xref: /freebsd/sys/net/if_bridge.c (revision e9dcd83155b39327497e7a2577d8990074144ff3)
1 /*	$NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $	*/
2 
3 /*-
4  * SPDX-License-Identifier: BSD-4-Clause
5  *
6  * Copyright 2001 Wasabi Systems, Inc.
7  * All rights reserved.
8  *
9  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  * 3. All advertising materials mentioning features or use of this software
20  *    must display the following acknowledgement:
21  *	This product includes software developed for the NetBSD Project by
22  *	Wasabi Systems, Inc.
23  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
24  *    or promote products derived from this software without specific prior
25  *    written permission.
26  *
27  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
28  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
31  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37  * POSSIBILITY OF SUCH DAMAGE.
38  */
39 
40 /*
41  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
42  * All rights reserved.
43  *
44  * Redistribution and use in source and binary forms, with or without
45  * modification, are permitted provided that the following conditions
46  * are met:
47  * 1. Redistributions of source code must retain the above copyright
48  *    notice, this list of conditions and the following disclaimer.
49  * 2. Redistributions in binary form must reproduce the above copyright
50  *    notice, this list of conditions and the following disclaimer in the
51  *    documentation and/or other materials provided with the distribution.
52  *
53  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
54  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
55  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
56  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
57  * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
58  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
59  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
61  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
62  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
63  * POSSIBILITY OF SUCH DAMAGE.
64  *
65  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
66  */
67 
68 /*
69  * Network interface bridge support.
70  *
71  * TODO:
72  *
73  *	- Currently only supports Ethernet-like interfaces (Ethernet,
74  *	  802.11, VLANs on Ethernet, etc.)  Figure out a nice way
75  *	  to bridge other types of interfaces (maybe consider
76  *	  heterogeneous bridges).
77  */
78 
79 #include <sys/cdefs.h>
80 __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 VNET_DEFINE_STATIC(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 
344 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
345 #define	VLANTAGOF(_m)	\
346     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
347 
348 static struct bstp_cb_ops bridge_ops = {
349 	.bcb_state = bridge_state_change,
350 	.bcb_rtage = bridge_rtable_expire
351 };
352 
353 SYSCTL_DECL(_net_link);
354 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
355 
356 /* only pass IP[46] packets when pfil is enabled */
357 VNET_DEFINE_STATIC(int, pfil_onlyip) = 1;
358 #define	V_pfil_onlyip	VNET(pfil_onlyip)
359 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
360     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
361     "Only pass IP packets when pfil is enabled");
362 
363 /* run pfil hooks on the bridge interface */
364 VNET_DEFINE_STATIC(int, pfil_bridge) = 1;
365 #define	V_pfil_bridge	VNET(pfil_bridge)
366 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
367     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
368     "Packet filter on the bridge interface");
369 
370 /* layer2 filter with ipfw */
371 VNET_DEFINE_STATIC(int, pfil_ipfw);
372 #define	V_pfil_ipfw	VNET(pfil_ipfw)
373 
374 /* layer2 ARP filter with ipfw */
375 VNET_DEFINE_STATIC(int, pfil_ipfw_arp);
376 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
377 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
378     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
379     "Filter ARP packets through IPFW layer2");
380 
381 /* run pfil hooks on the member interface */
382 VNET_DEFINE_STATIC(int, pfil_member) = 1;
383 #define	V_pfil_member	VNET(pfil_member)
384 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
385     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
386     "Packet filter on the member interface");
387 
388 /* run pfil hooks on the physical interface for locally destined packets */
389 VNET_DEFINE_STATIC(int, pfil_local_phys);
390 #define	V_pfil_local_phys	VNET(pfil_local_phys)
391 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
392     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
393     "Packet filter on the physical interface for locally destined packets");
394 
395 /* log STP state changes */
396 VNET_DEFINE_STATIC(int, log_stp);
397 #define	V_log_stp	VNET(log_stp)
398 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
399     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
400     "Log STP state changes");
401 
402 /* share MAC with first bridge member */
403 VNET_DEFINE_STATIC(int, bridge_inherit_mac);
404 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
405 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
406     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
407     "Inherit MAC address from the first bridge member");
408 
409 VNET_DEFINE_STATIC(int, allow_llz_overlap) = 0;
410 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
411 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
412     CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
413     "Allow overlap of link-local scope "
414     "zones of a bridge interface and the member interfaces");
415 
416 struct bridge_control {
417 	int	(*bc_func)(struct bridge_softc *, void *);
418 	int	bc_argsize;
419 	int	bc_flags;
420 };
421 
422 #define	BC_F_COPYIN		0x01	/* copy arguments in */
423 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
424 #define	BC_F_SUSER		0x04	/* do super-user check */
425 
426 const struct bridge_control bridge_control_table[] = {
427 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
428 	  BC_F_COPYIN|BC_F_SUSER },
429 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
430 	  BC_F_COPYIN|BC_F_SUSER },
431 
432 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
433 	  BC_F_COPYIN|BC_F_COPYOUT },
434 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
435 	  BC_F_COPYIN|BC_F_SUSER },
436 
437 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
438 	  BC_F_COPYIN|BC_F_SUSER },
439 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
440 	  BC_F_COPYOUT },
441 
442 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
443 	  BC_F_COPYIN|BC_F_COPYOUT },
444 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
445 	  BC_F_COPYIN|BC_F_COPYOUT },
446 
447 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
448 	  BC_F_COPYIN|BC_F_SUSER },
449 
450 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
451 	  BC_F_COPYIN|BC_F_SUSER },
452 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
453 	  BC_F_COPYOUT },
454 
455 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
456 	  BC_F_COPYIN|BC_F_SUSER },
457 
458 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
459 	  BC_F_COPYIN|BC_F_SUSER },
460 
461 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
462 	  BC_F_COPYOUT },
463 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
464 	  BC_F_COPYIN|BC_F_SUSER },
465 
466 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
467 	  BC_F_COPYOUT },
468 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
469 	  BC_F_COPYIN|BC_F_SUSER },
470 
471 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
472 	  BC_F_COPYOUT },
473 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
474 	  BC_F_COPYIN|BC_F_SUSER },
475 
476 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
477 	  BC_F_COPYOUT },
478 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
479 	  BC_F_COPYIN|BC_F_SUSER },
480 
481 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
482 	  BC_F_COPYIN|BC_F_SUSER },
483 
484 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
485 	  BC_F_COPYIN|BC_F_SUSER },
486 
487 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
488 	  BC_F_COPYIN|BC_F_SUSER },
489 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
490 	  BC_F_COPYIN|BC_F_SUSER },
491 
492 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
493 	  BC_F_COPYOUT },
494 
495 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
496 	  BC_F_COPYOUT },
497 
498 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
499 	  BC_F_COPYIN|BC_F_COPYOUT },
500 
501 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
502 	  BC_F_COPYIN|BC_F_SUSER },
503 
504 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
505 	  BC_F_COPYIN|BC_F_SUSER },
506 
507 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
508 	  BC_F_COPYIN|BC_F_SUSER },
509 
510 };
511 const int bridge_control_table_size = nitems(bridge_control_table);
512 
513 VNET_DEFINE_STATIC(LIST_HEAD(, bridge_softc), bridge_list);
514 #define	V_bridge_list	VNET(bridge_list)
515 #define	BRIDGE_LIST_LOCK_INIT(x)	mtx_init(&V_bridge_list_mtx,	\
516 					    "if_bridge list", NULL, MTX_DEF)
517 #define	BRIDGE_LIST_LOCK_DESTROY(x)	mtx_destroy(&V_bridge_list_mtx)
518 #define	BRIDGE_LIST_LOCK(x)		mtx_lock(&V_bridge_list_mtx)
519 #define	BRIDGE_LIST_UNLOCK(x)		mtx_unlock(&V_bridge_list_mtx)
520 
521 VNET_DEFINE_STATIC(struct if_clone *, bridge_cloner);
522 #define	V_bridge_cloner	VNET(bridge_cloner)
523 
524 static const char bridge_name[] = "bridge";
525 
526 static void
527 vnet_bridge_init(const void *unused __unused)
528 {
529 
530 	BRIDGE_LIST_LOCK_INIT();
531 	LIST_INIT(&V_bridge_list);
532 	V_bridge_cloner = if_clone_simple(bridge_name,
533 	    bridge_clone_create, bridge_clone_destroy, 0);
534 }
535 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
536     vnet_bridge_init, NULL);
537 
538 static void
539 vnet_bridge_uninit(const void *unused __unused)
540 {
541 
542 	if_clone_detach(V_bridge_cloner);
543 	V_bridge_cloner = NULL;
544 	BRIDGE_LIST_LOCK_DESTROY();
545 }
546 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PSEUDO, SI_ORDER_ANY,
547     vnet_bridge_uninit, NULL);
548 
549 static int
550 bridge_modevent(module_t mod, int type, void *data)
551 {
552 
553 	switch (type) {
554 	case MOD_LOAD:
555 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
556 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
557 		    UMA_ALIGN_PTR, 0);
558 		bridge_dn_p = bridge_dummynet;
559 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
560 		    ifnet_departure_event, bridge_ifdetach, NULL,
561 		    EVENTHANDLER_PRI_ANY);
562 		break;
563 	case MOD_UNLOAD:
564 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
565 		    bridge_detach_cookie);
566 		uma_zdestroy(bridge_rtnode_zone);
567 		bridge_dn_p = NULL;
568 		break;
569 	default:
570 		return (EOPNOTSUPP);
571 	}
572 	return (0);
573 }
574 
575 static moduledata_t bridge_mod = {
576 	"if_bridge",
577 	bridge_modevent,
578 	0
579 };
580 
581 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
582 MODULE_VERSION(if_bridge, 1);
583 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
584 
585 /*
586  * handler for net.link.bridge.ipfw
587  */
588 static int
589 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
590 {
591 	int enable = V_pfil_ipfw;
592 	int error;
593 
594 	error = sysctl_handle_int(oidp, &enable, 0, req);
595 	enable &= 1;
596 
597 	if (enable != V_pfil_ipfw) {
598 		V_pfil_ipfw = enable;
599 
600 		/*
601 		 * Disable pfil so that ipfw doesnt run twice, if the user
602 		 * really wants both then they can re-enable pfil_bridge and/or
603 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
604 		 * layer2 type.
