xref: /freebsd/sys/net/if_bridge.c (revision eb69d1f144a6fcc765d1b9d44a5ae8082353e70b)
1 /*	$NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $	*/
2 
3 /*-
4  * SPDX-License-Identifier: BSD-4-Clause
5  *
6  * Copyright 2001 Wasabi Systems, Inc.
7  * All rights reserved.
8  *
9  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  * 3. All advertising materials mentioning features or use of this software
20  *    must display the following acknowledgement:
21  *	This product includes software developed for the NetBSD Project by
22  *	Wasabi Systems, Inc.
23  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
24  *    or promote products derived from this software without specific prior
25  *    written permission.
26  *
27  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
28  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
31  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37  * POSSIBILITY OF SUCH DAMAGE.
38  */
39 
40 /*
41  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
42  * All rights reserved.
43  *
44  * Redistribution and use in source and binary forms, with or without
45  * modification, are permitted provided that the following conditions
46  * are met:
47  * 1. Redistributions of source code must retain the above copyright
48  *    notice, this list of conditions and the following disclaimer.
49  * 2. Redistributions in binary form must reproduce the above copyright
50  *    notice, this list of conditions and the following disclaimer in the
51  *    documentation and/or other materials provided with the distribution.
52  *
53  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
54  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
55  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
56  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
57  * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
58  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
59  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
61  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
62  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
63  * POSSIBILITY OF SUCH DAMAGE.
64  *
65  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
66  */
67 
68 /*
69  * Network interface bridge support.
70  *
71  * TODO:
72  *
73  *	- Currently only supports Ethernet-like interfaces (Ethernet,
74  *	  802.11, VLANs on Ethernet, etc.)  Figure out a nice way
75  *	  to bridge other types of interfaces (FDDI-FDDI, and maybe
76  *	  consider heterogeneous bridges).
77  */
78 
79 #include <sys/cdefs.h>
80 __FBSDID("$FreeBSD$");
81 
82 #include "opt_inet.h"
83 #include "opt_inet6.h"
84 
85 #include <sys/param.h>
86 #include <sys/eventhandler.h>
87 #include <sys/mbuf.h>
88 #include <sys/malloc.h>
89 #include <sys/protosw.h>
90 #include <sys/systm.h>
91 #include <sys/jail.h>
92 #include <sys/time.h>
93 #include <sys/socket.h> /* for net/if.h */
94 #include <sys/sockio.h>
95 #include <sys/ctype.h>  /* string functions */
96 #include <sys/kernel.h>
97 #include <sys/random.h>
98 #include <sys/syslog.h>
99 #include <sys/sysctl.h>
100 #include <vm/uma.h>
101 #include <sys/module.h>
102 #include <sys/priv.h>
103 #include <sys/proc.h>
104 #include <sys/lock.h>
105 #include <sys/mutex.h>
106 
107 #include <net/bpf.h>
108 #include <net/if.h>
109 #include <net/if_clone.h>
110 #include <net/if_dl.h>
111 #include <net/if_types.h>
112 #include <net/if_var.h>
113 #include <net/pfil.h>
114 #include <net/vnet.h>
115 
116 #include <netinet/in.h>
117 #include <netinet/in_systm.h>
118 #include <netinet/in_var.h>
119 #include <netinet/ip.h>
120 #include <netinet/ip_var.h>
121 #ifdef INET6
122 #include <netinet/ip6.h>
123 #include <netinet6/ip6_var.h>
124 #include <netinet6/in6_ifattach.h>
125 #endif
126 #if defined(INET) || defined(INET6)
127 #include <netinet/ip_carp.h>
128 #endif
129 #include <machine/in_cksum.h>
130 #include <netinet/if_ether.h>
131 #include <net/bridgestp.h>
132 #include <net/if_bridgevar.h>
133 #include <net/if_llc.h>
134 #include <net/if_vlan_var.h>
135 
136 #include <net/route.h>
137 
138 /*
139  * Size of the route hash table.  Must be a power of two.
140  */
141 #ifndef BRIDGE_RTHASH_SIZE
142 #define	BRIDGE_RTHASH_SIZE		1024
143 #endif
144 
145 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
146 
147 /*
148  * Default maximum number of addresses to cache.
149  */
150 #ifndef BRIDGE_RTABLE_MAX
151 #define	BRIDGE_RTABLE_MAX		2000
152 #endif
153 
154 /*
155  * Timeout (in seconds) for entries learned dynamically.
156  */
157 #ifndef BRIDGE_RTABLE_TIMEOUT
158 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
159 #endif
160 
161 /*
162  * Number of seconds between walks of the route list.
163  */
164 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
165 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
166 #endif
167 
168 /*
169  * List of capabilities to possibly mask on the member interface.
170  */
171 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM|\
172 					 IFCAP_TXCSUM_IPV6)
173 
174 /*
175  * List of capabilities to strip
176  */
177 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
178 
179 /*
180  * Bridge interface list entry.
181  */
182 struct bridge_iflist {
183 	LIST_ENTRY(bridge_iflist) bif_next;
184 	struct ifnet		*bif_ifp;	/* member if */
185 	struct bstp_port	bif_stp;	/* STP state */
186 	uint32_t		bif_flags;	/* member if flags */
187 	int			bif_savedcaps;	/* saved capabilities */
188 	uint32_t		bif_addrmax;	/* max # of addresses */
189 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
190 	uint32_t		bif_addrexceeded;/* # of address violations */
191 };
192 
193 /*
194  * Bridge route node.
195  */
196 struct bridge_rtnode {
197 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
198 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
199 	struct bridge_iflist	*brt_dst;	/* destination if */
200 	unsigned long		brt_expire;	/* expiration time */
201 	uint8_t			brt_flags;	/* address flags */
202 	uint8_t			brt_addr[ETHER_ADDR_LEN];
203 	uint16_t		brt_vlan;	/* vlan id */
204 };
205 #define	brt_ifp			brt_dst->bif_ifp
206 
207 /*
208  * Software state for each bridge.
209  */
210 struct bridge_softc {
211 	struct ifnet		*sc_ifp;	/* make this an interface */
212 	LIST_ENTRY(bridge_softc) sc_list;
213 	struct mtx		sc_mtx;
214 	struct cv		sc_cv;
215 	uint32_t		sc_brtmax;	/* max # of addresses */
216 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
217 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
218 	struct callout		sc_brcallout;	/* bridge callout */
219 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
220 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
221 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
222 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
223 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
224 	uint32_t		sc_rthash_key;	/* key for hash */
225 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
226 	struct bstp_state	sc_stp;		/* STP state */
227 	uint32_t		sc_brtexceeded;	/* # of cache drops */
228 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
229 	u_char			sc_defaddr[6];	/* Default MAC address */
230 };
231 
232 static VNET_DEFINE(struct mtx, bridge_list_mtx);
233 #define	V_bridge_list_mtx	VNET(bridge_list_mtx)
234 static eventhandler_tag bridge_detach_cookie;
235 
236 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
237 
238 uma_zone_t bridge_rtnode_zone;
239 
240 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
241 static void	bridge_clone_destroy(struct ifnet *);
242 
243 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
244 static void	bridge_mutecaps(struct bridge_softc *);
245 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
246 		    int);
247 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
248 static void	bridge_init(void *);
249 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
250 static void	bridge_stop(struct ifnet *, int);
251 static int	bridge_transmit(struct ifnet *, struct mbuf *);
252 static void	bridge_qflush(struct ifnet *);
253 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
254 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
255 		    struct rtentry *);
256 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
257 		    struct mbuf *);
258 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
259 
260 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
261 		    struct mbuf *m);
262 
263 static void	bridge_timer(void *);
264 
265 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
266 		    struct mbuf *, int);
267 static void	bridge_span(struct bridge_softc *, struct mbuf *);
268 
269 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
270 		    uint16_t, struct bridge_iflist *, int, uint8_t);
271 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
272 		    uint16_t);
273 static void	bridge_rttrim(struct bridge_softc *);
274 static void	bridge_rtage(struct bridge_softc *);
275 static void	bridge_rtflush(struct bridge_softc *, int);
276 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
277 		    uint16_t);
278 
279 static void	bridge_rtable_init(struct bridge_softc *);
280 static void	bridge_rtable_fini(struct bridge_softc *);
281 
282 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
283 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
284 		    const uint8_t *, uint16_t);
285 static int	bridge_rtnode_insert(struct bridge_softc *,
286 		    struct bridge_rtnode *);
287 static void	bridge_rtnode_destroy(struct bridge_softc *,
288 		    struct bridge_rtnode *);
289 static void	bridge_rtable_expire(struct ifnet *, int);
290 static void	bridge_state_change(struct ifnet *, int);
291 
292 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
293 		    const char *name);
294 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
295 		    struct ifnet *ifp);
296 static void	bridge_delete_member(struct bridge_softc *,
297 		    struct bridge_iflist *, int);
298 static void	bridge_delete_span(struct bridge_softc *,
299 		    struct bridge_iflist *);
300 
301 static int	bridge_ioctl_add(struct bridge_softc *, void *);
302 static int	bridge_ioctl_del(struct bridge_softc *, void *);
303 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
304 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
305 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
306 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
307 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
308 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
309 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
310 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
311 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
312 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
313 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
314 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
315 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
316 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
317 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
318 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
319 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
320 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
321 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
322 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
323 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
324 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
325 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
326 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
327 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
328 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
329 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
330 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
331 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
332 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
333 		    int);
334 static int	bridge_ip_checkbasic(struct mbuf **mp);
335 #ifdef INET6
336 static int	bridge_ip6_checkbasic(struct mbuf **mp);
337 #endif /* INET6 */
338 static int	bridge_fragment(struct ifnet *, struct mbuf **mp,
339 		    struct ether_header *, int, struct llc *);
340 static void	bridge_linkstate(struct ifnet *ifp);
341 static void	bridge_linkcheck(struct bridge_softc *sc);
342 
343 extern void (*bridge_linkstate_p)(struct ifnet *ifp);
344 
345 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
346 #define	VLANTAGOF(_m)	\
347     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
348 
349 static struct bstp_cb_ops bridge_ops = {
350 	.bcb_state = bridge_state_change,
351 	.bcb_rtage = bridge_rtable_expire
352 };
353 
354 SYSCTL_DECL(_net_link);
355 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
356 
357 /* only pass IP[46] packets when pfil is enabled */
358 static VNET_DEFINE(int, pfil_onlyip) = 1;
359 #define	V_pfil_onlyip	VNET(pfil_onlyip)
360 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
361     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
362     "Only pass IP packets when pfil is enabled");
363 
364 /* run pfil hooks on the bridge interface */
365 static VNET_DEFINE(int, pfil_bridge) = 1;
366 #define	V_pfil_bridge	VNET(pfil_bridge)
367 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
368     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
369     "Packet filter on the bridge interface");
370 
371 /* layer2 filter with ipfw */
372 static VNET_DEFINE(int, pfil_ipfw);
373 #define	V_pfil_ipfw	VNET(pfil_ipfw)
374 
375 /* layer2 ARP filter with ipfw */
376 static VNET_DEFINE(int, pfil_ipfw_arp);
377 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
378 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
379     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
380     "Filter ARP packets through IPFW layer2");
381 
382 /* run pfil hooks on the member interface */
383 static VNET_DEFINE(int, pfil_member) = 1;
384 #define	V_pfil_member	VNET(pfil_member)
385 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
386     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
387     "Packet filter on the member interface");
388 
