xref: /freebsd/sys/net/if_bridge.c (revision 052d159a8b83f03d7dc5eb31cd9a9b4a6fe3d9da)
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, mask, stuck;
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 		mask = BRIDGE_IFCAPS_MASK | BRIDGE_IFCAPS_STRIP;
942 		stuck = ifp->if_capenable & mask & ~set;
943 		if (stuck != 0)
944 			if_printf(sc->sc_ifp,
945 			    "can't disable some capabilities on %s: 0x%x\n",
946 			    ifp->if_xname, stuck);
947 	}
948 }
949 
950 /*
951  * bridge_lookup_member:
952  *
953  *	Lookup a bridge member interface.
954  */
955 static struct bridge_iflist *
956 bridge_lookup_member(struct bridge_softc *sc, const char *name)
957 {
958 	struct bridge_iflist *bif;
959 	struct ifnet *ifp;
960 
961 	BRIDGE_LOCK_ASSERT(sc);
962 
963 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
964 		ifp = bif->bif_ifp;
965 		if (strcmp(ifp->if_xname, name) == 0)
966 			return (bif);
967 	}
968 
969 	return (NULL);
970 }
971 
972 /*
973  * bridge_lookup_member_if:
974  *
975  *	Lookup a bridge member interface by ifnet*.
976  */
977 static struct bridge_iflist *
978 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
979 {
980 	struct bridge_iflist *bif;
981 
982 	BRIDGE_LOCK_ASSERT(sc);
983 
984 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
985 		if (bif->bif_ifp == member_ifp)
986 			return (bif);
987 	}
988 
989 	return (NULL);
990 }
991 
992 /*
993  * bridge_delete_member:
994  *
995  *	Delete the specified member interface.
996  */
997 static void
998 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
999     int gone)
1000 {
1001 	struct ifnet *ifs = bif->bif_ifp;
1002 	struct ifnet *fif = NULL;
1003 
1004 	BRIDGE_LOCK_ASSERT(sc);
1005 
1006 	if (bif->bif_flags & IFBIF_STP)
1007 		bstp_disable(&bif->bif_stp);
1008 
1009 	ifs->if_bridge = NULL;
1010 	BRIDGE_XLOCK(sc);
1011 	LIST_REMOVE(bif, bif_next);
1012 	BRIDGE_XDROP(sc);
1013 
1014 	/*
1015 	 * If removing the interface that gave the bridge its mac address, set
1016 	 * the mac address of the bridge to the address of the next member, or
1017 	 * to its default address if no members are left.
1018 	 */
1019 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1020 		if (LIST_EMPTY(&sc->sc_iflist)) {
1021 			bcopy(sc->sc_defaddr,
1022 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1023 			sc->sc_ifaddr = NULL;
1024 		} else {
1025 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
1026 			bcopy(IF_LLADDR(fif),
1027 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1028 			sc->sc_ifaddr = fif;
1029 		}
1030 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1031 	}
1032 
1033 	bridge_linkcheck(sc);
1034 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1035 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1036 	KASSERT(bif->bif_addrcnt == 0,
1037 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1038 
1039 	ifs->if_bridge_output = NULL;
1040 	ifs->if_bridge_input = NULL;
1041 	ifs->if_bridge_linkstate = NULL;
1042 	BRIDGE_UNLOCK(sc);
1043 	if (!gone) {
1044 		switch (ifs->if_type) {
1045 		case IFT_ETHER:
1046 		case IFT_L2VLAN:
1047 			/*
1048 			 * Take the interface out of promiscuous mode, but only
1049 			 * if it was promiscuous in the first place. It might
1050 			 * not be if we're in the bridge_ioctl_add() error path.
1051 			 */
1052 			if (ifs->if_flags & IFF_PROMISC)
1053 				(void) ifpromisc(ifs, 0);
1054 			break;
1055 
1056 		case IFT_GIF:
1057 			break;
1058 
1059 		default:
1060 #ifdef DIAGNOSTIC
1061 			panic("bridge_delete_member: impossible");
1062 #endif
1063 			break;
1064 		}
1065 		/* reneable any interface capabilities */
1066 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1067 	}
1068 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1069 	BRIDGE_LOCK(sc);
1070 	free(bif, M_DEVBUF);
1071 }
1072 
1073 /*
1074  * bridge_delete_span:
1075  *
1076  *	Delete the specified span interface.
1077  */
1078 static void
1079 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1080 {
1081 	BRIDGE_LOCK_ASSERT(sc);
1082 
1083 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1084 	    ("%s: not a span interface", __func__));
1085 
1086 	LIST_REMOVE(bif, bif_next);
1087 	free(bif, M_DEVBUF);
1088 }
1089 
1090 static int
1091 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1092 {
1093 	struct ifbreq *req = arg;
1094 	struct bridge_iflist *bif = NULL;
1095 	struct ifnet *ifs;
1096 	int error = 0;
1097 
1098 	ifs = ifunit(req->ifbr_ifsname);
1099 	if (ifs == NULL)
1100 		return (ENOENT);
1101 	if (ifs->if_ioctl == NULL)	/* must be supported */
1102 		return (EINVAL);
1103 
1104 	/* If it's in the span list, it can't be a member. */
1105 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1106 		if (ifs == bif->bif_ifp)
1107 			return (EBUSY);
1108 
1109 	if (ifs->if_bridge == sc)
1110 		return (EEXIST);
1111 
1112 	if (ifs->if_bridge != NULL)
1113 		return (EBUSY);
1114 
1115 	switch (ifs->if_type) {
1116 	case IFT_ETHER:
1117 	case IFT_L2VLAN:
1118 	case IFT_GIF:
1119 		/* permitted interface types */
1120 		break;
1121 	default:
1122 		return (EINVAL);
1123 	}
1124 
1125 #ifdef INET6
1126 	/*
1127 	 * Two valid inet6 addresses with link-local scope must not be
1128 	 * on the parent interface and the member interfaces at the
1129 	 * same time.  This restriction is needed to prevent violation
1130 	 * of link-local scope zone.  Attempts to add a member
1131 	 * interface which has inet6 addresses when the parent has
1132 	 * inet6 triggers removal of all inet6 addresses on the member
1133 	 * interface.
1134 	 */
1135 
1136 	/* Check if the parent interface has a link-local scope addr. */
1137 	if (V_allow_llz_overlap == 0 &&
1138 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1139 		/*
1140 		 * If any, remove all inet6 addresses from the member
1141 		 * interfaces.
1142 		 */
1143 		BRIDGE_XLOCK(sc);
1144 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1145  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1146 				BRIDGE_UNLOCK(sc);
1147 				in6_ifdetach(bif->bif_ifp);
1148 				BRIDGE_LOCK(sc);
1149 				if_printf(sc->sc_ifp,
1150 				    "IPv6 addresses on %s have been removed "
1151 				    "before adding it as a member to prevent "
1152 				    "IPv6 address scope violation.\n",
1153 				    bif->bif_ifp->if_xname);
1154 			}
1155 		}
1156 		BRIDGE_XDROP(sc);
1157 		if (in6ifa_llaonifp(ifs)) {
1158 			BRIDGE_UNLOCK(sc);
1159 			in6_ifdetach(ifs);
1160 			BRIDGE_LOCK(sc);
1161 			if_printf(sc->sc_ifp,
1162 			    "IPv6 addresses on %s have been removed "
1163 			    "before adding it as a member to prevent "
1164 			    "IPv6 address scope violation.\n",
1165 			    ifs->if_xname);
1166 		}
1167 	}
1168 #endif
1169 	/* Allow the first Ethernet member to define the MTU */
1170 	if (LIST_EMPTY(&sc->sc_iflist))
1171 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1172 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1173 		if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1174 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1175 		return (EINVAL);
1176 	}
1177 
1178 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1179 	if (bif == NULL)
1180 		return (ENOMEM);
1181 
1182 	bif->bif_ifp = ifs;
1183 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1184 	bif->bif_savedcaps = ifs->if_capenable;
1185 
1186 	/*
1187 	 * Assign the interface's MAC address to the bridge if it's the first
1188 	 * member and the MAC address of the bridge has not been changed from
1189 	 * the default randomly generated one.
1190 	 */
1191 	if (V_bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1192 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1193 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1194 		sc->sc_ifaddr = ifs;
1195 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1196 	}
1197 
1198 	ifs->if_bridge = sc;
1199 	ifs->if_bridge_output = bridge_output;
1200 	ifs->if_bridge_input = bridge_input;
1201 	ifs->if_bridge_linkstate = bridge_linkstate;
1202 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1203 	/*
1204 	 * XXX: XLOCK HERE!?!
