xref: /freebsd/sys/net/if_bridge.c (revision 193d9e768ba63fcfb187cfd17f461f7d41345048)
1 /*	$NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $	*/
2 
3 /*
4  * Copyright 2001 Wasabi Systems, Inc.
5  * All rights reserved.
6  *
7  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed for the NetBSD Project by
20  *	Wasabi Systems, Inc.
21  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
22  *    or promote products derived from this software without specific prior
23  *    written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
27  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
28  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
29  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35  * POSSIBILITY OF SUCH DAMAGE.
36  */
37 
38 /*
39  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
40  * All rights reserved.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  *
51  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
52  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
53  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
54  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
55  * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
56  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
57  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
58  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
59  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
60  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
61  * POSSIBILITY OF SUCH DAMAGE.
62  *
63  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
64  */
65 
66 /*
67  * Network interface bridge support.
68  *
69  * TODO:
70  *
71  *	- Currently only supports Ethernet-like interfaces (Ethernet,
72  *	  802.11, VLANs on Ethernet, etc.)  Figure out a nice way
73  *	  to bridge other types of interfaces (FDDI-FDDI, and maybe
74  *	  consider heterogeneous bridges).
75  */
76 
77 #include <sys/cdefs.h>
78 __FBSDID("$FreeBSD$");
79 
80 #include "opt_inet.h"
81 #include "opt_inet6.h"
82 
83 #include <sys/param.h>
84 #include <sys/eventhandler.h>
85 #include <sys/mbuf.h>
86 #include <sys/malloc.h>
87 #include <sys/protosw.h>
88 #include <sys/systm.h>
89 #include <sys/jail.h>
90 #include <sys/time.h>
91 #include <sys/socket.h> /* for net/if.h */
92 #include <sys/sockio.h>
93 #include <sys/ctype.h>  /* string functions */
94 #include <sys/kernel.h>
95 #include <sys/random.h>
96 #include <sys/syslog.h>
97 #include <sys/sysctl.h>
98 #include <vm/uma.h>
99 #include <sys/module.h>
100 #include <sys/priv.h>
101 #include <sys/proc.h>
102 #include <sys/lock.h>
103 #include <sys/mutex.h>
104 
105 #include <net/bpf.h>
106 #include <net/if.h>
107 #include <net/if_clone.h>
108 #include <net/if_dl.h>
109 #include <net/if_types.h>
110 #include <net/if_var.h>
111 #include <net/pfil.h>
112 #include <net/vnet.h>
113 
114 #include <netinet/in.h>
115 #include <netinet/in_systm.h>
116 #include <netinet/in_var.h>
117 #include <netinet/ip.h>
118 #include <netinet/ip_var.h>
119 #ifdef INET6
120 #include <netinet/ip6.h>
121 #include <netinet6/ip6_var.h>
122 #include <netinet6/in6_ifattach.h>
123 #endif
124 #if defined(INET) || defined(INET6)
125 #include <netinet/ip_carp.h>
126 #endif
127 #include <machine/in_cksum.h>
128 #include <netinet/if_ether.h>
129 #include <net/bridgestp.h>
130 #include <net/if_bridgevar.h>
131 #include <net/if_llc.h>
132 #include <net/if_vlan_var.h>
133 
134 #include <net/route.h>
135 
136 /*
137  * Size of the route hash table.  Must be a power of two.
138  */
139 #ifndef BRIDGE_RTHASH_SIZE
140 #define	BRIDGE_RTHASH_SIZE		1024
141 #endif
142 
143 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
144 
145 /*
146  * Default maximum number of addresses to cache.
147  */
148 #ifndef BRIDGE_RTABLE_MAX
149 #define	BRIDGE_RTABLE_MAX		2000
150 #endif
151 
152 /*
153  * Timeout (in seconds) for entries learned dynamically.
154  */
155 #ifndef BRIDGE_RTABLE_TIMEOUT
156 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
157 #endif
158 
159 /*
160  * Number of seconds between walks of the route list.
161  */
162 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
163 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
164 #endif
165 
166 /*
167  * List of capabilities to possibly mask on the member interface.
168  */
169 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
170 					 IFCAP_TXCSUM_IPV6)
171 
172 /*
173  * List of capabilities to strip
174  */
175 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
176 
177 /*
178  * Bridge interface list entry.
179  */
180 struct bridge_iflist {
181 	LIST_ENTRY(bridge_iflist) bif_next;
182 	struct ifnet		*bif_ifp;	/* member if */
183 	struct bstp_port	bif_stp;	/* STP state */
184 	uint32_t		bif_flags;	/* member if flags */
185 	int			bif_savedcaps;	/* saved capabilities */
186 	uint32_t		bif_addrmax;	/* max # of addresses */
187 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
188 	uint32_t		bif_addrexceeded;/* # of address violations */
189 };
190 
191 /*
192  * Bridge route node.
193  */
194 struct bridge_rtnode {
195 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
196 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
197 	struct bridge_iflist	*brt_dst;	/* destination if */
198 	unsigned long		brt_expire;	/* expiration time */
199 	uint8_t			brt_flags;	/* address flags */
200 	uint8_t			brt_addr[ETHER_ADDR_LEN];
201 	uint16_t		brt_vlan;	/* vlan id */
202 };
203 #define	brt_ifp			brt_dst->bif_ifp
204 
205 /*
206  * Software state for each bridge.
207  */
208 struct bridge_softc {
209 	struct ifnet		*sc_ifp;	/* make this an interface */
210 	LIST_ENTRY(bridge_softc) sc_list;
211 	struct mtx		sc_mtx;
212 	struct cv		sc_cv;
213 	uint32_t		sc_brtmax;	/* max # of addresses */
214 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
215 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
216 	struct callout		sc_brcallout;	/* bridge callout */
217 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
218 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
219 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
220 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
221 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
222 	uint32_t		sc_rthash_key;	/* key for hash */
223 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
224 	struct bstp_state	sc_stp;		/* STP state */
225 	uint32_t		sc_brtexceeded;	/* # of cache drops */
226 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
227 	u_char			sc_defaddr[6];	/* Default MAC address */
228 };
229 
230 static VNET_DEFINE(struct mtx, bridge_list_mtx);
231 #define	V_bridge_list_mtx	VNET(bridge_list_mtx)
232 static eventhandler_tag bridge_detach_cookie;
233 
234 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
235 
236 uma_zone_t bridge_rtnode_zone;
237 
238 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
239 static void	bridge_clone_destroy(struct ifnet *);
240 
241 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
242 static void	bridge_mutecaps(struct bridge_softc *);
243 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
244 		    int);
245 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
246 static void	bridge_init(void *);
247 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
248 static void	bridge_stop(struct ifnet *, int);
249 static int	bridge_transmit(struct ifnet *, struct mbuf *);
250 static void	bridge_qflush(struct ifnet *);
251 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
252 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
253 		    struct rtentry *);
254 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
255 		    struct mbuf *);
256 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
257 
258 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
259 		    struct mbuf *m);
260 
261 static void	bridge_timer(void *);
262 
263 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
264 		    struct mbuf *, int);
265 static void	bridge_span(struct bridge_softc *, struct mbuf *);
266 
267 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
268 		    uint16_t, struct bridge_iflist *, int, uint8_t);
269 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
270 		    uint16_t);
271 static void	bridge_rttrim(struct bridge_softc *);
272 static void	bridge_rtage(struct bridge_softc *);
273 static void	bridge_rtflush(struct bridge_softc *, int);
274 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
275 		    uint16_t);
276 
277 static void	bridge_rtable_init(struct bridge_softc *);
278 static void	bridge_rtable_fini(struct bridge_softc *);
279 
280 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
281 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
282 		    const uint8_t *, uint16_t);
283 static int	bridge_rtnode_insert(struct bridge_softc *,
284 		    struct bridge_rtnode *);
285 static void	bridge_rtnode_destroy(struct bridge_softc *,
286 		    struct bridge_rtnode *);
287 static void	bridge_rtable_expire(struct ifnet *, int);
288 static void	bridge_state_change(struct ifnet *, int);
289 
290 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
291 		    const char *name);
292 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
293 		    struct ifnet *ifp);
294 static void	bridge_delete_member(struct bridge_softc *,
295 		    struct bridge_iflist *, int);
296 static void	bridge_delete_span(struct bridge_softc *,
297 		    struct bridge_iflist *);
298 
299 static int	bridge_ioctl_add(struct bridge_softc *, void *);
300 static int	bridge_ioctl_del(struct bridge_softc *, void *);
301 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
302 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
303 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
304 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
305 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
306 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
307 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
308 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
309 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
310 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
311 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
312 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
313 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
314 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
315 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
316 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
317 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
318 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
319 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
320 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
321 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
322 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
323 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
324 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
325 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
326 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
327 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
328 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
329 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
330 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
331 		    int);
332 static int	bridge_ip_checkbasic(struct mbuf **mp);
333 #ifdef INET6
334 static int	bridge_ip6_checkbasic(struct mbuf **mp);
335 #endif /* INET6 */
336 static int	bridge_fragment(struct ifnet *, struct mbuf **mp,
337 		    struct ether_header *, int, struct llc *);
338 static void	bridge_linkstate(struct ifnet *ifp);
339 static void	bridge_linkcheck(struct bridge_softc *sc);
340 
341 extern void (*bridge_linkstate_p)(struct ifnet *ifp);
342 
343 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
344 #define	VLANTAGOF(_m)	\
345     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
346 
347 static struct bstp_cb_ops bridge_ops = {
348 	.bcb_state = bridge_state_change,
349 	.bcb_rtage = bridge_rtable_expire
350 };
351 
352 SYSCTL_DECL(_net_link);
353 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
354 
355 /* only pass IP[46] packets when pfil is enabled */
356 static VNET_DEFINE(int, pfil_onlyip) = 1;
357 #define	V_pfil_onlyip	VNET(pfil_onlyip)
358 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
359     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
360     "Only pass IP packets when pfil is enabled");
361 
362 /* run pfil hooks on the bridge interface */
363 static VNET_DEFINE(int, pfil_bridge) = 1;
364 #define	V_pfil_bridge	VNET(pfil_bridge)
365 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
366     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
367     "Packet filter on the bridge interface");
368 
369 /* layer2 filter with ipfw */
370 static VNET_DEFINE(int, pfil_ipfw);
371 #define	V_pfil_ipfw	VNET(pfil_ipfw)
372 
373 /* layer2 ARP filter with ipfw */
374 static VNET_DEFINE(int, pfil_ipfw_arp);
375 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
376 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
377     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
378     "Filter ARP packets through IPFW layer2");
379 
380 /* run pfil hooks on the member interface */
381 static VNET_DEFINE(int, pfil_member) = 1;
382 #define	V_pfil_member	VNET(pfil_member)
383 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
384     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
385     "Packet filter on the member interface");
386 
387 /* run pfil hooks on the physical interface for locally destined packets */
388 static VNET_DEFINE(int, pfil_local_phys);
389 #define	V_pfil_local_phys	VNET(pfil_local_phys)
390 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
391     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
392     "Packet filter on the physical interface for locally destined packets");
393 
394 /* log STP state changes */
395 static VNET_DEFINE(int, log_stp);
396 #define	V_log_stp	VNET(log_stp)
397 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
398     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
399     "Log STP state changes");
400 
401 /* share MAC with first bridge member */
402 static VNET_DEFINE(int, bridge_inherit_mac);
