */
224 	int			bif_savedcaps;	/* saved capabilities */
225 	uint32_t		bif_addrmax;	/* max # of addresses */
226 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
227 	uint32_t		bif_addrexceeded;/* # of address violations */
228 	struct epoch_context	bif_epoch_ctx;
229 };
230 
231 /*
232  * Bridge route node.
233  */
234 struct bridge_rtnode {
235 	CK_LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
236 	CK_LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
237 	struct bridge_iflist	*brt_dst;	/* destination if */
238 	unsigned long		brt_expire;	/* expiration time */
239 	uint8_t			brt_flags;	/* address flags */
240 	uint8_t			brt_addr[ETHER_ADDR_LEN];
241 	uint16_t		brt_vlan;	/* vlan id */
242 	struct	vnet		*brt_vnet;
243 	struct	epoch_context	brt_epoch_ctx;
244 };
245 #define	brt_ifp			brt_dst->bif_ifp
246 
247 /*
248  * Software state for each bridge.
249  */
250 struct bridge_softc {
251 	struct ifnet		*sc_ifp;	/* make this an interface */
252 	LIST_ENTRY(bridge_softc) sc_list;
253 	struct sx		sc_sx;
254 	struct mtx		sc_rt_mtx;
255 	uint32_t		sc_brtmax;	/* max # of addresses */
256 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
257 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
258 	struct callout		sc_brcallout;	/* bridge callout */
259 	CK_LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
260 	CK_LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
261 	CK_LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
262 	uint32_t		sc_rthash_key;	/* key for hash */
263 	CK_LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
264 	struct bstp_state	sc_stp;		/* STP state */
265 	uint32_t		sc_brtexceeded;	/* # of cache drops */
266 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
267 	struct ether_addr	sc_defaddr;	/* Default MAC address */
268 	if_input_fn_t		sc_if_input;	/* Saved copy of if_input */
269 	struct epoch_context	sc_epoch_ctx;
270 };
271 
272 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
273 #define	V_bridge_list_sx	VNET(bridge_list_sx)
274 static eventhandler_tag bridge_detach_cookie;
275 
276 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
277 
278 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
279 #define	V_bridge_rtnode_zone	VNET(bridge_rtnode_zone)
280 
281 static int	bridge_clone_create(struct if_clone *, char *, size_t,
282 		    struct ifc_data *, struct ifnet **);
283 static int	bridge_clone_destroy(struct if_clone *, struct ifnet *, uint32_t);
284 
285 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
286 static void	bridge_mutecaps(struct bridge_softc *);
287 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
288 		    int);
289 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
290 static void	bridge_init(void *);
291 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
292 static void	bridge_stop(struct ifnet *, int);
293 static int	bridge_transmit(struct ifnet *, struct mbuf *);
294 #ifdef ALTQ
295 static void	bridge_altq_start(if_t);
296 static int	bridge_altq_transmit(if_t, struct mbuf *);
297 #endif
298 static void	bridge_qflush(struct ifnet *);
299 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
300 static void	bridge_inject(struct ifnet *, struct mbuf *);
301 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
302 		    struct rtentry *);
303 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
304 		    struct mbuf *);
305 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
306 
307 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
308 		    struct mbuf *m);
309 
310 static void	bridge_timer(void *);
311 
312 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
313 		    struct mbuf *, int);
314 static void	bridge_span(struct bridge_softc *, struct mbuf *);
315 
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 *,
319 		    uint16_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 *,
324 		    uint16_t);
325 
326 static void	bridge_rtable_init(struct bridge_softc *);
327 static void	bridge_rtable_fini(struct bridge_softc *);
328 
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);
338 
339 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
340 		    const char *name);
341 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
342 		    struct ifnet *ifp);
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 *);
347 
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 *,
380 		    int);
381 static int	bridge_ip_checkbasic(struct mbuf **mp);
382 #ifdef INET6
383 static int	bridge_ip6_checkbasic(struct mbuf **mp);
384 #endif /* INET6 */
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);
389 
390 /*
391  * Use the "null" value from IEEE 802.1Q-2014 Table 9-2
392  * to indicate untagged frames.
393  */
394 #define	VLANTAGOF(_m)	\
395     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : DOT1Q_VID_NULL
396 
397 static struct bstp_cb_ops bridge_ops = {
398 	.bcb_state = bridge_state_change,
399 	.bcb_rtage = bridge_rtable_expire
400 };
401 
402 SYSCTL_DECL(_net_link);
403 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
404     "Bridge");
405 
406 /* only pass IP[46] packets when pfil is enabled */
407 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
408 #define	V_pfil_onlyip	VNET(pfil_onlyip)
409 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
410     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
411     "Only pass IP packets when pfil is enabled");
412 
413 /* run pfil hooks on the bridge interface */
414 VNET_DEFINE_STATIC(int, pfil_bridge) = 0;
415 #define	V_pfil_bridge	VNET(pfil_bridge)
416 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
417     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
418     "Packet filter on the bridge interface");
419 
420 /* layer2 filter with ipfw */
421 VNET_DEFINE_STATIC(int, pfil_ipfw);
422 #define	V_pfil_ipfw	VNET(pfil_ipfw)
423 
424 /* layer2 ARP filter with ipfw */
425 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
426 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
427 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
428     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
429     "Filter ARP packets through IPFW layer2");
430 
431 /* run pfil hooks on the member interface */
432 VNET_DEFINE_STATIC(int, pfil_member) = 0;
433 #define	V_pfil_member	VNET(pfil_member)
434 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
435     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
436     "Packet filter on the member interface");
437 
438 /* run pfil hooks on the physical interface for locally destined packets */
439 VNET_DEFINE_STATIC(int, pfil_local_phys);
440 #define	V_pfil_local_phys	VNET(pfil_local_phys)
441 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
442     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
443     "Packet filter on the physical interface for locally destined packets");
444 
445 /* log STP state changes */
446 VNET_DEFINE_STATIC(int, log_stp);
447 #define	V_log_stp	VNET(log_stp)
448 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
449     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
450     "Log STP state changes");
451 
452 /* share MAC with first bridge member */
453 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
454 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
455 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
456     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
457     "Inherit MAC address from the first bridge member");
458 
459 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
460 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
461 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
462     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
463     "Allow overlap of link-local scope "
464     "zones of a bridge interface and the member interfaces");
465 
466 /* log MAC address port flapping */
467 VNET_DEFINE_STATIC(bool, log_mac_flap) = true;
468 #define	V_log_mac_flap	VNET(log_mac_flap)
469 SYSCTL_BOOL(_net_link_bridge, OID_AUTO, log_mac_flap,
470     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(log_mac_flap), true,
471     "Log MAC address port flapping");
472 
473 VNET_DEFINE_STATIC(int, log_interval) = 5;
474 VNET_DEFINE_STATIC(int, log_count) = 0;
475 VNET_DEFINE_STATIC(struct timeval, log_last) = { 0 };
476 
477 #define	V_log_interval	VNET(log_interval)
478 #define	V_log_count	VNET(log_count)
479 #define	V_log_last	VNET(log_last)
480 
481 struct bridge_control {
482 	int	(*bc_func)(struct bridge_softc *, void *);
483 	int	bc_argsize;
484 	int	bc_flags;
485 };
486 
487 #define	BC_F_COPYIN		0x01	/* copy arguments in */
488 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
489 #define	BC_F_SUSER		0x04	/* do super-user check */
490 
491 static const struct bridge_control bridge_control_table[] = {
492 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
493 	  BC_F_COPYIN|BC_F_SUSER },
494 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
495 	  BC_F_COPYIN|BC_F_SUSER },
496 
497 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
498 	  BC_F_COPYIN|BC_F_COPYOUT },
499 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
500 	  BC_F_COPYIN|BC_F_SUSER },
501 
502 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
503 	  BC_F_COPYIN|BC_F_SUSER },
504 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
505 	  BC_F_COPYOUT },
506 
507 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
508 	  BC_F_COPYIN|BC_F_COPYOUT },
509 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
510 	  BC_F_COPYIN|BC_F_COPYOUT },
511 
512 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
513 	  BC_F_COPYIN|BC_F_SUSER },
514 
515 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
516 	  BC_F_COPYIN|BC_F_SUSER },
517 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
518 	  BC_F_COPYOUT },
519 
520 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
521 	  BC_F_COPYIN|BC_F_SUSER },
522 
523 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
524 	  BC_F_COPYIN|BC_F_SUSER },
525 
526 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
527 	  BC_F_COPYOUT },
528 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
529 	  BC_F_COPYIN|BC_F_SUSER },
530 
531 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
532 	  BC_F_COPYOUT },
533 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
534 	  BC_F_COPYIN|BC_F_SUSER },
535 
536 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
537 	  BC_F_COPYOUT },
538 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
539 	  BC_F_COPYIN|BC_F_SUSER },
540 
541 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
542 	  BC_F_COPYOUT },
543 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
544 	  BC_F_COPYIN|BC_F_SUSER },
545 
546 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
547 	  BC_F_COPYIN|BC_F_SUSER },
548 
549 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
550 	  BC_F_COPYIN|BC_F_SUSER },
551 
552 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
553 	  BC_F_COPYIN|BC_F_SUSER },
554 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
555 	  BC_F_COPYIN|BC_F_SUSER },
556 
557 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
558 	  BC_F_COPYOUT },
559 
560 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
561 	  BC_F_COPYOUT },
562 
563 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
564 	  BC_F_COPYIN|BC_F_COPYOUT },
565 
566 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
567 	  BC_F_COPYIN|BC_F_SUSER },
568 
569 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
570 	  BC_F_COPYIN|BC_F_SUSER },
571 
572 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
573 	  BC_F_COPYIN|BC_F_SUSER },
574 
575 };
576 static const int bridge_control_table_size = nitems(bridge_control_table);
577 
578 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
579 #define	V_bridge_list	VNET(bridge_list)
580 #define	BRIDGE_LIST_LOCK_INIT(x)	sx_init(&V_bridge_list_sx,	\
581 					    "if_bridge list")
582 #define	BRIDGE_LIST_LOCK_DESTROY(x)	sx_destroy(&V_bridge_list_sx)
583 #define	BRIDGE_LIST_LOCK(x)		sx_xlock(&V_bridge_list_sx)
584 #define	BRIDGE_LIST_UNLOCK(x)		sx_xunlock(&V_bridge_list_sx)
585 
586 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
587 #define	V_bridge_cloner	VNET(bridge_cloner)
588 
589 static const char bridge_name[] = "bridge";
590 
591 static void
592 vnet_bridge_init(const void *unused __unused)
593 {
594 
595 	V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
596 	    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
597 	    UMA_ALIGN_PTR, 0);
598 	BRIDGE_LIST_LOCK_INIT();
599 	LIST_INIT(&V_bridge_list);
600 
601 	struct if_clone_addreq req = {
602 		.create_f = bridge_clone_create,
603 		.destroy_f = bridge_clone_destroy,
604 		.flags = IFC_F_AUTOUNIT,
605 	};
606 	V_bridge_cloner = ifc_attach_cloner(bridge_name, &req);
607 }
608 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
609     vnet_bridge_init, NULL);
610 
611 static void
612 vnet_bridge_uninit(const void *unused __unused)
613 {
614 
615 	ifc_detach_cloner(V_bridge_cloner);
616 	V_bridge_cloner = NULL;
617 	BRIDGE_LIST_LOCK_DESTROY();
618 
619 	/* Callbacks may use the UMA zone. */
620 	NET_EPOCH_DRAIN_CALLBACKS();
621 
622 	uma_zdestroy(V_bridge_rtnode_zone);
623 }
624 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
625     vnet_bridge_uninit, NULL);
626 
627 static int
628 bridge_modevent(module_t mod, int type, void *data)
629 {
630 
631 	switch (type) {
632 	case MOD_LOAD:
633 		bridge_dn_p = bridge_dummynet;
634 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
635 		    ifnet_departure_event, bridge_ifdetach, NULL,
636 		    EVENTHANDLER_PRI_ANY);
637 		break;
638 	case MOD_UNLOAD:
639 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
640 		    bridge_detach_cookie);
641 		bridge_dn_p = NULL;
642 		break;
643 	default:
644 		return (EOPNOTSUPP);
645 	}
646 	return (0);
647 }
648 
649 static moduledata_t bridge_mod = {
650 	"if_bridge",
651 	bridge_modevent,
652 	0
653 };
654 
655 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
656 MODULE_VERSION(if_bridge, 1);
657 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
658 
659 /*
660  * handler for net.link.bridge.ipfw
661  */
662 static int
663 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
664 {
665 	int enable = V_pfil_ipfw;
666 	int error;
667 
668 	error = sysctl_handle_int(oidp, &enable, 0, req);
669 	enable &= 1;
670 
671 	if (enable != V_pfil_ipfw) {
672 		V_pfil_ipfw = enable;
673 
674 		/*
675 		 * Disable pfil so that ipfw doesnt run twice, if the user
676 		 * really wants both then they can re-enable pfil_bridge and/or
677 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
678 		 * layer2 type.
679 		 */
680 		if (V_pfil_ipfw) {
681 			V_pfil_onlyip = 0;
682 			V_pfil_bridge = 0;
683 			V_pfil_member = 0;
684 		}
685 	}
686 
687 	return (error);
688 }
689 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
690     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
691     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
692     "Layer2 filter with IPFW");
693 
694 #ifdef VIMAGE
695 static void
696 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
697 {
698 	struct bridge_softc *sc = ifp->if_softc;
699 	struct bridge_iflist *bif;
700 
701 	BRIDGE_LOCK(sc);
702 
703 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
704 		bridge_delete_member(sc, bif, 0);
705 
706 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
707 		bridge_delete_span(sc, bif);
708 	}
709 
710 	BRIDGE_UNLOCK(sc);
711 
712 	ether_reassign(ifp, newvnet, arg);
713 }
714 #endif
715 
716 /*
717  * bridge_clone_create:
718  *
719  *	Create a new bridge instance.