605 		 */
606 		if (V_pfil_ipfw) {
607 			V_pfil_onlyip = 0;
608 			V_pfil_bridge = 0;
609 			V_pfil_member = 0;
610 		}
611 	}
612 
613 	return (error);
614 }
615 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
616     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
617     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
618     "Layer2 filter with IPFW");
619 
620 /*
621  * bridge_clone_create:
622  *
623  *	Create a new bridge instance.
624  */
625 static int
626 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
627 {
628 	struct bridge_softc *sc, *sc2;
629 	struct ifnet *bifp, *ifp;
630 	int fb, retry;
631 	unsigned long hostid;
632 
633 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
634 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
635 	if (ifp == NULL) {
636 		free(sc, M_DEVBUF);
637 		return (ENOSPC);
638 	}
639 
640 	BRIDGE_LOCK_INIT(sc);
641 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
642 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
643 
644 	/* Initialize our routing table. */
645 	bridge_rtable_init(sc);
646 
647 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
648 
649 	LIST_INIT(&sc->sc_iflist);
650 	LIST_INIT(&sc->sc_spanlist);
651 
652 	ifp->if_softc = sc;
653 	if_initname(ifp, bridge_name, unit);
654 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
655 	ifp->if_ioctl = bridge_ioctl;
656 	ifp->if_transmit = bridge_transmit;
657 	ifp->if_qflush = bridge_qflush;
658 	ifp->if_init = bridge_init;
659 	ifp->if_type = IFT_BRIDGE;
660 
661 	/*
662 	 * Generate an ethernet address with a locally administered address.
663 	 *
664 	 * Since we are using random ethernet addresses for the bridge, it is
665 	 * possible that we might have address collisions, so make sure that
666 	 * this hardware address isn't already in use on another bridge.
667 	 * The first try uses the hostid and falls back to arc4rand().
668 	 */
669 	fb = 0;
670 	getcredhostid(curthread->td_ucred, &hostid);
671 	do {
672 		if (fb || hostid == 0) {
673 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
674 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
675 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
676 		} else {
677 			sc->sc_defaddr[0] = 0x2;
678 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
679 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
680 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
681 			sc->sc_defaddr[4] =  hostid        & 0xff;
682 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
683 		}
684 
685 		fb = 1;
686 		retry = 0;
687 		BRIDGE_LIST_LOCK();
688 		LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
689 			bifp = sc2->sc_ifp;
690 			if (memcmp(sc->sc_defaddr,
691 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
692 				retry = 1;
693 				break;
694 			}
695 		}
696 		BRIDGE_LIST_UNLOCK();
697 	} while (retry == 1);
698 
699 	bstp_attach(&sc->sc_stp, &bridge_ops);
700 	ether_ifattach(ifp, sc->sc_defaddr);
701 	/* Now undo some of the damage... */
702 	ifp->if_baudrate = 0;
703 	ifp->if_type = IFT_BRIDGE;
704 
705 	BRIDGE_LIST_LOCK();
706 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
707 	BRIDGE_LIST_UNLOCK();
708 
709 	return (0);
710 }
711 
712 /*
713  * bridge_clone_destroy:
714  *
715  *	Destroy a bridge instance.
716  */
717 static void
718 bridge_clone_destroy(struct ifnet *ifp)
719 {
720 	struct bridge_softc *sc = ifp->if_softc;
721 	struct bridge_iflist *bif;
722 
723 	BRIDGE_LOCK(sc);
724 
725 	bridge_stop(ifp, 1);
726 	ifp->if_flags &= ~IFF_UP;
727 
728 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
729 		bridge_delete_member(sc, bif, 0);
730 
731 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
732 		bridge_delete_span(sc, bif);
733 	}
734 
735 	BRIDGE_UNLOCK(sc);
736 
737 	callout_drain(&sc->sc_brcallout);
738 
739 	BRIDGE_LIST_LOCK();
740 	LIST_REMOVE(sc, sc_list);
741 	BRIDGE_LIST_UNLOCK();
742 
743 	bstp_detach(&sc->sc_stp);
744 	ether_ifdetach(ifp);
745 	if_free(ifp);
746 
747 	/* Tear down the routing table. */
748 	bridge_rtable_fini(sc);
749 
750 	BRIDGE_LOCK_DESTROY(sc);
751 	free(sc, M_DEVBUF);
752 }
753 
754 /*
755  * bridge_ioctl:
756  *
757  *	Handle a control request from the operator.
758  */
759 static int
760 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
761 {
762 	struct bridge_softc *sc = ifp->if_softc;
763 	struct ifreq *ifr = (struct ifreq *)data;
764 	struct bridge_iflist *bif;
765 	struct thread *td = curthread;
766 	union {
767 		struct ifbreq ifbreq;
768 		struct ifbifconf ifbifconf;
769 		struct ifbareq ifbareq;
770 		struct ifbaconf ifbaconf;
771 		struct ifbrparam ifbrparam;
772 		struct ifbropreq ifbropreq;
773 	} args;
774 	struct ifdrv *ifd = (struct ifdrv *) data;
775 	const struct bridge_control *bc;
776 	int error = 0;
777 
778 	switch (cmd) {
779 
780 	case SIOCADDMULTI:
781 	case SIOCDELMULTI:
782 		break;
783 
784 	case SIOCGDRVSPEC:
785 	case SIOCSDRVSPEC:
786 		if (ifd->ifd_cmd >= bridge_control_table_size) {
787 			error = EINVAL;
788 			break;
789 		}
790 		bc = &bridge_control_table[ifd->ifd_cmd];
791 
792 		if (cmd == SIOCGDRVSPEC &&
793 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
794 			error = EINVAL;
795 			break;
796 		}
797 		else if (cmd == SIOCSDRVSPEC &&
798 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
799 			error = EINVAL;
800 			break;
801 		}
802 
803 		if (bc->bc_flags & BC_F_SUSER) {
804 			error = priv_check(td, PRIV_NET_BRIDGE);
805 			if (error)
806 				break;
807 		}
808 
809 		if (ifd->ifd_len != bc->bc_argsize ||
810 		    ifd->ifd_len > sizeof(args)) {
811 			error = EINVAL;
812 			break;
813 		}
814 
815 		bzero(&args, sizeof(args));
816 		if (bc->bc_flags & BC_F_COPYIN) {
817 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
818 			if (error)
819 				break;
820 		}
821 
822 		BRIDGE_LOCK(sc);
823 		error = (*bc->bc_func)(sc, &args);
824 		BRIDGE_UNLOCK(sc);
825 		if (error)
826 			break;
827 
828 		if (bc->bc_flags & BC_F_COPYOUT)
829 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
830 
831 		break;
832 
833 	case SIOCSIFFLAGS:
834 		if (!(ifp->if_flags & IFF_UP) &&
835 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
836 			/*
837 			 * If interface is marked down and it is running,
838 			 * then stop and disable it.
839 			 */
840 			BRIDGE_LOCK(sc);
841 			bridge_stop(ifp, 1);
842 			BRIDGE_UNLOCK(sc);
843 		} else if ((ifp->if_flags & IFF_UP) &&
844 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
845 			/*
846 			 * If interface is marked up and it is stopped, then
847 			 * start it.
848 			 */
849 			(*ifp->if_init)(sc);
850 		}
851 		break;
852 
853 	case SIOCSIFMTU:
854 		if (ifr->ifr_mtu < 576) {
855 			error = EINVAL;
856 			break;
857 		}
858 		if (LIST_EMPTY(&sc->sc_iflist)) {
859 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
860 			break;
861 		}
862 		BRIDGE_LOCK(sc);
863 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
864 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
865 				log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
866 				    " != %d\n", sc->sc_ifp->if_xname,
867 				    bif->bif_ifp->if_mtu,
868 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
869 				error = EINVAL;
870 				break;
871 			}
872 		}
873 		if (!error)
874 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
875 		BRIDGE_UNLOCK(sc);
876 		break;
877 	default:
878 		/*
879 		 * drop the lock as ether_ioctl() will call bridge_start() and
880 		 * cause the lock to be recursed.
881 		 */
882 		error = ether_ioctl(ifp, cmd, data);
883 		break;
884 	}
885 
886 	return (error);
887 }
888 
889 /*
890  * bridge_mutecaps:
891  *
892  *	Clear or restore unwanted capabilities on the member interface
893  */
894 static void
895 bridge_mutecaps(struct bridge_softc *sc)
896 {
897 	struct bridge_iflist *bif;
898 	int enabled, mask;
899 
900 	/* Initial bitmask of capabilities to test */
901 	mask = BRIDGE_IFCAPS_MASK;
902 
903 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
904 		/* Every member must support it or its disabled */
905 		mask &= bif->bif_savedcaps;
906 	}
907 
908 	BRIDGE_XLOCK(sc);
909 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
910 		enabled = bif->bif_ifp->if_capenable;
911 		enabled &= ~BRIDGE_IFCAPS_STRIP;
912 		/* strip off mask bits and enable them again if allowed */
913 		enabled &= ~BRIDGE_IFCAPS_MASK;
914 		enabled |= mask;
915 		BRIDGE_UNLOCK(sc);
916 		bridge_set_ifcap(sc, bif, enabled);
917 		BRIDGE_LOCK(sc);
918 	}
919 	BRIDGE_XDROP(sc);
920 
921 }
922 
923 static void
924 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
925 {
926 	struct ifnet *ifp = bif->bif_ifp;
927 	struct ifreq ifr;
928 	int error;
929 
930 	BRIDGE_UNLOCK_ASSERT(sc);
931 
932 	bzero(&ifr, sizeof(ifr));
933 	ifr.ifr_reqcap = set;
934 
935 	if (ifp->if_capenable != set) {
936 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
937 		if (error)
938 			if_printf(sc->sc_ifp,
939 			    "error setting capabilities on %s: %d\n",
940 			    ifp->if_xname, error);
941 		if ((ifp->if_capenable & ~set) != 0)
942 			if_printf(sc->sc_ifp,
943 			    "can't disable some capabilities on %s: 0x%x\n",
944 			    ifp->if_xname, ifp->if_capenable & ~set);
945 	}
946 }
947 
948 /*
949  * bridge_lookup_member:
950  *
951  *	Lookup a bridge member interface.