1205 	 *
1206 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1207 	 */
1208 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1209 
1210 	/* Set interface capabilities to the intersection set of all members */
1211 	bridge_mutecaps(sc);
1212 	bridge_linkcheck(sc);
1213 
1214 	/* Place the interface into promiscuous mode */
1215 	switch (ifs->if_type) {
1216 		case IFT_ETHER:
1217 		case IFT_L2VLAN:
1218 			BRIDGE_UNLOCK(sc);
1219 			error = ifpromisc(ifs, 1);
1220 			BRIDGE_LOCK(sc);
1221 			break;
1222 	}
1223 
1224 	if (error)
1225 		bridge_delete_member(sc, bif, 0);
1226 	return (error);
1227 }
1228 
1229 static int
1230 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1231 {
1232 	struct ifbreq *req = arg;
1233 	struct bridge_iflist *bif;
1234 
1235 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1236 	if (bif == NULL)
1237 		return (ENOENT);
1238 
1239 	bridge_delete_member(sc, bif, 0);
1240 
1241 	return (0);
1242 }
1243 
1244 static int
1245 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1246 {
1247 	struct ifbreq *req = arg;
1248 	struct bridge_iflist *bif;
1249 	struct bstp_port *bp;
1250 
1251 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1252 	if (bif == NULL)
1253 		return (ENOENT);
1254 
1255 	bp = &bif->bif_stp;
1256 	req->ifbr_ifsflags = bif->bif_flags;
1257 	req->ifbr_state = bp->bp_state;
1258 	req->ifbr_priority = bp->bp_priority;
1259 	req->ifbr_path_cost = bp->bp_path_cost;
1260 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1261 	req->ifbr_proto = bp->bp_protover;
1262 	req->ifbr_role = bp->bp_role;
1263 	req->ifbr_stpflags = bp->bp_flags;
1264 	req->ifbr_addrcnt = bif->bif_addrcnt;
1265 	req->ifbr_addrmax = bif->bif_addrmax;
1266 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1267 
1268 	/* Copy STP state options as flags */
1269 	if (bp->bp_operedge)
1270 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1271 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1272 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1273 	if (bp->bp_ptp_link)
1274 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1275 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1276 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1277 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1278 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1279 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1280 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1281 	return (0);
1282 }
1283 
1284 static int
1285 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1286 {
1287 	struct ifbreq *req = arg;
1288 	struct bridge_iflist *bif;
1289 	struct bstp_port *bp;
1290 	int error;
1291 
1292 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1293 	if (bif == NULL)
1294 		return (ENOENT);
1295 	bp = &bif->bif_stp;
1296 
1297 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1298 		/* SPAN is readonly */
1299 		return (EINVAL);
1300 
1301 	if (req->ifbr_ifsflags & IFBIF_STP) {
1302 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1303 			error = bstp_enable(&bif->bif_stp);
1304 			if (error)
1305 				return (error);
1306 		}
1307 	} else {
1308 		if ((bif->bif_flags & IFBIF_STP) != 0)
1309 			bstp_disable(&bif->bif_stp);
1310 	}
1311 
1312 	/* Pass on STP flags */
1313 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1314 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1315 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1316 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1317 
1318 	/* Save the bits relating to the bridge */
1319 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1320 
1321 	return (0);
1322 }
1323 
1324 static int
1325 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1326 {
1327 	struct ifbrparam *param = arg;
1328 
1329 	sc->sc_brtmax = param->ifbrp_csize;
1330 	bridge_rttrim(sc);
1331 
1332 	return (0);
1333 }
1334 
1335 static int
1336 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1337 {
1338 	struct ifbrparam *param = arg;
1339 
1340 	param->ifbrp_csize = sc->sc_brtmax;
1341 
1342 	return (0);
1343 }
1344 
1345 static int
1346 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1347 {
1348 	struct ifbifconf *bifc = arg;
1349 	struct bridge_iflist *bif;
1350 	struct ifbreq breq;
1351 	char *buf, *outbuf;
1352 	int count, buflen, len, error = 0;
1353 
1354 	count = 0;
1355 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1356 		count++;
1357 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1358 		count++;
1359 
1360 	buflen = sizeof(breq) * count;
1361 	if (bifc->ifbic_len == 0) {
1362 		bifc->ifbic_len = buflen;
1363 		return (0);
1364 	}
1365 	BRIDGE_UNLOCK(sc);
1366 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1367 	BRIDGE_LOCK(sc);
1368 
1369 	count = 0;
1370 	buf = outbuf;
1371 	len = min(bifc->ifbic_len, buflen);
1372 	bzero(&breq, sizeof(breq));
1373 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1374 		if (len < sizeof(breq))
1375 			break;
1376 
1377 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1378 		    sizeof(breq.ifbr_ifsname));
1379 		/* Fill in the ifbreq structure */
1380 		error = bridge_ioctl_gifflags(sc, &breq);
1381 		if (error)
1382 			break;
1383 		memcpy(buf, &breq, sizeof(breq));
1384 		count++;
1385 		buf += sizeof(breq);
1386 		len -= sizeof(breq);
1387 	}
1388 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1389 		if (len < sizeof(breq))
1390 			break;
1391 
1392 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1393 		    sizeof(breq.ifbr_ifsname));
1394 		breq.ifbr_ifsflags = bif->bif_flags;
1395 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1396 		memcpy(buf, &breq, sizeof(breq));
1397 		count++;
1398 		buf += sizeof(breq);
1399 		len -= sizeof(breq);
1400 	}
1401 
1402 	BRIDGE_UNLOCK(sc);
1403 	bifc->ifbic_len = sizeof(breq) * count;
1404 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1405 	BRIDGE_LOCK(sc);
1406 	free(outbuf, M_TEMP);
1407 	return (error);
1408 }
1409 
1410 static int
1411 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1412 {
1413 	struct ifbaconf *bac = arg;
1414 	struct bridge_rtnode *brt;
1415 	struct ifbareq bareq;
1416 	char *buf, *outbuf;
1417 	int count, buflen, len, error = 0;
1418 
1419 	if (bac->ifbac_len == 0)
1420 		return (0);
1421 
1422 	count = 0;
1423 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1424 		count++;
1425 	buflen = sizeof(bareq) * count;
1426 
1427 	BRIDGE_UNLOCK(sc);
1428 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1429 	BRIDGE_LOCK(sc);
1430 
1431 	count = 0;
1432 	buf = outbuf;
1433 	len = min(bac->ifbac_len, buflen);
1434 	bzero(&bareq, sizeof(bareq));
1435 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1436 		if (len < sizeof(bareq))
1437 			goto out;
1438 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1439 		    sizeof(bareq.ifba_ifsname));
1440 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1441 		bareq.ifba_vlan = brt->brt_vlan;
1442 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1443 				time_uptime < brt->brt_expire)
1444 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1445 		else
1446 			bareq.ifba_expire = 0;
1447 		bareq.ifba_flags = brt->brt_flags;
1448 
1449 		memcpy(buf, &bareq, sizeof(bareq));
1450 		count++;
1451 		buf += sizeof(bareq);
1452 		len -= sizeof(bareq);
1453 	}
1454 out:
1455 	BRIDGE_UNLOCK(sc);
1456 	bac->ifbac_len = sizeof(bareq) * count;
1457 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1458 	BRIDGE_LOCK(sc);
1459 	free(outbuf, M_TEMP);
1460 	return (error);
1461 }
1462 
1463 static int
1464 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1465 {
1466 	struct ifbareq *req = arg;
1467 	struct bridge_iflist *bif;
1468 	int error;
1469 
1470 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1471 	if (bif == NULL)
1472 		return (ENOENT);
1473 
1474 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1475 	    req->ifba_flags);
1476 
1477 	return (error);
1478 }
1479 
1480 static int
1481 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1482 {
1483 	struct ifbrparam *param = arg;
1484 
1485 	sc->sc_brttimeout = param->ifbrp_ctime;
1486 	return (0);
1487 }
1488 
1489 static int
1490 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1491 {
1492 	struct ifbrparam *param = arg;
1493 
1494 	param->ifbrp_ctime = sc->sc_brttimeout;
1495 	return (0);
1496 }
1497 
1498 static int
1499 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1500 {
1501 	struct ifbareq *req = arg;
1502 
1503 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1504 }
1505 
1506 static int
1507 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1508 {
1509 	struct ifbreq *req = arg;
1510 
1511 	bridge_rtflush(sc, req->ifbr_ifsflags);
1512 	return (0);
1513 }
1514 
1515 static int
1516 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1517 {
1518 	struct ifbrparam *param = arg;
1519 	struct bstp_state *bs = &sc->sc_stp;
1520 
1521 	param->ifbrp_prio = bs->bs_bridge_priority;
1522 	return (0);
1523 }
1524 
1525 static int
1526 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1527 {
1528 	struct ifbrparam *param = arg;
1529 
1530 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1531 }
1532 
1533 static int
1534 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1535 {
1536 	struct ifbrparam *param = arg;
1537 	struct bstp_state *bs = &sc->sc_stp;
1538 
1539 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1540 	return (0);
1541 }
1542 
1543 static int
1544 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1545 {
1546 	struct ifbrparam *param = arg;
1547 
1548 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1549 }
1550 
1551 static int
1552 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1553 {
1554 	struct ifbrparam *param = arg;
1555 	struct bstp_state *bs = &sc->sc_stp;
1556 
1557 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1558 	return (0);
1559 }
1560 
1561 static int
1562 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1563 {
1564 	struct ifbrparam *param = arg;
1565 
1566 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1567 }
1568 
1569 static int
1570 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1571 {
1572 	struct ifbrparam *param = arg;
1573 	struct bstp_state *bs = &sc->sc_stp;
1574 
1575 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1576 	return (0);
1577 }
1578 
1579 static int
1580 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1581 {
1582 	struct ifbrparam *param = arg;
1583 
1584 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1585 }
1586 
1587 static int
1588 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1589 {
1590 	struct ifbreq *req = arg;
1591 	struct bridge_iflist *bif;
1592 
1593 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1594 	if (bif == NULL)