403 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
404 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
405     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
406     "Inherit MAC address from the first bridge member");
407 
408 static VNET_DEFINE(int, allow_llz_overlap) = 0;
409 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
410 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
411     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
412     "Allow overlap of link-local scope "
413     "zones of a bridge interface and the member interfaces");
414 
415 struct bridge_control {
416 	int	(*bc_func)(struct bridge_softc *, void *);
417 	int	bc_argsize;
418 	int	bc_flags;
419 };
420 
421 #define	BC_F_COPYIN		0x01	/* copy arguments in */
422 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
423 #define	BC_F_SUSER		0x04	/* do super-user check */
424 
425 const struct bridge_control bridge_control_table[] = {
426 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
427 	  BC_F_COPYIN|BC_F_SUSER },
428 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
429 	  BC_F_COPYIN|BC_F_SUSER },
430 
431 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
432 	  BC_F_COPYIN|BC_F_COPYOUT },
433 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
434 	  BC_F_COPYIN|BC_F_SUSER },
435 
436 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
437 	  BC_F_COPYIN|BC_F_SUSER },
438 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
439 	  BC_F_COPYOUT },
440 
441 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
442 	  BC_F_COPYIN|BC_F_COPYOUT },
443 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
444 	  BC_F_COPYIN|BC_F_COPYOUT },
445 
446 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
447 	  BC_F_COPYIN|BC_F_SUSER },
448 
449 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
450 	  BC_F_COPYIN|BC_F_SUSER },
451 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
452 	  BC_F_COPYOUT },
453 
454 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
455 	  BC_F_COPYIN|BC_F_SUSER },
456 
457 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
458 	  BC_F_COPYIN|BC_F_SUSER },
459 
460 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
461 	  BC_F_COPYOUT },
462 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
463 	  BC_F_COPYIN|BC_F_SUSER },
464 
465 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
466 	  BC_F_COPYOUT },
467 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
468 	  BC_F_COPYIN|BC_F_SUSER },
469 
470 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
471 	  BC_F_COPYOUT },
472 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
473 	  BC_F_COPYIN|BC_F_SUSER },
474 
475 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
476 	  BC_F_COPYOUT },
477 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
478 	  BC_F_COPYIN|BC_F_SUSER },
479 
480 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
481 	  BC_F_COPYIN|BC_F_SUSER },
482 
483 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
484 	  BC_F_COPYIN|BC_F_SUSER },
485 
486 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
487 	  BC_F_COPYIN|BC_F_SUSER },
488 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
489 	  BC_F_COPYIN|BC_F_SUSER },
490 
491 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
492 	  BC_F_COPYOUT },
493 
494 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
495 	  BC_F_COPYOUT },
496 
497 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
498 	  BC_F_COPYIN|BC_F_COPYOUT },
499 
500 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
501 	  BC_F_COPYIN|BC_F_SUSER },
502 
503 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
504 	  BC_F_COPYIN|BC_F_SUSER },
505 
506 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
507 	  BC_F_COPYIN|BC_F_SUSER },
508 
509 };
510 const int bridge_control_table_size = nitems(bridge_control_table);
511 
512 static VNET_DEFINE(LIST_HEAD(, bridge_softc), bridge_list);
513 #define	V_bridge_list	VNET(bridge_list)
514 #define	BRIDGE_LIST_LOCK_INIT(x)	mtx_init(&V_bridge_list_mtx,	\
515 					    "if_bridge list", NULL, MTX_DEF)
516 #define	BRIDGE_LIST_LOCK_DESTROY(x)	mtx_destroy(&V_bridge_list_mtx)
517 #define	BRIDGE_LIST_LOCK(x)		mtx_lock(&V_bridge_list_mtx)
518 #define	BRIDGE_LIST_UNLOCK(x)		mtx_unlock(&V_bridge_list_mtx)
519 
520 static VNET_DEFINE(struct if_clone *, bridge_cloner);
521 #define	V_bridge_cloner	VNET(bridge_cloner)
522 
523 static const char bridge_name[] = "bridge";
524 
525 static void
526 vnet_bridge_init(const void *unused __unused)
527 {
528 
529 	BRIDGE_LIST_LOCK_INIT();
530 	LIST_INIT(&V_bridge_list);
531 	V_bridge_cloner = if_clone_simple(bridge_name,
532 	    bridge_clone_create, bridge_clone_destroy, 0);
533 }
534 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
535     vnet_bridge_init, NULL);
536 
537 static void
538 vnet_bridge_uninit(const void *unused __unused)
539 {
540 
541 	if_clone_detach(V_bridge_cloner);
542 	V_bridge_cloner = NULL;
543 	BRIDGE_LIST_LOCK_DESTROY();
544 }
545 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
546     vnet_bridge_uninit, NULL);
547 
548 static int
549 bridge_modevent(module_t mod, int type, void *data)
550 {
551 
552 	switch (type) {
553 	case MOD_LOAD:
554 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
555 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
556 		    UMA_ALIGN_PTR, 0);
557 		bridge_input_p = bridge_input;
558 		bridge_output_p = bridge_output;
559 		bridge_dn_p = bridge_dummynet;
560 		bridge_linkstate_p = bridge_linkstate;
561 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
562 		    ifnet_departure_event, bridge_ifdetach, NULL,
563 		    EVENTHANDLER_PRI_ANY);
564 		break;
565 	case MOD_UNLOAD:
566 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
567 		    bridge_detach_cookie);
568 		uma_zdestroy(bridge_rtnode_zone);
569 		bridge_input_p = NULL;
570 		bridge_output_p = NULL;
571 		bridge_dn_p = NULL;
572 		bridge_linkstate_p = NULL;
573 		break;
574 	default:
575 		return (EOPNOTSUPP);
576 	}
577 	return (0);
578 }
579 
580 static moduledata_t bridge_mod = {
581 	"if_bridge",
582 	bridge_modevent,
583 	0
584 };
585 
586 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
587 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
588 
589 /*
590  * handler for net.link.bridge.ipfw
591  */
592 static int
593 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
594 {
595 	int enable = V_pfil_ipfw;
596 	int error;
597 
598 	error = sysctl_handle_int(oidp, &enable, 0, req);
599 	enable &= 1;
600 
601 	if (enable != V_pfil_ipfw) {
602 		V_pfil_ipfw = enable;
603 
604 		/*
605 		 * Disable pfil so that ipfw doesnt run twice, if the user
606 		 * really wants both then they can re-enable pfil_bridge and/or
607 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
608 		 * layer2 type.
609 		 */
610 		if (V_pfil_ipfw) {
611 			V_pfil_onlyip = 0;
612 			V_pfil_bridge = 0;
613 			V_pfil_member = 0;
614 		}
615 	}
616 
617 	return (error);
618 }
619 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
620     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
621     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
622     "Layer2 filter with IPFW");
623 
624 /*
625  * bridge_clone_create:
626  *
627  *	Create a new bridge instance.
628  */
629 static int
630 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
631 {
632 	struct bridge_softc *sc, *sc2;
633 	struct ifnet *bifp, *ifp;
634 	int fb, retry;
635 	unsigned long hostid;
636 
637 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
638 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
639 	if (ifp == NULL) {
640 		free(sc, M_DEVBUF);
641 		return (ENOSPC);
642 	}
643 
644 	BRIDGE_LOCK_INIT(sc);
645 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
646 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
647 
648 	/* Initialize our routing table. */
649 	bridge_rtable_init(sc);
650 
651 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
652 
653 	LIST_INIT(&sc->sc_iflist);
654 	LIST_INIT(&sc->sc_spanlist);
655 
656 	ifp->if_softc = sc;
657 	if_initname(ifp, bridge_name, unit);
658 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
659 	ifp->if_ioctl = bridge_ioctl;
660 	ifp->if_transmit = bridge_transmit;
661 	ifp->if_qflush = bridge_qflush;
662 	ifp->if_init = bridge_init;
663 	ifp->if_type = IFT_BRIDGE;
664 
665 	/*
666 	 * Generate an ethernet address with a locally administered address.
667 	 *
668 	 * Since we are using random ethernet addresses for the bridge, it is
669 	 * possible that we might have address collisions, so make sure that
670 	 * this hardware address isn't already in use on another bridge.
671 	 * The first try uses the hostid and falls back to arc4rand().
672 	 */
673 	fb = 0;
674 	getcredhostid(curthread->td_ucred, &hostid);
675 	do {
676 		if (fb || hostid == 0) {
677 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
678 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
679 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
680 		} else {
681 			sc->sc_defaddr[0] = 0x2;
682 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
683 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
684 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
685 			sc->sc_defaddr[4] =  hostid        & 0xff;
686 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
687 		}
688 
689 		fb = 1;
690 		retry = 0;
691 		BRIDGE_LIST_LOCK();
692 		LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
693 			bifp = sc2->sc_ifp;
694 			if (memcmp(sc->sc_defaddr,
695 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
696 				retry = 1;
697 				break;
698 			}
699 		}
700 		BRIDGE_LIST_UNLOCK();
701 	} while (retry == 1);
702 
703 	bstp_attach(&sc->sc_stp, &bridge_ops);
704 	ether_ifattach(ifp, sc->sc_defaddr);
705 	/* Now undo some of the damage... */
706 	ifp->if_baudrate = 0;
707 	ifp->if_type = IFT_BRIDGE;
708 
709 	BRIDGE_LIST_LOCK();
710 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
711 	BRIDGE_LIST_UNLOCK();
712 
713 	return (0);
714 }
715 
716 /*
717  * bridge_clone_destroy:
718  *
719  *	Destroy a bridge instance.
720  */
721 static void
722 bridge_clone_destroy(struct ifnet *ifp)
723 {
724 	struct bridge_softc *sc = ifp->if_softc;
725 	struct bridge_iflist *bif;
726 
727 	BRIDGE_LOCK(sc);
728 
729 	bridge_stop(ifp, 1);
730 	ifp->if_flags &= ~IFF_UP;
731 
732 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
733 		bridge_delete_member(sc, bif, 0);
734 
735 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
736 		bridge_delete_span(sc, bif);
737 	}
738 
739 	BRIDGE_UNLOCK(sc);
740 
741 	callout_drain(&sc->sc_brcallout);
742 
743 	BRIDGE_LIST_LOCK();
744 	LIST_REMOVE(sc, sc_list);
745 	BRIDGE_LIST_UNLOCK();
746 
747 	bstp_detach(&sc->sc_stp);
748 	ether_ifdetach(ifp);
749 	if_free(ifp);
750 
751 	/* Tear down the routing table. */
752 	bridge_rtable_fini(sc);
753 
754 	BRIDGE_LOCK_DESTROY(sc);
755 	free(sc, M_DEVBUF);
756 }
757 
758 /*
759  * bridge_ioctl:
760  *
761  *	Handle a control request from the operator.
762  */
763 static int
764 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
765 {
766 	struct bridge_softc *sc = ifp->if_softc;
767 	struct ifreq *ifr = (struct ifreq *)data;
768 	struct bridge_iflist *bif;
769 	struct thread *td = curthread;
770 	union {
771 		struct ifbreq ifbreq;
772 		struct ifbifconf ifbifconf;
773 		struct ifbareq ifbareq;
774 		struct ifbaconf ifbaconf;
775 		struct ifbrparam ifbrparam;
776 		struct ifbropreq ifbropreq;
777 	} args;
778 	struct ifdrv *ifd = (struct ifdrv *) data;
779 	const struct bridge_control *bc;
780 	int error = 0;
781 
782 	switch (cmd) {
783 
784 	case SIOCADDMULTI:
785 	case SIOCDELMULTI:
786 		break;
787 
788 	case SIOCGDRVSPEC:
789 	case SIOCSDRVSPEC:
790 		if (ifd->ifd_cmd >= bridge_control_table_size) {
791 			error = EINVAL;
792 			break;
793 		}
794 		bc = &bridge_control_table[ifd->ifd_cmd];
795 
796 		if (cmd == SIOCGDRVSPEC &&
797 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
798 			error = EINVAL;
799 			break;
800 		}
801 		else if (cmd == SIOCSDRVSPEC &&
802 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
803 			error = EINVAL;
804 			break;
805 		}
806 
807 		if (bc->bc_flags & BC_F_SUSER) {
808 			error = priv_check(td, PRIV_NET_BRIDGE);
809 			if (error)
810 				break;
811 		}
812 
813 		if (ifd->ifd_len != bc->bc_argsize ||
814 		    ifd->ifd_len > sizeof(args)) {
815 			error = EINVAL;
816 			break;
817 		}
818 
819 		bzero(&args, sizeof(args));
820 		if (bc->bc_flags & BC_F_COPYIN) {
821 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
822 			if (error)
823 				break;
824 		}
825 
826 		BRIDGE_LOCK(sc);
827 		error = (*bc->bc_func)(sc, &args);
828 		BRIDGE_UNLOCK(sc);
829 		if (error)
830 			break;
831 
832 		if (bc->bc_flags & BC_F_COPYOUT)
833 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
834 
835 		break;
836 
837 	case SIOCSIFFLAGS:
838 		if (!(ifp->if_flags & IFF_UP) &&
839 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
840 			/*
841 			 * If interface is marked down and it is running,
842 			 * then stop and disable it.
843 			 */
844 			BRIDGE_LOCK(sc);
845 			bridge_stop(ifp, 1);
846 			BRIDGE_UNLOCK(sc);
847 		} else if ((ifp->if_flags & IFF_UP) &&
848 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
849 			/*
850 			 * If interface is marked up and it is stopped, then
851 			 * start it.
852 			 */
853 			(*ifp->if_init)(sc);
854 		}
855 		break;
856 
857 	case SIOCSIFMTU:
858 		if (ifr->ifr_mtu < 576) {
859 			error = EINVAL;
860 			break;
861 		}
862 		if (LIST_EMPTY(&sc->sc_iflist)) {
863 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
864 			break;
865 		}
866 		BRIDGE_LOCK(sc);
867 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
868 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
869 				log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
870 				    " != %d\n", sc->sc_ifp->if_xname,
871 				    bif->bif_ifp->if_mtu,
872 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
873 				error = EINVAL;
874 				break;
875 			}
876 		}
877 		if (!error)
878 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
879 		BRIDGE_UNLOCK(sc);
880 		break;
881 	default:
882 		/*
883 		 * drop the lock as ether_ioctl() will call bridge_start() and
884 		 * cause the lock to be recursed.