720  */
721 static int
722 bridge_clone_create(struct if_clone *ifc, char *name, size_t len,
723     struct ifc_data *ifd, struct ifnet **ifpp)
724 {
725 	struct bridge_softc *sc;
726 	struct ifnet *ifp;
727 
728 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
729 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
730 	if (ifp == NULL) {
731 		free(sc, M_DEVBUF);
732 		return (ENOSPC);
733 	}
734 
735 	BRIDGE_LOCK_INIT(sc);
736 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
737 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
738 
739 	/* Initialize our routing table. */
740 	bridge_rtable_init(sc);
741 
742 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
743 
744 	CK_LIST_INIT(&sc->sc_iflist);
745 	CK_LIST_INIT(&sc->sc_spanlist);
746 
747 	ifp->if_softc = sc;
748 	if_initname(ifp, bridge_name, ifd->unit);
749 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
750 	ifp->if_ioctl = bridge_ioctl;
751 #ifdef ALTQ
752 	ifp->if_start = bridge_altq_start;
753 	ifp->if_transmit = bridge_altq_transmit;
754 	IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
755 	ifp->if_snd.ifq_drv_maxlen = 0;
756 	IFQ_SET_READY(&ifp->if_snd);
757 #else
758 	ifp->if_transmit = bridge_transmit;
759 #endif
760 	ifp->if_qflush = bridge_qflush;
761 	ifp->if_init = bridge_init;
762 	ifp->if_type = IFT_BRIDGE;
763 
764 	ether_gen_addr(ifp, &sc->sc_defaddr);
765 
766 	bstp_attach(&sc->sc_stp, &bridge_ops);
767 	ether_ifattach(ifp, sc->sc_defaddr.octet);
768 	/* Now undo some of the damage... */
769 	ifp->if_baudrate = 0;
770 	ifp->if_type = IFT_BRIDGE;
771 #ifdef VIMAGE
772 	ifp->if_reassign = bridge_reassign;
773 #endif
774 	sc->sc_if_input = ifp->if_input;	/* ether_input */
775 	ifp->if_input = bridge_inject;
776 
777 	/*
778 	 * Allow BRIDGE_INPUT() to pass in packets originating from the bridge
779 	 * itself via bridge_inject().  This is required for netmap but
780 	 * otherwise has no effect.
781 	 */
782 	ifp->if_bridge_input = bridge_input;
783 
784 	BRIDGE_LIST_LOCK();
785 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
786 	BRIDGE_LIST_UNLOCK();
787 	*ifpp = ifp;
788 
789 	return (0);
790 }
791 
792 static void
793 bridge_clone_destroy_cb(struct epoch_context *ctx)
794 {
795 	struct bridge_softc *sc;
796 
797 	sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
798 
799 	BRIDGE_LOCK_DESTROY(sc);
800 	free(sc, M_DEVBUF);
801 }
802 
803 /*
804  * bridge_clone_destroy:
805  *
806  *	Destroy a bridge instance.
807  */
808 static int
809 bridge_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags)
810 {
811 	struct bridge_softc *sc = ifp->if_softc;
812 	struct bridge_iflist *bif;
813 	struct epoch_tracker et;
814 
815 	BRIDGE_LOCK(sc);
816 
817 	bridge_stop(ifp, 1);
818 	ifp->if_flags &= ~IFF_UP;
819 
820 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
821 		bridge_delete_member(sc, bif, 0);
822 
823 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
824 		bridge_delete_span(sc, bif);
825 	}
826 
827 	/* Tear down the routing table. */
828 	bridge_rtable_fini(sc);
829 
830 	BRIDGE_UNLOCK(sc);
831 
832 	NET_EPOCH_ENTER(et);
833 
834 	callout_drain(&sc->sc_brcallout);
835 
836 	BRIDGE_LIST_LOCK();
837 	LIST_REMOVE(sc, sc_list);
838 	BRIDGE_LIST_UNLOCK();
839 
840 	bstp_detach(&sc->sc_stp);
841 #ifdef ALTQ
842 	IFQ_PURGE(&ifp->if_snd);
843 #endif
844 	NET_EPOCH_EXIT(et);
845 
846 	ether_ifdetach(ifp);
847 	if_free(ifp);
848 
849 	NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
850 
851 	return (0);
852 }
853 
854 /*
855  * bridge_ioctl:
856  *
857  *	Handle a control request from the operator.
858  */
859 static int
860 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
861 {
862 	struct bridge_softc *sc = ifp->if_softc;
863 	struct ifreq *ifr = (struct ifreq *)data;
864 	struct bridge_iflist *bif;
865 	struct thread *td = curthread;
866 	union {
867 		struct ifbreq ifbreq;
868 		struct ifbifconf ifbifconf;
869 		struct ifbareq ifbareq;
870 		struct ifbaconf ifbaconf;
871 		struct ifbrparam ifbrparam;
872 		struct ifbropreq ifbropreq;
873 	} args;
874 	struct ifdrv *ifd = (struct ifdrv *) data;
875 	const struct bridge_control *bc;
876 	int error = 0, oldmtu;
877 
878 	BRIDGE_LOCK(sc);
879 
880 	switch (cmd) {
881 	case SIOCADDMULTI:
882 	case SIOCDELMULTI:
883 		break;
884 
885 	case SIOCGDRVSPEC:
886 	case SIOCSDRVSPEC:
887 		if (ifd->ifd_cmd >= bridge_control_table_size) {
888 			error = EINVAL;
889 			break;
890 		}
891 		bc = &bridge_control_table[ifd->ifd_cmd];
892 
893 		if (cmd == SIOCGDRVSPEC &&
894 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
895 			error = EINVAL;
896 			break;
897 		}
898 		else if (cmd == SIOCSDRVSPEC &&
899 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
900 			error = EINVAL;
901 			break;
902 		}
903 
904 		if (bc->bc_flags & BC_F_SUSER) {
905 			error = priv_check(td, PRIV_NET_BRIDGE);
906 			if (error)
907 				break;
908 		}
909 
910 		if (ifd->ifd_len != bc->bc_argsize ||
911 		    ifd->ifd_len > sizeof(args)) {
912 			error = EINVAL;
913 			break;
914 		}
915 
916 		bzero(&args, sizeof(args));
917 		if (bc->bc_flags & BC_F_COPYIN) {
918 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
919 			if (error)
920 				break;
921 		}
922 
923 		oldmtu = ifp->if_mtu;
924 		error = (*bc->bc_func)(sc, &args);
925 		if (error)
926 			break;
927 
928 		/*
929 		 * Bridge MTU may change during addition of the first port.
930 		 * If it did, do network layer specific procedure.
931 		 */
932 		if (ifp->if_mtu != oldmtu)
933 			if_notifymtu(ifp);
934 
935 		if (bc->bc_flags & BC_F_COPYOUT)
936 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
937 
938 		break;
939 
940 	case SIOCSIFFLAGS:
941 		if (!(ifp->if_flags & IFF_UP) &&
942 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
943 			/*
944 			 * If interface is marked down and it is running,
945 			 * then stop and disable it.
946 			 */
947 			bridge_stop(ifp, 1);
948 		} else if ((ifp->if_flags & IFF_UP) &&
949 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
950 			/*
951 			 * If interface is marked up and it is stopped, then
952 			 * start it.
953 			 */
954 			BRIDGE_UNLOCK(sc);
955 			(*ifp->if_init)(sc);
956 			BRIDGE_LOCK(sc);
957 		}
958 		break;
959 
960 	case SIOCSIFMTU:
961 		oldmtu = sc->sc_ifp->if_mtu;
962 
963 		if (ifr->ifr_mtu < IF_MINMTU) {
964 			error = EINVAL;
965 			break;
966 		}
967 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
968 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
969 			break;
970 		}
971 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
972 			error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
973 			    SIOCSIFMTU, (caddr_t)ifr);
974 			if (error != 0) {
975 				log(LOG_NOTICE, "%s: invalid MTU: %u for"
976 				    " member %s\n", sc->sc_ifp->if_xname,
977 				    ifr->ifr_mtu,
978 				    bif->bif_ifp->if_xname);
979 				error = EINVAL;
980 				break;
981 			}
982 		}
983 		if (error) {
984 			/* Restore the previous MTU on all member interfaces. */
985 			ifr->ifr_mtu = oldmtu;
986 			CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
987 				(*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
988 				    SIOCSIFMTU, (caddr_t)ifr);
989 			}
990 		} else {
991 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
992 		}
993 		break;
994 	default:
995 		/*
996 		 * drop the lock as ether_ioctl() will call bridge_start() and
997 		 * cause the lock to be recursed.
998 		 */
999 		BRIDGE_UNLOCK(sc);
1000 		error = ether_ioctl(ifp, cmd, data);
1001 		BRIDGE_LOCK(sc);
1002 		break;
1003 	}
1004 
1005 	BRIDGE_UNLOCK(sc);
1006 
1007 	return (error);
1008 }
1009 
1010 /*
1011  * bridge_mutecaps:
1012  *
1013  *	Clear or restore unwanted capabilities on the member interface
1014  */
1015 static void
1016 bridge_mutecaps(struct bridge_softc *sc)
1017 {
1018 	struct bridge_iflist *bif;
1019 	int enabled, mask;
1020 
1021 	BRIDGE_LOCK_ASSERT(sc);
1022 
1023 	/* Initial bitmask of capabilities to test */
1024 	mask = BRIDGE_IFCAPS_MASK;
1025 
1026 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1027 		/* Every member must support it or its disabled */
1028 		mask &= bif->bif_savedcaps;
1029 	}
1030 
1031 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1032 		enabled = bif->bif_ifp->if_capenable;
1033 		enabled &= ~BRIDGE_IFCAPS_STRIP;
1034 		/* strip off mask bits and enable them again if allowed */
1035 		enabled &= ~BRIDGE_IFCAPS_MASK;
1036 		enabled |= mask;
1037 		bridge_set_ifcap(sc, bif, enabled);
1038 	}
1039 }
1040 
1041 static void
1042 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1043 {
1044 	struct ifnet *ifp = bif->bif_ifp;
1045 	struct ifreq ifr;
1046 	int error, mask, stuck;
1047 
1048 	bzero(&ifr, sizeof(ifr));
1049 	ifr.ifr_reqcap = set;
1050 
1051 	if (ifp->if_capenable != set) {
1052 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1053 		if (error)
1054 			if_printf(sc->sc_ifp,
1055 			    "error setting capabilities on %s: %d\n",
1056 			    ifp->if_xname, error);
1057 		mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1058 		stuck = ifp->if_capenable & mask & ~set;
1059 		if (stuck != 0)
1060 			if_printf(sc->sc_ifp,
1061 			    "can't disable some capabilities on %s: 0x%x\n",
1062 			    ifp->if_xname, stuck);
1063 	}
1064 }
1065 
1066 /*
1067  * bridge_lookup_member:
1068  *
1069  *	Lookup a bridge member interface.
1070  */
1071 static struct bridge_iflist *
1072 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1073 {
1074 	struct bridge_iflist *bif;
1075 	struct ifnet *ifp;
1076 
1077 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1078 
1079 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1080 		ifp = bif->bif_ifp;
1081 		if (strcmp(ifp->if_xname, name) == 0)
1082 			return (bif);
1083 	}
1084 
1085 	return (NULL);
1086 }
1087 
1088 /*
1089  * bridge_lookup_member_if:
1090  *
1091  *	Lookup a bridge member interface by ifnet*.
1092  */
1093 static struct bridge_iflist *
1094 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1095 {
1096 	struct bridge_iflist *bif;
1097 
1098 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1099 
1100 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1101 		if (bif->bif_ifp == member_ifp)
1102 			return (bif);
1103 	}
1104 
1105 	return (NULL);
1106 }
1107 
1108 static void
1109 bridge_delete_member_cb(struct epoch_context *ctx)
1110 {
1111 	struct bridge_iflist *bif;
1112 
1113 	bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1114 
1115 	free(bif, M_DEVBUF);
1116 }
1117 
1118 /*
1119  * bridge_delete_member:
1120  *
1121  *	Delete the specified member interface.
1122  */
1123 static void
1124 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1125     int gone)
1126 {
1127 	struct ifnet *ifs = bif->bif_ifp;
1128 	struct ifnet *fif = NULL;
1129 	struct bridge_iflist *bifl;
1130 
1131 	BRIDGE_LOCK_ASSERT(sc);
1132 
1133 	if (bif->bif_flags & IFBIF_STP)
1134 		bstp_disable(&bif->bif_stp);
1135 
1136 	ifs->if_bridge = NULL;
1137 	CK_LIST_REMOVE(bif, bif_next);
1138 
1139 	/*
1140 	 * If removing the interface that gave the bridge its mac address, set
1141 	 * the mac address of the bridge to the address of the next member, or
1142 	 * to its default address if no members are left.