1595 		return (ENOENT);
1596 
1597 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1598 }
1599 
1600 static int
1601 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1602 {
1603 	struct ifbreq *req = arg;
1604 	struct bridge_iflist *bif;
1605 
1606 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1607 	if (bif == NULL)
1608 		return (ENOENT);
1609 
1610 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1611 }
1612 
1613 static int
1614 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1615 {
1616 	struct ifbreq *req = arg;
1617 	struct bridge_iflist *bif;
1618 
1619 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1620 	if (bif == NULL)
1621 		return (ENOENT);
1622 
1623 	bif->bif_addrmax = req->ifbr_addrmax;
1624 	return (0);
1625 }
1626 
1627 static int
1628 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1629 {
1630 	struct ifbreq *req = arg;
1631 	struct bridge_iflist *bif = NULL;
1632 	struct ifnet *ifs;
1633 
1634 	ifs = ifunit(req->ifbr_ifsname);
1635 	if (ifs == NULL)
1636 		return (ENOENT);
1637 
1638 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1639 		if (ifs == bif->bif_ifp)
1640 			return (EBUSY);
1641 
1642 	if (ifs->if_bridge != NULL)
1643 		return (EBUSY);
1644 
1645 	switch (ifs->if_type) {
1646 		case IFT_ETHER:
1647 		case IFT_GIF:
1648 		case IFT_L2VLAN:
1649 			break;
1650 		default:
1651 			return (EINVAL);
1652 	}
1653 
1654 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1655 	if (bif == NULL)
1656 		return (ENOMEM);
1657 
1658 	bif->bif_ifp = ifs;
1659 	bif->bif_flags = IFBIF_SPAN;
1660 
1661 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1662 
1663 	return (0);
1664 }
1665 
1666 static int
1667 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1668 {
1669 	struct ifbreq *req = arg;
1670 	struct bridge_iflist *bif;
1671 	struct ifnet *ifs;
1672 
1673 	ifs = ifunit(req->ifbr_ifsname);
1674 	if (ifs == NULL)
1675 		return (ENOENT);
1676 
1677 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1678 		if (ifs == bif->bif_ifp)
1679 			break;
1680 
1681 	if (bif == NULL)
1682 		return (ENOENT);
1683 
1684 	bridge_delete_span(sc, bif);
1685 
1686 	return (0);
1687 }
1688 
1689 static int
1690 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1691 {
1692 	struct ifbropreq *req = arg;
1693 	struct bstp_state *bs = &sc->sc_stp;
1694 	struct bstp_port *root_port;
1695 
1696 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1697 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1698 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1699 
1700 	root_port = bs->bs_root_port;
1701 	if (root_port == NULL)
1702 		req->ifbop_root_port = 0;
1703 	else
1704 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1705 
1706 	req->ifbop_holdcount = bs->bs_txholdcount;
1707 	req->ifbop_priority = bs->bs_bridge_priority;
1708 	req->ifbop_protocol = bs->bs_protover;
1709 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1710 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1711 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1712 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1713 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1714 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1715 
1716 	return (0);
1717 }
1718 
1719 static int
1720 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1721 {
1722 	struct ifbrparam *param = arg;
1723 
1724 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1725 	return (0);
1726 }
1727 
1728 static int
1729 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1730 {
1731 	struct ifbpstpconf *bifstp = arg;
1732 	struct bridge_iflist *bif;
1733 	struct bstp_port *bp;
1734 	struct ifbpstpreq bpreq;
1735 	char *buf, *outbuf;
1736 	int count, buflen, len, error = 0;
1737 
1738 	count = 0;
1739 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1740 		if ((bif->bif_flags & IFBIF_STP) != 0)
1741 			count++;
1742 	}
1743 
1744 	buflen = sizeof(bpreq) * count;
1745 	if (bifstp->ifbpstp_len == 0) {
1746 		bifstp->ifbpstp_len = buflen;
1747 		return (0);
1748 	}
1749 
1750 	BRIDGE_UNLOCK(sc);
1751 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1752 	BRIDGE_LOCK(sc);
1753 
1754 	count = 0;
1755 	buf = outbuf;
1756 	len = min(bifstp->ifbpstp_len, buflen);
1757 	bzero(&bpreq, sizeof(bpreq));
1758 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1759 		if (len < sizeof(bpreq))
1760 			break;
1761 
1762 		if ((bif->bif_flags & IFBIF_STP) == 0)
1763 			continue;
1764 
1765 		bp = &bif->bif_stp;
1766 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1767 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1768 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1769 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1770 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1771 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1772 
1773 		memcpy(buf, &bpreq, sizeof(bpreq));
1774 		count++;
1775 		buf += sizeof(bpreq);
1776 		len -= sizeof(bpreq);
1777 	}
1778 
1779 	BRIDGE_UNLOCK(sc);
1780 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1781 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1782 	BRIDGE_LOCK(sc);
1783 	free(outbuf, M_TEMP);
1784 	return (error);
1785 }
1786 
1787 static int
1788 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1789 {
1790 	struct ifbrparam *param = arg;
1791 
1792 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1793 }
1794 
1795 static int
1796 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1797 {
1798 	struct ifbrparam *param = arg;
1799 
1800 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1801 }
1802 
1803 /*
1804  * bridge_ifdetach:
1805  *
1806  *	Detach an interface from a bridge.  Called when a member
1807  *	interface is detaching.
1808  */
1809 static void
1810 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1811 {
1812 	struct bridge_softc *sc = ifp->if_bridge;
1813 	struct bridge_iflist *bif;
1814 
1815 	if (ifp->if_flags & IFF_RENAMING)
1816 		return;
1817 	if (V_bridge_cloner == NULL) {
1818 		/*
1819 		 * This detach handler can be called after
1820 		 * vnet_bridge_uninit().  Just return in that case.
1821 		 */
1822 		return;
1823 	}
1824 	/* Check if the interface is a bridge member */
1825 	if (sc != NULL) {
1826 		BRIDGE_LOCK(sc);
1827 
1828 		bif = bridge_lookup_member_if(sc, ifp);
1829 		if (bif != NULL)
1830 			bridge_delete_member(sc, bif, 1);
1831 
1832 		BRIDGE_UNLOCK(sc);
1833 		return;
1834 	}
1835 
1836 	/* Check if the interface is a span port */
1837 	BRIDGE_LIST_LOCK();
1838 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1839 		BRIDGE_LOCK(sc);
1840 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1841 			if (ifp == bif->bif_ifp) {
1842 				bridge_delete_span(sc, bif);
1843 				break;
1844 			}
1845 
1846 		BRIDGE_UNLOCK(sc);
1847 	}
1848 	BRIDGE_LIST_UNLOCK();
1849 }
1850 
1851 /*
1852  * bridge_init:
1853  *
1854  *	Initialize a bridge interface.
1855  */
1856 static void
1857 bridge_init(void *xsc)
1858 {
1859 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1860 	struct ifnet *ifp = sc->sc_ifp;
1861 
1862 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1863 		return;
1864 
1865 	BRIDGE_LOCK(sc);
1866 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1867 	    bridge_timer, sc);
1868 
1869 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1870 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1871 
1872 	BRIDGE_UNLOCK(sc);
1873 }
1874 
1875 /*
1876  * bridge_stop:
1877  *
1878  *	Stop the bridge interface.
1879  */
1880 static void
1881 bridge_stop(struct ifnet *ifp, int disable)
1882 {
1883 	struct bridge_softc *sc = ifp->if_softc;
1884 
1885 	BRIDGE_LOCK_ASSERT(sc);
1886 
1887 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1888 		return;
1889 
1890 	callout_stop(&sc->sc_brcallout);
1891 	bstp_stop(&sc->sc_stp);
1892 
1893 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1894 
1895 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1896 }
1897 
1898 /*
1899  * bridge_enqueue:
1900  *
1901  *	Enqueue a packet on a bridge member interface.
1902  *
1903  */
1904 static int
1905 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1906 {
1907 	int len, err = 0;
1908 	short mflags;
1909 	struct mbuf *m0;
1910 
1911 	/* We may be sending a fragment so traverse the mbuf */
1912 	for (; m; m = m0) {
1913 		m0 = m->m_nextpkt;
1914 		m->m_nextpkt = NULL;
1915 		len = m->m_pkthdr.len;
1916 		mflags = m->m_flags;
1917 
1918 		/*
1919 		 * If underlying interface can not do VLAN tag insertion itself
1920 		 * then attach a packet tag that holds it.
1921 		 */
1922 		if ((m->m_flags & M_VLANTAG) &&
1923 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1924 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1925 			if (m == NULL) {
1926 				if_printf(dst_ifp,
1927 				    "unable to prepend VLAN header\n");
1928 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1929 				continue;
1930 			}
1931 			m->m_flags &= ~M_VLANTAG;
1932 		}
1933 
1934 		M_ASSERTPKTHDR(m); /* We shouldn't transmit mbuf without pkthdr */
1935 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1936 			m_freem(m0);
1937 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1938 			break;
1939 		}
1940 
1941 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1942 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1943 		if (mflags & M_MCAST)
1944 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
1945 	}
1946 
1947 	return (err);
1948 }
1949 
1950 /*
1951  * bridge_dummynet:
1952  *
1953  * 	Receive a queued packet from dummynet and pass it on to the output
1954  * 	interface.
1955  *
1956  *	The mbuf has the Ethernet header already attached.
1957  */
1958 static void
1959 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1960 {
1961 	struct bridge_softc *sc;
1962 
1963 	sc = ifp->if_bridge;
1964 
1965 	/*
1966 	 * The packet didnt originate from a member interface. This should only
1967 	 * ever happen if a member interface is removed while packets are
1968 	 * queued for it.
1969 	 */
1970 	if (sc == NULL) {
1971 		m_freem(m);
1972 		return;
1973 	}
1974 
1975 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
1976 #ifdef INET6
1977 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
1978 #endif
1979 	    ) {
1980 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1981 			return;
1982 		if (m == NULL)
1983 			return;
1984 	}
1985 
1986 	bridge_enqueue(sc, ifp, m);
1987 }
1988 
1989 /*
1990  * bridge_output:
1991  *
1992  *	Send output from a bridge member interface.  This
1993  *	performs the bridging function for locally originated
1994  *	packets.
1995  *
1996  *	The mbuf has the Ethernet header already attached.  We must
1997  *	enqueue or free the mbuf before returning.
