xref: /freebsd/sys/net/if_bridge.c (revision a98ff317388a00b992f1bf8404dee596f9383f5e)
1 /*	$NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $	*/
2 
3 /*
4  * Copyright 2001 Wasabi Systems, Inc.
5  * All rights reserved.
6  *
7  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed for the NetBSD Project by
20  *	Wasabi Systems, Inc.
21  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
22  *    or promote products derived from this software without specific prior
23  *    written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
27  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
28  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
29  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35  * POSSIBILITY OF SUCH DAMAGE.
36  */
37 
38 /*
39  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
40  * All rights reserved.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  *
51  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
52  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
53  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
54  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
55  * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
56  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
57  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
58  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
59  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
60  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
61  * POSSIBILITY OF SUCH DAMAGE.
62  *
63  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
64  */
65 
66 /*
67  * Network interface bridge support.
68  *
69  * TODO:
70  *
71  *	- Currently only supports Ethernet-like interfaces (Ethernet,
72  *	  802.11, VLANs on Ethernet, etc.)  Figure out a nice way
73  *	  to bridge other types of interfaces (FDDI-FDDI, and maybe
74  *	  consider heterogenous bridges).
75  */
76 
77 #include <sys/cdefs.h>
78 __FBSDID("$FreeBSD$");
79 
80 #include "opt_inet.h"
81 #include "opt_inet6.h"
82 
83 #include <sys/param.h>
84 #include <sys/mbuf.h>
85 #include <sys/malloc.h>
86 #include <sys/protosw.h>
87 #include <sys/systm.h>
88 #include <sys/jail.h>
89 #include <sys/time.h>
90 #include <sys/socket.h> /* for net/if.h */
91 #include <sys/sockio.h>
92 #include <sys/ctype.h>  /* string functions */
93 #include <sys/kernel.h>
94 #include <sys/random.h>
95 #include <sys/syslog.h>
96 #include <sys/sysctl.h>
97 #include <vm/uma.h>
98 #include <sys/module.h>
99 #include <sys/priv.h>
100 #include <sys/proc.h>
101 #include <sys/lock.h>
102 #include <sys/mutex.h>
103 
104 #include <net/bpf.h>
105 #include <net/if.h>
106 #include <net/if_clone.h>
107 #include <net/if_dl.h>
108 #include <net/if_types.h>
109 #include <net/if_var.h>
110 #include <net/pfil.h>
111 #include <net/vnet.h>
112 
113 #include <netinet/in.h> /* for struct arpcom */
114 #include <netinet/in_systm.h>
115 #include <netinet/in_var.h>
116 #include <netinet/ip.h>
117 #include <netinet/ip_var.h>
118 #ifdef INET6
119 #include <netinet/ip6.h>
120 #include <netinet6/ip6_var.h>
121 #include <netinet6/in6_ifattach.h>
122 #endif
123 #if defined(INET) || defined(INET6)
124 #include <netinet/ip_carp.h>
125 #endif
126 #include <machine/in_cksum.h>
127 #include <netinet/if_ether.h> /* for struct arpcom */
128 #include <net/bridgestp.h>
129 #include <net/if_bridgevar.h>
130 #include <net/if_llc.h>
131 #include <net/if_vlan_var.h>
132 
133 #include <net/route.h>
134 
135 /*
136  * Size of the route hash table.  Must be a power of two.
137  */
138 #ifndef BRIDGE_RTHASH_SIZE
139 #define	BRIDGE_RTHASH_SIZE		1024
140 #endif
141 
142 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
143 
144 /*
145  * Default maximum number of addresses to cache.
146  */
147 #ifndef BRIDGE_RTABLE_MAX
148 #define	BRIDGE_RTABLE_MAX		2000
149 #endif
150 
151 /*
152  * Timeout (in seconds) for entries learned dynamically.
153  */
154 #ifndef BRIDGE_RTABLE_TIMEOUT
155 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
156 #endif
157 
158 /*
159  * Number of seconds between walks of the route list.
160  */
161 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
162 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
163 #endif
164 
165 /*
166  * List of capabilities to possibly mask on the member interface.
167  */
168 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
169 
170 /*
171  * List of capabilities to strip
172  */
173 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
174 
175 /*
176  * Bridge interface list entry.
177  */
178 struct bridge_iflist {
179 	LIST_ENTRY(bridge_iflist) bif_next;
180 	struct ifnet		*bif_ifp;	/* member if */
181 	struct bstp_port	bif_stp;	/* STP state */
182 	uint32_t		bif_flags;	/* member if flags */
183 	int			bif_savedcaps;	/* saved capabilities */
184 	uint32_t		bif_addrmax;	/* max # of addresses */
185 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
186 	uint32_t		bif_addrexceeded;/* # of address violations */
187 };
188 
189 /*
190  * Bridge route node.
191  */
192 struct bridge_rtnode {
193 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
194 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
195 	struct bridge_iflist	*brt_dst;	/* destination if */
196 	unsigned long		brt_expire;	/* expiration time */
197 	uint8_t			brt_flags;	/* address flags */
198 	uint8_t			brt_addr[ETHER_ADDR_LEN];
199 	uint16_t		brt_vlan;	/* vlan id */
200 };
201 #define	brt_ifp			brt_dst->bif_ifp
202 
203 /*
204  * Software state for each bridge.
205  */
206 struct bridge_softc {
207 	struct ifnet		*sc_ifp;	/* make this an interface */
208 	LIST_ENTRY(bridge_softc) sc_list;
209 	struct mtx		sc_mtx;
210 	struct cv		sc_cv;
211 	uint32_t		sc_brtmax;	/* max # of addresses */
212 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
213 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
214 	struct callout		sc_brcallout;	/* bridge callout */
215 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
216 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
217 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
218 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
219 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
220 	uint32_t		sc_rthash_key;	/* key for hash */
221 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
222 	struct bstp_state	sc_stp;		/* STP state */
223 	uint32_t		sc_brtexceeded;	/* # of cache drops */
224 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
225 	u_char			sc_defaddr[6];	/* Default MAC address */
226 };
227 
228 static struct mtx 	bridge_list_mtx;
229 eventhandler_tag	bridge_detach_cookie = NULL;
230 
231 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
232 
233 uma_zone_t bridge_rtnode_zone;
234 
235 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
236 static void	bridge_clone_destroy(struct ifnet *);
237 
238 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
239 static void	bridge_mutecaps(struct bridge_softc *);
240 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
241 		    int);
242 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
243 static void	bridge_init(void *);
244 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
245 static void	bridge_stop(struct ifnet *, int);
246 static int	bridge_transmit(struct ifnet *, struct mbuf *);
247 static void	bridge_qflush(struct ifnet *);
248 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
249 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
250 		    struct rtentry *);
251 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
252 		    struct mbuf *);
253 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
254 
255 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
256 		    struct mbuf *m);
257 
258 static void	bridge_timer(void *);
259 
260 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
261 		    struct mbuf *, int);
262 static void	bridge_span(struct bridge_softc *, struct mbuf *);
263 
264 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
265 		    uint16_t, struct bridge_iflist *, int, uint8_t);
266 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
267 		    uint16_t);
268 static void	bridge_rttrim(struct bridge_softc *);
269 static void	bridge_rtage(struct bridge_softc *);
270 static void	bridge_rtflush(struct bridge_softc *, int);
271 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
272 		    uint16_t);
273 
274 static void	bridge_rtable_init(struct bridge_softc *);
275 static void	bridge_rtable_fini(struct bridge_softc *);
276 
277 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
278 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
279 		    const uint8_t *, uint16_t);
280 static int	bridge_rtnode_insert(struct bridge_softc *,
281 		    struct bridge_rtnode *);
282 static void	bridge_rtnode_destroy(struct bridge_softc *,
283 		    struct bridge_rtnode *);
284 static void	bridge_rtable_expire(struct ifnet *, int);
285 static void	bridge_state_change(struct ifnet *, int);
286 
287 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
288 		    const char *name);
289 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
290 		    struct ifnet *ifp);
291 static void	bridge_delete_member(struct bridge_softc *,
292 		    struct bridge_iflist *, int);
293 static void	bridge_delete_span(struct bridge_softc *,
294 		    struct bridge_iflist *);
295 
296 static int	bridge_ioctl_add(struct bridge_softc *, void *);
297 static int	bridge_ioctl_del(struct bridge_softc *, void *);
298 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
299 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
300 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
301 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
302 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
303 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
304 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
305 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
306 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
307 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
308 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
309 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
310 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
311 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
312 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
313 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
314 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
315 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
316 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
317 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
318 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
319 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
320 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
321 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
322 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
323 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
324 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
325 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
326 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
327 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
328 		    int);
329 static int	bridge_ip_checkbasic(struct mbuf **mp);
330 #ifdef INET6
331 static int	bridge_ip6_checkbasic(struct mbuf **mp);
332 #endif /* INET6 */
333 static int	bridge_fragment(struct ifnet *, struct mbuf *,
334 		    struct ether_header *, int, struct llc *);
335 static void	bridge_linkstate(struct ifnet *ifp);
336 static void	bridge_linkcheck(struct bridge_softc *sc);
337 
338 extern void (*bridge_linkstate_p)(struct ifnet *ifp);
339 
340 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
341 #define	VLANTAGOF(_m)	\
342     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