885 		 */
886 		error = ether_ioctl(ifp, cmd, data);
887 		break;
888 	}
889 
890 	return (error);
891 }
892 
893 /*
894  * bridge_mutecaps:
895  *
896  *	Clear or restore unwanted capabilities on the member interface
897  */
898 static void
899 bridge_mutecaps(struct bridge_softc *sc)
900 {
901 	struct bridge_iflist *bif;
902 	int enabled, mask;
903 
904 	/* Initial bitmask of capabilities to test */
905 	mask = BRIDGE_IFCAPS_MASK;
906 
907 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
908 		/* Every member must support it or its disabled */
909 		mask &= bif->bif_savedcaps;
910 	}
911 
912 	BRIDGE_XLOCK(sc);
913 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
914 		enabled = bif->bif_ifp->if_capenable;
915 		enabled &= ~BRIDGE_IFCAPS_STRIP;
916 		/* strip off mask bits and enable them again if allowed */
917 		enabled &= ~BRIDGE_IFCAPS_MASK;
918 		enabled |= mask;
919 		BRIDGE_UNLOCK(sc);
920 		bridge_set_ifcap(sc, bif, enabled);
921 		BRIDGE_LOCK(sc);
922 	}
923 	BRIDGE_XDROP(sc);
924 
925 }
926 
927 static void
928 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
929 {
930 	struct ifnet *ifp = bif->bif_ifp;
931 	struct ifreq ifr;
932 	int error;
933 
934 	BRIDGE_UNLOCK_ASSERT(sc);
935 
936 	bzero(&ifr, sizeof(ifr));
937 	ifr.ifr_reqcap = set;
938 
939 	if (ifp->if_capenable != set) {
940 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
941 		if (error)
942 			if_printf(sc->sc_ifp,
943 			    "error setting interface capabilities on %s\n",
944 			    ifp->if_xname);
945 	}
946 }
947 
948 /*
949  * bridge_lookup_member:
950  *
951  *	Lookup a bridge member interface.
952  */
953 static struct bridge_iflist *
954 bridge_lookup_member(struct bridge_softc *sc, const char *name)
955 {
956 	struct bridge_iflist *bif;
957 	struct ifnet *ifp;
958 
959 	BRIDGE_LOCK_ASSERT(sc);
960 
961 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
962 		ifp = bif->bif_ifp;
963 		if (strcmp(ifp->if_xname, name) == 0)
964 			return (bif);
965 	}
966 
967 	return (NULL);
968 }
969 
970 /*
971  * bridge_lookup_member_if:
972  *
973  *	Lookup a bridge member interface by ifnet*.
974  */
975 static struct bridge_iflist *
976 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
977 {
978 	struct bridge_iflist *bif;
979 
980 	BRIDGE_LOCK_ASSERT(sc);
981 
982 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
983 		if (bif->bif_ifp == member_ifp)
984 			return (bif);
985 	}
986 
987 	return (NULL);
988 }
989 
990 /*
991  * bridge_delete_member:
992  *
993  *	Delete the specified member interface.
994  */
995 static void
996 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
997     int gone)
998 {
999 	struct ifnet *ifs = bif->bif_ifp;
1000 	struct ifnet *fif = NULL;
1001 
1002 	BRIDGE_LOCK_ASSERT(sc);
1003 
1004 	if (bif->bif_flags & IFBIF_STP)
1005 		bstp_disable(&bif->bif_stp);
1006 
1007 	ifs->if_bridge = NULL;
1008 	BRIDGE_XLOCK(sc);
1009 	LIST_REMOVE(bif, bif_next);
1010 	BRIDGE_XDROP(sc);
1011 
1012 	/*
1013 	 * If removing the interface that gave the bridge its mac address, set
1014 	 * the mac address of the bridge to the address of the next member, or
1015 	 * to its default address if no members are left.
1016 	 */
1017 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1018 		if (LIST_EMPTY(&sc->sc_iflist)) {
1019 			bcopy(sc->sc_defaddr,
1020 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1021 			sc->sc_ifaddr = NULL;
1022 		} else {
1023 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
1024 			bcopy(IF_LLADDR(fif),
1025 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1026 			sc->sc_ifaddr = fif;
1027 		}
1028 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1029 	}
1030 
1031 	bridge_linkcheck(sc);
1032 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1033 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1034 	KASSERT(bif->bif_addrcnt == 0,
1035 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1036 
1037 	BRIDGE_UNLOCK(sc);
1038 	if (!gone) {
1039 		switch (ifs->if_type) {
1040 		case IFT_ETHER:
1041 		case IFT_L2VLAN:
1042 			/*
1043 			 * Take the interface out of promiscuous mode, but only
1044 			 * if it was promiscuous in the first place. It might
1045 			 * not be if we're in the bridge_ioctl_add() error path.
1046 			 */
1047 			if (ifs->if_flags & IFF_PROMISC)
1048 				(void) ifpromisc(ifs, 0);
1049 			break;
1050 
1051 		case IFT_GIF:
1052 			break;
1053 
1054 		default:
1055 #ifdef DIAGNOSTIC
1056 			panic("bridge_delete_member: impossible");
1057 #endif
1058 			break;
1059 		}
1060 		/* reneable any interface capabilities */
1061 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1062 	}
1063 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1064 	BRIDGE_LOCK(sc);
1065 	free(bif, M_DEVBUF);
1066 }
1067 
1068 /*
1069  * bridge_delete_span:
1070  *
1071  *	Delete the specified span interface.
1072  */
1073 static void
1074 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1075 {
1076 	BRIDGE_LOCK_ASSERT(sc);
1077 
1078 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1079 	    ("%s: not a span interface", __func__));
1080 
1081 	LIST_REMOVE(bif, bif_next);
1082 	free(bif, M_DEVBUF);
1083 }
1084 
1085 static int
1086 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1087 {
1088 	struct ifbreq *req = arg;
1089 	struct bridge_iflist *bif = NULL;
1090 	struct ifnet *ifs;
1091 	int error = 0;
1092 
1093 	ifs = ifunit(req->ifbr_ifsname);
1094 	if (ifs == NULL)
1095 		return (ENOENT);
1096 	if (ifs->if_ioctl == NULL)	/* must be supported */
1097 		return (EINVAL);
1098 
1099 	/* If it's in the span list, it can't be a member. */
1100 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1101 		if (ifs == bif->bif_ifp)
1102 			return (EBUSY);
1103 
1104 	if (ifs->if_bridge == sc)
1105 		return (EEXIST);
1106 
1107 	if (ifs->if_bridge != NULL)
1108 		return (EBUSY);
1109 
1110 	switch (ifs->if_type) {
1111 	case IFT_ETHER:
1112 	case IFT_L2VLAN:
1113 	case IFT_GIF:
1114 		/* permitted interface types */
1115 		break;
1116 	default:
1117 		return (EINVAL);
1118 	}
1119 
1120 #ifdef INET6
1121 	/*
1122 	 * Two valid inet6 addresses with link-local scope must not be
1123 	 * on the parent interface and the member interfaces at the
1124 	 * same time.  This restriction is needed to prevent violation
1125 	 * of link-local scope zone.  Attempts to add a member
1126 	 * interface which has inet6 addresses when the parent has
1127 	 * inet6 triggers removal of all inet6 addresses on the member
1128 	 * interface.
1129 	 */
1130 
1131 	/* Check if the parent interface has a link-local scope addr. */
1132 	if (V_allow_llz_overlap == 0 &&
1133 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1134 		/*
1135 		 * If any, remove all inet6 addresses from the member
1136 		 * interfaces.
1137 		 */
1138 		BRIDGE_XLOCK(sc);
1139 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1140  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1141 				BRIDGE_UNLOCK(sc);
1142 				in6_ifdetach(bif->bif_ifp);
1143 				BRIDGE_LOCK(sc);
1144 				if_printf(sc->sc_ifp,
1145 				    "IPv6 addresses on %s have been removed "
1146 				    "before adding it as a member to prevent "
1147 				    "IPv6 address scope violation.\n",
1148 				    bif->bif_ifp->if_xname);
1149 			}
1150 		}
1151 		BRIDGE_XDROP(sc);
1152 		if (in6ifa_llaonifp(ifs)) {
1153 			BRIDGE_UNLOCK(sc);
1154 			in6_ifdetach(ifs);
1155 			BRIDGE_LOCK(sc);
1156 			if_printf(sc->sc_ifp,
1157 			    "IPv6 addresses on %s have been removed "
1158 			    "before adding it as a member to prevent "
1159 			    "IPv6 address scope violation.\n",
1160 			    ifs->if_xname);
1161 		}
1162 	}
1163 #endif
1164 	/* Allow the first Ethernet member to define the MTU */
1165 	if (LIST_EMPTY(&sc->sc_iflist))
1166 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1167 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1168 		if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1169 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1170 		return (EINVAL);
1171 	}
1172 
1173 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1174 	if (bif == NULL)
1175 		return (ENOMEM);
1176 
1177 	bif->bif_ifp = ifs;
1178 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1179 	bif->bif_savedcaps = ifs->if_capenable;
1180 
1181 	/*
1182 	 * Assign the interface's MAC address to the bridge if it's the first
1183 	 * member and the MAC address of the bridge has not been changed from
1184 	 * the default randomly generated one.
1185 	 */
1186 	if (V_bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1187 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1188 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1189 		sc->sc_ifaddr = ifs;
1190 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1191 	}
1192 
1193 	ifs->if_bridge = sc;
1194 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1195 	/*
1196 	 * XXX: XLOCK HERE!?!
