xref: /freebsd/sys/net/if_bridge.c (revision 9764aa1ccad08a7ec53ed9b80741b9553f3fa4e6)
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 "opt_inet.h"
80 #include "opt_inet6.h"
81 
82 #define	EXTERR_CATEGORY	EXTERR_CAT_BRIDGE
83 
84 #include <sys/param.h>
85 #include <sys/ctype.h>  /* string functions */
86 #include <sys/eventhandler.h>
87 #include <sys/exterrvar.h>
88 #include <sys/jail.h>
89 #include <sys/kernel.h>
90 #include <sys/lock.h>
91 #include <sys/malloc.h>
92 #include <sys/mbuf.h>
93 #include <sys/module.h>
94 #include <sys/mutex.h>
95 #include <sys/priv.h>
96 #include <sys/proc.h>
97 #include <sys/protosw.h>
98 #include <sys/random.h>
99 #include <sys/systm.h>
100 #include <sys/socket.h> /* for net/if.h */
101 #include <sys/sockio.h>
102 #include <sys/syslog.h>
103 #include <sys/sysctl.h>
104 #include <sys/time.h>
105 
106 #include <vm/uma.h>
107 
108 #include <net/bpf.h>
109 #include <net/if.h>
110 #include <net/if_clone.h>
111 #include <net/if_dl.h>
112 #include <net/if_types.h>
113 #include <net/if_var.h>
114 #include <net/if_private.h>
115 #include <net/pfil.h>
116 #include <net/vnet.h>
117 
118 #include <netinet/in.h>
119 #include <netinet/in_systm.h>
120 #include <netinet/in_var.h>
121 #include <netinet/ip.h>
122 #include <netinet/ip_var.h>
123 #ifdef INET6
124 #include <netinet/ip6.h>
125 #include <netinet6/ip6_var.h>
126 #include <netinet6/in6_ifattach.h>
127 #endif
128 #if defined(INET) || defined(INET6)
129 #include <netinet/ip_carp.h>
130 #endif
131 #include <machine/in_cksum.h>
132 #include <netinet/if_ether.h>
133 #include <net/bridgestp.h>
134 #include <net/if_bridgevar.h>
135 #include <net/if_llc.h>
136 #include <net/if_vlan_var.h>
137 
138 #include <net/route.h>
139 
140 /*
141  * At various points in the code we need to know if we're hooked into the INET
142  * and/or INET6 pfil.  Define some macros to do that based on which IP versions
143  * are enabled in the kernel.  This avoids littering the rest of the code with
144  * #ifnet INET6 to avoid referencing V_inet6_pfil_head.
145  */
146 #ifdef INET6
147 #define		PFIL_HOOKED_IN_INET6	PFIL_HOOKED_IN(V_inet6_pfil_head)
148 #define		PFIL_HOOKED_OUT_INET6	PFIL_HOOKED_OUT(V_inet6_pfil_head)
149 #else
150 #define		PFIL_HOOKED_IN_INET6	false
151 #define		PFIL_HOOKED_OUT_INET6	false
152 #endif
153 
154 #ifdef INET
155 #define		PFIL_HOOKED_IN_INET	PFIL_HOOKED_IN(V_inet_pfil_head)
156 #define		PFIL_HOOKED_OUT_INET	PFIL_HOOKED_OUT(V_inet_pfil_head)
157 #else
158 #define		PFIL_HOOKED_IN_INET	false
159 #define		PFIL_HOOKED_OUT_INET	false
160 #endif
161 
162 #define		PFIL_HOOKED_IN_46	(PFIL_HOOKED_IN_INET6 || PFIL_HOOKED_IN_INET)
163 #define		PFIL_HOOKED_OUT_46	(PFIL_HOOKED_OUT_INET6 || PFIL_HOOKED_OUT_INET)
164 
165 /*
166  * Size of the route hash table.  Must be a power of two.
167  */
168 #ifndef BRIDGE_RTHASH_SIZE
169 #define	BRIDGE_RTHASH_SIZE		1024
170 #endif
171 
172 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
173 
174 /*
175  * Default maximum number of addresses to cache.
176  */
177 #ifndef BRIDGE_RTABLE_MAX
178 #define	BRIDGE_RTABLE_MAX		2000
179 #endif
180 
181 /*
182  * Timeout (in seconds) for entries learned dynamically.
183  */
184 #ifndef BRIDGE_RTABLE_TIMEOUT
185 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
186 #endif
187 
188 /*
189  * Number of seconds between walks of the route list.
190  */
191 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
192 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
193 #endif
194 
195 /*
196  * List of capabilities to possibly mask on the member interface.
197  */
198 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
199 					 IFCAP_TXCSUM_IPV6|IFCAP_MEXTPG)
200 
201 /*
202  * List of capabilities to strip
203  */
204 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
205 
206 /*
207  * Bridge locking
208  *
209  * The bridge relies heavily on the epoch(9) system to protect its data
210  * structures. This means we can safely use CK_LISTs while in NET_EPOCH, but we
211  * must ensure there is only one writer at a time.
212  *
213  * That is: for read accesses we only need to be in NET_EPOCH, but for write
214  * accesses we must hold:
215  *
216  *  - BRIDGE_RT_LOCK, for any change to bridge_rtnodes
217  *  - BRIDGE_LOCK, for any other change
218  *
219  * The BRIDGE_LOCK is a sleepable lock, because it is held across ioctl()
220  * calls to bridge member interfaces and these ioctl()s can sleep.
221  * The BRIDGE_RT_LOCK is a non-sleepable mutex, because it is sometimes
222  * required while we're in NET_EPOCH and then we're not allowed to sleep.
223  */
224 #define BRIDGE_LOCK_INIT(_sc)		do {			\
225 	sx_init(&(_sc)->sc_sx, "if_bridge");			\
226 	mtx_init(&(_sc)->sc_rt_mtx, "if_bridge rt", NULL, MTX_DEF);	\
227 } while (0)
228 #define BRIDGE_LOCK_DESTROY(_sc)	do {	\
229 	sx_destroy(&(_sc)->sc_sx);		\
230 	mtx_destroy(&(_sc)->sc_rt_mtx);		\
231 } while (0)
232 #define BRIDGE_LOCK(_sc)		sx_xlock(&(_sc)->sc_sx)
233 #define BRIDGE_UNLOCK(_sc)		sx_xunlock(&(_sc)->sc_sx)
234 #define BRIDGE_LOCK_ASSERT(_sc)		sx_assert(&(_sc)->sc_sx, SX_XLOCKED)
235 #define BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(_sc)	\
236 	    MPASS(in_epoch(net_epoch_preempt) || sx_xlocked(&(_sc)->sc_sx))
237 #define BRIDGE_UNLOCK_ASSERT(_sc)	sx_assert(&(_sc)->sc_sx, SX_UNLOCKED)
238 #define BRIDGE_RT_LOCK(_sc)		mtx_lock(&(_sc)->sc_rt_mtx)
239 #define BRIDGE_RT_UNLOCK(_sc)		mtx_unlock(&(_sc)->sc_rt_mtx)
240 #define BRIDGE_RT_LOCK_ASSERT(_sc)	mtx_assert(&(_sc)->sc_rt_mtx, MA_OWNED)
241 #define BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(_sc)	\
242 	    MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(_sc)->sc_rt_mtx))
243 
244 struct bridge_softc;
245 
246 /*
247  * Bridge interface list entry.
248  */
249 struct bridge_iflist {
250 	CK_LIST_ENTRY(bridge_iflist) bif_next;
251 	struct ifnet		*bif_ifp;	/* member if */
252 	struct bridge_softc	*bif_sc;	/* parent bridge */
253 	struct bstp_port	bif_stp;	/* STP state */
254 	uint32_t		bif_flags;	/* member if flags */
255 	int			bif_savedcaps;	/* saved capabilities */
256 	uint32_t		bif_addrmax;	/* max # of addresses */
257 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
258 	uint32_t		bif_addrexceeded;/* # of address violations */
259 	struct epoch_context	bif_epoch_ctx;
260 	ether_vlanid_t		bif_pvid;	/* port vlan id */
261 	ifbvlan_set_t		bif_vlan_set;	/* if allowed tagged vlans */
262 	uint16_t		bif_vlanproto;	/* vlan protocol */
263 };
264 
265 /*
266  * Bridge route node.
267  */
268 struct bridge_rtnode {
269 	CK_LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
270 	CK_LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
271 	struct bridge_iflist	*brt_dst;	/* destination if */
272 	unsigned long		brt_expire;	/* expiration time */
273 	uint8_t			brt_flags;	/* address flags */
274 	uint8_t			brt_addr[ETHER_ADDR_LEN];
275 	ether_vlanid_t		brt_vlan;	/* vlan id */
276 	struct	vnet		*brt_vnet;
277 	struct	epoch_context	brt_epoch_ctx;
278 };
279 #define	brt_ifp			brt_dst->bif_ifp
280 
281 /*
282  * Software state for each bridge.
283  */
284 struct bridge_softc {
285 	struct ifnet		*sc_ifp;	/* make this an interface */
286 	LIST_ENTRY(bridge_softc) sc_list;
287 	struct sx		sc_sx;
288 	struct mtx		sc_rt_mtx;
289 	uint32_t		sc_brtmax;	/* max # of addresses */
290 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
291 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
292 	struct callout		sc_brcallout;	/* bridge callout */
293 	CK_LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
294 	CK_LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
295 	CK_LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
296 	uint32_t		sc_rthash_key;	/* key for hash */
297 	CK_LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
298 	struct bstp_state	sc_stp;		/* STP state */
299 	uint32_t		sc_brtexceeded;	/* # of cache drops */
300 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
301 	struct ether_addr	sc_defaddr;	/* Default MAC address */
302 	if_input_fn_t		sc_if_input;	/* Saved copy of if_input */
303 	struct epoch_context	sc_epoch_ctx;
304 	ifbr_flags_t		sc_flags;	/* bridge flags */
305 	ether_vlanid_t		sc_defpvid;	/* default PVID */
306 };
307 
308 VNET_DEFINE_STATIC(struct sx, bridge_list_sx);
309 #define	V_bridge_list_sx	VNET(bridge_list_sx)
310 static eventhandler_tag bridge_detach_cookie;
311 
312 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
313 
314 VNET_DEFINE_STATIC(uma_zone_t, bridge_rtnode_zone);
315 #define	V_bridge_rtnode_zone	VNET(bridge_rtnode_zone)
316 
317 static int	bridge_clone_create(struct if_clone *, char *, size_t,
318 		    struct ifc_data *, struct ifnet **);
319 static int	bridge_clone_destroy(struct if_clone *, struct ifnet *, uint32_t);
320 
321 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
322 static void	bridge_mutecaps(struct bridge_softc *);
323 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
324 		    int);
325 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
326 static void	bridge_init(void *);
327 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
328 static bool	bridge_same(const void *, const void *);
329 static void	*bridge_get_softc(struct ifnet *);
330 static void	bridge_stop(struct ifnet *, int);
331 static int	bridge_transmit(struct ifnet *, struct mbuf *);
332 #ifdef ALTQ
333 static void	bridge_altq_start(if_t);
334 static int	bridge_altq_transmit(if_t, struct mbuf *);
335 #endif
336 static void	bridge_qflush(struct ifnet *);
337 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
338 static void	bridge_inject(struct ifnet *, struct mbuf *);
339 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
340 		    struct rtentry *);
341 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
342 		    struct mbuf *, struct bridge_iflist *);
343 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
344 
345 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
346 		    struct mbuf *m);
347 static bool	bridge_member_ifaddrs(void);
348 static void	bridge_timer(void *);
349 
350 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
351 		    struct mbuf *, int);
352 static void	bridge_span(struct bridge_softc *, struct mbuf *);
353 
354 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
355 		    ether_vlanid_t, struct bridge_iflist *, int, uint8_t);
356 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
357 		    ether_vlanid_t);
358 static void	bridge_rttrim(struct bridge_softc *);
359 static void	bridge_rtage(struct bridge_softc *);
360 static void	bridge_rtflush(struct bridge_softc *, int);
361 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
362 		    ether_vlanid_t);
363 static bool	bridge_vfilter_in(const struct bridge_iflist *, struct mbuf *);
364 static bool	bridge_vfilter_out(const struct bridge_iflist *,
365 		    const struct mbuf *);
366 
367 static void	bridge_rtable_init(struct bridge_softc *);
368 static void	bridge_rtable_fini(struct bridge_softc *);
369 
370 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
371 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
372 		    const uint8_t *, ether_vlanid_t);
373 static int	bridge_rtnode_insert(struct bridge_softc *,
374 		    struct bridge_rtnode *);
375 static void	bridge_rtnode_destroy(struct bridge_softc *,
376 		    struct bridge_rtnode *);
377 static void	bridge_rtable_expire(struct ifnet *, int);
378 static void	bridge_state_change(struct ifnet *, int);
379 
380 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
381 		    const char *name);
382 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
383 		    struct ifnet *ifp);
384 static void	bridge_delete_member(struct bridge_softc *,
385 		    struct bridge_iflist *, int);
386 static void	bridge_delete_span(struct bridge_softc *,
387 		    struct bridge_iflist *);
388 
389 static int	bridge_ioctl_add(struct bridge_softc *, void *);
390 static int	bridge_ioctl_del(struct bridge_softc *, void *);
391 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
392 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
393 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
394 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
395 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
396 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
397 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
398 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
399 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
400 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
401 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
402 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
403 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
404 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
405 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
406 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
407 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
408 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
409 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
410 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
411 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
412 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
413 static int	bridge_ioctl_sifpvid(struct bridge_softc *, void *);
414 static int	bridge_ioctl_sifvlanset(struct bridge_softc *, void *);
415 static int	bridge_ioctl_gifvlanset(struct bridge_softc *, void *);
416 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
417 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
418 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
419 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
420 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
421 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
422 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
423 static int	bridge_ioctl_gflags(struct bridge_softc *, void *);
424 static int	bridge_ioctl_sflags(struct bridge_softc *, void *);
425 static int	bridge_ioctl_gdefpvid(struct bridge_softc *, void *);
426 static int	bridge_ioctl_sdefpvid(struct bridge_softc *, void *);
427 static int	bridge_ioctl_svlanproto(struct bridge_softc *, void *);
428 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
429 		    int);
430 #ifdef INET
431 static int	bridge_ip_checkbasic(struct mbuf **mp);
432 static int	bridge_fragment(struct ifnet *, struct mbuf **mp,
433 		    struct ether_header *, int, struct llc *);
434 #endif /* INET */
435 #ifdef INET6
436 static int	bridge_ip6_checkbasic(struct mbuf **mp);
437 #endif /* INET6 */
438 static void	bridge_linkstate(struct ifnet *ifp);
439 static void	bridge_linkcheck(struct bridge_softc *sc);
440 
441 /*
442  * Use the "null" value from IEEE 802.1Q-2014 Table 9-2
443  * to indicate untagged frames.