389 /* run pfil hooks on the physical interface for locally destined packets */
390 static VNET_DEFINE(int, pfil_local_phys);
391 #define	V_pfil_local_phys	VNET(pfil_local_phys)
392 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
393     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
394     "Packet filter on the physical interface for locally destined packets");
395 
396 /* log STP state changes */
397 static VNET_DEFINE(int, log_stp);
398 #define	V_log_stp	VNET(log_stp)
399 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
400     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
401     "Log STP state changes");
402 
403 /* share MAC with first bridge member */
404 static VNET_DEFINE(int, bridge_inherit_mac);
405 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
406 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
407     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
408     "Inherit MAC address from the first bridge member");
409 
410 static VNET_DEFINE(int, allow_llz_overlap) = 0;
411 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
412 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
413     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
414     "Allow overlap of link-local scope "
415     "zones of a bridge interface and the member interfaces");
416 
417 struct bridge_control {
418 	int	(*bc_func)(struct bridge_softc *, void *);
419 	int	bc_argsize;
420 	int	bc_flags;
421 };
422 
423 #define	BC_F_COPYIN		0x01	/* copy arguments in */
424 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
425 #define	BC_F_SUSER		0x04	/* do super-user check */
426 
427 const struct bridge_control bridge_control_table[] = {
428 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
429 	  BC_F_COPYIN|BC_F_SUSER },
430 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
431 	  BC_F_COPYIN|BC_F_SUSER },
432 
433 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
434 	  BC_F_COPYIN|BC_F_COPYOUT },
435 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
436 	  BC_F_COPYIN|BC_F_SUSER },
437 
438 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
439 	  BC_F_COPYIN|BC_F_SUSER },
440 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
441 	  BC_F_COPYOUT },
442 
443 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
444 	  BC_F_COPYIN|BC_F_COPYOUT },
445 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
446 	  BC_F_COPYIN|BC_F_COPYOUT },
447 
448 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
449 	  BC_F_COPYIN|BC_F_SUSER },
450 
451 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
452 	  BC_F_COPYIN|BC_F_SUSER },
453 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
454 	  BC_F_COPYOUT },
455 
456 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
457 	  BC_F_COPYIN|BC_F_SUSER },
458 
459 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
460 	  BC_F_COPYIN|BC_F_SUSER },
461 
462 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
463 	  BC_F_COPYOUT },
464 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
465 	  BC_F_COPYIN|BC_F_SUSER },
466 
467 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
468 	  BC_F_COPYOUT },
469 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
470 	  BC_F_COPYIN|BC_F_SUSER },
471 
472 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
473 	  BC_F_COPYOUT },
474 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
475 	  BC_F_COPYIN|BC_F_SUSER },
476 
477 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
478 	  BC_F_COPYOUT },
479 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
480 	  BC_F_COPYIN|BC_F_SUSER },
481 
482 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
483 	  BC_F_COPYIN|BC_F_SUSER },
484 
485 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
486 	  BC_F_COPYIN|BC_F_SUSER },
487 
488 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
489 	  BC_F_COPYIN|BC_F_SUSER },
490 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
491 	  BC_F_COPYIN|BC_F_SUSER },
492 
493 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
494 	  BC_F_COPYOUT },
495 
496 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
497 	  BC_F_COPYOUT },
498 
499 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
500 	  BC_F_COPYIN|BC_F_COPYOUT },
501 
502 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
503 	  BC_F_COPYIN|BC_F_SUSER },
504 
505 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
506 	  BC_F_COPYIN|BC_F_SUSER },
507 
508 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
509 	  BC_F_COPYIN|BC_F_SUSER },
510 
511 };
512 const int bridge_control_table_size = nitems(bridge_control_table);
513 
514 static VNET_DEFINE(LIST_HEAD(, bridge_softc), bridge_list);
515 #define	V_bridge_list	VNET(bridge_list)
516 #define	BRIDGE_LIST_LOCK_INIT(x)	mtx_init(&V_bridge_list_mtx,	\
517 					    "if_bridge list", NULL, MTX_DEF)
518 #define	BRIDGE_LIST_LOCK_DESTROY(x)	mtx_destroy(&V_bridge_list_mtx)
519 #define	BRIDGE_LIST_LOCK(x)		mtx_lock(&V_bridge_list_mtx)
520 #define	BRIDGE_LIST_UNLOCK(x)		mtx_unlock(&V_bridge_list_mtx)
521 
522 static VNET_DEFINE(struct if_clone *, bridge_cloner);
523 #define	V_bridge_cloner	VNET(bridge_cloner)
524 
525 static const char bridge_name[] = "bridge";
526 
527 static void
528 vnet_bridge_init(const void *unused __unused)
529 {
530 
531 	BRIDGE_LIST_LOCK_INIT();
532 	LIST_INIT(&V_bridge_list);
533 	V_bridge_cloner = if_clone_simple(bridge_name,
534 	    bridge_clone_create, bridge_clone_destroy, 0);
535 }
536 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
537     vnet_bridge_init, NULL);
538 
539 static void
540 vnet_bridge_uninit(const void *unused __unused)
541 {
542 
543 	if_clone_detach(V_bridge_cloner);
544 	V_bridge_cloner = NULL;
545 	BRIDGE_LIST_LOCK_DESTROY();
546 }
547 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
548     vnet_bridge_uninit, NULL);
549 
550 static int
551 bridge_modevent(module_t mod, int type, void *data)
552 {
553 
554 	switch (type) {
555 	case MOD_LOAD:
556 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
557 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
558 		    UMA_ALIGN_PTR, 0);
559 		bridge_input_p = bridge_input;
560 		bridge_output_p = bridge_output;
561 		bridge_dn_p = bridge_dummynet;
562 		bridge_linkstate_p = bridge_linkstate;
563 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
564 		    ifnet_departure_event, bridge_ifdetach, NULL,
565 		    EVENTHANDLER_PRI_ANY);
566 		break;
567 	case MOD_UNLOAD:
568 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
569 		    bridge_detach_cookie);
570 		uma_zdestroy(bridge_rtnode_zone);
571 		bridge_input_p = NULL;
572 		bridge_output_p = NULL;
573 		bridge_dn_p = NULL;
574 		bridge_linkstate_p = NULL;
575 		break;
576 	default:
577 		return (EOPNOTSUPP);
578 	}
579 	return (0);
580 }
581 
582 static moduledata_t bridge_mod = {
583 	"if_bridge",
584 	bridge_modevent,
585 	0
586 };
587 
588 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
589 MODULE_VERSION(if_bridge, 1);
590 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
591 
592 /*
593  * handler for net.link.bridge.ipfw
594  */
595 static int
596 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
597 {
598 	int enable = V_pfil_ipfw;
599 	int error;
600 
601 	error = sysctl_handle_int(oidp, &enable, 0, req);
602 	enable &= 1;
603 
604 	if (enable != V_pfil_ipfw) {
605 		V_pfil_ipfw = enable;
606 
607 		/*
608 		 * Disable pfil so that ipfw doesnt run twice, if the user
609 		 * really wants both then they can re-enable pfil_bridge and/or
610 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
611 		 * layer2 type.
612 		 */
613 		if (V_pfil_ipfw) {
614 			V_pfil_onlyip = 0;
615 			V_pfil_bridge = 0;
616 			V_pfil_member = 0;
617 		}
618 	}
619 
620 	return (error);
621 }
622 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
623     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
624     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
625     "Layer2 filter with IPFW");
626 
627 /*
628  * bridge_clone_create:
629  *
630  *	Create a new bridge instance.
631  */
632 static int
633 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
634 {
635 	struct bridge_softc *sc, *sc2;
636 	struct ifnet *bifp, *ifp;
637 	int fb, retry;
638 	unsigned long hostid;
639 
640 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
641 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
642 	if (ifp == NULL) {
643 		free(sc, M_DEVBUF);
644 		return (ENOSPC);
645 	}
646 
647 	BRIDGE_LOCK_INIT(sc);
648 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
649 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
650 
651 	/* Initialize our routing table. */
652 	bridge_rtable_init(sc);
653 
654 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
655 
656 	LIST_INIT(&sc->sc_iflist);
657 	LIST_INIT(&sc->sc_spanlist);
658 
659 	ifp->if_softc = sc;
660 	if_initname(ifp, bridge_name, unit);
661 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
662 	ifp->if_ioctl = bridge_ioctl;
663 	ifp->if_transmit = bridge_transmit;
664 	ifp->if_qflush = bridge_qflush;
665 	ifp->if_init = bridge_init;
666 	ifp->if_type = IFT_BRIDGE;
667 
668 	/*
669 	 * Generate an ethernet address with a locally administered address.
670 	 *
671 	 * Since we are using random ethernet addresses for the bridge, it is
672 	 * possible that we might have address collisions, so make sure that
673 	 * this hardware address isn't already in use on another bridge.
674 	 * The first try uses the hostid and falls back to arc4rand().
675 	 */
676 	fb = 0;
677 	getcredhostid(curthread->td_ucred, &hostid);
678 	do {
679 		if (fb || hostid == 0) {
680 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
681 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
682 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
683 		} else {
684 			sc->sc_defaddr[0] = 0x2;
685 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
686 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
687 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
688 			sc->sc_defaddr[4] =  hostid        & 0xff;
689 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
690 		}
691 
692 		fb = 1;
693 		retry = 0;
694 		BRIDGE_LIST_LOCK();
695 		LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
696 			bifp = sc2->sc_ifp;
697 			if (memcmp(sc->sc_defaddr,
698 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
699 				retry = 1;
700 				break;
701 			}
702 		}
703 		BRIDGE_LIST_UNLOCK();
704 	} while (retry == 1);
705 
706 	bstp_attach(&sc->sc_stp, &bridge_ops);
707 	ether_ifattach(ifp, sc->sc_defaddr);
708 	/* Now undo some of the damage... */
709 	ifp->if_baudrate = 0;
710 	ifp->if_type = IFT_BRIDGE;
711 
712 	BRIDGE_LIST_LOCK();
713 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
714 	BRIDGE_LIST_UNLOCK();
715 
716 	return (0);
717 }
718 
719 /*
720  * bridge_clone_destroy:
721  *
722  *	Destroy a bridge instance.
723  */
724 static void
725 bridge_clone_destroy(struct ifnet *ifp)
726 {
727 	struct bridge_softc *sc = ifp->if_softc;
728 	struct bridge_iflist *bif;
729 
730 	BRIDGE_LOCK(sc);
731 
732 	bridge_stop(ifp, 1);
733 	ifp->if_flags &= ~IFF_UP;
734 
735 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
736 		bridge_delete_member(sc, bif, 0);
737 
738 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
739 		bridge_delete_span(sc, bif);
740 	}
741 
742 	BRIDGE_UNLOCK(sc);
743 
744 	callout_drain(&sc->sc_brcallout);
745 
746 	BRIDGE_LIST_LOCK();
747 	LIST_REMOVE(sc, sc_list);
748 	BRIDGE_LIST_UNLOCK();
749 
750 	bstp_detach(&sc->sc_stp);
751 	ether_ifdetach(ifp);
752 	if_free(ifp);
753 
754 	/* Tear down the routing table. */
755 	bridge_rtable_fini(sc);
756 
757 	BRIDGE_LOCK_DESTROY(sc);
758 	free(sc, M_DEVBUF);
759 }
760 
761 /*
762  * bridge_ioctl:
763  *
764  *	Handle a control request from the operator.
765  */
766 static int
767 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
768 {
769 	struct bridge_softc *sc = ifp->if_softc;
770 	struct ifreq *ifr = (struct ifreq *)data;
771 	struct bridge_iflist *bif;
772 	struct thread *td = curthread;
773 	union {
774 		struct ifbreq ifbreq;
775 		struct ifbifconf ifbifconf;
776 		struct ifbareq ifbareq;
777 		struct ifbaconf ifbaconf;
778 		struct ifbrparam ifbrparam;
779 		struct ifbropreq ifbropreq;
780 	} args;
781 	struct ifdrv *ifd = (struct ifdrv *) data;
782 	const struct bridge_control *bc;
783 	int error = 0;
784 
785 	switch (cmd) {
786 
787 	case SIOCADDMULTI:
788 	case SIOCDELMULTI:
789 		break;
790 
791 	case SIOCGDRVSPEC:
792 	case SIOCSDRVSPEC:
793 		if (ifd->ifd_cmd >= bridge_control_table_size) {
794 			error = EINVAL;
795 			break;
796 		}
797 		bc = &bridge_control_table[ifd->ifd_cmd];
798 
799 		if (cmd == SIOCGDRVSPEC &&
800 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
801 			error = EINVAL;
802 			break;
803 		}
804 		else if (cmd == SIOCSDRVSPEC &&
805 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
806 			error = EINVAL;
807 			break;
808 		}
809 
810 		if (bc->bc_flags & BC_F_SUSER) {
811 			error = priv_check(td, PRIV_NET_BRIDGE);
812 			if (error)
813 				break;
814 		}
815 
816 		if (ifd->ifd_len != bc->bc_argsize ||
817 		    ifd->ifd_len > sizeof(args)) {
818 			error = EINVAL;
819 			break;
820 		}
821 
822 		bzero(&args, sizeof(args));
823 		if (bc->bc_flags & BC_F_COPYIN) {
824 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
825 			if (error)
826 				break;
827 		}
828 
829 		BRIDGE_LOCK(sc);
830 		error = (*bc->bc_func)(sc, &args);
831 		BRIDGE_UNLOCK(sc);
832 		if (error)
833 			break;
834 
835 		if (bc->bc_flags & BC_F_COPYOUT)
836 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
837 
838 		break;
839 
840 	case SIOCSIFFLAGS:
841 		if (!(ifp->if_flags & IFF_UP) &&
842 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
843 			/*
844 			 * If interface is marked down and it is running,
845 			 * then stop and disable it.
846 			 */
847 			BRIDGE_LOCK(sc);
848 			bridge_stop(ifp, 1);
849 			BRIDGE_UNLOCK(sc);
850 		} else if ((ifp->if_flags & IFF_UP) &&
851 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
852 			/*
853 			 * If interface is marked up and it is stopped, then
854 			 * start it.