343 
344 static struct bstp_cb_ops bridge_ops = {
345 	.bcb_state = bridge_state_change,
346 	.bcb_rtage = bridge_rtable_expire
347 };
348 
349 SYSCTL_DECL(_net_link);
350 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
351 
352 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
353 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
354 static int pfil_member = 1; /* run pfil hooks on the member interface */
355 static int pfil_ipfw = 0;   /* layer2 filter with ipfw */
356 static int pfil_ipfw_arp = 0;   /* layer2 filter with ipfw */
357 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
358                                    locally destined packets */
359 static int log_stp   = 0;   /* log STP state changes */
360 static int bridge_inherit_mac = 0;   /* share MAC with first bridge member */
361 TUNABLE_INT("net.link.bridge.pfil_onlyip", &pfil_onlyip);
362 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
363     &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
364 TUNABLE_INT("net.link.bridge.ipfw_arp", &pfil_ipfw_arp);
365 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
366     &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
367 TUNABLE_INT("net.link.bridge.pfil_bridge", &pfil_bridge);
368 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
369     &pfil_bridge, 0, "Packet filter on the bridge interface");
370 TUNABLE_INT("net.link.bridge.pfil_member", &pfil_member);
371 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
372     &pfil_member, 0, "Packet filter on the member interface");
373 TUNABLE_INT("net.link.bridge.pfil_local_phys", &pfil_local_phys);
374 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
375     &pfil_local_phys, 0,
376     "Packet filter on the physical interface for locally destined packets");
377 TUNABLE_INT("net.link.bridge.log_stp", &log_stp);
378 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
379     &log_stp, 0, "Log STP state changes");
380 TUNABLE_INT("net.link.bridge.inherit_mac", &bridge_inherit_mac);
381 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
382     &bridge_inherit_mac, 0,
383     "Inherit MAC address from the first bridge member");
384 
385 struct bridge_control {
386 	int	(*bc_func)(struct bridge_softc *, void *);
387 	int	bc_argsize;
388 	int	bc_flags;
389 };
390 
391 #define	BC_F_COPYIN		0x01	/* copy arguments in */
392 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
393 #define	BC_F_SUSER		0x04	/* do super-user check */
394 
395 const struct bridge_control bridge_control_table[] = {
396 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
397 	  BC_F_COPYIN|BC_F_SUSER },
398 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
399 	  BC_F_COPYIN|BC_F_SUSER },
400 
401 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
402 	  BC_F_COPYIN|BC_F_COPYOUT },
403 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
404 	  BC_F_COPYIN|BC_F_SUSER },
405 
406 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
407 	  BC_F_COPYIN|BC_F_SUSER },
408 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
409 	  BC_F_COPYOUT },
410 
411 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
412 	  BC_F_COPYIN|BC_F_COPYOUT },
413 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
414 	  BC_F_COPYIN|BC_F_COPYOUT },
415 
416 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
417 	  BC_F_COPYIN|BC_F_SUSER },
418 
419 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
420 	  BC_F_COPYIN|BC_F_SUSER },
421 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
422 	  BC_F_COPYOUT },
423 
424 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
425 	  BC_F_COPYIN|BC_F_SUSER },
426 
427 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
428 	  BC_F_COPYIN|BC_F_SUSER },
429 
430 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
431 	  BC_F_COPYOUT },
432 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
433 	  BC_F_COPYIN|BC_F_SUSER },
434 
435 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
436 	  BC_F_COPYOUT },
437 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
438 	  BC_F_COPYIN|BC_F_SUSER },
439 
440 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
441 	  BC_F_COPYOUT },
442 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
443 	  BC_F_COPYIN|BC_F_SUSER },
444 
445 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
446 	  BC_F_COPYOUT },
447 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
448 	  BC_F_COPYIN|BC_F_SUSER },
449 
450 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
451 	  BC_F_COPYIN|BC_F_SUSER },
452 
453 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
454 	  BC_F_COPYIN|BC_F_SUSER },
455 
456 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
457 	  BC_F_COPYIN|BC_F_SUSER },
458 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
459 	  BC_F_COPYIN|BC_F_SUSER },
460 
461 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
462 	  BC_F_COPYOUT },
463 
464 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
465 	  BC_F_COPYOUT },
466 
467 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
468 	  BC_F_COPYIN|BC_F_COPYOUT },
469 
470 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
471 	  BC_F_COPYIN|BC_F_SUSER },
472 
473 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
474 	  BC_F_COPYIN|BC_F_SUSER },
475 
476 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
477 	  BC_F_COPYIN|BC_F_SUSER },
478 
479 };
480 const int bridge_control_table_size =
481     sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
482 
483 LIST_HEAD(, bridge_softc) bridge_list;
484 
485 static struct if_clone *bridge_cloner;
486 static const char bridge_name[] = "bridge";
487 
488 static int
489 bridge_modevent(module_t mod, int type, void *data)
490 {
491 
492 	switch (type) {
493 	case MOD_LOAD:
494 		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
495 		bridge_cloner = if_clone_simple(bridge_name,
496 		    bridge_clone_create, bridge_clone_destroy, 0);
497 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
498 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
499 		    UMA_ALIGN_PTR, 0);
500 		LIST_INIT(&bridge_list);
501 		bridge_input_p = bridge_input;
502 		bridge_output_p = bridge_output;
503 		bridge_dn_p = bridge_dummynet;
504 		bridge_linkstate_p = bridge_linkstate;
505 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
506 		    ifnet_departure_event, bridge_ifdetach, NULL,
507 		    EVENTHANDLER_PRI_ANY);
508 		break;
509 	case MOD_UNLOAD:
510 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
511 		    bridge_detach_cookie);
512 		if_clone_detach(bridge_cloner);
513 		uma_zdestroy(bridge_rtnode_zone);
514 		bridge_input_p = NULL;
515 		bridge_output_p = NULL;
516 		bridge_dn_p = NULL;
517 		bridge_linkstate_p = NULL;
518 		mtx_destroy(&bridge_list_mtx);
519 		break;
520 	default:
521 		return (EOPNOTSUPP);
522 	}
523 	return (0);
524 }
525 
526 static moduledata_t bridge_mod = {
527 	"if_bridge",
528 	bridge_modevent,
529 	0
530 };
531 
532 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
533 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
534 
535 /*
536  * handler for net.link.bridge.ipfw
537  */
538 static int
539 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
540 {
541 	int enable = pfil_ipfw;
542 	int error;
543 
544 	error = sysctl_handle_int(oidp, &enable, 0, req);
545 	enable = (enable) ? 1 : 0;
546 
547 	if (enable != pfil_ipfw) {
548 		pfil_ipfw = enable;
549 
550 		/*
551 		 * Disable pfil so that ipfw doesnt run twice, if the user
552 		 * really wants both then they can re-enable pfil_bridge and/or
553 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
554 		 * layer2 type.
555 		 */
556 		if (pfil_ipfw) {
557 			pfil_onlyip = 0;
558 			pfil_bridge = 0;
559 			pfil_member = 0;
560 		}
561 	}
562 
563 	return (error);
564 }
565 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
566 	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
567 
568 /*
569  * bridge_clone_create:
570  *
571  *	Create a new bridge instance.
572  */
573 static int
574 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
575 {
576 	struct bridge_softc *sc, *sc2;
577 	struct ifnet *bifp, *ifp;
578 	int fb, retry;
579 	unsigned long hostid;
580 
581 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
582 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
583 	if (ifp == NULL) {
584 		free(sc, M_DEVBUF);
585 		return (ENOSPC);
586 	}
587 
588 	BRIDGE_LOCK_INIT(sc);
589 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
590 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
591 
592 	/* Initialize our routing table. */
593 	bridge_rtable_init(sc);
594 
595 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
596 
597 	LIST_INIT(&sc->sc_iflist);
598 	LIST_INIT(&sc->sc_spanlist);
599 
600 	ifp->if_softc = sc;
601 	if_initname(ifp, bridge_name, unit);
602 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
603 	ifp->if_ioctl = bridge_ioctl;
604 	ifp->if_transmit = bridge_transmit;
605 	ifp->if_qflush = bridge_qflush;
606 	ifp->if_init = bridge_init;
607 	ifp->if_type = IFT_BRIDGE;
608 
609 	/*
610 	 * Generate an ethernet address with a locally administered address.
611 	 *
612 	 * Since we are using random ethernet addresses for the bridge, it is
613 	 * possible that we might have address collisions, so make sure that
614 	 * this hardware address isn't already in use on another bridge.
615 	 * The first try uses the hostid and falls back to arc4rand().
616 	 */
617 	fb = 0;
618 	getcredhostid(curthread->td_ucred, &hostid);
619 	do {
620 		if (fb || hostid == 0) {
621 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
622 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
623 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
624 		} else {
625 			sc->sc_defaddr[0] = 0x2;
626 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
627 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
628 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
629 			sc->sc_defaddr[4] =  hostid        & 0xff;
630 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
631 		}
632 
633 		fb = 1;
634 		retry = 0;
635 		mtx_lock(&bridge_list_mtx);
636 		LIST_FOREACH(sc2, &bridge_list, sc_list) {
637 			bifp = sc2->sc_ifp;
638 			if (memcmp(sc->sc_defaddr,
639 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
640 				retry = 1;
641 				break;
642 			}
643 		}
644 		mtx_unlock(&bridge_list_mtx);
645 	} while (retry == 1);
646 
647 	bstp_attach(&sc->sc_stp, &bridge_ops);
648 	ether_ifattach(ifp, sc->sc_defaddr);
649 	/* Now undo some of the damage... */
650 	ifp->if_baudrate = 0;
651 	ifp->if_type = IFT_BRIDGE;
652 
653 	mtx_lock(&bridge_list_mtx);
654 	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
655 	mtx_unlock(&bridge_list_mtx);
656 
657 	return (0);
658 }
659 
660 /*
661  * bridge_clone_destroy:
662  *
663  *	Destroy a bridge instance.
664  */
665 static void
666 bridge_clone_destroy(struct ifnet *ifp)
667 {
668 	struct bridge_softc *sc = ifp->if_softc;
669 	struct bridge_iflist *bif;
670 
671 	BRIDGE_LOCK(sc);
672 
673 	bridge_stop(ifp, 1);
674 	ifp->if_flags &= ~IFF_UP;
675 
676 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
677 		bridge_delete_member(sc, bif, 0);
678 
679 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
680 		bridge_delete_span(sc, bif);
681 	}
682 
683 	BRIDGE_UNLOCK(sc);
684 
685 	callout_drain(&sc->sc_brcallout);
686 
687 	mtx_lock(&bridge_list_mtx);
688 	LIST_REMOVE(sc, sc_list);
689 	mtx_unlock(&bridge_list_mtx);
690 
691 	bstp_detach(&sc->sc_stp);
692 	ether_ifdetach(ifp);
693 	if_free(ifp);
694 
695 	/* Tear down the routing table. */
696 	bridge_rtable_fini(sc);
697 
698 	BRIDGE_LOCK_DESTROY(sc);
699 	free(sc, M_DEVBUF);
700 }
701 
702 /*
703  * bridge_ioctl:
704  *
705  *	Handle a control request from the operator.