444  */
445 #define	VLANTAGOF(_m)	\
446     ((_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : DOT1Q_VID_NULL)
447 
448 static struct bstp_cb_ops bridge_ops = {
449 	.bcb_state = bridge_state_change,
450 	.bcb_rtage = bridge_rtable_expire
451 };
452 
453 SYSCTL_DECL(_net_link);
454 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
455     "Bridge");
456 
457 /* only pass IP[46] packets when pfil is enabled */
458 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
459 #define	V_pfil_onlyip	VNET(pfil_onlyip)
460 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
461     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
462     "Only pass IP packets when pfil is enabled");
463 
464 /* run pfil hooks on the bridge interface */
465 VNET_DEFINE_STATIC(int, pfil_bridge) = 0;
466 #define	V_pfil_bridge	VNET(pfil_bridge)
467 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
468     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
469     "Packet filter on the bridge interface");
470 
471 /* layer2 filter with ipfw */
472 VNET_DEFINE_STATIC(int, pfil_ipfw);
473 #define	V_pfil_ipfw	VNET(pfil_ipfw)
474 
475 /* layer2 ARP filter with ipfw */
476 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
477 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
478 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
479     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
480     "Filter ARP packets through IPFW layer2");
481 
482 /* run pfil hooks on the member interface */
483 VNET_DEFINE_STATIC(int, pfil_member) = 0;
484 #define	V_pfil_member	VNET(pfil_member)
485 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
486     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
487     "Packet filter on the member interface");
488 
489 /* run pfil hooks on the physical interface for locally destined packets */
490 VNET_DEFINE_STATIC(int, pfil_local_phys);
491 #define	V_pfil_local_phys	VNET(pfil_local_phys)
492 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
493     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
494     "Packet filter on the physical interface for locally destined packets");
495 
496 /* log STP state changes */
497 VNET_DEFINE_STATIC(int, log_stp);
498 #define	V_log_stp	VNET(log_stp)
499 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
500     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
501     "Log STP state changes");
502 
503 /* share MAC with first bridge member */
504 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
505 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
506 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
507     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
508     "Inherit MAC address from the first bridge member");
509 
510 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
511 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
512 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
513     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
514     "Allow overlap of link-local scope "
515     "zones of a bridge interface and the member interfaces");
516 
517 /* log MAC address port flapping */
518 VNET_DEFINE_STATIC(bool, log_mac_flap) = true;
519 #define	V_log_mac_flap	VNET(log_mac_flap)
520 SYSCTL_BOOL(_net_link_bridge, OID_AUTO, log_mac_flap,
521     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(log_mac_flap), true,
522     "Log MAC address port flapping");
523 
524 /* allow IP addresses on bridge members */
525 VNET_DEFINE_STATIC(bool, member_ifaddrs) = false;
526 #define	V_member_ifaddrs	VNET(member_ifaddrs)
527 SYSCTL_BOOL(_net_link_bridge, OID_AUTO, member_ifaddrs,
528     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(member_ifaddrs), false,
529     "Allow layer 3 addresses on bridge members");
530 
531 static bool
532 bridge_member_ifaddrs(void)
533 {
534 	return (V_member_ifaddrs);
535 }
536 
537 VNET_DEFINE_STATIC(int, log_interval) = 5;
538 VNET_DEFINE_STATIC(int, log_count) = 0;
539 VNET_DEFINE_STATIC(struct timeval, log_last) = { 0 };
540 
541 #define	V_log_interval	VNET(log_interval)
542 #define	V_log_count	VNET(log_count)
543 #define	V_log_last	VNET(log_last)
544 
545 struct bridge_control {
546 	int	(*bc_func)(struct bridge_softc *, void *);
547 	int	bc_argsize;
548 	int	bc_flags;
549 };
550 
551 #define	BC_F_COPYIN		0x01	/* copy arguments in */
552 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
553 #define	BC_F_SUSER		0x04	/* do super-user check */
554 
555 static const struct bridge_control bridge_control_table[] = {
556 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
557 	  BC_F_COPYIN|BC_F_SUSER },
558 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
559 	  BC_F_COPYIN|BC_F_SUSER },
560 
561 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
562 	  BC_F_COPYIN|BC_F_COPYOUT },
563 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
564 	  BC_F_COPYIN|BC_F_SUSER },
565 
566 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
567 	  BC_F_COPYIN|BC_F_SUSER },
568 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
569 	  BC_F_COPYOUT },
570 
571 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
572 	  BC_F_COPYIN|BC_F_COPYOUT },
573 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
574 	  BC_F_COPYIN|BC_F_COPYOUT },
575 
576 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
577 	  BC_F_COPYIN|BC_F_SUSER },
578 
579 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
580 	  BC_F_COPYIN|BC_F_SUSER },
581 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
582 	  BC_F_COPYOUT },
583 
584 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
585 	  BC_F_COPYIN|BC_F_SUSER },
586 
587 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
588 	  BC_F_COPYIN|BC_F_SUSER },
589 
590 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
591 	  BC_F_COPYOUT },
592 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
593 	  BC_F_COPYIN|BC_F_SUSER },
594 
595 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
596 	  BC_F_COPYOUT },
597 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
598 	  BC_F_COPYIN|BC_F_SUSER },
599 
600 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
601 	  BC_F_COPYOUT },
602 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
603 	  BC_F_COPYIN|BC_F_SUSER },
604 
605 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
606 	  BC_F_COPYOUT },
607 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
608 	  BC_F_COPYIN|BC_F_SUSER },
609 
610 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
611 	  BC_F_COPYIN|BC_F_SUSER },
612 
613 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
614 	  BC_F_COPYIN|BC_F_SUSER },
615 
616 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
617 	  BC_F_COPYIN|BC_F_SUSER },
618 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
619 	  BC_F_COPYIN|BC_F_SUSER },
620 
621 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
622 	  BC_F_COPYOUT },
623 
624 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
625 	  BC_F_COPYOUT },
626 
627 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
628 	  BC_F_COPYIN|BC_F_COPYOUT },
629 
630 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
631 	  BC_F_COPYIN|BC_F_SUSER },
632 
633 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
634 	  BC_F_COPYIN|BC_F_SUSER },
635 
636 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
637 	  BC_F_COPYIN|BC_F_SUSER },
638 
639 	{ bridge_ioctl_sifpvid,		sizeof(struct ifbreq),
640 	  BC_F_COPYIN|BC_F_SUSER },
641 
642 	{ bridge_ioctl_sifvlanset,	sizeof(struct ifbif_vlan_req),
643 	  BC_F_COPYIN|BC_F_SUSER },
644 
645 	{ bridge_ioctl_gifvlanset,	sizeof(struct ifbif_vlan_req),
646 	  BC_F_COPYIN|BC_F_COPYOUT },
647 
648 	{ bridge_ioctl_gflags,		sizeof(struct ifbrparam),
649 	  BC_F_COPYOUT },
650 
651 	{ bridge_ioctl_sflags,		sizeof(struct ifbrparam),
652 	  BC_F_COPYIN|BC_F_SUSER },
653 
654 	{ bridge_ioctl_gdefpvid,	sizeof(struct ifbrparam),
655 	  BC_F_COPYOUT },
656 
657 	{ bridge_ioctl_sdefpvid,	sizeof(struct ifbrparam),
658 	  BC_F_COPYIN|BC_F_SUSER },
659 
660 	{ bridge_ioctl_svlanproto,	sizeof(struct ifbreq),
661 	  BC_F_COPYIN|BC_F_SUSER },
662 };
663 static const int bridge_control_table_size = nitems(bridge_control_table);
664 
665 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list) =
666     LIST_HEAD_INITIALIZER();
667 #define	V_bridge_list	VNET(bridge_list)
668 #define	BRIDGE_LIST_LOCK_INIT(x)	sx_init(&V_bridge_list_sx,	\
669 					    "if_bridge list")
670 #define	BRIDGE_LIST_LOCK_DESTROY(x)	sx_destroy(&V_bridge_list_sx)
671 #define	BRIDGE_LIST_LOCK(x)		sx_xlock(&V_bridge_list_sx)
672 #define	BRIDGE_LIST_UNLOCK(x)		sx_xunlock(&V_bridge_list_sx)
673 
674 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
675 #define	V_bridge_cloner	VNET(bridge_cloner)
676 
677 static const char bridge_name[] = "bridge";
678 
679 static void
680 vnet_bridge_init(const void *unused __unused)
681 {
682 
683 	V_bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
684 	    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
685 	    UMA_ALIGN_PTR, 0);
686 	BRIDGE_LIST_LOCK_INIT();
687 
688 	struct if_clone_addreq req = {
689 		.create_f = bridge_clone_create,
690 		.destroy_f = bridge_clone_destroy,
691 		.flags = IFC_F_AUTOUNIT,
692 	};
693 	V_bridge_cloner = ifc_attach_cloner(bridge_name, &req);
694 }
695 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
696     vnet_bridge_init, NULL);
697 
698 static void
699 vnet_bridge_uninit(const void *unused __unused)
700 {
701 
702 	ifc_detach_cloner(V_bridge_cloner);
703 	V_bridge_cloner = NULL;
704 	BRIDGE_LIST_LOCK_DESTROY();
705 
706 	/* Callbacks may use the UMA zone. */
707 	NET_EPOCH_DRAIN_CALLBACKS();
708 
709 	uma_zdestroy(V_bridge_rtnode_zone);
710 }
711 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
712     vnet_bridge_uninit, NULL);
713 
714 static int
715 bridge_modevent(module_t mod, int type, void *data)
716 {
717 
718 	switch (type) {
719 	case MOD_LOAD:
720 		bridge_dn_p = bridge_dummynet;
721 		bridge_same_p = bridge_same;
722 		bridge_get_softc_p = bridge_get_softc;
723 		bridge_member_ifaddrs_p = bridge_member_ifaddrs;
724 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
725 		    ifnet_departure_event, bridge_ifdetach, NULL,
726 		    EVENTHANDLER_PRI_ANY);
727 		break;
728 	case MOD_UNLOAD:
729 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
730 		    bridge_detach_cookie);
731 		bridge_dn_p = NULL;
732 		bridge_same_p = NULL;
733 		bridge_get_softc_p = NULL;
734 		bridge_member_ifaddrs_p = NULL;
735 		break;
736 	default:
737 		return (EOPNOTSUPP);
738 	}
739 	return (0);
740 }
741 
742 static moduledata_t bridge_mod = {
743 	"if_bridge",
744 	bridge_modevent,
745 	0
746 };
747 
748 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
749 MODULE_VERSION(if_bridge, 1);
750 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
751 
752 /*
753  * handler for net.link.bridge.ipfw
754  */
755 static int
756 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
757 {
758 	int enable = V_pfil_ipfw;
759 	int error;
760 
761 	error = sysctl_handle_int(oidp, &enable, 0, req);
762 	enable &= 1;
763 
764 	if (enable != V_pfil_ipfw) {
765 		V_pfil_ipfw = enable;
766 
767 		/*
768 		 * Disable pfil so that ipfw doesnt run twice, if the user
769 		 * really wants both then they can re-enable pfil_bridge and/or
770 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
771 		 * layer2 type.
772 		 */
773 		if (V_pfil_ipfw) {
774 			V_pfil_onlyip = 0;
775 			V_pfil_bridge = 0;
776 			V_pfil_member = 0;
777 		}
778 	}
779 
780 	return (error);
781 }
782 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
783     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET | CTLFLAG_NEEDGIANT,
784     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
785     "Layer2 filter with IPFW");
786 
787 #ifdef VIMAGE
788 static void
789 bridge_reassign(struct ifnet *ifp, struct vnet *newvnet, char *arg)
790 {
791 	struct bridge_softc *sc = ifp->if_softc;
792 	struct bridge_iflist *bif;
793 
794 	BRIDGE_LOCK(sc);
795 
796 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
797 		bridge_delete_member(sc, bif, 0);
798 
799 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
800 		bridge_delete_span(sc, bif);
801 	}
802 
803 	BRIDGE_UNLOCK(sc);
804 
805 	ether_reassign(ifp, newvnet, arg);
806 }
807 #endif
808 
809 /*
810  * bridge_get_softc:
811  *
812  * Return the bridge softc for an ifnet.
813  */
814 static void *
815 bridge_get_softc(struct ifnet *ifp)
816 {
817 	struct bridge_iflist *bif;
818 
819 	NET_EPOCH_ASSERT();
820 
821 	bif = ifp->if_bridge;
822 	if (bif == NULL)
823 		return (NULL);
824 	return (bif->bif_sc);
825 }
826 
827 /*
828  * bridge_same:
829  *
830  * Return true if two interfaces are in the same bridge.  This is only used by
831  * bridgestp via bridge_same_p.
832  */
833 static bool
834 bridge_same(const void *bifap, const void *bifbp)
835 {
836 	const struct bridge_iflist *bifa = bifap, *bifb = bifbp;
837 
838 	NET_EPOCH_ASSERT();
839 
840 	if (bifa == NULL || bifb == NULL)
841 		return (false);
842 
843 	return (bifa->bif_sc == bifb->bif_sc);
844 }
845 
846 /*
847  * bridge_clone_create:
848  *
849  *	Create a new bridge instance.
850  */
851 static int
852 bridge_clone_create(struct if_clone *ifc, char *name, size_t len,
853     struct ifc_data *ifd, struct ifnet **ifpp)
854 {
855 	struct bridge_softc *sc;
856 	struct ifnet *ifp;
857 
858 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
859 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
860 
861 	BRIDGE_LOCK_INIT(sc);
862 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
863 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
864 
865 	/* Initialize our routing table. */
866 	bridge_rtable_init(sc);
867 
868 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_rt_mtx, 0);
869 
870 	CK_LIST_INIT(&sc->sc_iflist);
871 	CK_LIST_INIT(&sc->sc_spanlist);
872 
873 	ifp->if_softc = sc;
874 	if_initname(ifp, bridge_name, ifd->unit);
875 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
876 	ifp->if_capabilities = ifp->if_capenable = IFCAP_VLAN_HWTAGGING;
877 	ifp->if_ioctl = bridge_ioctl;
878 #ifdef ALTQ
879 	ifp->if_start = bridge_altq_start;
880 	ifp->if_transmit = bridge_altq_transmit;
881 	IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
882 	ifp->if_snd.ifq_drv_maxlen = 0;
883 	IFQ_SET_READY(&ifp->if_snd);
884 #else
885 	ifp->if_transmit = bridge_transmit;
886 #endif
887 	ifp->if_qflush = bridge_qflush;
888 	ifp->if_init = bridge_init;
889 	ifp->if_type = IFT_BRIDGE;
890 
891 	ether_gen_addr(ifp, &sc->sc_defaddr);
892 
893 	bstp_attach(&sc->sc_stp, &bridge_ops);
894 	ether_ifattach(ifp, sc->sc_defaddr.octet);
895 	/* Now undo some of the damage... */
896 	ifp->if_baudrate = 0;
897 #ifdef VIMAGE
898 	ifp->if_reassign = bridge_reassign;
899 #endif
900 	sc->sc_if_input = ifp->if_input;	/* ether_input */
901 	ifp->if_input = bridge_inject;
902 
903 	/*
904 	 * Allow BRIDGE_INPUT() to pass in packets originating from the bridge
905 	 * itself via bridge_inject().  This is required for netmap but
906 	 * otherwise has no effect.
907 	 */
908 	ifp->if_bridge_input = bridge_input;
909 
910 	BRIDGE_LIST_LOCK();
911 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
912 	BRIDGE_LIST_UNLOCK();
913 	*ifpp = ifp;
914 
915 	return (0);
916 }
917 
918 static void
919 bridge_clone_destroy_cb(struct epoch_context *ctx)
920 {
921 	struct bridge_softc *sc;
922 
923 	sc = __containerof(ctx, struct bridge_softc, sc_epoch_ctx);
924 
925 	BRIDGE_LOCK_DESTROY(sc);
926 	free(sc, M_DEVBUF);
927 }
928 
929 /*
930  * bridge_clone_destroy:
931  *
932  *	Destroy a bridge instance.
933  */
934 static int
935 bridge_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags)
936 {
937 	struct bridge_softc *sc = ifp->if_softc;
938 	struct bridge_iflist *bif;
939 	struct epoch_tracker et;
940 
941 	BRIDGE_LOCK(sc);
942 
943 	bridge_stop(ifp, 1);
944 	ifp->if_flags &= ~IFF_UP;
945 
946 	while ((bif = CK_LIST_FIRST(&sc->sc_iflist)) != NULL)
947 		bridge_delete_member(sc, bif, 0);
948 
949 	while ((bif = CK_LIST_FIRST(&sc->sc_spanlist)) != NULL) {
950 		bridge_delete_span(sc, bif);
951 	}
952 
953 	/* Tear down the routing table. */
954 	bridge_rtable_fini(sc);
955 
956 	BRIDGE_UNLOCK(sc);
957 
958 	NET_EPOCH_ENTER(et);
959 
960 	callout_drain(&sc->sc_brcallout);
961 
962 	BRIDGE_LIST_LOCK();
963 	LIST_REMOVE(sc, sc_list);
964 	BRIDGE_LIST_UNLOCK();
965 
966 	bstp_detach(&sc->sc_stp);
967 #ifdef ALTQ
968 	IFQ_PURGE(&ifp->if_snd);
969 #endif
970 	NET_EPOCH_EXIT(et);
971 
972 	ether_ifdetach(ifp);
973 	if_free(ifp);
974 
975 	NET_EPOCH_CALL(bridge_clone_destroy_cb, &sc->sc_epoch_ctx);
976 
977 	return (0);
978 }
979 
980 /*
981  * bridge_ioctl:
982  *
983  *	Handle a control request from the operator.
984  */
985 static int
986 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
987 {
988 	struct bridge_softc *sc = ifp->if_softc;
989 	struct ifreq *ifr = (struct ifreq *)data;
990 	struct bridge_iflist *bif;
991 	struct thread *td = curthread;
992 	union {
993 		struct ifbreq ifbreq;
994 		struct ifbifconf ifbifconf;
995 		struct ifbareq ifbareq;
996 		struct ifbaconf ifbaconf;
997 		struct ifbrparam ifbrparam;
998 		struct ifbropreq ifbropreq;
999 		struct ifbif_vlan_req ifvlanreq;
1000 	} args;
1001 	struct ifdrv *ifd = (struct ifdrv *) data;
1002 	const struct bridge_control *bc;
1003 	int error = 0, oldmtu;
1004 
1005 	BRIDGE_LOCK(sc);
1006 
1007 	switch (cmd) {
1008 	case SIOCADDMULTI:
1009 	case SIOCDELMULTI:
1010 		break;
1011 
1012 	case SIOCGDRVSPEC:
1013 	case SIOCSDRVSPEC:
1014 		if (ifd->ifd_cmd >= bridge_control_table_size) {
1015 			error = EXTERROR(EINVAL, "Invalid control command");
1016 			break;
1017 		}
1018 		bc = &bridge_control_table[ifd->ifd_cmd];
1019 
1020 		if (cmd == SIOCGDRVSPEC &&
1021 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
1022 			error = EXTERROR(EINVAL,
1023 			    "Inappropriate ioctl for command "
1024 			    "(expected SIOCSDRVSPEC)");
1025 			break;
1026 		}
1027 		else if (cmd == SIOCSDRVSPEC &&
1028 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
1029 			error = EXTERROR(EINVAL,
1030 			    "Inappropriate ioctl for command "
1031 			    "(expected SIOCGDRVSPEC)");
1032 			break;
1033 		}
1034 
1035 		if (bc->bc_flags & BC_F_SUSER) {
1036 			error = priv_check(td, PRIV_NET_BRIDGE);
1037 			if (error) {
1038 				EXTERROR(error, "PRIV_NET_BRIDGE required");
1039 				break;
1040 			}
1041 		}
1042 
1043 		if (ifd->ifd_len != bc->bc_argsize ||
1044 		    ifd->ifd_len > sizeof(args)) {
1045 			error = EXTERROR(EINVAL, "Invalid argument size");
1046 			break;
1047 		}
1048 
1049 		bzero(&args, sizeof(args));
1050 		if (bc->bc_flags & BC_F_COPYIN) {
1051 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
1052 			if (error)
1053 				break;
1054 		}
1055 
1056 		oldmtu = ifp->if_mtu;
1057 		error = (*bc->bc_func)(sc, &args);
1058 		if (error)
1059 			break;
1060 
1061 		/*
1062 		 * Bridge MTU may change during addition of the first port.
1063 		 * If it did, do network layer specific procedure.
1064 		 */
1065 		if (ifp->if_mtu != oldmtu)
1066 			if_notifymtu(ifp);
1067 
1068 		if (bc->bc_flags & BC_F_COPYOUT)
1069 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
1070 
1071 		break;
1072 
1073 	case SIOCSIFFLAGS:
1074 		if (!(ifp->if_flags & IFF_UP) &&
1075 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
1076 			/*
1077 			 * If interface is marked down and it is running,
1078 			 * then stop and disable it.
1079 			 */
1080 			bridge_stop(ifp, 1);
1081 		} else if ((ifp->if_flags & IFF_UP) &&
1082 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
1083 			/*
1084 			 * If interface is marked up and it is stopped, then
1085 			 * start it.