855 			 */
856 			(*ifp->if_init)(sc);
857 		}
858 		break;
859 
860 	case SIOCSIFMTU:
861 		if (ifr->ifr_mtu < 576) {
862 			error = EINVAL;
863 			break;
864 		}
865 		if (LIST_EMPTY(&sc->sc_iflist)) {
866 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
867 			break;
868 		}
869 		BRIDGE_LOCK(sc);
870 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
871 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
872 				log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
873 				    " != %d\n", sc->sc_ifp->if_xname,
874 				    bif->bif_ifp->if_mtu,
875 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
876 				error = EINVAL;
877 				break;
878 			}
879 		}
880 		if (!error)
881 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
882 		BRIDGE_UNLOCK(sc);
883 		break;
884 	default:
885 		/*
886 		 * drop the lock as ether_ioctl() will call bridge_start() and
887 		 * cause the lock to be recursed.
888 		 */
889 		error = ether_ioctl(ifp, cmd, data);
890 		break;
891 	}
892 
893 	return (error);
894 }
895 
896 /*
897  * bridge_mutecaps:
898  *
899  *	Clear or restore unwanted capabilities on the member interface
900  */
901 static void
902 bridge_mutecaps(struct bridge_softc *sc)
903 {
904 	struct bridge_iflist *bif;
905 	int enabled, mask;
906 
907 	/* Initial bitmask of capabilities to test */
908 	mask = BRIDGE_IFCAPS_MASK;
909 
910 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
911 		/* Every member must support it or its disabled */
912 		mask &= bif->bif_savedcaps;
913 	}
914 
915 	BRIDGE_XLOCK(sc);
916 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
917 		enabled = bif->bif_ifp->if_capenable;
918 		enabled &= ~BRIDGE_IFCAPS_STRIP;
919 		/* strip off mask bits and enable them again if allowed */
920 		enabled &= ~BRIDGE_IFCAPS_MASK;
921 		enabled |= mask;
922 		BRIDGE_UNLOCK(sc);
923 		bridge_set_ifcap(sc, bif, enabled);
924 		BRIDGE_LOCK(sc);
925 	}
926 	BRIDGE_XDROP(sc);
927 
928 }
929 
930 static void
931 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
932 {
933 	struct ifnet *ifp = bif->bif_ifp;
934 	struct ifreq ifr;
935 	int error;
936 
937 	BRIDGE_UNLOCK_ASSERT(sc);
938 
939 	bzero(&ifr, sizeof(ifr));
940 	ifr.ifr_reqcap = set;
941 
942 	if (ifp->if_capenable != set) {
943 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
944 		if (error)
945 			if_printf(sc->sc_ifp,
946 			    "error setting capabilities on %s: %d\n",
947 			    ifp->if_xname, error);
948 		if ((ifp->if_capenable & ~set) != 0)
949 			if_printf(sc->sc_ifp,
950 			    "can't disable some capabilities on %s: 0x%x\n",
951 			    ifp->if_xname, ifp->if_capenable & ~set);
952 	}
953 }
954 
955 /*
956  * bridge_lookup_member:
957  *
958  *	Lookup a bridge member interface.
959  */
960 static struct bridge_iflist *
961 bridge_lookup_member(struct bridge_softc *sc, const char *name)
962 {
963 	struct bridge_iflist *bif;
964 	struct ifnet *ifp;
965 
966 	BRIDGE_LOCK_ASSERT(sc);
967 
968 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
969 		ifp = bif->bif_ifp;
970 		if (strcmp(ifp->if_xname, name) == 0)
971 			return (bif);
972 	}
973 
974 	return (NULL);
975 }
976 
977 /*
978  * bridge_lookup_member_if:
979  *
980  *	Lookup a bridge member interface by ifnet*.
981  */
982 static struct bridge_iflist *
983 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
984 {
985 	struct bridge_iflist *bif;
986 
987 	BRIDGE_LOCK_ASSERT(sc);
988 
989 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
990 		if (bif->bif_ifp == member_ifp)
991 			return (bif);
992 	}
993 
994 	return (NULL);
995 }
996 
997 /*
998  * bridge_delete_member:
999  *
1000  *	Delete the specified member interface.
1001  */
1002 static void
1003 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
1004     int gone)
1005 {
1006 	struct ifnet *ifs = bif->bif_ifp;
1007 	struct ifnet *fif = NULL;
1008 
1009 	BRIDGE_LOCK_ASSERT(sc);
1010 
1011 	if (bif->bif_flags & IFBIF_STP)
1012 		bstp_disable(&bif->bif_stp);
1013 
1014 	ifs->if_bridge = NULL;
1015 	BRIDGE_XLOCK(sc);
1016 	LIST_REMOVE(bif, bif_next);
1017 	BRIDGE_XDROP(sc);
1018 
1019 	/*
1020 	 * If removing the interface that gave the bridge its mac address, set
1021 	 * the mac address of the bridge to the address of the next member, or
1022 	 * to its default address if no members are left.
1023 	 */
1024 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1025 		if (LIST_EMPTY(&sc->sc_iflist)) {
1026 			bcopy(sc->sc_defaddr,
1027 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1028 			sc->sc_ifaddr = NULL;
1029 		} else {
1030 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
1031 			bcopy(IF_LLADDR(fif),
1032 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1033 			sc->sc_ifaddr = fif;
1034 		}
1035 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1036 	}
1037 
1038 	bridge_linkcheck(sc);
1039 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1040 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1041 	KASSERT(bif->bif_addrcnt == 0,
1042 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1043 
1044 	BRIDGE_UNLOCK(sc);
1045 	if (!gone) {
1046 		switch (ifs->if_type) {
1047 		case IFT_ETHER:
1048 		case IFT_L2VLAN:
1049 			/*
1050 			 * Take the interface out of promiscuous mode, but only
1051 			 * if it was promiscuous in the first place. It might
1052 			 * not be if we're in the bridge_ioctl_add() error path.
1053 			 */
1054 			if (ifs->if_flags & IFF_PROMISC)
1055 				(void) ifpromisc(ifs, 0);
1056 			break;
1057 
1058 		case IFT_GIF:
1059 			break;
1060 
1061 		default:
1062 #ifdef DIAGNOSTIC
1063 			panic("bridge_delete_member: impossible");
1064 #endif
1065 			break;
1066 		}
1067 		/* reneable any interface capabilities */
1068 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1069 	}
1070 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1071 	BRIDGE_LOCK(sc);
1072 	free(bif, M_DEVBUF);
1073 }
1074 
1075 /*
1076  * bridge_delete_span:
1077  *
1078  *	Delete the specified span interface.
1079  */
1080 static void
1081 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1082 {
1083 	BRIDGE_LOCK_ASSERT(sc);
1084 
1085 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1086 	    ("%s: not a span interface", __func__));
1087 
1088 	LIST_REMOVE(bif, bif_next);
1089 	free(bif, M_DEVBUF);
1090 }
1091 
1092 static int
1093 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1094 {
1095 	struct ifbreq *req = arg;
1096 	struct bridge_iflist *bif = NULL;
1097 	struct ifnet *ifs;
1098 	int error = 0;
1099 
1100 	ifs = ifunit(req->ifbr_ifsname);
1101 	if (ifs == NULL)
1102 		return (ENOENT);
1103 	if (ifs->if_ioctl == NULL)	/* must be supported */
1104 		return (EINVAL);
1105 
1106 	/* If it's in the span list, it can't be a member. */
1107 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1108 		if (ifs == bif->bif_ifp)
1109 			return (EBUSY);
1110 
1111 	if (ifs->if_bridge == sc)
1112 		return (EEXIST);
1113 
1114 	if (ifs->if_bridge != NULL)
1115 		return (EBUSY);
1116 
1117 	switch (ifs->if_type) {
1118 	case IFT_ETHER:
1119 	case IFT_L2VLAN:
1120 	case IFT_GIF:
1121 		/* permitted interface types */
1122 		break;
1123 	default:
1124 		return (EINVAL);
1125 	}
1126 
1127 #ifdef INET6
1128 	/*
1129 	 * Two valid inet6 addresses with link-local scope must not be
1130 	 * on the parent interface and the member interfaces at the
1131 	 * same time.  This restriction is needed to prevent violation
1132 	 * of link-local scope zone.  Attempts to add a member
1133 	 * interface which has inet6 addresses when the parent has
1134 	 * inet6 triggers removal of all inet6 addresses on the member
1135 	 * interface.
1136 	 */
1137 
1138 	/* Check if the parent interface has a link-local scope addr. */
1139 	if (V_allow_llz_overlap == 0 &&
1140 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1141 		/*
1142 		 * If any, remove all inet6 addresses from the member
1143 		 * interfaces.
1144 		 */
1145 		BRIDGE_XLOCK(sc);
1146 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1147  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1148 				BRIDGE_UNLOCK(sc);
1149 				in6_ifdetach(bif->bif_ifp);
1150 				BRIDGE_LOCK(sc);
1151 				if_printf(sc->sc_ifp,
1152 				    "IPv6 addresses on %s have been removed "
1153 				    "before adding it as a member to prevent "
1154 				    "IPv6 address scope violation.\n",
1155 				    bif->bif_ifp->if_xname);
1156 			}
1157 		}
1158 		BRIDGE_XDROP(sc);
1159 		if (in6ifa_llaonifp(ifs)) {
1160 			BRIDGE_UNLOCK(sc);
1161 			in6_ifdetach(ifs);
1162 			BRIDGE_LOCK(sc);
1163 			if_printf(sc->sc_ifp,
1164 			    "IPv6 addresses on %s have been removed "
1165 			    "before adding it as a member to prevent "
1166 			    "IPv6 address scope violation.\n",
1167 			    ifs->if_xname);
1168 		}
1169 	}
1170 #endif
1171 	/* Allow the first Ethernet member to define the MTU */
1172 	if (LIST_EMPTY(&sc->sc_iflist))
1173 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1174 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1175 		if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1176 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1177 		return (EINVAL);
1178 	}
1179 
1180 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1181 	if (bif == NULL)
1182 		return (ENOMEM);
1183 
1184 	bif->bif_ifp = ifs;
1185 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1186 	bif->bif_savedcaps = ifs->if_capenable;
1187 
1188 	/*
1189 	 * Assign the interface's MAC address to the bridge if it's the first
1190 	 * member and the MAC address of the bridge has not been changed from
1191 	 * the default randomly generated one.
1192 	 */
1193 	if (V_bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1194 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1195 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1196 		sc->sc_ifaddr = ifs;
1197 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1198 	}
1199 
1200 	ifs->if_bridge = sc;
1201 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1202 	/*
1203 	 * XXX: XLOCK HERE!?!