706  */
707 static int
708 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
709 {
710 	struct bridge_softc *sc = ifp->if_softc;
711 	struct ifreq *ifr = (struct ifreq *)data;
712 	struct bridge_iflist *bif;
713 	struct thread *td = curthread;
714 	union {
715 		struct ifbreq ifbreq;
716 		struct ifbifconf ifbifconf;
717 		struct ifbareq ifbareq;
718 		struct ifbaconf ifbaconf;
719 		struct ifbrparam ifbrparam;
720 		struct ifbropreq ifbropreq;
721 	} args;
722 	struct ifdrv *ifd = (struct ifdrv *) data;
723 	const struct bridge_control *bc;
724 	int error = 0;
725 
726 	switch (cmd) {
727 
728 	case SIOCADDMULTI:
729 	case SIOCDELMULTI:
730 		break;
731 
732 	case SIOCGDRVSPEC:
733 	case SIOCSDRVSPEC:
734 		if (ifd->ifd_cmd >= bridge_control_table_size) {
735 			error = EINVAL;
736 			break;
737 		}
738 		bc = &bridge_control_table[ifd->ifd_cmd];
739 
740 		if (cmd == SIOCGDRVSPEC &&
741 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
742 			error = EINVAL;
743 			break;
744 		}
745 		else if (cmd == SIOCSDRVSPEC &&
746 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
747 			error = EINVAL;
748 			break;
749 		}
750 
751 		if (bc->bc_flags & BC_F_SUSER) {
752 			error = priv_check(td, PRIV_NET_BRIDGE);
753 			if (error)
754 				break;
755 		}
756 
757 		if (ifd->ifd_len != bc->bc_argsize ||
758 		    ifd->ifd_len > sizeof(args)) {
759 			error = EINVAL;
760 			break;
761 		}
762 
763 		bzero(&args, sizeof(args));
764 		if (bc->bc_flags & BC_F_COPYIN) {
765 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
766 			if (error)
767 				break;
768 		}
769 
770 		BRIDGE_LOCK(sc);
771 		error = (*bc->bc_func)(sc, &args);
772 		BRIDGE_UNLOCK(sc);
773 		if (error)
774 			break;
775 
776 		if (bc->bc_flags & BC_F_COPYOUT)
777 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
778 
779 		break;
780 
781 	case SIOCSIFFLAGS:
782 		if (!(ifp->if_flags & IFF_UP) &&
783 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
784 			/*
785 			 * If interface is marked down and it is running,
786 			 * then stop and disable it.
787 			 */
788 			BRIDGE_LOCK(sc);
789 			bridge_stop(ifp, 1);
790 			BRIDGE_UNLOCK(sc);
791 		} else if ((ifp->if_flags & IFF_UP) &&
792 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
793 			/*
794 			 * If interface is marked up and it is stopped, then
795 			 * start it.
796 			 */
797 			(*ifp->if_init)(sc);
798 		}
799 		break;
800 
801 	case SIOCSIFMTU:
802 		if (ifr->ifr_mtu < 576) {
803 			error = EINVAL;
804 			break;
805 		}
806 		if (LIST_EMPTY(&sc->sc_iflist)) {
807 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
808 			break;
809 		}
810 		BRIDGE_LOCK(sc);
811 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
812 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
813 				log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
814 				    " != %d\n", sc->sc_ifp->if_xname,
815 				    bif->bif_ifp->if_mtu,
816 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
817 				error = EINVAL;
818 				break;
819 			}
820 		}
821 		if (!error)
822 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
823 		BRIDGE_UNLOCK(sc);
824 		break;
825 	default:
826 		/*
827 		 * drop the lock as ether_ioctl() will call bridge_start() and
828 		 * cause the lock to be recursed.
829 		 */
830 		error = ether_ioctl(ifp, cmd, data);
831 		break;
832 	}
833 
834 	return (error);
835 }
836 
837 /*
838  * bridge_mutecaps:
839  *
840  *	Clear or restore unwanted capabilities on the member interface
841  */
842 static void
843 bridge_mutecaps(struct bridge_softc *sc)
844 {
845 	struct bridge_iflist *bif;
846 	int enabled, mask;
847 
848 	/* Initial bitmask of capabilities to test */
849 	mask = BRIDGE_IFCAPS_MASK;
850 
851 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
852 		/* Every member must support it or its disabled */
853 		mask &= bif->bif_savedcaps;
854 	}
855 
856 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
857 		enabled = bif->bif_ifp->if_capenable;
858 		enabled &= ~BRIDGE_IFCAPS_STRIP;
859 		/* strip off mask bits and enable them again if allowed */
860 		enabled &= ~BRIDGE_IFCAPS_MASK;
861 		enabled |= mask;
862 		bridge_set_ifcap(sc, bif, enabled);
863 	}
864 
865 }
866 
867 static void
868 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
869 {
870 	struct ifnet *ifp = bif->bif_ifp;
871 	struct ifreq ifr;
872 	int error;
873 
874 	bzero(&ifr, sizeof(ifr));
875 	ifr.ifr_reqcap = set;
876 
877 	if (ifp->if_capenable != set) {
878 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
879 		if (error)
880 			if_printf(sc->sc_ifp,
881 			    "error setting interface capabilities on %s\n",
882 			    ifp->if_xname);
883 	}
884 }
885 
886 /*
887  * bridge_lookup_member:
888  *
889  *	Lookup a bridge member interface.
890  */
891 static struct bridge_iflist *
892 bridge_lookup_member(struct bridge_softc *sc, const char *name)
893 {
894 	struct bridge_iflist *bif;
895 	struct ifnet *ifp;
896 
897 	BRIDGE_LOCK_ASSERT(sc);
898 
899 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
900 		ifp = bif->bif_ifp;
901 		if (strcmp(ifp->if_xname, name) == 0)
902 			return (bif);
903 	}
904 
905 	return (NULL);
906 }
907 
908 /*
909  * bridge_lookup_member_if:
910  *
911  *	Lookup a bridge member interface by ifnet*.
912  */
913 static struct bridge_iflist *
914 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
915 {
916 	struct bridge_iflist *bif;
917 
918 	BRIDGE_LOCK_ASSERT(sc);
919 
920 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
921 		if (bif->bif_ifp == member_ifp)
922 			return (bif);
923 	}
924 
925 	return (NULL);
926 }
927 
928 /*
929  * bridge_delete_member:
930  *
931  *	Delete the specified member interface.
932  */
933 static void
934 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
935     int gone)
936 {
937 	struct ifnet *ifs = bif->bif_ifp;
938 	struct ifnet *fif = NULL;
939 
940 	BRIDGE_LOCK_ASSERT(sc);
941 
942 	if (bif->bif_flags & IFBIF_STP)
943 		bstp_disable(&bif->bif_stp);
944 
945 	ifs->if_bridge = NULL;
946 	BRIDGE_XLOCK(sc);
947 	LIST_REMOVE(bif, bif_next);
948 	BRIDGE_XDROP(sc);
949 
950 	/*
951 	 * If removing the interface that gave the bridge its mac address, set
952 	 * the mac address of the bridge to the address of the next member, or
953 	 * to its default address if no members are left.
954 	 */
955 	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
956 		if (LIST_EMPTY(&sc->sc_iflist)) {
957 			bcopy(sc->sc_defaddr,
958 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
959 			sc->sc_ifaddr = NULL;
960 		} else {
961 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
962 			bcopy(IF_LLADDR(fif),
963 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
964 			sc->sc_ifaddr = fif;
965 		}
966 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
967 	}
968 
969 	bridge_linkcheck(sc);
970 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
971 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
972 	KASSERT(bif->bif_addrcnt == 0,
973 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
974 
975 	BRIDGE_UNLOCK(sc);
976 	if (!gone) {
977 		switch (ifs->if_type) {
978 		case IFT_ETHER:
979 		case IFT_L2VLAN:
980 			/*
981 			 * Take the interface out of promiscuous mode.
982 			 */
983 			(void) ifpromisc(ifs, 0);
984 			break;
985 
986 		case IFT_GIF:
987 			break;
988 
989 		default:
990 #ifdef DIAGNOSTIC
991 			panic("bridge_delete_member: impossible");
992 #endif
993 			break;
994 		}
995 		/* reneable any interface capabilities */
996 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
997 	}
998 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
999 	BRIDGE_LOCK(sc);
1000 	free(bif, M_DEVBUF);
1001 }
1002 
1003 /*
1004  * bridge_delete_span:
1005  *
1006  *	Delete the specified span interface.
1007  */
1008 static void
1009 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1010 {
1011 	BRIDGE_LOCK_ASSERT(sc);
1012 
1013 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1014 	    ("%s: not a span interface", __func__));
1015 
1016 	LIST_REMOVE(bif, bif_next);
1017 	free(bif, M_DEVBUF);
1018 }
1019 
1020 static int
1021 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1022 {
1023 	struct ifbreq *req = arg;
1024 	struct bridge_iflist *bif = NULL;
1025 	struct ifnet *ifs;
1026 	int error = 0;
1027 
1028 	ifs = ifunit(req->ifbr_ifsname);
1029 	if (ifs == NULL)
1030 		return (ENOENT);
1031 	if (ifs->if_ioctl == NULL)	/* must be supported */
1032 		return (EINVAL);
1033 
1034 	/* If it's in the span list, it can't be a member. */
1035 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1036 		if (ifs == bif->bif_ifp)
1037 			return (EBUSY);
1038 
1039 	if (ifs->if_bridge == sc)
1040 		return (EEXIST);
1041 
1042 	if (ifs->if_bridge != NULL)
1043 		return (EBUSY);
1044 
1045 	switch (ifs->if_type) {
1046 	case IFT_ETHER:
1047 	case IFT_L2VLAN:
1048 	case IFT_GIF:
1049 		/* permitted interface types */
1050 		break;
1051 	default:
1052 		return (EINVAL);
1053 	}
1054 
1055 #ifdef INET6
1056 	/*
1057 	 * Two valid inet6 addresses with link-local scope must not be
1058 	 * on the parent interface and the member interfaces at the
1059 	 * same time.  This restriction is needed to prevent violation
1060 	 * of link-local scope zone.  Attempts to add a member
1061 	 * interface which has inet6 addresses when the parent has
1062 	 * inet6 triggers removal of all inet6 addresses on the member
1063 	 * interface.
1064 	 */
1065 
1066 	/* Check if the parent interface has a link-local scope addr. */
1067 	if (in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1068 		/*
1069 		 * If any, remove all inet6 addresses from the member
1070 		 * interfaces.
1071 		 */
1072 		BRIDGE_XLOCK(sc);
1073 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1074  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1075 				BRIDGE_UNLOCK(sc);
1076 				in6_ifdetach(bif->bif_ifp);
1077 				BRIDGE_LOCK(sc);
1078 				if_printf(sc->sc_ifp,
1079 				    "IPv6 addresses on %s have been removed "
1080 				    "before adding it as a member to prevent "
1081 				    "IPv6 address scope violation.\n",
1082 				    bif->bif_ifp->if_xname);
1083 			}
1084 		}
1085 		BRIDGE_XDROP(sc);
1086 		if (in6ifa_llaonifp(ifs)) {
1087 			BRIDGE_UNLOCK(sc);
1088 			in6_ifdetach(ifs);
1089 			BRIDGE_LOCK(sc);
1090 			if_printf(sc->sc_ifp,
1091 			    "IPv6 addresses on %s have been removed "
1092 			    "before adding it as a member to prevent "
1093 			    "IPv6 address scope violation.\n",
1094 			    ifs->if_xname);
1095 		}
1096 	} else {
1097 		struct in6_ifaddr *ia6_m, *ia6_s;
1098 		/*
1099 		 * If not, check whether one of the existing member
1100 		 * interfaces have inet6 address.  If any, remove
1101 		 * inet6 addresses on the interface to be added.