1143 	 */
1144 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1145 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1146 			bcopy(&sc->sc_defaddr,
1147 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1148 			sc->sc_ifaddr = NULL;
1149 		} else {
1150 			bifl = CK_LIST_FIRST(&sc->sc_iflist);
1151 			fif = bifl->bif_ifp;
1152 			bcopy(IF_LLADDR(fif),
1153 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1154 			sc->sc_ifaddr = fif;
1155 		}
1156 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1157 	}
1158 
1159 	bridge_linkcheck(sc);
1160 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1161 	BRIDGE_RT_LOCK(sc);
1162 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1163 	BRIDGE_RT_UNLOCK(sc);
1164 	KASSERT(bif->bif_addrcnt == 0,
1165 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1166 
1167 	ifs->if_bridge_output = NULL;
1168 	ifs->if_bridge_input = NULL;
1169 	ifs->if_bridge_linkstate = NULL;
1170 	if (!gone) {
1171 		switch (ifs->if_type) {
1172 		case IFT_ETHER:
1173 		case IFT_L2VLAN:
1174 			/*
1175 			 * Take the interface out of promiscuous mode, but only
1176 			 * if it was promiscuous in the first place. It might
1177 			 * not be if we're in the bridge_ioctl_add() error path.
1178 			 */
1179 			if (ifs->if_flags & IFF_PROMISC)
1180 				(void) ifpromisc(ifs, 0);
1181 			break;
1182 
1183 		case IFT_GIF:
1184 			break;
1185 
1186 		default:
1187 #ifdef DIAGNOSTIC
1188 			panic("bridge_delete_member: impossible");
1189 #endif
1190 			break;
1191 		}
1192 		/* reneable any interface capabilities */
1193 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1194 	}
1195 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1196 
1197 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1198 }
1199 
1200 /*
1201  * bridge_delete_span:
1202  *
1203  *	Delete the specified span interface.
1204  */
1205 static void
1206 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1207 {
1208 	BRIDGE_LOCK_ASSERT(sc);
1209 
1210 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1211 	    ("%s: not a span interface", __func__));
1212 
1213 	CK_LIST_REMOVE(bif, bif_next);
1214 
1215 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1216 }
1217 
1218 static int
1219 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1220 {
1221 	struct ifbreq *req = arg;
1222 	struct bridge_iflist *bif = NULL;
1223 	struct ifnet *ifs;
1224 	int error = 0;
1225 
1226 	ifs = ifunit(req->ifbr_ifsname);
1227 	if (ifs == NULL)
1228 		return (ENOENT);
1229 	if (ifs->if_ioctl == NULL)	/* must be supported */
1230 		return (EINVAL);
1231 
1232 	/* If it's in the span list, it can't be a member. */
1233 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1234 		if (ifs == bif->bif_ifp)
1235 			return (EBUSY);
1236 
1237 	if (ifs->if_bridge == sc)
1238 		return (EEXIST);
1239 
1240 	if (ifs->if_bridge != NULL)
1241 		return (EBUSY);
1242 
1243 	switch (ifs->if_type) {
1244 	case IFT_ETHER:
1245 	case IFT_L2VLAN:
1246 	case IFT_GIF:
1247 		/* permitted interface types */
1248 		break;
1249 	default:
1250 		return (EINVAL);
1251 	}
1252 
1253 #ifdef INET6
1254 	/*
1255 	 * Two valid inet6 addresses with link-local scope must not be
1256 	 * on the parent interface and the member interfaces at the
1257 	 * same time.  This restriction is needed to prevent violation
1258 	 * of link-local scope zone.  Attempts to add a member
1259 	 * interface which has inet6 addresses when the parent has
1260 	 * inet6 triggers removal of all inet6 addresses on the member
1261 	 * interface.
1262 	 */
1263 
1264 	/* Check if the parent interface has a link-local scope addr. */
1265 	if (V_allow_llz_overlap == 0 &&
1266 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1267 		/*
1268 		 * If any, remove all inet6 addresses from the member
1269 		 * interfaces.
1270 		 */
1271 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1272  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1273 				in6_ifdetach(bif->bif_ifp);
1274 				if_printf(sc->sc_ifp,
1275 				    "IPv6 addresses on %s have been removed "
1276 				    "before adding it as a member to prevent "
1277 				    "IPv6 address scope violation.\n",
1278 				    bif->bif_ifp->if_xname);
1279 			}
1280 		}
1281 		if (in6ifa_llaonifp(ifs)) {
1282 			in6_ifdetach(ifs);
1283 			if_printf(sc->sc_ifp,
1284 			    "IPv6 addresses on %s have been removed "
1285 			    "before adding it as a member to prevent "
1286 			    "IPv6 address scope violation.\n",
1287 			    ifs->if_xname);
1288 		}
1289 	}
1290 #endif
1291 	/* Allow the first Ethernet member to define the MTU */
1292 	if (CK_LIST_EMPTY(&sc->sc_iflist))
1293 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1294 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1295 		struct ifreq ifr;
1296 
1297 		snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s",
1298 		    ifs->if_xname);
1299 		ifr.ifr_mtu = sc->sc_ifp->if_mtu;
1300 
1301 		error = (*ifs->if_ioctl)(ifs,
1302 		    SIOCSIFMTU, (caddr_t)&ifr);
1303 		if (error != 0) {
1304 			log(LOG_NOTICE, "%s: invalid MTU: %u for"
1305 			    " new member %s\n", sc->sc_ifp->if_xname,
1306 			    ifr.ifr_mtu,
1307 			    ifs->if_xname);
1308 			return (EINVAL);
1309 		}
1310 	}
1311 
1312 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1313 	if (bif == NULL)
1314 		return (ENOMEM);
1315 
1316 	bif->bif_ifp = ifs;
1317 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1318 	bif->bif_savedcaps = ifs->if_capenable;
1319 
1320 	/*
1321 	 * Assign the interface's MAC address to the bridge if it's the first
1322 	 * member and the MAC address of the bridge has not been changed from
1323 	 * the default randomly generated one.
1324 	 */
1325 	if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1326 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1327 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1328 		sc->sc_ifaddr = ifs;
1329 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1330 	}
1331 
1332 	ifs->if_bridge = sc;
1333 	ifs->if_bridge_output = bridge_output;
1334 	ifs->if_bridge_input = bridge_input;
1335 	ifs->if_bridge_linkstate = bridge_linkstate;
1336 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1337 	/*
1338 	 * XXX: XLOCK HERE!?!
1339 	 *
1340 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1341 	 */
1342 	CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1343 
1344 	/* Set interface capabilities to the intersection set of all members */
1345 	bridge_mutecaps(sc);
1346 	bridge_linkcheck(sc);
1347 
1348 	/* Place the interface into promiscuous mode */
1349 	switch (ifs->if_type) {
1350 		case IFT_ETHER:
1351 		case IFT_L2VLAN:
1352 			error = ifpromisc(ifs, 1);
1353 			break;
1354 	}
1355 
1356 	if (error)
1357 		bridge_delete_member(sc, bif, 0);
1358 	return (error);
1359 }
1360 
1361 static int
1362 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1363 {
1364 	struct ifbreq *req = arg;
1365 	struct bridge_iflist *bif;
1366 
1367 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1368 	if (bif == NULL)
1369 		return (ENOENT);
1370 
1371 	bridge_delete_member(sc, bif, 0);
1372 
1373 	return (0);
1374 }
1375 
1376 static int
1377 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1378 {
1379 	struct ifbreq *req = arg;
1380 	struct bridge_iflist *bif;
1381 	struct bstp_port *bp;
1382 
1383 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1384 	if (bif == NULL)
1385 		return (ENOENT);
1386 
1387 	bp = &bif->bif_stp;
1388 	req->ifbr_ifsflags = bif->bif_flags;
1389 	req->ifbr_state = bp->bp_state;
1390 	req->ifbr_priority = bp->bp_priority;
1391 	req->ifbr_path_cost = bp->bp_path_cost;
1392 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1393 	req->ifbr_proto = bp->bp_protover;
1394 	req->ifbr_role = bp->bp_role;
1395 	req->ifbr_stpflags = bp->bp_flags;
1396 	req->ifbr_addrcnt = bif->bif_addrcnt;
1397 	req->ifbr_addrmax = bif->bif_addrmax;
1398 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1399 
1400 	/* Copy STP state options as flags */
1401 	if (bp->bp_operedge)
1402 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1403 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1404 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1405 	if (bp->bp_ptp_link)
1406 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1407 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1408 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1409 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1410 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1411 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1412 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1413 	return (0);
1414 }
1415 
1416 static int
1417 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1418 {
1419 	struct epoch_tracker et;
1420 	struct ifbreq *req = arg;
1421 	struct bridge_iflist *bif;
1422 	struct bstp_port *bp;
1423 	int error;
1424 
1425 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1426 	if (bif == NULL)
1427 		return (ENOENT);
1428 	bp = &bif->bif_stp;
1429 
1430 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1431 		/* SPAN is readonly */
1432 		return (EINVAL);
1433 
1434 	NET_EPOCH_ENTER(et);
1435 
1436 	if (req->ifbr_ifsflags & IFBIF_STP) {
1437 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1438 			error = bstp_enable(&bif->bif_stp);
1439 			if (error) {
1440 				NET_EPOCH_EXIT(et);
1441 				return (error);
1442 			}
1443 		}
1444 	} else {
1445 		if ((bif->bif_flags & IFBIF_STP) != 0)
1446 			bstp_disable(&bif->bif_stp);
1447 	}
1448 
1449 	/* Pass on STP flags */
1450 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1451 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1452 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1453 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1454 
1455 	/* Save the bits relating to the bridge */
1456 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1457 
1458 	NET_EPOCH_EXIT(et);
1459 
1460 	return (0);
1461 }
1462 
1463 static int
1464 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1465 {
1466 	struct ifbrparam *param = arg;
1467 
1468 	sc->sc_brtmax = param->ifbrp_csize;
1469 	bridge_rttrim(sc);
1470 
1471 	return (0);
1472 }
1473 
1474 static int
1475 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1476 {
1477 	struct ifbrparam *param = arg;
1478 
1479 	param->ifbrp_csize = sc->sc_brtmax;
1480 
1481 	return (0);
1482 }
1483 
1484 static int
1485 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1486 {
1487 	struct ifbifconf *bifc = arg;
1488 	struct bridge_iflist *bif;
1489 	struct ifbreq breq;
1490 	char *buf, *outbuf;
1491 	int count, buflen, len, error = 0;
1492 
1493 	count = 0;
1494 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1495 		count++;
1496 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1497 		count++;
1498 
1499 	buflen = sizeof(breq) * count;
1500 	if (bifc->ifbic_len == 0) {
1501 		bifc->ifbic_len = buflen;
1502 		return (0);
1503 	}
1504 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1505 	if (outbuf == NULL)
1506 		return (ENOMEM);
1507 
1508 	count = 0;
1509 	buf = outbuf;
1510 	len = min(bifc->ifbic_len, buflen);
1511 	bzero(&breq, sizeof(breq));
1512 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1513 		if (len < sizeof(breq))
1514 			break;
1515 
1516 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1517 		    sizeof(breq.ifbr_ifsname));
1518 		/* Fill in the ifbreq structure */
1519 		error = bridge_ioctl_gifflags(sc, &breq);
1520 		if (error)
1521 			break;
1522 		memcpy(buf, &breq, sizeof(breq));
1523 		count++;
1524 		buf += sizeof(breq);
1525 		len -= sizeof(breq);
1526 	}
1527 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1528 		if (len < sizeof(breq))
1529 			break;
1530 
1531 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1532 		    sizeof(breq.ifbr_ifsname));
1533 		breq.ifbr_ifsflags = bif->bif_flags;
1534 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1535 		memcpy(buf, &breq, sizeof(breq));
1536 		count++;
1537 		buf += sizeof(breq);
1538 		len -= sizeof(breq);
1539 	}
1540 
1541 	bifc->ifbic_len = sizeof(breq) * count;
1542 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1543 	free(outbuf, M_TEMP);
1544 	return (error);
1545 }
1546 
1547 static int
1548 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1549 {
1550 	struct ifbaconf *bac = arg;
1551 	struct bridge_rtnode *brt;
1552 	struct ifbareq bareq;
1553 	char *buf, *outbuf;
1554 	int count, buflen, len, error = 0;
1555 
1556 	if (bac->ifbac_len == 0)
1557 		return (0);
1558 
1559 	count = 0;
1560 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1561 		count++;
1562 	buflen = sizeof(bareq) * count;
1563 
1564 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1565 	if (outbuf == NULL)
1566 		return (ENOMEM);
1567 
1568 	count = 0;
1569 	buf = outbuf;
1570 	len = min(bac->ifbac_len, buflen);
1571 	bzero(&bareq, sizeof(bareq));
1572 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1573 		if (len < sizeof(bareq))
1574 			goto out;
1575 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1576 		    sizeof(bareq.ifba_ifsname));
1577 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1578 		bareq.ifba_vlan = brt->brt_vlan;
1579 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1580 				time_uptime < brt->brt_expire)
1581 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1582 		else
1583 			bareq.ifba_expire = 0;
1584 		bareq.ifba_flags = brt->brt_flags;
1585 
1586 		memcpy(buf, &bareq, sizeof(bareq));
1587 		count++;
1588 		buf += sizeof(bareq);
1589 		len -= sizeof(bareq);
1590 	}
1591 out:
1592 	bac->ifbac_len = sizeof(bareq) * count;
1593 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1594 	free(outbuf, M_TEMP);
1595 	return (error);
1596 }
1597 
1598 static int
1599 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1600 {
1601 	struct ifbareq *req = arg;
1602 	struct bridge_iflist *bif;
1603 	struct epoch_tracker et;
1604 	int error;
1605 
1606 	NET_EPOCH_ENTER(et);
1607 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1608 	if (bif == NULL) {