1086 			 */
1087 			BRIDGE_UNLOCK(sc);
1088 			(*ifp->if_init)(sc);
1089 			BRIDGE_LOCK(sc);
1090 		}
1091 		break;
1092 
1093 	case SIOCSIFMTU:
1094 		oldmtu = sc->sc_ifp->if_mtu;
1095 
1096 		if (ifr->ifr_mtu < IF_MINMTU) {
1097 			error = EXTERROR(EINVAL,
1098 			    "Requested MTU is lower than IF_MINMTU");
1099 			break;
1100 		}
1101 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1102 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
1103 			break;
1104 		}
1105 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1106 			error = (*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
1107 			    SIOCSIFMTU, (caddr_t)ifr);
1108 			if (error != 0) {
1109 				log(LOG_NOTICE, "%s: invalid MTU: %u for"
1110 				    " member %s\n", sc->sc_ifp->if_xname,
1111 				    ifr->ifr_mtu,
1112 				    bif->bif_ifp->if_xname);
1113 				error = EINVAL;
1114 				break;
1115 			}
1116 		}
1117 		if (error) {
1118 			/* Restore the previous MTU on all member interfaces. */
1119 			ifr->ifr_mtu = oldmtu;
1120 			CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1121 				(*bif->bif_ifp->if_ioctl)(bif->bif_ifp,
1122 				    SIOCSIFMTU, (caddr_t)ifr);
1123 			}
1124 			EXTERROR(error,
1125 			    "Failed to set MTU on member interface");
1126 		} else {
1127 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
1128 		}
1129 		break;
1130 	default:
1131 		/*
1132 		 * drop the lock as ether_ioctl() will call bridge_start() and
1133 		 * cause the lock to be recursed.
1134 		 */
1135 		BRIDGE_UNLOCK(sc);
1136 		error = ether_ioctl(ifp, cmd, data);
1137 		BRIDGE_LOCK(sc);
1138 		break;
1139 	}
1140 
1141 	BRIDGE_UNLOCK(sc);
1142 
1143 	return (error);
1144 }
1145 
1146 /*
1147  * bridge_mutecaps:
1148  *
1149  *	Clear or restore unwanted capabilities on the member interface
1150  */
1151 static void
1152 bridge_mutecaps(struct bridge_softc *sc)
1153 {
1154 	struct bridge_iflist *bif;
1155 	int enabled, mask;
1156 
1157 	BRIDGE_LOCK_ASSERT(sc);
1158 
1159 	/* Initial bitmask of capabilities to test */
1160 	mask = BRIDGE_IFCAPS_MASK;
1161 
1162 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1163 		/* Every member must support it or it's disabled */
1164 		mask &= bif->bif_savedcaps;
1165 	}
1166 
1167 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1168 		enabled = bif->bif_ifp->if_capenable;
1169 		enabled &= ~BRIDGE_IFCAPS_STRIP;
1170 		/* Strip off mask bits and enable them again if allowed */
1171 		enabled &= ~BRIDGE_IFCAPS_MASK;
1172 		enabled |= mask;
1173 		bridge_set_ifcap(sc, bif, enabled);
1174 	}
1175 }
1176 
1177 static void
1178 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
1179 {
1180 	struct ifnet *ifp = bif->bif_ifp;
1181 	struct ifreq ifr;
1182 	int error, mask, stuck;
1183 
1184 	bzero(&ifr, sizeof(ifr));
1185 	ifr.ifr_reqcap = set;
1186 
1187 	if (ifp->if_capenable != set) {
1188 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
1189 		if (error)
1190 			if_printf(sc->sc_ifp,
1191 			    "error setting capabilities on %s: %d\n",
1192 			    ifp->if_xname, error);
1193 		mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
1194 		stuck = ifp->if_capenable & mask & ~set;
1195 		if (stuck != 0)
1196 			if_printf(sc->sc_ifp,
1197 			    "can't disable some capabilities on %s: 0x%x\n",
1198 			    ifp->if_xname, stuck);
1199 	}
1200 }
1201 
1202 /*
1203  * bridge_lookup_member:
1204  *
1205  *	Lookup a bridge member interface.
1206  */
1207 static struct bridge_iflist *
1208 bridge_lookup_member(struct bridge_softc *sc, const char *name)
1209 {
1210 	struct bridge_iflist *bif;
1211 	struct ifnet *ifp;
1212 
1213 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1214 
1215 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1216 		ifp = bif->bif_ifp;
1217 		if (strcmp(ifp->if_xname, name) == 0)
1218 			return (bif);
1219 	}
1220 
1221 	return (NULL);
1222 }
1223 
1224 /*
1225  * bridge_lookup_member_if:
1226  *
1227  *	Lookup a bridge member interface by ifnet*.
1228  */
1229 static struct bridge_iflist *
1230 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
1231 {
1232 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
1233 	return (member_ifp->if_bridge);
1234 }
1235 
1236 static void
1237 bridge_delete_member_cb(struct epoch_context *ctx)
1238 {
1239 	struct bridge_iflist *bif;
1240 
1241 	bif = __containerof(ctx, struct bridge_iflist, bif_epoch_ctx);
1242 
1243 	free(bif, M_DEVBUF);
1244 }
1245 
1246 /*
1247  * bridge_delete_member:
1248  *
1249  *	Delete the specified member interface.
1250  */
1251 static void
1252 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1253     int gone)
1254 {
1255 	struct ifnet *ifs = bif->bif_ifp;
1256 	struct ifnet *fif = NULL;
1257 	struct bridge_iflist *bifl;
1258 
1259 	BRIDGE_LOCK_ASSERT(sc);
1260 
1261 	if (bif->bif_flags & IFBIF_STP)
1262 		bstp_disable(&bif->bif_stp);
1263 
1264 	ifs->if_bridge = NULL;
1265 	CK_LIST_REMOVE(bif, bif_next);
1266 
1267 	/*
1268 	 * If removing the interface that gave the bridge its mac address, set
1269 	 * the mac address of the bridge to the address of the next member, or
1270 	 * to its default address if no members are left.
1271 	 */
1272 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1273 		if (CK_LIST_EMPTY(&sc->sc_iflist)) {
1274 			bcopy(&sc->sc_defaddr,
1275 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1276 			sc->sc_ifaddr = NULL;
1277 		} else {
1278 			bifl = CK_LIST_FIRST(&sc->sc_iflist);
1279 			fif = bifl->bif_ifp;
1280 			bcopy(IF_LLADDR(fif),
1281 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1282 			sc->sc_ifaddr = fif;
1283 		}
1284 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1285 	}
1286 
1287 	bridge_linkcheck(sc);
1288 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1289 	BRIDGE_RT_LOCK(sc);
1290 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1291 	BRIDGE_RT_UNLOCK(sc);
1292 	KASSERT(bif->bif_addrcnt == 0,
1293 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1294 
1295 	ifs->if_bridge_output = NULL;
1296 	ifs->if_bridge_input = NULL;
1297 	ifs->if_bridge_linkstate = NULL;
1298 	if (!gone) {
1299 		switch (ifs->if_type) {
1300 		case IFT_ETHER:
1301 		case IFT_L2VLAN:
1302 			/*
1303 			 * Take the interface out of promiscuous mode, but only
1304 			 * if it was promiscuous in the first place. It might
1305 			 * not be if we're in the bridge_ioctl_add() error path.
1306 			 */
1307 			if (ifs->if_flags & IFF_PROMISC)
1308 				(void) ifpromisc(ifs, 0);
1309 			break;
1310 
1311 		case IFT_GIF:
1312 			break;
1313 
1314 		default:
1315 #ifdef DIAGNOSTIC
1316 			panic("bridge_delete_member: impossible");
1317 #endif
1318 			break;
1319 		}
1320 		/* Re-enable any interface capabilities */
1321 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1322 	}
1323 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1324 
1325 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1326 }
1327 
1328 /*
1329  * bridge_delete_span:
1330  *
1331  *	Delete the specified span interface.
1332  */
1333 static void
1334 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1335 {
1336 	BRIDGE_LOCK_ASSERT(sc);
1337 
1338 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1339 	    ("%s: not a span interface", __func__));
1340 
1341 	CK_LIST_REMOVE(bif, bif_next);
1342 
1343 	NET_EPOCH_CALL(bridge_delete_member_cb, &bif->bif_epoch_ctx);
1344 }
1345 
1346 static int
1347 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1348 {
1349 	struct ifbreq *req = arg;
1350 	struct bridge_iflist *bif = NULL;
1351 	struct ifnet *ifs;
1352 	int error = 0;
1353 
1354 	ifs = ifunit(req->ifbr_ifsname);
1355 	if (ifs == NULL)
1356 		return (EXTERROR(ENOENT, "No such interface",
1357 		    req->ifbr_ifsname));
1358 	if (ifs->if_ioctl == NULL)	/* must be supported */
1359 		return (EXTERROR(EINVAL, "Interface must support ioctl(2)"));
1360 
1361 	/*
1362 	 * If the new interface is a vlan(4), it could be a bridge SVI.
1363 	 * Don't allow such things to be added to bridges.
1364 	 */
1365 	if (ifs->if_type == IFT_L2VLAN) {
1366 		struct ifnet *parent;
1367 		struct epoch_tracker et;
1368 		bool is_bridge;
1369 
1370 		/*
1371 		 * Entering NET_EPOCH with BRIDGE_LOCK held, but this is okay
1372 		 * since we don't sleep here.
1373 		 */
1374 		NET_EPOCH_ENTER(et);
1375 		parent = VLAN_TRUNKDEV(ifs);
1376 		is_bridge = (parent != NULL && parent->if_type == IFT_BRIDGE);
1377 		NET_EPOCH_EXIT(et);
1378 
1379 		if (is_bridge)
1380 			return (EXTERROR(EINVAL,
1381 			    "Bridge SVI cannot be added to a bridge"));
1382 	}
1383 
1384 	/* If it's in the span list, it can't be a member. */
1385 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1386 		if (ifs == bif->bif_ifp)
1387 			return (EXTERROR(EBUSY,
1388 			    "Span interface cannot be a member"));
1389 
1390 	if (ifs->if_bridge) {
1391 		struct bridge_iflist *sbif = ifs->if_bridge;
1392 		if (sbif->bif_sc == sc)
1393 			return (EXTERROR(EEXIST,
1394 			    "Interface is already a member of this bridge"));
1395 
1396 		return (EXTERROR(EBUSY,
1397 		    "Interface is already a member of another bridge"));
1398 	}
1399 
1400 	switch (ifs->if_type) {
1401 	case IFT_ETHER:
1402 	case IFT_L2VLAN:
1403 	case IFT_GIF:
1404 		/* permitted interface types */
1405 		break;
1406 	default:
1407 		return (EXTERROR(EINVAL, "Unsupported interface type"));
1408 	}
1409 
1410 #ifdef INET6
1411 	/*
1412 	 * Two valid inet6 addresses with link-local scope must not be
1413 	 * on the parent interface and the member interfaces at the
1414 	 * same time.  This restriction is needed to prevent violation
1415 	 * of link-local scope zone.  Attempts to add a member
1416 	 * interface which has inet6 addresses when the parent has
1417 	 * inet6 triggers removal of all inet6 addresses on the member
1418 	 * interface.
1419 	 */
1420 
1421 	/* Check if the parent interface has a link-local scope addr. */
1422 	if (V_allow_llz_overlap == 0 &&
1423 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1424 		/*
1425 		 * If any, remove all inet6 addresses from the member
1426 		 * interfaces.
1427 		 */
1428 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1429  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1430 				in6_ifdetach(bif->bif_ifp);
1431 				if_printf(sc->sc_ifp,
1432 				    "IPv6 addresses on %s have been removed "
1433 				    "before adding it as a member to prevent "
1434 				    "IPv6 address scope violation.\n",
1435 				    bif->bif_ifp->if_xname);
1436 			}
1437 		}
1438 		if (in6ifa_llaonifp(ifs)) {
1439 			in6_ifdetach(ifs);
1440 			if_printf(sc->sc_ifp,
1441 			    "IPv6 addresses on %s have been removed "
1442 			    "before adding it as a member to prevent "
1443 			    "IPv6 address scope violation.\n",
1444 			    ifs->if_xname);
1445 		}
1446 	}
1447 #endif
1448 
1449 	/*
1450 	 * If member_ifaddrs is disabled, do not allow an Ethernet-like
1451 	 * interface with assigned IP addresses to be added to a bridge.
1452 	 */
1453 	if (!V_member_ifaddrs && ifs->if_type != IFT_GIF) {
1454 		struct ifaddr *ifa;
1455 
1456 		CK_STAILQ_FOREACH(ifa, &ifs->if_addrhead, ifa_link) {
1457 #ifdef INET
1458 			if (ifa->ifa_addr->sa_family == AF_INET)
1459 				return (EXTERROR(EINVAL,
1460 				    "Member interface may not have "
1461 				    "an IPv4 address configured"));
1462 #endif
1463 #ifdef INET6
1464 			if (ifa->ifa_addr->sa_family == AF_INET6)
1465 				return (EXTERROR(EINVAL,
1466 				    "Member interface may not have "
1467 				    "an IPv6 address configured"));
1468 #endif
1469 		}
1470 	}
1471 
1472 	/* Allow the first Ethernet member to define the MTU */
1473 	if (CK_LIST_EMPTY(&sc->sc_iflist))
1474 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1475 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1476 		struct ifreq ifr;
1477 
1478 		snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s",
1479 		    ifs->if_xname);
1480 		ifr.ifr_mtu = sc->sc_ifp->if_mtu;
1481 
1482 		error = (*ifs->if_ioctl)(ifs,
1483 		    SIOCSIFMTU, (caddr_t)&ifr);
1484 		if (error != 0) {
1485 			log(LOG_NOTICE, "%s: invalid MTU: %u for"
1486 			    " new member %s\n", sc->sc_ifp->if_xname,
1487 			    ifr.ifr_mtu,
1488 			    ifs->if_xname);
1489 			return (EXTERROR(EINVAL,
1490 			    "Failed to set MTU on new member"));
1491 		}
1492 	}
1493 
1494 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1495 	if (bif == NULL)
1496 		return (ENOMEM);
1497 
1498 	bif->bif_sc = sc;
1499 	bif->bif_ifp = ifs;
1500 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1501 	bif->bif_savedcaps = ifs->if_capenable;
1502 	bif->bif_vlanproto = ETHERTYPE_VLAN;
1503 	bif->bif_pvid = sc->sc_defpvid;
1504 	if (sc->sc_flags & IFBRF_DEFQINQ)
1505 		bif->bif_flags |= IFBIF_QINQ;
1506 
1507 	/*
1508 	 * Assign the interface's MAC address to the bridge if it's the first
1509 	 * member and the MAC address of the bridge has not been changed from
1510 	 * the default randomly generated one.
1511 	 */
1512 	if (V_bridge_inherit_mac && CK_LIST_EMPTY(&sc->sc_iflist) &&
1513 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr.octet, ETHER_ADDR_LEN)) {
1514 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1515 		sc->sc_ifaddr = ifs;
1516 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1517 	}
1518 
1519 	ifs->if_bridge = bif;
1520 	ifs->if_bridge_output = bridge_output;
1521 	ifs->if_bridge_input = bridge_input;
1522 	ifs->if_bridge_linkstate = bridge_linkstate;
1523 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1524 	/*
1525 	 * XXX: XLOCK HERE!?!