1102 		 */
1103 		ia6_m = NULL;
1104 		BRIDGE_XLOCK(sc);
1105 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1106 			ia6_m = in6ifa_llaonifp(bif->bif_ifp);
1107 			if (ia6_m != NULL)
1108 				break;
1109 		}
1110 		BRIDGE_XDROP(sc);
1111 		ia6_s = in6ifa_llaonifp(ifs);
1112 
1113 		if (ia6_m != NULL && ia6_s != NULL) {
1114 			BRIDGE_UNLOCK(sc);
1115 			in6_ifdetach(ifs);
1116 			BRIDGE_LOCK(sc);
1117 			if_printf(sc->sc_ifp, "IPv6 addresses on %s have "
1118 				  "been removed before adding it as a member "
1119 				  "to prevent IPv6 address scope violation.\n",
1120 				  ifs->if_xname);
1121 		}
1122 	}
1123 #endif
1124 	/* Allow the first Ethernet member to define the MTU */
1125 	if (LIST_EMPTY(&sc->sc_iflist))
1126 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1127 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1128 		if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1129 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1130 		return (EINVAL);
1131 	}
1132 
1133 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1134 	if (bif == NULL)
1135 		return (ENOMEM);
1136 
1137 	bif->bif_ifp = ifs;
1138 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1139 	bif->bif_savedcaps = ifs->if_capenable;
1140 
1141 	/*
1142 	 * Assign the interface's MAC address to the bridge if it's the first
1143 	 * member and the MAC address of the bridge has not been changed from
1144 	 * the default randomly generated one.
1145 	 */
1146 	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1147 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1148 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1149 		sc->sc_ifaddr = ifs;
1150 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1151 	}
1152 
1153 	ifs->if_bridge = sc;
1154 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1155 	/*
1156 	 * XXX: XLOCK HERE!?!
1157 	 *
1158 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1159 	 */
1160 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1161 
1162 	/* Set interface capabilities to the intersection set of all members */
1163 	bridge_mutecaps(sc);
1164 	bridge_linkcheck(sc);
1165 
1166 	/* Place the interface into promiscuous mode */
1167 	switch (ifs->if_type) {
1168 		case IFT_ETHER:
1169 		case IFT_L2VLAN:
1170 			BRIDGE_UNLOCK(sc);
1171 			error = ifpromisc(ifs, 1);
1172 			BRIDGE_LOCK(sc);
1173 			break;
1174 	}
1175 
1176 	if (error) {
1177 		bridge_delete_member(sc, bif, 0);
1178 		free(bif, M_DEVBUF);
1179 	}
1180 	return (error);
1181 }
1182 
1183 static int
1184 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1185 {
1186 	struct ifbreq *req = arg;
1187 	struct bridge_iflist *bif;
1188 
1189 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1190 	if (bif == NULL)
1191 		return (ENOENT);
1192 
1193 	bridge_delete_member(sc, bif, 0);
1194 
1195 	return (0);
1196 }
1197 
1198 static int
1199 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1200 {
1201 	struct ifbreq *req = arg;
1202 	struct bridge_iflist *bif;
1203 	struct bstp_port *bp;
1204 
1205 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1206 	if (bif == NULL)
1207 		return (ENOENT);
1208 
1209 	bp = &bif->bif_stp;
1210 	req->ifbr_ifsflags = bif->bif_flags;
1211 	req->ifbr_state = bp->bp_state;
1212 	req->ifbr_priority = bp->bp_priority;
1213 	req->ifbr_path_cost = bp->bp_path_cost;
1214 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1215 	req->ifbr_proto = bp->bp_protover;
1216 	req->ifbr_role = bp->bp_role;
1217 	req->ifbr_stpflags = bp->bp_flags;
1218 	req->ifbr_addrcnt = bif->bif_addrcnt;
1219 	req->ifbr_addrmax = bif->bif_addrmax;
1220 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1221 
1222 	/* Copy STP state options as flags */
1223 	if (bp->bp_operedge)
1224 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1225 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1226 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1227 	if (bp->bp_ptp_link)
1228 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1229 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1230 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1231 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1232 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1233 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1234 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1235 	return (0);
1236 }
1237 
1238 static int
1239 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1240 {
1241 	struct ifbreq *req = arg;
1242 	struct bridge_iflist *bif;
1243 	struct bstp_port *bp;
1244 	int error;
1245 
1246 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1247 	if (bif == NULL)
1248 		return (ENOENT);
1249 	bp = &bif->bif_stp;
1250 
1251 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1252 		/* SPAN is readonly */
1253 		return (EINVAL);
1254 
1255 	if (req->ifbr_ifsflags & IFBIF_STP) {
1256 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1257 			error = bstp_enable(&bif->bif_stp);
1258 			if (error)
1259 				return (error);
1260 		}
1261 	} else {
1262 		if ((bif->bif_flags & IFBIF_STP) != 0)
1263 			bstp_disable(&bif->bif_stp);
1264 	}
1265 
1266 	/* Pass on STP flags */
1267 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1268 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1269 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1270 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1271 
1272 	/* Save the bits relating to the bridge */
1273 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1274 
1275 	return (0);
1276 }
1277 
1278 static int
1279 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1280 {
1281 	struct ifbrparam *param = arg;
1282 
1283 	sc->sc_brtmax = param->ifbrp_csize;
1284 	bridge_rttrim(sc);
1285 
1286 	return (0);
1287 }
1288 
1289 static int
1290 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1291 {
1292 	struct ifbrparam *param = arg;
1293 
1294 	param->ifbrp_csize = sc->sc_brtmax;
1295 
1296 	return (0);
1297 }
1298 
1299 static int
1300 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1301 {
1302 	struct ifbifconf *bifc = arg;
1303 	struct bridge_iflist *bif;
1304 	struct ifbreq breq;
1305 	char *buf, *outbuf;
1306 	int count, buflen, len, error = 0;
1307 
1308 	count = 0;
1309 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1310 		count++;
1311 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1312 		count++;
1313 
1314 	buflen = sizeof(breq) * count;
1315 	if (bifc->ifbic_len == 0) {
1316 		bifc->ifbic_len = buflen;
1317 		return (0);
1318 	}
1319 	BRIDGE_UNLOCK(sc);
1320 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1321 	BRIDGE_LOCK(sc);
1322 
1323 	count = 0;
1324 	buf = outbuf;
1325 	len = min(bifc->ifbic_len, buflen);
1326 	bzero(&breq, sizeof(breq));
1327 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1328 		if (len < sizeof(breq))
1329 			break;
1330 
1331 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1332 		    sizeof(breq.ifbr_ifsname));
1333 		/* Fill in the ifbreq structure */
1334 		error = bridge_ioctl_gifflags(sc, &breq);
1335 		if (error)
1336 			break;
1337 		memcpy(buf, &breq, sizeof(breq));
1338 		count++;
1339 		buf += sizeof(breq);
1340 		len -= sizeof(breq);
1341 	}
1342 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1343 		if (len < sizeof(breq))
1344 			break;
1345 
1346 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1347 		    sizeof(breq.ifbr_ifsname));
1348 		breq.ifbr_ifsflags = bif->bif_flags;
1349 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1350 		memcpy(buf, &breq, sizeof(breq));
1351 		count++;
1352 		buf += sizeof(breq);
1353 		len -= sizeof(breq);
1354 	}
1355 
1356 	BRIDGE_UNLOCK(sc);
1357 	bifc->ifbic_len = sizeof(breq) * count;
1358 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1359 	BRIDGE_LOCK(sc);
1360 	free(outbuf, M_TEMP);
1361 	return (error);
1362 }
1363 
1364 static int
1365 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1366 {
1367 	struct ifbaconf *bac = arg;
1368 	struct bridge_rtnode *brt;
1369 	struct ifbareq bareq;
1370 	char *buf, *outbuf;
1371 	int count, buflen, len, error = 0;
1372 
1373 	if (bac->ifbac_len == 0)
1374 		return (0);
1375 
1376 	count = 0;
1377 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1378 		count++;
1379 	buflen = sizeof(bareq) * count;
1380 
1381 	BRIDGE_UNLOCK(sc);
1382 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1383 	BRIDGE_LOCK(sc);
1384 
1385 	count = 0;
1386 	buf = outbuf;
1387 	len = min(bac->ifbac_len, buflen);
1388 	bzero(&bareq, sizeof(bareq));
1389 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1390 		if (len < sizeof(bareq))
1391 			goto out;
1392 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1393 		    sizeof(bareq.ifba_ifsname));
1394 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1395 		bareq.ifba_vlan = brt->brt_vlan;
1396 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1397 				time_uptime < brt->brt_expire)
1398 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1399 		else
1400 			bareq.ifba_expire = 0;
1401 		bareq.ifba_flags = brt->brt_flags;
1402 
1403 		memcpy(buf, &bareq, sizeof(bareq));
1404 		count++;
1405 		buf += sizeof(bareq);
1406 		len -= sizeof(bareq);
1407 	}
1408 out:
1409 	BRIDGE_UNLOCK(sc);
1410 	bac->ifbac_len = sizeof(bareq) * count;
1411 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1412 	BRIDGE_LOCK(sc);
1413 	free(outbuf, M_TEMP);
1414 	return (error);
1415 }
1416 
1417 static int
1418 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1419 {
1420 	struct ifbareq *req = arg;
1421 	struct bridge_iflist *bif;
1422 	int error;
1423 
1424 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1425 	if (bif == NULL)
1426 		return (ENOENT);
1427 
1428 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1429 	    req->ifba_flags);
1430 
1431 	return (error);
1432 }
1433 
1434 static int
1435 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1436 {
1437 	struct ifbrparam *param = arg;
1438 
1439 	sc->sc_brttimeout = param->ifbrp_ctime;
1440 	return (0);
1441 }
1442 
1443 static int
1444 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1445 {
1446 	struct ifbrparam *param = arg;
1447 
1448 	param->ifbrp_ctime = sc->sc_brttimeout;
1449 	return (0);
1450 }
1451 
1452 static int
1453 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1454 {
1455 	struct ifbareq *req = arg;
1456 
1457 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1458 }
1459 
1460 static int
1461 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1462 {