1197 	 *
1198 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1199 	 */
1200 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1201 
1202 	/* Set interface capabilities to the intersection set of all members */
1203 	bridge_mutecaps(sc);
1204 	bridge_linkcheck(sc);
1205 
1206 	/* Place the interface into promiscuous mode */
1207 	switch (ifs->if_type) {
1208 		case IFT_ETHER:
1209 		case IFT_L2VLAN:
1210 			BRIDGE_UNLOCK(sc);
1211 			error = ifpromisc(ifs, 1);
1212 			BRIDGE_LOCK(sc);
1213 			break;
1214 	}
1215 
1216 	if (error)
1217 		bridge_delete_member(sc, bif, 0);
1218 	return (error);
1219 }
1220 
1221 static int
1222 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1223 {
1224 	struct ifbreq *req = arg;
1225 	struct bridge_iflist *bif;
1226 
1227 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1228 	if (bif == NULL)
1229 		return (ENOENT);
1230 
1231 	bridge_delete_member(sc, bif, 0);
1232 
1233 	return (0);
1234 }
1235 
1236 static int
1237 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1238 {
1239 	struct ifbreq *req = arg;
1240 	struct bridge_iflist *bif;
1241 	struct bstp_port *bp;
1242 
1243 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1244 	if (bif == NULL)
1245 		return (ENOENT);
1246 
1247 	bp = &bif->bif_stp;
1248 	req->ifbr_ifsflags = bif->bif_flags;
1249 	req->ifbr_state = bp->bp_state;
1250 	req->ifbr_priority = bp->bp_priority;
1251 	req->ifbr_path_cost = bp->bp_path_cost;
1252 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1253 	req->ifbr_proto = bp->bp_protover;
1254 	req->ifbr_role = bp->bp_role;
1255 	req->ifbr_stpflags = bp->bp_flags;
1256 	req->ifbr_addrcnt = bif->bif_addrcnt;
1257 	req->ifbr_addrmax = bif->bif_addrmax;
1258 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1259 
1260 	/* Copy STP state options as flags */
1261 	if (bp->bp_operedge)
1262 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1263 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1264 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1265 	if (bp->bp_ptp_link)
1266 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1267 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1268 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1269 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1270 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1271 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1272 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1273 	return (0);
1274 }
1275 
1276 static int
1277 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1278 {
1279 	struct ifbreq *req = arg;
1280 	struct bridge_iflist *bif;
1281 	struct bstp_port *bp;
1282 	int error;
1283 
1284 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1285 	if (bif == NULL)
1286 		return (ENOENT);
1287 	bp = &bif->bif_stp;
1288 
1289 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1290 		/* SPAN is readonly */
1291 		return (EINVAL);
1292 
1293 	if (req->ifbr_ifsflags & IFBIF_STP) {
1294 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1295 			error = bstp_enable(&bif->bif_stp);
1296 			if (error)
1297 				return (error);
1298 		}
1299 	} else {
1300 		if ((bif->bif_flags & IFBIF_STP) != 0)
1301 			bstp_disable(&bif->bif_stp);
1302 	}
1303 
1304 	/* Pass on STP flags */
1305 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1306 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1307 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1308 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1309 
1310 	/* Save the bits relating to the bridge */
1311 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1312 
1313 	return (0);
1314 }
1315 
1316 static int
1317 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1318 {
1319 	struct ifbrparam *param = arg;
1320 
1321 	sc->sc_brtmax = param->ifbrp_csize;
1322 	bridge_rttrim(sc);
1323 
1324 	return (0);
1325 }
1326 
1327 static int
1328 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1329 {
1330 	struct ifbrparam *param = arg;
1331 
1332 	param->ifbrp_csize = sc->sc_brtmax;
1333 
1334 	return (0);
1335 }
1336 
1337 static int
1338 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1339 {
1340 	struct ifbifconf *bifc = arg;
1341 	struct bridge_iflist *bif;
1342 	struct ifbreq breq;
1343 	char *buf, *outbuf;
1344 	int count, buflen, len, error = 0;
1345 
1346 	count = 0;
1347 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1348 		count++;
1349 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1350 		count++;
1351 
1352 	buflen = sizeof(breq) * count;
1353 	if (bifc->ifbic_len == 0) {
1354 		bifc->ifbic_len = buflen;
1355 		return (0);
1356 	}
1357 	BRIDGE_UNLOCK(sc);
1358 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1359 	BRIDGE_LOCK(sc);
1360 
1361 	count = 0;
1362 	buf = outbuf;
1363 	len = min(bifc->ifbic_len, buflen);
1364 	bzero(&breq, sizeof(breq));
1365 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1366 		if (len < sizeof(breq))
1367 			break;
1368 
1369 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1370 		    sizeof(breq.ifbr_ifsname));
1371 		/* Fill in the ifbreq structure */
1372 		error = bridge_ioctl_gifflags(sc, &breq);
1373 		if (error)
1374 			break;
1375 		memcpy(buf, &breq, sizeof(breq));
1376 		count++;
1377 		buf += sizeof(breq);
1378 		len -= sizeof(breq);
1379 	}
1380 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1381 		if (len < sizeof(breq))
1382 			break;
1383 
1384 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1385 		    sizeof(breq.ifbr_ifsname));
1386 		breq.ifbr_ifsflags = bif->bif_flags;
1387 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1388 		memcpy(buf, &breq, sizeof(breq));
1389 		count++;
1390 		buf += sizeof(breq);
1391 		len -= sizeof(breq);
1392 	}
1393 
1394 	BRIDGE_UNLOCK(sc);
1395 	bifc->ifbic_len = sizeof(breq) * count;
1396 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1397 	BRIDGE_LOCK(sc);
1398 	free(outbuf, M_TEMP);
1399 	return (error);
1400 }
1401 
1402 static int
1403 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1404 {
1405 	struct ifbaconf *bac = arg;
1406 	struct bridge_rtnode *brt;
1407 	struct ifbareq bareq;
1408 	char *buf, *outbuf;
1409 	int count, buflen, len, error = 0;
1410 
1411 	if (bac->ifbac_len == 0)
1412 		return (0);
1413 
1414 	count = 0;
1415 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1416 		count++;
1417 	buflen = sizeof(bareq) * count;
1418 
1419 	BRIDGE_UNLOCK(sc);
1420 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1421 	BRIDGE_LOCK(sc);
1422 
1423 	count = 0;
1424 	buf = outbuf;
1425 	len = min(bac->ifbac_len, buflen);
1426 	bzero(&bareq, sizeof(bareq));
1427 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1428 		if (len < sizeof(bareq))
1429 			goto out;
1430 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1431 		    sizeof(bareq.ifba_ifsname));
1432 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1433 		bareq.ifba_vlan = brt->brt_vlan;
1434 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1435 				time_uptime < brt->brt_expire)
1436 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1437 		else
1438 			bareq.ifba_expire = 0;
1439 		bareq.ifba_flags = brt->brt_flags;
1440 
1441 		memcpy(buf, &bareq, sizeof(bareq));
1442 		count++;
1443 		buf += sizeof(bareq);
1444 		len -= sizeof(bareq);
1445 	}
1446 out:
1447 	BRIDGE_UNLOCK(sc);
1448 	bac->ifbac_len = sizeof(bareq) * count;
1449 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1450 	BRIDGE_LOCK(sc);
1451 	free(outbuf, M_TEMP);
1452 	return (error);
1453 }
1454 
1455 static int
1456 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1457 {
1458 	struct ifbareq *req = arg;
1459 	struct bridge_iflist *bif;
1460 	int error;
1461 
1462 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1463 	if (bif == NULL)
1464 		return (ENOENT);
1465 
1466 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1467 	    req->ifba_flags);
1468 
1469 	return (error);
1470 }
1471 
1472 static int
1473 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1474 {
1475 	struct ifbrparam *param = arg;
1476 
1477 	sc->sc_brttimeout = param->ifbrp_ctime;
1478 	return (0);
1479 }
1480 
1481 static int
1482 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1483 {
1484 	struct ifbrparam *param = arg;
1485 
1486 	param->ifbrp_ctime = sc->sc_brttimeout;
1487 	return (0);
1488 }
1489 
1490 static int
1491 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1492 {
1493 	struct ifbareq *req = arg;
1494 
1495 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1496 }
1497 
1498 static int
1499 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1500 {
1501 	struct ifbreq *req = arg;
1502 
1503 	bridge_rtflush(sc, req->ifbr_ifsflags);
1504 	return (0);
1505 }
1506 
1507 static int
1508 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1509 {
1510 	struct ifbrparam *param = arg;
1511 	struct bstp_state *bs = &sc->sc_stp;
1512 
1513 	param->ifbrp_prio = bs->bs_bridge_priority;
1514 	return (0);
1515 }
1516 
1517 static int
1518 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1519 {
1520 	struct ifbrparam *param = arg;
1521 
1522 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1523 }
1524 
1525 static int
1526 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1527 {
1528 	struct ifbrparam *param = arg;
1529 	struct bstp_state *bs = &sc->sc_stp;
1530 
1531 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1532 	return (0);
1533 }
1534 
1535 static int
1536 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1537 {
1538 	struct ifbrparam *param = arg;
1539 
1540 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1541 }
1542 
1543 static int
1544 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1545 {
1546 	struct ifbrparam *param = arg;
1547 	struct bstp_state *bs = &sc->sc_stp;
1548 
1549 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1550 	return (0);
1551 }
1552 
1553 static int
1554 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1555 {
1556 	struct ifbrparam *param = arg;
1557 
1558 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1559 }
1560 
1561 static int
1562 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1563 {
1564 	struct ifbrparam *param = arg;
1565 	struct bstp_state *bs = &sc->sc_stp;
1566 
1567 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1568 	return (0);
1569 }
1570 
1571 static int
1572 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1573 {
1574 	struct ifbrparam *param = arg;
1575 
1576 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1577 }
1578 
1579 static int
1580 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1581 {
1582 	struct ifbreq *req = arg;
1583 	struct bridge_iflist *bif;
1584 
1585 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1586 	if (bif == NULL)
1587 		return (ENOENT);
1588 
1589 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1590 }
1591 
1592 static int
1593 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1594 {
1595 	struct ifbreq *req = arg;
1596 	struct bridge_iflist *bif;
1597 
1598 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1599 	if (bif == NULL)
1600 		return (ENOENT);
1601 
1602 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1603 }
1604 
1605 static int
1606 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1607 {
1608 	struct ifbreq *req = arg;
1609 	struct bridge_iflist *bif;
1610 
1611 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1612 	if (bif == NULL)
1613 		return (ENOENT);
1614 
1615 	bif->bif_addrmax = req->ifbr_addrmax;
1616 	return (0);
1617 }
1618 
1619 static int
1620 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1621 {
1622 	struct ifbreq *req = arg;
1623 	struct bridge_iflist *bif = NULL;
1624 	struct ifnet *ifs;
1625 
1626 	ifs = ifunit(req->ifbr_ifsname);
1627 	if (ifs == NULL)
1628 		return (ENOENT);
1629 
1630 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1631 		if (ifs == bif->bif_ifp)
1632 			return (EBUSY);
1633 
1634 	if (ifs->if_bridge != NULL)
1635 		return (EBUSY);
1636 
1637 	switch (ifs->if_type) {
1638 		case IFT_ETHER:
1639 		case IFT_GIF:
1640 		case IFT_L2VLAN:
1641 			break;
1642 		default:
1643 			return (EINVAL);
1644 	}
1645 
1646 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1647 	if (bif == NULL)
1648 		return (ENOMEM);
1649 
1650 	bif->bif_ifp = ifs;
1651 	bif->bif_flags = IFBIF_SPAN;
1652 
1653 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1654 
1655 	return (0);
1656 }
1657 
1658 static int
1659 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1660 {
1661 	struct ifbreq *req = arg;
1662 	struct bridge_iflist *bif;
1663 	struct ifnet *ifs;
1664 
1665 	ifs = ifunit(req->ifbr_ifsname);
1666 	if (ifs == NULL)
1667 		return (ENOENT);
1668 
1669 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1670 		if (ifs == bif->bif_ifp)
1671 			break;
1672 
1673 	if (bif == NULL)
1674 		return (ENOENT);
1675 
1676 	bridge_delete_span(sc, bif);
1677 
1678 	return (0);
1679 }
1680 
1681 static int
1682 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1683 {
1684 	struct ifbropreq *req = arg;
1685 	struct bstp_state *bs = &sc->sc_stp;
1686 	struct bstp_port *root_port;
1687 
1688 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1689 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1690 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1691 
1692 	root_port = bs->bs_root_port;
1693 	if (root_port == NULL)
1694 		req->ifbop_root_port = 0;
1695 	else
1696 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1697 
1698 	req->ifbop_holdcount = bs->bs_txholdcount;
1699 	req->ifbop_priority = bs->bs_bridge_priority;
1700 	req->ifbop_protocol = bs->bs_protover;
1701 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1702 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1703 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1704 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1705 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1706 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1707 
1708 	return (0);
1709 }
1710 
1711 static int
1712 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1713 {
1714 	struct ifbrparam *param = arg;
1715 
1716 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1717 	return (0);
1718 }
1719 
1720 static int
1721 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1722 {
1723 	struct ifbpstpconf *bifstp = arg;
1724 	struct bridge_iflist *bif;
1725 	struct bstp_port *bp;
1726 	struct ifbpstpreq bpreq;
1727 	char *buf, *outbuf;
1728 	int count, buflen, len, error = 0;
1729 
1730 	count = 0;
1731 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1732 		if ((bif->bif_flags & IFBIF_STP) != 0)
1733 			count++;
1734 	}
1735 
1736 	buflen = sizeof(bpreq) * count;
1737 	if (bifstp->ifbpstp_len == 0) {
1738 		bifstp->ifbpstp_len = buflen;
1739 		return (0);
1740 	}
1741 
1742 	BRIDGE_UNLOCK(sc);
1743 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1744 	BRIDGE_LOCK(sc);
1745 
1746 	count = 0;
1747 	buf = outbuf;
1748 	len = min(bifstp->ifbpstp_len, buflen);
1749 	bzero(&bpreq, sizeof(bpreq));
1750 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1751 		if (len < sizeof(bpreq))
1752 			break;
1753 
1754 		if ((bif->bif_flags & IFBIF_STP) == 0)
1755 			continue;
1756 
1757 		bp = &bif->bif_stp;
1758 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1759 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1760 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1761 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1762 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1763 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1764 
1765 		memcpy(buf, &bpreq, sizeof(bpreq));
1766 		count++;
1767 		buf += sizeof(bpreq);
1768 		len -= sizeof(bpreq);
1769 	}
1770 
1771 	BRIDGE_UNLOCK(sc);
1772 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1773 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1774 	BRIDGE_LOCK(sc);
1775 	free(outbuf, M_TEMP);
1776 	return (error);
1777 }
1778 
1779 static int
1780 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1781 {
1782 	struct ifbrparam *param = arg;
1783 
1784 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1785 }
1786 
1787 static int
1788 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1789 {
1790 	struct ifbrparam *param = arg;
1791 
1792 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1793 }
1794 
1795 /*
1796  * bridge_ifdetach:
1797  *
1798  *	Detach an interface from a bridge.  Called when a member
1799  *	interface is detaching.
1800  */
1801 static void
1802 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1803 {
1804 	struct bridge_softc *sc = ifp->if_bridge;
1805 	struct bridge_iflist *bif;
1806 
1807 	if (ifp->if_flags & IFF_RENAMING)
1808 		return;
1809 	if (V_bridge_cloner == NULL) {
1810 		/*
1811 		 * This detach handler can be called after
1812 		 * vnet_bridge_uninit().  Just return in that case.