1998  */
1999 static int
2000 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
2001     struct rtentry *rt)
2002 {
2003 	struct ether_header *eh;
2004 	struct ifnet *dst_if;
2005 	struct bridge_softc *sc;
2006 	uint16_t vlan;
2007 
2008 	if (m->m_len < ETHER_HDR_LEN) {
2009 		m = m_pullup(m, ETHER_HDR_LEN);
2010 		if (m == NULL)
2011 			return (0);
2012 	}
2013 
2014 	eh = mtod(m, struct ether_header *);
2015 	sc = ifp->if_bridge;
2016 	vlan = VLANTAGOF(m);
2017 
2018 	BRIDGE_LOCK(sc);
2019 
2020 	/*
2021 	 * If bridge is down, but the original output interface is up,
2022 	 * go ahead and send out that interface.  Otherwise, the packet
2023 	 * is dropped below.
2024 	 */
2025 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2026 		dst_if = ifp;
2027 		goto sendunicast;
2028 	}
2029 
2030 	/*
2031 	 * If the packet is a multicast, or we don't know a better way to
2032 	 * get there, send to all interfaces.
2033 	 */
2034 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2035 		dst_if = NULL;
2036 	else
2037 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2038 	if (dst_if == NULL) {
2039 		struct bridge_iflist *bif;
2040 		struct mbuf *mc;
2041 		int error = 0, used = 0;
2042 
2043 		bridge_span(sc, m);
2044 
2045 		BRIDGE_LOCK2REF(sc, error);
2046 		if (error) {
2047 			m_freem(m);
2048 			return (0);
2049 		}
2050 
2051 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2052 			dst_if = bif->bif_ifp;
2053 
2054 			if (dst_if->if_type == IFT_GIF)
2055 				continue;
2056 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2057 				continue;
2058 
2059 			/*
2060 			 * If this is not the original output interface,
2061 			 * and the interface is participating in spanning
2062 			 * tree, make sure the port is in a state that
2063 			 * allows forwarding.
2064 			 */
2065 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2066 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2067 				continue;
2068 
2069 			if (LIST_NEXT(bif, bif_next) == NULL) {
2070 				used = 1;
2071 				mc = m;
2072 			} else {
2073 				mc = m_copypacket(m, M_NOWAIT);
2074 				if (mc == NULL) {
2075 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2076 					continue;
2077 				}
2078 			}
2079 
2080 			bridge_enqueue(sc, dst_if, mc);
2081 		}
2082 		if (used == 0)
2083 			m_freem(m);
2084 		BRIDGE_UNREF(sc);
2085 		return (0);
2086 	}
2087 
2088 sendunicast:
2089 	/*
2090 	 * XXX Spanning tree consideration here?
2091 	 */
2092 
2093 	bridge_span(sc, m);
2094 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2095 		m_freem(m);
2096 		BRIDGE_UNLOCK(sc);
2097 		return (0);
2098 	}
2099 
2100 	BRIDGE_UNLOCK(sc);
2101 	bridge_enqueue(sc, dst_if, m);
2102 	return (0);
2103 }
2104 
2105 /*
2106  * bridge_transmit:
2107  *
2108  *	Do output on a bridge.
2109  *
2110  */
2111 static int
2112 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2113 {
2114 	struct bridge_softc *sc;
2115 	struct ether_header *eh;
2116 	struct ifnet *dst_if;
2117 	int error = 0;
2118 
2119 	sc = ifp->if_softc;
2120 
2121 	ETHER_BPF_MTAP(ifp, m);
2122 
2123 	eh = mtod(m, struct ether_header *);
2124 
2125 	BRIDGE_LOCK(sc);
2126 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2127 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2128 		BRIDGE_UNLOCK(sc);
2129 		error = bridge_enqueue(sc, dst_if, m);
2130 	} else
2131 		bridge_broadcast(sc, ifp, m, 0);
2132 
2133 	return (error);
2134 }
2135 
2136 /*
2137  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2138  */
2139 static void
2140 bridge_qflush(struct ifnet *ifp __unused)
2141 {
2142 }
2143 
2144 /*
2145  * bridge_forward:
2146  *
2147  *	The forwarding function of the bridge.
2148  *
2149  *	NOTE: Releases the lock on return.
2150  */
2151 static void
2152 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2153     struct mbuf *m)
2154 {
2155 	struct bridge_iflist *dbif;
2156 	struct ifnet *src_if, *dst_if, *ifp;
2157 	struct ether_header *eh;
2158 	uint16_t vlan;
2159 	uint8_t *dst;
2160 	int error;
2161 
2162 	src_if = m->m_pkthdr.rcvif;
2163 	ifp = sc->sc_ifp;
2164 
2165 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2166 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2167 	vlan = VLANTAGOF(m);
2168 
2169 	if ((sbif->bif_flags & IFBIF_STP) &&
2170 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2171 		goto drop;
2172 
2173 	eh = mtod(m, struct ether_header *);
2174 	dst = eh->ether_dhost;
2175 
2176 	/* If the interface is learning, record the address. */
2177 	if (sbif->bif_flags & IFBIF_LEARNING) {
2178 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2179 		    sbif, 0, IFBAF_DYNAMIC);
2180 		/*
2181 		 * If the interface has addresses limits then deny any source
2182 		 * that is not in the cache.
2183 		 */
2184 		if (error && sbif->bif_addrmax)
2185 			goto drop;
2186 	}
2187 
2188 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2189 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2190 		goto drop;
2191 
2192 	/*
2193 	 * At this point, the port either doesn't participate
2194 	 * in spanning tree or it is in the forwarding state.
2195 	 */
2196 
2197 	/*
2198 	 * If the packet is unicast, destined for someone on
2199 	 * "this" side of the bridge, drop it.
2200 	 */
2201 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2202 		dst_if = bridge_rtlookup(sc, dst, vlan);
2203 		if (src_if == dst_if)
2204 			goto drop;
2205 	} else {
2206 		/*
2207 		 * Check if its a reserved multicast address, any address
2208 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2209 		 * bridge.
2210 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2211 		 */
2212 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2213 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2214 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2215 			goto drop;
2216 
2217 		/* ...forward it to all interfaces. */
2218 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2219 		dst_if = NULL;
2220 	}
2221 
2222 	/*
2223 	 * If we have a destination interface which is a member of our bridge,
2224 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2225 	 * For broadcast or multicast packets, don't bother because it will
2226 	 * be reinjected into ether_input. We do this before we pass the packets
2227 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2228 	 * drop the packet, or possibly modify it, making it difficult to debug
2229 	 * firewall issues on the bridge.
2230 	 */
2231 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2232 		ETHER_BPF_MTAP(ifp, m);
2233 
2234 	/* run the packet filter */
2235 	if (PFIL_HOOKED_IN(V_inet_pfil_head)
2236 #ifdef INET6
2237 	    || PFIL_HOOKED_IN(V_inet6_pfil_head)
2238 #endif
2239 	    ) {
2240 		BRIDGE_UNLOCK(sc);
2241 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2242 			return;
2243 		if (m == NULL)
2244 			return;
2245 		BRIDGE_LOCK(sc);
2246 	}
2247 
2248 	if (dst_if == NULL) {
2249 		bridge_broadcast(sc, src_if, m, 1);
2250 		return;
2251 	}
2252 
2253 	/*
2254 	 * At this point, we're dealing with a unicast frame
2255 	 * going to a different interface.
2256 	 */
2257 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2258 		goto drop;
2259 
2260 	dbif = bridge_lookup_member_if(sc, dst_if);
2261 	if (dbif == NULL)
2262 		/* Not a member of the bridge (anymore?) */
2263 		goto drop;
2264 
2265 	/* Private segments can not talk to each other */
2266 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2267 		goto drop;
2268 
2269 	if ((dbif->bif_flags & IFBIF_STP) &&
2270 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2271 		goto drop;
2272 
2273 	BRIDGE_UNLOCK(sc);
2274 
2275 	if (PFIL_HOOKED_OUT(V_inet_pfil_head)
2276 #ifdef INET6
2277 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2278 #endif
2279 	    ) {
2280 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2281 			return;
2282 		if (m == NULL)
2283 			return;
2284 	}
2285 
2286 	bridge_enqueue(sc, dst_if, m);
2287 	return;
2288 
2289 drop:
2290 	BRIDGE_UNLOCK(sc);
2291 	m_freem(m);
2292 }
2293 
2294 /*
2295  * bridge_input:
2296  *
2297  *	Receive input from a member interface.  Queue the packet for
2298  *	bridging if it is not for us.
2299  */
2300 static struct mbuf *
2301 bridge_input(struct ifnet *ifp, struct mbuf *m)
2302 {
2303 	struct bridge_softc *sc = ifp->if_bridge;
2304 	struct bridge_iflist *bif, *bif2;
2305 	struct ifnet *bifp;
2306 	struct ether_header *eh;
2307 	struct mbuf *mc, *mc2;
2308 	uint16_t vlan;
2309 	int error;
2310 
2311 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2312 		return (m);
2313 
2314 	bifp = sc->sc_ifp;
2315 	vlan = VLANTAGOF(m);
2316 
2317 	/*
2318 	 * Implement support for bridge monitoring. If this flag has been
2319 	 * set on this interface, discard the packet once we push it through
2320 	 * the bpf(4) machinery, but before we do, increment the byte and
2321 	 * packet counters associated with this interface.