1463 	struct ifbreq *req = arg;
1464 
1465 	bridge_rtflush(sc, req->ifbr_ifsflags);
1466 	return (0);
1467 }
1468 
1469 static int
1470 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1471 {
1472 	struct ifbrparam *param = arg;
1473 	struct bstp_state *bs = &sc->sc_stp;
1474 
1475 	param->ifbrp_prio = bs->bs_bridge_priority;
1476 	return (0);
1477 }
1478 
1479 static int
1480 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1481 {
1482 	struct ifbrparam *param = arg;
1483 
1484 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1485 }
1486 
1487 static int
1488 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1489 {
1490 	struct ifbrparam *param = arg;
1491 	struct bstp_state *bs = &sc->sc_stp;
1492 
1493 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1494 	return (0);
1495 }
1496 
1497 static int
1498 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1499 {
1500 	struct ifbrparam *param = arg;
1501 
1502 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1503 }
1504 
1505 static int
1506 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1507 {
1508 	struct ifbrparam *param = arg;
1509 	struct bstp_state *bs = &sc->sc_stp;
1510 
1511 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1512 	return (0);
1513 }
1514 
1515 static int
1516 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1517 {
1518 	struct ifbrparam *param = arg;
1519 
1520 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1521 }
1522 
1523 static int
1524 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1525 {
1526 	struct ifbrparam *param = arg;
1527 	struct bstp_state *bs = &sc->sc_stp;
1528 
1529 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1530 	return (0);
1531 }
1532 
1533 static int
1534 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1535 {
1536 	struct ifbrparam *param = arg;
1537 
1538 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1539 }
1540 
1541 static int
1542 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1543 {
1544 	struct ifbreq *req = arg;
1545 	struct bridge_iflist *bif;
1546 
1547 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1548 	if (bif == NULL)
1549 		return (ENOENT);
1550 
1551 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1552 }
1553 
1554 static int
1555 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1556 {
1557 	struct ifbreq *req = arg;
1558 	struct bridge_iflist *bif;
1559 
1560 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1561 	if (bif == NULL)
1562 		return (ENOENT);
1563 
1564 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1565 }
1566 
1567 static int
1568 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1569 {
1570 	struct ifbreq *req = arg;
1571 	struct bridge_iflist *bif;
1572 
1573 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1574 	if (bif == NULL)
1575 		return (ENOENT);
1576 
1577 	bif->bif_addrmax = req->ifbr_addrmax;
1578 	return (0);
1579 }
1580 
1581 static int
1582 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1583 {
1584 	struct ifbreq *req = arg;
1585 	struct bridge_iflist *bif = NULL;
1586 	struct ifnet *ifs;
1587 
1588 	ifs = ifunit(req->ifbr_ifsname);
1589 	if (ifs == NULL)
1590 		return (ENOENT);
1591 
1592 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1593 		if (ifs == bif->bif_ifp)
1594 			return (EBUSY);
1595 
1596 	if (ifs->if_bridge != NULL)
1597 		return (EBUSY);
1598 
1599 	switch (ifs->if_type) {
1600 		case IFT_ETHER:
1601 		case IFT_GIF:
1602 		case IFT_L2VLAN:
1603 			break;
1604 		default:
1605 			return (EINVAL);
1606 	}
1607 
1608 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1609 	if (bif == NULL)
1610 		return (ENOMEM);
1611 
1612 	bif->bif_ifp = ifs;
1613 	bif->bif_flags = IFBIF_SPAN;
1614 
1615 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1616 
1617 	return (0);
1618 }
1619 
1620 static int
1621 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1622 {
1623 	struct ifbreq *req = arg;
1624 	struct bridge_iflist *bif;
1625 	struct ifnet *ifs;
1626 
1627 	ifs = ifunit(req->ifbr_ifsname);
1628 	if (ifs == NULL)
1629 		return (ENOENT);
1630 
1631 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1632 		if (ifs == bif->bif_ifp)
1633 			break;
1634 
1635 	if (bif == NULL)
1636 		return (ENOENT);
1637 
1638 	bridge_delete_span(sc, bif);
1639 
1640 	return (0);
1641 }
1642 
1643 static int
1644 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1645 {
1646 	struct ifbropreq *req = arg;
1647 	struct bstp_state *bs = &sc->sc_stp;
1648 	struct bstp_port *root_port;
1649 
1650 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1651 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1652 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1653 
1654 	root_port = bs->bs_root_port;
1655 	if (root_port == NULL)
1656 		req->ifbop_root_port = 0;
1657 	else
1658 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1659 
1660 	req->ifbop_holdcount = bs->bs_txholdcount;
1661 	req->ifbop_priority = bs->bs_bridge_priority;
1662 	req->ifbop_protocol = bs->bs_protover;
1663 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1664 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1665 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1666 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1667 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1668 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1669 
1670 	return (0);
1671 }
1672 
1673 static int
1674 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1675 {
1676 	struct ifbrparam *param = arg;
1677 
1678 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1679 	return (0);
1680 }
1681 
1682 static int
1683 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1684 {
1685 	struct ifbpstpconf *bifstp = arg;
1686 	struct bridge_iflist *bif;
1687 	struct bstp_port *bp;
1688 	struct ifbpstpreq bpreq;
1689 	char *buf, *outbuf;
1690 	int count, buflen, len, error = 0;
1691 
1692 	count = 0;
1693 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1694 		if ((bif->bif_flags & IFBIF_STP) != 0)
1695 			count++;
1696 	}
1697 
1698 	buflen = sizeof(bpreq) * count;
1699 	if (bifstp->ifbpstp_len == 0) {
1700 		bifstp->ifbpstp_len = buflen;
1701 		return (0);
1702 	}
1703 
1704 	BRIDGE_UNLOCK(sc);
1705 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1706 	BRIDGE_LOCK(sc);
1707 
1708 	count = 0;
1709 	buf = outbuf;
1710 	len = min(bifstp->ifbpstp_len, buflen);
1711 	bzero(&bpreq, sizeof(bpreq));
1712 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1713 		if (len < sizeof(bpreq))
1714 			break;
1715 
1716 		if ((bif->bif_flags & IFBIF_STP) == 0)
1717 			continue;
1718 
1719 		bp = &bif->bif_stp;
1720 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1721 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1722 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1723 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1724 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1725 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1726 
1727 		memcpy(buf, &bpreq, sizeof(bpreq));
1728 		count++;
1729 		buf += sizeof(bpreq);
1730 		len -= sizeof(bpreq);
1731 	}
1732 
1733 	BRIDGE_UNLOCK(sc);
1734 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1735 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1736 	BRIDGE_LOCK(sc);
1737 	free(outbuf, M_TEMP);
1738 	return (error);
1739 }
1740 
1741 static int
1742 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1743 {
1744 	struct ifbrparam *param = arg;
1745 
1746 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1747 }
1748 
1749 static int
1750 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1751 {
1752 	struct ifbrparam *param = arg;
1753 
1754 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1755 }
1756 
1757 /*
1758  * bridge_ifdetach:
1759  *
1760  *	Detach an interface from a bridge.  Called when a member
1761  *	interface is detaching.
1762  */
1763 static void
1764 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1765 {
1766 	struct bridge_softc *sc = ifp->if_bridge;
1767 	struct bridge_iflist *bif;
1768 
1769 	if (ifp->if_flags & IFF_RENAMING)
1770 		return;
1771 
1772 	/* Check if the interface is a bridge member */
1773 	if (sc != NULL) {
1774 		BRIDGE_LOCK(sc);
1775 
1776 		bif = bridge_lookup_member_if(sc, ifp);
1777 		if (bif != NULL)
1778 			bridge_delete_member(sc, bif, 1);
1779 
1780 		BRIDGE_UNLOCK(sc);
1781 		return;
1782 	}
1783 
1784 	/* Check if the interface is a span port */
1785 	mtx_lock(&bridge_list_mtx);
1786 	LIST_FOREACH(sc, &bridge_list, sc_list) {
1787 		BRIDGE_LOCK(sc);
1788 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1789 			if (ifp == bif->bif_ifp) {
1790 				bridge_delete_span(sc, bif);
1791 				break;
1792 			}
1793 
1794 		BRIDGE_UNLOCK(sc);
1795 	}
1796 	mtx_unlock(&bridge_list_mtx);
1797 }
1798 
1799 /*
1800  * bridge_init:
1801  *
1802  *	Initialize a bridge interface.
1803  */
1804 static void
1805 bridge_init(void *xsc)
1806 {
1807 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1808 	struct ifnet *ifp = sc->sc_ifp;
1809 
1810 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1811 		return;
1812 
1813 	BRIDGE_LOCK(sc);
1814 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1815 	    bridge_timer, sc);
1816 
1817 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1818 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1819 
1820 	BRIDGE_UNLOCK(sc);
1821 }
1822 
1823 /*
1824  * bridge_stop:
1825  *
1826  *	Stop the bridge interface.
1827  */
1828 static void
1829 bridge_stop(struct ifnet *ifp, int disable)
1830 {
1831 	struct bridge_softc *sc = ifp->if_softc;
1832 
1833 	BRIDGE_LOCK_ASSERT(sc);
1834 
1835 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1836 		return;
1837 
1838 	callout_stop(&sc->sc_brcallout);
1839 	bstp_stop(&sc->sc_stp);
1840 
1841 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1842 
1843 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1844 }
1845 
1846 /*
1847  * bridge_enqueue:
1848  *
1849  *	Enqueue a packet on a bridge member interface.
1850  *
1851  */
1852 static int
1853 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1854 {
1855 	int len, err = 0;
1856 	short mflags;
1857 	struct mbuf *m0;
1858 
1859 	/* We may be sending a fragment so traverse the mbuf */
1860 	for (; m; m = m0) {
1861 		m0 = m->m_nextpkt;
1862 		m->m_nextpkt = NULL;
1863 		len = m->m_pkthdr.len;
1864 		mflags = m->m_flags;
1865 
1866 		/*
1867 		 * If underlying interface can not do VLAN tag insertion itself
1868 		 * then attach a packet tag that holds it.