1813 		 */
1814 		return;
1815 	}
1816 	/* Check if the interface is a bridge member */
1817 	if (sc != NULL) {
1818 		BRIDGE_LOCK(sc);
1819 
1820 		bif = bridge_lookup_member_if(sc, ifp);
1821 		if (bif != NULL)
1822 			bridge_delete_member(sc, bif, 1);
1823 
1824 		BRIDGE_UNLOCK(sc);
1825 		return;
1826 	}
1827 
1828 	/* Check if the interface is a span port */
1829 	BRIDGE_LIST_LOCK();
1830 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1831 		BRIDGE_LOCK(sc);
1832 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1833 			if (ifp == bif->bif_ifp) {
1834 				bridge_delete_span(sc, bif);
1835 				break;
1836 			}
1837 
1838 		BRIDGE_UNLOCK(sc);
1839 	}
1840 	BRIDGE_LIST_UNLOCK();
1841 }
1842 
1843 /*
1844  * bridge_init:
1845  *
1846  *	Initialize a bridge interface.
1847  */
1848 static void
1849 bridge_init(void *xsc)
1850 {
1851 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1852 	struct ifnet *ifp = sc->sc_ifp;
1853 
1854 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1855 		return;
1856 
1857 	BRIDGE_LOCK(sc);
1858 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1859 	    bridge_timer, sc);
1860 
1861 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1862 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1863 
1864 	BRIDGE_UNLOCK(sc);
1865 }
1866 
1867 /*
1868  * bridge_stop:
1869  *
1870  *	Stop the bridge interface.
1871  */
1872 static void
1873 bridge_stop(struct ifnet *ifp, int disable)
1874 {
1875 	struct bridge_softc *sc = ifp->if_softc;
1876 
1877 	BRIDGE_LOCK_ASSERT(sc);
1878 
1879 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1880 		return;
1881 
1882 	callout_stop(&sc->sc_brcallout);
1883 	bstp_stop(&sc->sc_stp);
1884 
1885 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1886 
1887 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1888 }
1889 
1890 /*
1891  * bridge_enqueue:
1892  *
1893  *	Enqueue a packet on a bridge member interface.
1894  *
1895  */
1896 static int
1897 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1898 {
1899 	int len, err = 0;
1900 	short mflags;
1901 	struct mbuf *m0;
1902 
1903 	/* We may be sending a fragment so traverse the mbuf */
1904 	for (; m; m = m0) {
1905 		m0 = m->m_nextpkt;
1906 		m->m_nextpkt = NULL;
1907 		len = m->m_pkthdr.len;
1908 		mflags = m->m_flags;
1909 
1910 		/*
1911 		 * If underlying interface can not do VLAN tag insertion itself
1912 		 * then attach a packet tag that holds it.
1913 		 */
1914 		if ((m->m_flags & M_VLANTAG) &&
1915 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1916 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1917 			if (m == NULL) {
1918 				if_printf(dst_ifp,
1919 				    "unable to prepend VLAN header\n");
1920 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1921 				continue;
1922 			}
1923 			m->m_flags &= ~M_VLANTAG;
1924 		}
1925 
1926 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1927 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1928 			m_freem(m0);
1929 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1930 			break;
1931 		}
1932 
1933 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1934 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1935 		if (mflags & M_MCAST)
1936 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
1937 	}
1938 
1939 	return (err);
1940 }
1941 
1942 /*
1943  * bridge_dummynet:
1944  *
1945  * 	Receive a queued packet from dummynet and pass it on to the output
1946  * 	interface.
1947  *
1948  *	The mbuf has the Ethernet header already attached.
1949  */
1950 static void
1951 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1952 {
1953 	struct bridge_softc *sc;
1954 
1955 	sc = ifp->if_bridge;
1956 
1957 	/*
1958 	 * The packet didnt originate from a member interface. This should only
1959 	 * ever happen if a member interface is removed while packets are
1960 	 * queued for it.
1961 	 */
1962 	if (sc == NULL) {
1963 		m_freem(m);
1964 		return;
1965 	}
1966 
1967 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1968 #ifdef INET6
1969 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1970 #endif
1971 	    ) {
1972 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1973 			return;
1974 		if (m == NULL)
1975 			return;
1976 	}
1977 
1978 	bridge_enqueue(sc, ifp, m);
1979 }
1980 
1981 /*
1982  * bridge_output:
1983  *
1984  *	Send output from a bridge member interface.  This
1985  *	performs the bridging function for locally originated
1986  *	packets.
1987  *
1988  *	The mbuf has the Ethernet header already attached.  We must
1989  *	enqueue or free the mbuf before returning.
1990  */
1991 static int
1992 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1993     struct rtentry *rt)
1994 {
1995 	struct ether_header *eh;
1996 	struct ifnet *dst_if;
1997 	struct bridge_softc *sc;
1998 	uint16_t vlan;
1999 
2000 	if (m->m_len < ETHER_HDR_LEN) {
2001 		m = m_pullup(m, ETHER_HDR_LEN);
2002 		if (m == NULL)
2003 			return (0);
2004 	}
2005 
2006 	eh = mtod(m, struct ether_header *);
2007 	sc = ifp->if_bridge;
2008 	vlan = VLANTAGOF(m);
2009 
2010 	BRIDGE_LOCK(sc);
2011 
2012 	/*
2013 	 * If bridge is down, but the original output interface is up,
2014 	 * go ahead and send out that interface.  Otherwise, the packet
2015 	 * is dropped below.
2016 	 */
2017 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2018 		dst_if = ifp;
2019 		goto sendunicast;
2020 	}
2021 
2022 	/*
2023 	 * If the packet is a multicast, or we don't know a better way to
2024 	 * get there, send to all interfaces.
2025 	 */
2026 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2027 		dst_if = NULL;
2028 	else
2029 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2030 	if (dst_if == NULL) {
2031 		struct bridge_iflist *bif;
2032 		struct mbuf *mc;
2033 		int error = 0, used = 0;
2034 
2035 		bridge_span(sc, m);
2036 
2037 		BRIDGE_LOCK2REF(sc, error);
2038 		if (error) {
2039 			m_freem(m);
2040 			return (0);
2041 		}
2042 
2043 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2044 			dst_if = bif->bif_ifp;
2045 
2046 			if (dst_if->if_type == IFT_GIF)
2047 				continue;
2048 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2049 				continue;
2050 
2051 			/*
2052 			 * If this is not the original output interface,
2053 			 * and the interface is participating in spanning
2054 			 * tree, make sure the port is in a state that
2055 			 * allows forwarding.
2056 			 */
2057 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2058 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2059 				continue;
2060 
2061 			if (LIST_NEXT(bif, bif_next) == NULL) {
2062 				used = 1;
2063 				mc = m;
2064 			} else {
2065 				mc = m_copypacket(m, M_NOWAIT);
2066 				if (mc == NULL) {
2067 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2068 					continue;
2069 				}
2070 			}
2071 
2072 			bridge_enqueue(sc, dst_if, mc);
2073 		}
2074 		if (used == 0)
2075 			m_freem(m);
2076 		BRIDGE_UNREF(sc);
2077 		return (0);
2078 	}
2079 
2080 sendunicast:
2081 	/*
2082 	 * XXX Spanning tree consideration here?
2083 	 */
2084 
2085 	bridge_span(sc, m);
2086 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2087 		m_freem(m);
2088 		BRIDGE_UNLOCK(sc);
2089 		return (0);
2090 	}
2091 
2092 	BRIDGE_UNLOCK(sc);
2093 	bridge_enqueue(sc, dst_if, m);
2094 	return (0);
2095 }
2096 
2097 /*
2098  * bridge_transmit:
2099  *
2100  *	Do output on a bridge.
2101  *
2102  */
2103 static int
2104 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2105 {
2106 	struct bridge_softc *sc;
2107 	struct ether_header *eh;
2108 	struct ifnet *dst_if;
2109 	int error = 0;
2110 
2111 	sc = ifp->if_softc;
2112 
2113 	ETHER_BPF_MTAP(ifp, m);
2114 
2115 	eh = mtod(m, struct ether_header *);
2116 
2117 	BRIDGE_LOCK(sc);
2118 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2119 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2120 		BRIDGE_UNLOCK(sc);
2121 		error = bridge_enqueue(sc, dst_if, m);
2122 	} else
2123 		bridge_broadcast(sc, ifp, m, 0);
2124 
2125 	return (error);
2126 }
2127 
2128 /*
2129  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2130  */
2131 static void
2132 bridge_qflush(struct ifnet *ifp __unused)
2133 {
2134 }
2135 
2136 /*
2137  * bridge_forward:
2138  *
2139  *	The forwarding function of the bridge.
2140  *
2141  *	NOTE: Releases the lock on return.
2142  */
2143 static void
2144 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2145     struct mbuf *m)
2146 {
2147 	struct bridge_iflist *dbif;
2148 	struct ifnet *src_if, *dst_if, *ifp;
2149 	struct ether_header *eh;
2150 	uint16_t vlan;
2151 	uint8_t *dst;
2152 	int error;
2153 
2154 	src_if = m->m_pkthdr.rcvif;
2155 	ifp = sc->sc_ifp;
2156 
2157 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2158 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2159 	vlan = VLANTAGOF(m);
2160 
2161 	if ((sbif->bif_flags & IFBIF_STP) &&
2162 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2163 		goto drop;
2164 
2165 	eh = mtod(m, struct ether_header *);
2166 	dst = eh->ether_dhost;
2167 
2168 	/* If the interface is learning, record the address. */
2169 	if (sbif->bif_flags & IFBIF_LEARNING) {
2170 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2171 		    sbif, 0, IFBAF_DYNAMIC);
2172 		/*
2173 		 * If the interface has addresses limits then deny any source
2174 		 * that is not in the cache.
2175 		 */
2176 		if (error && sbif->bif_addrmax)
2177 			goto drop;
2178 	}
2179 
2180 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2181 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2182 		goto drop;
2183 
2184 	/*
2185 	 * At this point, the port either doesn't participate
2186 	 * in spanning tree or it is in the forwarding state.
2187 	 */
2188 
2189 	/*
2190 	 * If the packet is unicast, destined for someone on
2191 	 * "this" side of the bridge, drop it.
2192 	 */
2193 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2194 		dst_if = bridge_rtlookup(sc, dst, vlan);
2195 		if (src_if == dst_if)
2196 			goto drop;
2197 	} else {
2198 		/*
2199 		 * Check if its a reserved multicast address, any address
2200 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2201 		 * bridge.
2202 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2203 		 */
2204 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2205 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2206 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2207 			goto drop;
2208 
2209 		/* ...forward it to all interfaces. */
2210 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2211 		dst_if = NULL;
2212 	}
2213 
2214 	/*
2215 	 * If we have a destination interface which is a member of our bridge,
2216 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2217 	 * For broadcast or multicast packets, don't bother because it will
2218 	 * be reinjected into ether_input. We do this before we pass the packets
2219 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2220 	 * drop the packet, or possibly modify it, making it difficult to debug
2221 	 * firewall issues on the bridge.
2222 	 */
2223 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2224 		ETHER_BPF_MTAP(ifp, m);
2225 
2226 	/* run the packet filter */
2227 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2228 #ifdef INET6
2229 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2230 #endif
2231 	    ) {
2232 		BRIDGE_UNLOCK(sc);
2233 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2234 			return;
2235 		if (m == NULL)
2236 			return;
2237 		BRIDGE_LOCK(sc);
2238 	}
2239 
2240 	if (dst_if == NULL) {
2241 		bridge_broadcast(sc, src_if, m, 1);
2242 		return;
2243 	}
2244 
2245 	/*
2246 	 * At this point, we're dealing with a unicast frame
2247 	 * going to a different interface.
2248 	 */
2249 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2250 		goto drop;
2251 
2252 	dbif = bridge_lookup_member_if(sc, dst_if);
2253 	if (dbif == NULL)
2254 		/* Not a member of the bridge (anymore?) */
2255 		goto drop;
2256 
2257 	/* Private segments can not talk to each other */
2258 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2259 		goto drop;
2260 
2261 	if ((dbif->bif_flags & IFBIF_STP) &&
2262 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2263 		goto drop;
2264 
2265 	BRIDGE_UNLOCK(sc);
2266 
2267 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2268 #ifdef INET6
2269 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2270 #endif
2271 	    ) {
2272 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2273 			return;
2274 		if (m == NULL)
2275 			return;
2276 	}
2277 
2278 	bridge_enqueue(sc, dst_if, m);
2279 	return;
2280 
2281 drop:
2282 	BRIDGE_UNLOCK(sc);
2283 	m_freem(m);
2284 }
2285 
2286 /*
2287  * bridge_input:
2288  *
2289  *	Receive input from a member interface.  Queue the packet for
2290  *	bridging if it is not for us.