952  */
953 static struct bridge_iflist *
954 bridge_lookup_member(struct bridge_softc *sc, const char *name)
955 {
956 	struct bridge_iflist *bif;
957 	struct ifnet *ifp;
958 
959 	BRIDGE_LOCK_ASSERT(sc);
960 
961 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
962 		ifp = bif->bif_ifp;
963 		if (strcmp(ifp->if_xname, name) == 0)
964 			return (bif);
965 	}
966 
967 	return (NULL);
968 }
969 
970 /*
971  * bridge_lookup_member_if:
972  *
973  *	Lookup a bridge member interface by ifnet*.
974  */
975 static struct bridge_iflist *
976 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
977 {
978 	struct bridge_iflist *bif;
979 
980 	BRIDGE_LOCK_ASSERT(sc);
981 
982 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
983 		if (bif->bif_ifp == member_ifp)
984 			return (bif);
985 	}
986 
987 	return (NULL);
988 }
989 
990 /*
991  * bridge_delete_member:
992  *
993  *	Delete the specified member interface.
994  */
995 static void
996 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
997     int gone)
998 {
999 	struct ifnet *ifs = bif->bif_ifp;
1000 	struct ifnet *fif = NULL;
1001 
1002 	BRIDGE_LOCK_ASSERT(sc);
1003 
1004 	if (bif->bif_flags & IFBIF_STP)
1005 		bstp_disable(&bif->bif_stp);
1006 
1007 	ifs->if_bridge = NULL;
1008 	BRIDGE_XLOCK(sc);
1009 	LIST_REMOVE(bif, bif_next);
1010 	BRIDGE_XDROP(sc);
1011 
1012 	/*
1013 	 * If removing the interface that gave the bridge its mac address, set
1014 	 * the mac address of the bridge to the address of the next member, or
1015 	 * to its default address if no members are left.
1016 	 */
1017 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1018 		if (LIST_EMPTY(&sc->sc_iflist)) {
1019 			bcopy(sc->sc_defaddr,
1020 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1021 			sc->sc_ifaddr = NULL;
1022 		} else {
1023 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
1024 			bcopy(IF_LLADDR(fif),
1025 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1026 			sc->sc_ifaddr = fif;
1027 		}
1028 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1029 	}
1030 
1031 	bridge_linkcheck(sc);
1032 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1033 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1034 	KASSERT(bif->bif_addrcnt == 0,
1035 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1036 
1037 	ifs->if_bridge_output = NULL;
1038 	ifs->if_bridge_input = NULL;
1039 	ifs->if_bridge_linkstate = NULL;
1040 	BRIDGE_UNLOCK(sc);
1041 	if (!gone) {
1042 		switch (ifs->if_type) {
1043 		case IFT_ETHER:
1044 		case IFT_L2VLAN:
1045 			/*
1046 			 * Take the interface out of promiscuous mode, but only
1047 			 * if it was promiscuous in the first place. It might
1048 			 * not be if we're in the bridge_ioctl_add() error path.
1049 			 */
1050 			if (ifs->if_flags & IFF_PROMISC)
1051 				(void) ifpromisc(ifs, 0);
1052 			break;
1053 
1054 		case IFT_GIF:
1055 			break;
1056 
1057 		default:
1058 #ifdef DIAGNOSTIC
1059 			panic("bridge_delete_member: impossible");
1060 #endif
1061 			break;
1062 		}
1063 		/* reneable any interface capabilities */
1064 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1065 	}
1066 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1067 	BRIDGE_LOCK(sc);
1068 	free(bif, M_DEVBUF);
1069 }
1070 
1071 /*
1072  * bridge_delete_span:
1073  *
1074  *	Delete the specified span interface.
1075  */
1076 static void
1077 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1078 {
1079 	BRIDGE_LOCK_ASSERT(sc);
1080 
1081 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1082 	    ("%s: not a span interface", __func__));
1083 
1084 	LIST_REMOVE(bif, bif_next);
1085 	free(bif, M_DEVBUF);
1086 }
1087 
1088 static int
1089 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1090 {
1091 	struct ifbreq *req = arg;
1092 	struct bridge_iflist *bif = NULL;
1093 	struct ifnet *ifs;
1094 	int error = 0;
1095 
1096 	ifs = ifunit(req->ifbr_ifsname);
1097 	if (ifs == NULL)
1098 		return (ENOENT);
1099 	if (ifs->if_ioctl == NULL)	/* must be supported */
1100 		return (EINVAL);
1101 
1102 	/* If it's in the span list, it can't be a member. */
1103 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1104 		if (ifs == bif->bif_ifp)
1105 			return (EBUSY);
1106 
1107 	if (ifs->if_bridge == sc)
1108 		return (EEXIST);
1109 
1110 	if (ifs->if_bridge != NULL)
1111 		return (EBUSY);
1112 
1113 	switch (ifs->if_type) {
1114 	case IFT_ETHER:
1115 	case IFT_L2VLAN:
1116 	case IFT_GIF:
1117 		/* permitted interface types */
1118 		break;
1119 	default:
1120 		return (EINVAL);
1121 	}
1122 
1123 #ifdef INET6
1124 	/*
1125 	 * Two valid inet6 addresses with link-local scope must not be
1126 	 * on the parent interface and the member interfaces at the
1127 	 * same time.  This restriction is needed to prevent violation
1128 	 * of link-local scope zone.  Attempts to add a member
1129 	 * interface which has inet6 addresses when the parent has
1130 	 * inet6 triggers removal of all inet6 addresses on the member
1131 	 * interface.
1132 	 */
1133 
1134 	/* Check if the parent interface has a link-local scope addr. */
1135 	if (V_allow_llz_overlap == 0 &&
1136 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1137 		/*
1138 		 * If any, remove all inet6 addresses from the member
1139 		 * interfaces.
1140 		 */
1141 		BRIDGE_XLOCK(sc);
1142 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1143  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1144 				BRIDGE_UNLOCK(sc);
1145 				in6_ifdetach(bif->bif_ifp);
1146 				BRIDGE_LOCK(sc);
1147 				if_printf(sc->sc_ifp,
1148 				    "IPv6 addresses on %s have been removed "
1149 				    "before adding it as a member to prevent "
1150 				    "IPv6 address scope violation.\n",
1151 				    bif->bif_ifp->if_xname);
1152 			}
1153 		}
1154 		BRIDGE_XDROP(sc);
1155 		if (in6ifa_llaonifp(ifs)) {
1156 			BRIDGE_UNLOCK(sc);
1157 			in6_ifdetach(ifs);
1158 			BRIDGE_LOCK(sc);
1159 			if_printf(sc->sc_ifp,
1160 			    "IPv6 addresses on %s have been removed "
1161 			    "before adding it as a member to prevent "
1162 			    "IPv6 address scope violation.\n",
1163 			    ifs->if_xname);
1164 		}
1165 	}
1166 #endif
1167 	/* Allow the first Ethernet member to define the MTU */
1168 	if (LIST_EMPTY(&sc->sc_iflist))
1169 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1170 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1171 		if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1172 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1173 		return (EINVAL);
1174 	}
1175 
1176 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1177 	if (bif == NULL)
1178 		return (ENOMEM);
1179 
1180 	bif->bif_ifp = ifs;
1181 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1182 	bif->bif_savedcaps = ifs->if_capenable;
1183 
1184 	/*
1185 	 * Assign the interface's MAC address to the bridge if it's the first
1186 	 * member and the MAC address of the bridge has not been changed from
1187 	 * the default randomly generated one.
1188 	 */
1189 	if (V_bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1190 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1191 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1192 		sc->sc_ifaddr = ifs;
1193 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1194 	}
1195 
1196 	ifs->if_bridge = sc;
1197 	ifs->if_bridge_output = bridge_output;
1198 	ifs->if_bridge_input = bridge_input;
1199 	ifs->if_bridge_linkstate = bridge_linkstate;
1200 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1201 	/*
1202 	 * XXX: XLOCK HERE!?!