1609 		NET_EPOCH_EXIT(et);
1610 		return (ENOENT);
1611 	}
1612 
1613 	/* bridge_rtupdate() may acquire the lock. */
1614 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1615 	    req->ifba_flags);
1616 	NET_EPOCH_EXIT(et);
1617 
1618 	return (error);
1619 }
1620 
1621 static int
1622 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1623 {
1624 	struct ifbrparam *param = arg;
1625 
1626 	sc->sc_brttimeout = param->ifbrp_ctime;
1627 	return (0);
1628 }
1629 
1630 static int
1631 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1632 {
1633 	struct ifbrparam *param = arg;
1634 
1635 	param->ifbrp_ctime = sc->sc_brttimeout;
1636 	return (0);
1637 }
1638 
1639 static int
1640 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1641 {
1642 	struct ifbareq *req = arg;
1643 	int vlan = req->ifba_vlan;
1644 
1645 	/* Userspace uses '0' to mean 'any vlan' */
1646 	if (vlan == 0)
1647 		vlan = DOT1Q_VID_RSVD_IMPL;
1648 
1649 	return (bridge_rtdaddr(sc, req->ifba_dst, vlan));
1650 }
1651 
1652 static int
1653 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1654 {
1655 	struct ifbreq *req = arg;
1656 
1657 	BRIDGE_RT_LOCK(sc);
1658 	bridge_rtflush(sc, req->ifbr_ifsflags);
1659 	BRIDGE_RT_UNLOCK(sc);
1660 
1661 	return (0);
1662 }
1663 
1664 static int
1665 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1666 {
1667 	struct ifbrparam *param = arg;
1668 	struct bstp_state *bs = &sc->sc_stp;
1669 
1670 	param->ifbrp_prio = bs->bs_bridge_priority;
1671 	return (0);
1672 }
1673 
1674 static int
1675 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1676 {
1677 	struct ifbrparam *param = arg;
1678 
1679 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1680 }
1681 
1682 static int
1683 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1684 {
1685 	struct ifbrparam *param = arg;
1686 	struct bstp_state *bs = &sc->sc_stp;
1687 
1688 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1689 	return (0);
1690 }
1691 
1692 static int
1693 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1694 {
1695 	struct ifbrparam *param = arg;
1696 
1697 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1698 }
1699 
1700 static int
1701 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1702 {
1703 	struct ifbrparam *param = arg;
1704 	struct bstp_state *bs = &sc->sc_stp;
1705 
1706 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1707 	return (0);
1708 }
1709 
1710 static int
1711 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1712 {
1713 	struct ifbrparam *param = arg;
1714 
1715 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1716 }
1717 
1718 static int
1719 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1720 {
1721 	struct ifbrparam *param = arg;
1722 	struct bstp_state *bs = &sc->sc_stp;
1723 
1724 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1725 	return (0);
1726 }
1727 
1728 static int
1729 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1730 {
1731 	struct ifbrparam *param = arg;
1732 
1733 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1734 }
1735 
1736 static int
1737 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1738 {
1739 	struct ifbreq *req = arg;
1740 	struct bridge_iflist *bif;
1741 
1742 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1743 	if (bif == NULL)
1744 		return (ENOENT);
1745 
1746 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1747 }
1748 
1749 static int
1750 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1751 {
1752 	struct ifbreq *req = arg;
1753 	struct bridge_iflist *bif;
1754 
1755 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1756 	if (bif == NULL)
1757 		return (ENOENT);
1758 
1759 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1760 }
1761 
1762 static int
1763 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1764 {
1765 	struct ifbreq *req = arg;
1766 	struct bridge_iflist *bif;
1767 
1768 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1769 	if (bif == NULL)
1770 		return (ENOENT);
1771 
1772 	bif->bif_addrmax = req->ifbr_addrmax;
1773 	return (0);
1774 }
1775 
1776 static int
1777 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1778 {
1779 	struct ifbreq *req = arg;
1780 	struct bridge_iflist *bif = NULL;
1781 	struct ifnet *ifs;
1782 
1783 	ifs = ifunit(req->ifbr_ifsname);
1784 	if (ifs == NULL)
1785 		return (ENOENT);
1786 
1787 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1788 		if (ifs == bif->bif_ifp)
1789 			return (EBUSY);
1790 
1791 	if (ifs->if_bridge != NULL)
1792 		return (EBUSY);
1793 
1794 	switch (ifs->if_type) {
1795 		case IFT_ETHER:
1796 		case IFT_GIF:
1797 		case IFT_L2VLAN:
1798 			break;
1799 		default:
1800 			return (EINVAL);
1801 	}
1802 
1803 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1804 	if (bif == NULL)
1805 		return (ENOMEM);
1806 
1807 	bif->bif_ifp = ifs;
1808 	bif->bif_flags = IFBIF_SPAN;
1809 
1810 	CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1811 
1812 	return (0);
1813 }
1814 
1815 static int
1816 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1817 {
1818 	struct ifbreq *req = arg;
1819 	struct bridge_iflist *bif;
1820 	struct ifnet *ifs;
1821 
1822 	ifs = ifunit(req->ifbr_ifsname);
1823 	if (ifs == NULL)
1824 		return (ENOENT);
1825 
1826 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1827 		if (ifs == bif->bif_ifp)
1828 			break;
1829 
1830 	if (bif == NULL)
1831 		return (ENOENT);
1832 
1833 	bridge_delete_span(sc, bif);
1834 
1835 	return (0);
1836 }
1837 
1838 static int
1839 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1840 {
1841 	struct ifbropreq *req = arg;
1842 	struct bstp_state *bs = &sc->sc_stp;
1843 	struct bstp_port *root_port;
1844 
1845 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1846 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1847 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1848 
1849 	root_port = bs->bs_root_port;
1850 	if (root_port == NULL)
1851 		req->ifbop_root_port = 0;
1852 	else
1853 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1854 
1855 	req->ifbop_holdcount = bs->bs_txholdcount;
1856 	req->ifbop_priority = bs->bs_bridge_priority;
1857 	req->ifbop_protocol = bs->bs_protover;
1858 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1859 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1860 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1861 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1862 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1863 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1864 
1865 	return (0);
1866 }
1867 
1868 static int
1869 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1870 {
1871 	struct ifbrparam *param = arg;
1872 
1873 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1874 	return (0);
1875 }
1876 
1877 static int
1878 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1879 {
1880 	struct ifbpstpconf *bifstp = arg;
1881 	struct bridge_iflist *bif;
1882 	struct bstp_port *bp;
1883 	struct ifbpstpreq bpreq;
1884 	char *buf, *outbuf;
1885 	int count, buflen, len, error = 0;
1886 
1887 	count = 0;
1888 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1889 		if ((bif->bif_flags & IFBIF_STP) != 0)
1890 			count++;
1891 	}
1892 
1893 	buflen = sizeof(bpreq) * count;
1894 	if (bifstp->ifbpstp_len == 0) {
1895 		bifstp->ifbpstp_len = buflen;
1896 		return (0);
1897 	}
1898 
1899 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1900 	if (outbuf == NULL)
1901 		return (ENOMEM);
1902 
1903 	count = 0;
1904 	buf = outbuf;
1905 	len = min(bifstp->ifbpstp_len, buflen);
1906 	bzero(&bpreq, sizeof(bpreq));
1907 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1908 		if (len < sizeof(bpreq))
1909 			break;
1910 
1911 		if ((bif->bif_flags & IFBIF_STP) == 0)
1912 			continue;
1913 
1914 		bp = &bif->bif_stp;
1915 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1916 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1917 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1918 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1919 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1920 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1921 
1922 		memcpy(buf, &bpreq, sizeof(bpreq));
1923 		count++;
1924 		buf += sizeof(bpreq);
1925 		len -= sizeof(bpreq);
1926 	}
1927 
1928 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1929 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1930 	free(outbuf, M_TEMP);
1931 	return (error);
1932 }
1933 
1934 static int
1935 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1936 {
1937 	struct ifbrparam *param = arg;
1938 
1939 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1940 }
1941 
1942 static int
1943 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1944 {
1945 	struct ifbrparam *param = arg;
1946 
1947 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1948 }
1949 
1950 /*
1951  * bridge_ifdetach:
1952  *
1953  *	Detach an interface from a bridge.  Called when a member
1954  *	interface is detaching.
1955  */
1956 static void
1957 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1958 {
1959 	struct bridge_softc *sc = ifp->if_bridge;
1960 	struct bridge_iflist *bif;
1961 
1962 	if (ifp->if_flags & IFF_RENAMING)
1963 		return;
1964 	if (V_bridge_cloner == NULL) {
1965 		/*
1966 		 * This detach handler can be called after
1967 		 * vnet_bridge_uninit().  Just return in that case.
1968 		 */
1969 		return;
1970 	}
1971 	/* Check if the interface is a bridge member */
1972 	if (sc != NULL) {
1973 		BRIDGE_LOCK(sc);
1974 
1975 		bif = bridge_lookup_member_if(sc, ifp);
1976 		if (bif != NULL)
1977 			bridge_delete_member(sc, bif, 1);
1978 
1979 		BRIDGE_UNLOCK(sc);
1980 		return;
1981 	}
1982 
1983 	/* Check if the interface is a span port */
1984 	BRIDGE_LIST_LOCK();
1985 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1986 		BRIDGE_LOCK(sc);
1987 		CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1988 			if (ifp == bif->bif_ifp) {
1989 				bridge_delete_span(sc, bif);
1990 				break;
1991 			}
1992 
1993 		BRIDGE_UNLOCK(sc);
1994 	}
1995 	BRIDGE_LIST_UNLOCK();
1996 }
1997 
1998 /*
1999  * bridge_init:
2000  *
2001  *	Initialize a bridge interface.
2002  */
2003 static void
2004 bridge_init(void *xsc)
2005 {
2006 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
2007 	struct ifnet *ifp = sc->sc_ifp;
2008 
2009 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
2010 		return;
2011 
2012 	BRIDGE_LOCK(sc);
2013 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
2014 	    bridge_timer, sc);
2015 
2016 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
2017 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
2018 
2019 	BRIDGE_UNLOCK(sc);
2020 }
2021 
2022 /*
2023  * bridge_stop:
2024  *
2025  *	Stop the bridge interface.
2026  */
2027 static void
2028 bridge_stop(struct ifnet *ifp, int disable)
2029 {
2030 	struct bridge_softc *sc = ifp->if_softc;
2031 
2032 	BRIDGE_LOCK_ASSERT(sc);
2033 
2034 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2035 		return;
2036 
2037 	BRIDGE_RT_LOCK(sc);
2038 	callout_stop(&sc->sc_brcallout);
2039 
2040 	bstp_stop(&sc->sc_stp);
2041 
2042 	bridge_rtflush(sc, IFBF_FLUSHDYN);
2043 	BRIDGE_RT_UNLOCK(sc);
2044 
2045 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2046 }
2047 
2048 /*
2049  * bridge_enqueue:
2050  *
2051  *	Enqueue a packet on a bridge member interface.
2052  *
2053  */
2054 static int
2055 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
2056 {
2057 	int len, err = 0;
2058 	short mflags;
2059 	struct mbuf *m0;
2060 
2061 	/* We may be sending a fragment so traverse the mbuf */
2062 	for (; m; m = m0) {
2063 		m0 = m->m_nextpkt;
2064 		m->m_nextpkt = NULL;
2065 		len = m->m_pkthdr.len;
2066 		mflags = m->m_flags;
2067 
2068 		/*
2069 		 * If underlying interface can not do VLAN tag insertion itself
2070 		 * then attach a packet tag that holds it.
2071 		 */
2072 		if ((m->m_flags & M_VLANTAG) &&
2073 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
2074 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
2075 			if (m == NULL) {
2076 				if_printf(dst_ifp,
2077 				    "unable to prepend VLAN header\n");
2078 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2079 				continue;
2080 			}
2081 			m->m_flags &= ~M_VLANTAG;
2082 		}
2083 
2084 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2085 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2086 			int n;
2087 
2088 			for (m = m0, n = 1; m != NULL; m = m0, n++) {
2089 				m0 = m->m_nextpkt;
2090 				m_freem(m);
2091 			}
2092 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, n);
2093 			break;
2094 		}
2095 
2096 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2097 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2098 		if (mflags & M_MCAST)
2099 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2100 	}
2101 
2102 	return (err);
2103 }
2104 
2105 /*
2106  * bridge_dummynet:
2107  *
2108  * 	Receive a queued packet from dummynet and pass it on to the output
2109  * 	interface.
2110  *
2111  *	The mbuf has the Ethernet header already attached.
2112  */
2113 static void
2114 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2115 {
2116 	struct bridge_softc *sc;
2117 
2118 	sc = ifp->if_bridge;
2119 
2120 	/*
2121 	 * The packet didnt originate from a member interface. This should only
2122 	 * ever happen if a member interface is removed while packets are
2123 	 * queued for it.
2124 	 */
2125 	if (sc == NULL) {
2126 		m_freem(m);
2127 		return;
2128 	}
2129 
2130 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2131 #ifdef INET6
2132 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2133 #endif
2134 	    ) {
2135 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2136 			return;
2137 		if (m == NULL)
2138 			return;
2139 	}
2140 
2141 	bridge_enqueue(sc, ifp, m);
2142 }
2143 
2144 /*
2145  * bridge_output:
2146  *
2147  *	Send output from a bridge member interface.  This
2148  *	performs the bridging function for locally originated
2149  *	packets.