1526 	 *
1527 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1528 	 */
1529 	CK_LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1530 
1531 	/* Set interface capabilities to the intersection set of all members */
1532 	bridge_mutecaps(sc);
1533 	bridge_linkcheck(sc);
1534 
1535 	/* Place the interface into promiscuous mode */
1536 	switch (ifs->if_type) {
1537 		case IFT_ETHER:
1538 		case IFT_L2VLAN:
1539 			error = ifpromisc(ifs, 1);
1540 			break;
1541 	}
1542 
1543 	if (error)
1544 		bridge_delete_member(sc, bif, 0);
1545 	return (error);
1546 }
1547 
1548 static int
1549 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1550 {
1551 	struct ifbreq *req = arg;
1552 	struct bridge_iflist *bif;
1553 
1554 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1555 	if (bif == NULL)
1556 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1557 
1558 	bridge_delete_member(sc, bif, 0);
1559 
1560 	return (0);
1561 }
1562 
1563 static int
1564 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1565 {
1566 	struct ifbreq *req = arg;
1567 	struct bridge_iflist *bif;
1568 	struct bstp_port *bp;
1569 
1570 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1571 	if (bif == NULL)
1572 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1573 
1574 	bp = &bif->bif_stp;
1575 	req->ifbr_ifsflags = bif->bif_flags;
1576 	req->ifbr_state = bp->bp_state;
1577 	req->ifbr_priority = bp->bp_priority;
1578 	req->ifbr_path_cost = bp->bp_path_cost;
1579 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1580 	req->ifbr_proto = bp->bp_protover;
1581 	req->ifbr_role = bp->bp_role;
1582 	req->ifbr_stpflags = bp->bp_flags;
1583 	req->ifbr_addrcnt = bif->bif_addrcnt;
1584 	req->ifbr_addrmax = bif->bif_addrmax;
1585 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1586 	req->ifbr_pvid = bif->bif_pvid;
1587 	req->ifbr_vlanproto = bif->bif_vlanproto;
1588 
1589 	/* Copy STP state options as flags */
1590 	if (bp->bp_operedge)
1591 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1592 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1593 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1594 	if (bp->bp_ptp_link)
1595 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1596 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1597 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1598 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1599 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1600 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1601 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1602 	return (0);
1603 }
1604 
1605 static int
1606 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1607 {
1608 	struct epoch_tracker et;
1609 	struct ifbreq *req = arg;
1610 	struct bridge_iflist *bif;
1611 	struct bstp_port *bp;
1612 	int error;
1613 
1614 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1615 	if (bif == NULL)
1616 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1617 	bp = &bif->bif_stp;
1618 
1619 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1620 		/* SPAN is readonly */
1621 		return (EXTERROR(EINVAL, "Span interface cannot be modified"));
1622 
1623 	NET_EPOCH_ENTER(et);
1624 
1625 	if (req->ifbr_ifsflags & IFBIF_STP) {
1626 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1627 			error = bstp_enable(&bif->bif_stp);
1628 			if (error) {
1629 				NET_EPOCH_EXIT(et);
1630 				return (EXTERROR(error,
1631 				    "Failed to enable STP"));
1632 			}
1633 		}
1634 	} else {
1635 		if ((bif->bif_flags & IFBIF_STP) != 0)
1636 			bstp_disable(&bif->bif_stp);
1637 	}
1638 
1639 	/* Pass on STP flags */
1640 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1641 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1642 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1643 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1644 
1645 	/* Save the bits relating to the bridge */
1646 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1647 
1648 	NET_EPOCH_EXIT(et);
1649 
1650 	return (0);
1651 }
1652 
1653 static int
1654 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1655 {
1656 	struct ifbrparam *param = arg;
1657 
1658 	sc->sc_brtmax = param->ifbrp_csize;
1659 	bridge_rttrim(sc);
1660 
1661 	return (0);
1662 }
1663 
1664 static int
1665 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1666 {
1667 	struct ifbrparam *param = arg;
1668 
1669 	param->ifbrp_csize = sc->sc_brtmax;
1670 
1671 	return (0);
1672 }
1673 
1674 static int
1675 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1676 {
1677 	struct ifbifconf *bifc = arg;
1678 	struct bridge_iflist *bif;
1679 	struct ifbreq breq;
1680 	char *buf, *outbuf;
1681 	int count, buflen, len, error = 0;
1682 
1683 	count = 0;
1684 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1685 		count++;
1686 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1687 		count++;
1688 
1689 	buflen = sizeof(breq) * count;
1690 	if (bifc->ifbic_len == 0) {
1691 		bifc->ifbic_len = buflen;
1692 		return (0);
1693 	}
1694 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1695 	if (outbuf == NULL)
1696 		return (ENOMEM);
1697 
1698 	count = 0;
1699 	buf = outbuf;
1700 	len = min(bifc->ifbic_len, buflen);
1701 	bzero(&breq, sizeof(breq));
1702 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1703 		if (len < sizeof(breq))
1704 			break;
1705 
1706 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1707 		    sizeof(breq.ifbr_ifsname));
1708 		/* Fill in the ifbreq structure */
1709 		error = bridge_ioctl_gifflags(sc, &breq);
1710 		if (error)
1711 			break;
1712 		memcpy(buf, &breq, sizeof(breq));
1713 		count++;
1714 		buf += sizeof(breq);
1715 		len -= sizeof(breq);
1716 	}
1717 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1718 		if (len < sizeof(breq))
1719 			break;
1720 
1721 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1722 		    sizeof(breq.ifbr_ifsname));
1723 		breq.ifbr_ifsflags = bif->bif_flags;
1724 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1725 		memcpy(buf, &breq, sizeof(breq));
1726 		count++;
1727 		buf += sizeof(breq);
1728 		len -= sizeof(breq);
1729 	}
1730 
1731 	bifc->ifbic_len = sizeof(breq) * count;
1732 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1733 	free(outbuf, M_TEMP);
1734 	return (error);
1735 }
1736 
1737 static int
1738 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1739 {
1740 	struct ifbaconf *bac = arg;
1741 	struct bridge_rtnode *brt;
1742 	struct ifbareq bareq;
1743 	char *buf, *outbuf;
1744 	int count, buflen, len, error = 0;
1745 
1746 	if (bac->ifbac_len == 0)
1747 		return (0);
1748 
1749 	count = 0;
1750 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1751 		count++;
1752 	buflen = sizeof(bareq) * count;
1753 
1754 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
1755 	if (outbuf == NULL)
1756 		return (ENOMEM);
1757 
1758 	count = 0;
1759 	buf = outbuf;
1760 	len = min(bac->ifbac_len, buflen);
1761 	bzero(&bareq, sizeof(bareq));
1762 	CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1763 		if (len < sizeof(bareq))
1764 			goto out;
1765 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1766 		    sizeof(bareq.ifba_ifsname));
1767 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1768 		bareq.ifba_vlan = brt->brt_vlan;
1769 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1770 				time_uptime < brt->brt_expire)
1771 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1772 		else
1773 			bareq.ifba_expire = 0;
1774 		bareq.ifba_flags = brt->brt_flags;
1775 
1776 		memcpy(buf, &bareq, sizeof(bareq));
1777 		count++;
1778 		buf += sizeof(bareq);
1779 		len -= sizeof(bareq);
1780 	}
1781 out:
1782 	bac->ifbac_len = sizeof(bareq) * count;
1783 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1784 	free(outbuf, M_TEMP);
1785 	return (error);
1786 }
1787 
1788 static int
1789 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1790 {
1791 	struct ifbareq *req = arg;
1792 	struct bridge_iflist *bif;
1793 	struct epoch_tracker et;
1794 	int error;
1795 
1796 	NET_EPOCH_ENTER(et);
1797 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1798 	if (bif == NULL) {
1799 		NET_EPOCH_EXIT(et);
1800 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1801 	}
1802 
1803 	/* bridge_rtupdate() may acquire the lock. */
1804 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1805 	    req->ifba_flags);
1806 	NET_EPOCH_EXIT(et);
1807 
1808 	return (error);
1809 }
1810 
1811 static int
1812 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1813 {
1814 	struct ifbrparam *param = arg;
1815 
1816 	sc->sc_brttimeout = param->ifbrp_ctime;
1817 	return (0);
1818 }
1819 
1820 static int
1821 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1822 {
1823 	struct ifbrparam *param = arg;
1824 
1825 	param->ifbrp_ctime = sc->sc_brttimeout;
1826 	return (0);
1827 }
1828 
1829 static int
1830 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1831 {
1832 	struct ifbareq *req = arg;
1833 	int vlan = req->ifba_vlan;
1834 
1835 	/* Userspace uses '0' to mean 'any vlan' */
1836 	if (vlan == 0)
1837 		vlan = DOT1Q_VID_RSVD_IMPL;
1838 
1839 	return (bridge_rtdaddr(sc, req->ifba_dst, vlan));
1840 }
1841 
1842 static int
1843 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1844 {
1845 	struct ifbreq *req = arg;
1846 
1847 	BRIDGE_RT_LOCK(sc);
1848 	bridge_rtflush(sc, req->ifbr_ifsflags);
1849 	BRIDGE_RT_UNLOCK(sc);
1850 
1851 	return (0);
1852 }
1853 
1854 static int
1855 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1856 {
1857 	struct ifbrparam *param = arg;
1858 	struct bstp_state *bs = &sc->sc_stp;
1859 
1860 	param->ifbrp_prio = bs->bs_bridge_priority;
1861 	return (0);
1862 }
1863 
1864 static int
1865 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1866 {
1867 	struct ifbrparam *param = arg;
1868 
1869 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1870 }
1871 
1872 static int
1873 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1874 {
1875 	struct ifbrparam *param = arg;
1876 	struct bstp_state *bs = &sc->sc_stp;
1877 
1878 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1879 	return (0);
1880 }
1881 
1882 static int
1883 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1884 {
1885 	struct ifbrparam *param = arg;
1886 
1887 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1888 }
1889 
1890 static int
1891 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1892 {
1893 	struct ifbrparam *param = arg;
1894 	struct bstp_state *bs = &sc->sc_stp;
1895 
1896 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1897 	return (0);
1898 }
1899 
1900 static int
1901 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1902 {
1903 	struct ifbrparam *param = arg;
1904 
1905 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1906 }
1907 
1908 static int
1909 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1910 {
1911 	struct ifbrparam *param = arg;
1912 	struct bstp_state *bs = &sc->sc_stp;
1913 
1914 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1915 	return (0);
1916 }
1917 
1918 static int
1919 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1920 {
1921 	struct ifbrparam *param = arg;
1922 
1923 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1924 }
1925 
1926 static int
1927 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1928 {
1929 	struct ifbreq *req = arg;
1930 	struct bridge_iflist *bif;
1931 
1932 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1933 	if (bif == NULL)
1934 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1935 
1936 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1937 }
1938 
1939 static int
1940 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1941 {
1942 	struct ifbreq *req = arg;
1943 	struct bridge_iflist *bif;
1944 
1945 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1946 	if (bif == NULL)
1947 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1948 
1949 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1950 }
1951 
1952 static int
1953 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1954 {
1955 	struct ifbreq *req = arg;
1956 	struct bridge_iflist *bif;
1957 
1958 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1959 	if (bif == NULL)
1960 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1961 
1962 	bif->bif_addrmax = req->ifbr_addrmax;
1963 	return (0);
1964 }
1965 
1966 static int
1967 bridge_ioctl_sifpvid(struct bridge_softc *sc, void *arg)
1968 {
1969 	struct ifbreq *req = arg;
1970 	struct bridge_iflist *bif;
1971 
1972 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1973 	if (bif == NULL)
1974 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1975 
1976 	if (req->ifbr_pvid > DOT1Q_VID_MAX)
1977 		return (EXTERROR(EINVAL, "Invalid VLAN ID"));
1978 
1979 	bif->bif_pvid = req->ifbr_pvid;
1980 	return (0);
1981 }
1982 
1983 static int
1984 bridge_ioctl_sifvlanset(struct bridge_softc *sc, void *arg)
1985 {
1986 	struct ifbif_vlan_req *req = arg;
1987 	struct bridge_iflist *bif;
1988 
1989 	if ((sc->sc_flags & IFBRF_VLANFILTER) == 0)
1990 		return (EXTERROR(EINVAL, "VLAN filtering not enabled"));
1991 
1992 	bif = bridge_lookup_member(sc, req->bv_ifname);
1993 	if (bif == NULL)
1994 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
1995 
1996 	/* Reject invalid VIDs. */
1997 	if (BRVLAN_TEST(&req->bv_set, DOT1Q_VID_NULL) ||
1998 	    BRVLAN_TEST(&req->bv_set, DOT1Q_VID_RSVD_IMPL))
1999 		return (EXTERROR(EINVAL, "Invalid VLAN ID in set"));
2000 
2001 	switch (req->bv_op) {
2002 		/* Replace the existing vlan set with the new set */
2003 	case BRDG_VLAN_OP_SET:
2004 		BIT_COPY(BRVLAN_SETSIZE, &req->bv_set, &bif->bif_vlan_set);
2005 		break;
2006 
2007 		/* Modify the existing vlan set to add the given vlans */
2008 	case BRDG_VLAN_OP_ADD:
2009 		BIT_OR(BRVLAN_SETSIZE, &bif->bif_vlan_set, &req->bv_set);
2010 		break;
2011 
2012 		/* Modify the existing vlan set to remove the given vlans */
2013 	case BRDG_VLAN_OP_DEL:
2014 		BIT_ANDNOT(BRVLAN_SETSIZE, &bif->bif_vlan_set, &req->bv_set);
2015 		break;
2016 
2017 		/* Invalid or unknown operation */
2018 	default:
2019 		return (EXTERROR(EINVAL,
2020 		    "Unsupported BRDGSIFVLANSET operation"));
2021 	}
2022 
2023 	return (0);
2024 }
2025 
2026 static int
2027 bridge_ioctl_gifvlanset(struct bridge_softc *sc, void *arg)
2028 {
2029 	struct ifbif_vlan_req *req = arg;
2030 	struct bridge_iflist *bif;
2031 
2032 	bif = bridge_lookup_member(sc, req->bv_ifname);
2033 	if (bif == NULL)
2034 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
2035 
2036 	BIT_COPY(BRVLAN_SETSIZE, &bif->bif_vlan_set, &req->bv_set);
2037 	return (0);
2038 }
2039 
2040 static int
2041 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
2042 {
2043 	struct ifbreq *req = arg;
2044 	struct bridge_iflist *bif = NULL;
2045 	struct ifnet *ifs;
2046 
2047 	ifs = ifunit(req->ifbr_ifsname);
2048 	if (ifs == NULL)
2049 		return (EXTERROR(ENOENT, "No such interface"));
2050 
2051 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
2052 		if (ifs == bif->bif_ifp)
2053 			return (EXTERROR(EBUSY,
2054 			    "Interface is already a span port"));
2055 
2056 	if (ifs->if_bridge != NULL)
2057 		return (EXTERROR(EEXIST,
2058 		    "Interface is already a bridge member"));
2059 
2060 	switch (ifs->if_type) {
2061 		case IFT_ETHER:
2062 		case IFT_GIF:
2063 		case IFT_L2VLAN:
2064 			break;
2065 		default:
2066 			return (EXTERROR(EINVAL, "Unsupported interface type"));
2067 	}
2068 
2069 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
2070 	if (bif == NULL)
2071 		return (ENOMEM);
2072 
2073 	bif->bif_ifp = ifs;
2074 	bif->bif_flags = IFBIF_SPAN;
2075 
2076 	CK_LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
2077 
2078 	return (0);
2079 }
2080 
2081 static int
2082 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
2083 {
2084 	struct ifbreq *req = arg;
2085 	struct bridge_iflist *bif;
2086 	struct ifnet *ifs;
2087 
2088 	ifs = ifunit(req->ifbr_ifsname);
2089 	if (ifs == NULL)
2090 		return (EXTERROR(ENOENT, "No such interface"));
2091 
2092 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
2093 		if (ifs == bif->bif_ifp)
2094 			break;
2095 
2096 	if (bif == NULL)
2097 		return (EXTERROR(ENOENT, "Interface is not a span port"));
2098 
2099 	bridge_delete_span(sc, bif);
2100 
2101 	return (0);
2102 }
2103 
2104 static int
2105 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
2106 {
2107 	struct ifbropreq *req = arg;
2108 	struct bstp_state *bs = &sc->sc_stp;
2109 	struct bstp_port *root_port;
2110 
2111 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
2112 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
2113 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
2114 
2115 	root_port = bs->bs_root_port;
2116 	if (root_port == NULL)
2117 		req->ifbop_root_port = 0;
2118 	else
2119 		req->ifbop_root_port = root_port->bp_ifp->if_index;
2120 
2121 	req->ifbop_holdcount = bs->bs_txholdcount;
2122 	req->ifbop_priority = bs->bs_bridge_priority;
2123 	req->ifbop_protocol = bs->bs_protover;
2124 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
2125 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
2126 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
2127 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
2128 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
2129 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
2130 
2131 	return (0);
2132 }
2133 
2134 static int
2135 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
2136 {
2137 	struct ifbrparam *param = arg;
2138 
2139 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
2140 	return (0);
2141 }
2142 
2143 static int
2144 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
2145 {
2146 	struct ifbpstpconf *bifstp = arg;
2147 	struct bridge_iflist *bif;
2148 	struct bstp_port *bp;
2149 	struct ifbpstpreq bpreq;
2150 	char *buf, *outbuf;
2151 	int count, buflen, len, error = 0;
2152 
2153 	count = 0;
2154 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2155 		if ((bif->bif_flags & IFBIF_STP) != 0)
2156 			count++;
2157 	}
2158 
2159 	buflen = sizeof(bpreq) * count;
2160 	if (bifstp->ifbpstp_len == 0) {
2161 		bifstp->ifbpstp_len = buflen;
2162 		return (0);
2163 	}
2164 
2165 	outbuf = malloc(buflen, M_TEMP, M_NOWAIT | M_ZERO);
2166 	if (outbuf == NULL)
2167 		return (ENOMEM);
2168 
2169 	count = 0;
2170 	buf = outbuf;
2171 	len = min(bifstp->ifbpstp_len, buflen);
2172 	bzero(&bpreq, sizeof(bpreq));
2173 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2174 		if (len < sizeof(bpreq))
2175 			break;
2176 
2177 		if ((bif->bif_flags & IFBIF_STP) == 0)
2178 			continue;
2179 
2180 		bp = &bif->bif_stp;
2181 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
2182 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
2183 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
2184 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
2185 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
2186 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
2187 
2188 		memcpy(buf, &bpreq, sizeof(bpreq));
2189 		count++;
2190 		buf += sizeof(bpreq);
2191 		len -= sizeof(bpreq);
2192 	}
2193 
2194 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
2195 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
2196 	free(outbuf, M_TEMP);
2197 	return (error);
2198 }
2199 
2200 static int
2201 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
2202 {
2203 	struct ifbrparam *param = arg;
2204 
2205 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
2206 }
2207 
2208 static int
2209 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
2210 {
2211 	struct ifbrparam *param = arg;
2212 
2213 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
2214 }
2215 
2216 static int
2217 bridge_ioctl_gflags(struct bridge_softc *sc, void *arg)
2218 {
2219 	struct ifbrparam *param = arg;
2220 
2221 	param->ifbrp_flags = sc->sc_flags;
2222 
2223 	return (0);
2224 }
2225 
2226 static int
2227 bridge_ioctl_sflags(struct bridge_softc *sc, void *arg)
2228 {
2229 	struct ifbrparam *param = arg;
2230 
2231 	sc->sc_flags = param->ifbrp_flags;
2232 
2233 	return (0);
2234 }
2235 
2236 static int
2237 bridge_ioctl_gdefpvid(struct bridge_softc *sc, void *arg)
2238 {
2239 	struct ifbrparam *param = arg;
2240 
2241 	param->ifbrp_defpvid = sc->sc_defpvid;
2242 
2243 	return (0);
2244 }
2245 
2246 static int
2247 bridge_ioctl_sdefpvid(struct bridge_softc *sc, void *arg)
2248 {
2249 	struct ifbrparam *param = arg;
2250 
2251 	/* Reject invalid VIDs, but allow 0 to mean 'none'. */
2252 	if (param->ifbrp_defpvid > DOT1Q_VID_MAX)
2253 		return (EINVAL);
2254 
2255 	sc->sc_defpvid = param->ifbrp_defpvid;
2256 
2257 	return (0);
2258 }
2259 
2260 static int
2261 bridge_ioctl_svlanproto(struct bridge_softc *sc, void *arg)
2262 {
2263 	struct ifbreq *req = arg;
2264 	struct bridge_iflist *bif;
2265 
2266 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
2267 	if (bif == NULL)
2268 		return (EXTERROR(ENOENT, "Interface is not a bridge member"));
2269 
2270 	if (req->ifbr_vlanproto != ETHERTYPE_VLAN &&
2271 	    req->ifbr_vlanproto != ETHERTYPE_QINQ)
2272 		return (EXTERROR(EINVAL, "Invalid VLAN protocol"));
2273 
2274 	bif->bif_vlanproto = req->ifbr_vlanproto;
2275 
2276 	return (0);
2277 }
2278 /*
2279  * bridge_ifdetach:
2280  *
2281  *	Detach an interface from a bridge.  Called when a member
2282  *	interface is detaching.