1203 	 *
1204 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1205 	 */
1206 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1207 
1208 	/* Set interface capabilities to the intersection set of all members */
1209 	bridge_mutecaps(sc);
1210 	bridge_linkcheck(sc);
1211 
1212 	/* Place the interface into promiscuous mode */
1213 	switch (ifs->if_type) {
1214 		case IFT_ETHER:
1215 		case IFT_L2VLAN:
1216 			BRIDGE_UNLOCK(sc);
1217 			error = ifpromisc(ifs, 1);
1218 			BRIDGE_LOCK(sc);
1219 			break;
1220 	}
1221 
1222 	if (error)
1223 		bridge_delete_member(sc, bif, 0);
1224 	return (error);
1225 }
1226 
1227 static int
1228 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1229 {
1230 	struct ifbreq *req = arg;
1231 	struct bridge_iflist *bif;
1232 
1233 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1234 	if (bif == NULL)
1235 		return (ENOENT);
1236 
1237 	bridge_delete_member(sc, bif, 0);
1238 
1239 	return (0);
1240 }
1241 
1242 static int
1243 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1244 {
1245 	struct ifbreq *req = arg;
1246 	struct bridge_iflist *bif;
1247 	struct bstp_port *bp;
1248 
1249 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1250 	if (bif == NULL)
1251 		return (ENOENT);
1252 
1253 	bp = &bif->bif_stp;
1254 	req->ifbr_ifsflags = bif->bif_flags;
1255 	req->ifbr_state = bp->bp_state;
1256 	req->ifbr_priority = bp->bp_priority;
1257 	req->ifbr_path_cost = bp->bp_path_cost;
1258 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1259 	req->ifbr_proto = bp->bp_protover;
1260 	req->ifbr_role = bp->bp_role;
1261 	req->ifbr_stpflags = bp->bp_flags;
1262 	req->ifbr_addrcnt = bif->bif_addrcnt;
1263 	req->ifbr_addrmax = bif->bif_addrmax;
1264 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1265 
1266 	/* Copy STP state options as flags */
1267 	if (bp->bp_operedge)
1268 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1269 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1270 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1271 	if (bp->bp_ptp_link)
1272 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1273 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1274 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1275 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1276 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1277 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1278 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1279 	return (0);
1280 }
1281 
1282 static int
1283 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1284 {
1285 	struct ifbreq *req = arg;
1286 	struct bridge_iflist *bif;
1287 	struct bstp_port *bp;
1288 	int error;
1289 
1290 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1291 	if (bif == NULL)
1292 		return (ENOENT);
1293 	bp = &bif->bif_stp;
1294 
1295 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1296 		/* SPAN is readonly */
1297 		return (EINVAL);
1298 
1299 	if (req->ifbr_ifsflags & IFBIF_STP) {
1300 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1301 			error = bstp_enable(&bif->bif_stp);
1302 			if (error)
1303 				return (error);
1304 		}
1305 	} else {
1306 		if ((bif->bif_flags & IFBIF_STP) != 0)
1307 			bstp_disable(&bif->bif_stp);
1308 	}
1309 
1310 	/* Pass on STP flags */
1311 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1312 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1313 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1314 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1315 
1316 	/* Save the bits relating to the bridge */
1317 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1318 
1319 	return (0);
1320 }
1321 
1322 static int
1323 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1324 {
1325 	struct ifbrparam *param = arg;
1326 
1327 	sc->sc_brtmax = param->ifbrp_csize;
1328 	bridge_rttrim(sc);
1329 
1330 	return (0);
1331 }
1332 
1333 static int
1334 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1335 {
1336 	struct ifbrparam *param = arg;
1337 
1338 	param->ifbrp_csize = sc->sc_brtmax;
1339 
1340 	return (0);
1341 }
1342 
1343 static int
1344 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1345 {
1346 	struct ifbifconf *bifc = arg;
1347 	struct bridge_iflist *bif;
1348 	struct ifbreq breq;
1349 	char *buf, *outbuf;
1350 	int count, buflen, len, error = 0;
1351 
1352 	count = 0;
1353 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1354 		count++;
1355 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1356 		count++;
1357 
1358 	buflen = sizeof(breq) * count;
1359 	if (bifc->ifbic_len == 0) {
1360 		bifc->ifbic_len = buflen;
1361 		return (0);
1362 	}
1363 	BRIDGE_UNLOCK(sc);
1364 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1365 	BRIDGE_LOCK(sc);
1366 
1367 	count = 0;
1368 	buf = outbuf;
1369 	len = min(bifc->ifbic_len, buflen);
1370 	bzero(&breq, sizeof(breq));
1371 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1372 		if (len < sizeof(breq))
1373 			break;
1374 
1375 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1376 		    sizeof(breq.ifbr_ifsname));
1377 		/* Fill in the ifbreq structure */
1378 		error = bridge_ioctl_gifflags(sc, &breq);
1379 		if (error)
1380 			break;
1381 		memcpy(buf, &breq, sizeof(breq));
1382 		count++;
1383 		buf += sizeof(breq);
1384 		len -= sizeof(breq);
1385 	}
1386 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1387 		if (len < sizeof(breq))
1388 			break;
1389 
1390 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1391 		    sizeof(breq.ifbr_ifsname));
1392 		breq.ifbr_ifsflags = bif->bif_flags;
1393 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1394 		memcpy(buf, &breq, sizeof(breq));
1395 		count++;
1396 		buf += sizeof(breq);
1397 		len -= sizeof(breq);
1398 	}
1399 
1400 	BRIDGE_UNLOCK(sc);
1401 	bifc->ifbic_len = sizeof(breq) * count;
1402 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1403 	BRIDGE_LOCK(sc);
1404 	free(outbuf, M_TEMP);
1405 	return (error);
1406 }
1407 
1408 static int
1409 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1410 {
1411 	struct ifbaconf *bac = arg;
1412 	struct bridge_rtnode *brt;
1413 	struct ifbareq bareq;
1414 	char *buf, *outbuf;
1415 	int count, buflen, len, error = 0;
1416 
1417 	if (bac->ifbac_len == 0)
1418 		return (0);
1419 
1420 	count = 0;
1421 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1422 		count++;
1423 	buflen = sizeof(bareq) * count;
1424 
1425 	BRIDGE_UNLOCK(sc);
1426 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1427 	BRIDGE_LOCK(sc);
1428 
1429 	count = 0;
1430 	buf = outbuf;
1431 	len = min(bac->ifbac_len, buflen);
1432 	bzero(&bareq, sizeof(bareq));
1433 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1434 		if (len < sizeof(bareq))
1435 			goto out;
1436 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1437 		    sizeof(bareq.ifba_ifsname));
1438 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1439 		bareq.ifba_vlan = brt->brt_vlan;
1440 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1441 				time_uptime < brt->brt_expire)
1442 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1443 		else
1444 			bareq.ifba_expire = 0;
1445 		bareq.ifba_flags = brt->brt_flags;
1446 
1447 		memcpy(buf, &bareq, sizeof(bareq));
1448 		count++;
1449 		buf += sizeof(bareq);
1450 		len -= sizeof(bareq);
1451 	}
1452 out:
1453 	BRIDGE_UNLOCK(sc);
1454 	bac->ifbac_len = sizeof(bareq) * count;
1455 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1456 	BRIDGE_LOCK(sc);
1457 	free(outbuf, M_TEMP);
1458 	return (error);
1459 }
1460 
1461 static int
1462 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1463 {
1464 	struct ifbareq *req = arg;
1465 	struct bridge_iflist *bif;
1466 	int error;
1467 
1468 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1469 	if (bif == NULL)
1470 		return (ENOENT);
1471 
1472 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1473 	    req->ifba_flags);
1474 
1475 	return (error);
1476 }
1477 
1478 static int
1479 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1480 {
1481 	struct ifbrparam *param = arg;
1482 
1483 	sc->sc_brttimeout = param->ifbrp_ctime;
1484 	return (0);
1485 }
1486 
1487 static int
1488 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1489 {
1490 	struct ifbrparam *param = arg;
1491 
1492 	param->ifbrp_ctime = sc->sc_brttimeout;
1493 	return (0);
1494 }
1495 
1496 static int
1497 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1498 {
1499 	struct ifbareq *req = arg;
1500 
1501 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1502 }
1503 
1504 static int
1505 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1506 {
1507 	struct ifbreq *req = arg;
1508 
1509 	bridge_rtflush(sc, req->ifbr_ifsflags);
1510 	return (0);
1511 }
1512 
1513 static int
1514 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1515 {
1516 	struct ifbrparam *param = arg;
1517 	struct bstp_state *bs = &sc->sc_stp;
1518 
1519 	param->ifbrp_prio = bs->bs_bridge_priority;
1520 	return (0);
1521 }
1522 
1523 static int
1524 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1525 {
1526 	struct ifbrparam *param = arg;
1527 
1528 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1529 }
1530 
1531 static int
1532 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1533 {
1534 	struct ifbrparam *param = arg;
1535 	struct bstp_state *bs = &sc->sc_stp;
1536 
1537 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1538 	return (0);
1539 }
1540 
1541 static int
1542 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1543 {
1544 	struct ifbrparam *param = arg;
1545 
1546 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1547 }
1548 
1549 static int
1550 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1551 {
1552 	struct ifbrparam *param = arg;
1553 	struct bstp_state *bs = &sc->sc_stp;
1554 
1555 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1556 	return (0);
1557 }
1558 
1559 static int
1560 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1561 {
1562 	struct ifbrparam *param = arg;
1563 
1564 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1565 }
1566 
1567 static int
1568 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1569 {
1570 	struct ifbrparam *param = arg;
1571 	struct bstp_state *bs = &sc->sc_stp;
1572 
1573 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1574 	return (0);
1575 }
1576 
1577 static int
1578 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1579 {
1580 	struct ifbrparam *param = arg;
1581 
1582 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1583 }
1584 
1585 static int
1586 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1587 {
1588 	struct ifbreq *req = arg;
1589 	struct bridge_iflist *bif;
1590 
1591 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1592 	if (bif == NULL)
1593 		return (ENOENT);
1594 
1595 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1596 }
1597 
1598 static int
1599 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1600 {
1601 	struct ifbreq *req = arg;
1602 	struct bridge_iflist *bif;
1603 
1604 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1605 	if (bif == NULL)
1606 		return (ENOENT);
1607 
1608 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1609 }
1610 
1611 static int
1612 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1613 {
1614 	struct ifbreq *req = arg;
1615 	struct bridge_iflist *bif;
1616 
1617 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1618 	if (bif == NULL)
1619 		return (ENOENT);
1620 
1621 	bif->bif_addrmax = req->ifbr_addrmax;
1622 	return (0);
1623 }
1624 
1625 static int
1626 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1627 {
1628 	struct ifbreq *req = arg;
1629 	struct bridge_iflist *bif = NULL;
1630 	struct ifnet *ifs;
1631 
1632 	ifs = ifunit(req->ifbr_ifsname);
1633 	if (ifs == NULL)
1634 		return (ENOENT);
1635 
1636 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1637 		if (ifs == bif->bif_ifp)
1638 			return (EBUSY);
1639 
1640 	if (ifs->if_bridge != NULL)
1641 		return (EBUSY);
1642 
1643 	switch (ifs->if_type) {
1644 		case IFT_ETHER:
1645 		case IFT_GIF:
1646 		case IFT_L2VLAN:
1647 			break;
1648 		default:
1649 			return (EINVAL);
1650 	}
1651 
1652 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1653 	if (bif == NULL)
1654 		return (ENOMEM);
1655 
1656 	bif->bif_ifp = ifs;
1657 	bif->bif_flags = IFBIF_SPAN;
1658 
1659 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1660 
1661 	return (0);
1662 }
1663 
1664 static int
1665 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1666 {
1667 	struct ifbreq *req = arg;
1668 	struct bridge_iflist *bif;
1669 	struct ifnet *ifs;
1670 
1671 	ifs = ifunit(req->ifbr_ifsname);
1672 	if (ifs == NULL)
1673 		return (ENOENT);
1674 
1675 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1676 		if (ifs == bif->bif_ifp)
1677 			break;
1678 
1679 	if (bif == NULL)
1680 		return (ENOENT);
1681 
1682 	bridge_delete_span(sc, bif);
1683 
1684 	return (0);
1685 }
1686 
1687 static int
1688 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1689 {
1690 	struct ifbropreq *req = arg;
1691 	struct bstp_state *bs = &sc->sc_stp;
1692 	struct bstp_port *root_port;
1693 
1694 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1695 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1696 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1697 
1698 	root_port = bs->bs_root_port;
1699 	if (root_port == NULL)
1700 		req->ifbop_root_port = 0;
1701 	else
1702 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1703 
1704 	req->ifbop_holdcount = bs->bs_txholdcount;
1705 	req->ifbop_priority = bs->bs_bridge_priority;
1706 	req->ifbop_protocol = bs->bs_protover;
1707 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1708 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1709 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1710 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1711 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1712 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1713 
1714 	return (0);
1715 }
1716 
1717 static int
1718 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1719 {
1720 	struct ifbrparam *param = arg;
1721 
1722 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1723 	return (0);
1724 }
1725 
1726 static int
1727 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1728 {
1729 	struct ifbpstpconf *bifstp = arg;
1730 	struct bridge_iflist *bif;
1731 	struct bstp_port *bp;
1732 	struct ifbpstpreq bpreq;
1733 	char *buf, *outbuf;
1734 	int count, buflen, len, error = 0;
1735 
1736 	count = 0;
1737 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1738 		if ((bif->bif_flags & IFBIF_STP) != 0)
1739 			count++;
1740 	}
1741 
1742 	buflen = sizeof(bpreq) * count;
1743 	if (bifstp->ifbpstp_len == 0) {
1744 		bifstp->ifbpstp_len = buflen;
1745 		return (0);
1746 	}
1747 
1748 	BRIDGE_UNLOCK(sc);
1749 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1750 	BRIDGE_LOCK(sc);
1751 
1752 	count = 0;
1753 	buf = outbuf;
1754 	len = min(bifstp->ifbpstp_len, buflen);
1755 	bzero(&bpreq, sizeof(bpreq));
1756 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1757 		if (len < sizeof(bpreq))
1758 			break;
1759 
1760 		if ((bif->bif_flags & IFBIF_STP) == 0)
1761 			continue;
1762 
1763 		bp = &bif->bif_stp;
1764 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1765 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1766 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1767 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1768 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1769 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1770 
1771 		memcpy(buf, &bpreq, sizeof(bpreq));
1772 		count++;
1773 		buf += sizeof(bpreq);
1774 		len -= sizeof(bpreq);
1775 	}
1776 
1777 	BRIDGE_UNLOCK(sc);
1778 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1779 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1780 	BRIDGE_LOCK(sc);
1781 	free(outbuf, M_TEMP);
1782 	return (error);
1783 }
1784 
1785 static int
1786 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1787 {
1788 	struct ifbrparam *param = arg;
1789 
1790 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1791 }
1792 
1793 static int
1794 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1795 {
1796 	struct ifbrparam *param = arg;
1797 
1798 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1799 }
1800 
1801 /*
1802  * bridge_ifdetach:
1803  *
1804  *	Detach an interface from a bridge.  Called when a member
1805  *	interface is detaching.