1204 	 *
1205 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1206 	 */
1207 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1208 
1209 	/* Set interface capabilities to the intersection set of all members */
1210 	bridge_mutecaps(sc);
1211 	bridge_linkcheck(sc);
1212 
1213 	/* Place the interface into promiscuous mode */
1214 	switch (ifs->if_type) {
1215 		case IFT_ETHER:
1216 		case IFT_L2VLAN:
1217 			BRIDGE_UNLOCK(sc);
1218 			error = ifpromisc(ifs, 1);
1219 			BRIDGE_LOCK(sc);
1220 			break;
1221 	}
1222 
1223 	if (error)
1224 		bridge_delete_member(sc, bif, 0);
1225 	return (error);
1226 }
1227 
1228 static int
1229 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1230 {
1231 	struct ifbreq *req = arg;
1232 	struct bridge_iflist *bif;
1233 
1234 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1235 	if (bif == NULL)
1236 		return (ENOENT);
1237 
1238 	bridge_delete_member(sc, bif, 0);
1239 
1240 	return (0);
1241 }
1242 
1243 static int
1244 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1245 {
1246 	struct ifbreq *req = arg;
1247 	struct bridge_iflist *bif;
1248 	struct bstp_port *bp;
1249 
1250 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1251 	if (bif == NULL)
1252 		return (ENOENT);
1253 
1254 	bp = &bif->bif_stp;
1255 	req->ifbr_ifsflags = bif->bif_flags;
1256 	req->ifbr_state = bp->bp_state;
1257 	req->ifbr_priority = bp->bp_priority;
1258 	req->ifbr_path_cost = bp->bp_path_cost;
1259 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1260 	req->ifbr_proto = bp->bp_protover;
1261 	req->ifbr_role = bp->bp_role;
1262 	req->ifbr_stpflags = bp->bp_flags;
1263 	req->ifbr_addrcnt = bif->bif_addrcnt;
1264 	req->ifbr_addrmax = bif->bif_addrmax;
1265 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1266 
1267 	/* Copy STP state options as flags */
1268 	if (bp->bp_operedge)
1269 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1270 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1271 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1272 	if (bp->bp_ptp_link)
1273 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1274 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1275 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1276 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1277 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1278 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1279 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1280 	return (0);
1281 }
1282 
1283 static int
1284 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1285 {
1286 	struct ifbreq *req = arg;
1287 	struct bridge_iflist *bif;
1288 	struct bstp_port *bp;
1289 	int error;
1290 
1291 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1292 	if (bif == NULL)
1293 		return (ENOENT);
1294 	bp = &bif->bif_stp;
1295 
1296 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1297 		/* SPAN is readonly */
1298 		return (EINVAL);
1299 
1300 	if (req->ifbr_ifsflags & IFBIF_STP) {
1301 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1302 			error = bstp_enable(&bif->bif_stp);
1303 			if (error)
1304 				return (error);
1305 		}
1306 	} else {
1307 		if ((bif->bif_flags & IFBIF_STP) != 0)
1308 			bstp_disable(&bif->bif_stp);
1309 	}
1310 
1311 	/* Pass on STP flags */
1312 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1313 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1314 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1315 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1316 
1317 	/* Save the bits relating to the bridge */
1318 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1319 
1320 	return (0);
1321 }
1322 
1323 static int
1324 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1325 {
1326 	struct ifbrparam *param = arg;
1327 
1328 	sc->sc_brtmax = param->ifbrp_csize;
1329 	bridge_rttrim(sc);
1330 
1331 	return (0);
1332 }
1333 
1334 static int
1335 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1336 {
1337 	struct ifbrparam *param = arg;
1338 
1339 	param->ifbrp_csize = sc->sc_brtmax;
1340 
1341 	return (0);
1342 }
1343 
1344 static int
1345 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1346 {
1347 	struct ifbifconf *bifc = arg;
1348 	struct bridge_iflist *bif;
1349 	struct ifbreq breq;
1350 	char *buf, *outbuf;
1351 	int count, buflen, len, error = 0;
1352 
1353 	count = 0;
1354 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1355 		count++;
1356 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1357 		count++;
1358 
1359 	buflen = sizeof(breq) * count;
1360 	if (bifc->ifbic_len == 0) {
1361 		bifc->ifbic_len = buflen;
1362 		return (0);
1363 	}
1364 	BRIDGE_UNLOCK(sc);
1365 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1366 	BRIDGE_LOCK(sc);
1367 
1368 	count = 0;
1369 	buf = outbuf;
1370 	len = min(bifc->ifbic_len, buflen);
1371 	bzero(&breq, sizeof(breq));
1372 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1373 		if (len < sizeof(breq))
1374 			break;
1375 
1376 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1377 		    sizeof(breq.ifbr_ifsname));
1378 		/* Fill in the ifbreq structure */
1379 		error = bridge_ioctl_gifflags(sc, &breq);
1380 		if (error)
1381 			break;
1382 		memcpy(buf, &breq, sizeof(breq));
1383 		count++;
1384 		buf += sizeof(breq);
1385 		len -= sizeof(breq);
1386 	}
1387 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1388 		if (len < sizeof(breq))
1389 			break;
1390 
1391 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1392 		    sizeof(breq.ifbr_ifsname));
1393 		breq.ifbr_ifsflags = bif->bif_flags;
1394 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1395 		memcpy(buf, &breq, sizeof(breq));
1396 		count++;
1397 		buf += sizeof(breq);
1398 		len -= sizeof(breq);
1399 	}
1400 
1401 	BRIDGE_UNLOCK(sc);
1402 	bifc->ifbic_len = sizeof(breq) * count;
1403 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1404 	BRIDGE_LOCK(sc);
1405 	free(outbuf, M_TEMP);
1406 	return (error);
1407 }
1408 
1409 static int
1410 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1411 {
1412 	struct ifbaconf *bac = arg;
1413 	struct bridge_rtnode *brt;
1414 	struct ifbareq bareq;
1415 	char *buf, *outbuf;
1416 	int count, buflen, len, error = 0;
1417 
1418 	if (bac->ifbac_len == 0)
1419 		return (0);
1420 
1421 	count = 0;
1422 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1423 		count++;
1424 	buflen = sizeof(bareq) * count;
1425 
1426 	BRIDGE_UNLOCK(sc);
1427 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1428 	BRIDGE_LOCK(sc);
1429 
1430 	count = 0;
1431 	buf = outbuf;
1432 	len = min(bac->ifbac_len, buflen);
1433 	bzero(&bareq, sizeof(bareq));
1434 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1435 		if (len < sizeof(bareq))
1436 			goto out;
1437 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1438 		    sizeof(bareq.ifba_ifsname));
1439 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1440 		bareq.ifba_vlan = brt->brt_vlan;
1441 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1442 				time_uptime < brt->brt_expire)
1443 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1444 		else
1445 			bareq.ifba_expire = 0;
1446 		bareq.ifba_flags = brt->brt_flags;
1447 
1448 		memcpy(buf, &bareq, sizeof(bareq));
1449 		count++;
1450 		buf += sizeof(bareq);
1451 		len -= sizeof(bareq);
1452 	}
1453 out:
1454 	BRIDGE_UNLOCK(sc);
1455 	bac->ifbac_len = sizeof(bareq) * count;
1456 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1457 	BRIDGE_LOCK(sc);
1458 	free(outbuf, M_TEMP);
1459 	return (error);
1460 }
1461 
1462 static int
1463 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1464 {
1465 	struct ifbareq *req = arg;
1466 	struct bridge_iflist *bif;
1467 	int error;
1468 
1469 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1470 	if (bif == NULL)
1471 		return (ENOENT);
1472 
1473 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1474 	    req->ifba_flags);
1475 
1476 	return (error);
1477 }
1478 
1479 static int
1480 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1481 {
1482 	struct ifbrparam *param = arg;
1483 
1484 	sc->sc_brttimeout = param->ifbrp_ctime;
1485 	return (0);
1486 }
1487 
1488 static int
1489 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1490 {
1491 	struct ifbrparam *param = arg;
1492 
1493 	param->ifbrp_ctime = sc->sc_brttimeout;
1494 	return (0);
1495 }
1496 
1497 static int
1498 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1499 {
1500 	struct ifbareq *req = arg;
1501 
1502 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1503 }
1504 
1505 static int
1506 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1507 {
1508 	struct ifbreq *req = arg;
1509 
1510 	bridge_rtflush(sc, req->ifbr_ifsflags);
1511 	return (0);
1512 }
1513 
1514 static int
1515 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1516 {
1517 	struct ifbrparam *param = arg;
1518 	struct bstp_state *bs = &sc->sc_stp;
1519 
1520 	param->ifbrp_prio = bs->bs_bridge_priority;
1521 	return (0);
1522 }
1523 
1524 static int
1525 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1526 {
1527 	struct ifbrparam *param = arg;
1528 
1529 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1530 }
1531 
1532 static int
1533 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1534 {
1535 	struct ifbrparam *param = arg;
1536 	struct bstp_state *bs = &sc->sc_stp;
1537 
1538 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1539 	return (0);
1540 }
1541 
1542 static int
1543 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1544 {
1545 	struct ifbrparam *param = arg;
1546 
1547 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1548 }
1549 
1550 static int
1551 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1552 {
1553 	struct ifbrparam *param = arg;
1554 	struct bstp_state *bs = &sc->sc_stp;
1555 
1556 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1557 	return (0);
1558 }
1559 
1560 static int
1561 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1562 {
1563 	struct ifbrparam *param = arg;
1564 
1565 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1566 }
1567 
1568 static int
1569 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1570 {
1571 	struct ifbrparam *param = arg;
1572 	struct bstp_state *bs = &sc->sc_stp;
1573 
1574 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1575 	return (0);
1576 }
1577 
1578 static int
1579 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1580 {
1581 	struct ifbrparam *param = arg;
1582 
1583 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1584 }
1585 
1586 static int
1587 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1588 {
1589 	struct ifbreq *req = arg;
1590 	struct bridge_iflist *bif;
1591 
1592 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1593 	if (bif == NULL)
1594 		return (ENOENT);
1595 
1596 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1597 }
1598 
1599 static int
1600 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1601 {
1602 	struct ifbreq *req = arg;
1603 	struct bridge_iflist *bif;
1604 
1605 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1606 	if (bif == NULL)
1607 		return (ENOENT);
1608 
1609 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1610 }
1611 
1612 static int
1613 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1614 {
1615 	struct ifbreq *req = arg;
1616 	struct bridge_iflist *bif;
1617 
1618 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1619 	if (bif == NULL)
1620 		return (ENOENT);
1621 
1622 	bif->bif_addrmax = req->ifbr_addrmax;
1623 	return (0);
1624 }
1625 
1626 static int
1627 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1628 {
1629 	struct ifbreq *req = arg;
1630 	struct bridge_iflist *bif = NULL;
1631 	struct ifnet *ifs;
1632 
1633 	ifs = ifunit(req->ifbr_ifsname);
1634 	if (ifs == NULL)
1635 		return (ENOENT);
1636 
1637 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1638 		if (ifs == bif->bif_ifp)
1639 			return (EBUSY);
1640 
1641 	if (ifs->if_bridge != NULL)
1642 		return (EBUSY);
1643 
1644 	switch (ifs->if_type) {
1645 		case IFT_ETHER:
1646 		case IFT_GIF:
1647 		case IFT_L2VLAN:
1648 			break;
1649 		default:
1650 			return (EINVAL);
1651 	}
1652 
1653 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1654 	if (bif == NULL)
1655 		return (ENOMEM);
1656 
1657 	bif->bif_ifp = ifs;
1658 	bif->bif_flags = IFBIF_SPAN;
1659 
1660 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1661 
1662 	return (0);
1663 }
1664 
1665 static int
1666 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1667 {
1668 	struct ifbreq *req = arg;
1669 	struct bridge_iflist *bif;
1670 	struct ifnet *ifs;
1671 
1672 	ifs = ifunit(req->ifbr_ifsname);
1673 	if (ifs == NULL)
1674 		return (ENOENT);
1675 
1676 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1677 		if (ifs == bif->bif_ifp)
1678 			break;
1679 
1680 	if (bif == NULL)
1681 		return (ENOENT);
1682 
1683 	bridge_delete_span(sc, bif);
1684 
1685 	return (0);
1686 }
1687 
1688 static int
1689 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1690 {
1691 	struct ifbropreq *req = arg;
1692 	struct bstp_state *bs = &sc->sc_stp;
1693 	struct bstp_port *root_port;
1694 
1695 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1696 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1697 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1698 
1699 	root_port = bs->bs_root_port;
1700 	if (root_port == NULL)
1701 		req->ifbop_root_port = 0;
1702 	else
1703 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1704 
1705 	req->ifbop_holdcount = bs->bs_txholdcount;
1706 	req->ifbop_priority = bs->bs_bridge_priority;
1707 	req->ifbop_protocol = bs->bs_protover;
1708 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1709 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1710 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1711 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1712 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1713 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1714 
1715 	return (0);
1716 }
1717 
1718 static int
1719 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1720 {
1721 	struct ifbrparam *param = arg;
1722 
1723 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1724 	return (0);
1725 }
1726 
1727 static int
1728 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1729 {
1730 	struct ifbpstpconf *bifstp = arg;
1731 	struct bridge_iflist *bif;
1732 	struct bstp_port *bp;
1733 	struct ifbpstpreq bpreq;
1734 	char *buf, *outbuf;
1735 	int count, buflen, len, error = 0;
1736 
1737 	count = 0;
1738 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1739 		if ((bif->bif_flags & IFBIF_STP) != 0)
1740 			count++;
1741 	}
1742 
1743 	buflen = sizeof(bpreq) * count;
1744 	if (bifstp->ifbpstp_len == 0) {
1745 		bifstp->ifbpstp_len = buflen;
1746 		return (0);
1747 	}
1748 
1749 	BRIDGE_UNLOCK(sc);
1750 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1751 	BRIDGE_LOCK(sc);
1752 
1753 	count = 0;
1754 	buf = outbuf;
1755 	len = min(bifstp->ifbpstp_len, buflen);
1756 	bzero(&bpreq, sizeof(bpreq));
1757 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1758 		if (len < sizeof(bpreq))
1759 			break;
1760 
1761 		if ((bif->bif_flags & IFBIF_STP) == 0)
1762 			continue;
1763 
1764 		bp = &bif->bif_stp;
1765 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1766 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1767 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1768 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1769 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1770 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1771 
1772 		memcpy(buf, &bpreq, sizeof(bpreq));
1773 		count++;
1774 		buf += sizeof(bpreq);
1775 		len -= sizeof(bpreq);
1776 	}
1777 
1778 	BRIDGE_UNLOCK(sc);
1779 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1780 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1781 	BRIDGE_LOCK(sc);
1782 	free(outbuf, M_TEMP);
1783 	return (error);
1784 }
1785 
1786 static int
1787 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1788 {
1789 	struct ifbrparam *param = arg;
1790 
1791 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1792 }
1793 
1794 static int
1795 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1796 {
1797 	struct ifbrparam *param = arg;
1798 
1799 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1800 }
1801 
1802 /*
1803  * bridge_ifdetach:
1804  *
1805  *	Detach an interface from a bridge.  Called when a member
1806  *	interface is detaching.