2322 	 */
2323 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2324 		m->m_pkthdr.rcvif  = bifp;
2325 		ETHER_BPF_MTAP(bifp, m);
2326 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2327 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2328 		m_freem(m);
2329 		return (NULL);
2330 	}
2331 	BRIDGE_LOCK(sc);
2332 	bif = bridge_lookup_member_if(sc, ifp);
2333 	if (bif == NULL) {
2334 		BRIDGE_UNLOCK(sc);
2335 		return (m);
2336 	}
2337 
2338 	eh = mtod(m, struct ether_header *);
2339 
2340 	bridge_span(sc, m);
2341 
2342 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2343 		/* Tap off 802.1D packets; they do not get forwarded. */
2344 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2345 		    ETHER_ADDR_LEN) == 0) {
2346 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2347 			BRIDGE_UNLOCK(sc);
2348 			return (NULL);
2349 		}
2350 
2351 		if ((bif->bif_flags & IFBIF_STP) &&
2352 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2353 			BRIDGE_UNLOCK(sc);
2354 			return (m);
2355 		}
2356 
2357 		/*
2358 		 * Make a deep copy of the packet and enqueue the copy
2359 		 * for bridge processing; return the original packet for
2360 		 * local processing.
2361 		 */
2362 		mc = m_dup(m, M_NOWAIT);
2363 		if (mc == NULL) {
2364 			BRIDGE_UNLOCK(sc);
2365 			return (m);
2366 		}
2367 
2368 		/* Perform the bridge forwarding function with the copy. */
2369 		bridge_forward(sc, bif, mc);
2370 
2371 		/*
2372 		 * Reinject the mbuf as arriving on the bridge so we have a
2373 		 * chance at claiming multicast packets. We can not loop back
2374 		 * here from ether_input as a bridge is never a member of a
2375 		 * bridge.
2376 		 */
2377 		KASSERT(bifp->if_bridge == NULL,
2378 		    ("loop created in bridge_input"));
2379 		mc2 = m_dup(m, M_NOWAIT);
2380 		if (mc2 != NULL) {
2381 			/* Keep the layer3 header aligned */
2382 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2383 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2384 		}
2385 		if (mc2 != NULL) {
2386 			mc2->m_pkthdr.rcvif = bifp;
2387 			(*bifp->if_input)(bifp, mc2);
2388 		}
2389 
2390 		/* Return the original packet for local processing. */
2391 		return (m);
2392 	}
2393 
2394 	if ((bif->bif_flags & IFBIF_STP) &&
2395 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2396 		BRIDGE_UNLOCK(sc);
2397 		return (m);
2398 	}
2399 
2400 #if (defined(INET) || defined(INET6))
2401 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2402 	|| ((iface)->if_carp \
2403 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2404 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2405 	|| ((iface)->if_carp \
2406 	    && (*carp_forus_p)((iface), eh->ether_shost))
2407 #else
2408 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2409 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2410 #endif
2411 
2412 #ifdef INET6
2413 #   define OR_PFIL_HOOKED_INET6 \
2414 	|| PFIL_HOOKED_IN(V_inet6_pfil_head)
2415 #else
2416 #   define OR_PFIL_HOOKED_INET6
2417 #endif
2418 
2419 #define GRAB_OUR_PACKETS(iface) \
2420 	if ((iface)->if_type == IFT_GIF) \
2421 		continue; \
2422 	/* It is destined for us. */ \
2423 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2424 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2425 	    ) {								\
2426 		if ((iface)->if_type == IFT_BRIDGE) {			\
2427 			ETHER_BPF_MTAP(iface, m);			\
2428 			if_inc_counter(iface, IFCOUNTER_IPACKETS, 1);				\
2429 			if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len);		\
2430 			/* Filter on the physical interface. */		\
2431 			if (V_pfil_local_phys &&			\
2432 			    (PFIL_HOOKED_IN(V_inet_pfil_head)		\
2433 			     OR_PFIL_HOOKED_INET6)) {			\
2434 				if (bridge_pfil(&m, NULL, ifp,		\
2435 				    PFIL_IN) != 0 || m == NULL) {	\
2436 					BRIDGE_UNLOCK(sc);		\
2437 					return (NULL);			\
2438 				}					\
2439 				eh = mtod(m, struct ether_header *);	\
2440 			}						\
2441 		}							\
2442 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2443 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2444 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2445 			if (error && bif->bif_addrmax) {		\
2446 				BRIDGE_UNLOCK(sc);			\
2447 				m_freem(m);				\
2448 				return (NULL);				\
2449 			}						\
2450 		}							\
2451 		m->m_pkthdr.rcvif = iface;				\
2452 		BRIDGE_UNLOCK(sc);					\
2453 		return (m);						\
2454 	}								\
2455 									\
2456 	/* We just received a packet that we sent out. */		\
2457 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2458 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2459 	    ) {								\
2460 		BRIDGE_UNLOCK(sc);					\
2461 		m_freem(m);						\
2462 		return (NULL);						\
2463 	}
2464 
2465 	/*
2466 	 * Unicast.  Make sure it's not for the bridge.
2467 	 */
2468 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2469 
2470 	/*
2471 	 * Give a chance for ifp at first priority. This will help when	the
2472 	 * packet comes through the interface like VLAN's with the same MACs
2473 	 * on several interfaces from the same bridge. This also will save
2474 	 * some CPU cycles in case the destination interface and the input
2475 	 * interface (eq ifp) are the same.
2476 	 */
2477 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2478 
2479 	/* Now check the all bridge members. */
2480 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2481 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2482 	}
2483 
2484 #undef OR_CARP_CHECK_WE_ARE_DST
2485 #undef OR_CARP_CHECK_WE_ARE_SRC
2486 #undef OR_PFIL_HOOKED_INET6
2487 #undef GRAB_OUR_PACKETS
2488 
2489 	/* Perform the bridge forwarding function. */
2490 	bridge_forward(sc, bif, m);
2491 
2492 	return (NULL);
2493 }
2494 
2495 /*
2496  * bridge_broadcast:
2497  *
2498  *	Send a frame to all interfaces that are members of
2499  *	the bridge, except for the one on which the packet
2500  *	arrived.
2501  *
2502  *	NOTE: Releases the lock on return.
2503  */
2504 static void
2505 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2506     struct mbuf *m, int runfilt)
2507 {
2508 	struct bridge_iflist *dbif, *sbif;
2509 	struct mbuf *mc;
2510 	struct ifnet *dst_if;
2511 	int error = 0, used = 0, i;
2512 
2513 	sbif = bridge_lookup_member_if(sc, src_if);
2514 
2515 	BRIDGE_LOCK2REF(sc, error);
2516 	if (error) {
2517 		m_freem(m);
2518 		return;
2519 	}
2520 
2521 	/* Filter on the bridge interface before broadcasting */
2522 	if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2523 #ifdef INET6
2524 	    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2525 #endif
2526 	    )) {
2527 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2528 			goto out;
2529 		if (m == NULL)
2530 			goto out;
2531 	}
2532 
2533 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2534 		dst_if = dbif->bif_ifp;
2535 		if (dst_if == src_if)
2536 			continue;
2537 
2538 		/* Private segments can not talk to each other */
2539 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2540 			continue;
2541 
2542 		if ((dbif->bif_flags & IFBIF_STP) &&
2543 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2544 			continue;
2545 
2546 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2547 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2548 			continue;
2549 
2550 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2551 			continue;
2552 
2553 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2554 			mc = m;
2555 			used = 1;
2556 		} else {
2557 			mc = m_dup(m, M_NOWAIT);
2558 			if (mc == NULL) {
2559 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2560 				continue;
2561 			}
2562 		}
2563 
2564 		/*
2565 		 * Filter on the output interface. Pass a NULL bridge interface
2566 		 * pointer so we do not redundantly filter on the bridge for
2567 		 * each interface we broadcast on.
2568 		 */
2569 		if (runfilt && (PFIL_HOOKED_OUT(V_inet_pfil_head)
2570 #ifdef INET6
2571 		    || PFIL_HOOKED_OUT(V_inet6_pfil_head)
2572 #endif
2573 		    )) {
2574 			if (used == 0) {
2575 				/* Keep the layer3 header aligned */
2576 				i = min(mc->m_pkthdr.len, max_protohdr);
2577 				mc = m_copyup(mc, i, ETHER_ALIGN);
2578 				if (mc == NULL) {
2579 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2580 					continue;
2581 				}
2582 			}
2583 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2584 				continue;
2585 			if (mc == NULL)
2586 				continue;
2587 		}
2588 
2589 		bridge_enqueue(sc, dst_if, mc);
2590 	}
2591 	if (used == 0)
2592 		m_freem(m);
2593 
2594 out:
2595 	BRIDGE_UNREF(sc);
2596 }
2597 
2598 /*
2599  * bridge_span:
2600  *
2601  *	Duplicate a packet out one or more interfaces that are in span mode,
2602  *	the original mbuf is unmodified.
2603  */
2604 static void
2605 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2606 {
2607 	struct bridge_iflist *bif;
2608 	struct ifnet *dst_if;
2609 	struct mbuf *mc;
2610 
2611 	if (LIST_EMPTY(&sc->sc_spanlist))
2612 		return;
2613 
2614 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2615 		dst_if = bif->bif_ifp;
2616 
2617 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2618 			continue;
2619 
2620 		mc = m_copypacket(m, M_NOWAIT);
2621 		if (mc == NULL) {
2622 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2623 			continue;
2624 		}
2625 
2626 		bridge_enqueue(sc, dst_if, mc);
2627 	}
2628 }
2629 
2630 /*
2631  * bridge_rtupdate:
2632  *
2633  *	Add a bridge routing entry.