2291  */
2292 static struct mbuf *
2293 bridge_input(struct ifnet *ifp, struct mbuf *m)
2294 {
2295 	struct bridge_softc *sc = ifp->if_bridge;
2296 	struct bridge_iflist *bif, *bif2;
2297 	struct ifnet *bifp;
2298 	struct ether_header *eh;
2299 	struct mbuf *mc, *mc2;
2300 	uint16_t vlan;
2301 	int error;
2302 
2303 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2304 		return (m);
2305 
2306 	bifp = sc->sc_ifp;
2307 	vlan = VLANTAGOF(m);
2308 
2309 	/*
2310 	 * Implement support for bridge monitoring. If this flag has been
2311 	 * set on this interface, discard the packet once we push it through
2312 	 * the bpf(4) machinery, but before we do, increment the byte and
2313 	 * packet counters associated with this interface.
2314 	 */
2315 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2316 		m->m_pkthdr.rcvif  = bifp;
2317 		ETHER_BPF_MTAP(bifp, m);
2318 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2319 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2320 		m_freem(m);
2321 		return (NULL);
2322 	}
2323 	BRIDGE_LOCK(sc);
2324 	bif = bridge_lookup_member_if(sc, ifp);
2325 	if (bif == NULL) {
2326 		BRIDGE_UNLOCK(sc);
2327 		return (m);
2328 	}
2329 
2330 	eh = mtod(m, struct ether_header *);
2331 
2332 	bridge_span(sc, m);
2333 
2334 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2335 		/* Tap off 802.1D packets; they do not get forwarded. */
2336 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2337 		    ETHER_ADDR_LEN) == 0) {
2338 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2339 			BRIDGE_UNLOCK(sc);
2340 			return (NULL);
2341 		}
2342 
2343 		if ((bif->bif_flags & IFBIF_STP) &&
2344 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2345 			BRIDGE_UNLOCK(sc);
2346 			return (m);
2347 		}
2348 
2349 		/*
2350 		 * Make a deep copy of the packet and enqueue the copy
2351 		 * for bridge processing; return the original packet for
2352 		 * local processing.
2353 		 */
2354 		mc = m_dup(m, M_NOWAIT);
2355 		if (mc == NULL) {
2356 			BRIDGE_UNLOCK(sc);
2357 			return (m);
2358 		}
2359 
2360 		/* Perform the bridge forwarding function with the copy. */
2361 		bridge_forward(sc, bif, mc);
2362 
2363 		/*
2364 		 * Reinject the mbuf as arriving on the bridge so we have a
2365 		 * chance at claiming multicast packets. We can not loop back
2366 		 * here from ether_input as a bridge is never a member of a
2367 		 * bridge.
2368 		 */
2369 		KASSERT(bifp->if_bridge == NULL,
2370 		    ("loop created in bridge_input"));
2371 		mc2 = m_dup(m, M_NOWAIT);
2372 		if (mc2 != NULL) {
2373 			/* Keep the layer3 header aligned */
2374 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2375 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2376 		}
2377 		if (mc2 != NULL) {
2378 			mc2->m_pkthdr.rcvif = bifp;
2379 			(*bifp->if_input)(bifp, mc2);
2380 		}
2381 
2382 		/* Return the original packet for local processing. */
2383 		return (m);
2384 	}
2385 
2386 	if ((bif->bif_flags & IFBIF_STP) &&
2387 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2388 		BRIDGE_UNLOCK(sc);
2389 		return (m);
2390 	}
2391 
2392 #if (defined(INET) || defined(INET6))
2393 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2394 	|| ((iface)->if_carp \
2395 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2396 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2397 	|| ((iface)->if_carp \
2398 	    && (*carp_forus_p)((iface), eh->ether_shost))
2399 #else
2400 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2401 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2402 #endif
2403 
2404 #ifdef INET6
2405 #   define OR_PFIL_HOOKED_INET6 \
2406 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2407 #else
2408 #   define OR_PFIL_HOOKED_INET6
2409 #endif
2410 
2411 #define GRAB_OUR_PACKETS(iface) \
2412 	if ((iface)->if_type == IFT_GIF) \
2413 		continue; \
2414 	/* It is destined for us. */ \
2415 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2416 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2417 	    ) {								\
2418 		if ((iface)->if_type == IFT_BRIDGE) {			\
2419 			ETHER_BPF_MTAP(iface, m);			\
2420 			if_inc_counter(iface, IFCOUNTER_IPACKETS, 1);				\
2421 			if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len);		\
2422 			/* Filter on the physical interface. */		\
2423 			if (V_pfil_local_phys &&			\
2424 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2425 			     OR_PFIL_HOOKED_INET6)) {			\
2426 				if (bridge_pfil(&m, NULL, ifp,		\
2427 				    PFIL_IN) != 0 || m == NULL) {	\
2428 					BRIDGE_UNLOCK(sc);		\
2429 					return (NULL);			\
2430 				}					\
2431 				eh = mtod(m, struct ether_header *);	\
2432 			}						\
2433 		}							\
2434 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2435 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2436 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2437 			if (error && bif->bif_addrmax) {		\
2438 				BRIDGE_UNLOCK(sc);			\
2439 				m_freem(m);				\
2440 				return (NULL);				\
2441 			}						\
2442 		}							\
2443 		m->m_pkthdr.rcvif = iface;				\
2444 		BRIDGE_UNLOCK(sc);					\
2445 		return (m);						\
2446 	}								\
2447 									\
2448 	/* We just received a packet that we sent out. */		\
2449 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2450 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2451 	    ) {								\
2452 		BRIDGE_UNLOCK(sc);					\
2453 		m_freem(m);						\
2454 		return (NULL);						\
2455 	}
2456 
2457 	/*
2458 	 * Unicast.  Make sure it's not for the bridge.
2459 	 */
2460 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2461 
2462 	/*
2463 	 * Give a chance for ifp at first priority. This will help when	the
2464 	 * packet comes through the interface like VLAN's with the same MACs
2465 	 * on several interfaces from the same bridge. This also will save
2466 	 * some CPU cycles in case the destination interface and the input
2467 	 * interface (eq ifp) are the same.
2468 	 */
2469 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2470 
2471 	/* Now check the all bridge members. */
2472 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2473 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2474 	}
2475 
2476 #undef OR_CARP_CHECK_WE_ARE_DST
2477 #undef OR_CARP_CHECK_WE_ARE_SRC
2478 #undef OR_PFIL_HOOKED_INET6
2479 #undef GRAB_OUR_PACKETS
2480 
2481 	/* Perform the bridge forwarding function. */
2482 	bridge_forward(sc, bif, m);
2483 
2484 	return (NULL);
2485 }
2486 
2487 /*
2488  * bridge_broadcast:
2489  *
2490  *	Send a frame to all interfaces that are members of
2491  *	the bridge, except for the one on which the packet
2492  *	arrived.
2493  *
2494  *	NOTE: Releases the lock on return.
2495  */
2496 static void
2497 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2498     struct mbuf *m, int runfilt)
2499 {
2500 	struct bridge_iflist *dbif, *sbif;
2501 	struct mbuf *mc;
2502 	struct ifnet *dst_if;
2503 	int error = 0, used = 0, i;
2504 
2505 	sbif = bridge_lookup_member_if(sc, src_if);
2506 
2507 	BRIDGE_LOCK2REF(sc, error);
2508 	if (error) {
2509 		m_freem(m);
2510 		return;
2511 	}
2512 
2513 	/* Filter on the bridge interface before broadcasting */
2514 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2515 #ifdef INET6
2516 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2517 #endif
2518 	    )) {
2519 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2520 			goto out;
2521 		if (m == NULL)
2522 			goto out;
2523 	}
2524 
2525 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2526 		dst_if = dbif->bif_ifp;
2527 		if (dst_if == src_if)
2528 			continue;
2529 
2530 		/* Private segments can not talk to each other */
2531 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2532 			continue;
2533 
2534 		if ((dbif->bif_flags & IFBIF_STP) &&
2535 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2536 			continue;
2537 
2538 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2539 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2540 			continue;
2541 
2542 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2543 			continue;
2544 
2545 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2546 			mc = m;
2547 			used = 1;
2548 		} else {
2549 			mc = m_dup(m, M_NOWAIT);
2550 			if (mc == NULL) {
2551 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2552 				continue;
2553 			}
2554 		}
2555 
2556 		/*
2557 		 * Filter on the output interface. Pass a NULL bridge interface
2558 		 * pointer so we do not redundantly filter on the bridge for
2559 		 * each interface we broadcast on.
2560 		 */
2561 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2562 #ifdef INET6
2563 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2564 #endif
2565 		    )) {
2566 			if (used == 0) {
2567 				/* Keep the layer3 header aligned */
2568 				i = min(mc->m_pkthdr.len, max_protohdr);
2569 				mc = m_copyup(mc, i, ETHER_ALIGN);
2570 				if (mc == NULL) {
2571 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2572 					continue;
2573 				}
2574 			}
2575 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2576 				continue;
2577 			if (mc == NULL)
2578 				continue;
2579 		}
2580 
2581 		bridge_enqueue(sc, dst_if, mc);
2582 	}
2583 	if (used == 0)
2584 		m_freem(m);
2585 
2586 out:
2587 	BRIDGE_UNREF(sc);
2588 }
2589 
2590 /*
2591  * bridge_span:
2592  *
2593  *	Duplicate a packet out one or more interfaces that are in span mode,
2594  *	the original mbuf is unmodified.
2595  */
2596 static void
2597 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2598 {
2599 	struct bridge_iflist *bif;
2600 	struct ifnet *dst_if;
2601 	struct mbuf *mc;
2602 
2603 	if (LIST_EMPTY(&sc->sc_spanlist))
2604 		return;
2605 
2606 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2607 		dst_if = bif->bif_ifp;
2608 
2609 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2610 			continue;
2611 
2612 		mc = m_copypacket(m, M_NOWAIT);
2613 		if (mc == NULL) {
2614 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2615 			continue;
2616 		}
2617 
2618 		bridge_enqueue(sc, dst_if, mc);
2619 	}
2620 }
2621 
2622 /*
2623  * bridge_rtupdate:
2624  *
2625  *	Add a bridge routing entry.
2626  */
2627 static int
2628 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2629     struct bridge_iflist *bif, int setflags, uint8_t flags)
2630 {
2631 	struct bridge_rtnode *brt;
2632 	int error;
2633 
2634 	BRIDGE_LOCK_ASSERT(sc);
2635 
2636 	/* Check the source address is valid and not multicast. */
2637 	if (ETHER_IS_MULTICAST(dst) ||
2638 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2639 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2640 		return (EINVAL);
2641 
2642 	/* 802.1p frames map to vlan 1 */
2643 	if (vlan == 0)
2644 		vlan = 1;
2645 
2646 	/*
2647 	 * A route for this destination might already exist.  If so,
2648 	 * update it, otherwise create a new one.
2649 	 */
2650 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2651 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2652 			sc->sc_brtexceeded++;
2653 			return (ENOSPC);
2654 		}
2655 		/* Check per interface address limits (if enabled) */
2656 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2657 			bif->bif_addrexceeded++;
2658 			return (ENOSPC);
2659 		}
2660 
2661 		/*
2662 		 * Allocate a new bridge forwarding node, and
2663 		 * initialize the expiration time and Ethernet
2664 		 * address.
2665 		 */
2666 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2667 		if (brt == NULL)
2668 			return (ENOMEM);
2669 
2670 		if (bif->bif_flags & IFBIF_STICKY)
2671 			brt->brt_flags = IFBAF_STICKY;
2672 		else
2673 			brt->brt_flags = IFBAF_DYNAMIC;
2674 
2675 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2676 		brt->brt_vlan = vlan;
2677 
2678 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2679 			uma_zfree(bridge_rtnode_zone, brt);
2680 			return (error);
2681 		}
2682 		brt->brt_dst = bif;
2683 		bif->bif_addrcnt++;
2684 	}
2685 
2686 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2687 	    brt->brt_dst != bif) {
2688 		brt->brt_dst->bif_addrcnt--;
2689 		brt->brt_dst = bif;
2690 		brt->brt_dst->bif_addrcnt++;
2691 	}
2692 
2693 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2694 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2695 	if (setflags)
2696 		brt->brt_flags = flags;
2697 
2698 	return (0);
2699 }
2700 
2701 /*
2702  * bridge_rtlookup:
2703  *
2704  *	Lookup the destination interface for an address.