1806  */
1807 static void
1808 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1809 {
1810 	struct bridge_softc *sc = ifp->if_bridge;
1811 	struct bridge_iflist *bif;
1812 
1813 	if (ifp->if_flags & IFF_RENAMING)
1814 		return;
1815 	if (V_bridge_cloner == NULL) {
1816 		/*
1817 		 * This detach handler can be called after
1818 		 * vnet_bridge_uninit().  Just return in that case.
1819 		 */
1820 		return;
1821 	}
1822 	/* Check if the interface is a bridge member */
1823 	if (sc != NULL) {
1824 		BRIDGE_LOCK(sc);
1825 
1826 		bif = bridge_lookup_member_if(sc, ifp);
1827 		if (bif != NULL)
1828 			bridge_delete_member(sc, bif, 1);
1829 
1830 		BRIDGE_UNLOCK(sc);
1831 		return;
1832 	}
1833 
1834 	/* Check if the interface is a span port */
1835 	BRIDGE_LIST_LOCK();
1836 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1837 		BRIDGE_LOCK(sc);
1838 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1839 			if (ifp == bif->bif_ifp) {
1840 				bridge_delete_span(sc, bif);
1841 				break;
1842 			}
1843 
1844 		BRIDGE_UNLOCK(sc);
1845 	}
1846 	BRIDGE_LIST_UNLOCK();
1847 }
1848 
1849 /*
1850  * bridge_init:
1851  *
1852  *	Initialize a bridge interface.
1853  */
1854 static void
1855 bridge_init(void *xsc)
1856 {
1857 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1858 	struct ifnet *ifp = sc->sc_ifp;
1859 
1860 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1861 		return;
1862 
1863 	BRIDGE_LOCK(sc);
1864 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1865 	    bridge_timer, sc);
1866 
1867 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1868 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1869 
1870 	BRIDGE_UNLOCK(sc);
1871 }
1872 
1873 /*
1874  * bridge_stop:
1875  *
1876  *	Stop the bridge interface.
1877  */
1878 static void
1879 bridge_stop(struct ifnet *ifp, int disable)
1880 {
1881 	struct bridge_softc *sc = ifp->if_softc;
1882 
1883 	BRIDGE_LOCK_ASSERT(sc);
1884 
1885 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1886 		return;
1887 
1888 	callout_stop(&sc->sc_brcallout);
1889 	bstp_stop(&sc->sc_stp);
1890 
1891 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1892 
1893 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1894 }
1895 
1896 /*
1897  * bridge_enqueue:
1898  *
1899  *	Enqueue a packet on a bridge member interface.
1900  *
1901  */
1902 static int
1903 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1904 {
1905 	int len, err = 0;
1906 	short mflags;
1907 	struct mbuf *m0;
1908 
1909 	/* We may be sending a fragment so traverse the mbuf */
1910 	for (; m; m = m0) {
1911 		m0 = m->m_nextpkt;
1912 		m->m_nextpkt = NULL;
1913 		len = m->m_pkthdr.len;
1914 		mflags = m->m_flags;
1915 
1916 		/*
1917 		 * If underlying interface can not do VLAN tag insertion itself
1918 		 * then attach a packet tag that holds it.
1919 		 */
1920 		if ((m->m_flags & M_VLANTAG) &&
1921 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1922 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1923 			if (m == NULL) {
1924 				if_printf(dst_ifp,
1925 				    "unable to prepend VLAN header\n");
1926 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1927 				continue;
1928 			}
1929 			m->m_flags &= ~M_VLANTAG;
1930 		}
1931 
1932 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1933 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1934 			m_freem(m0);
1935 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1936 			break;
1937 		}
1938 
1939 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1940 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1941 		if (mflags & M_MCAST)
1942 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
1943 	}
1944 
1945 	return (err);
1946 }
1947 
1948 /*
1949  * bridge_dummynet:
1950  *
1951  * 	Receive a queued packet from dummynet and pass it on to the output
1952  * 	interface.
1953  *
1954  *	The mbuf has the Ethernet header already attached.
1955  */
1956 static void
1957 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1958 {
1959 	struct bridge_softc *sc;
1960 
1961 	sc = ifp->if_bridge;
1962 
1963 	/*
1964 	 * The packet didnt originate from a member interface. This should only
1965 	 * ever happen if a member interface is removed while packets are
1966 	 * queued for it.
1967 	 */
1968 	if (sc == NULL) {
1969 		m_freem(m);
1970 		return;
1971 	}
1972 
1973 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1974 #ifdef INET6
1975 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1976 #endif
1977 	    ) {
1978 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1979 			return;
1980 		if (m == NULL)
1981 			return;
1982 	}
1983 
1984 	bridge_enqueue(sc, ifp, m);
1985 }
1986 
1987 /*
1988  * bridge_output:
1989  *
1990  *	Send output from a bridge member interface.  This
1991  *	performs the bridging function for locally originated
1992  *	packets.
1993  *
1994  *	The mbuf has the Ethernet header already attached.  We must
1995  *	enqueue or free the mbuf before returning.
1996  */
1997 static int
1998 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1999     struct rtentry *rt)
2000 {
2001 	struct ether_header *eh;
2002 	struct ifnet *dst_if;
2003 	struct bridge_softc *sc;
2004 	uint16_t vlan;
2005 
2006 	if (m->m_len < ETHER_HDR_LEN) {
2007 		m = m_pullup(m, ETHER_HDR_LEN);
2008 		if (m == NULL)
2009 			return (0);
2010 	}
2011 
2012 	eh = mtod(m, struct ether_header *);
2013 	sc = ifp->if_bridge;
2014 	vlan = VLANTAGOF(m);
2015 
2016 	BRIDGE_LOCK(sc);
2017 
2018 	/*
2019 	 * If bridge is down, but the original output interface is up,
2020 	 * go ahead and send out that interface.  Otherwise, the packet
2021 	 * is dropped below.
2022 	 */
2023 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2024 		dst_if = ifp;
2025 		goto sendunicast;
2026 	}
2027 
2028 	/*
2029 	 * If the packet is a multicast, or we don't know a better way to
2030 	 * get there, send to all interfaces.
2031 	 */
2032 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2033 		dst_if = NULL;
2034 	else
2035 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2036 	if (dst_if == NULL) {
2037 		struct bridge_iflist *bif;
2038 		struct mbuf *mc;
2039 		int error = 0, used = 0;
2040 
2041 		bridge_span(sc, m);
2042 
2043 		BRIDGE_LOCK2REF(sc, error);
2044 		if (error) {
2045 			m_freem(m);
2046 			return (0);
2047 		}
2048 
2049 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2050 			dst_if = bif->bif_ifp;
2051 
2052 			if (dst_if->if_type == IFT_GIF)
2053 				continue;
2054 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2055 				continue;
2056 
2057 			/*
2058 			 * If this is not the original output interface,
2059 			 * and the interface is participating in spanning
2060 			 * tree, make sure the port is in a state that
2061 			 * allows forwarding.
2062 			 */
2063 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2064 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2065 				continue;
2066 
2067 			if (LIST_NEXT(bif, bif_next) == NULL) {
2068 				used = 1;
2069 				mc = m;
2070 			} else {
2071 				mc = m_copypacket(m, M_NOWAIT);
2072 				if (mc == NULL) {
2073 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2074 					continue;
2075 				}
2076 			}
2077 
2078 			bridge_enqueue(sc, dst_if, mc);
2079 		}
2080 		if (used == 0)
2081 			m_freem(m);
2082 		BRIDGE_UNREF(sc);
2083 		return (0);
2084 	}
2085 
2086 sendunicast:
2087 	/*
2088 	 * XXX Spanning tree consideration here?