2634  */
2635 static int
2636 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2637     struct bridge_iflist *bif, int setflags, uint8_t flags)
2638 {
2639 	struct bridge_rtnode *brt;
2640 	int error;
2641 
2642 	BRIDGE_LOCK_ASSERT(sc);
2643 
2644 	/* Check the source address is valid and not multicast. */
2645 	if (ETHER_IS_MULTICAST(dst) ||
2646 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2647 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2648 		return (EINVAL);
2649 
2650 	/* 802.1p frames map to vlan 1 */
2651 	if (vlan == 0)
2652 		vlan = 1;
2653 
2654 	/*
2655 	 * A route for this destination might already exist.  If so,
2656 	 * update it, otherwise create a new one.
2657 	 */
2658 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2659 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2660 			sc->sc_brtexceeded++;
2661 			return (ENOSPC);
2662 		}
2663 		/* Check per interface address limits (if enabled) */
2664 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2665 			bif->bif_addrexceeded++;
2666 			return (ENOSPC);
2667 		}
2668 
2669 		/*
2670 		 * Allocate a new bridge forwarding node, and
2671 		 * initialize the expiration time and Ethernet
2672 		 * address.
2673 		 */
2674 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2675 		if (brt == NULL)
2676 			return (ENOMEM);
2677 
2678 		if (bif->bif_flags & IFBIF_STICKY)
2679 			brt->brt_flags = IFBAF_STICKY;
2680 		else
2681 			brt->brt_flags = IFBAF_DYNAMIC;
2682 
2683 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2684 		brt->brt_vlan = vlan;
2685 
2686 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2687 			uma_zfree(bridge_rtnode_zone, brt);
2688 			return (error);
2689 		}
2690 		brt->brt_dst = bif;
2691 		bif->bif_addrcnt++;
2692 	}
2693 
2694 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2695 	    brt->brt_dst != bif) {
2696 		brt->brt_dst->bif_addrcnt--;
2697 		brt->brt_dst = bif;
2698 		brt->brt_dst->bif_addrcnt++;
2699 	}
2700 
2701 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2702 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2703 	if (setflags)
2704 		brt->brt_flags = flags;
2705 
2706 	return (0);
2707 }
2708 
2709 /*
2710  * bridge_rtlookup:
2711  *
2712  *	Lookup the destination interface for an address.
2713  */
2714 static struct ifnet *
2715 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2716 {
2717 	struct bridge_rtnode *brt;
2718 
2719 	BRIDGE_LOCK_ASSERT(sc);
2720 
2721 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2722 		return (NULL);
2723 
2724 	return (brt->brt_ifp);
2725 }
2726 
2727 /*
2728  * bridge_rttrim:
2729  *
2730  *	Trim the routine table so that we have a number
2731  *	of routing entries less than or equal to the
2732  *	maximum number.
2733  */
2734 static void
2735 bridge_rttrim(struct bridge_softc *sc)
2736 {
2737 	struct bridge_rtnode *brt, *nbrt;
2738 
2739 	BRIDGE_LOCK_ASSERT(sc);
2740 
2741 	/* Make sure we actually need to do this. */
2742 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2743 		return;
2744 
2745 	/* Force an aging cycle; this might trim enough addresses. */
2746 	bridge_rtage(sc);
2747 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2748 		return;
2749 
2750 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2751 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2752 			bridge_rtnode_destroy(sc, brt);
2753 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2754 				return;
2755 		}
2756 	}
2757 }
2758 
2759 /*
2760  * bridge_timer:
2761  *
2762  *	Aging timer for the bridge.
2763  */
2764 static void
2765 bridge_timer(void *arg)
2766 {
2767 	struct bridge_softc *sc = arg;
2768 
2769 	BRIDGE_LOCK_ASSERT(sc);
2770 
2771 	bridge_rtage(sc);
2772 
2773 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2774 		callout_reset(&sc->sc_brcallout,
2775 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2776 }
2777 
2778 /*
2779  * bridge_rtage:
2780  *
2781  *	Perform an aging cycle.
2782  */
2783 static void
2784 bridge_rtage(struct bridge_softc *sc)
2785 {
2786 	struct bridge_rtnode *brt, *nbrt;
2787 
2788 	BRIDGE_LOCK_ASSERT(sc);
2789 
2790 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2791 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2792 			if (time_uptime >= brt->brt_expire)
2793 				bridge_rtnode_destroy(sc, brt);
2794 		}
2795 	}
2796 }
2797 
2798 /*
2799  * bridge_rtflush:
2800  *
2801  *	Remove all dynamic addresses from the bridge.
2802  */
2803 static void
2804 bridge_rtflush(struct bridge_softc *sc, int full)
2805 {
2806 	struct bridge_rtnode *brt, *nbrt;
2807 
2808 	BRIDGE_LOCK_ASSERT(sc);
2809 
2810 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2811 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2812 			bridge_rtnode_destroy(sc, brt);
2813 	}
2814 }
2815 
2816 /*
2817  * bridge_rtdaddr:
2818  *
2819  *	Remove an address from the table.
2820  */
2821 static int
2822 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2823 {
2824 	struct bridge_rtnode *brt;
2825 	int found = 0;
2826 
2827 	BRIDGE_LOCK_ASSERT(sc);
2828 
2829 	/*
2830 	 * If vlan is zero then we want to delete for all vlans so the lookup
2831 	 * may return more than one.
2832 	 */
2833 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2834 		bridge_rtnode_destroy(sc, brt);
2835 		found = 1;
2836 	}
2837 
2838 	return (found ? 0 : ENOENT);
2839 }
2840 
2841 /*
2842  * bridge_rtdelete:
2843  *
2844  *	Delete routes to a speicifc member interface.
2845  */
2846 static void
2847 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2848 {
2849 	struct bridge_rtnode *brt, *nbrt;
2850 
2851 	BRIDGE_LOCK_ASSERT(sc);
2852 
2853 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2854 		if (brt->brt_ifp == ifp && (full ||
2855 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2856 			bridge_rtnode_destroy(sc, brt);
2857 	}
2858 }
2859 
2860 /*
2861  * bridge_rtable_init:
2862  *
2863  *	Initialize the route table for this bridge.
2864  */
2865 static void
2866 bridge_rtable_init(struct bridge_softc *sc)
2867 {
2868 	int i;
2869 
2870 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2871 	    M_DEVBUF, M_WAITOK);
2872 
2873 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2874 		LIST_INIT(&sc->sc_rthash[i]);
2875 
2876 	sc->sc_rthash_key = arc4random();
2877 	LIST_INIT(&sc->sc_rtlist);
2878 }
2879 
2880 /*
2881  * bridge_rtable_fini:
2882  *
2883  *	Deconstruct the route table for this bridge.
2884  */
2885 static void
2886 bridge_rtable_fini(struct bridge_softc *sc)
2887 {
2888 
2889 	KASSERT(sc->sc_brtcnt == 0,
2890 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2891 	free(sc->sc_rthash, M_DEVBUF);
2892 }
2893 
2894 /*
2895  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2896  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2897  */
2898 #define	mix(a, b, c)							\
2899 do {									\
2900 	a -= b; a -= c; a ^= (c >> 13);					\
2901 	b -= c; b -= a; b ^= (a << 8);					\
2902 	c -= a; c -= b; c ^= (b >> 13);					\
2903 	a -= b; a -= c; a ^= (c >> 12);					\
2904 	b -= c; b -= a; b ^= (a << 16);					\
2905 	c -= a; c -= b; c ^= (b >> 5);					\
2906 	a -= b; a -= c; a ^= (c >> 3);					\
2907 	b -= c; b -= a; b ^= (a << 10);					\
2908 	c -= a; c -= b; c ^= (b >> 15);					\
2909 } while (/*CONSTCOND*/0)
2910 
2911 static __inline uint32_t
2912 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2913 {
2914 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2915 
2916 	b += addr[5] << 8;
2917 	b += addr[4];
2918 	a += addr[3] << 24;
2919 	a += addr[2] << 16;
2920 	a += addr[1] << 8;
2921 	a += addr[0];
2922 
2923 	mix(a, b, c);
2924 
2925 	return (c & BRIDGE_RTHASH_MASK);
2926 }
2927 
2928 #undef mix
2929 
2930 static int
2931 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2932 {
2933 	int i, d;
2934 
2935 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2936 		d = ((int)a[i]) - ((int)b[i]);
2937 	}
2938 
2939 	return (d);
2940 }
2941 
2942 /*
2943  * bridge_rtnode_lookup:
2944  *
2945  *	Look up a bridge route node for the specified destination. Compare the
2946  *	vlan id or if zero then just return the first match.
2947  */
2948 static struct bridge_rtnode *
2949 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2950 {
2951 	struct bridge_rtnode *brt;
2952 	uint32_t hash;
2953 	int dir;
2954 
2955 	BRIDGE_LOCK_ASSERT(sc);
2956 
2957 	hash = bridge_rthash(sc, addr);
2958 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2959 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2960 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2961 			return (brt);
2962 		if (dir > 0)
2963 			return (NULL);
2964 	}
2965 
2966 	return (NULL);
2967 }
2968 
2969 /*
2970  * bridge_rtnode_insert:
2971  *
2972  *	Insert the specified bridge node into the route table.  We
2973  *	assume the entry is not already in the table.
2974  */
2975 static int
2976 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2977 {
2978 	struct bridge_rtnode *lbrt;
2979 	uint32_t hash;
2980 	int dir;
2981 
2982 	BRIDGE_LOCK_ASSERT(sc);
2983 
2984 	hash = bridge_rthash(sc, brt->brt_addr);
2985 
2986 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2987 	if (lbrt == NULL) {
2988 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2989 		goto out;
2990 	}
2991 
2992 	do {
2993 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2994 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2995 			return (EEXIST);
2996 		if (dir > 0) {
2997 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2998 			goto out;
2999 		}
3000 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
3001 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
3002 			goto out;
3003 		}
3004 		lbrt = LIST_NEXT(lbrt, brt_hash);
3005 	} while (lbrt != NULL);
3006 
3007 #ifdef DIAGNOSTIC
3008 	panic("bridge_rtnode_insert: impossible");
3009 #endif
3010 
3011 out:
3012 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
3013 	sc->sc_brtcnt++;
3014 
3015 	return (0);
3016 }
3017 
3018 /*
3019  * bridge_rtnode_destroy:
3020  *
3021  *	Destroy a bridge rtnode.