1807  */
1808 static void
1809 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1810 {
1811 	struct bridge_softc *sc = ifp->if_bridge;
1812 	struct bridge_iflist *bif;
1813 
1814 	if (ifp->if_flags & IFF_RENAMING)
1815 		return;
1816 	if (V_bridge_cloner == NULL) {
1817 		/*
1818 		 * This detach handler can be called after
1819 		 * vnet_bridge_uninit().  Just return in that case.
1820 		 */
1821 		return;
1822 	}
1823 	/* Check if the interface is a bridge member */
1824 	if (sc != NULL) {
1825 		BRIDGE_LOCK(sc);
1826 
1827 		bif = bridge_lookup_member_if(sc, ifp);
1828 		if (bif != NULL)
1829 			bridge_delete_member(sc, bif, 1);
1830 
1831 		BRIDGE_UNLOCK(sc);
1832 		return;
1833 	}
1834 
1835 	/* Check if the interface is a span port */
1836 	BRIDGE_LIST_LOCK();
1837 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1838 		BRIDGE_LOCK(sc);
1839 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1840 			if (ifp == bif->bif_ifp) {
1841 				bridge_delete_span(sc, bif);
1842 				break;
1843 			}
1844 
1845 		BRIDGE_UNLOCK(sc);
1846 	}
1847 	BRIDGE_LIST_UNLOCK();
1848 }
1849 
1850 /*
1851  * bridge_init:
1852  *
1853  *	Initialize a bridge interface.
1854  */
1855 static void
1856 bridge_init(void *xsc)
1857 {
1858 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1859 	struct ifnet *ifp = sc->sc_ifp;
1860 
1861 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1862 		return;
1863 
1864 	BRIDGE_LOCK(sc);
1865 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1866 	    bridge_timer, sc);
1867 
1868 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1869 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1870 
1871 	BRIDGE_UNLOCK(sc);
1872 }
1873 
1874 /*
1875  * bridge_stop:
1876  *
1877  *	Stop the bridge interface.
1878  */
1879 static void
1880 bridge_stop(struct ifnet *ifp, int disable)
1881 {
1882 	struct bridge_softc *sc = ifp->if_softc;
1883 
1884 	BRIDGE_LOCK_ASSERT(sc);
1885 
1886 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1887 		return;
1888 
1889 	callout_stop(&sc->sc_brcallout);
1890 	bstp_stop(&sc->sc_stp);
1891 
1892 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1893 
1894 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1895 }
1896 
1897 /*
1898  * bridge_enqueue:
1899  *
1900  *	Enqueue a packet on a bridge member interface.
1901  *
1902  */
1903 static int
1904 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1905 {
1906 	int len, err = 0;
1907 	short mflags;
1908 	struct mbuf *m0;
1909 
1910 	/* We may be sending a fragment so traverse the mbuf */
1911 	for (; m; m = m0) {
1912 		m0 = m->m_nextpkt;
1913 		m->m_nextpkt = NULL;
1914 		len = m->m_pkthdr.len;
1915 		mflags = m->m_flags;
1916 
1917 		/*
1918 		 * If underlying interface can not do VLAN tag insertion itself
1919 		 * then attach a packet tag that holds it.
1920 		 */
1921 		if ((m->m_flags & M_VLANTAG) &&
1922 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1923 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1924 			if (m == NULL) {
1925 				if_printf(dst_ifp,
1926 				    "unable to prepend VLAN header\n");
1927 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1928 				continue;
1929 			}
1930 			m->m_flags &= ~M_VLANTAG;
1931 		}
1932 
1933 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1934 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1935 			m_freem(m0);
1936 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1937 			break;
1938 		}
1939 
1940 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1941 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1942 		if (mflags & M_MCAST)
1943 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
1944 	}
1945 
1946 	return (err);
1947 }
1948 
1949 /*
1950  * bridge_dummynet:
1951  *
1952  * 	Receive a queued packet from dummynet and pass it on to the output
1953  * 	interface.
1954  *
1955  *	The mbuf has the Ethernet header already attached.
1956  */
1957 static void
1958 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1959 {
1960 	struct bridge_softc *sc;
1961 
1962 	sc = ifp->if_bridge;
1963 
1964 	/*
1965 	 * The packet didnt originate from a member interface. This should only
1966 	 * ever happen if a member interface is removed while packets are
1967 	 * queued for it.
1968 	 */
1969 	if (sc == NULL) {
1970 		m_freem(m);
1971 		return;
1972 	}
1973 
1974 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1975 #ifdef INET6
1976 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1977 #endif
1978 	    ) {
1979 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1980 			return;
1981 		if (m == NULL)
1982 			return;
1983 	}
1984 
1985 	bridge_enqueue(sc, ifp, m);
1986 }
1987 
1988 /*
1989  * bridge_output:
1990  *
1991  *	Send output from a bridge member interface.  This
1992  *	performs the bridging function for locally originated
1993  *	packets.
1994  *
1995  *	The mbuf has the Ethernet header already attached.  We must
1996  *	enqueue or free the mbuf before returning.
1997  */
1998 static int
1999 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2000     struct rtentry *rt)
2001 {
2002 	struct ether_header *eh;
2003 	struct ifnet *dst_if;
2004 	struct bridge_softc *sc;
2005 	uint16_t vlan;
2006 
2007 	if (m->m_len < ETHER_HDR_LEN) {
2008 		m = m_pullup(m, ETHER_HDR_LEN);
2009 		if (m == NULL)
2010 			return (0);
2011 	}
2012 
2013 	eh = mtod(m, struct ether_header *);
2014 	sc = ifp->if_bridge;
2015 	vlan = VLANTAGOF(m);
2016 
2017 	BRIDGE_LOCK(sc);
2018 
2019 	/*
2020 	 * If bridge is down, but the original output interface is up,
2021 	 * go ahead and send out that interface.  Otherwise, the packet
2022 	 * is dropped below.
2023 	 */
2024 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2025 		dst_if = ifp;
2026 		goto sendunicast;
2027 	}
2028 
2029 	/*
2030 	 * If the packet is a multicast, or we don't know a better way to
2031 	 * get there, send to all interfaces.
2032 	 */
2033 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2034 		dst_if = NULL;
2035 	else
2036 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2037 	if (dst_if == NULL) {
2038 		struct bridge_iflist *bif;
2039 		struct mbuf *mc;
2040 		int error = 0, used = 0;
2041 
2042 		bridge_span(sc, m);
2043 
2044 		BRIDGE_LOCK2REF(sc, error);
2045 		if (error) {
2046 			m_freem(m);
2047 			return (0);
2048 		}
2049 
2050 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2051 			dst_if = bif->bif_ifp;
2052 
2053 			if (dst_if->if_type == IFT_GIF)
2054 				continue;
2055 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2056 				continue;
2057 
2058 			/*
2059 			 * If this is not the original output interface,
2060 			 * and the interface is participating in spanning
2061 			 * tree, make sure the port is in a state that
2062 			 * allows forwarding.
2063 			 */
2064 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2065 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2066 				continue;
2067 
2068 			if (LIST_NEXT(bif, bif_next) == NULL) {
2069 				used = 1;
2070 				mc = m;
2071 			} else {
2072 				mc = m_copypacket(m, M_NOWAIT);
2073 				if (mc == NULL) {
2074 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2075 					continue;
2076 				}
2077 			}
2078 
2079 			bridge_enqueue(sc, dst_if, mc);
2080 		}
2081 		if (used == 0)
2082 			m_freem(m);
2083 		BRIDGE_UNREF(sc);
2084 		return (0);
2085 	}
2086 
2087 sendunicast:
2088 	/*
2089 	 * XXX Spanning tree consideration here?
2090 	 */
2091 
2092 	bridge_span(sc, m);
2093 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2094 		m_freem(m);
2095 		BRIDGE_UNLOCK(sc);
2096 		return (0);
2097 	}
2098 
2099 	BRIDGE_UNLOCK(sc);
2100 	bridge_enqueue(sc, dst_if, m);
2101 	return (0);
2102 }
2103 
2104 /*
2105  * bridge_transmit:
2106  *
2107  *	Do output on a bridge.
2108  *
2109  */
2110 static int
2111 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2112 {
2113 	struct bridge_softc *sc;
2114 	struct ether_header *eh;
2115 	struct ifnet *dst_if;
2116 	int error = 0;
2117 
2118 	sc = ifp->if_softc;
2119 
2120 	ETHER_BPF_MTAP(ifp, m);
2121 
2122 	eh = mtod(m, struct ether_header *);
2123 
2124 	BRIDGE_LOCK(sc);
2125 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2126 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2127 		BRIDGE_UNLOCK(sc);
2128 		error = bridge_enqueue(sc, dst_if, m);
2129 	} else
2130 		bridge_broadcast(sc, ifp, m, 0);
2131 
2132 	return (error);
2133 }
2134 
2135 /*
2136  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2137  */
2138 static void
2139 bridge_qflush(struct ifnet *ifp __unused)
2140 {
2141 }
2142 
2143 /*
2144  * bridge_forward:
2145  *
2146  *	The forwarding function of the bridge.
2147  *
2148  *	NOTE: Releases the lock on return.
2149  */
2150 static void
2151 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2152     struct mbuf *m)
2153 {
2154 	struct bridge_iflist *dbif;
2155 	struct ifnet *src_if, *dst_if, *ifp;
2156 	struct ether_header *eh;
2157 	uint16_t vlan;
2158 	uint8_t *dst;
2159 	int error;
2160 
2161 	src_if = m->m_pkthdr.rcvif;
2162 	ifp = sc->sc_ifp;
2163 
2164 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2165 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2166 	vlan = VLANTAGOF(m);
2167 
2168 	if ((sbif->bif_flags & IFBIF_STP) &&
2169 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2170 		goto drop;
2171 
2172 	eh = mtod(m, struct ether_header *);
2173 	dst = eh->ether_dhost;
2174 
2175 	/* If the interface is learning, record the address. */
2176 	if (sbif->bif_flags & IFBIF_LEARNING) {
2177 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2178 		    sbif, 0, IFBAF_DYNAMIC);
2179 		/*
2180 		 * If the interface has addresses limits then deny any source
2181 		 * that is not in the cache.
2182 		 */
2183 		if (error && sbif->bif_addrmax)
2184 			goto drop;
2185 	}
2186 
2187 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2188 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2189 		goto drop;
2190 
2191 	/*
2192 	 * At this point, the port either doesn't participate
2193 	 * in spanning tree or it is in the forwarding state.
2194 	 */
2195 
2196 	/*
2197 	 * If the packet is unicast, destined for someone on
2198 	 * "this" side of the bridge, drop it.
2199 	 */
2200 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2201 		dst_if = bridge_rtlookup(sc, dst, vlan);
2202 		if (src_if == dst_if)
2203 			goto drop;
2204 	} else {
2205 		/*
2206 		 * Check if its a reserved multicast address, any address
2207 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2208 		 * bridge.
2209 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2210 		 */
2211 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2212 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2213 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2214 			goto drop;
2215 
2216 		/* ...forward it to all interfaces. */
2217 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2218 		dst_if = NULL;
2219 	}
2220 
2221 	/*
2222 	 * If we have a destination interface which is a member of our bridge,
2223 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2224 	 * For broadcast or multicast packets, don't bother because it will
2225 	 * be reinjected into ether_input. We do this before we pass the packets
2226 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2227 	 * drop the packet, or possibly modify it, making it difficult to debug
2228 	 * firewall issues on the bridge.
2229 	 */
2230 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2231 		ETHER_BPF_MTAP(ifp, m);
2232 
2233 	/* run the packet filter */
2234 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2235 #ifdef INET6
2236 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2237 #endif
2238 	    ) {
2239 		BRIDGE_UNLOCK(sc);
2240 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2241 			return;
2242 		if (m == NULL)
2243 			return;
2244 		BRIDGE_LOCK(sc);
2245 	}
2246 
2247 	if (dst_if == NULL) {
2248 		bridge_broadcast(sc, src_if, m, 1);
2249 		return;
2250 	}
2251 
2252 	/*
2253 	 * At this point, we're dealing with a unicast frame
2254 	 * going to a different interface.
2255 	 */
2256 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2257 		goto drop;
2258 
2259 	dbif = bridge_lookup_member_if(sc, dst_if);
2260 	if (dbif == NULL)
2261 		/* Not a member of the bridge (anymore?) */
2262 		goto drop;
2263 
2264 	/* Private segments can not talk to each other */
2265 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2266 		goto drop;
2267 
2268 	if ((dbif->bif_flags & IFBIF_STP) &&
2269 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2270 		goto drop;
2271 
2272 	BRIDGE_UNLOCK(sc);
2273 
2274 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2275 #ifdef INET6
2276 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2277 #endif
2278 	    ) {
2279 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2280 			return;
2281 		if (m == NULL)
2282 			return;
2283 	}
2284 
2285 	bridge_enqueue(sc, dst_if, m);
2286 	return;
2287 
2288 drop:
2289 	BRIDGE_UNLOCK(sc);
2290 	m_freem(m);
2291 }
2292 
2293 /*
2294  * bridge_input:
2295  *
2296  *	Receive input from a member interface.  Queue the packet for
2297  *	bridging if it is not for us.