1869 		 */
1870 		if ((m->m_flags & M_VLANTAG) &&
1871 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1872 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1873 			if (m == NULL) {
1874 				if_printf(dst_ifp,
1875 				    "unable to prepend VLAN header\n");
1876 				dst_ifp->if_oerrors++;
1877 				continue;
1878 			}
1879 			m->m_flags &= ~M_VLANTAG;
1880 		}
1881 
1882 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1883 			m_freem(m0);
1884 			sc->sc_ifp->if_oerrors++;
1885 			break;
1886 		}
1887 
1888 		sc->sc_ifp->if_opackets++;
1889 		sc->sc_ifp->if_obytes += len;
1890 		if (mflags & M_MCAST)
1891 			sc->sc_ifp->if_omcasts++;
1892 	}
1893 
1894 	return (err);
1895 }
1896 
1897 /*
1898  * bridge_dummynet:
1899  *
1900  * 	Receive a queued packet from dummynet and pass it on to the output
1901  * 	interface.
1902  *
1903  *	The mbuf has the Ethernet header already attached.
1904  */
1905 static void
1906 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1907 {
1908 	struct bridge_softc *sc;
1909 
1910 	sc = ifp->if_bridge;
1911 
1912 	/*
1913 	 * The packet didnt originate from a member interface. This should only
1914 	 * ever happen if a member interface is removed while packets are
1915 	 * queued for it.
1916 	 */
1917 	if (sc == NULL) {
1918 		m_freem(m);
1919 		return;
1920 	}
1921 
1922 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1923 #ifdef INET6
1924 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1925 #endif
1926 	    ) {
1927 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1928 			return;
1929 		if (m == NULL)
1930 			return;
1931 	}
1932 
1933 	bridge_enqueue(sc, ifp, m);
1934 }
1935 
1936 /*
1937  * bridge_output:
1938  *
1939  *	Send output from a bridge member interface.  This
1940  *	performs the bridging function for locally originated
1941  *	packets.
1942  *
1943  *	The mbuf has the Ethernet header already attached.  We must
1944  *	enqueue or free the mbuf before returning.
1945  */
1946 static int
1947 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1948     struct rtentry *rt)
1949 {
1950 	struct ether_header *eh;
1951 	struct ifnet *dst_if;
1952 	struct bridge_softc *sc;
1953 	uint16_t vlan;
1954 
1955 	if (m->m_len < ETHER_HDR_LEN) {
1956 		m = m_pullup(m, ETHER_HDR_LEN);
1957 		if (m == NULL)
1958 			return (0);
1959 	}
1960 
1961 	eh = mtod(m, struct ether_header *);
1962 	sc = ifp->if_bridge;
1963 	vlan = VLANTAGOF(m);
1964 
1965 	BRIDGE_LOCK(sc);
1966 
1967 	/*
1968 	 * If bridge is down, but the original output interface is up,
1969 	 * go ahead and send out that interface.  Otherwise, the packet
1970 	 * is dropped below.
1971 	 */
1972 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1973 		dst_if = ifp;
1974 		goto sendunicast;
1975 	}
1976 
1977 	/*
1978 	 * If the packet is a multicast, or we don't know a better way to
1979 	 * get there, send to all interfaces.
1980 	 */
1981 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1982 		dst_if = NULL;
1983 	else
1984 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1985 	if (dst_if == NULL) {
1986 		struct bridge_iflist *bif;
1987 		struct mbuf *mc;
1988 		int error = 0, used = 0;
1989 
1990 		bridge_span(sc, m);
1991 
1992 		BRIDGE_LOCK2REF(sc, error);
1993 		if (error) {
1994 			m_freem(m);
1995 			return (0);
1996 		}
1997 
1998 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1999 			dst_if = bif->bif_ifp;
2000 
2001 			if (dst_if->if_type == IFT_GIF)
2002 				continue;
2003 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2004 				continue;
2005 
2006 			/*
2007 			 * If this is not the original output interface,
2008 			 * and the interface is participating in spanning
2009 			 * tree, make sure the port is in a state that
2010 			 * allows forwarding.
2011 			 */
2012 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2013 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2014 				continue;
2015 
2016 			if (LIST_NEXT(bif, bif_next) == NULL) {
2017 				used = 1;
2018 				mc = m;
2019 			} else {
2020 				mc = m_copypacket(m, M_NOWAIT);
2021 				if (mc == NULL) {
2022 					sc->sc_ifp->if_oerrors++;
2023 					continue;
2024 				}
2025 			}
2026 
2027 			bridge_enqueue(sc, dst_if, mc);
2028 		}
2029 		if (used == 0)
2030 			m_freem(m);
2031 		BRIDGE_UNREF(sc);
2032 		return (0);
2033 	}
2034 
2035 sendunicast:
2036 	/*
2037 	 * XXX Spanning tree consideration here?
2038 	 */
2039 
2040 	bridge_span(sc, m);
2041 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2042 		m_freem(m);
2043 		BRIDGE_UNLOCK(sc);
2044 		return (0);
2045 	}
2046 
2047 	BRIDGE_UNLOCK(sc);
2048 	bridge_enqueue(sc, dst_if, m);
2049 	return (0);
2050 }
2051 
2052 /*
2053  * bridge_transmit:
2054  *
2055  *	Do output on a bridge.
2056  *
2057  */
2058 static int
2059 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2060 {
2061 	struct bridge_softc *sc;
2062 	struct ether_header *eh;
2063 	struct ifnet *dst_if;
2064 	int error = 0;
2065 
2066 	sc = ifp->if_softc;
2067 
2068 	ETHER_BPF_MTAP(ifp, m);
2069 
2070 	eh = mtod(m, struct ether_header *);
2071 
2072 	BRIDGE_LOCK(sc);
2073 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2074 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2075 		BRIDGE_UNLOCK(sc);
2076 		error = bridge_enqueue(sc, dst_if, m);
2077 	} else
2078 		bridge_broadcast(sc, ifp, m, 0);
2079 
2080 	return (error);
2081 }
2082 
2083 /*
2084  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2085  */
2086 static void
2087 bridge_qflush(struct ifnet *ifp __unused)
2088 {
2089 }
2090 
2091 /*
2092  * bridge_forward:
2093  *
2094  *	The forwarding function of the bridge.
2095  *
2096  *	NOTE: Releases the lock on return.
2097  */
2098 static void
2099 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2100     struct mbuf *m)
2101 {
2102 	struct bridge_iflist *dbif;
2103 	struct ifnet *src_if, *dst_if, *ifp;
2104 	struct ether_header *eh;
2105 	uint16_t vlan;
2106 	uint8_t *dst;
2107 	int error;
2108 
2109 	src_if = m->m_pkthdr.rcvif;
2110 	ifp = sc->sc_ifp;
2111 
2112 	ifp->if_ipackets++;
2113 	ifp->if_ibytes += m->m_pkthdr.len;
2114 	vlan = VLANTAGOF(m);
2115 
2116 	if ((sbif->bif_flags & IFBIF_STP) &&
2117 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2118 		goto drop;
2119 
2120 	eh = mtod(m, struct ether_header *);
2121 	dst = eh->ether_dhost;
2122 
2123 	/* If the interface is learning, record the address. */
2124 	if (sbif->bif_flags & IFBIF_LEARNING) {
2125 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2126 		    sbif, 0, IFBAF_DYNAMIC);
2127 		/*
2128 		 * If the interface has addresses limits then deny any source
2129 		 * that is not in the cache.
2130 		 */
2131 		if (error && sbif->bif_addrmax)
2132 			goto drop;
2133 	}
2134 
2135 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2136 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2137 		goto drop;
2138 
2139 	/*
2140 	 * At this point, the port either doesn't participate
2141 	 * in spanning tree or it is in the forwarding state.
2142 	 */
2143 
2144 	/*
2145 	 * If the packet is unicast, destined for someone on
2146 	 * "this" side of the bridge, drop it.
2147 	 */
2148 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2149 		dst_if = bridge_rtlookup(sc, dst, vlan);
2150 		if (src_if == dst_if)
2151 			goto drop;
2152 	} else {
2153 		/*
2154 		 * Check if its a reserved multicast address, any address
2155 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2156 		 * bridge.
2157 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2158 		 */
2159 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2160 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2161 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2162 			goto drop;
2163 
2164 		/* ...forward it to all interfaces. */
2165 		ifp->if_imcasts++;
2166 		dst_if = NULL;
2167 	}
2168 
2169 	/*
2170 	 * If we have a destination interface which is a member of our bridge,
2171 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2172 	 * For broadcast or multicast packets, don't bother because it will
2173 	 * be reinjected into ether_input. We do this before we pass the packets
2174 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2175 	 * drop the packet, or possibly modify it, making it difficult to debug
2176 	 * firewall issues on the bridge.
2177 	 */
2178 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2179 		ETHER_BPF_MTAP(ifp, m);
2180 
2181 	/* run the packet filter */
2182 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2183 #ifdef INET6
2184 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2185 #endif
2186 	    ) {
2187 		BRIDGE_UNLOCK(sc);
2188 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2189 			return;
2190 		if (m == NULL)
2191 			return;
2192 		BRIDGE_LOCK(sc);
2193 	}
2194 
2195 	if (dst_if == NULL) {
2196 		bridge_broadcast(sc, src_if, m, 1);
2197 		return;
2198 	}
2199 
2200 	/*
2201 	 * At this point, we're dealing with a unicast frame
2202 	 * going to a different interface.
2203 	 */
2204 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2205 		goto drop;
2206 
2207 	dbif = bridge_lookup_member_if(sc, dst_if);
2208 	if (dbif == NULL)
2209 		/* Not a member of the bridge (anymore?) */
2210 		goto drop;
2211 
2212 	/* Private segments can not talk to each other */
2213 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2214 		goto drop;
2215 
2216 	if ((dbif->bif_flags & IFBIF_STP) &&
2217 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2218 		goto drop;
2219 
2220 	BRIDGE_UNLOCK(sc);
2221 
2222 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2223 #ifdef INET6
2224 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2225 #endif
2226 	    ) {
2227 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2228 			return;
2229 		if (m == NULL)
2230 			return;
2231 	}
2232 
2233 	bridge_enqueue(sc, dst_if, m);
2234 	return;
2235 
2236 drop:
2237 	BRIDGE_UNLOCK(sc);
2238 	m_freem(m);
2239 }
2240 
2241 /*
2242  * bridge_input:
2243  *
2244  *	Receive input from a member interface.  Queue the packet for
2245  *	bridging if it is not for us.
2246  */
2247 static struct mbuf *
2248 bridge_input(struct ifnet *ifp, struct mbuf *m)
2249 {
2250 	struct bridge_softc *sc = ifp->if_bridge;
2251 	struct bridge_iflist *bif, *bif2;
2252 	struct ifnet *bifp;
2253 	struct ether_header *eh;
2254 	struct mbuf *mc, *mc2;
2255 	uint16_t vlan;
2256 	int error;
2257 
2258 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2259 		return (m);
2260 
2261 	bifp = sc->sc_ifp;
2262 	vlan = VLANTAGOF(m);
2263 
2264 	/*
2265 	 * Implement support for bridge monitoring. If this flag has been
2266 	 * set on this interface, discard the packet once we push it through
2267 	 * the bpf(4) machinery, but before we do, increment the byte and
2268 	 * packet counters associated with this interface.