2150  *
2151  *	The mbuf has the Ethernet header already attached.  We must
2152  *	enqueue or free the mbuf before returning.
2153  */
2154 static int
2155 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2156     struct rtentry *rt)
2157 {
2158 	struct ether_header *eh;
2159 	struct ifnet *bifp, *dst_if;
2160 	struct bridge_softc *sc;
2161 	uint16_t vlan;
2162 
2163 	NET_EPOCH_ASSERT();
2164 
2165 	if (m->m_len < ETHER_HDR_LEN) {
2166 		m = m_pullup(m, ETHER_HDR_LEN);
2167 		if (m == NULL)
2168 			return (0);
2169 	}
2170 
2171 	eh = mtod(m, struct ether_header *);
2172 	sc = ifp->if_bridge;
2173 	vlan = VLANTAGOF(m);
2174 
2175 	bifp = sc->sc_ifp;
2176 
2177 	/*
2178 	 * If bridge is down, but the original output interface is up,
2179 	 * go ahead and send out that interface.  Otherwise, the packet
2180 	 * is dropped below.
2181 	 */
2182 	if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2183 		dst_if = ifp;
2184 		goto sendunicast;
2185 	}
2186 
2187 	/*
2188 	 * If the packet is a multicast, or we don't know a better way to
2189 	 * get there, send to all interfaces.
2190 	 */
2191 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2192 		dst_if = NULL;
2193 	else
2194 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2195 	/* Tap any traffic not passing back out the originating interface */
2196 	if (dst_if != ifp)
2197 		ETHER_BPF_MTAP(bifp, m);
2198 	if (dst_if == NULL) {
2199 		struct bridge_iflist *bif;
2200 		struct mbuf *mc;
2201 		int used = 0;
2202 
2203 		bridge_span(sc, m);
2204 
2205 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2206 			dst_if = bif->bif_ifp;
2207 
2208 			if (dst_if->if_type == IFT_GIF)
2209 				continue;
2210 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2211 				continue;
2212 
2213 			/*
2214 			 * If this is not the original output interface,
2215 			 * and the interface is participating in spanning
2216 			 * tree, make sure the port is in a state that
2217 			 * allows forwarding.
2218 			 */
2219 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2220 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2221 				continue;
2222 
2223 			if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2224 				used = 1;
2225 				mc = m;
2226 			} else {
2227 				mc = m_dup(m, M_NOWAIT);
2228 				if (mc == NULL) {
2229 					if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2230 					continue;
2231 				}
2232 			}
2233 
2234 			bridge_enqueue(sc, dst_if, mc);
2235 		}
2236 		if (used == 0)
2237 			m_freem(m);
2238 		return (0);
2239 	}
2240 
2241 sendunicast:
2242 	/*
2243 	 * XXX Spanning tree consideration here?
2244 	 */
2245 
2246 	bridge_span(sc, m);
2247 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2248 		m_freem(m);
2249 		return (0);
2250 	}
2251 
2252 	bridge_enqueue(sc, dst_if, m);
2253 	return (0);
2254 }
2255 
2256 /*
2257  * bridge_transmit:
2258  *
2259  *	Do output on a bridge.
2260  *
2261  */
2262 static int
2263 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2264 {
2265 	struct bridge_softc *sc;
2266 	struct ether_header *eh;
2267 	struct ifnet *dst_if;
2268 	int error = 0;
2269 
2270 	sc = ifp->if_softc;
2271 
2272 	ETHER_BPF_MTAP(ifp, m);
2273 
2274 	eh = mtod(m, struct ether_header *);
2275 
2276 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2277 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, DOT1Q_VID_NULL)) !=
2278 	    NULL) {
2279 		error = bridge_enqueue(sc, dst_if, m);
2280 	} else
2281 		bridge_broadcast(sc, ifp, m, 0);
2282 
2283 	return (error);
2284 }
2285 
2286 #ifdef ALTQ
2287 static void
2288 bridge_altq_start(if_t ifp)
2289 {
2290 	struct ifaltq *ifq = &ifp->if_snd;
2291 	struct mbuf *m;
2292 
2293 	IFQ_LOCK(ifq);
2294 	IFQ_DEQUEUE_NOLOCK(ifq, m);
2295 	while (m != NULL) {
2296 		bridge_transmit(ifp, m);
2297 		IFQ_DEQUEUE_NOLOCK(ifq, m);
2298 	}
2299 	IFQ_UNLOCK(ifq);
2300 }
2301 
2302 static int
2303 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2304 {
2305 	int err;
2306 
2307 	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2308 		IFQ_ENQUEUE(&ifp->if_snd, m, err);
2309 		if (err == 0)
2310 			bridge_altq_start(ifp);
2311 	} else
2312 		err = bridge_transmit(ifp, m);
2313 
2314 	return (err);
2315 }
2316 #endif	/* ALTQ */
2317 
2318 /*
2319  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2320  */
2321 static void
2322 bridge_qflush(struct ifnet *ifp __unused)
2323 {
2324 }
2325 
2326 /*
2327  * bridge_forward:
2328  *
2329  *	The forwarding function of the bridge.
2330  *
2331  *	NOTE: Releases the lock on return.
2332  */
2333 static void
2334 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2335     struct mbuf *m)
2336 {
2337 	struct bridge_iflist *dbif;
2338 	struct ifnet *src_if, *dst_if, *ifp;
2339 	struct ether_header *eh;
2340 	uint16_t vlan;
2341 	uint8_t *dst;
2342 	int error;
2343 
2344 	NET_EPOCH_ASSERT();
2345 
2346 	src_if = m->m_pkthdr.rcvif;
2347 	ifp = sc->sc_ifp;
2348 
2349 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2350 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2351 	vlan = VLANTAGOF(m);
2352 
2353 	if ((sbif->bif_flags & IFBIF_STP) &&
2354 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2355 		goto drop;
2356 
2357 	eh = mtod(m, struct ether_header *);
2358 	dst = eh->ether_dhost;
2359 
2360 	/* If the interface is learning, record the address. */
2361 	if (sbif->bif_flags & IFBIF_LEARNING) {
2362 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2363 		    sbif, 0, IFBAF_DYNAMIC);
2364 		/*
2365 		 * If the interface has addresses limits then deny any source
2366 		 * that is not in the cache.
2367 		 */
2368 		if (error && sbif->bif_addrmax)
2369 			goto drop;
2370 	}
2371 
2372 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2373 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2374 		goto drop;
2375 
2376 #ifdef DEV_NETMAP
2377 	/*
2378 	 * Hand the packet to netmap only if it wasn't injected by netmap
2379 	 * itself.
2380 	 */
2381 	if ((m->m_flags & M_BRIDGE_INJECT) == 0 &&
2382 	    (if_getcapenable(ifp) & IFCAP_NETMAP) != 0) {
2383 		ifp->if_input(ifp, m);
2384 		return;
2385 	}
2386 	m->m_flags &= ~M_BRIDGE_INJECT;
2387 #endif
2388 
2389 	/*
2390 	 * At this point, the port either doesn't participate
2391 	 * in spanning tree or it is in the forwarding state.
2392 	 */
2393 
2394 	/*
2395 	 * If the packet is unicast, destined for someone on
2396 	 * "this" side of the bridge, drop it.
2397 	 */
2398 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2399 		dst_if = bridge_rtlookup(sc, dst, vlan);
2400 		if (src_if == dst_if)
2401 			goto drop;
2402 	} else {
2403 		/*
2404 		 * Check if its a reserved multicast address, any address
2405 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2406 		 * bridge.
2407 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2408 		 */
2409 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2410 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2411 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2412 			goto drop;
2413 
2414 		/* ...forward it to all interfaces. */
2415 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2416 		dst_if = NULL;
2417 	}
2418 
2419 	/*
2420 	 * If we have a destination interface which is a member of our bridge,
2421 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2422 	 * For broadcast or multicast packets, don't bother because it will
2423 	 * be reinjected into ether_input. We do this before we pass the packets
2424 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2425 	 * drop the packet, or possibly modify it, making it difficult to debug
2426 	 * firewall issues on the bridge.
2427 	 */
2428 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2429 		ETHER_BPF_MTAP(ifp, m);
2430 
2431 	/* run the packet filter */
2432 	if (PFIL_HOOKED_IN(V_inet_pfil_head)
2433 #ifdef INET6
2434 	    || PFIL_HOOKED_IN(V_inet6_pfil_head)
2435 #endif
2436 	    ) {
2437 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2438 			return;
2439 		if (m == NULL)
2440 			return;
2441 	}
2442 
2443 	if (dst_if == NULL) {
2444 		bridge_broadcast(sc, src_if, m, 1);
2445 		return;
2446 	}
2447 
2448 	/*
2449 	 * At this point, we're dealing with a unicast frame
2450 	 * going to a different interface.
2451 	 */
2452 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2453 		goto drop;
2454 
2455 	dbif = bridge_lookup_member_if(sc, dst_if);
2456 	if (dbif == NULL)
2457 		/* Not a member of the bridge (anymore?) */
2458 		goto drop;
2459 
2460 	/* Private segments can not talk to each other */
2461 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2462 		goto drop;
2463 
2464 	if ((dbif->bif_flags & IFBIF_STP) &&
2465 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2466 		goto drop;
2467 
2468 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2469 #ifdef INET6
2470 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2471 #endif
2472 	    ) {
2473 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2474 			return;
2475 		if (m == NULL)
2476 			return;
2477 	}
2478 
2479 	bridge_enqueue(sc, dst_if, m);
2480 	return;
2481 
2482 drop:
2483 	m_freem(m);
2484 }
2485 
2486 /*
2487  * bridge_input:
2488  *
2489  *	Receive input from a member interface.  Queue the packet for
2490  *	bridging if it is not for us.
2491  */
2492 static struct mbuf *
2493 bridge_input(struct ifnet *ifp, struct mbuf *m)
2494 {
2495 	struct bridge_softc *sc;
2496 	struct bridge_iflist *bif, *bif2;
2497 	struct ifnet *bifp;
2498 	struct ether_header *eh;
2499 	struct mbuf *mc, *mc2;
2500 	uint16_t vlan;
2501 	int error;
2502 
2503 	NET_EPOCH_ASSERT();
2504 
2505 	eh = mtod(m, struct ether_header *);
2506 	vlan = VLANTAGOF(m);
2507 
2508 	sc = ifp->if_bridge;
2509 	if (sc == NULL) {
2510 		/*
2511 		 * This packet originated from the bridge itself, so it must
2512 		 * have been transmitted by netmap.  Derive the "source"
2513 		 * interface from the source address and drop the packet if the
2514 		 * source address isn't known.
2515 		 */
2516 		KASSERT((m->m_flags & M_BRIDGE_INJECT) != 0,
2517 		    ("%s: ifnet %p missing a bridge softc", __func__, ifp));
2518 		sc = if_getsoftc(ifp);
2519 		ifp = bridge_rtlookup(sc, eh->ether_shost, vlan);
2520 		if (ifp == NULL) {
2521 			if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1);
2522 			m_freem(m);
2523 			return (NULL);
2524 		}
2525 		m->m_pkthdr.rcvif = ifp;
2526 	}
2527 	bifp = sc->sc_ifp;
2528 	if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2529 		return (m);
2530 
2531 	/*
2532 	 * Implement support for bridge monitoring. If this flag has been
2533 	 * set on this interface, discard the packet once we push it through
2534 	 * the bpf(4) machinery, but before we do, increment the byte and
2535 	 * packet counters associated with this interface.
2536 	 */
2537 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2538 		m->m_pkthdr.rcvif  = bifp;
2539 		ETHER_BPF_MTAP(bifp, m);
2540 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2541 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2542 		m_freem(m);
2543 		return (NULL);
2544 	}
2545 	bif = bridge_lookup_member_if(sc, ifp);
2546 	if (bif == NULL) {
2547 		return (m);
2548 	}
2549 
2550 	bridge_span(sc, m);
2551 
2552 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2553 		/* Tap off 802.1D packets; they do not get forwarded. */
2554 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2555 		    ETHER_ADDR_LEN) == 0) {
2556 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2557 			return (NULL);
2558 		}
2559 
2560 		if ((bif->bif_flags & IFBIF_STP) &&
2561 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2562 			return (m);
2563 		}
2564 
2565 		/*
2566 		 * Make a deep copy of the packet and enqueue the copy
2567 		 * for bridge processing; return the original packet for
2568 		 * local processing.
2569 		 */
2570 		mc = m_dup(m, M_NOWAIT);
2571 		if (mc == NULL) {
2572 			return (m);
2573 		}
2574 
2575 		/* Perform the bridge forwarding function with the copy. */
2576 		bridge_forward(sc, bif, mc);
2577 
2578 #ifdef DEV_NETMAP
2579 		/*
2580 		 * If netmap is enabled and has not already seen this packet,
2581 		 * then it will be consumed by bridge_forward().
2582 		 */
2583 		if ((if_getcapenable(bifp) & IFCAP_NETMAP) != 0 &&
2584 		    (m->m_flags & M_BRIDGE_INJECT) == 0) {
2585 			m_freem(m);
2586 			return (NULL);
2587 		}
2588 #endif
2589 
2590 		/*
2591 		 * Reinject the mbuf as arriving on the bridge so we have a
2592 		 * chance at claiming multicast packets. We can not loop back
2593 		 * here from ether_input as a bridge is never a member of a
2594 		 * bridge.