2089 	 */
2090 
2091 	bridge_span(sc, m);
2092 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2093 		m_freem(m);
2094 		BRIDGE_UNLOCK(sc);
2095 		return (0);
2096 	}
2097 
2098 	BRIDGE_UNLOCK(sc);
2099 	bridge_enqueue(sc, dst_if, m);
2100 	return (0);
2101 }
2102 
2103 /*
2104  * bridge_transmit:
2105  *
2106  *	Do output on a bridge.
2107  *
2108  */
2109 static int
2110 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2111 {
2112 	struct bridge_softc *sc;
2113 	struct ether_header *eh;
2114 	struct ifnet *dst_if;
2115 	int error = 0;
2116 
2117 	sc = ifp->if_softc;
2118 
2119 	ETHER_BPF_MTAP(ifp, m);
2120 
2121 	eh = mtod(m, struct ether_header *);
2122 
2123 	BRIDGE_LOCK(sc);
2124 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2125 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2126 		BRIDGE_UNLOCK(sc);
2127 		error = bridge_enqueue(sc, dst_if, m);
2128 	} else
2129 		bridge_broadcast(sc, ifp, m, 0);
2130 
2131 	return (error);
2132 }
2133 
2134 /*
2135  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2136  */
2137 static void
2138 bridge_qflush(struct ifnet *ifp __unused)
2139 {
2140 }
2141 
2142 /*
2143  * bridge_forward:
2144  *
2145  *	The forwarding function of the bridge.
2146  *
2147  *	NOTE: Releases the lock on return.
2148  */
2149 static void
2150 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2151     struct mbuf *m)
2152 {
2153 	struct bridge_iflist *dbif;
2154 	struct ifnet *src_if, *dst_if, *ifp;
2155 	struct ether_header *eh;
2156 	uint16_t vlan;
2157 	uint8_t *dst;
2158 	int error;
2159 
2160 	src_if = m->m_pkthdr.rcvif;
2161 	ifp = sc->sc_ifp;
2162 
2163 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2164 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2165 	vlan = VLANTAGOF(m);
2166 
2167 	if ((sbif->bif_flags & IFBIF_STP) &&
2168 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2169 		goto drop;
2170 
2171 	eh = mtod(m, struct ether_header *);
2172 	dst = eh->ether_dhost;
2173 
2174 	/* If the interface is learning, record the address. */
2175 	if (sbif->bif_flags & IFBIF_LEARNING) {
2176 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2177 		    sbif, 0, IFBAF_DYNAMIC);
2178 		/*
2179 		 * If the interface has addresses limits then deny any source
2180 		 * that is not in the cache.
2181 		 */
2182 		if (error && sbif->bif_addrmax)
2183 			goto drop;
2184 	}
2185 
2186 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2187 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2188 		goto drop;
2189 
2190 	/*
2191 	 * At this point, the port either doesn't participate
2192 	 * in spanning tree or it is in the forwarding state.
2193 	 */
2194 
2195 	/*
2196 	 * If the packet is unicast, destined for someone on
2197 	 * "this" side of the bridge, drop it.
2198 	 */
2199 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2200 		dst_if = bridge_rtlookup(sc, dst, vlan);
2201 		if (src_if == dst_if)
2202 			goto drop;
2203 	} else {
2204 		/*
2205 		 * Check if its a reserved multicast address, any address
2206 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2207 		 * bridge.
2208 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2209 		 */
2210 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2211 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2212 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2213 			goto drop;
2214 
2215 		/* ...forward it to all interfaces. */
2216 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2217 		dst_if = NULL;
2218 	}
2219 
2220 	/*
2221 	 * If we have a destination interface which is a member of our bridge,
2222 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2223 	 * For broadcast or multicast packets, don't bother because it will
2224 	 * be reinjected into ether_input. We do this before we pass the packets
2225 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2226 	 * drop the packet, or possibly modify it, making it difficult to debug
2227 	 * firewall issues on the bridge.
2228 	 */
2229 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2230 		ETHER_BPF_MTAP(ifp, m);
2231 
2232 	/* run the packet filter */
2233 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2234 #ifdef INET6
2235 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2236 #endif
2237 	    ) {
2238 		BRIDGE_UNLOCK(sc);
2239 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2240 			return;
2241 		if (m == NULL)
2242 			return;
2243 		BRIDGE_LOCK(sc);
2244 	}
2245 
2246 	if (dst_if == NULL) {
2247 		bridge_broadcast(sc, src_if, m, 1);
2248 		return;
2249 	}
2250 
2251 	/*
2252 	 * At this point, we're dealing with a unicast frame
2253 	 * going to a different interface.
2254 	 */
2255 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2256 		goto drop;
2257 
2258 	dbif = bridge_lookup_member_if(sc, dst_if);
2259 	if (dbif == NULL)
2260 		/* Not a member of the bridge (anymore?) */
2261 		goto drop;
2262 
2263 	/* Private segments can not talk to each other */
2264 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2265 		goto drop;
2266 
2267 	if ((dbif->bif_flags & IFBIF_STP) &&
2268 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2269 		goto drop;
2270 
2271 	BRIDGE_UNLOCK(sc);
2272 
2273 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2274 #ifdef INET6
2275 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2276 #endif
2277 	    ) {
2278 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2279 			return;
2280 		if (m == NULL)
2281 			return;
2282 	}
2283 
2284 	bridge_enqueue(sc, dst_if, m);
2285 	return;
2286 
2287 drop:
2288 	BRIDGE_UNLOCK(sc);
2289 	m_freem(m);
2290 }
2291 
2292 /*
2293  * bridge_input:
2294  *
2295  *	Receive input from a member interface.  Queue the packet for
2296  *	bridging if it is not for us.
2297  */
2298 static struct mbuf *
2299 bridge_input(struct ifnet *ifp, struct mbuf *m)
2300 {
2301 	struct bridge_softc *sc = ifp->if_bridge;
2302 	struct bridge_iflist *bif, *bif2;
2303 	struct ifnet *bifp;
2304 	struct ether_header *eh;
2305 	struct mbuf *mc, *mc2;
2306 	uint16_t vlan;
2307 	int error;
2308 
2309 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2310 		return (m);
2311 
2312 	bifp = sc->sc_ifp;
2313 	vlan = VLANTAGOF(m);
2314 
2315 	/*
2316 	 * Implement support for bridge monitoring. If this flag has been
2317 	 * set on this interface, discard the packet once we push it through
2318 	 * the bpf(4) machinery, but before we do, increment the byte and
2319 	 * packet counters associated with this interface.
2320 	 */
2321 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2322 		m->m_pkthdr.rcvif  = bifp;
2323 		ETHER_BPF_MTAP(bifp, m);
2324 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2325 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2326 		m_freem(m);
2327 		return (NULL);
2328 	}
2329 	BRIDGE_LOCK(sc);
2330 	bif = bridge_lookup_member_if(sc, ifp);
2331 	if (bif == NULL) {
2332 		BRIDGE_UNLOCK(sc);
2333 		return (m);
2334 	}
2335 
2336 	eh = mtod(m, struct ether_header *);
2337 
2338 	bridge_span(sc, m);
2339 
2340 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2341 		/* Tap off 802.1D packets; they do not get forwarded. */
2342 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2343 		    ETHER_ADDR_LEN) == 0) {
2344 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2345 			BRIDGE_UNLOCK(sc);
2346 			return (NULL);
2347 		}
2348 
2349 		if ((bif->bif_flags & IFBIF_STP) &&
2350 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2351 			BRIDGE_UNLOCK(sc);
2352 			return (m);
2353 		}
2354 
2355 		/*
2356 		 * Make a deep copy of the packet and enqueue the copy
2357 		 * for bridge processing; return the original packet for
2358 		 * local processing.
2359 		 */
2360 		mc = m_dup(m, M_NOWAIT);
2361 		if (mc == NULL) {
2362 			BRIDGE_UNLOCK(sc);
2363 			return (m);
2364 		}
2365 
2366 		/* Perform the bridge forwarding function with the copy. */
2367 		bridge_forward(sc, bif, mc);
2368 
2369 		/*
2370 		 * Reinject the mbuf as arriving on the bridge so we have a
2371 		 * chance at claiming multicast packets. We can not loop back
2372 		 * here from ether_input as a bridge is never a member of a
2373 		 * bridge.
2374 		 */
2375 		KASSERT(bifp->if_bridge == NULL,
2376 		    ("loop created in bridge_input"));
2377 		mc2 = m_dup(m, M_NOWAIT);
2378 		if (mc2 != NULL) {
2379 			/* Keep the layer3 header aligned */
2380 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2381 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2382 		}
2383 		if (mc2 != NULL) {
2384 			mc2->m_pkthdr.rcvif = bifp;
2385 			(*bifp->if_input)(bifp, mc2);
2386 		}
2387 
2388 		/* Return the original packet for local processing. */
2389 		return (m);
2390 	}
2391 
2392 	if ((bif->bif_flags & IFBIF_STP) &&
2393 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2394 		BRIDGE_UNLOCK(sc);
2395 		return (m);
2396 	}
2397 
2398 #if (defined(INET) || defined(INET6))
2399 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2400 	|| ((iface)->if_carp \
2401 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2402 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2403 	|| ((iface)->if_carp \
2404 	    && (*carp_forus_p)((iface), eh->ether_shost))
2405 #else
2406 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2407 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2408 #endif
2409 
2410 #ifdef INET6
2411 #   define OR_PFIL_HOOKED_INET6 \
2412 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2413 #else
2414 #   define OR_PFIL_HOOKED_INET6
2415 #endif
2416 
2417 #define GRAB_OUR_PACKETS(iface) \
2418 	if ((iface)->if_type == IFT_GIF) \
2419 		continue; \
2420 	/* It is destined for us. */ \
2421 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2422 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2423 	    ) {								\
2424 		if ((iface)->if_type == IFT_BRIDGE) {			\
2425 			ETHER_BPF_MTAP(iface, m);			\
2426 			if_inc_counter(iface, IFCOUNTER_IPACKETS, 1);				\
2427 			if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len);		\
2428 			/* Filter on the physical interface. */		\
2429 			if (V_pfil_local_phys &&			\
2430 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2431 			     OR_PFIL_HOOKED_INET6)) {			\
2432 				if (bridge_pfil(&m, NULL, ifp,		\
2433 				    PFIL_IN) != 0 || m == NULL) {	\
2434 					BRIDGE_UNLOCK(sc);		\
2435 					return (NULL);			\
2436 				}					\
2437 				eh = mtod(m, struct ether_header *);	\
2438 			}						\
2439 		}							\
2440 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2441 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2442 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2443 			if (error && bif->bif_addrmax) {		\
2444 				BRIDGE_UNLOCK(sc);			\
2445 				m_freem(m);				\
2446 				return (NULL);				\
2447 			}						\
2448 		}							\
2449 		m->m_pkthdr.rcvif = iface;				\
2450 		BRIDGE_UNLOCK(sc);					\
2451 		return (m);						\
2452 	}								\
2453 									\
2454 	/* We just received a packet that we sent out. */		\
2455 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2456 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2457 	    ) {								\
2458 		BRIDGE_UNLOCK(sc);					\
2459 		m_freem(m);						\
2460 		return (NULL);						\
2461 	}
2462 
2463 	/*
2464 	 * Unicast.  Make sure it's not for the bridge.