2298  */
2299 static struct mbuf *
2300 bridge_input(struct ifnet *ifp, struct mbuf *m)
2301 {
2302 	struct bridge_softc *sc = ifp->if_bridge;
2303 	struct bridge_iflist *bif, *bif2;
2304 	struct ifnet *bifp;
2305 	struct ether_header *eh;
2306 	struct mbuf *mc, *mc2;
2307 	uint16_t vlan;
2308 	int error;
2309 
2310 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2311 		return (m);
2312 
2313 	bifp = sc->sc_ifp;
2314 	vlan = VLANTAGOF(m);
2315 
2316 	/*
2317 	 * Implement support for bridge monitoring. If this flag has been
2318 	 * set on this interface, discard the packet once we push it through
2319 	 * the bpf(4) machinery, but before we do, increment the byte and
2320 	 * packet counters associated with this interface.
2321 	 */
2322 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2323 		m->m_pkthdr.rcvif  = bifp;
2324 		ETHER_BPF_MTAP(bifp, m);
2325 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2326 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2327 		m_freem(m);
2328 		return (NULL);
2329 	}
2330 	BRIDGE_LOCK(sc);
2331 	bif = bridge_lookup_member_if(sc, ifp);
2332 	if (bif == NULL) {
2333 		BRIDGE_UNLOCK(sc);
2334 		return (m);
2335 	}
2336 
2337 	eh = mtod(m, struct ether_header *);
2338 
2339 	bridge_span(sc, m);
2340 
2341 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2342 		/* Tap off 802.1D packets; they do not get forwarded. */
2343 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2344 		    ETHER_ADDR_LEN) == 0) {
2345 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2346 			BRIDGE_UNLOCK(sc);
2347 			return (NULL);
2348 		}
2349 
2350 		if ((bif->bif_flags & IFBIF_STP) &&
2351 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2352 			BRIDGE_UNLOCK(sc);
2353 			return (m);
2354 		}
2355 
2356 		/*
2357 		 * Make a deep copy of the packet and enqueue the copy
2358 		 * for bridge processing; return the original packet for
2359 		 * local processing.
2360 		 */
2361 		mc = m_dup(m, M_NOWAIT);
2362 		if (mc == NULL) {
2363 			BRIDGE_UNLOCK(sc);
2364 			return (m);
2365 		}
2366 
2367 		/* Perform the bridge forwarding function with the copy. */
2368 		bridge_forward(sc, bif, mc);
2369 
2370 		/*
2371 		 * Reinject the mbuf as arriving on the bridge so we have a
2372 		 * chance at claiming multicast packets. We can not loop back
2373 		 * here from ether_input as a bridge is never a member of a
2374 		 * bridge.
2375 		 */
2376 		KASSERT(bifp->if_bridge == NULL,
2377 		    ("loop created in bridge_input"));
2378 		mc2 = m_dup(m, M_NOWAIT);
2379 		if (mc2 != NULL) {
2380 			/* Keep the layer3 header aligned */
2381 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2382 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2383 		}
2384 		if (mc2 != NULL) {
2385 			mc2->m_pkthdr.rcvif = bifp;
2386 			(*bifp->if_input)(bifp, mc2);
2387 		}
2388 
2389 		/* Return the original packet for local processing. */
2390 		return (m);
2391 	}
2392 
2393 	if ((bif->bif_flags & IFBIF_STP) &&
2394 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2395 		BRIDGE_UNLOCK(sc);
2396 		return (m);
2397 	}
2398 
2399 #if (defined(INET) || defined(INET6))
2400 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2401 	|| ((iface)->if_carp \
2402 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2403 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2404 	|| ((iface)->if_carp \
2405 	    && (*carp_forus_p)((iface), eh->ether_shost))
2406 #else
2407 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2408 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2409 #endif
2410 
2411 #ifdef INET6
2412 #   define OR_PFIL_HOOKED_INET6 \
2413 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2414 #else
2415 #   define OR_PFIL_HOOKED_INET6
2416 #endif
2417 
2418 #define GRAB_OUR_PACKETS(iface) \
2419 	if ((iface)->if_type == IFT_GIF) \
2420 		continue; \
2421 	/* It is destined for us. */ \
2422 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2423 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2424 	    ) {								\
2425 		if ((iface)->if_type == IFT_BRIDGE) {			\
2426 			ETHER_BPF_MTAP(iface, m);			\
2427 			if_inc_counter(iface, IFCOUNTER_IPACKETS, 1);				\
2428 			if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len);		\
2429 			/* Filter on the physical interface. */		\
2430 			if (V_pfil_local_phys &&			\
2431 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2432 			     OR_PFIL_HOOKED_INET6)) {			\
2433 				if (bridge_pfil(&m, NULL, ifp,		\
2434 				    PFIL_IN) != 0 || m == NULL) {	\
2435 					BRIDGE_UNLOCK(sc);		\
2436 					return (NULL);			\
2437 				}					\
2438 				eh = mtod(m, struct ether_header *);	\
2439 			}						\
2440 		}							\
2441 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2442 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2443 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2444 			if (error && bif->bif_addrmax) {		\
2445 				BRIDGE_UNLOCK(sc);			\
2446 				m_freem(m);				\
2447 				return (NULL);				\
2448 			}						\
2449 		}							\
2450 		m->m_pkthdr.rcvif = iface;				\
2451 		BRIDGE_UNLOCK(sc);					\
2452 		return (m);						\
2453 	}								\
2454 									\
2455 	/* We just received a packet that we sent out. */		\
2456 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2457 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2458 	    ) {								\
2459 		BRIDGE_UNLOCK(sc);					\
2460 		m_freem(m);						\
2461 		return (NULL);						\
2462 	}
2463 
2464 	/*
2465 	 * Unicast.  Make sure it's not for the bridge.
2466 	 */
2467 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2468 
2469 	/*
2470 	 * Give a chance for ifp at first priority. This will help when	the
2471 	 * packet comes through the interface like VLAN's with the same MACs
2472 	 * on several interfaces from the same bridge. This also will save
2473 	 * some CPU cycles in case the destination interface and the input
2474 	 * interface (eq ifp) are the same.
2475 	 */
2476 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2477 
2478 	/* Now check the all bridge members. */
2479 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2480 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2481 	}
2482 
2483 #undef OR_CARP_CHECK_WE_ARE_DST
2484 #undef OR_CARP_CHECK_WE_ARE_SRC
2485 #undef OR_PFIL_HOOKED_INET6
2486 #undef GRAB_OUR_PACKETS
2487 
2488 	/* Perform the bridge forwarding function. */
2489 	bridge_forward(sc, bif, m);
2490 
2491 	return (NULL);
2492 }
2493 
2494 /*
2495  * bridge_broadcast:
2496  *
2497  *	Send a frame to all interfaces that are members of
2498  *	the bridge, except for the one on which the packet
2499  *	arrived.
2500  *
2501  *	NOTE: Releases the lock on return.
2502  */
2503 static void
2504 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2505     struct mbuf *m, int runfilt)
2506 {
2507 	struct bridge_iflist *dbif, *sbif;
2508 	struct mbuf *mc;
2509 	struct ifnet *dst_if;
2510 	int error = 0, used = 0, i;
2511 
2512 	sbif = bridge_lookup_member_if(sc, src_if);
2513 
2514 	BRIDGE_LOCK2REF(sc, error);
2515 	if (error) {
2516 		m_freem(m);
2517 		return;
2518 	}
2519 
2520 	/* Filter on the bridge interface before broadcasting */
2521 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2522 #ifdef INET6
2523 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2524 #endif
2525 	    )) {
2526 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2527 			goto out;
2528 		if (m == NULL)
2529 			goto out;
2530 	}
2531 
2532 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2533 		dst_if = dbif->bif_ifp;
2534 		if (dst_if == src_if)
2535 			continue;
2536 
2537 		/* Private segments can not talk to each other */
2538 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2539 			continue;
2540 
2541 		if ((dbif->bif_flags & IFBIF_STP) &&
2542 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2543 			continue;
2544 
2545 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2546 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2547 			continue;
2548 
2549 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2550 			continue;
2551 
2552 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2553 			mc = m;
2554 			used = 1;
2555 		} else {
2556 			mc = m_dup(m, M_NOWAIT);
2557 			if (mc == NULL) {
2558 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2559 				continue;
2560 			}
2561 		}
2562 
2563 		/*
2564 		 * Filter on the output interface. Pass a NULL bridge interface
2565 		 * pointer so we do not redundantly filter on the bridge for
2566 		 * each interface we broadcast on.
2567 		 */
2568 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2569 #ifdef INET6
2570 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2571 #endif
2572 		    )) {
2573 			if (used == 0) {
2574 				/* Keep the layer3 header aligned */
2575 				i = min(mc->m_pkthdr.len, max_protohdr);
2576 				mc = m_copyup(mc, i, ETHER_ALIGN);
2577 				if (mc == NULL) {
2578 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2579 					continue;
2580 				}
2581 			}
2582 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2583 				continue;
2584 			if (mc == NULL)
2585 				continue;
2586 		}
2587 
2588 		bridge_enqueue(sc, dst_if, mc);
2589 	}
2590 	if (used == 0)
2591 		m_freem(m);
2592 
2593 out:
2594 	BRIDGE_UNREF(sc);
2595 }
2596 
2597 /*
2598  * bridge_span:
2599  *
2600  *	Duplicate a packet out one or more interfaces that are in span mode,
2601  *	the original mbuf is unmodified.
2602  */
2603 static void
2604 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2605 {
2606 	struct bridge_iflist *bif;
2607 	struct ifnet *dst_if;
2608 	struct mbuf *mc;
2609 
2610 	if (LIST_EMPTY(&sc->sc_spanlist))
2611 		return;
2612 
2613 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2614 		dst_if = bif->bif_ifp;
2615 
2616 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2617 			continue;
2618 
2619 		mc = m_copypacket(m, M_NOWAIT);
2620 		if (mc == NULL) {
2621 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2622 			continue;
2623 		}
2624 
2625 		bridge_enqueue(sc, dst_if, mc);
2626 	}
2627 }
2628 
2629 /*
2630  * bridge_rtupdate:
2631  *
2632  *	Add a bridge routing entry.
2633  */
2634 static int
2635 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2636     struct bridge_iflist *bif, int setflags, uint8_t flags)
2637 {
2638 	struct bridge_rtnode *brt;
2639 	int error;
2640 
2641 	BRIDGE_LOCK_ASSERT(sc);
2642 
2643 	/* Check the source address is valid and not multicast. */
2644 	if (ETHER_IS_MULTICAST(dst) ||
2645 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2646 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2647 		return (EINVAL);
2648 
2649 	/* 802.1p frames map to vlan 1 */
2650 	if (vlan == 0)
2651 		vlan = 1;
2652 
2653 	/*
2654 	 * A route for this destination might already exist.  If so,
2655 	 * update it, otherwise create a new one.
2656 	 */
2657 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2658 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2659 			sc->sc_brtexceeded++;
2660 			return (ENOSPC);
2661 		}
2662 		/* Check per interface address limits (if enabled) */
2663 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2664 			bif->bif_addrexceeded++;
2665 			return (ENOSPC);
2666 		}
2667 
2668 		/*
2669 		 * Allocate a new bridge forwarding node, and
2670 		 * initialize the expiration time and Ethernet
2671 		 * address.
2672 		 */
2673 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2674 		if (brt == NULL)
2675 			return (ENOMEM);
2676 
2677 		if (bif->bif_flags & IFBIF_STICKY)
2678 			brt->brt_flags = IFBAF_STICKY;
2679 		else
2680 			brt->brt_flags = IFBAF_DYNAMIC;
2681 
2682 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2683 		brt->brt_vlan = vlan;
2684 
2685 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2686 			uma_zfree(bridge_rtnode_zone, brt);
2687 			return (error);
2688 		}
2689 		brt->brt_dst = bif;
2690 		bif->bif_addrcnt++;
2691 	}
2692 
2693 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2694 	    brt->brt_dst != bif) {
2695 		brt->brt_dst->bif_addrcnt--;
2696 		brt->brt_dst = bif;
2697 		brt->brt_dst->bif_addrcnt++;
2698 	}
2699 
2700 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2701 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2702 	if (setflags)
2703 		brt->brt_flags = flags;
2704 
2705 	return (0);
2706 }
2707 
2708 /*
2709  * bridge_rtlookup:
2710  *
2711  *	Lookup the destination interface for an address.