2283  */
2284 static void
2285 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
2286 {
2287 	struct bridge_iflist *bif = ifp->if_bridge;
2288 	struct bridge_softc *sc = NULL;
2289 
2290 	if (bif)
2291 		sc = bif->bif_sc;
2292 
2293 	if (ifp->if_flags & IFF_RENAMING)
2294 		return;
2295 	if (V_bridge_cloner == NULL) {
2296 		/*
2297 		 * This detach handler can be called after
2298 		 * vnet_bridge_uninit().  Just return in that case.
2299 		 */
2300 		return;
2301 	}
2302 	/* Check if the interface is a bridge member */
2303 	if (sc != NULL) {
2304 		BRIDGE_LOCK(sc);
2305 		bridge_delete_member(sc, bif, 1);
2306 		BRIDGE_UNLOCK(sc);
2307 		return;
2308 	}
2309 
2310 	/* Check if the interface is a span port */
2311 	BRIDGE_LIST_LOCK();
2312 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
2313 		BRIDGE_LOCK(sc);
2314 		CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
2315 			if (ifp == bif->bif_ifp) {
2316 				bridge_delete_span(sc, bif);
2317 				break;
2318 			}
2319 
2320 		BRIDGE_UNLOCK(sc);
2321 	}
2322 	BRIDGE_LIST_UNLOCK();
2323 }
2324 
2325 /*
2326  * bridge_init:
2327  *
2328  *	Initialize a bridge interface.
2329  */
2330 static void
2331 bridge_init(void *xsc)
2332 {
2333 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
2334 	struct ifnet *ifp = sc->sc_ifp;
2335 
2336 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
2337 		return;
2338 
2339 	BRIDGE_LOCK(sc);
2340 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
2341 	    bridge_timer, sc);
2342 
2343 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
2344 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
2345 
2346 	BRIDGE_UNLOCK(sc);
2347 }
2348 
2349 /*
2350  * bridge_stop:
2351  *
2352  *	Stop the bridge interface.
2353  */
2354 static void
2355 bridge_stop(struct ifnet *ifp, int disable)
2356 {
2357 	struct bridge_softc *sc = ifp->if_softc;
2358 
2359 	BRIDGE_LOCK_ASSERT(sc);
2360 
2361 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2362 		return;
2363 
2364 	BRIDGE_RT_LOCK(sc);
2365 	callout_stop(&sc->sc_brcallout);
2366 
2367 	bstp_stop(&sc->sc_stp);
2368 
2369 	bridge_rtflush(sc, IFBF_FLUSHDYN);
2370 	BRIDGE_RT_UNLOCK(sc);
2371 
2372 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
2373 }
2374 
2375 /*
2376  * bridge_enqueue:
2377  *
2378  *	Enqueue a packet on a bridge member interface.
2379  *
2380  */
2381 static int
2382 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m,
2383     struct bridge_iflist *bif)
2384 {
2385 	int len, err = 0;
2386 	short mflags;
2387 	struct mbuf *m0;
2388 
2389 	/*
2390 	 * Find the bridge member port this packet is being sent on, if the
2391 	 * caller didn't already provide it.
2392 	 */
2393 	if (bif == NULL)
2394 		bif = bridge_lookup_member_if(sc, dst_ifp);
2395 	if (bif == NULL) {
2396 		/* Perhaps the interface was removed from the bridge */
2397 		m_freem(m);
2398 		return (EINVAL);
2399 	}
2400 
2401 	/* We may be sending a fragment so traverse the mbuf */
2402 	for (; m; m = m0) {
2403 		m0 = m->m_nextpkt;
2404 		m->m_nextpkt = NULL;
2405 		len = m->m_pkthdr.len;
2406 		mflags = m->m_flags;
2407 
2408 		/*
2409 		 * If the native VLAN ID of the outgoing interface matches the
2410 		 * VLAN ID of the frame, remove the VLAN tag.
2411 		 */
2412 		if (bif->bif_pvid != DOT1Q_VID_NULL &&
2413 		    VLANTAGOF(m) == bif->bif_pvid) {
2414 			m->m_flags &= ~M_VLANTAG;
2415 			m->m_pkthdr.ether_vtag = 0;
2416 		}
2417 
2418 		/*
2419 		 * There are two cases where we have to insert our own tag:
2420 		 * if the member interface doesn't support hardware tagging,
2421 		 * or if the tag proto is not 802.1q.
2422 		 */
2423 		if ((m->m_flags & M_VLANTAG) &&
2424 		    ((dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0 ||
2425 		      bif->bif_vlanproto != ETHERTYPE_VLAN)) {
2426 			m = ether_vlanencap_proto(m, m->m_pkthdr.ether_vtag,
2427 			    bif->bif_vlanproto);
2428 			if (m == NULL) {
2429 				if_printf(dst_ifp,
2430 				    "unable to prepend VLAN header\n");
2431 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
2432 				continue;
2433 			}
2434 			m->m_flags &= ~M_VLANTAG;
2435 		}
2436 
2437 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
2438 		/*
2439 		 * XXXZL: gif(4) requires the af to be saved in csum_data field
2440 		 * so that gif_transmit() routine can pull it back.
2441 		 */
2442 		if (dst_ifp->if_type == IFT_GIF)
2443 			m->m_pkthdr.csum_data = AF_LINK;
2444 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
2445 			int n;
2446 
2447 			for (m = m0, n = 1; m != NULL; m = m0, n++) {
2448 				m0 = m->m_nextpkt;
2449 				m_freem(m);
2450 			}
2451 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, n);
2452 			break;
2453 		}
2454 
2455 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
2456 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
2457 		if (mflags & M_MCAST)
2458 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
2459 	}
2460 
2461 	return (err);
2462 }
2463 
2464 /*
2465  * bridge_dummynet:
2466  *
2467  * 	Receive a queued packet from dummynet and pass it on to the output
2468  * 	interface.
2469  *
2470  *	The mbuf has the Ethernet header already attached.
2471  */
2472 static void
2473 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
2474 {
2475 	struct bridge_iflist *bif = ifp->if_bridge;
2476 	struct bridge_softc *sc = NULL;
2477 
2478 	if (bif)
2479 		sc = bif->bif_sc;
2480 
2481 	/*
2482 	 * The packet didnt originate from a member interface. This should only
2483 	 * ever happen if a member interface is removed while packets are
2484 	 * queued for it.
2485 	 */
2486 	if (sc == NULL) {
2487 		m_freem(m);
2488 		return;
2489 	}
2490 
2491 	if (PFIL_HOOKED_OUT_46) {
2492 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
2493 			return;
2494 		if (m == NULL)
2495 			return;
2496 	}
2497 
2498 	bridge_enqueue(sc, ifp, m, NULL);
2499 }
2500 
2501 /*
2502  * bridge_output:
2503  *
2504  *	Send output from a bridge member interface.  This
2505  *	performs the bridging function for locally originated
2506  *	packets.
2507  *
2508  *	The mbuf has the Ethernet header already attached.  We must
2509  *	enqueue or free the mbuf before returning.
2510  */
2511 static int
2512 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2513     struct rtentry *rt)
2514 {
2515 	struct ether_header *eh;
2516 	struct bridge_iflist *sbif;
2517 	struct ifnet *bifp, *dst_if;
2518 	struct bridge_softc *sc;
2519 	ether_vlanid_t vlan;
2520 
2521 	NET_EPOCH_ASSERT();
2522 
2523 	if (m->m_len < ETHER_HDR_LEN) {
2524 		m = m_pullup(m, ETHER_HDR_LEN);
2525 		if (m == NULL)
2526 			return (0);
2527 	}
2528 
2529 	sbif = ifp->if_bridge;
2530 	sc = sbif->bif_sc;
2531 	bifp = sc->sc_ifp;
2532 
2533 	eh = mtod(m, struct ether_header *);
2534 	vlan = VLANTAGOF(m);
2535 
2536 	/*
2537 	 * If bridge is down, but the original output interface is up,
2538 	 * go ahead and send out that interface.  Otherwise, the packet
2539 	 * is dropped below.
2540 	 */
2541 	if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2542 		dst_if = ifp;
2543 		goto sendunicast;
2544 	}
2545 
2546 	/*
2547 	 * If the packet is a multicast, or we don't know a better way to
2548 	 * get there, send to all interfaces.
2549 	 */
2550 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2551 		dst_if = NULL;
2552 	else
2553 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2554 	/* Tap any traffic not passing back out the originating interface */
2555 	if (dst_if != ifp)
2556 		ETHER_BPF_MTAP(bifp, m);
2557 	if (dst_if == NULL) {
2558 		struct bridge_iflist *bif;
2559 		struct mbuf *mc;
2560 		int used = 0;
2561 
2562 		bridge_span(sc, m);
2563 
2564 		CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2565 			dst_if = bif->bif_ifp;
2566 
2567 			if (dst_if->if_type == IFT_GIF)
2568 				continue;
2569 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2570 				continue;
2571 
2572 			/*
2573 			 * If this is not the original output interface,
2574 			 * and the interface is participating in spanning
2575 			 * tree, make sure the port is in a state that
2576 			 * allows forwarding.
2577 			 */
2578 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2579 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2580 				continue;
2581 
2582 			if (CK_LIST_NEXT(bif, bif_next) == NULL) {
2583 				used = 1;
2584 				mc = m;
2585 			} else {
2586 				mc = m_dup(m, M_NOWAIT);
2587 				if (mc == NULL) {
2588 					if_inc_counter(bifp, IFCOUNTER_OERRORS, 1);
2589 					continue;
2590 				}
2591 			}
2592 
2593 			bridge_enqueue(sc, dst_if, mc, bif);
2594 		}
2595 		if (used == 0)
2596 			m_freem(m);
2597 		return (0);
2598 	}
2599 
2600 sendunicast:
2601 	/*
2602 	 * XXX Spanning tree consideration here?
2603 	 */
2604 
2605 	bridge_span(sc, m);
2606 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2607 		m_freem(m);
2608 		return (0);
2609 	}
2610 
2611 	bridge_enqueue(sc, dst_if, m, NULL);
2612 	return (0);
2613 }
2614 
2615 /*
2616  * bridge_transmit:
2617  *
2618  *	Do output on a bridge.
2619  *
2620  */
2621 static int
2622 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2623 {
2624 	struct bridge_softc *sc;
2625 	struct ether_header *eh;
2626 	struct ifnet *dst_if;
2627 	int error = 0;
2628 	ether_vlanid_t vlan;
2629 
2630 	sc = ifp->if_softc;
2631 
2632 	ETHER_BPF_MTAP(ifp, m);
2633 
2634 	eh = mtod(m, struct ether_header *);
2635 	vlan = VLANTAGOF(m);
2636 
2637 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2638 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan)) != NULL) {
2639 		error = bridge_enqueue(sc, dst_if, m, NULL);
2640 	} else
2641 		bridge_broadcast(sc, ifp, m, 0);
2642 
2643 	return (error);
2644 }
2645 
2646 #ifdef ALTQ
2647 static void
2648 bridge_altq_start(if_t ifp)
2649 {
2650 	struct ifaltq *ifq = &ifp->if_snd;
2651 	struct mbuf *m;
2652 
2653 	IFQ_LOCK(ifq);
2654 	IFQ_DEQUEUE_NOLOCK(ifq, m);
2655 	while (m != NULL) {
2656 		bridge_transmit(ifp, m);
2657 		IFQ_DEQUEUE_NOLOCK(ifq, m);
2658 	}
2659 	IFQ_UNLOCK(ifq);
2660 }
2661 
2662 static int
2663 bridge_altq_transmit(if_t ifp, struct mbuf *m)
2664 {
2665 	int err;
2666 
2667 	if (ALTQ_IS_ENABLED(&ifp->if_snd)) {
2668 		IFQ_ENQUEUE(&ifp->if_snd, m, err);
2669 		if (err == 0)
2670 			bridge_altq_start(ifp);
2671 	} else
2672 		err = bridge_transmit(ifp, m);
2673 
2674 	return (err);
2675 }
2676 #endif	/* ALTQ */
2677 
2678 /*
2679  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2680  */
2681 static void
2682 bridge_qflush(struct ifnet *ifp __unused)
2683 {
2684 }
2685 
2686 /*
2687  * bridge_forward:
2688  *
2689  *	The forwarding function of the bridge.
2690  *
2691  *	NOTE: Releases the lock on return.
2692  */
2693 static void
2694 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2695     struct mbuf *m)
2696 {
2697 	struct bridge_iflist *dbif;
2698 	struct ifnet *src_if, *dst_if, *ifp;
2699 	struct ether_header *eh;
2700 	uint8_t *dst;
2701 	int error;
2702 	ether_vlanid_t vlan;
2703 
2704 	NET_EPOCH_ASSERT();
2705 
2706 	src_if = m->m_pkthdr.rcvif;
2707 	ifp = sc->sc_ifp;
2708 	vlan = VLANTAGOF(m);
2709 
2710 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2711 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2712 
2713 	if ((sbif->bif_flags & IFBIF_STP) &&
2714 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2715 		goto drop;
2716 
2717 	eh = mtod(m, struct ether_header *);
2718 	dst = eh->ether_dhost;
2719 
2720 	/* If the interface is learning, record the address. */
2721 	if (sbif->bif_flags & IFBIF_LEARNING) {
2722 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2723 		    sbif, 0, IFBAF_DYNAMIC);
2724 		/*
2725 		 * If the interface has addresses limits then deny any source
2726 		 * that is not in the cache.
2727 		 */
2728 		if (error && sbif->bif_addrmax)
2729 			goto drop;
2730 	}
2731 
2732 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2733 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2734 		goto drop;
2735 
2736 #ifdef DEV_NETMAP
2737 	/*
2738 	 * Hand the packet to netmap only if it wasn't injected by netmap
2739 	 * itself.
2740 	 */
2741 	if ((m->m_flags & M_BRIDGE_INJECT) == 0 &&
2742 	    (if_getcapenable(ifp) & IFCAP_NETMAP) != 0) {
2743 		ifp->if_input(ifp, m);
2744 		return;
2745 	}
2746 	m->m_flags &= ~M_BRIDGE_INJECT;
2747 #endif
2748 
2749 	/*
2750 	 * At this point, the port either doesn't participate
2751 	 * in spanning tree or it is in the forwarding state.
2752 	 */
2753 
2754 	/*
2755 	 * If the packet is unicast, destined for someone on
2756 	 * "this" side of the bridge, drop it.
2757 	 */
2758 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2759 		dst_if = bridge_rtlookup(sc, dst, vlan);
2760 		if (src_if == dst_if)
2761 			goto drop;
2762 	} else {
2763 		/*
2764 		 * Check if its a reserved multicast address, any address
2765 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2766 		 * bridge.
2767 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2768 		 */
2769 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2770 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2771 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2772 			goto drop;
2773 
2774 		/* ...forward it to all interfaces. */
2775 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2776 		dst_if = NULL;
2777 	}
2778 
2779 	/*
2780 	 * If we have a destination interface which is a member of our bridge,
2781 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2782 	 * For broadcast or multicast packets, don't bother because it will
2783 	 * be reinjected into ether_input. We do this before we pass the packets
2784 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2785 	 * drop the packet, or possibly modify it, making it difficult to debug
2786 	 * firewall issues on the bridge.
2787 	 */
2788 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2789 		ETHER_BPF_MTAP(ifp, m);
2790 
2791 	/* run the packet filter */
2792 	if (PFIL_HOOKED_IN_46) {
2793 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2794 			return;
2795 		if (m == NULL)
2796 			return;
2797 	}
2798 
2799 	if (dst_if == NULL) {
2800 		bridge_broadcast(sc, src_if, m, 1);
2801 		return;
2802 	}
2803 
2804 	/*
2805 	 * At this point, we're dealing with a unicast frame
2806 	 * going to a different interface.
2807 	 */
2808 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2809 		goto drop;
2810 
2811 	dbif = bridge_lookup_member_if(sc, dst_if);
2812 	if (dbif == NULL)
2813 		/* Not a member of the bridge (anymore?) */
2814 		goto drop;
2815 
2816 	/* Private segments can not talk to each other */
2817 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2818 		goto drop;
2819 
2820 	/* Do VLAN filtering. */
2821 	if (!bridge_vfilter_out(dbif, m))
2822 		goto drop;
2823 
2824 	if ((dbif->bif_flags & IFBIF_STP) &&
2825 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2826 		goto drop;
2827 
2828 	if (PFIL_HOOKED_OUT_46) {
2829 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2830 			return;
2831 		if (m == NULL)
2832 			return;
2833 	}
2834 
2835 	bridge_enqueue(sc, dst_if, m, dbif);
2836 	return;
2837 
2838 drop:
2839 	m_freem(m);
2840 }
2841 
2842 /*
2843  * bridge_input:
2844  *
2845  *	Receive input from a member interface.  Queue the packet for
2846  *	bridging if it is not for us.