2465 	 */
2466 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2467 
2468 	/*
2469 	 * Give a chance for ifp at first priority. This will help when	the
2470 	 * packet comes through the interface like VLAN's with the same MACs
2471 	 * on several interfaces from the same bridge. This also will save
2472 	 * some CPU cycles in case the destination interface and the input
2473 	 * interface (eq ifp) are the same.
2474 	 */
2475 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2476 
2477 	/* Now check the all bridge members. */
2478 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2479 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2480 	}
2481 
2482 #undef OR_CARP_CHECK_WE_ARE_DST
2483 #undef OR_CARP_CHECK_WE_ARE_SRC
2484 #undef OR_PFIL_HOOKED_INET6
2485 #undef GRAB_OUR_PACKETS
2486 
2487 	/* Perform the bridge forwarding function. */
2488 	bridge_forward(sc, bif, m);
2489 
2490 	return (NULL);
2491 }
2492 
2493 /*
2494  * bridge_broadcast:
2495  *
2496  *	Send a frame to all interfaces that are members of
2497  *	the bridge, except for the one on which the packet
2498  *	arrived.
2499  *
2500  *	NOTE: Releases the lock on return.
2501  */
2502 static void
2503 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2504     struct mbuf *m, int runfilt)
2505 {
2506 	struct bridge_iflist *dbif, *sbif;
2507 	struct mbuf *mc;
2508 	struct ifnet *dst_if;
2509 	int error = 0, used = 0, i;
2510 
2511 	sbif = bridge_lookup_member_if(sc, src_if);
2512 
2513 	BRIDGE_LOCK2REF(sc, error);
2514 	if (error) {
2515 		m_freem(m);
2516 		return;
2517 	}
2518 
2519 	/* Filter on the bridge interface before broadcasting */
2520 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2521 #ifdef INET6
2522 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2523 #endif
2524 	    )) {
2525 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2526 			goto out;
2527 		if (m == NULL)
2528 			goto out;
2529 	}
2530 
2531 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2532 		dst_if = dbif->bif_ifp;
2533 		if (dst_if == src_if)
2534 			continue;
2535 
2536 		/* Private segments can not talk to each other */
2537 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2538 			continue;
2539 
2540 		if ((dbif->bif_flags & IFBIF_STP) &&
2541 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2542 			continue;
2543 
2544 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2545 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2546 			continue;
2547 
2548 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2549 			continue;
2550 
2551 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2552 			mc = m;
2553 			used = 1;
2554 		} else {
2555 			mc = m_dup(m, M_NOWAIT);
2556 			if (mc == NULL) {
2557 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2558 				continue;
2559 			}
2560 		}
2561 
2562 		/*
2563 		 * Filter on the output interface. Pass a NULL bridge interface
2564 		 * pointer so we do not redundantly filter on the bridge for
2565 		 * each interface we broadcast on.
2566 		 */
2567 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2568 #ifdef INET6
2569 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2570 #endif
2571 		    )) {
2572 			if (used == 0) {
2573 				/* Keep the layer3 header aligned */
2574 				i = min(mc->m_pkthdr.len, max_protohdr);
2575 				mc = m_copyup(mc, i, ETHER_ALIGN);
2576 				if (mc == NULL) {
2577 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2578 					continue;
2579 				}
2580 			}
2581 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2582 				continue;
2583 			if (mc == NULL)
2584 				continue;
2585 		}
2586 
2587 		bridge_enqueue(sc, dst_if, mc);
2588 	}
2589 	if (used == 0)
2590 		m_freem(m);
2591 
2592 out:
2593 	BRIDGE_UNREF(sc);
2594 }
2595 
2596 /*
2597  * bridge_span:
2598  *
2599  *	Duplicate a packet out one or more interfaces that are in span mode,
2600  *	the original mbuf is unmodified.
2601  */
2602 static void
2603 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2604 {
2605 	struct bridge_iflist *bif;
2606 	struct ifnet *dst_if;
2607 	struct mbuf *mc;
2608 
2609 	if (LIST_EMPTY(&sc->sc_spanlist))
2610 		return;
2611 
2612 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2613 		dst_if = bif->bif_ifp;
2614 
2615 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2616 			continue;
2617 
2618 		mc = m_copypacket(m, M_NOWAIT);
2619 		if (mc == NULL) {
2620 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2621 			continue;
2622 		}
2623 
2624 		bridge_enqueue(sc, dst_if, mc);
2625 	}
2626 }
2627 
2628 /*
2629  * bridge_rtupdate:
2630  *
2631  *	Add a bridge routing entry.
2632  */
2633 static int
2634 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2635     struct bridge_iflist *bif, int setflags, uint8_t flags)
2636 {
2637 	struct bridge_rtnode *brt;
2638 	int error;
2639 
2640 	BRIDGE_LOCK_ASSERT(sc);
2641 
2642 	/* Check the source address is valid and not multicast. */
2643 	if (ETHER_IS_MULTICAST(dst) ||
2644 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2645 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2646 		return (EINVAL);
2647 
2648 	/* 802.1p frames map to vlan 1 */
2649 	if (vlan == 0)
2650 		vlan = 1;
2651 
2652 	/*
2653 	 * A route for this destination might already exist.  If so,
2654 	 * update it, otherwise create a new one.
2655 	 */
2656 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2657 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2658 			sc->sc_brtexceeded++;
2659 			return (ENOSPC);
2660 		}
2661 		/* Check per interface address limits (if enabled) */
2662 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2663 			bif->bif_addrexceeded++;
2664 			return (ENOSPC);
2665 		}
2666 
2667 		/*
2668 		 * Allocate a new bridge forwarding node, and
2669 		 * initialize the expiration time and Ethernet
2670 		 * address.
2671 		 */
2672 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2673 		if (brt == NULL)
2674 			return (ENOMEM);
2675 
2676 		if (bif->bif_flags & IFBIF_STICKY)
2677 			brt->brt_flags = IFBAF_STICKY;
2678 		else
2679 			brt->brt_flags = IFBAF_DYNAMIC;
2680 
2681 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2682 		brt->brt_vlan = vlan;
2683 
2684 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2685 			uma_zfree(bridge_rtnode_zone, brt);
2686 			return (error);
2687 		}
2688 		brt->brt_dst = bif;
2689 		bif->bif_addrcnt++;
2690 	}
2691 
2692 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2693 	    brt->brt_dst != bif) {
2694 		brt->brt_dst->bif_addrcnt--;
2695 		brt->brt_dst = bif;
2696 		brt->brt_dst->bif_addrcnt++;
2697 	}
2698 
2699 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2700 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2701 	if (setflags)
2702 		brt->brt_flags = flags;
2703 
2704 	return (0);
2705 }
2706 
2707 /*
2708  * bridge_rtlookup:
2709  *
2710  *	Lookup the destination interface for an address.
2711  */
2712 static struct ifnet *
2713 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2714 {
2715 	struct bridge_rtnode *brt;
2716 
2717 	BRIDGE_LOCK_ASSERT(sc);
2718 
2719 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2720 		return (NULL);
2721 
2722 	return (brt->brt_ifp);
2723 }
2724 
2725 /*
2726  * bridge_rttrim:
2727  *
2728  *	Trim the routine table so that we have a number
2729  *	of routing entries less than or equal to the
2730  *	maximum number.
2731  */
2732 static void
2733 bridge_rttrim(struct bridge_softc *sc)
2734 {
2735 	struct bridge_rtnode *brt, *nbrt;
2736 
2737 	BRIDGE_LOCK_ASSERT(sc);
2738 
2739 	/* Make sure we actually need to do this. */
2740 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2741 		return;
2742 
2743 	/* Force an aging cycle; this might trim enough addresses. */
2744 	bridge_rtage(sc);
2745 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2746 		return;
2747 
2748 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2749 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2750 			bridge_rtnode_destroy(sc, brt);
2751 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2752 				return;
2753 		}
2754 	}
2755 }
2756 
2757 /*
2758  * bridge_timer:
2759  *
2760  *	Aging timer for the bridge.
2761  */
2762 static void
2763 bridge_timer(void *arg)
2764 {
2765 	struct bridge_softc *sc = arg;
2766 
2767 	BRIDGE_LOCK_ASSERT(sc);
2768 
2769 	bridge_rtage(sc);
2770 
2771 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2772 		callout_reset(&sc->sc_brcallout,
2773 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2774 }
2775 
2776 /*
2777  * bridge_rtage:
2778  *
2779  *	Perform an aging cycle.
2780  */
2781 static void
2782 bridge_rtage(struct bridge_softc *sc)
2783 {
2784 	struct bridge_rtnode *brt, *nbrt;
2785 
2786 	BRIDGE_LOCK_ASSERT(sc);
2787 
2788 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2789 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2790 			if (time_uptime >= brt->brt_expire)
2791 				bridge_rtnode_destroy(sc, brt);
2792 		}
2793 	}
2794 }
2795 
2796 /*
2797  * bridge_rtflush:
2798  *
2799  *	Remove all dynamic addresses from the bridge.
2800  */
2801 static void
2802 bridge_rtflush(struct bridge_softc *sc, int full)
2803 {
2804 	struct bridge_rtnode *brt, *nbrt;
2805 
2806 	BRIDGE_LOCK_ASSERT(sc);
2807 
2808 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2809 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2810 			bridge_rtnode_destroy(sc, brt);
2811 	}
2812 }
2813 
2814 /*
2815  * bridge_rtdaddr:
2816  *
2817  *	Remove an address from the table.