2269 	 */
2270 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2271 		m->m_pkthdr.rcvif  = bifp;
2272 		ETHER_BPF_MTAP(bifp, m);
2273 		bifp->if_ipackets++;
2274 		bifp->if_ibytes += m->m_pkthdr.len;
2275 		m_freem(m);
2276 		return (NULL);
2277 	}
2278 	BRIDGE_LOCK(sc);
2279 	bif = bridge_lookup_member_if(sc, ifp);
2280 	if (bif == NULL) {
2281 		BRIDGE_UNLOCK(sc);
2282 		return (m);
2283 	}
2284 
2285 	eh = mtod(m, struct ether_header *);
2286 
2287 	bridge_span(sc, m);
2288 
2289 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2290 		/* Tap off 802.1D packets; they do not get forwarded. */
2291 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2292 		    ETHER_ADDR_LEN) == 0) {
2293 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2294 			BRIDGE_UNLOCK(sc);
2295 			return (NULL);
2296 		}
2297 
2298 		if ((bif->bif_flags & IFBIF_STP) &&
2299 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2300 			BRIDGE_UNLOCK(sc);
2301 			return (m);
2302 		}
2303 
2304 		/*
2305 		 * Make a deep copy of the packet and enqueue the copy
2306 		 * for bridge processing; return the original packet for
2307 		 * local processing.
2308 		 */
2309 		mc = m_dup(m, M_NOWAIT);
2310 		if (mc == NULL) {
2311 			BRIDGE_UNLOCK(sc);
2312 			return (m);
2313 		}
2314 
2315 		/* Perform the bridge forwarding function with the copy. */
2316 		bridge_forward(sc, bif, mc);
2317 
2318 		/*
2319 		 * Reinject the mbuf as arriving on the bridge so we have a
2320 		 * chance at claiming multicast packets. We can not loop back
2321 		 * here from ether_input as a bridge is never a member of a
2322 		 * bridge.
2323 		 */
2324 		KASSERT(bifp->if_bridge == NULL,
2325 		    ("loop created in bridge_input"));
2326 		mc2 = m_dup(m, M_NOWAIT);
2327 		if (mc2 != NULL) {
2328 			/* Keep the layer3 header aligned */
2329 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2330 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2331 		}
2332 		if (mc2 != NULL) {
2333 			mc2->m_pkthdr.rcvif = bifp;
2334 			(*bifp->if_input)(bifp, mc2);
2335 		}
2336 
2337 		/* Return the original packet for local processing. */
2338 		return (m);
2339 	}
2340 
2341 	if ((bif->bif_flags & IFBIF_STP) &&
2342 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2343 		BRIDGE_UNLOCK(sc);
2344 		return (m);
2345 	}
2346 
2347 #if (defined(INET) || defined(INET6))
2348 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2349 	|| ((iface)->if_carp \
2350 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2351 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2352 	|| ((iface)->if_carp \
2353 	    && (*carp_forus_p)((iface), eh->ether_shost))
2354 #else
2355 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2356 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2357 #endif
2358 
2359 #ifdef INET6
2360 #   define OR_PFIL_HOOKED_INET6 \
2361 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2362 #else
2363 #   define OR_PFIL_HOOKED_INET6
2364 #endif
2365 
2366 #define GRAB_OUR_PACKETS(iface) \
2367 	if ((iface)->if_type == IFT_GIF) \
2368 		continue; \
2369 	/* It is destined for us. */ \
2370 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2371 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2372 	    ) {								\
2373 		if ((iface)->if_type == IFT_BRIDGE) {			\
2374 			ETHER_BPF_MTAP(iface, m);			\
2375 			iface->if_ipackets++;				\
2376 			iface->if_ibytes += m->m_pkthdr.len;		\
2377 			/* Filter on the physical interface. */		\
2378 			if (pfil_local_phys &&				\
2379 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2380 			     OR_PFIL_HOOKED_INET6)) {			\
2381 				if (bridge_pfil(&m, NULL, ifp,		\
2382 				    PFIL_IN) != 0 || m == NULL) {	\
2383 					BRIDGE_UNLOCK(sc);		\
2384 					return (NULL);			\
2385 				}					\
2386 				eh = mtod(m, struct ether_header *);	\
2387 			}						\
2388 		}							\
2389 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2390 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2391 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2392 			if (error && bif->bif_addrmax) {		\
2393 				BRIDGE_UNLOCK(sc);			\
2394 				m_freem(m);				\
2395 				return (NULL);				\
2396 			}						\
2397 		}							\
2398 		m->m_pkthdr.rcvif = iface;				\
2399 		BRIDGE_UNLOCK(sc);					\
2400 		return (m);						\
2401 	}								\
2402 									\
2403 	/* We just received a packet that we sent out. */		\
2404 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2405 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2406 	    ) {								\
2407 		BRIDGE_UNLOCK(sc);					\
2408 		m_freem(m);						\
2409 		return (NULL);						\
2410 	}
2411 
2412 	/*
2413 	 * Unicast.  Make sure it's not for the bridge.
2414 	 */
2415 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2416 
2417 	/*
2418 	 * Give a chance for ifp at first priority. This will help when	the
2419 	 * packet comes through the interface like VLAN's with the same MACs
2420 	 * on several interfaces from the same bridge. This also will save
2421 	 * some CPU cycles in case the destination interface and the input
2422 	 * interface (eq ifp) are the same.
2423 	 */
2424 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2425 
2426 	/* Now check the all bridge members. */
2427 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2428 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2429 	}
2430 
2431 #undef OR_CARP_CHECK_WE_ARE_DST
2432 #undef OR_CARP_CHECK_WE_ARE_SRC
2433 #undef OR_PFIL_HOOKED_INET6
2434 #undef GRAB_OUR_PACKETS
2435 
2436 	/* Perform the bridge forwarding function. */
2437 	bridge_forward(sc, bif, m);
2438 
2439 	return (NULL);
2440 }
2441 
2442 /*
2443  * bridge_broadcast:
2444  *
2445  *	Send a frame to all interfaces that are members of
2446  *	the bridge, except for the one on which the packet
2447  *	arrived.
2448  *
2449  *	NOTE: Releases the lock on return.
2450  */
2451 static void
2452 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2453     struct mbuf *m, int runfilt)
2454 {
2455 	struct bridge_iflist *dbif, *sbif;
2456 	struct mbuf *mc;
2457 	struct ifnet *dst_if;
2458 	int error = 0, used = 0, i;
2459 
2460 	sbif = bridge_lookup_member_if(sc, src_if);
2461 
2462 	BRIDGE_LOCK2REF(sc, error);
2463 	if (error) {
2464 		m_freem(m);
2465 		return;
2466 	}
2467 
2468 	/* Filter on the bridge interface before broadcasting */
2469 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2470 #ifdef INET6
2471 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2472 #endif
2473 	    )) {
2474 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2475 			goto out;
2476 		if (m == NULL)
2477 			goto out;
2478 	}
2479 
2480 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2481 		dst_if = dbif->bif_ifp;
2482 		if (dst_if == src_if)
2483 			continue;
2484 
2485 		/* Private segments can not talk to each other */
2486 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2487 			continue;
2488 
2489 		if ((dbif->bif_flags & IFBIF_STP) &&
2490 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2491 			continue;
2492 
2493 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2494 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2495 			continue;
2496 
2497 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2498 			continue;
2499 
2500 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2501 			mc = m;
2502 			used = 1;
2503 		} else {
2504 			mc = m_dup(m, M_NOWAIT);
2505 			if (mc == NULL) {
2506 				sc->sc_ifp->if_oerrors++;
2507 				continue;
2508 			}
2509 		}
2510 
2511 		/*
2512 		 * Filter on the output interface. Pass a NULL bridge interface
2513 		 * pointer so we do not redundantly filter on the bridge for
2514 		 * each interface we broadcast on.
2515 		 */
2516 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2517 #ifdef INET6
2518 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2519 #endif
2520 		    )) {
2521 			if (used == 0) {
2522 				/* Keep the layer3 header aligned */
2523 				i = min(mc->m_pkthdr.len, max_protohdr);
2524 				mc = m_copyup(mc, i, ETHER_ALIGN);
2525 				if (mc == NULL) {
2526 					sc->sc_ifp->if_oerrors++;
2527 					continue;
2528 				}
2529 			}
2530 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2531 				continue;
2532 			if (mc == NULL)
2533 				continue;
2534 		}
2535 
2536 		bridge_enqueue(sc, dst_if, mc);
2537 	}
2538 	if (used == 0)
2539 		m_freem(m);
2540 
2541 out:
2542 	BRIDGE_UNREF(sc);
2543 }
2544 
2545 /*
2546  * bridge_span:
2547  *
2548  *	Duplicate a packet out one or more interfaces that are in span mode,
2549  *	the original mbuf is unmodified.
2550  */
2551 static void
2552 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2553 {
2554 	struct bridge_iflist *bif;
2555 	struct ifnet *dst_if;
2556 	struct mbuf *mc;
2557 
2558 	if (LIST_EMPTY(&sc->sc_spanlist))
2559 		return;
2560 
2561 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2562 		dst_if = bif->bif_ifp;
2563 
2564 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2565 			continue;
2566 
2567 		mc = m_copypacket(m, M_NOWAIT);
2568 		if (mc == NULL) {
2569 			sc->sc_ifp->if_oerrors++;
2570 			continue;
2571 		}
2572 
2573 		bridge_enqueue(sc, dst_if, mc);
2574 	}
2575 }
2576 
2577 /*
2578  * bridge_rtupdate:
2579  *
2580  *	Add a bridge routing entry.
2581  */
2582 static int
2583 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2584     struct bridge_iflist *bif, int setflags, uint8_t flags)
2585 {
2586 	struct bridge_rtnode *brt;
2587 	int error;
2588 
2589 	BRIDGE_LOCK_ASSERT(sc);
2590 
2591 	/* Check the source address is valid and not multicast. */
2592 	if (ETHER_IS_MULTICAST(dst) ||
2593 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2594 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2595 		return (EINVAL);
2596 
2597 	/* 802.1p frames map to vlan 1 */
2598 	if (vlan == 0)
2599 		vlan = 1;
2600 
2601 	/*
2602 	 * A route for this destination might already exist.  If so,
2603 	 * update it, otherwise create a new one.