2712  */
2713 static struct ifnet *
2714 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2715 {
2716 	struct bridge_rtnode *brt;
2717 
2718 	BRIDGE_LOCK_ASSERT(sc);
2719 
2720 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2721 		return (NULL);
2722 
2723 	return (brt->brt_ifp);
2724 }
2725 
2726 /*
2727  * bridge_rttrim:
2728  *
2729  *	Trim the routine table so that we have a number
2730  *	of routing entries less than or equal to the
2731  *	maximum number.
2732  */
2733 static void
2734 bridge_rttrim(struct bridge_softc *sc)
2735 {
2736 	struct bridge_rtnode *brt, *nbrt;
2737 
2738 	BRIDGE_LOCK_ASSERT(sc);
2739 
2740 	/* Make sure we actually need to do this. */
2741 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2742 		return;
2743 
2744 	/* Force an aging cycle; this might trim enough addresses. */
2745 	bridge_rtage(sc);
2746 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2747 		return;
2748 
2749 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2750 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2751 			bridge_rtnode_destroy(sc, brt);
2752 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2753 				return;
2754 		}
2755 	}
2756 }
2757 
2758 /*
2759  * bridge_timer:
2760  *
2761  *	Aging timer for the bridge.
2762  */
2763 static void
2764 bridge_timer(void *arg)
2765 {
2766 	struct bridge_softc *sc = arg;
2767 
2768 	BRIDGE_LOCK_ASSERT(sc);
2769 
2770 	bridge_rtage(sc);
2771 
2772 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2773 		callout_reset(&sc->sc_brcallout,
2774 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2775 }
2776 
2777 /*
2778  * bridge_rtage:
2779  *
2780  *	Perform an aging cycle.
2781  */
2782 static void
2783 bridge_rtage(struct bridge_softc *sc)
2784 {
2785 	struct bridge_rtnode *brt, *nbrt;
2786 
2787 	BRIDGE_LOCK_ASSERT(sc);
2788 
2789 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2790 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2791 			if (time_uptime >= brt->brt_expire)
2792 				bridge_rtnode_destroy(sc, brt);
2793 		}
2794 	}
2795 }
2796 
2797 /*
2798  * bridge_rtflush:
2799  *
2800  *	Remove all dynamic addresses from the bridge.
2801  */
2802 static void
2803 bridge_rtflush(struct bridge_softc *sc, int full)
2804 {
2805 	struct bridge_rtnode *brt, *nbrt;
2806 
2807 	BRIDGE_LOCK_ASSERT(sc);
2808 
2809 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2810 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2811 			bridge_rtnode_destroy(sc, brt);
2812 	}
2813 }
2814 
2815 /*
2816  * bridge_rtdaddr:
2817  *
2818  *	Remove an address from the table.
2819  */
2820 static int
2821 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2822 {
2823 	struct bridge_rtnode *brt;
2824 	int found = 0;
2825 
2826 	BRIDGE_LOCK_ASSERT(sc);
2827 
2828 	/*
2829 	 * If vlan is zero then we want to delete for all vlans so the lookup
2830 	 * may return more than one.
2831 	 */
2832 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2833 		bridge_rtnode_destroy(sc, brt);
2834 		found = 1;
2835 	}
2836 
2837 	return (found ? 0 : ENOENT);
2838 }
2839 
2840 /*
2841  * bridge_rtdelete:
2842  *
2843  *	Delete routes to a speicifc member interface.
2844  */
2845 static void
2846 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2847 {
2848 	struct bridge_rtnode *brt, *nbrt;
2849 
2850 	BRIDGE_LOCK_ASSERT(sc);
2851 
2852 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2853 		if (brt->brt_ifp == ifp && (full ||
2854 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2855 			bridge_rtnode_destroy(sc, brt);
2856 	}
2857 }
2858 
2859 /*
2860  * bridge_rtable_init:
2861  *
2862  *	Initialize the route table for this bridge.
2863  */
2864 static void
2865 bridge_rtable_init(struct bridge_softc *sc)
2866 {
2867 	int i;
2868 
2869 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2870 	    M_DEVBUF, M_WAITOK);
2871 
2872 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2873 		LIST_INIT(&sc->sc_rthash[i]);
2874 
2875 	sc->sc_rthash_key = arc4random();
2876 	LIST_INIT(&sc->sc_rtlist);
2877 }
2878 
2879 /*
2880  * bridge_rtable_fini:
2881  *
2882  *	Deconstruct the route table for this bridge.
2883  */
2884 static void
2885 bridge_rtable_fini(struct bridge_softc *sc)
2886 {
2887 
2888 	KASSERT(sc->sc_brtcnt == 0,
2889 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2890 	free(sc->sc_rthash, M_DEVBUF);
2891 }
2892 
2893 /*
2894  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2895  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2896  */
2897 #define	mix(a, b, c)							\
2898 do {									\
2899 	a -= b; a -= c; a ^= (c >> 13);					\
2900 	b -= c; b -= a; b ^= (a << 8);					\
2901 	c -= a; c -= b; c ^= (b >> 13);					\
2902 	a -= b; a -= c; a ^= (c >> 12);					\
2903 	b -= c; b -= a; b ^= (a << 16);					\
2904 	c -= a; c -= b; c ^= (b >> 5);					\
2905 	a -= b; a -= c; a ^= (c >> 3);					\
2906 	b -= c; b -= a; b ^= (a << 10);					\
2907 	c -= a; c -= b; c ^= (b >> 15);					\
2908 } while (/*CONSTCOND*/0)
2909 
2910 static __inline uint32_t
2911 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2912 {
2913 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2914 
2915 	b += addr[5] << 8;
2916 	b += addr[4];
2917 	a += addr[3] << 24;
2918 	a += addr[2] << 16;
2919 	a += addr[1] << 8;
2920 	a += addr[0];
2921 
2922 	mix(a, b, c);
2923 
2924 	return (c & BRIDGE_RTHASH_MASK);
2925 }
2926 
2927 #undef mix
2928 
2929 static int
2930 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2931 {
2932 	int i, d;
2933 
2934 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2935 		d = ((int)a[i]) - ((int)b[i]);
2936 	}
2937 
2938 	return (d);
2939 }
2940 
2941 /*
2942  * bridge_rtnode_lookup:
2943  *
2944  *	Look up a bridge route node for the specified destination. Compare the
2945  *	vlan id or if zero then just return the first match.
2946  */
2947 static struct bridge_rtnode *
2948 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2949 {
2950 	struct bridge_rtnode *brt;
2951 	uint32_t hash;
2952 	int dir;
2953 
2954 	BRIDGE_LOCK_ASSERT(sc);
2955 
2956 	hash = bridge_rthash(sc, addr);
2957 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2958 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2959 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2960 			return (brt);
2961 		if (dir > 0)
2962 			return (NULL);
2963 	}
2964 
2965 	return (NULL);
2966 }
2967 
2968 /*
2969  * bridge_rtnode_insert:
2970  *
2971  *	Insert the specified bridge node into the route table.  We
2972  *	assume the entry is not already in the table.
2973  */
2974 static int
2975 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2976 {
2977 	struct bridge_rtnode *lbrt;
2978 	uint32_t hash;
2979 	int dir;
2980 
2981 	BRIDGE_LOCK_ASSERT(sc);
2982 
2983 	hash = bridge_rthash(sc, brt->brt_addr);
2984 
2985 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2986 	if (lbrt == NULL) {
2987 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2988 		goto out;
2989 	}
2990 
2991 	do {
2992 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2993 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2994 			return (EEXIST);
2995 		if (dir > 0) {
2996 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2997 			goto out;
2998 		}
2999 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
3000 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3001 			goto out;
3002 		}
3003 		lbrt = LIST_NEXT(lbrt, brt_hash);
3004 	} while (lbrt != NULL);
3005 
3006 #ifdef DIAGNOSTIC
3007 	panic("bridge_rtnode_insert: impossible");
3008 #endif
3009 
3010 out:
3011 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3012 	sc->sc_brtcnt++;
3013 
3014 	return (0);
3015 }
3016 
3017 /*
3018  * bridge_rtnode_destroy:
3019  *
3020  *	Destroy a bridge rtnode.
3021  */
3022 static void
3023 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3024 {
3025 	BRIDGE_LOCK_ASSERT(sc);
3026 
3027 	LIST_REMOVE(brt, brt_hash);
3028 
3029 	LIST_REMOVE(brt, brt_list);
3030 	sc->sc_brtcnt--;
3031 	brt->brt_dst->bif_addrcnt--;
3032 	uma_zfree(bridge_rtnode_zone, brt);
3033 }
3034 
3035 /*
3036  * bridge_rtable_expire:
3037  *
3038  *	Set the expiry time for all routes on an interface.
3039  */
3040 static void
3041 bridge_rtable_expire(struct ifnet *ifp, int age)
3042 {
3043 	struct bridge_softc *sc = ifp->if_bridge;
3044 	struct bridge_rtnode *brt;
3045 
3046 	BRIDGE_LOCK(sc);
3047 
3048 	/*
3049 	 * If the age is zero then flush, otherwise set all the expiry times to
3050 	 * age for the interface
3051 	 */
3052 	if (age == 0)
3053 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3054 	else {
3055 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3056 			/* Cap the expiry time to 'age' */
3057 			if (brt->brt_ifp == ifp &&
3058 			    brt->brt_expire > time_uptime + age &&
3059 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3060 				brt->brt_expire = time_uptime + age;
3061 		}
3062 	}
3063 	BRIDGE_UNLOCK(sc);
3064 }
3065 
3066 /*
3067  * bridge_state_change:
3068  *
3069  *	Callback from the bridgestp code when a port changes states.
3070  */
3071 static void
3072 bridge_state_change(struct ifnet *ifp, int state)
3073 {
3074 	struct bridge_softc *sc = ifp->if_bridge;
3075 	static const char *stpstates[] = {
3076 		"disabled",
3077 		"listening",
3078 		"learning",
3079 		"forwarding",
3080 		"blocking",
3081 		"discarding"
3082 	};
3083 
3084 	CURVNET_SET(ifp->if_vnet);
3085 	if (V_log_stp)
3086 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3087 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3088 	CURVNET_RESTORE();
3089 }
3090 
3091 /*
3092  * Send bridge packets through pfil if they are one of the types pfil can deal
3093  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3094  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3095  * that interface.
3096  */
3097 static int
3098 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3099 {
3100 	int snap, error, i, hlen;
3101 	struct ether_header *eh1, eh2;
3102 	struct ip *ip;
3103 	struct llc llc1;
3104 	u_int16_t ether_type;
3105 
3106 	snap = 0;
3107 	error = -1;	/* Default error if not error == 0 */
3108 
3109 #if 0
3110 	/* we may return with the IP fields swapped, ensure its not shared */
3111 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3112 #endif
3113 
3114 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3115 		return (0); /* filtering is disabled */
3116 
3117 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3118 	if ((*mp)->m_len < i) {
3119 	    *mp = m_pullup(*mp, i);
3120 	    if (*mp == NULL) {
3121 		printf("%s: m_pullup failed\n", __func__);
3122 		return (-1);
3123 	    }
3124 	}
3125 
3126 	eh1 = mtod(*mp, struct ether_header *);
3127 	ether_type = ntohs(eh1->ether_type);
3128 
3129 	/*
3130 	 * Check for SNAP/LLC.
3131 	 */
3132 	if (ether_type < ETHERMTU) {
3133 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3134 
3135 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3136 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3137 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3138 		    llc2->llc_control == LLC_UI) {
3139 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3140 			snap = 1;
3141 		}
3142 	}
3143 
3144 	/*
3145 	 * If we're trying to filter bridge traffic, don't look at anything
3146 	 * other than IP and ARP traffic.  If the filter doesn't understand
3147 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3148 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3149 	 * but of course we don't have an AppleTalk filter to begin with.
3150 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3151 	 * ARP traffic.)
3152 	 */
3153 	switch (ether_type) {
3154 		case ETHERTYPE_ARP:
3155 		case ETHERTYPE_REVARP:
3156 			if (V_pfil_ipfw_arp == 0)
3157 				return (0); /* Automatically pass */
3158 			break;
3159 
3160 		case ETHERTYPE_IP:
3161 #ifdef INET6
3162 		case ETHERTYPE_IPV6:
3163 #endif /* INET6 */
3164 			break;
3165 		default:
3166 			/*
3167 			 * Check to see if the user wants to pass non-ip
3168 			 * packets, these will not be checked by pfil(9) and
3169 			 * passed unconditionally so the default is to drop.