2818  */
2819 static int
2820 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2821 {
2822 	struct bridge_rtnode *brt;
2823 	int found = 0;
2824 
2825 	BRIDGE_LOCK_ASSERT(sc);
2826 
2827 	/*
2828 	 * If vlan is zero then we want to delete for all vlans so the lookup
2829 	 * may return more than one.
2830 	 */
2831 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2832 		bridge_rtnode_destroy(sc, brt);
2833 		found = 1;
2834 	}
2835 
2836 	return (found ? 0 : ENOENT);
2837 }
2838 
2839 /*
2840  * bridge_rtdelete:
2841  *
2842  *	Delete routes to a speicifc member interface.
2843  */
2844 static void
2845 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2846 {
2847 	struct bridge_rtnode *brt, *nbrt;
2848 
2849 	BRIDGE_LOCK_ASSERT(sc);
2850 
2851 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2852 		if (brt->brt_ifp == ifp && (full ||
2853 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2854 			bridge_rtnode_destroy(sc, brt);
2855 	}
2856 }
2857 
2858 /*
2859  * bridge_rtable_init:
2860  *
2861  *	Initialize the route table for this bridge.
2862  */
2863 static void
2864 bridge_rtable_init(struct bridge_softc *sc)
2865 {
2866 	int i;
2867 
2868 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2869 	    M_DEVBUF, M_WAITOK);
2870 
2871 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2872 		LIST_INIT(&sc->sc_rthash[i]);
2873 
2874 	sc->sc_rthash_key = arc4random();
2875 	LIST_INIT(&sc->sc_rtlist);
2876 }
2877 
2878 /*
2879  * bridge_rtable_fini:
2880  *
2881  *	Deconstruct the route table for this bridge.
2882  */
2883 static void
2884 bridge_rtable_fini(struct bridge_softc *sc)
2885 {
2886 
2887 	KASSERT(sc->sc_brtcnt == 0,
2888 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2889 	free(sc->sc_rthash, M_DEVBUF);
2890 }
2891 
2892 /*
2893  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2894  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2895  */
2896 #define	mix(a, b, c)							\
2897 do {									\
2898 	a -= b; a -= c; a ^= (c >> 13);					\
2899 	b -= c; b -= a; b ^= (a << 8);					\
2900 	c -= a; c -= b; c ^= (b >> 13);					\
2901 	a -= b; a -= c; a ^= (c >> 12);					\
2902 	b -= c; b -= a; b ^= (a << 16);					\
2903 	c -= a; c -= b; c ^= (b >> 5);					\
2904 	a -= b; a -= c; a ^= (c >> 3);					\
2905 	b -= c; b -= a; b ^= (a << 10);					\
2906 	c -= a; c -= b; c ^= (b >> 15);					\
2907 } while (/*CONSTCOND*/0)
2908 
2909 static __inline uint32_t
2910 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2911 {
2912 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2913 
2914 	b += addr[5] << 8;
2915 	b += addr[4];
2916 	a += addr[3] << 24;
2917 	a += addr[2] << 16;
2918 	a += addr[1] << 8;
2919 	a += addr[0];
2920 
2921 	mix(a, b, c);
2922 
2923 	return (c & BRIDGE_RTHASH_MASK);
2924 }
2925 
2926 #undef mix
2927 
2928 static int
2929 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2930 {
2931 	int i, d;
2932 
2933 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2934 		d = ((int)a[i]) - ((int)b[i]);
2935 	}
2936 
2937 	return (d);
2938 }
2939 
2940 /*
2941  * bridge_rtnode_lookup:
2942  *
2943  *	Look up a bridge route node for the specified destination. Compare the
2944  *	vlan id or if zero then just return the first match.
2945  */
2946 static struct bridge_rtnode *
2947 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2948 {
2949 	struct bridge_rtnode *brt;
2950 	uint32_t hash;
2951 	int dir;
2952 
2953 	BRIDGE_LOCK_ASSERT(sc);
2954 
2955 	hash = bridge_rthash(sc, addr);
2956 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2957 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2958 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2959 			return (brt);
2960 		if (dir > 0)
2961 			return (NULL);
2962 	}
2963 
2964 	return (NULL);
2965 }
2966 
2967 /*
2968  * bridge_rtnode_insert:
2969  *
2970  *	Insert the specified bridge node into the route table.  We
2971  *	assume the entry is not already in the table.
2972  */
2973 static int
2974 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2975 {
2976 	struct bridge_rtnode *lbrt;
2977 	uint32_t hash;
2978 	int dir;
2979 
2980 	BRIDGE_LOCK_ASSERT(sc);
2981 
2982 	hash = bridge_rthash(sc, brt->brt_addr);
2983 
2984 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2985 	if (lbrt == NULL) {
2986 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2987 		goto out;
2988 	}
2989 
2990 	do {
2991 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2992 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2993 			return (EEXIST);
2994 		if (dir > 0) {
2995 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2996 			goto out;
2997 		}
2998 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2999 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3000 			goto out;
3001 		}
3002 		lbrt = LIST_NEXT(lbrt, brt_hash);
3003 	} while (lbrt != NULL);
3004 
3005 #ifdef DIAGNOSTIC
3006 	panic("bridge_rtnode_insert: impossible");
3007 #endif
3008 
3009 out:
3010 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3011 	sc->sc_brtcnt++;
3012 
3013 	return (0);
3014 }
3015 
3016 /*
3017  * bridge_rtnode_destroy:
3018  *
3019  *	Destroy a bridge rtnode.
3020  */
3021 static void
3022 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3023 {
3024 	BRIDGE_LOCK_ASSERT(sc);
3025 
3026 	LIST_REMOVE(brt, brt_hash);
3027 
3028 	LIST_REMOVE(brt, brt_list);
3029 	sc->sc_brtcnt--;
3030 	brt->brt_dst->bif_addrcnt--;
3031 	uma_zfree(bridge_rtnode_zone, brt);
3032 }
3033 
3034 /*
3035  * bridge_rtable_expire:
3036  *
3037  *	Set the expiry time for all routes on an interface.
3038  */
3039 static void
3040 bridge_rtable_expire(struct ifnet *ifp, int age)
3041 {
3042 	struct bridge_softc *sc = ifp->if_bridge;
3043 	struct bridge_rtnode *brt;
3044 
3045 	BRIDGE_LOCK(sc);
3046 
3047 	/*
3048 	 * If the age is zero then flush, otherwise set all the expiry times to
3049 	 * age for the interface
3050 	 */
3051 	if (age == 0)
3052 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3053 	else {
3054 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3055 			/* Cap the expiry time to 'age' */
3056 			if (brt->brt_ifp == ifp &&
3057 			    brt->brt_expire > time_uptime + age &&
3058 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3059 				brt->brt_expire = time_uptime + age;
3060 		}
3061 	}
3062 	BRIDGE_UNLOCK(sc);
3063 }
3064 
3065 /*
3066  * bridge_state_change:
3067  *
3068  *	Callback from the bridgestp code when a port changes states.
3069  */
3070 static void
3071 bridge_state_change(struct ifnet *ifp, int state)
3072 {
3073 	struct bridge_softc *sc = ifp->if_bridge;
3074 	static const char *stpstates[] = {
3075 		"disabled",
3076 		"listening",
3077 		"learning",
3078 		"forwarding",
3079 		"blocking",
3080 		"discarding"
3081 	};
3082 
3083 	CURVNET_SET(ifp->if_vnet);
3084 	if (V_log_stp)
3085 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3086 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3087 	CURVNET_RESTORE();
3088 }
3089 
3090 /*
3091  * Send bridge packets through pfil if they are one of the types pfil can deal
3092  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3093  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3094  * that interface.
3095  */
3096 static int
3097 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3098 {
3099 	int snap, error, i, hlen;
3100 	struct ether_header *eh1, eh2;
3101 	struct ip *ip;
3102 	struct llc llc1;
3103 	u_int16_t ether_type;
3104 
3105 	snap = 0;
3106 	error = -1;	/* Default error if not error == 0 */
3107 
3108 #if 0
3109 	/* we may return with the IP fields swapped, ensure its not shared */
3110 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3111 #endif
3112 
3113 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3114 		return (0); /* filtering is disabled */
3115 
3116 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3117 	if ((*mp)->m_len < i) {
3118 	    *mp = m_pullup(*mp, i);
3119 	    if (*mp == NULL) {
3120 		printf("%s: m_pullup failed\n", __func__);
3121 		return (-1);
3122 	    }
3123 	}
3124 
3125 	eh1 = mtod(*mp, struct ether_header *);
3126 	ether_type = ntohs(eh1->ether_type);
3127 
3128 	/*
3129 	 * Check for SNAP/LLC.
3130 	 */
3131 	if (ether_type < ETHERMTU) {
3132 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3133 
3134 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3135 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3136 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3137 		    llc2->llc_control == LLC_UI) {
3138 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3139 			snap = 1;
3140 		}
3141 	}
3142 
3143 	/*
3144 	 * If we're trying to filter bridge traffic, don't look at anything
3145 	 * other than IP and ARP traffic.  If the filter doesn't understand
3146 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3147 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3148 	 * but of course we don't have an AppleTalk filter to begin with.
3149 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3150 	 * ARP traffic.)
3151 	 */
3152 	switch (ether_type) {
3153 		case ETHERTYPE_ARP:
3154 		case ETHERTYPE_REVARP:
3155 			if (V_pfil_ipfw_arp == 0)
3156 				return (0); /* Automatically pass */
3157 			break;
3158 
3159 		case ETHERTYPE_IP:
3160 #ifdef INET6
3161 		case ETHERTYPE_IPV6:
3162 #endif /* INET6 */
3163 			break;
3164 		default:
3165 			/*
3166 			 * Check to see if the user wants to pass non-ip
3167 			 * packets, these will not be checked by pfil(9) and
3168 			 * passed unconditionally so the default is to drop.
3169 			 */
3170 			if (V_pfil_onlyip)
3171 				goto bad;
3172 	}
3173 
3174 	/* Run the packet through pfil before stripping link headers */
3175 	if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3176 			dir == PFIL_OUT && ifp != NULL) {
3177 
3178 		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, 0,
3179 		    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, 0, 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, 0, 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, 0, 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, 0, 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, 0, 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, 0, 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