2705  */
2706 static struct ifnet *
2707 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2708 {
2709 	struct bridge_rtnode *brt;
2710 
2711 	BRIDGE_LOCK_ASSERT(sc);
2712 
2713 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2714 		return (NULL);
2715 
2716 	return (brt->brt_ifp);
2717 }
2718 
2719 /*
2720  * bridge_rttrim:
2721  *
2722  *	Trim the routine table so that we have a number
2723  *	of routing entries less than or equal to the
2724  *	maximum number.
2725  */
2726 static void
2727 bridge_rttrim(struct bridge_softc *sc)
2728 {
2729 	struct bridge_rtnode *brt, *nbrt;
2730 
2731 	BRIDGE_LOCK_ASSERT(sc);
2732 
2733 	/* Make sure we actually need to do this. */
2734 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2735 		return;
2736 
2737 	/* Force an aging cycle; this might trim enough addresses. */
2738 	bridge_rtage(sc);
2739 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2740 		return;
2741 
2742 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2743 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2744 			bridge_rtnode_destroy(sc, brt);
2745 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2746 				return;
2747 		}
2748 	}
2749 }
2750 
2751 /*
2752  * bridge_timer:
2753  *
2754  *	Aging timer for the bridge.
2755  */
2756 static void
2757 bridge_timer(void *arg)
2758 {
2759 	struct bridge_softc *sc = arg;
2760 
2761 	BRIDGE_LOCK_ASSERT(sc);
2762 
2763 	bridge_rtage(sc);
2764 
2765 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2766 		callout_reset(&sc->sc_brcallout,
2767 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2768 }
2769 
2770 /*
2771  * bridge_rtage:
2772  *
2773  *	Perform an aging cycle.
2774  */
2775 static void
2776 bridge_rtage(struct bridge_softc *sc)
2777 {
2778 	struct bridge_rtnode *brt, *nbrt;
2779 
2780 	BRIDGE_LOCK_ASSERT(sc);
2781 
2782 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2783 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2784 			if (time_uptime >= brt->brt_expire)
2785 				bridge_rtnode_destroy(sc, brt);
2786 		}
2787 	}
2788 }
2789 
2790 /*
2791  * bridge_rtflush:
2792  *
2793  *	Remove all dynamic addresses from the bridge.
2794  */
2795 static void
2796 bridge_rtflush(struct bridge_softc *sc, int full)
2797 {
2798 	struct bridge_rtnode *brt, *nbrt;
2799 
2800 	BRIDGE_LOCK_ASSERT(sc);
2801 
2802 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2803 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2804 			bridge_rtnode_destroy(sc, brt);
2805 	}
2806 }
2807 
2808 /*
2809  * bridge_rtdaddr:
2810  *
2811  *	Remove an address from the table.
2812  */
2813 static int
2814 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2815 {
2816 	struct bridge_rtnode *brt;
2817 	int found = 0;
2818 
2819 	BRIDGE_LOCK_ASSERT(sc);
2820 
2821 	/*
2822 	 * If vlan is zero then we want to delete for all vlans so the lookup
2823 	 * may return more than one.
2824 	 */
2825 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2826 		bridge_rtnode_destroy(sc, brt);
2827 		found = 1;
2828 	}
2829 
2830 	return (found ? 0 : ENOENT);
2831 }
2832 
2833 /*
2834  * bridge_rtdelete:
2835  *
2836  *	Delete routes to a speicifc member interface.
2837  */
2838 static void
2839 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2840 {
2841 	struct bridge_rtnode *brt, *nbrt;
2842 
2843 	BRIDGE_LOCK_ASSERT(sc);
2844 
2845 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2846 		if (brt->brt_ifp == ifp && (full ||
2847 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2848 			bridge_rtnode_destroy(sc, brt);
2849 	}
2850 }
2851 
2852 /*
2853  * bridge_rtable_init:
2854  *
2855  *	Initialize the route table for this bridge.
2856  */
2857 static void
2858 bridge_rtable_init(struct bridge_softc *sc)
2859 {
2860 	int i;
2861 
2862 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2863 	    M_DEVBUF, M_WAITOK);
2864 
2865 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2866 		LIST_INIT(&sc->sc_rthash[i]);
2867 
2868 	sc->sc_rthash_key = arc4random();
2869 	LIST_INIT(&sc->sc_rtlist);
2870 }
2871 
2872 /*
2873  * bridge_rtable_fini:
2874  *
2875  *	Deconstruct the route table for this bridge.
2876  */
2877 static void
2878 bridge_rtable_fini(struct bridge_softc *sc)
2879 {
2880 
2881 	KASSERT(sc->sc_brtcnt == 0,
2882 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2883 	free(sc->sc_rthash, M_DEVBUF);
2884 }
2885 
2886 /*
2887  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2888  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2889  */
2890 #define	mix(a, b, c)							\
2891 do {									\
2892 	a -= b; a -= c; a ^= (c >> 13);					\
2893 	b -= c; b -= a; b ^= (a << 8);					\
2894 	c -= a; c -= b; c ^= (b >> 13);					\
2895 	a -= b; a -= c; a ^= (c >> 12);					\
2896 	b -= c; b -= a; b ^= (a << 16);					\
2897 	c -= a; c -= b; c ^= (b >> 5);					\
2898 	a -= b; a -= c; a ^= (c >> 3);					\
2899 	b -= c; b -= a; b ^= (a << 10);					\
2900 	c -= a; c -= b; c ^= (b >> 15);					\
2901 } while (/*CONSTCOND*/0)
2902 
2903 static __inline uint32_t
2904 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2905 {
2906 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2907 
2908 	b += addr[5] << 8;
2909 	b += addr[4];
2910 	a += addr[3] << 24;
2911 	a += addr[2] << 16;
2912 	a += addr[1] << 8;
2913 	a += addr[0];
2914 
2915 	mix(a, b, c);
2916 
2917 	return (c & BRIDGE_RTHASH_MASK);
2918 }
2919 
2920 #undef mix
2921 
2922 static int
2923 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2924 {
2925 	int i, d;
2926 
2927 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2928 		d = ((int)a[i]) - ((int)b[i]);
2929 	}
2930 
2931 	return (d);
2932 }
2933 
2934 /*
2935  * bridge_rtnode_lookup:
2936  *
2937  *	Look up a bridge route node for the specified destination. Compare the
2938  *	vlan id or if zero then just return the first match.
2939  */
2940 static struct bridge_rtnode *
2941 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2942 {
2943 	struct bridge_rtnode *brt;
2944 	uint32_t hash;
2945 	int dir;
2946 
2947 	BRIDGE_LOCK_ASSERT(sc);
2948 
2949 	hash = bridge_rthash(sc, addr);
2950 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2951 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2952 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2953 			return (brt);
2954 		if (dir > 0)
2955 			return (NULL);
2956 	}
2957 
2958 	return (NULL);
2959 }
2960 
2961 /*
2962  * bridge_rtnode_insert:
2963  *
2964  *	Insert the specified bridge node into the route table.  We
2965  *	assume the entry is not already in the table.
2966  */
2967 static int
2968 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2969 {
2970 	struct bridge_rtnode *lbrt;
2971 	uint32_t hash;
2972 	int dir;
2973 
2974 	BRIDGE_LOCK_ASSERT(sc);
2975 
2976 	hash = bridge_rthash(sc, brt->brt_addr);
2977 
2978 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2979 	if (lbrt == NULL) {
2980 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2981 		goto out;
2982 	}
2983 
2984 	do {
2985 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2986 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2987 			return (EEXIST);
2988 		if (dir > 0) {
2989 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2990 			goto out;
2991 		}
2992 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2993 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2994 			goto out;
2995 		}
2996 		lbrt = LIST_NEXT(lbrt, brt_hash);
2997 	} while (lbrt != NULL);
2998 
2999 #ifdef DIAGNOSTIC
3000 	panic("bridge_rtnode_insert: impossible");
3001 #endif
3002 
3003 out:
3004 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3005 	sc->sc_brtcnt++;
3006 
3007 	return (0);
3008 }
3009 
3010 /*
3011  * bridge_rtnode_destroy:
3012  *
3013  *	Destroy a bridge rtnode.
3014  */
3015 static void
3016 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3017 {
3018 	BRIDGE_LOCK_ASSERT(sc);
3019 
3020 	LIST_REMOVE(brt, brt_hash);
3021 
3022 	LIST_REMOVE(brt, brt_list);
3023 	sc->sc_brtcnt--;
3024 	brt->brt_dst->bif_addrcnt--;
3025 	uma_zfree(bridge_rtnode_zone, brt);
3026 }
3027 
3028 /*
3029  * bridge_rtable_expire:
3030  *
3031  *	Set the expiry time for all routes on an interface.
3032  */
3033 static void
3034 bridge_rtable_expire(struct ifnet *ifp, int age)
3035 {
3036 	struct bridge_softc *sc = ifp->if_bridge;
3037 	struct bridge_rtnode *brt;
3038 
3039 	BRIDGE_LOCK(sc);
3040 
3041 	/*
3042 	 * If the age is zero then flush, otherwise set all the expiry times to
3043 	 * age for the interface
3044 	 */
3045 	if (age == 0)
3046 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3047 	else {
3048 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3049 			/* Cap the expiry time to 'age' */
3050 			if (brt->brt_ifp == ifp &&
3051 			    brt->brt_expire > time_uptime + age &&
3052 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3053 				brt->brt_expire = time_uptime + age;
3054 		}
3055 	}
3056 	BRIDGE_UNLOCK(sc);
3057 }
3058 
3059 /*
3060  * bridge_state_change:
3061  *
3062  *	Callback from the bridgestp code when a port changes states.
3063  */
3064 static void
3065 bridge_state_change(struct ifnet *ifp, int state)
3066 {
3067 	struct bridge_softc *sc = ifp->if_bridge;
3068 	static const char *stpstates[] = {
3069 		"disabled",
3070 		"listening",
3071 		"learning",
3072 		"forwarding",
3073 		"blocking",
3074 		"discarding"
3075 	};
3076 
3077 	CURVNET_SET(ifp->if_vnet);
3078 	if (V_log_stp)
3079 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3080 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3081 	CURVNET_RESTORE();
3082 }
3083 
3084 /*
3085  * Send bridge packets through pfil if they are one of the types pfil can deal
3086  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3087  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3088  * that interface.
3089  */
3090 static int
3091 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3092 {
3093 	int snap, error, i, hlen;
3094 	struct ether_header *eh1, eh2;
3095 	struct ip *ip;
3096 	struct llc llc1;
3097 	u_int16_t ether_type;
3098 
3099 	snap = 0;
3100 	error = -1;	/* Default error if not error == 0 */
3101 
3102 #if 0
3103 	/* we may return with the IP fields swapped, ensure its not shared */
3104 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3105 #endif
3106 
3107 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3108 		return (0); /* filtering is disabled */
3109 
3110 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3111 	if ((*mp)->m_len < i) {
3112 	    *mp = m_pullup(*mp, i);
3113 	    if (*mp == NULL) {
3114 		printf("%s: m_pullup failed\n", __func__);
3115 		return (-1);
3116 	    }
3117 	}
3118 
3119 	eh1 = mtod(*mp, struct ether_header *);
3120 	ether_type = ntohs(eh1->ether_type);
3121 
3122 	/*
3123 	 * Check for SNAP/LLC.
3124 	 */
3125 	if (ether_type < ETHERMTU) {
3126 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3127 
3128 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3129 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3130 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3131 		    llc2->llc_control == LLC_UI) {
3132 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3133 			snap = 1;
3134 		}
3135 	}
3136 
3137 	/*
3138 	 * If we're trying to filter bridge traffic, don't look at anything
3139 	 * other than IP and ARP traffic.  If the filter doesn't understand
3140 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3141 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3142 	 * but of course we don't have an AppleTalk filter to begin with.
3143 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3144 	 * ARP traffic.)
3145 	 */
3146 	switch (ether_type) {
3147 		case ETHERTYPE_ARP:
3148 		case ETHERTYPE_REVARP:
3149 			if (V_pfil_ipfw_arp == 0)
3150 				return (0); /* Automatically pass */
3151 			break;
3152 
3153 		case ETHERTYPE_IP:
3154 #ifdef INET6
3155 		case ETHERTYPE_IPV6:
3156 #endif /* INET6 */
3157 			break;
3158 		default:
3159 			/*
3160 			 * Check to see if the user wants to pass non-ip
3161 			 * packets, these will not be checked by pfil(9) and
3162 			 * passed unconditionally so the default is to drop.