2595 		 */
2596 		KASSERT(bifp->if_bridge == NULL,
2597 		    ("loop created in bridge_input"));
2598 		mc2 = m_dup(m, M_NOWAIT);
2599 		if (mc2 != NULL) {
2600 			/* Keep the layer3 header aligned */
2601 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2602 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2603 		}
2604 		if (mc2 != NULL) {
2605 			mc2->m_pkthdr.rcvif = bifp;
2606 			mc2->m_flags &= ~M_BRIDGE_INJECT;
2607 			sc->sc_if_input(bifp, mc2);
2608 		}
2609 
2610 		/* Return the original packet for local processing. */
2611 		return (m);
2612 	}
2613 
2614 	if ((bif->bif_flags & IFBIF_STP) &&
2615 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2616 		return (m);
2617 	}
2618 
2619 #if defined(INET) || defined(INET6)
2620 #define	CARP_CHECK_WE_ARE_DST(iface) \
2621 	((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_dhost))
2622 #define	CARP_CHECK_WE_ARE_SRC(iface) \
2623 	((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_shost))
2624 #else
2625 #define	CARP_CHECK_WE_ARE_DST(iface)	false
2626 #define	CARP_CHECK_WE_ARE_SRC(iface)	false
2627 #endif
2628 
2629 #ifdef INET6
2630 #define	PFIL_HOOKED_INET6	PFIL_HOOKED_IN(V_inet6_pfil_head)
2631 #else
2632 #define	PFIL_HOOKED_INET6	false
2633 #endif
2634 
2635 #ifdef DEV_NETMAP
2636 #define	GRAB_FOR_NETMAP(ifp, m) do {					\
2637 	if ((if_getcapenable(ifp) & IFCAP_NETMAP) != 0 &&		\
2638 	    ((m)->m_flags & M_BRIDGE_INJECT) == 0) {			\
2639 		(ifp)->if_input(ifp, m);				\
2640 		return (NULL);						\
2641 	}								\
2642 } while (0)
2643 #else
2644 #define	GRAB_FOR_NETMAP(ifp, m)
2645 #endif
2646 
2647 #define GRAB_OUR_PACKETS(iface)						\
2648 	if ((iface)->if_type == IFT_GIF)				\
2649 		continue;						\
2650 	/* It is destined for us. */					\
2651 	if (memcmp(IF_LLADDR(iface), eh->ether_dhost, ETHER_ADDR_LEN) == 0 || \
2652 	    CARP_CHECK_WE_ARE_DST(iface)) {				\
2653 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2654 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2655 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2656 			if (error && bif->bif_addrmax) {		\
2657 				m_freem(m);				\
2658 				return (NULL);				\
2659 			}						\
2660 		}							\
2661 		m->m_pkthdr.rcvif = iface;				\
2662 		if ((iface) == ifp) {					\
2663 			/* Skip bridge processing... src == dest */	\
2664 			return (m);					\
2665 		}							\
2666 		/* It's passing over or to the bridge, locally. */	\
2667 		ETHER_BPF_MTAP(bifp, m);				\
2668 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);		\
2669 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);\
2670 		/* Hand the packet over to netmap if necessary. */	\
2671 		GRAB_FOR_NETMAP(bifp, m);				\
2672 		/* Filter on the physical interface. */			\
2673 		if (V_pfil_local_phys && (PFIL_HOOKED_IN(V_inet_pfil_head) || \
2674 		    PFIL_HOOKED_INET6)) {				\
2675 			if (bridge_pfil(&m, NULL, ifp,			\
2676 			    PFIL_IN) != 0 || m == NULL) {		\
2677 				return (NULL);				\
2678 			}						\
2679 		}							\
2680 		if ((iface) != bifp)					\
2681 			ETHER_BPF_MTAP(iface, m);			\
2682 		return (m);						\
2683 	}								\
2684 									\
2685 	/* We just received a packet that we sent out. */		\
2686 	if (memcmp(IF_LLADDR(iface), eh->ether_shost, ETHER_ADDR_LEN) == 0 || \
2687 	    CARP_CHECK_WE_ARE_SRC(iface)) {				\
2688 		m_freem(m);						\
2689 		return (NULL);						\
2690 	}
2691 
2692 	/*
2693 	 * Unicast.  Make sure it's not for the bridge.
2694 	 */
2695 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2696 
2697 	/*
2698 	 * Give a chance for ifp at first priority. This will help when	the
2699 	 * packet comes through the interface like VLAN's with the same MACs
2700 	 * on several interfaces from the same bridge. This also will save
2701 	 * some CPU cycles in case the destination interface and the input
2702 	 * interface (eq ifp) are the same.
2703 	 */
2704 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2705 
2706 	/* Now check the all bridge members. */
2707 	CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2708 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2709 	}
2710 
2711 #undef CARP_CHECK_WE_ARE_DST
2712 #undef CARP_CHECK_WE_ARE_SRC
2713 #undef PFIL_HOOKED_INET6
2714 #undef GRAB_FOR_NETMAP
2715 #undef GRAB_OUR_PACKETS
2716 
2717 	/* Perform the bridge forwarding function. */
2718 	bridge_forward(sc, bif, m);
2719 
2720 	return (NULL);
2721 }
2722 
2723 /*
2724  * Inject a packet back into the host ethernet stack.  This will generally only
2725  * be used by netmap when an application writes to the host TX ring.  The
2726  * M_BRIDGE_INJECT flag ensures that the packet is re-routed to the bridge
2727  * interface after ethernet processing.
2728  */
2729 static void
2730 bridge_inject(struct ifnet *ifp, struct mbuf *m)
2731 {
2732 	struct bridge_softc *sc;
2733 
2734 	KASSERT((if_getcapenable(ifp) & IFCAP_NETMAP) != 0,
2735 	    ("%s: iface %s is not running in netmap mode",
2736 	    __func__, if_name(ifp)));
2737 	KASSERT((m->m_flags & M_BRIDGE_INJECT) == 0,
2738 	    ("%s: mbuf %p has M_BRIDGE_INJECT set", __func__, m));
2739 
2740 	m->m_flags |= M_BRIDGE_INJECT;
2741 	sc = if_getsoftc(ifp);
2742 	sc->sc_if_input(ifp, m);
2743 }
2744 
2745 /*
2746  * bridge_broadcast:
2747  *
2748  *	Send a frame to all interfaces that are members of
2749  *	the bridge, except for the one on which the packet
2750  *	arrived.
2751  *
2752  *	NOTE: Releases the lock on return.
2753  */
2754 static void
2755 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2756     struct mbuf *m, int runfilt)
2757 {
2758 	struct bridge_iflist *dbif, *sbif;
2759 	struct mbuf *mc;
2760 	struct ifnet *dst_if;
2761 	int used = 0, i;
2762 
2763 	NET_EPOCH_ASSERT();
2764 
2765 	sbif = bridge_lookup_member_if(sc, src_if);
2766 
2767 	/* Filter on the bridge interface before broadcasting */
2768 	if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2769 #ifdef INET6
2770 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2771 #endif
2772 	    )) {
2773 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2774 			return;
2775 		if (m == NULL)
2776 			return;
2777 	}
2778 
2779 	CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2780 		dst_if = dbif->bif_ifp;
2781 		if (dst_if == src_if)
2782 			continue;
2783 
2784 		/* Private segments can not talk to each other */
2785 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2786 			continue;
2787 
2788 		if ((dbif->bif_flags & IFBIF_STP) &&
2789 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2790 			continue;
2791 
2792 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2793 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2794 			continue;
2795 
2796 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2797 			continue;
2798 
2799 		if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
2800 			mc = m;
2801 			used = 1;
2802 		} else {
2803 			mc = m_dup(m, M_NOWAIT);
2804 			if (mc == NULL) {
2805 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2806 				continue;
2807 			}
2808 		}
2809 
2810 		/*
2811 		 * Filter on the output interface. Pass a NULL bridge interface
2812 		 * pointer so we do not redundantly filter on the bridge for
2813 		 * each interface we broadcast on.
2814 		 */
2815 		if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2816 #ifdef INET6
2817 		    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2818 #endif
2819 		    )) {
2820 			if (used == 0) {
2821 				/* Keep the layer3 header aligned */
2822 				i = min(mc->m_pkthdr.len, max_protohdr);
2823 				mc = m_copyup(mc, i, ETHER_ALIGN);
2824 				if (mc == NULL) {
2825 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2826 					continue;
2827 				}
2828 			}
2829 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2830 				continue;
2831 			if (mc == NULL)
2832 				continue;
2833 		}
2834 
2835 		bridge_enqueue(sc, dst_if, mc);
2836 	}
2837 	if (used == 0)
2838 		m_freem(m);
2839 }
2840 
2841 /*
2842  * bridge_span:
2843  *
2844  *	Duplicate a packet out one or more interfaces that are in span mode,
2845  *	the original mbuf is unmodified.
2846  */
2847 static void
2848 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2849 {
2850 	struct bridge_iflist *bif;
2851 	struct ifnet *dst_if;
2852 	struct mbuf *mc;
2853 
2854 	NET_EPOCH_ASSERT();
2855 
2856 	if (CK_LIST_EMPTY(&sc->sc_spanlist))
2857 		return;
2858 
2859 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2860 		dst_if = bif->bif_ifp;
2861 
2862 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2863 			continue;
2864 
2865 		mc = m_dup(m, M_NOWAIT);
2866 		if (mc == NULL) {
2867 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2868 			continue;
2869 		}
2870 
2871 		bridge_enqueue(sc, dst_if, mc);
2872 	}
2873 }
2874 
2875 /*
2876  * bridge_rtupdate:
2877  *
2878  *	Add a bridge routing entry.
2879  */
2880 static int
2881 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2882     struct bridge_iflist *bif, int setflags, uint8_t flags)
2883 {
2884 	struct bridge_rtnode *brt;
2885 	struct bridge_iflist *obif;
2886 	int error;
2887 
2888 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
2889 
2890 	/* Check the source address is valid and not multicast. */
2891 	if (ETHER_IS_MULTICAST(dst) ||
2892 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2893 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2894 		return (EINVAL);
2895 
2896 	/*
2897 	 * A route for this destination might already exist.  If so,
2898 	 * update it, otherwise create a new one.
2899 	 */
2900 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2901 		BRIDGE_RT_LOCK(sc);
2902 
2903 		/* Check again, now that we have the lock. There could have
2904 		 * been a race and we only want to insert this once. */
2905 		if (bridge_rtnode_lookup(sc, dst, vlan) != NULL) {
2906 			BRIDGE_RT_UNLOCK(sc);
2907 			return (0);
2908 		}
2909 
2910 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2911 			sc->sc_brtexceeded++;
2912 			BRIDGE_RT_UNLOCK(sc);
2913 			return (ENOSPC);
2914 		}
2915 		/* Check per interface address limits (if enabled) */
2916 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2917 			bif->bif_addrexceeded++;
2918 			BRIDGE_RT_UNLOCK(sc);
2919 			return (ENOSPC);
2920 		}
2921 
2922 		/*
2923 		 * Allocate a new bridge forwarding node, and
2924 		 * initialize the expiration time and Ethernet
2925 		 * address.
2926 		 */
2927 		brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2928 		if (brt == NULL) {
2929 			BRIDGE_RT_UNLOCK(sc);
2930 			return (ENOMEM);
2931 		}
2932 		brt->brt_vnet = curvnet;
2933 
2934 		if (bif->bif_flags & IFBIF_STICKY)
2935 			brt->brt_flags = IFBAF_STICKY;
2936 		else
2937 			brt->brt_flags = IFBAF_DYNAMIC;
2938 
2939 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2940 		brt->brt_vlan = vlan;
2941 
2942 		brt->brt_dst = bif;
2943 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2944 			uma_zfree(V_bridge_rtnode_zone, brt);
2945 			BRIDGE_RT_UNLOCK(sc);
2946 			return (error);
2947 		}
2948 		bif->bif_addrcnt++;
2949 
2950 		BRIDGE_RT_UNLOCK(sc);
2951 	}
2952 
2953 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2954 	    (obif = brt->brt_dst) != bif) {
2955 		MPASS(obif != NULL);
2956 
2957 		BRIDGE_RT_LOCK(sc);
2958 		brt->brt_dst->bif_addrcnt--;
2959 		brt->brt_dst = bif;
2960 		brt->brt_dst->bif_addrcnt++;
2961 		BRIDGE_RT_UNLOCK(sc);
2962 
2963 		if (V_log_mac_flap &&
2964 		    ppsratecheck(&V_log_last, &V_log_count, V_log_interval)) {
2965 			log(LOG_NOTICE,
2966 			    "%s: mac address %6D vlan %d moved from %s to %s\n",
2967 			    sc->sc_ifp->if_xname,
2968 			    &brt->brt_addr[0], ":",
2969 			    brt->brt_vlan,
2970 			    obif->bif_ifp->if_xname,
2971 			    bif->bif_ifp->if_xname);
2972 		}
2973 	}
2974 
2975 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2976 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2977 	if (setflags)
2978 		brt->brt_flags = flags;
2979 
2980 	return (0);
2981 }
2982 
2983 /*
2984  * bridge_rtlookup:
2985  *
2986  *	Lookup the destination interface for an address.
2987  */
2988 static struct ifnet *
2989 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2990 {
2991 	struct bridge_rtnode *brt;
2992 
2993 	NET_EPOCH_ASSERT();
2994 
2995 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2996 		return (NULL);
2997 
2998 	return (brt->brt_ifp);
2999 }
3000 
3001 /*
3002  * bridge_rttrim:
3003  *
3004  *	Trim the routine table so that we have a number
3005  *	of routing entries less than or equal to the
3006  *	maximum number.