2847  */
2848 static struct mbuf *
2849 bridge_input(struct ifnet *ifp, struct mbuf *m)
2850 {
2851 	struct bridge_softc *sc = NULL;
2852 	struct bridge_iflist *bif, *bif2;
2853 	struct ifnet *bifp;
2854 	struct ether_header *eh;
2855 	struct mbuf *mc, *mc2;
2856 	ether_vlanid_t vlan;
2857 	int error;
2858 
2859 	NET_EPOCH_ASSERT();
2860 
2861 	/* We need the Ethernet header later, so make sure we have it now. */
2862 	if (m->m_len < ETHER_HDR_LEN) {
2863 		m = m_pullup(m, ETHER_HDR_LEN);
2864 		if (m == NULL) {
2865 			if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1);
2866 			m_freem(m);
2867 			return (NULL);
2868 		}
2869 	}
2870 
2871 	eh = mtod(m, struct ether_header *);
2872 	vlan = VLANTAGOF(m);
2873 
2874 	/*
2875 	 * If this frame has a VLAN tag and the receiving interface has a
2876 	 * vlan(4) trunk, then it is is destined for vlan(4), not for us.
2877 	 * This means if vlan(4) and bridge(4) are configured on the same
2878 	 * interface, vlan(4) is preferred, which is what users typically
2879 	 * expect.
2880 	 */
2881 	if (vlan != DOT1Q_VID_NULL && ifp->if_vlantrunk != NULL)
2882 		return (m);
2883 
2884 	bif = ifp->if_bridge;
2885 	if (bif)
2886 		sc = bif->bif_sc;
2887 
2888 	if (sc == NULL) {
2889 		/*
2890 		 * This packet originated from the bridge itself, so it must
2891 		 * have been transmitted by netmap.  Derive the "source"
2892 		 * interface from the source address and drop the packet if the
2893 		 * source address isn't known.
2894 		 */
2895 		KASSERT((m->m_flags & M_BRIDGE_INJECT) != 0,
2896 		    ("%s: ifnet %p missing a bridge softc", __func__, ifp));
2897 		sc = if_getsoftc(ifp);
2898 		ifp = bridge_rtlookup(sc, eh->ether_shost, vlan);
2899 		if (ifp == NULL) {
2900 			if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1);
2901 			m_freem(m);
2902 			return (NULL);
2903 		}
2904 		m->m_pkthdr.rcvif = ifp;
2905 	}
2906 	bifp = sc->sc_ifp;
2907 	if ((bifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2908 		return (m);
2909 
2910 	/*
2911 	 * Implement support for bridge monitoring. If this flag has been
2912 	 * set on this interface, discard the packet once we push it through
2913 	 * the bpf(4) machinery, but before we do, increment the byte and
2914 	 * packet counters associated with this interface.
2915 	 */
2916 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2917 		m->m_pkthdr.rcvif  = bifp;
2918 		ETHER_BPF_MTAP(bifp, m);
2919 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2920 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2921 		m_freem(m);
2922 		return (NULL);
2923 	}
2924 
2925 	/* Do VLAN filtering. */
2926 	if (!bridge_vfilter_in(bif, m)) {
2927 		if_inc_counter(sc->sc_ifp, IFCOUNTER_IERRORS, 1);
2928 		m_freem(m);
2929 		return (NULL);
2930 	}
2931 	/* bridge_vfilter_in() may add a tag */
2932 	vlan = VLANTAGOF(m);
2933 
2934 	bridge_span(sc, m);
2935 
2936 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2937 		/* Tap off 802.1D packets; they do not get forwarded. */
2938 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2939 		    ETHER_ADDR_LEN) == 0) {
2940 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2941 			return (NULL);
2942 		}
2943 
2944 		if ((bif->bif_flags & IFBIF_STP) &&
2945 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2946 			return (m);
2947 		}
2948 
2949 		/*
2950 		 * Make a deep copy of the packet and enqueue the copy
2951 		 * for bridge processing; return the original packet for
2952 		 * local processing.
2953 		 */
2954 		mc = m_dup(m, M_NOWAIT);
2955 		if (mc == NULL) {
2956 			return (m);
2957 		}
2958 
2959 		/* Perform the bridge forwarding function with the copy. */
2960 		bridge_forward(sc, bif, mc);
2961 
2962 #ifdef DEV_NETMAP
2963 		/*
2964 		 * If netmap is enabled and has not already seen this packet,
2965 		 * then it will be consumed by bridge_forward().
2966 		 */
2967 		if ((if_getcapenable(bifp) & IFCAP_NETMAP) != 0 &&
2968 		    (m->m_flags & M_BRIDGE_INJECT) == 0) {
2969 			m_freem(m);
2970 			return (NULL);
2971 		}
2972 #endif
2973 
2974 		/*
2975 		 * Reinject the mbuf as arriving on the bridge so we have a
2976 		 * chance at claiming multicast packets. We can not loop back
2977 		 * here from ether_input as a bridge is never a member of a
2978 		 * bridge.
2979 		 */
2980 		KASSERT(bifp->if_bridge == NULL,
2981 		    ("loop created in bridge_input"));
2982 		mc2 = m_dup(m, M_NOWAIT);
2983 		if (mc2 != NULL) {
2984 			/* Keep the layer3 header aligned */
2985 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2986 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2987 		}
2988 		if (mc2 != NULL) {
2989 			mc2->m_pkthdr.rcvif = bifp;
2990 			mc2->m_flags &= ~M_BRIDGE_INJECT;
2991 			sc->sc_if_input(bifp, mc2);
2992 		}
2993 
2994 		/* Return the original packet for local processing. */
2995 		return (m);
2996 	}
2997 
2998 	if ((bif->bif_flags & IFBIF_STP) &&
2999 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
3000 		return (m);
3001 	}
3002 
3003 #if defined(INET) || defined(INET6)
3004 #define	CARP_CHECK_WE_ARE_DST(iface) \
3005 	((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_dhost))
3006 #define	CARP_CHECK_WE_ARE_SRC(iface) \
3007 	((iface)->if_carp && (*carp_forus_p)((iface), eh->ether_shost))
3008 #else
3009 #define	CARP_CHECK_WE_ARE_DST(iface)	false
3010 #define	CARP_CHECK_WE_ARE_SRC(iface)	false
3011 #endif
3012 
3013 #ifdef DEV_NETMAP
3014 #define	GRAB_FOR_NETMAP(ifp, m) do {					\
3015 	if ((if_getcapenable(ifp) & IFCAP_NETMAP) != 0 &&		\
3016 	    ((m)->m_flags & M_BRIDGE_INJECT) == 0) {			\
3017 		(ifp)->if_input(ifp, m);				\
3018 		return (NULL);						\
3019 	}								\
3020 } while (0)
3021 #else
3022 #define	GRAB_FOR_NETMAP(ifp, m)
3023 #endif
3024 
3025 #define GRAB_OUR_PACKETS(iface)						\
3026 	if ((iface)->if_type == IFT_GIF)				\
3027 		continue;						\
3028 	/* It is destined for us. */					\
3029 	if (memcmp(IF_LLADDR(iface), eh->ether_dhost, ETHER_ADDR_LEN) == 0 || \
3030 	    CARP_CHECK_WE_ARE_DST(iface)) {				\
3031 		if (bif->bif_flags & IFBIF_LEARNING) {			\
3032 			error = bridge_rtupdate(sc, eh->ether_shost,	\
3033 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
3034 			if (error && bif->bif_addrmax) {		\
3035 				m_freem(m);				\
3036 				return (NULL);				\
3037 			}						\
3038 		}							\
3039 		m->m_pkthdr.rcvif = iface;				\
3040 		if ((iface) == ifp) {					\
3041 			/* Skip bridge processing... src == dest */	\
3042 			return (m);					\
3043 		}							\
3044 		/* It's passing over or to the bridge, locally. */	\
3045 		ETHER_BPF_MTAP(bifp, m);				\
3046 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);		\
3047 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);\
3048 		/* Hand the packet over to netmap if necessary. */	\
3049 		GRAB_FOR_NETMAP(bifp, m);				\
3050 		/* Filter on the physical interface. */			\
3051 		if (V_pfil_local_phys && PFIL_HOOKED_IN_46) {		\
3052 			if (bridge_pfil(&m, NULL, ifp,			\
3053 			    PFIL_IN) != 0 || m == NULL) {		\
3054 				return (NULL);				\
3055 			}						\
3056 		}							\
3057 		if ((iface) != bifp)					\
3058 			ETHER_BPF_MTAP(iface, m);			\
3059 		/* Pass tagged packets to if_vlan, if it's loaded */	\
3060 		if (VLANTAGOF(m) != 0) {				\
3061 			if (bifp->if_vlantrunk == NULL) {		\
3062 				m_freem(m);				\
3063 				return (NULL);				\
3064 			}						\
3065 			(*vlan_input_p)(bifp, m);			\
3066 			return (NULL);					\
3067 		}							\
3068 		return (m);						\
3069 	}								\
3070 									\
3071 	/* We just received a packet that we sent out. */		\
3072 	if (memcmp(IF_LLADDR(iface), eh->ether_shost, ETHER_ADDR_LEN) == 0 || \
3073 	    CARP_CHECK_WE_ARE_SRC(iface)) {				\
3074 		m_freem(m);						\
3075 		return (NULL);						\
3076 	}
3077 
3078 	/*
3079 	 * Unicast.  Make sure it's not for the bridge.
3080 	 */
3081 	do { GRAB_OUR_PACKETS(bifp) } while (0);
3082 
3083 	/*
3084 	 * If member_ifaddrs is enabled, see if the packet is destined for
3085 	 * one of the members' addresses.
3086 	 */
3087 	if (V_member_ifaddrs) {
3088 		/* Check the interface the packet arrived on. */
3089 		do { GRAB_OUR_PACKETS(ifp) } while (0);
3090 
3091 		CK_LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
3092 			GRAB_OUR_PACKETS(bif2->bif_ifp)
3093 		}
3094 	}
3095 
3096 #undef CARP_CHECK_WE_ARE_DST
3097 #undef CARP_CHECK_WE_ARE_SRC
3098 #undef GRAB_FOR_NETMAP
3099 #undef GRAB_OUR_PACKETS
3100 
3101 	/* Perform the bridge forwarding function. */
3102 	bridge_forward(sc, bif, m);
3103 
3104 	return (NULL);
3105 }
3106 
3107 /*
3108  * Inject a packet back into the host ethernet stack.  This will generally only
3109  * be used by netmap when an application writes to the host TX ring.  The
3110  * M_BRIDGE_INJECT flag ensures that the packet is re-routed to the bridge
3111  * interface after ethernet processing.
3112  */
3113 static void
3114 bridge_inject(struct ifnet *ifp, struct mbuf *m)
3115 {
3116 	struct bridge_softc *sc;
3117 
3118 	if (ifp->if_type == IFT_L2VLAN) {
3119 		/*
3120 		 * vlan(4) gives us the vlan ifnet, so we need to get the
3121 		 * bridge softc to get a pointer to ether_input to send the
3122 		 * packet to.
3123 		 */
3124 		struct ifnet *bifp = NULL;
3125 
3126 		if (vlan_trunkdev_p == NULL) {
3127 			m_freem(m);
3128 			return;
3129 		}
3130 
3131 		bifp = vlan_trunkdev_p(ifp);
3132 		if (bifp == NULL) {
3133 			m_freem(m);
3134 			return;
3135 		}
3136 
3137 		sc = if_getsoftc(bifp);
3138 		sc->sc_if_input(ifp, m);
3139 		return;
3140 	}
3141 
3142 	KASSERT((if_getcapenable(ifp) & IFCAP_NETMAP) != 0,
3143 	    ("%s: iface %s is not running in netmap mode",
3144 	    __func__, if_name(ifp)));
3145 	KASSERT((m->m_flags & M_BRIDGE_INJECT) == 0,
3146 	    ("%s: mbuf %p has M_BRIDGE_INJECT set", __func__, m));
3147 
3148 	m->m_flags |= M_BRIDGE_INJECT;
3149 	sc = if_getsoftc(ifp);
3150 	sc->sc_if_input(ifp, m);
3151 }
3152 
3153 /*
3154  * bridge_broadcast:
3155  *
3156  *	Send a frame to all interfaces that are members of
3157  *	the bridge, except for the one on which the packet
3158  *	arrived.
3159  *
3160  *	NOTE: Releases the lock on return.
3161  */
3162 static void
3163 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
3164     struct mbuf *m, int runfilt)
3165 {
3166 	struct bridge_iflist *dbif, *sbif;
3167 	struct mbuf *mc;
3168 	struct ifnet *dst_if;
3169 	int used = 0, i;
3170 
3171 	NET_EPOCH_ASSERT();
3172 
3173 	sbif = bridge_lookup_member_if(sc, src_if);
3174 
3175 	/* Filter on the bridge interface before broadcasting */
3176 	if (runfilt && PFIL_HOOKED_OUT_46) {
3177 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
3178 			return;
3179 		if (m == NULL)
3180 			return;
3181 	}
3182 
3183 	CK_LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
3184 		dst_if = dbif->bif_ifp;
3185 		if (dst_if == src_if)
3186 			continue;
3187 
3188 		/* Private segments can not talk to each other */
3189 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
3190 			continue;
3191 
3192 		/* Do VLAN filtering. */
3193 		if (!bridge_vfilter_out(dbif, m))
3194 			continue;
3195 
3196 		if ((dbif->bif_flags & IFBIF_STP) &&
3197 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
3198 			continue;
3199 
3200 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
3201 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
3202 			continue;
3203 
3204 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
3205 			continue;
3206 
3207 		if (CK_LIST_NEXT(dbif, bif_next) == NULL) {
3208 			mc = m;
3209 			used = 1;
3210 		} else {
3211 			mc = m_dup(m, M_NOWAIT);
3212 			if (mc == NULL) {
3213 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
3214 				continue;
3215 			}
3216 		}
3217 
3218 		/*
3219 		 * Filter on the output interface. Pass a NULL bridge interface
3220 		 * pointer so we do not redundantly filter on the bridge for
3221 		 * each interface we broadcast on.
3222 		 */
3223 		if (runfilt && PFIL_HOOKED_OUT_46) {
3224 			if (used == 0) {
3225 				/* Keep the layer3 header aligned */
3226 				i = min(mc->m_pkthdr.len, max_protohdr);
3227 				mc = m_copyup(mc, i, ETHER_ALIGN);
3228 				if (mc == NULL) {
3229 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
3230 					continue;
3231 				}
3232 			}
3233 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
3234 				continue;
3235 			if (mc == NULL)
3236 				continue;
3237 		}
3238 
3239 		bridge_enqueue(sc, dst_if, mc, dbif);
3240 	}
3241 	if (used == 0)
3242 		m_freem(m);
3243 }
3244 
3245 /*
3246  * bridge_span:
3247  *
3248  *	Duplicate a packet out one or more interfaces that are in span mode,
3249  *	the original mbuf is unmodified.
3250  */
3251 static void
3252 bridge_span(struct bridge_softc *sc, struct mbuf *m)
3253 {
3254 	struct bridge_iflist *bif;
3255 	struct ifnet *dst_if;
3256 	struct mbuf *mc;
3257 
3258 	NET_EPOCH_ASSERT();
3259 
3260 	if (CK_LIST_EMPTY(&sc->sc_spanlist))
3261 		return;
3262 
3263 	CK_LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
3264 		dst_if = bif->bif_ifp;
3265 
3266 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
3267 			continue;
3268 
3269 		mc = m_dup(m, M_NOWAIT);
3270 		if (mc == NULL) {
3271 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
3272 			continue;
3273 		}
3274 
3275 		bridge_enqueue(sc, dst_if, mc, bif);
3276 	}
3277 }
3278 
3279 /*
3280  * Incoming VLAN filtering.  Given a frame and the member interface it was
3281  * received on, decide whether the port configuration allows it.
3282  */
3283 static bool
3284 bridge_vfilter_in(const struct bridge_iflist *sbif, struct mbuf *m)
3285 {
3286 	ether_vlanid_t vlan;
3287 
3288 	vlan = VLANTAGOF(m);
3289 	/* Make sure the vlan id is reasonable. */
3290 	if (vlan > DOT1Q_VID_MAX)
3291 		return (false);
3292 
3293 	/*
3294 	 * If VLAN filtering isn't enabled, pass everything, but add a tag
3295 	 * if the port has a pvid configured.
3296 	 */
3297 	if ((sbif->bif_sc->sc_flags & IFBRF_VLANFILTER) == 0) {
3298 		if (vlan == DOT1Q_VID_NULL &&
3299 		    sbif->bif_pvid != DOT1Q_VID_NULL) {
3300 			m->m_pkthdr.ether_vtag = sbif->bif_pvid;
3301 			m->m_flags |= M_VLANTAG;
3302 		}
3303 
3304 		return (true);
3305 	}
3306 
3307 	/* If Q-in-Q is disabled, check for stacked tags. */
3308 	if ((sbif->bif_flags & IFBIF_QINQ) == 0) {
3309 		struct ether_header *eh;
3310 		uint16_t proto;
3311 
3312 		eh = mtod(m, struct ether_header *);
3313 		proto = ntohs(eh->ether_type);
3314 
3315 		if (proto == ETHERTYPE_VLAN || proto == ETHERTYPE_QINQ)
3316 			return (false);
3317 	}
3318 
3319 	if (vlan == DOT1Q_VID_NULL) {
3320 		/*
3321 		 * The frame doesn't have a tag.  If the interface does not
3322 		 * have an untagged vlan configured, drop the frame.
3323 		 */
3324 		if (sbif->bif_pvid == DOT1Q_VID_NULL)
3325 			return (false);
3326 
3327 		/*
3328 		 * Otherwise, insert a new tag based on the interface's
3329 		 * untagged vlan id.
3330 		 */
3331 		m->m_pkthdr.ether_vtag = sbif->bif_pvid;
3332 		m->m_flags |= M_VLANTAG;
3333 	} else {
3334 		/*
3335 		 * The frame has a tag, so check it matches the interface's
3336 		 * vlan access list.  We explicitly do not accept tagged
3337 		 * frames for the untagged vlan id here (unless it's also
3338 		 * in the access list).