3163 			 */
3164 			if (V_pfil_onlyip)
3165 				goto bad;
3166 	}
3167 
3168 	/* Run the packet through pfil before stripping link headers */
3169 	if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3170 			dir == PFIL_OUT && ifp != NULL) {
3171 
3172 		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3173 
3174 		if (*mp == NULL || error != 0) /* packet consumed by filter */
3175 			return (error);
3176 	}
3177 
3178 	/* Strip off the Ethernet header and keep a copy. */
3179 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3180 	m_adj(*mp, ETHER_HDR_LEN);
3181 
3182 	/* Strip off snap header, if present */
3183 	if (snap) {
3184 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3185 		m_adj(*mp, sizeof(struct llc));
3186 	}
3187 
3188 	/*
3189 	 * Check the IP header for alignment and errors
3190 	 */
3191 	if (dir == PFIL_IN) {
3192 		switch (ether_type) {
3193 			case ETHERTYPE_IP:
3194 				error = bridge_ip_checkbasic(mp);
3195 				break;
3196 #ifdef INET6
3197 			case ETHERTYPE_IPV6:
3198 				error = bridge_ip6_checkbasic(mp);
3199 				break;
3200 #endif /* INET6 */
3201 			default:
3202 				error = 0;
3203 		}
3204 		if (error)
3205 			goto bad;
3206 	}
3207 
3208 	error = 0;
3209 
3210 	/*
3211 	 * Run the packet through pfil
3212 	 */
3213 	switch (ether_type) {
3214 	case ETHERTYPE_IP:
3215 		/*
3216 		 * Run pfil on the member interface and the bridge, both can
3217 		 * be skipped by clearing pfil_member or pfil_bridge.
3218 		 *
3219 		 * Keep the order:
3220 		 *   in_if -> bridge_if -> out_if
3221 		 */
3222 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3223 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3224 					dir, NULL);
3225 
3226 		if (*mp == NULL || error != 0) /* filter may consume */
3227 			break;
3228 
3229 		if (V_pfil_member && ifp != NULL)
3230 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3231 					dir, NULL);
3232 
3233 		if (*mp == NULL || error != 0) /* filter may consume */
3234 			break;
3235 
3236 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3237 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3238 					dir, NULL);
3239 
3240 		if (*mp == NULL || error != 0) /* filter may consume */
3241 			break;
3242 
3243 		/* check if we need to fragment the packet */
3244 		/* bridge_fragment generates a mbuf chain of packets */
3245 		/* that already include eth headers */
3246 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3247 			i = (*mp)->m_pkthdr.len;
3248 			if (i > ifp->if_mtu) {
3249 				error = bridge_fragment(ifp, mp, &eh2, snap,
3250 					    &llc1);
3251 				return (error);
3252 			}
3253 		}
3254 
3255 		/* Recalculate the ip checksum. */
3256 		ip = mtod(*mp, struct ip *);
3257 		hlen = ip->ip_hl << 2;
3258 		if (hlen < sizeof(struct ip))
3259 			goto bad;
3260 		if (hlen > (*mp)->m_len) {
3261 			if ((*mp = m_pullup(*mp, hlen)) == NULL)
3262 				goto bad;
3263 			ip = mtod(*mp, struct ip *);
3264 			if (ip == NULL)
3265 				goto bad;
3266 		}
3267 		ip->ip_sum = 0;
3268 		if (hlen == sizeof(struct ip))
3269 			ip->ip_sum = in_cksum_hdr(ip);
3270 		else
3271 			ip->ip_sum = in_cksum(*mp, hlen);
3272 
3273 		break;
3274 #ifdef INET6
3275 	case ETHERTYPE_IPV6:
3276 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3277 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3278 					dir, NULL);
3279 
3280 		if (*mp == NULL || error != 0) /* filter may consume */
3281 			break;
3282 
3283 		if (V_pfil_member && ifp != NULL)
3284 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3285 					dir, NULL);
3286 
3287 		if (*mp == NULL || error != 0) /* filter may consume */
3288 			break;
3289 
3290 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3291 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3292 					dir, NULL);
3293 		break;
3294 #endif
3295 	default:
3296 		error = 0;
3297 		break;
3298 	}
3299 
3300 	if (*mp == NULL)
3301 		return (error);
3302 	if (error != 0)
3303 		goto bad;
3304 
3305 	error = -1;
3306 
3307 	/*
3308 	 * Finally, put everything back the way it was and return
3309 	 */
3310 	if (snap) {
3311 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3312 		if (*mp == NULL)
3313 			return (error);
3314 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3315 	}
3316 
3317 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3318 	if (*mp == NULL)
3319 		return (error);
3320 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3321 
3322 	return (0);
3323 
3324 bad:
3325 	m_freem(*mp);
3326 	*mp = NULL;
3327 	return (error);
3328 }
3329 
3330 /*
3331  * Perform basic checks on header size since
3332  * pfil assumes ip_input has already processed
3333  * it for it.  Cut-and-pasted from ip_input.c.
3334  * Given how simple the IPv6 version is,
3335  * does the IPv4 version really need to be
3336  * this complicated?
3337  *
3338  * XXX Should we update ipstat here, or not?
3339  * XXX Right now we update ipstat but not
3340  * XXX csum_counter.
3341  */
3342 static int
3343 bridge_ip_checkbasic(struct mbuf **mp)
3344 {
3345 	struct mbuf *m = *mp;
3346 	struct ip *ip;
3347 	int len, hlen;
3348 	u_short sum;
3349 
3350 	if (*mp == NULL)
3351 		return (-1);
3352 
3353 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3354 		if ((m = m_copyup(m, sizeof(struct ip),
3355 			(max_linkhdr + 3) & ~3)) == NULL) {
3356 			/* XXXJRT new stat, please */
3357 			KMOD_IPSTAT_INC(ips_toosmall);
3358 			goto bad;
3359 		}
3360 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3361 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3362 			KMOD_IPSTAT_INC(ips_toosmall);
3363 			goto bad;
3364 		}
3365 	}
3366 	ip = mtod(m, struct ip *);
3367 	if (ip == NULL) goto bad;
3368 
3369 	if (ip->ip_v != IPVERSION) {
3370 		KMOD_IPSTAT_INC(ips_badvers);
3371 		goto bad;
3372 	}
3373 	hlen = ip->ip_hl << 2;
3374 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3375 		KMOD_IPSTAT_INC(ips_badhlen);
3376 		goto bad;
3377 	}
3378 	if (hlen > m->m_len) {
3379 		if ((m = m_pullup(m, hlen)) == NULL) {
3380 			KMOD_IPSTAT_INC(ips_badhlen);
3381 			goto bad;
3382 		}
3383 		ip = mtod(m, struct ip *);
3384 		if (ip == NULL) goto bad;
3385 	}
3386 
3387 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3388 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3389 	} else {
3390 		if (hlen == sizeof(struct ip)) {
3391 			sum = in_cksum_hdr(ip);
3392 		} else {
3393 			sum = in_cksum(m, hlen);
3394 		}
3395 	}
3396 	if (sum) {
3397 		KMOD_IPSTAT_INC(ips_badsum);
3398 		goto bad;
3399 	}
3400 
3401 	/* Retrieve the packet length. */
3402 	len = ntohs(ip->ip_len);
3403 
3404 	/*
3405 	 * Check for additional length bogosity
3406 	 */
3407 	if (len < hlen) {
3408 		KMOD_IPSTAT_INC(ips_badlen);
3409 		goto bad;
3410 	}
3411 
3412 	/*
3413 	 * Check that the amount of data in the buffers
3414 	 * is as at least much as the IP header would have us expect.
3415 	 * Drop packet if shorter than we expect.
3416 	 */
3417 	if (m->m_pkthdr.len < len) {
3418 		KMOD_IPSTAT_INC(ips_tooshort);
3419 		goto bad;
3420 	}
3421 
3422 	/* Checks out, proceed */
3423 	*mp = m;
3424 	return (0);
3425 
3426 bad:
3427 	*mp = m;
3428 	return (-1);
3429 }
3430 
3431 #ifdef INET6
3432 /*
3433  * Same as above, but for IPv6.
3434  * Cut-and-pasted from ip6_input.c.
3435  * XXX Should we update ip6stat, or not?
3436  */
3437 static int
3438 bridge_ip6_checkbasic(struct mbuf **mp)
3439 {
3440 	struct mbuf *m = *mp;
3441 	struct ip6_hdr *ip6;
3442 
3443 	/*
3444 	 * If the IPv6 header is not aligned, slurp it up into a new
3445 	 * mbuf with space for link headers, in the event we forward
3446 	 * it.  Otherwise, if it is aligned, make sure the entire base
3447 	 * IPv6 header is in the first mbuf of the chain.
3448 	 */
3449 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3450 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3451 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3452 			    (max_linkhdr + 3) & ~3)) == NULL) {
3453 			/* XXXJRT new stat, please */
3454 			IP6STAT_INC(ip6s_toosmall);
3455 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3456 			goto bad;
3457 		}
3458 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3459 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3460 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3461 			IP6STAT_INC(ip6s_toosmall);
3462 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3463 			goto bad;
3464 		}
3465 	}
3466 
3467 	ip6 = mtod(m, struct ip6_hdr *);
3468 
3469 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3470 		IP6STAT_INC(ip6s_badvers);
3471 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3472 		goto bad;
3473 	}
3474 
3475 	/* Checks out, proceed */
3476 	*mp = m;
3477 	return (0);
3478 
3479 bad:
3480 	*mp = m;
3481 	return (-1);
3482 }
3483 #endif /* INET6 */
3484 
3485 /*
3486  * bridge_fragment:
3487  *
3488  *	Fragment mbuf chain in multiple packets and prepend ethernet header.
3489  */
3490 static int
3491 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3492     int snap, struct llc *llc)
3493 {
3494 	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3495 	struct ip *ip;
3496 	int error = -1;
3497 
3498 	if (m->m_len < sizeof(struct ip) &&
3499 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3500 		goto dropit;
3501 	ip = mtod(m, struct ip *);
3502 
3503 	m->m_pkthdr.csum_flags |= CSUM_IP;
3504 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3505 	if (error)
3506 		goto dropit;
3507 
3508 	/*
3509 	 * Walk the chain and re-add the Ethernet header for
3510 	 * each mbuf packet.
3511 	 */
3512 	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3513 		nextpkt = mcur->m_nextpkt;
3514 		mcur->m_nextpkt = NULL;
3515 		if (snap) {
3516 			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3517 			if (mcur == NULL) {
3518 				error = ENOBUFS;
3519 				if (mprev != NULL)
3520 					mprev->m_nextpkt = nextpkt;
3521 				goto dropit;
3522 			}
3523 			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3524 		}
3525 
3526 		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3527 		if (mcur == NULL) {
3528 			error = ENOBUFS;
3529 			if (mprev != NULL)
3530 				mprev->m_nextpkt = nextpkt;
3531 			goto dropit;
3532 		}
3533 		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3534 
3535 		/*
3536 		 * The previous two M_PREPEND could have inserted one or two
3537 		 * mbufs in front so we have to update the previous packet's
3538 		 * m_nextpkt.
3539 		 */
3540 		mcur->m_nextpkt = nextpkt;
3541 		if (mprev != NULL)
3542 			mprev->m_nextpkt = mcur;
3543 		else {
3544 			/* The first mbuf in the original chain needs to be
3545 			 * updated. */
3546 			*mp = mcur;
3547 		}
3548 		mprev = mcur;
3549 	}
3550 
3551 	KMOD_IPSTAT_INC(ips_fragmented);
3552 	return (error);
3553 
3554 dropit:
3555 	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3556 		m = mcur->m_nextpkt;
3557 		m_freem(mcur);
3558 	}
3559 	return (error);
3560 }
3561 
3562 static void
3563 bridge_linkstate(struct ifnet *ifp)
3564 {
3565 	struct bridge_softc *sc = ifp->if_bridge;
3566 	struct bridge_iflist *bif;
3567 
3568 	BRIDGE_LOCK(sc);
3569 	bif = bridge_lookup_member_if(sc, ifp);
3570 	if (bif == NULL) {
3571 		BRIDGE_UNLOCK(sc);
3572 		return;
3573 	}
3574 	bridge_linkcheck(sc);
3575 	BRIDGE_UNLOCK(sc);
3576 
3577 	bstp_linkstate(&bif->bif_stp);
3578 }
3579 
3580 static void
3581 bridge_linkcheck(struct bridge_softc *sc)
3582 {
3583 	struct bridge_iflist *bif;
3584 	int new_link, hasls;
3585 
3586 	BRIDGE_LOCK_ASSERT(sc);
3587 	new_link = LINK_STATE_DOWN;
3588 	hasls = 0;
3589 	/* Our link is considered up if at least one of our ports is active */
3590 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3591 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3592 			hasls++;
3593 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3594 			new_link = LINK_STATE_UP;
3595 			break;
3596 		}
3597 	}
3598 	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3599 		/* If no interfaces support link-state then we default to up */
3600 		new_link = LINK_STATE_UP;
3601 	}
3602 	if_link_state_change(sc->sc_ifp, new_link);
3603 }
3604