3007  */
3008 static void
3009 bridge_rttrim(struct bridge_softc *sc)
3010 {
3011 	struct bridge_rtnode *brt, *nbrt;
3012 
3013 	NET_EPOCH_ASSERT();
3014 	BRIDGE_RT_LOCK_ASSERT(sc);
3015 
3016 	/* Make sure we actually need to do this. */
3017 	if (sc->sc_brtcnt <= sc->sc_brtmax)
3018 		return;
3019 
3020 	/* Force an aging cycle; this might trim enough addresses. */
3021 	bridge_rtage(sc);
3022 	if (sc->sc_brtcnt <= sc->sc_brtmax)
3023 		return;
3024 
3025 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3026 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
3027 			bridge_rtnode_destroy(sc, brt);
3028 			if (sc->sc_brtcnt <= sc->sc_brtmax)
3029 				return;
3030 		}
3031 	}
3032 }
3033 
3034 /*
3035  * bridge_timer:
3036  *
3037  *	Aging timer for the bridge.
3038  */
3039 static void
3040 bridge_timer(void *arg)
3041 {
3042 	struct bridge_softc *sc = arg;
3043 
3044 	BRIDGE_RT_LOCK_ASSERT(sc);
3045 
3046 	/* Destruction of rtnodes requires a proper vnet context */
3047 	CURVNET_SET(sc->sc_ifp->if_vnet);
3048 	bridge_rtage(sc);
3049 
3050 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
3051 		callout_reset(&sc->sc_brcallout,
3052 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
3053 	CURVNET_RESTORE();
3054 }
3055 
3056 /*
3057  * bridge_rtage:
3058  *
3059  *	Perform an aging cycle.
3060  */
3061 static void
3062 bridge_rtage(struct bridge_softc *sc)
3063 {
3064 	struct bridge_rtnode *brt, *nbrt;
3065 
3066 	BRIDGE_RT_LOCK_ASSERT(sc);
3067 
3068 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3069 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
3070 			if (time_uptime >= brt->brt_expire)
3071 				bridge_rtnode_destroy(sc, brt);
3072 		}
3073 	}
3074 }
3075 
3076 /*
3077  * bridge_rtflush:
3078  *
3079  *	Remove all dynamic addresses from the bridge.
3080  */
3081 static void
3082 bridge_rtflush(struct bridge_softc *sc, int full)
3083 {
3084 	struct bridge_rtnode *brt, *nbrt;
3085 
3086 	BRIDGE_RT_LOCK_ASSERT(sc);
3087 
3088 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3089 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3090 			bridge_rtnode_destroy(sc, brt);
3091 	}
3092 }
3093 
3094 /*
3095  * bridge_rtdaddr:
3096  *
3097  *	Remove an address from the table.
3098  */
3099 static int
3100 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3101 {
3102 	struct bridge_rtnode *brt;
3103 	int found = 0;
3104 
3105 	BRIDGE_RT_LOCK(sc);
3106 
3107 	/*
3108 	 * If vlan is DOT1Q_VID_RSVD_IMPL then we want to delete for all vlans
3109 	 * so the lookup may return more than one.
3110 	 */
3111 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
3112 		bridge_rtnode_destroy(sc, brt);
3113 		found = 1;
3114 	}
3115 
3116 	BRIDGE_RT_UNLOCK(sc);
3117 
3118 	return (found ? 0 : ENOENT);
3119 }
3120 
3121 /*
3122  * bridge_rtdelete:
3123  *
3124  *	Delete routes to a speicifc member interface.
3125  */
3126 static void
3127 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3128 {
3129 	struct bridge_rtnode *brt, *nbrt;
3130 
3131 	BRIDGE_RT_LOCK_ASSERT(sc);
3132 
3133 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3134 		if (brt->brt_ifp == ifp && (full ||
3135 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3136 			bridge_rtnode_destroy(sc, brt);
3137 	}
3138 }
3139 
3140 /*
3141  * bridge_rtable_init:
3142  *
3143  *	Initialize the route table for this bridge.
3144  */
3145 static void
3146 bridge_rtable_init(struct bridge_softc *sc)
3147 {
3148 	int i;
3149 
3150 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3151 	    M_DEVBUF, M_WAITOK);
3152 
3153 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3154 		CK_LIST_INIT(&sc->sc_rthash[i]);
3155 
3156 	sc->sc_rthash_key = arc4random();
3157 	CK_LIST_INIT(&sc->sc_rtlist);
3158 }
3159 
3160 /*
3161  * bridge_rtable_fini:
3162  *
3163  *	Deconstruct the route table for this bridge.
3164  */
3165 static void
3166 bridge_rtable_fini(struct bridge_softc *sc)
3167 {
3168 
3169 	KASSERT(sc->sc_brtcnt == 0,
3170 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3171 	free(sc->sc_rthash, M_DEVBUF);
3172 }
3173 
3174 /*
3175  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3176  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3177  */
3178 #define	mix(a, b, c)							\
3179 do {									\
3180 	a -= b; a -= c; a ^= (c >> 13);					\
3181 	b -= c; b -= a; b ^= (a << 8);					\
3182 	c -= a; c -= b; c ^= (b >> 13);					\
3183 	a -= b; a -= c; a ^= (c >> 12);					\
3184 	b -= c; b -= a; b ^= (a << 16);					\
3185 	c -= a; c -= b; c ^= (b >> 5);					\
3186 	a -= b; a -= c; a ^= (c >> 3);					\
3187 	b -= c; b -= a; b ^= (a << 10);					\
3188 	c -= a; c -= b; c ^= (b >> 15);					\
3189 } while (/*CONSTCOND*/0)
3190 
3191 static __inline uint32_t
3192 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3193 {
3194 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3195 
3196 	b += addr[5] << 8;
3197 	b += addr[4];
3198 	a += addr[3] << 24;
3199 	a += addr[2] << 16;
3200 	a += addr[1] << 8;
3201 	a += addr[0];
3202 
3203 	mix(a, b, c);
3204 
3205 	return (c & BRIDGE_RTHASH_MASK);
3206 }
3207 
3208 #undef mix
3209 
3210 static int
3211 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3212 {
3213 	int i, d;
3214 
3215 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3216 		d = ((int)a[i]) - ((int)b[i]);
3217 	}
3218 
3219 	return (d);
3220 }
3221 
3222 /*
3223  * bridge_rtnode_lookup:
3224  *
3225  *	Look up a bridge route node for the specified destination. Compare the
3226  *	vlan id or if zero then just return the first match.
3227  */
3228 static struct bridge_rtnode *
3229 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
3230 {
3231 	struct bridge_rtnode *brt;
3232 	uint32_t hash;
3233 	int dir;
3234 
3235 	BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3236 
3237 	hash = bridge_rthash(sc, addr);
3238 	CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3239 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3240 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == DOT1Q_VID_RSVD_IMPL))
3241 			return (brt);
3242 		if (dir > 0)
3243 			return (NULL);
3244 	}
3245 
3246 	return (NULL);
3247 }
3248 
3249 /*
3250  * bridge_rtnode_insert:
3251  *
3252  *	Insert the specified bridge node into the route table.  We
3253  *	assume the entry is not already in the table.
3254  */
3255 static int
3256 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3257 {
3258 	struct bridge_rtnode *lbrt;
3259 	uint32_t hash;
3260 	int dir;
3261 
3262 	BRIDGE_RT_LOCK_ASSERT(sc);
3263 
3264 	hash = bridge_rthash(sc, brt->brt_addr);
3265 
3266 	lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3267 	if (lbrt == NULL) {
3268 		CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3269 		goto out;
3270 	}
3271 
3272 	do {
3273 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3274 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3275 			return (EEXIST);
3276 		if (dir > 0) {
3277 			CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3278 			goto out;
3279 		}
3280 		if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3281 			CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3282 			goto out;
3283 		}
3284 		lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3285 	} while (lbrt != NULL);
3286 
3287 #ifdef DIAGNOSTIC
3288 	panic("bridge_rtnode_insert: impossible");
3289 #endif
3290 
3291 out:
3292 	CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3293 	sc->sc_brtcnt++;
3294 
3295 	return (0);
3296 }
3297 
3298 static void
3299 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3300 {
3301 	struct bridge_rtnode *brt;
3302 
3303 	brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3304 
3305 	CURVNET_SET(brt->brt_vnet);
3306 	uma_zfree(V_bridge_rtnode_zone, brt);
3307 	CURVNET_RESTORE();
3308 }
3309 
3310 /*
3311  * bridge_rtnode_destroy:
3312  *
3313  *	Destroy a bridge rtnode.
3314  */
3315 static void
3316 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3317 {
3318 	BRIDGE_RT_LOCK_ASSERT(sc);
3319 
3320 	CK_LIST_REMOVE(brt, brt_hash);
3321 
3322 	CK_LIST_REMOVE(brt, brt_list);
3323 	sc->sc_brtcnt--;
3324 	brt->brt_dst->bif_addrcnt--;
3325 
3326 	NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3327 }
3328 
3329 /*
3330  * bridge_rtable_expire:
3331  *
3332  *	Set the expiry time for all routes on an interface.
3333  */
3334 static void
3335 bridge_rtable_expire(struct ifnet *ifp, int age)
3336 {
3337 	struct bridge_softc *sc = ifp->if_bridge;
3338 	struct bridge_rtnode *brt;
3339 
3340 	CURVNET_SET(ifp->if_vnet);
3341 	BRIDGE_RT_LOCK(sc);
3342 
3343 	/*
3344 	 * If the age is zero then flush, otherwise set all the expiry times to
3345 	 * age for the interface
3346 	 */
3347 	if (age == 0)
3348 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3349 	else {
3350 		CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3351 			/* Cap the expiry time to 'age' */
3352 			if (brt->brt_ifp == ifp &&
3353 			    brt->brt_expire > time_uptime + age &&
3354 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3355 				brt->brt_expire = time_uptime + age;
3356 		}
3357 	}
3358 	BRIDGE_RT_UNLOCK(sc);
3359 	CURVNET_RESTORE();
3360 }
3361 
3362 /*
3363  * bridge_state_change:
3364  *
3365  *	Callback from the bridgestp code when a port changes states.
3366  */
3367 static void
3368 bridge_state_change(struct ifnet *ifp, int state)
3369 {
3370 	struct bridge_softc *sc = ifp->if_bridge;
3371 	static const char *stpstates[] = {
3372 		"disabled",
3373 		"listening",
3374 		"learning",
3375 		"forwarding",
3376 		"blocking",
3377 		"discarding"
3378 	};
3379 
3380 	CURVNET_SET(ifp->if_vnet);
3381 	if (V_log_stp)
3382 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3383 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3384 	CURVNET_RESTORE();
3385 }
3386 
3387 /*
3388  * Send bridge packets through pfil if they are one of the types pfil can deal
3389  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3390  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3391  * that interface.
3392  */
3393 static int
3394 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3395 {
3396 	int snap, error, i, hlen;
3397 	struct ether_header *eh1, eh2;
3398 	struct ip *ip;
3399 	struct llc llc1;
3400 	u_int16_t ether_type;
3401 	pfil_return_t rv;
3402 
3403 	snap = 0;
3404 	error = -1;	/* Default error if not error == 0 */
3405 
3406 #if 0
3407 	/* we may return with the IP fields swapped, ensure its not shared */
3408 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3409 #endif
3410 
3411 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3412 		return (0); /* filtering is disabled */
3413 
3414 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3415 	if ((*mp)->m_len < i) {
3416 	    *mp = m_pullup(*mp, i);
3417 	    if (*mp == NULL) {
3418 		printf("%s: m_pullup failed\n", __func__);
3419 		return (-1);
3420 	    }
3421 	}
3422 
3423 	eh1 = mtod(*mp, struct ether_header *);
3424 	ether_type = ntohs(eh1->ether_type);
3425 
3426 	/*
3427 	 * Check for SNAP/LLC.
3428 	 */
3429 	if (ether_type < ETHERMTU) {
3430 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3431 
3432 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3433 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3434 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3435 		    llc2->llc_control == LLC_UI) {
3436 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3437 			snap = 1;
3438 		}
3439 	}
3440 
3441 	/*
3442 	 * If we're trying to filter bridge traffic, don't look at anything
3443 	 * other than IP and ARP traffic.  If the filter doesn't understand
3444 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3445 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3446 	 * but of course we don't have an AppleTalk filter to begin with.
3447 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3448 	 * ARP traffic.)
3449 	 */
3450 	switch (ether_type) {
3451 		case ETHERTYPE_ARP:
3452 		case ETHERTYPE_REVARP:
3453 			if (V_pfil_ipfw_arp == 0)
3454 				return (0); /* Automatically pass */
3455 			break;
3456 
3457 		case ETHERTYPE_IP:
3458 #ifdef INET6
3459 		case ETHERTYPE_IPV6:
3460 #endif /* INET6 */
3461 			break;
3462 		default:
3463 			/*
3464 			 * Check to see if the user wants to pass non-ip
3465 			 * packets, these will not be checked by pfil(9) and
3466 			 * passed unconditionally so the default is to drop.