3339 		 */
3340 		if (!BRVLAN_TEST(&sbif->bif_vlan_set, vlan))
3341 			return (false);
3342 	}
3343 
3344 	/* Accept the frame. */
3345 	return (true);
3346 }
3347 
3348 /*
3349  * Outgoing VLAN filtering.  Given a frame, its vlan, and the member interface
3350  * we intend to send it to, decide whether the port configuration allows it to
3351  * be sent.
3352  */
3353 static bool
3354 bridge_vfilter_out(const struct bridge_iflist *dbif, const struct mbuf *m)
3355 {
3356 	struct ether_header *eh;
3357 	ether_vlanid_t vlan;
3358 
3359 	NET_EPOCH_ASSERT();
3360 
3361 	/* If VLAN filtering isn't enabled, pass everything. */
3362 	if ((dbif->bif_sc->sc_flags & IFBRF_VLANFILTER) == 0)
3363 		return (true);
3364 
3365 	vlan = VLANTAGOF(m);
3366 
3367 	/*
3368 	 * Always allow untagged 802.1D STP frames, even if they would
3369 	 * otherwise be dropped.  This is required for STP to work on
3370 	 * a filtering bridge.
3371 	 *
3372 	 * Tagged STP (Cisco PVST+) is a non-standard extension, so
3373 	 * handle those frames via the normal filtering path.
3374 	 */
3375 	eh = mtod(m, struct ether_header *);
3376 	if (vlan == DOT1Q_VID_NULL &&
3377 	    memcmp(eh->ether_dhost, bstp_etheraddr, ETHER_ADDR_LEN) == 0)
3378 		return (true);
3379 
3380 	/*
3381 	 * If the frame wasn't assigned to a vlan at ingress, drop it.
3382 	 * We can't forward these frames to filtering ports because we
3383 	 * don't know what VLAN they're supposed to be in.
3384 	 */
3385 	if (vlan == DOT1Q_VID_NULL)
3386 		return (false);
3387 
3388 	/*
3389 	 * If the frame's vlan matches the interfaces's untagged vlan,
3390 	 * allow it.
3391 	 */
3392 	if (vlan == dbif->bif_pvid)
3393 		return (true);
3394 
3395 	/*
3396 	 * If the frame's vlan is on the interface's tagged access list,
3397 	 * allow it.
3398 	 */
3399 	if (BRVLAN_TEST(&dbif->bif_vlan_set, vlan))
3400 		return (true);
3401 
3402 	/* The frame was not permitted, so drop it. */
3403 	return (false);
3404 }
3405 
3406 /*
3407  * bridge_rtupdate:
3408  *
3409  *	Add a bridge routing entry.
3410  */
3411 static int
3412 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst,
3413 		ether_vlanid_t vlan, struct bridge_iflist *bif,
3414 		int setflags, uint8_t flags)
3415 {
3416 	struct bridge_rtnode *brt;
3417 	struct bridge_iflist *obif;
3418 	int error;
3419 
3420 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
3421 
3422 	/* Check the source address is valid and not multicast. */
3423 	if (ETHER_IS_MULTICAST(dst))
3424 		return (EXTERROR(EINVAL, "Multicast address not permitted"));
3425 	if (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
3426 	    dst[3] == 0 && dst[4] == 0 && dst[5] == 0)
3427 		return (EXTERROR(EINVAL, "Zero address not permitted"));
3428 
3429 	/*
3430 	 * A route for this destination might already exist.  If so,
3431 	 * update it, otherwise create a new one.
3432 	 */
3433 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
3434 		BRIDGE_RT_LOCK(sc);
3435 
3436 		/* Check again, now that we have the lock. There could have
3437 		 * been a race and we only want to insert this once. */
3438 		if (bridge_rtnode_lookup(sc, dst, vlan) != NULL) {
3439 			BRIDGE_RT_UNLOCK(sc);
3440 			return (0);
3441 		}
3442 
3443 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
3444 			sc->sc_brtexceeded++;
3445 			BRIDGE_RT_UNLOCK(sc);
3446 			return (EXTERROR(ENOSPC, "Address table is full"));
3447 		}
3448 		/* Check per interface address limits (if enabled) */
3449 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
3450 			bif->bif_addrexceeded++;
3451 			BRIDGE_RT_UNLOCK(sc);
3452 			return (EXTERROR(ENOSPC,
3453 			    "Interface address limit exceeded"));
3454 		}
3455 
3456 		/*
3457 		 * Allocate a new bridge forwarding node, and
3458 		 * initialize the expiration time and Ethernet
3459 		 * address.
3460 		 */
3461 		brt = uma_zalloc(V_bridge_rtnode_zone, M_NOWAIT | M_ZERO);
3462 		if (brt == NULL) {
3463 			BRIDGE_RT_UNLOCK(sc);
3464 			return (EXTERROR(ENOMEM,
3465 			    "Cannot allocate address node"));
3466 		}
3467 		brt->brt_vnet = curvnet;
3468 
3469 		if (bif->bif_flags & IFBIF_STICKY)
3470 			brt->brt_flags = IFBAF_STICKY;
3471 		else
3472 			brt->brt_flags = IFBAF_DYNAMIC;
3473 
3474 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
3475 		brt->brt_vlan = vlan;
3476 
3477 		brt->brt_dst = bif;
3478 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
3479 			uma_zfree(V_bridge_rtnode_zone, brt);
3480 			BRIDGE_RT_UNLOCK(sc);
3481 			return (error);
3482 		}
3483 		bif->bif_addrcnt++;
3484 
3485 		BRIDGE_RT_UNLOCK(sc);
3486 	}
3487 
3488 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
3489 	    (obif = brt->brt_dst) != bif) {
3490 		MPASS(obif != NULL);
3491 
3492 		BRIDGE_RT_LOCK(sc);
3493 		brt->brt_dst->bif_addrcnt--;
3494 		brt->brt_dst = bif;
3495 		brt->brt_dst->bif_addrcnt++;
3496 		BRIDGE_RT_UNLOCK(sc);
3497 
3498 		if (V_log_mac_flap &&
3499 		    ppsratecheck(&V_log_last, &V_log_count, V_log_interval)) {
3500 			log(LOG_NOTICE,
3501 			    "%s: mac address %6D vlan %d moved from %s to %s\n",
3502 			    sc->sc_ifp->if_xname,
3503 			    &brt->brt_addr[0], ":",
3504 			    brt->brt_vlan,
3505 			    obif->bif_ifp->if_xname,
3506 			    bif->bif_ifp->if_xname);
3507 		}
3508 	}
3509 
3510 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3511 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
3512 	if (setflags)
3513 		brt->brt_flags = flags;
3514 
3515 	return (0);
3516 }
3517 
3518 /*
3519  * bridge_rtlookup:
3520  *
3521  *	Lookup the destination interface for an address.
3522  */
3523 static struct ifnet *
3524 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr,
3525 		ether_vlanid_t vlan)
3526 {
3527 	struct bridge_rtnode *brt;
3528 
3529 	NET_EPOCH_ASSERT();
3530 
3531 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
3532 		return (NULL);
3533 
3534 	return (brt->brt_ifp);
3535 }
3536 
3537 /*
3538  * bridge_rttrim:
3539  *
3540  *	Trim the routine table so that we have a number
3541  *	of routing entries less than or equal to the
3542  *	maximum number.
3543  */
3544 static void
3545 bridge_rttrim(struct bridge_softc *sc)
3546 {
3547 	struct bridge_rtnode *brt, *nbrt;
3548 
3549 	NET_EPOCH_ASSERT();
3550 	BRIDGE_RT_LOCK_ASSERT(sc);
3551 
3552 	/* Make sure we actually need to do this. */
3553 	if (sc->sc_brtcnt <= sc->sc_brtmax)
3554 		return;
3555 
3556 	/* Force an aging cycle; this might trim enough addresses. */
3557 	bridge_rtage(sc);
3558 	if (sc->sc_brtcnt <= sc->sc_brtmax)
3559 		return;
3560 
3561 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3562 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
3563 			bridge_rtnode_destroy(sc, brt);
3564 			if (sc->sc_brtcnt <= sc->sc_brtmax)
3565 				return;
3566 		}
3567 	}
3568 }
3569 
3570 /*
3571  * bridge_timer:
3572  *
3573  *	Aging timer for the bridge.
3574  */
3575 static void
3576 bridge_timer(void *arg)
3577 {
3578 	struct bridge_softc *sc = arg;
3579 
3580 	BRIDGE_RT_LOCK_ASSERT(sc);
3581 
3582 	/* Destruction of rtnodes requires a proper vnet context */
3583 	CURVNET_SET(sc->sc_ifp->if_vnet);
3584 	bridge_rtage(sc);
3585 
3586 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
3587 		callout_reset(&sc->sc_brcallout,
3588 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
3589 	CURVNET_RESTORE();
3590 }
3591 
3592 /*
3593  * bridge_rtage:
3594  *
3595  *	Perform an aging cycle.
3596  */
3597 static void
3598 bridge_rtage(struct bridge_softc *sc)
3599 {
3600 	struct bridge_rtnode *brt, *nbrt;
3601 
3602 	BRIDGE_RT_LOCK_ASSERT(sc);
3603 
3604 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3605 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
3606 			if (time_uptime >= brt->brt_expire)
3607 				bridge_rtnode_destroy(sc, brt);
3608 		}
3609 	}
3610 }
3611 
3612 /*
3613  * bridge_rtflush:
3614  *
3615  *	Remove all dynamic addresses from the bridge.
3616  */
3617 static void
3618 bridge_rtflush(struct bridge_softc *sc, int full)
3619 {
3620 	struct bridge_rtnode *brt, *nbrt;
3621 
3622 	BRIDGE_RT_LOCK_ASSERT(sc);
3623 
3624 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3625 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3626 			bridge_rtnode_destroy(sc, brt);
3627 	}
3628 }
3629 
3630 /*
3631  * bridge_rtdaddr:
3632  *
3633  *	Remove an address from the table.
3634  */
3635 static int
3636 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr,
3637 	       ether_vlanid_t vlan)
3638 {
3639 	struct bridge_rtnode *brt;
3640 	int found = 0;
3641 
3642 	BRIDGE_RT_LOCK(sc);
3643 
3644 	/*
3645 	 * If vlan is DOT1Q_VID_RSVD_IMPL then we want to delete for all vlans
3646 	 * so the lookup may return more than one.
3647 	 */
3648 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
3649 		bridge_rtnode_destroy(sc, brt);
3650 		found = 1;
3651 	}
3652 
3653 	BRIDGE_RT_UNLOCK(sc);
3654 
3655 	return (found ? 0 : ENOENT);
3656 }
3657 
3658 /*
3659  * bridge_rtdelete:
3660  *
3661  *	Delete routes to a speicifc member interface.
3662  */
3663 static void
3664 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
3665 {
3666 	struct bridge_rtnode *brt, *nbrt;
3667 
3668 	BRIDGE_RT_LOCK_ASSERT(sc);
3669 
3670 	CK_LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
3671 		if (brt->brt_ifp == ifp && (full ||
3672 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
3673 			bridge_rtnode_destroy(sc, brt);
3674 	}
3675 }
3676 
3677 /*
3678  * bridge_rtable_init:
3679  *
3680  *	Initialize the route table for this bridge.
3681  */
3682 static void
3683 bridge_rtable_init(struct bridge_softc *sc)
3684 {
3685 	int i;
3686 
3687 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
3688 	    M_DEVBUF, M_WAITOK);
3689 
3690 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
3691 		CK_LIST_INIT(&sc->sc_rthash[i]);
3692 
3693 	sc->sc_rthash_key = arc4random();
3694 	CK_LIST_INIT(&sc->sc_rtlist);
3695 }
3696 
3697 /*
3698  * bridge_rtable_fini:
3699  *
3700  *	Deconstruct the route table for this bridge.
3701  */
3702 static void
3703 bridge_rtable_fini(struct bridge_softc *sc)
3704 {
3705 
3706 	KASSERT(sc->sc_brtcnt == 0,
3707 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
3708 	free(sc->sc_rthash, M_DEVBUF);
3709 }
3710 
3711 /*
3712  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
3713  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
3714  */
3715 #define	mix(a, b, c)							\
3716 do {									\
3717 	a -= b; a -= c; a ^= (c >> 13);					\
3718 	b -= c; b -= a; b ^= (a << 8);					\
3719 	c -= a; c -= b; c ^= (b >> 13);					\
3720 	a -= b; a -= c; a ^= (c >> 12);					\
3721 	b -= c; b -= a; b ^= (a << 16);					\
3722 	c -= a; c -= b; c ^= (b >> 5);					\
3723 	a -= b; a -= c; a ^= (c >> 3);					\
3724 	b -= c; b -= a; b ^= (a << 10);					\
3725 	c -= a; c -= b; c ^= (b >> 15);					\
3726 } while (/*CONSTCOND*/0)
3727 
3728 static __inline uint32_t
3729 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
3730 {
3731 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
3732 
3733 	b += addr[5] << 8;
3734 	b += addr[4];
3735 	a += addr[3] << 24;
3736 	a += addr[2] << 16;
3737 	a += addr[1] << 8;
3738 	a += addr[0];
3739 
3740 	mix(a, b, c);
3741 
3742 	return (c & BRIDGE_RTHASH_MASK);
3743 }
3744 
3745 #undef mix
3746 
3747 static int
3748 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
3749 {
3750 	int i, d;
3751 
3752 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
3753 		d = ((int)a[i]) - ((int)b[i]);
3754 	}
3755 
3756 	return (d);
3757 }
3758 
3759 /*
3760  * bridge_rtnode_lookup:
3761  *
3762  *	Look up a bridge route node for the specified destination. Compare the
3763  *	vlan id or if zero then just return the first match.
3764  */
3765 static struct bridge_rtnode *
3766 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr,
3767 		     ether_vlanid_t vlan)
3768 {
3769 	struct bridge_rtnode *brt;
3770 	uint32_t hash;
3771 	int dir;
3772 
3773 	BRIDGE_RT_LOCK_OR_NET_EPOCH_ASSERT(sc);
3774 
3775 	hash = bridge_rthash(sc, addr);
3776 	CK_LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
3777 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
3778 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == DOT1Q_VID_RSVD_IMPL))
3779 			return (brt);
3780 		if (dir > 0)
3781 			return (NULL);
3782 	}
3783 
3784 	return (NULL);
3785 }
3786 
3787 /*
3788  * bridge_rtnode_insert:
3789  *
3790  *	Insert the specified bridge node into the route table.  We
3791  *	assume the entry is not already in the table.
3792  */
3793 static int
3794 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
3795 {
3796 	struct bridge_rtnode *lbrt;
3797 	uint32_t hash;
3798 	int dir;
3799 
3800 	BRIDGE_RT_LOCK_ASSERT(sc);
3801 
3802 	hash = bridge_rthash(sc, brt->brt_addr);
3803 
3804 	lbrt = CK_LIST_FIRST(&sc->sc_rthash[hash]);
3805 	if (lbrt == NULL) {
3806 		CK_LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
3807 		goto out;
3808 	}
3809 
3810 	do {
3811 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
3812 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
3813 			return (EXTERROR(EEXIST, "Address already exists"));
3814 		if (dir > 0) {
3815 			CK_LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
3816 			goto out;
3817 		}
3818 		if (CK_LIST_NEXT(lbrt, brt_hash) == NULL) {
3819 			CK_LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3820 			goto out;
3821 		}
3822 		lbrt = CK_LIST_NEXT(lbrt, brt_hash);
3823 	} while (lbrt != NULL);
3824 
3825 #ifdef DIAGNOSTIC
3826 	panic("bridge_rtnode_insert: impossible");
3827 #endif
3828 
3829 out:
3830 	CK_LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3831 	sc->sc_brtcnt++;
3832 
3833 	return (0);
3834 }
3835 
3836 static void
3837 bridge_rtnode_destroy_cb(struct epoch_context *ctx)
3838 {
3839 	struct bridge_rtnode *brt;
3840 
3841 	brt = __containerof(ctx, struct bridge_rtnode, brt_epoch_ctx);
3842 
3843 	CURVNET_SET(brt->brt_vnet);
3844 	uma_zfree(V_bridge_rtnode_zone, brt);
3845 	CURVNET_RESTORE();
3846 }
3847 
3848 /*
3849  * bridge_rtnode_destroy:
3850  *
3851  *	Destroy a bridge rtnode.
3852  */
3853 static void
3854 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3855 {
3856 	BRIDGE_RT_LOCK_ASSERT(sc);
3857 
3858 	CK_LIST_REMOVE(brt, brt_hash);
3859 
3860 	CK_LIST_REMOVE(brt, brt_list);
3861 	sc->sc_brtcnt--;
3862 	brt->brt_dst->bif_addrcnt--;
3863 
3864 	NET_EPOCH_CALL(bridge_rtnode_destroy_cb, &brt->brt_epoch_ctx);
3865 }
3866 
3867 /*
3868  * bridge_rtable_expire:
3869  *
3870  *	Set the expiry time for all routes on an interface.
3871  */
3872 static void
3873 bridge_rtable_expire(struct ifnet *ifp, int age)
3874 {
3875 	struct bridge_iflist *bif = NULL;
3876 	struct bridge_softc *sc = NULL;
3877 	struct bridge_rtnode *brt;
3878 
3879 	CURVNET_SET(ifp->if_vnet);
3880 
3881 	bif = ifp->if_bridge;
3882 	if (bif)
3883 		sc = bif->bif_sc;
3884 	MPASS(sc != NULL);
3885 	BRIDGE_RT_LOCK(sc);
3886 
3887 	/*
3888 	 * If the age is zero then flush, otherwise set all the expiry times to
3889 	 * age for the interface
3890 	 */
3891 	if (age == 0)
3892 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3893 	else {
3894 		CK_LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3895 			/* Cap the expiry time to 'age' */
3896 			if (brt->brt_ifp == ifp &&
3897 			    brt->brt_expire > time_uptime + age &&
3898 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3899 				brt->brt_expire = time_uptime + age;
3900 		}
3901 	}
3902 	BRIDGE_RT_UNLOCK(sc);
3903 	CURVNET_RESTORE();
3904 }
3905 
3906 /*
3907  * bridge_state_change:
3908  *
3909  *	Callback from the bridgestp code when a port changes states.