3170 			 */
3171 			if (V_pfil_onlyip)
3172 				goto bad;
3173 	}
3174 
3175 	/* Run the packet through pfil before stripping link headers */
3176 	if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3177 			dir == PFIL_OUT && ifp != NULL) {
3178 
3179 		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3180 
3181 		if (*mp == NULL || error != 0) /* packet consumed by filter */
3182 			return (error);
3183 	}
3184 
3185 	/* Strip off the Ethernet header and keep a copy. */
3186 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3187 	m_adj(*mp, ETHER_HDR_LEN);
3188 
3189 	/* Strip off snap header, if present */
3190 	if (snap) {
3191 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3192 		m_adj(*mp, sizeof(struct llc));
3193 	}
3194 
3195 	/*
3196 	 * Check the IP header for alignment and errors
3197 	 */
3198 	if (dir == PFIL_IN) {
3199 		switch (ether_type) {
3200 			case ETHERTYPE_IP:
3201 				error = bridge_ip_checkbasic(mp);
3202 				break;
3203 #ifdef INET6
3204 			case ETHERTYPE_IPV6:
3205 				error = bridge_ip6_checkbasic(mp);
3206 				break;
3207 #endif /* INET6 */
3208 			default:
3209 				error = 0;
3210 		}
3211 		if (error)
3212 			goto bad;
3213 	}
3214 
3215 	error = 0;
3216 
3217 	/*
3218 	 * Run the packet through pfil
3219 	 */
3220 	switch (ether_type) {
3221 	case ETHERTYPE_IP:
3222 		/*
3223 		 * Run pfil on the member interface and the bridge, both can
3224 		 * be skipped by clearing pfil_member or pfil_bridge.
3225 		 *
3226 		 * Keep the order:
3227 		 *   in_if -> bridge_if -> out_if
3228 		 */
3229 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3230 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3231 					dir, NULL);
3232 
3233 		if (*mp == NULL || error != 0) /* filter may consume */
3234 			break;
3235 
3236 		if (V_pfil_member && ifp != NULL)
3237 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3238 					dir, NULL);
3239 
3240 		if (*mp == NULL || error != 0) /* filter may consume */
3241 			break;
3242 
3243 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3244 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3245 					dir, NULL);
3246 
3247 		if (*mp == NULL || error != 0) /* filter may consume */
3248 			break;
3249 
3250 		/* check if we need to fragment the packet */
3251 		/* bridge_fragment generates a mbuf chain of packets */
3252 		/* that already include eth headers */
3253 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3254 			i = (*mp)->m_pkthdr.len;
3255 			if (i > ifp->if_mtu) {
3256 				error = bridge_fragment(ifp, mp, &eh2, snap,
3257 					    &llc1);
3258 				return (error);
3259 			}
3260 		}
3261 
3262 		/* Recalculate the ip checksum. */
3263 		ip = mtod(*mp, struct ip *);
3264 		hlen = ip->ip_hl << 2;
3265 		if (hlen < sizeof(struct ip))
3266 			goto bad;
3267 		if (hlen > (*mp)->m_len) {
3268 			if ((*mp = m_pullup(*mp, hlen)) == NULL)
3269 				goto bad;
3270 			ip = mtod(*mp, struct ip *);
3271 			if (ip == NULL)
3272 				goto bad;
3273 		}
3274 		ip->ip_sum = 0;
3275 		if (hlen == sizeof(struct ip))
3276 			ip->ip_sum = in_cksum_hdr(ip);
3277 		else
3278 			ip->ip_sum = in_cksum(*mp, hlen);
3279 
3280 		break;
3281 #ifdef INET6
3282 	case ETHERTYPE_IPV6:
3283 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3284 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3285 					dir, NULL);
3286 
3287 		if (*mp == NULL || error != 0) /* filter may consume */
3288 			break;
3289 
3290 		if (V_pfil_member && ifp != NULL)
3291 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3292 					dir, NULL);
3293 
3294 		if (*mp == NULL || error != 0) /* filter may consume */
3295 			break;
3296 
3297 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3298 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3299 					dir, NULL);
3300 		break;
3301 #endif
3302 	default:
3303 		error = 0;
3304 		break;
3305 	}
3306 
3307 	if (*mp == NULL)
3308 		return (error);
3309 	if (error != 0)
3310 		goto bad;
3311 
3312 	error = -1;
3313 
3314 	/*
3315 	 * Finally, put everything back the way it was and return
3316 	 */
3317 	if (snap) {
3318 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3319 		if (*mp == NULL)
3320 			return (error);
3321 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3322 	}
3323 
3324 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3325 	if (*mp == NULL)
3326 		return (error);
3327 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3328 
3329 	return (0);
3330 
3331 bad:
3332 	m_freem(*mp);
3333 	*mp = NULL;
3334 	return (error);
3335 }
3336 
3337 /*
3338  * Perform basic checks on header size since
3339  * pfil assumes ip_input has already processed
3340  * it for it.  Cut-and-pasted from ip_input.c.
3341  * Given how simple the IPv6 version is,
3342  * does the IPv4 version really need to be
3343  * this complicated?
3344  *
3345  * XXX Should we update ipstat here, or not?
3346  * XXX Right now we update ipstat but not
3347  * XXX csum_counter.
3348  */
3349 static int
3350 bridge_ip_checkbasic(struct mbuf **mp)
3351 {
3352 	struct mbuf *m = *mp;
3353 	struct ip *ip;
3354 	int len, hlen;
3355 	u_short sum;
3356 
3357 	if (*mp == NULL)
3358 		return (-1);
3359 
3360 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3361 		if ((m = m_copyup(m, sizeof(struct ip),
3362 			(max_linkhdr + 3) & ~3)) == NULL) {
3363 			/* XXXJRT new stat, please */
3364 			KMOD_IPSTAT_INC(ips_toosmall);
3365 			goto bad;
3366 		}
3367 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3368 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3369 			KMOD_IPSTAT_INC(ips_toosmall);
3370 			goto bad;
3371 		}
3372 	}
3373 	ip = mtod(m, struct ip *);
3374 	if (ip == NULL) goto bad;
3375 
3376 	if (ip->ip_v != IPVERSION) {
3377 		KMOD_IPSTAT_INC(ips_badvers);
3378 		goto bad;
3379 	}
3380 	hlen = ip->ip_hl << 2;
3381 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3382 		KMOD_IPSTAT_INC(ips_badhlen);
3383 		goto bad;
3384 	}
3385 	if (hlen > m->m_len) {
3386 		if ((m = m_pullup(m, hlen)) == NULL) {
3387 			KMOD_IPSTAT_INC(ips_badhlen);
3388 			goto bad;
3389 		}
3390 		ip = mtod(m, struct ip *);
3391 		if (ip == NULL) goto bad;
3392 	}
3393 
3394 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3395 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3396 	} else {
3397 		if (hlen == sizeof(struct ip)) {
3398 			sum = in_cksum_hdr(ip);
3399 		} else {
3400 			sum = in_cksum(m, hlen);
3401 		}
3402 	}
3403 	if (sum) {
3404 		KMOD_IPSTAT_INC(ips_badsum);
3405 		goto bad;
3406 	}
3407 
3408 	/* Retrieve the packet length. */
3409 	len = ntohs(ip->ip_len);
3410 
3411 	/*
3412 	 * Check for additional length bogosity
3413 	 */
3414 	if (len < hlen) {
3415 		KMOD_IPSTAT_INC(ips_badlen);
3416 		goto bad;
3417 	}
3418 
3419 	/*
3420 	 * Check that the amount of data in the buffers
3421 	 * is as at least much as the IP header would have us expect.
3422 	 * Drop packet if shorter than we expect.
3423 	 */
3424 	if (m->m_pkthdr.len < len) {
3425 		KMOD_IPSTAT_INC(ips_tooshort);
3426 		goto bad;
3427 	}
3428 
3429 	/* Checks out, proceed */
3430 	*mp = m;
3431 	return (0);
3432 
3433 bad:
3434 	*mp = m;
3435 	return (-1);
3436 }
3437 
3438 #ifdef INET6
3439 /*
3440  * Same as above, but for IPv6.
3441  * Cut-and-pasted from ip6_input.c.
3442  * XXX Should we update ip6stat, or not?
3443  */
3444 static int
3445 bridge_ip6_checkbasic(struct mbuf **mp)
3446 {
3447 	struct mbuf *m = *mp;
3448 	struct ip6_hdr *ip6;
3449 
3450 	/*
3451 	 * If the IPv6 header is not aligned, slurp it up into a new
3452 	 * mbuf with space for link headers, in the event we forward
3453 	 * it.  Otherwise, if it is aligned, make sure the entire base
3454 	 * IPv6 header is in the first mbuf of the chain.
3455 	 */
3456 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3457 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3458 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3459 			    (max_linkhdr + 3) & ~3)) == NULL) {
3460 			/* XXXJRT new stat, please */
3461 			IP6STAT_INC(ip6s_toosmall);
3462 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3463 			goto bad;
3464 		}
3465 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3466 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3467 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3468 			IP6STAT_INC(ip6s_toosmall);
3469 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3470 			goto bad;
3471 		}
3472 	}
3473 
3474 	ip6 = mtod(m, struct ip6_hdr *);
3475 
3476 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3477 		IP6STAT_INC(ip6s_badvers);
3478 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3479 		goto bad;
3480 	}
3481 
3482 	/* Checks out, proceed */
3483 	*mp = m;
3484 	return (0);
3485 
3486 bad:
3487 	*mp = m;
3488 	return (-1);
3489 }
3490 #endif /* INET6 */
3491 
3492 /*
3493  * bridge_fragment:
3494  *
3495  *	Fragment mbuf chain in multiple packets and prepend ethernet header.
3496  */
3497 static int
3498 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3499     int snap, struct llc *llc)
3500 {
3501 	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3502 	struct ip *ip;
3503 	int error = -1;
3504 
3505 	if (m->m_len < sizeof(struct ip) &&
3506 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3507 		goto dropit;
3508 	ip = mtod(m, struct ip *);
3509 
3510 	m->m_pkthdr.csum_flags |= CSUM_IP;
3511 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3512 	if (error)
3513 		goto dropit;
3514 
3515 	/*
3516 	 * Walk the chain and re-add the Ethernet header for
3517 	 * each mbuf packet.
3518 	 */
3519 	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3520 		nextpkt = mcur->m_nextpkt;
3521 		mcur->m_nextpkt = NULL;
3522 		if (snap) {
3523 			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3524 			if (mcur == NULL) {
3525 				error = ENOBUFS;
3526 				if (mprev != NULL)
3527 					mprev->m_nextpkt = nextpkt;
3528 				goto dropit;
3529 			}
3530 			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3531 		}
3532 
3533 		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3534 		if (mcur == NULL) {
3535 			error = ENOBUFS;
3536 			if (mprev != NULL)
3537 				mprev->m_nextpkt = nextpkt;
3538 			goto dropit;
3539 		}
3540 		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3541 
3542 		/*
3543 		 * The previous two M_PREPEND could have inserted one or two
3544 		 * mbufs in front so we have to update the previous packet's
3545 		 * m_nextpkt.
3546 		 */
3547 		mcur->m_nextpkt = nextpkt;
3548 		if (mprev != NULL)
3549 			mprev->m_nextpkt = mcur;
3550 		else {
3551 			/* The first mbuf in the original chain needs to be
3552 			 * updated. */
3553 			*mp = mcur;
3554 		}
3555 		mprev = mcur;
3556 	}
3557 
3558 	KMOD_IPSTAT_INC(ips_fragmented);
3559 	return (error);
3560 
3561 dropit:
3562 	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3563 		m = mcur->m_nextpkt;
3564 		m_freem(mcur);
3565 	}
3566 	return (error);
3567 }
3568 
3569 static void
3570 bridge_linkstate(struct ifnet *ifp)
3571 {
3572 	struct bridge_softc *sc = ifp->if_bridge;
3573 	struct bridge_iflist *bif;
3574 
3575 	BRIDGE_LOCK(sc);
3576 	bif = bridge_lookup_member_if(sc, ifp);
3577 	if (bif == NULL) {
3578 		BRIDGE_UNLOCK(sc);
3579 		return;
3580 	}
3581 	bridge_linkcheck(sc);
3582 	BRIDGE_UNLOCK(sc);
3583 
3584 	bstp_linkstate(&bif->bif_stp);
3585 }
3586 
3587 static void
3588 bridge_linkcheck(struct bridge_softc *sc)
3589 {
3590 	struct bridge_iflist *bif;
3591 	int new_link, hasls;
3592 
3593 	BRIDGE_LOCK_ASSERT(sc);
3594 	new_link = LINK_STATE_DOWN;
3595 	hasls = 0;
3596 	/* Our link is considered up if at least one of our ports is active */
3597 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3598 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3599 			hasls++;
3600 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3601 			new_link = LINK_STATE_UP;
3602 			break;
3603 		}
3604 	}
3605 	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3606 		/* If no interfaces support link-state then we default to up */
3607 		new_link = LINK_STATE_UP;
3608 	}
3609 	if_link_state_change(sc->sc_ifp, new_link);
3610 }
3611