3467 			 */
3468 			if (V_pfil_onlyip)
3469 				goto bad;
3470 	}
3471 
3472 	/* Run the packet through pfil before stripping link headers */
3473 	if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3474 	    dir == PFIL_OUT && ifp != NULL) {
3475 		switch (pfil_mbuf_out(V_link_pfil_head, mp, ifp, NULL)) {
3476 		case PFIL_DROPPED:
3477 			return (EACCES);
3478 		case PFIL_CONSUMED:
3479 			return (0);
3480 		}
3481 	}
3482 
3483 	/* Strip off the Ethernet header and keep a copy. */
3484 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3485 	m_adj(*mp, ETHER_HDR_LEN);
3486 
3487 	/* Strip off snap header, if present */
3488 	if (snap) {
3489 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3490 		m_adj(*mp, sizeof(struct llc));
3491 	}
3492 
3493 	/*
3494 	 * Check the IP header for alignment and errors
3495 	 */
3496 	if (dir == PFIL_IN) {
3497 		switch (ether_type) {
3498 			case ETHERTYPE_IP:
3499 				error = bridge_ip_checkbasic(mp);
3500 				break;
3501 #ifdef INET6
3502 			case ETHERTYPE_IPV6:
3503 				error = bridge_ip6_checkbasic(mp);
3504 				break;
3505 #endif /* INET6 */
3506 			default:
3507 				error = 0;
3508 		}
3509 		if (error)
3510 			goto bad;
3511 	}
3512 
3513 	error = 0;
3514 
3515 	/*
3516 	 * Run the packet through pfil
3517 	 */
3518 	rv = PFIL_PASS;
3519 	switch (ether_type) {
3520 	case ETHERTYPE_IP:
3521 		/*
3522 		 * Run pfil on the member interface and the bridge, both can
3523 		 * be skipped by clearing pfil_member or pfil_bridge.
3524 		 *
3525 		 * Keep the order:
3526 		 *   in_if -> bridge_if -> out_if
3527 		 */
3528 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3529 		    pfil_mbuf_out(V_inet_pfil_head, mp, bifp, NULL)) !=
3530 		    PFIL_PASS)
3531 			break;
3532 
3533 		if (V_pfil_member && ifp != NULL) {
3534 			rv = (dir == PFIL_OUT) ?
3535 			    pfil_mbuf_out(V_inet_pfil_head, mp, ifp, NULL) :
3536 			    pfil_mbuf_in(V_inet_pfil_head, mp, ifp, NULL);
3537 			if (rv != PFIL_PASS)
3538 				break;
3539 		}
3540 
3541 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3542 		    pfil_mbuf_in(V_inet_pfil_head, mp, bifp, NULL)) !=
3543 		    PFIL_PASS)
3544 			break;
3545 
3546 		/* check if we need to fragment the packet */
3547 		/* bridge_fragment generates a mbuf chain of packets */
3548 		/* that already include eth headers */
3549 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3550 			i = (*mp)->m_pkthdr.len;
3551 			if (i > ifp->if_mtu) {
3552 				error = bridge_fragment(ifp, mp, &eh2, snap,
3553 					    &llc1);
3554 				return (error);
3555 			}
3556 		}
3557 
3558 		/* Recalculate the ip checksum. */
3559 		ip = mtod(*mp, struct ip *);
3560 		hlen = ip->ip_hl << 2;
3561 		if (hlen < sizeof(struct ip))
3562 			goto bad;
3563 		if (hlen > (*mp)->m_len) {
3564 			if ((*mp = m_pullup(*mp, hlen)) == NULL)
3565 				goto bad;
3566 			ip = mtod(*mp, struct ip *);
3567 			if (ip == NULL)
3568 				goto bad;
3569 		}
3570 		ip->ip_sum = 0;
3571 		if (hlen == sizeof(struct ip))
3572 			ip->ip_sum = in_cksum_hdr(ip);
3573 		else
3574 			ip->ip_sum = in_cksum(*mp, hlen);
3575 
3576 		break;
3577 #ifdef INET6
3578 	case ETHERTYPE_IPV6:
3579 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3580 		    pfil_mbuf_out(V_inet6_pfil_head, mp, bifp, NULL)) !=
3581 		    PFIL_PASS)
3582 			break;
3583 
3584 		if (V_pfil_member && ifp != NULL) {
3585 			rv = (dir == PFIL_OUT) ?
3586 			    pfil_mbuf_out(V_inet6_pfil_head, mp, ifp, NULL) :
3587 			    pfil_mbuf_in(V_inet6_pfil_head, mp, ifp, NULL);
3588 			if (rv != PFIL_PASS)
3589 				break;
3590 		}
3591 
3592 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3593 		    pfil_mbuf_in(V_inet6_pfil_head, mp, bifp, NULL)) !=
3594 		    PFIL_PASS)
3595 			break;
3596 		break;
3597 #endif
3598 	}
3599 
3600 	switch (rv) {
3601 	case PFIL_CONSUMED:
3602 		return (0);
3603 	case PFIL_DROPPED:
3604 		return (EACCES);
3605 	default:
3606 		break;
3607 	}
3608 
3609 	error = -1;
3610 
3611 	/*
3612 	 * Finally, put everything back the way it was and return
3613 	 */
3614 	if (snap) {
3615 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3616 		if (*mp == NULL)
3617 			return (error);
3618 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3619 	}
3620 
3621 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3622 	if (*mp == NULL)
3623 		return (error);
3624 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3625 
3626 	return (0);
3627 
3628 bad:
3629 	m_freem(*mp);
3630 	*mp = NULL;
3631 	return (error);
3632 }
3633 
3634 /*
3635  * Perform basic checks on header size since
3636  * pfil assumes ip_input has already processed
3637  * it for it.  Cut-and-pasted from ip_input.c.
3638  * Given how simple the IPv6 version is,
3639  * does the IPv4 version really need to be
3640  * this complicated?
3641  *
3642  * XXX Should we update ipstat here, or not?
3643  * XXX Right now we update ipstat but not
3644  * XXX csum_counter.
3645  */
3646 static int
3647 bridge_ip_checkbasic(struct mbuf **mp)
3648 {
3649 	struct mbuf *m = *mp;
3650 	struct ip *ip;
3651 	int len, hlen;
3652 	u_short sum;
3653 
3654 	if (*mp == NULL)
3655 		return (-1);
3656 
3657 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3658 		if ((m = m_copyup(m, sizeof(struct ip),
3659 			(max_linkhdr + 3) & ~3)) == NULL) {
3660 			/* XXXJRT new stat, please */
3661 			KMOD_IPSTAT_INC(ips_toosmall);
3662 			goto bad;
3663 		}
3664 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3665 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3666 			KMOD_IPSTAT_INC(ips_toosmall);
3667 			goto bad;
3668 		}
3669 	}
3670 	ip = mtod(m, struct ip *);
3671 	if (ip == NULL) goto bad;
3672 
3673 	if (ip->ip_v != IPVERSION) {
3674 		KMOD_IPSTAT_INC(ips_badvers);
3675 		goto bad;
3676 	}
3677 	hlen = ip->ip_hl << 2;
3678 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3679 		KMOD_IPSTAT_INC(ips_badhlen);
3680 		goto bad;
3681 	}
3682 	if (hlen > m->m_len) {
3683 		if ((m = m_pullup(m, hlen)) == NULL) {
3684 			KMOD_IPSTAT_INC(ips_badhlen);
3685 			goto bad;
3686 		}
3687 		ip = mtod(m, struct ip *);
3688 		if (ip == NULL) goto bad;
3689 	}
3690 
3691 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3692 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3693 	} else {
3694 		if (hlen == sizeof(struct ip)) {
3695 			sum = in_cksum_hdr(ip);
3696 		} else {
3697 			sum = in_cksum(m, hlen);
3698 		}
3699 	}
3700 	if (sum) {
3701 		KMOD_IPSTAT_INC(ips_badsum);
3702 		goto bad;
3703 	}
3704 
3705 	/* Retrieve the packet length. */
3706 	len = ntohs(ip->ip_len);
3707 
3708 	/*
3709 	 * Check for additional length bogosity
3710 	 */
3711 	if (len < hlen) {
3712 		KMOD_IPSTAT_INC(ips_badlen);
3713 		goto bad;
3714 	}
3715 
3716 	/*
3717 	 * Check that the amount of data in the buffers
3718 	 * is as at least much as the IP header would have us expect.
3719 	 * Drop packet if shorter than we expect.
3720 	 */
3721 	if (m->m_pkthdr.len < len) {
3722 		KMOD_IPSTAT_INC(ips_tooshort);
3723 		goto bad;
3724 	}
3725 
3726 	/* Checks out, proceed */
3727 	*mp = m;
3728 	return (0);
3729 
3730 bad:
3731 	*mp = m;
3732 	return (-1);
3733 }
3734 
3735 #ifdef INET6
3736 /*
3737  * Same as above, but for IPv6.
3738  * Cut-and-pasted from ip6_input.c.
3739  * XXX Should we update ip6stat, or not?
3740  */
3741 static int
3742 bridge_ip6_checkbasic(struct mbuf **mp)
3743 {
3744 	struct mbuf *m = *mp;
3745 	struct ip6_hdr *ip6;
3746 
3747 	/*
3748 	 * If the IPv6 header is not aligned, slurp it up into a new
3749 	 * mbuf with space for link headers, in the event we forward
3750 	 * it.  Otherwise, if it is aligned, make sure the entire base
3751 	 * IPv6 header is in the first mbuf of the chain.
3752 	 */
3753 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3754 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3755 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3756 			    (max_linkhdr + 3) & ~3)) == NULL) {
3757 			/* XXXJRT new stat, please */
3758 			IP6STAT_INC(ip6s_toosmall);
3759 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3760 			goto bad;
3761 		}
3762 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3763 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3764 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3765 			IP6STAT_INC(ip6s_toosmall);
3766 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3767 			goto bad;
3768 		}
3769 	}
3770 
3771 	ip6 = mtod(m, struct ip6_hdr *);
3772 
3773 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3774 		IP6STAT_INC(ip6s_badvers);
3775 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3776 		goto bad;
3777 	}
3778 
3779 	/* Checks out, proceed */
3780 	*mp = m;
3781 	return (0);
3782 
3783 bad:
3784 	*mp = m;
3785 	return (-1);
3786 }
3787 #endif /* INET6 */
3788 
3789 /*
3790  * bridge_fragment:
3791  *
3792  *	Fragment mbuf chain in multiple packets and prepend ethernet header.
3793  */
3794 static int
3795 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3796     int snap, struct llc *llc)
3797 {
3798 	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3799 	struct ip *ip;
3800 	int error = -1;
3801 
3802 	if (m->m_len < sizeof(struct ip) &&
3803 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3804 		goto dropit;
3805 	ip = mtod(m, struct ip *);
3806 
3807 	m->m_pkthdr.csum_flags |= CSUM_IP;
3808 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3809 	if (error)
3810 		goto dropit;
3811 
3812 	/*
3813 	 * Walk the chain and re-add the Ethernet header for
3814 	 * each mbuf packet.
3815 	 */
3816 	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3817 		nextpkt = mcur->m_nextpkt;
3818 		mcur->m_nextpkt = NULL;
3819 		if (snap) {
3820 			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3821 			if (mcur == NULL) {
3822 				error = ENOBUFS;
3823 				if (mprev != NULL)
3824 					mprev->m_nextpkt = nextpkt;
3825 				goto dropit;
3826 			}
3827 			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3828 		}
3829 
3830 		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3831 		if (mcur == NULL) {
3832 			error = ENOBUFS;
3833 			if (mprev != NULL)
3834 				mprev->m_nextpkt = nextpkt;
3835 			goto dropit;
3836 		}
3837 		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3838 
3839 		/*
3840 		 * The previous two M_PREPEND could have inserted one or two
3841 		 * mbufs in front so we have to update the previous packet's
3842 		 * m_nextpkt.
3843 		 */
3844 		mcur->m_nextpkt = nextpkt;
3845 		if (mprev != NULL)
3846 			mprev->m_nextpkt = mcur;
3847 		else {
3848 			/* The first mbuf in the original chain needs to be
3849 			 * updated. */
3850 			*mp = mcur;
3851 		}
3852 		mprev = mcur;
3853 	}
3854 
3855 	KMOD_IPSTAT_INC(ips_fragmented);
3856 	return (error);
3857 
3858 dropit:
3859 	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3860 		m = mcur->m_nextpkt;
3861 		m_freem(mcur);
3862 	}
3863 	return (error);
3864 }
3865 
3866 static void
3867 bridge_linkstate(struct ifnet *ifp)
3868 {
3869 	struct bridge_softc *sc = ifp->if_bridge;
3870 	struct bridge_iflist *bif;
3871 	struct epoch_tracker et;
3872 
3873 	NET_EPOCH_ENTER(et);
3874 
3875 	bif = bridge_lookup_member_if(sc, ifp);
3876 	if (bif == NULL) {
3877 		NET_EPOCH_EXIT(et);
3878 		return;
3879 	}
3880 	bridge_linkcheck(sc);
3881 
3882 	bstp_linkstate(&bif->bif_stp);
3883 
3884 	NET_EPOCH_EXIT(et);
3885 }
3886 
3887 static void
3888 bridge_linkcheck(struct bridge_softc *sc)
3889 {
3890 	struct bridge_iflist *bif;
3891 	int new_link, hasls;
3892 
3893 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3894 
3895 	new_link = LINK_STATE_DOWN;
3896 	hasls = 0;
3897 	/* Our link is considered up if at least one of our ports is active */
3898 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3899 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3900 			hasls++;
3901 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3902 			new_link = LINK_STATE_UP;
3903 			break;
3904 		}
3905 	}
3906 	if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3907 		/* If no interfaces support link-state then we default to up */
3908 		new_link = LINK_STATE_UP;
3909 	}
3910 	if_link_state_change(sc->sc_ifp, new_link);
3911 }
3912