3022  */
3023 static void
3024 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3025 {
3026 	BRIDGE_LOCK_ASSERT(sc);
3027 
3028 	LIST_REMOVE(brt, brt_hash);
3029 
3030 	LIST_REMOVE(brt, brt_list);
3031 	sc->sc_brtcnt--;
3032 	brt->brt_dst->bif_addrcnt--;
3033 	uma_zfree(bridge_rtnode_zone, brt);
3034 }
3035 
3036 /*
3037  * bridge_rtable_expire:
3038  *
3039  *	Set the expiry time for all routes on an interface.
3040  */
3041 static void
3042 bridge_rtable_expire(struct ifnet *ifp, int age)
3043 {
3044 	struct bridge_softc *sc = ifp->if_bridge;
3045 	struct bridge_rtnode *brt;
3046 
3047 	BRIDGE_LOCK(sc);
3048 
3049 	/*
3050 	 * If the age is zero then flush, otherwise set all the expiry times to
3051 	 * age for the interface
3052 	 */
3053 	if (age == 0)
3054 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3055 	else {
3056 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3057 			/* Cap the expiry time to 'age' */
3058 			if (brt->brt_ifp == ifp &&
3059 			    brt->brt_expire > time_uptime + age &&
3060 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3061 				brt->brt_expire = time_uptime + age;
3062 		}
3063 	}
3064 	BRIDGE_UNLOCK(sc);
3065 }
3066 
3067 /*
3068  * bridge_state_change:
3069  *
3070  *	Callback from the bridgestp code when a port changes states.
3071  */
3072 static void
3073 bridge_state_change(struct ifnet *ifp, int state)
3074 {
3075 	struct bridge_softc *sc = ifp->if_bridge;
3076 	static const char *stpstates[] = {
3077 		"disabled",
3078 		"listening",
3079 		"learning",
3080 		"forwarding",
3081 		"blocking",
3082 		"discarding"
3083 	};
3084 
3085 	CURVNET_SET(ifp->if_vnet);
3086 	if (V_log_stp)
3087 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3088 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3089 	CURVNET_RESTORE();
3090 }
3091 
3092 /*
3093  * Send bridge packets through pfil if they are one of the types pfil can deal
3094  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3095  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3096  * that interface.
3097  */
3098 static int
3099 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3100 {
3101 	int snap, error, i, hlen;
3102 	struct ether_header *eh1, eh2;
3103 	struct ip *ip;
3104 	struct llc llc1;
3105 	u_int16_t ether_type;
3106 	pfil_return_t rv;
3107 
3108 	snap = 0;
3109 	error = -1;	/* Default error if not error == 0 */
3110 
3111 #if 0
3112 	/* we may return with the IP fields swapped, ensure its not shared */
3113 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3114 #endif
3115 
3116 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3117 		return (0); /* filtering is disabled */
3118 
3119 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3120 	if ((*mp)->m_len < i) {
3121 	    *mp = m_pullup(*mp, i);
3122 	    if (*mp == NULL) {
3123 		printf("%s: m_pullup failed\n", __func__);
3124 		return (-1);
3125 	    }
3126 	}
3127 
3128 	eh1 = mtod(*mp, struct ether_header *);
3129 	ether_type = ntohs(eh1->ether_type);
3130 
3131 	/*
3132 	 * Check for SNAP/LLC.
3133 	 */
3134 	if (ether_type < ETHERMTU) {
3135 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3136 
3137 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3138 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3139 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3140 		    llc2->llc_control == LLC_UI) {
3141 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3142 			snap = 1;
3143 		}
3144 	}
3145 
3146 	/*
3147 	 * If we're trying to filter bridge traffic, don't look at anything
3148 	 * other than IP and ARP traffic.  If the filter doesn't understand
3149 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3150 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3151 	 * but of course we don't have an AppleTalk filter to begin with.
3152 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3153 	 * ARP traffic.)
3154 	 */
3155 	switch (ether_type) {
3156 		case ETHERTYPE_ARP:
3157 		case ETHERTYPE_REVARP:
3158 			if (V_pfil_ipfw_arp == 0)
3159 				return (0); /* Automatically pass */
3160 			break;
3161 
3162 		case ETHERTYPE_IP:
3163 #ifdef INET6
3164 		case ETHERTYPE_IPV6:
3165 #endif /* INET6 */
3166 			break;
3167 		default:
3168 			/*
3169 			 * Check to see if the user wants to pass non-ip
3170 			 * packets, these will not be checked by pfil(9) and
3171 			 * passed unconditionally so the default is to drop.
3172 			 */
3173 			if (V_pfil_onlyip)
3174 				goto bad;
3175 	}
3176 
3177 	/* Run the packet through pfil before stripping link headers */
3178 	if (PFIL_HOOKED_OUT(V_link_pfil_head) && V_pfil_ipfw != 0 &&
3179 	    dir == PFIL_OUT && ifp != NULL) {
3180 		switch (pfil_run_hooks(V_link_pfil_head, mp, ifp, dir, NULL)) {
3181 		case PFIL_DROPPED:
3182 			return (EPERM);
3183 		case PFIL_CONSUMED:
3184 			return (0);
3185 		}
3186 	}
3187 
3188 	/* Strip off the Ethernet header and keep a copy. */
3189 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3190 	m_adj(*mp, ETHER_HDR_LEN);
3191 
3192 	/* Strip off snap header, if present */
3193 	if (snap) {
3194 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3195 		m_adj(*mp, sizeof(struct llc));
3196 	}
3197 
3198 	/*
3199 	 * Check the IP header for alignment and errors
3200 	 */
3201 	if (dir == PFIL_IN) {
3202 		switch (ether_type) {
3203 			case ETHERTYPE_IP:
3204 				error = bridge_ip_checkbasic(mp);
3205 				break;
3206 #ifdef INET6
3207 			case ETHERTYPE_IPV6:
3208 				error = bridge_ip6_checkbasic(mp);
3209 				break;
3210 #endif /* INET6 */
3211 			default:
3212 				error = 0;
3213 		}
3214 		if (error)
3215 			goto bad;
3216 	}
3217 
3218 	error = 0;
3219 
3220 	/*
3221 	 * Run the packet through pfil
3222 	 */
3223 	rv = PFIL_PASS;
3224 	switch (ether_type) {
3225 	case ETHERTYPE_IP:
3226 		/*
3227 		 * Run pfil on the member interface and the bridge, both can
3228 		 * be skipped by clearing pfil_member or pfil_bridge.
3229 		 *
3230 		 * Keep the order:
3231 		 *   in_if -> bridge_if -> out_if
3232 		 */
3233 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3234 		    pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3235 		    PFIL_PASS)
3236 			break;
3237 
3238 		if (V_pfil_member && ifp != NULL && (rv =
3239 		    pfil_run_hooks(V_inet_pfil_head, mp, ifp, dir, NULL)) !=
3240 		    PFIL_PASS)
3241 			break;
3242 
3243 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3244 		    pfil_run_hooks(V_inet_pfil_head, mp, bifp, dir, NULL)) !=
3245 		    PFIL_PASS)
3246 			break;
3247 
3248 		/* check if we need to fragment the packet */
3249 		/* bridge_fragment generates a mbuf chain of packets */
3250 		/* that already include eth headers */
3251 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3252 			i = (*mp)->m_pkthdr.len;
3253 			if (i > ifp->if_mtu) {
3254 				error = bridge_fragment(ifp, mp, &eh2, snap,
3255 					    &llc1);
3256 				return (error);
3257 			}
3258 		}
3259 
3260 		/* Recalculate the ip checksum. */
3261 		ip = mtod(*mp, struct ip *);
3262 		hlen = ip->ip_hl << 2;
3263 		if (hlen < sizeof(struct ip))
3264 			goto bad;
3265 		if (hlen > (*mp)->m_len) {
3266 			if ((*mp = m_pullup(*mp, hlen)) == NULL)
3267 				goto bad;
3268 			ip = mtod(*mp, struct ip *);
3269 			if (ip == NULL)
3270 				goto bad;
3271 		}
3272 		ip->ip_sum = 0;
3273 		if (hlen == sizeof(struct ip))
3274 			ip->ip_sum = in_cksum_hdr(ip);
3275 		else
3276 			ip->ip_sum = in_cksum(*mp, hlen);
3277 
3278 		break;
3279 #ifdef INET6
3280 	case ETHERTYPE_IPV6:
3281 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL && (rv =
3282 		    pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3283 		    PFIL_PASS)
3284 			break;
3285 
3286 		if (V_pfil_member && ifp != NULL && (rv =
3287 		    pfil_run_hooks(V_inet6_pfil_head, mp, ifp, dir, NULL)) !=
3288 		    PFIL_PASS)
3289 			break;
3290 
3291 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL && (rv =
3292 		    pfil_run_hooks(V_inet6_pfil_head, mp, bifp, dir, NULL)) !=
3293 		    PFIL_PASS)
3294 			break;
3295 		break;
3296 #endif
3297 	}
3298 
3299 	switch (rv) {
3300 	case PFIL_CONSUMED:
3301 		return (0);
3302 	case PFIL_DROPPED:
3303 		return (EPERM);
3304 	default:
3305 		break;
3306 	}
3307 
3308 	error = -1;
3309 
3310 	/*
3311 	 * Finally, put everything back the way it was and return
3312 	 */
3313 	if (snap) {
3314 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3315 		if (*mp == NULL)
3316 			return (error);
3317 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3318 	}
3319 
3320 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3321 	if (*mp == NULL)
3322 		return (error);
3323 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3324 
3325 	return (0);
3326 
3327 bad:
3328 	m_freem(*mp);
3329 	*mp = NULL;
3330 	return (error);
3331 }
3332 
3333 /*
3334  * Perform basic checks on header size since
3335  * pfil assumes ip_input has already processed
3336  * it for it.  Cut-and-pasted from ip_input.c.