3910  */
3911 static void
3912 bridge_state_change(struct ifnet *ifp, int state)
3913 {
3914 	struct bridge_iflist *bif = ifp->if_bridge;
3915 	struct bridge_softc *sc = bif->bif_sc;
3916 	static const char *stpstates[] = {
3917 		"disabled",
3918 		"listening",
3919 		"learning",
3920 		"forwarding",
3921 		"blocking",
3922 		"discarding"
3923 	};
3924 
3925 	CURVNET_SET(ifp->if_vnet);
3926 	if (V_log_stp)
3927 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3928 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3929 	CURVNET_RESTORE();
3930 }
3931 
3932 /*
3933  * Send bridge packets through pfil if they are one of the types pfil can deal
3934  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3935  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3936  * that interface.
3937  */
3938 static int
3939 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3940 {
3941 	int snap, error, i;
3942 	struct ether_header *eh1, eh2;
3943 	struct llc llc1;
3944 	u_int16_t ether_type;
3945 	pfil_return_t rv;
3946 #ifdef INET
3947 	struct ip *ip = NULL;
3948 	int hlen = 0;
3949 #endif
3950 
3951 	snap = 0;
3952 	error = -1;	/* Default error if not error == 0 */
3953 
3954 #if 0
3955 	/* we may return with the IP fields swapped, ensure its not shared */
3956 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3957 #endif
3958 
3959 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3960 		return (0); /* filtering is disabled */
3961 
3962 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3963 	if ((*mp)->m_len < i) {
3964 	    *mp = m_pullup(*mp, i);
3965 	    if (*mp == NULL) {
3966 		printf("%s: m_pullup failed\n", __func__);
3967 		return (-1);
3968 	    }
3969 	}
3970 
3971 	eh1 = mtod(*mp, struct ether_header *);
3972 	ether_type = ntohs(eh1->ether_type);
3973 
3974 	/*
3975 	 * Check for SNAP/LLC.
3976 	 */
3977 	if (ether_type < ETHERMTU) {
3978 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3979 
3980 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3981 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3982 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3983 		    llc2->llc_control == LLC_UI) {
3984 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3985 			snap = 1;
3986 		}
3987 	}
3988 
3989 	/*
3990 	 * If we're trying to filter bridge traffic, only look at traffic for
3991 	 * protocols available in the kernel (IPv4 and/or IPv6) to avoid
3992 	 * passing traffic for an unsupported protocol to the filter.  This is
3993 	 * lame since if we really wanted, say, an AppleTalk filter, we are
3994 	 * hosed, but of course we don't have an AppleTalk filter to begin
3995 	 * with.  (Note that since pfil doesn't understand ARP it will pass
3996 	 * *ALL* ARP traffic.)
3997 	 */
3998 	switch (ether_type) {
3999 #ifdef INET
4000 		case ETHERTYPE_ARP:
4001 		case ETHERTYPE_REVARP:
4002 			if (V_pfil_ipfw_arp == 0)
4003 				return (0); /* Automatically pass */
4004 
4005 			/* FALLTHROUGH */
4006 		case ETHERTYPE_IP:
4007 #endif
4008 #ifdef INET6
4009 		case ETHERTYPE_IPV6:
4010 #endif /* INET6 */
4011 			break;
4012 
4013 		default:
4014 			/*
4015 			 * We get here if the packet isn't from a supported
4016 			 * protocol.  Check to see if the user wants to pass
4017 			 * non-IP packets, these will not be checked by pfil(9)
4018 			 * and passed unconditionally so the default is to
4019 			 * drop.
4020 			 */
4021 			if (V_pfil_onlyip)
4022 				goto bad;
4023 	}
4024 
4025 	/* Run the packet through pfil before stripping link headers */
4026 	if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
4027 	    dir == PFIL_OUT && ifp != NULL) {
4028 		switch (pfil_mbuf_out(V_link_pfil_head, mp, ifp, NULL)) {
4029 		case PFIL_DROPPED:
4030 			return (EACCES);
4031 		case PFIL_CONSUMED:
4032 			return (0);
4033 		}
4034 	}
4035 
4036 	/* Strip off the Ethernet header and keep a copy. */
4037 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
4038 	m_adj(*mp, ETHER_HDR_LEN);
4039 
4040 	/* Strip off snap header, if present */
4041 	if (snap) {
4042 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
4043 		m_adj(*mp, sizeof(struct llc));
4044 	}
4045 
4046 	/*
4047 	 * Check the IP header for alignment and errors
4048 	 */
4049 	if (dir == PFIL_IN) {
4050 		switch (ether_type) {
4051 #ifdef INET
4052 			case ETHERTYPE_IP:
4053 				error = bridge_ip_checkbasic(mp);
4054 				break;
4055 #endif
4056 #ifdef INET6
4057 			case ETHERTYPE_IPV6:
4058 				error = bridge_ip6_checkbasic(mp);
4059 				break;
4060 #endif /* INET6 */
4061 			default:
4062 				error = 0;
4063 		}
4064 		if (error)
4065 			goto bad;
4066 	}
4067 
4068 	error = 0;
4069 
4070 	/*
4071 	 * Run the packet through pfil
4072 	 */
4073 	rv = PFIL_PASS;
4074 	switch (ether_type) {
4075 #ifdef INET
4076 	case ETHERTYPE_IP:
4077 		/*
4078 		 * Run pfil on the member interface and the bridge, both can
4079 		 * be skipped by clearing pfil_member or pfil_bridge.
4080 		 *
4081 		 * Keep the order:
4082 		 *   in_if -> bridge_if -> out_if
4083 		 */
4084 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
4085 		    pfil_mbuf_out(V_inet_pfil_head, mp, bifp, NULL)) !=
4086 		    PFIL_PASS)
4087 			break;
4088 
4089 		if (V_pfil_member && ifp != NULL) {
4090 			rv = (dir == PFIL_OUT) ?
4091 			    pfil_mbuf_out(V_inet_pfil_head, mp, ifp, NULL) :
4092 			    pfil_mbuf_in(V_inet_pfil_head, mp, ifp, NULL);
4093 			if (rv != PFIL_PASS)
4094 				break;
4095 		}
4096 
4097 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
4098 		    pfil_mbuf_in(V_inet_pfil_head, mp, bifp, NULL)) !=
4099 		    PFIL_PASS)
4100 			break;
4101 
4102 		/* check if we need to fragment the packet */
4103 		/* bridge_fragment generates a mbuf chain of packets */
4104 		/* that already include eth headers */
4105 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
4106 			i = (*mp)->m_pkthdr.len;
4107 			if (i > ifp->if_mtu) {
4108 				error = bridge_fragment(ifp, mp, &eh2, snap,
4109 					    &llc1);
4110 				return (error);
4111 			}
4112 		}
4113 
4114 		/* Recalculate the ip checksum. */
4115 		ip = mtod(*mp, struct ip *);
4116 		hlen = ip->ip_hl << 2;
4117 		if (hlen < sizeof(struct ip))
4118 			goto bad;
4119 		if (hlen > (*mp)->m_len) {
4120 			if ((*mp = m_pullup(*mp, hlen)) == NULL)
4121 				goto bad;
4122 			ip = mtod(*mp, struct ip *);
4123 			if (ip == NULL)
4124 				goto bad;
4125 		}
4126 		ip->ip_sum = 0;
4127 		if (hlen == sizeof(struct ip))
4128 			ip->ip_sum = in_cksum_hdr(ip);
4129 		else
4130 			ip->ip_sum = in_cksum(*mp, hlen);
4131 
4132 		break;
4133 #endif /* INET */
4134 #ifdef INET6
4135 	case ETHERTYPE_IPV6:
4136 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
4137 		    pfil_mbuf_out(V_inet6_pfil_head, mp, bifp, NULL)) !=
4138 		    PFIL_PASS)
4139 			break;
4140 
4141 		if (V_pfil_member && ifp != NULL) {
4142 			rv = (dir == PFIL_OUT) ?
4143 			    pfil_mbuf_out(V_inet6_pfil_head, mp, ifp, NULL) :
4144 			    pfil_mbuf_in(V_inet6_pfil_head, mp, ifp, NULL);
4145 			if (rv != PFIL_PASS)
4146 				break;
4147 		}
4148 
4149 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
4150 		    pfil_mbuf_in(V_inet6_pfil_head, mp, bifp, NULL)) !=
4151 		    PFIL_PASS)
4152 			break;
4153 		break;
4154 #endif
4155 	}
4156 
4157 	switch (rv) {
4158 	case PFIL_CONSUMED:
4159 		return (0);
4160 	case PFIL_DROPPED:
4161 		return (EACCES);
4162 	default:
4163 		break;
4164 	}
4165 
4166 	error = -1;
4167 
4168 	/*
4169 	 * Finally, put everything back the way it was and return
4170 	 */
4171 	if (snap) {
4172 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
4173 		if (*mp == NULL)
4174 			return (error);
4175 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
4176 	}
4177 
4178 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
4179 	if (*mp == NULL)
4180 		return (error);
4181 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
4182 
4183 	return (0);
4184 
4185 bad:
4186 	m_freem(*mp);
4187 	*mp = NULL;
4188 	return (error);
4189 }
4190 
4191 #ifdef INET
4192 /*
4193  * Perform basic checks on header size since
4194  * pfil assumes ip_input has already processed
4195  * it for it.  Cut-and-pasted from ip_input.c.
4196  * Given how simple the IPv6 version is,
4197  * does the IPv4 version really need to be
4198  * this complicated?
4199  *
4200  * XXX Should we update ipstat here, or not?
4201  * XXX Right now we update ipstat but not
4202  * XXX csum_counter.
4203  */
4204 static int
4205 bridge_ip_checkbasic(struct mbuf **mp)
4206 {
4207 	struct mbuf *m = *mp;
4208 	struct ip *ip;
4209 	int len, hlen;
4210 	u_short sum;
4211 
4212 	if (*mp == NULL)
4213 		return (-1);
4214 
4215 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
4216 		if ((m = m_copyup(m, sizeof(struct ip),
4217 			(max_linkhdr + 3) & ~3)) == NULL) {
4218 			/* XXXJRT new stat, please */
4219 			KMOD_IPSTAT_INC(ips_toosmall);
4220 			goto bad;
4221 		}
4222 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
4223 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
4224 			KMOD_IPSTAT_INC(ips_toosmall);
4225 			goto bad;
4226 		}
4227 	}
4228 	ip = mtod(m, struct ip *);
4229 	if (ip == NULL) goto bad;
4230 
4231 	if (ip->ip_v != IPVERSION) {
4232 		KMOD_IPSTAT_INC(ips_badvers);
4233 		goto bad;
4234 	}
4235 	hlen = ip->ip_hl << 2;
4236 	if (hlen < sizeof(struct ip)) { /* minimum header length */
4237 		KMOD_IPSTAT_INC(ips_badhlen);
4238 		goto bad;
4239 	}
4240 	if (hlen > m->m_len) {
4241 		if ((m = m_pullup(m, hlen)) == NULL) {
4242 			KMOD_IPSTAT_INC(ips_badhlen);
4243 			goto bad;
4244 		}
4245 		ip = mtod(m, struct ip *);
4246 		if (ip == NULL) goto bad;
4247 	}
4248 
4249 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
4250 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
4251 	} else {
4252 		if (hlen == sizeof(struct ip)) {
4253 			sum = in_cksum_hdr(ip);
4254 		} else {
4255 			sum = in_cksum(m, hlen);
4256 		}
4257 	}
4258 	if (sum) {
4259 		KMOD_IPSTAT_INC(ips_badsum);
4260 		goto bad;
4261 	}
4262 
4263 	/* Retrieve the packet length. */
4264 	len = ntohs(ip->ip_len);
4265 
4266 	/*
4267 	 * Check for additional length bogosity
4268 	 */
4269 	if (len < hlen) {
4270 		KMOD_IPSTAT_INC(ips_badlen);
4271 		goto bad;
4272 	}
4273 
4274 	/*
4275 	 * Check that the amount of data in the buffers
4276 	 * is as at least much as the IP header would have us expect.
4277 	 * Drop packet if shorter than we expect.
4278 	 */
4279 	if (m->m_pkthdr.len < len) {
4280 		KMOD_IPSTAT_INC(ips_tooshort);
4281 		goto bad;
4282 	}
4283 
4284 	/* Checks out, proceed */
4285 	*mp = m;
4286 	return (0);
4287 
4288 bad:
4289 	*mp = m;
4290 	return (-1);
4291 }
4292 #endif /* INET */
4293 
4294 #ifdef INET6
4295 /*
4296  * Same as above, but for IPv6.
4297  * Cut-and-pasted from ip6_input.c.
4298  * XXX Should we update ip6stat, or not?
4299  */
4300 static int
4301 bridge_ip6_checkbasic(struct mbuf **mp)
4302 {
4303 	struct mbuf *m = *mp;
4304 	struct ip6_hdr *ip6;
4305 
4306 	/*
4307 	 * If the IPv6 header is not aligned, slurp it up into a new
4308 	 * mbuf with space for link headers, in the event we forward
4309 	 * it.  Otherwise, if it is aligned, make sure the entire base
4310 	 * IPv6 header is in the first mbuf of the chain.
4311 	 */
4312 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
4313 		struct ifnet *inifp = m->m_pkthdr.rcvif;
4314 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
4315 			    (max_linkhdr + 3) & ~3)) == NULL) {
4316 			/* XXXJRT new stat, please */
4317 			IP6STAT_INC(ip6s_toosmall);
4318 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
4319 			goto bad;
4320 		}
4321 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
4322 		struct ifnet *inifp = m->m_pkthdr.rcvif;
4323 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
4324 			IP6STAT_INC(ip6s_toosmall);
4325 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
4326 			goto bad;
4327 		}
4328 	}
4329 
4330 	ip6 = mtod(m, struct ip6_hdr *);
4331 
4332 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
4333 		IP6STAT_INC(ip6s_badvers);
4334 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
4335 		goto bad;
4336 	}
4337 
4338 	/* Checks out, proceed */
4339 	*mp = m;
4340 	return (0);
4341 
4342 bad:
4343 	*mp = m;
4344 	return (-1);
4345 }
4346 #endif /* INET6 */
4347 
4348 #ifdef INET
4349 /*
4350  * bridge_fragment:
4351  *
4352  *	Fragment mbuf chain in multiple packets and prepend ethernet header.
4353  */
4354 static int
4355 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
4356     int snap, struct llc *llc)
4357 {
4358 	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
4359 	struct ip *ip;
4360 	int error = -1;
4361 
4362 	if (m->m_len < sizeof(struct ip) &&
4363 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
4364 		goto dropit;
4365 	ip = mtod(m, struct ip *);
4366 
4367 	m->m_pkthdr.csum_flags |= CSUM_IP;
4368 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
4369 	if (error)
4370 		goto dropit;
4371 
4372 	/*
4373 	 * Walk the chain and re-add the Ethernet header for
4374 	 * each mbuf packet.
4375 	 */
4376 	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
4377 		nextpkt = mcur->m_nextpkt;
4378 		mcur->m_nextpkt = NULL;
4379 		if (snap) {
4380 			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
4381 			if (mcur == NULL) {
4382 				error = ENOBUFS;
4383 				if (mprev != NULL)
4384 					mprev->m_nextpkt = nextpkt;
4385 				goto dropit;
4386 			}
4387 			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
4388 		}
4389 
4390 		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
4391 		if (mcur == NULL) {
4392 			error = ENOBUFS;
4393 			if (mprev != NULL)
4394 				mprev->m_nextpkt = nextpkt;
4395 			goto dropit;
4396 		}
4397 		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
4398 
4399 		/*
4400 		 * The previous two M_PREPEND could have inserted one or two
4401 		 * mbufs in front so we have to update the previous packet's
4402 		 * m_nextpkt.
4403 		 */
4404 		mcur->m_nextpkt = nextpkt;
4405 		if (mprev != NULL)
4406 			mprev->m_nextpkt = mcur;
4407 		else {
4408 			/* The first mbuf in the original chain needs to be
4409 			 * updated. */
4410 			*mp = mcur;
4411 		}
4412 		mprev = mcur;
4413 	}
4414 
4415 	KMOD_IPSTAT_INC(ips_fragmented);
4416 	return (error);
4417 
4418 dropit:
4419 	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
4420 		m = mcur->m_nextpkt;
4421 		m_freem(mcur);
4422 	}
4423 	return (error);
4424 }
4425 #endif /* INET */
4426 
4427 static void
4428 bridge_linkstate(struct ifnet *ifp)
4429 {
4430 	struct bridge_softc *sc = NULL;
4431 	struct bridge_iflist *bif;
4432 	struct epoch_tracker et;
4433 
4434 	NET_EPOCH_ENTER(et);
4435 
4436 	bif = ifp->if_bridge;
4437 	if (bif)
4438 		sc = bif->bif_sc;
4439 
4440 	if (sc != NULL) {
4441 		bridge_linkcheck(sc);
4442 		bstp_linkstate(&bif->bif_stp);
4443 	}
4444 
4445 	NET_EPOCH_EXIT(et);
4446 }
4447 
4448 static void
4449 bridge_linkcheck(struct bridge_softc *sc)
4450 {
4451 	struct bridge_iflist *bif;
4452 	int new_link, hasls;
4453 
4454 	BRIDGE_LOCK_OR_NET_EPOCH_ASSERT(sc);
4455 
4456 	new_link = LINK_STATE_DOWN;
4457 	hasls = 0;
4458 	/* Our link is considered up if at least one of our ports is active */
4459 	CK_LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
4460 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
4461 			hasls++;
4462 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
4463 			new_link = LINK_STATE_UP;
4464 			break;
4465 		}
4466 	}
4467 	if (!CK_LIST_EMPTY(&sc->sc_iflist) && !hasls) {
4468 		/* If no interfaces support link-state then we default to up */
4469 		new_link = LINK_STATE_UP;
4470 	}
4471 	if_link_state_change(sc->sc_ifp, new_link);
4472 }
4473