3337  * Given how simple the IPv6 version is,
3338  * does the IPv4 version really need to be
3339  * this complicated?
3340  *
3341  * XXX Should we update ipstat here, or not?
3342  * XXX Right now we update ipstat but not
3343  * XXX csum_counter.
3344  */
3345 static int
3346 bridge_ip_checkbasic(struct mbuf **mp)
3347 {
3348 	struct mbuf *m = *mp;
3349 	struct ip *ip;
3350 	int len, hlen;
3351 	u_short sum;
3352 
3353 	if (*mp == NULL)
3354 		return (-1);
3355 
3356 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3357 		if ((m = m_copyup(m, sizeof(struct ip),
3358 			(max_linkhdr + 3) & ~3)) == NULL) {
3359 			/* XXXJRT new stat, please */
3360 			KMOD_IPSTAT_INC(ips_toosmall);
3361 			goto bad;
3362 		}
3363 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3364 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3365 			KMOD_IPSTAT_INC(ips_toosmall);
3366 			goto bad;
3367 		}
3368 	}
3369 	ip = mtod(m, struct ip *);
3370 	if (ip == NULL) goto bad;
3371 
3372 	if (ip->ip_v != IPVERSION) {
3373 		KMOD_IPSTAT_INC(ips_badvers);
3374 		goto bad;
3375 	}
3376 	hlen = ip->ip_hl << 2;
3377 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3378 		KMOD_IPSTAT_INC(ips_badhlen);
3379 		goto bad;
3380 	}
3381 	if (hlen > m->m_len) {
3382 		if ((m = m_pullup(m, hlen)) == NULL) {
3383 			KMOD_IPSTAT_INC(ips_badhlen);
3384 			goto bad;
3385 		}
3386 		ip = mtod(m, struct ip *);
3387 		if (ip == NULL) goto bad;
3388 	}
3389 
3390 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3391 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3392 	} else {
3393 		if (hlen == sizeof(struct ip)) {
3394 			sum = in_cksum_hdr(ip);
3395 		} else {
3396 			sum = in_cksum(m, hlen);
3397 		}
3398 	}
3399 	if (sum) {
3400 		KMOD_IPSTAT_INC(ips_badsum);
3401 		goto bad;
3402 	}
3403 
3404 	/* Retrieve the packet length. */
3405 	len = ntohs(ip->ip_len);
3406 
3407 	/*
3408 	 * Check for additional length bogosity
3409 	 */
3410 	if (len < hlen) {
3411 		KMOD_IPSTAT_INC(ips_badlen);
3412 		goto bad;
3413 	}
3414 
3415 	/*
3416 	 * Check that the amount of data in the buffers
3417 	 * is as at least much as the IP header would have us expect.
3418 	 * Drop packet if shorter than we expect.
3419 	 */
3420 	if (m->m_pkthdr.len < len) {
3421 		KMOD_IPSTAT_INC(ips_tooshort);
3422 		goto bad;
3423 	}
3424 
3425 	/* Checks out, proceed */
3426 	*mp = m;
3427 	return (0);
3428 
3429 bad:
3430 	*mp = m;
3431 	return (-1);
3432 }
3433 
3434 #ifdef INET6
3435 /*
3436  * Same as above, but for IPv6.
3437  * Cut-and-pasted from ip6_input.c.
3438  * XXX Should we update ip6stat, or not?
3439  */
3440 static int
3441 bridge_ip6_checkbasic(struct mbuf **mp)
3442 {
3443 	struct mbuf *m = *mp;
3444 	struct ip6_hdr *ip6;
3445 
3446 	/*
3447 	 * If the IPv6 header is not aligned, slurp it up into a new
3448 	 * mbuf with space for link headers, in the event we forward
3449 	 * it.  Otherwise, if it is aligned, make sure the entire base
3450 	 * IPv6 header is in the first mbuf of the chain.
3451 	 */
3452 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3453 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3454 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3455 			    (max_linkhdr + 3) & ~3)) == NULL) {
3456 			/* XXXJRT new stat, please */
3457 			IP6STAT_INC(ip6s_toosmall);
3458 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3459 			goto bad;
3460 		}
3461 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3462 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3463 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3464 			IP6STAT_INC(ip6s_toosmall);
3465 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3466 			goto bad;
3467 		}
3468 	}
3469 
3470 	ip6 = mtod(m, struct ip6_hdr *);
3471 
3472 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3473 		IP6STAT_INC(ip6s_badvers);
3474 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3475 		goto bad;
3476 	}
3477 
3478 	/* Checks out, proceed */
3479 	*mp = m;
3480 	return (0);
3481 
3482 bad:
3483 	*mp = m;
3484 	return (-1);
3485 }
3486 #endif /* INET6 */
3487 
3488 /*
3489  * bridge_fragment:
3490  *
3491  *	Fragment mbuf chain in multiple packets and prepend ethernet header.
3492  */
3493 static int
3494 bridge_fragment(struct ifnet *ifp, struct mbuf **mp, struct ether_header *eh,
3495     int snap, struct llc *llc)
3496 {
3497 	struct mbuf *m = *mp, *nextpkt = NULL, *mprev = NULL, *mcur = NULL;
3498 	struct ip *ip;
3499 	int error = -1;
3500 
3501 	if (m->m_len < sizeof(struct ip) &&
3502 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3503 		goto dropit;
3504 	ip = mtod(m, struct ip *);
3505 
3506 	m->m_pkthdr.csum_flags |= CSUM_IP;
3507 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3508 	if (error)
3509 		goto dropit;
3510 
3511 	/*
3512 	 * Walk the chain and re-add the Ethernet header for
3513 	 * each mbuf packet.
3514 	 */
3515 	for (mcur = m; mcur; mcur = mcur->m_nextpkt) {
3516 		nextpkt = mcur->m_nextpkt;
3517 		mcur->m_nextpkt = NULL;
3518 		if (snap) {
3519 			M_PREPEND(mcur, sizeof(struct llc), M_NOWAIT);
3520 			if (mcur == NULL) {
3521 				error = ENOBUFS;
3522 				if (mprev != NULL)
3523 					mprev->m_nextpkt = nextpkt;
3524 				goto dropit;
3525 			}
3526 			bcopy(llc, mtod(mcur, caddr_t),sizeof(struct llc));
3527 		}
3528 
3529 		M_PREPEND(mcur, ETHER_HDR_LEN, M_NOWAIT);
3530 		if (mcur == NULL) {
3531 			error = ENOBUFS;
3532 			if (mprev != NULL)
3533 				mprev->m_nextpkt = nextpkt;
3534 			goto dropit;
3535 		}
3536 		bcopy(eh, mtod(mcur, caddr_t), ETHER_HDR_LEN);
3537 
3538 		/*
3539 		 * The previous two M_PREPEND could have inserted one or two
3540 		 * mbufs in front so we have to update the previous packet's
3541 		 * m_nextpkt.
3542 		 */
3543 		mcur->m_nextpkt = nextpkt;
3544 		if (mprev != NULL)
3545 			mprev->m_nextpkt = mcur;
3546 		else {
3547 			/* The first mbuf in the original chain needs to be
3548 			 * updated. */
3549 			*mp = mcur;
3550 		}
3551 		mprev = mcur;
3552 	}
3553 
3554 	KMOD_IPSTAT_INC(ips_fragmented);
3555 	return (error);
3556 
3557 dropit:
3558 	for (mcur = *mp; mcur; mcur = m) { /* droping the full packet chain */
3559 		m = mcur->m_nextpkt;
3560 		m_freem(mcur);
3561 	}
3562 	return (error);
3563 }
3564 
3565 static void
3566 bridge_linkstate(struct ifnet *ifp)
3567 {
3568 	struct bridge_softc *sc = ifp->if_bridge;
3569 	struct bridge_iflist *bif;
3570 
3571 	BRIDGE_LOCK(sc);
3572 	bif = bridge_lookup_member_if(sc, ifp);
3573 	if (bif == NULL) {
3574 		BRIDGE_UNLOCK(sc);
3575 		return;
3576 	}
3577 	bridge_linkcheck(sc);
3578 	BRIDGE_UNLOCK(sc);
3579 
3580 	bstp_linkstate(&bif->bif_stp);
3581 }
3582 
3583 static void
3584 bridge_linkcheck(struct bridge_softc *sc)
3585 {
3586 	struct bridge_iflist *bif;
3587 	int new_link, hasls;
3588 
3589 	BRIDGE_LOCK_ASSERT(sc);
3590 	new_link = LINK_STATE_DOWN;
3591 	hasls = 0;
3592 	/* Our link is considered up if at least one of our ports is active */
3593 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3594 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3595 			hasls++;
3596 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3597 			new_link = LINK_STATE_UP;
3598 			break;
3599 		}
3600 	}
3601 	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3602 		/* If no interfaces support link-state then we default to up */
3603 		new_link = LINK_STATE_UP;
3604 	}
3605 	if_link_state_change(sc->sc_ifp, new_link);
3606 }
3607