2604 	 */
2605 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2606 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2607 			sc->sc_brtexceeded++;
2608 			return (ENOSPC);
2609 		}
2610 		/* Check per interface address limits (if enabled) */
2611 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2612 			bif->bif_addrexceeded++;
2613 			return (ENOSPC);
2614 		}
2615 
2616 		/*
2617 		 * Allocate a new bridge forwarding node, and
2618 		 * initialize the expiration time and Ethernet
2619 		 * address.
2620 		 */
2621 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2622 		if (brt == NULL)
2623 			return (ENOMEM);
2624 
2625 		if (bif->bif_flags & IFBIF_STICKY)
2626 			brt->brt_flags = IFBAF_STICKY;
2627 		else
2628 			brt->brt_flags = IFBAF_DYNAMIC;
2629 
2630 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2631 		brt->brt_vlan = vlan;
2632 
2633 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2634 			uma_zfree(bridge_rtnode_zone, brt);
2635 			return (error);
2636 		}
2637 		brt->brt_dst = bif;
2638 		bif->bif_addrcnt++;
2639 	}
2640 
2641 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2642 	    brt->brt_dst != bif) {
2643 		brt->brt_dst->bif_addrcnt--;
2644 		brt->brt_dst = bif;
2645 		brt->brt_dst->bif_addrcnt++;
2646 	}
2647 
2648 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2649 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2650 	if (setflags)
2651 		brt->brt_flags = flags;
2652 
2653 	return (0);
2654 }
2655 
2656 /*
2657  * bridge_rtlookup:
2658  *
2659  *	Lookup the destination interface for an address.
2660  */
2661 static struct ifnet *
2662 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2663 {
2664 	struct bridge_rtnode *brt;
2665 
2666 	BRIDGE_LOCK_ASSERT(sc);
2667 
2668 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2669 		return (NULL);
2670 
2671 	return (brt->brt_ifp);
2672 }
2673 
2674 /*
2675  * bridge_rttrim:
2676  *
2677  *	Trim the routine table so that we have a number
2678  *	of routing entries less than or equal to the
2679  *	maximum number.
2680  */
2681 static void
2682 bridge_rttrim(struct bridge_softc *sc)
2683 {
2684 	struct bridge_rtnode *brt, *nbrt;
2685 
2686 	BRIDGE_LOCK_ASSERT(sc);
2687 
2688 	/* Make sure we actually need to do this. */
2689 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2690 		return;
2691 
2692 	/* Force an aging cycle; this might trim enough addresses. */
2693 	bridge_rtage(sc);
2694 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2695 		return;
2696 
2697 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2698 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2699 			bridge_rtnode_destroy(sc, brt);
2700 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2701 				return;
2702 		}
2703 	}
2704 }
2705 
2706 /*
2707  * bridge_timer:
2708  *
2709  *	Aging timer for the bridge.
2710  */
2711 static void
2712 bridge_timer(void *arg)
2713 {
2714 	struct bridge_softc *sc = arg;
2715 
2716 	BRIDGE_LOCK_ASSERT(sc);
2717 
2718 	bridge_rtage(sc);
2719 
2720 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2721 		callout_reset(&sc->sc_brcallout,
2722 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2723 }
2724 
2725 /*
2726  * bridge_rtage:
2727  *
2728  *	Perform an aging cycle.
2729  */
2730 static void
2731 bridge_rtage(struct bridge_softc *sc)
2732 {
2733 	struct bridge_rtnode *brt, *nbrt;
2734 
2735 	BRIDGE_LOCK_ASSERT(sc);
2736 
2737 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2738 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2739 			if (time_uptime >= brt->brt_expire)
2740 				bridge_rtnode_destroy(sc, brt);
2741 		}
2742 	}
2743 }
2744 
2745 /*
2746  * bridge_rtflush:
2747  *
2748  *	Remove all dynamic addresses from the bridge.
2749  */
2750 static void
2751 bridge_rtflush(struct bridge_softc *sc, int full)
2752 {
2753 	struct bridge_rtnode *brt, *nbrt;
2754 
2755 	BRIDGE_LOCK_ASSERT(sc);
2756 
2757 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2758 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2759 			bridge_rtnode_destroy(sc, brt);
2760 	}
2761 }
2762 
2763 /*
2764  * bridge_rtdaddr:
2765  *
2766  *	Remove an address from the table.
2767  */
2768 static int
2769 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2770 {
2771 	struct bridge_rtnode *brt;
2772 	int found = 0;
2773 
2774 	BRIDGE_LOCK_ASSERT(sc);
2775 
2776 	/*
2777 	 * If vlan is zero then we want to delete for all vlans so the lookup
2778 	 * may return more than one.
2779 	 */
2780 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2781 		bridge_rtnode_destroy(sc, brt);
2782 		found = 1;
2783 	}
2784 
2785 	return (found ? 0 : ENOENT);
2786 }
2787 
2788 /*
2789  * bridge_rtdelete:
2790  *
2791  *	Delete routes to a speicifc member interface.
2792  */
2793 static void
2794 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2795 {
2796 	struct bridge_rtnode *brt, *nbrt;
2797 
2798 	BRIDGE_LOCK_ASSERT(sc);
2799 
2800 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2801 		if (brt->brt_ifp == ifp && (full ||
2802 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2803 			bridge_rtnode_destroy(sc, brt);
2804 	}
2805 }
2806 
2807 /*
2808  * bridge_rtable_init:
2809  *
2810  *	Initialize the route table for this bridge.
2811  */
2812 static void
2813 bridge_rtable_init(struct bridge_softc *sc)
2814 {
2815 	int i;
2816 
2817 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2818 	    M_DEVBUF, M_WAITOK);
2819 
2820 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2821 		LIST_INIT(&sc->sc_rthash[i]);
2822 
2823 	sc->sc_rthash_key = arc4random();
2824 	LIST_INIT(&sc->sc_rtlist);
2825 }
2826 
2827 /*
2828  * bridge_rtable_fini:
2829  *
2830  *	Deconstruct the route table for this bridge.
2831  */
2832 static void
2833 bridge_rtable_fini(struct bridge_softc *sc)
2834 {
2835 
2836 	KASSERT(sc->sc_brtcnt == 0,
2837 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2838 	free(sc->sc_rthash, M_DEVBUF);
2839 }
2840 
2841 /*
2842  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2843  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2844  */
2845 #define	mix(a, b, c)							\
2846 do {									\
2847 	a -= b; a -= c; a ^= (c >> 13);					\
2848 	b -= c; b -= a; b ^= (a << 8);					\
2849 	c -= a; c -= b; c ^= (b >> 13);					\
2850 	a -= b; a -= c; a ^= (c >> 12);					\
2851 	b -= c; b -= a; b ^= (a << 16);					\
2852 	c -= a; c -= b; c ^= (b >> 5);					\
2853 	a -= b; a -= c; a ^= (c >> 3);					\
2854 	b -= c; b -= a; b ^= (a << 10);					\
2855 	c -= a; c -= b; c ^= (b >> 15);					\
2856 } while (/*CONSTCOND*/0)
2857 
2858 static __inline uint32_t
2859 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2860 {
2861 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2862 
2863 	b += addr[5] << 8;
2864 	b += addr[4];
2865 	a += addr[3] << 24;
2866 	a += addr[2] << 16;
2867 	a += addr[1] << 8;
2868 	a += addr[0];
2869 
2870 	mix(a, b, c);
2871 
2872 	return (c & BRIDGE_RTHASH_MASK);
2873 }
2874 
2875 #undef mix
2876 
2877 static int
2878 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2879 {
2880 	int i, d;
2881 
2882 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2883 		d = ((int)a[i]) - ((int)b[i]);
2884 	}
2885 
2886 	return (d);
2887 }
2888 
2889 /*
2890  * bridge_rtnode_lookup:
2891  *
2892  *	Look up a bridge route node for the specified destination. Compare the
2893  *	vlan id or if zero then just return the first match.
2894  */
2895 static struct bridge_rtnode *
2896 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2897 {
2898 	struct bridge_rtnode *brt;
2899 	uint32_t hash;
2900 	int dir;
2901 
2902 	BRIDGE_LOCK_ASSERT(sc);
2903 
2904 	hash = bridge_rthash(sc, addr);
2905 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2906 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2907 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2908 			return (brt);
2909 		if (dir > 0)
2910 			return (NULL);
2911 	}
2912 
2913 	return (NULL);
2914 }
2915 
2916 /*
2917  * bridge_rtnode_insert:
2918  *
2919  *	Insert the specified bridge node into the route table.  We
2920  *	assume the entry is not already in the table.
2921  */
2922 static int
2923 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2924 {
2925 	struct bridge_rtnode *lbrt;
2926 	uint32_t hash;
2927 	int dir;
2928 
2929 	BRIDGE_LOCK_ASSERT(sc);
2930 
2931 	hash = bridge_rthash(sc, brt->brt_addr);
2932 
2933 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2934 	if (lbrt == NULL) {
2935 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2936 		goto out;
2937 	}
2938 
2939 	do {
2940 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2941 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2942 			return (EEXIST);
2943 		if (dir > 0) {
2944 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2945 			goto out;
2946 		}
2947 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2948 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2949 			goto out;
2950 		}
2951 		lbrt = LIST_NEXT(lbrt, brt_hash);
2952 	} while (lbrt != NULL);
2953 
2954 #ifdef DIAGNOSTIC
2955 	panic("bridge_rtnode_insert: impossible");
2956 #endif
2957 
2958 out:
2959 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2960 	sc->sc_brtcnt++;
2961 
2962 	return (0);
2963 }
2964 
2965 /*
2966  * bridge_rtnode_destroy:
2967  *
2968  *	Destroy a bridge rtnode.
2969  */
2970 static void
2971 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2972 {
2973 	BRIDGE_LOCK_ASSERT(sc);
2974 
2975 	LIST_REMOVE(brt, brt_hash);
2976 
2977 	LIST_REMOVE(brt, brt_list);
2978 	sc->sc_brtcnt--;
2979 	brt->brt_dst->bif_addrcnt--;
2980 	uma_zfree(bridge_rtnode_zone, brt);
2981 }
2982 
2983 /*
2984  * bridge_rtable_expire:
2985  *
2986  *	Set the expiry time for all routes on an interface.
2987  */
2988 static void
2989 bridge_rtable_expire(struct ifnet *ifp, int age)
2990 {
2991 	struct bridge_softc *sc = ifp->if_bridge;
2992 	struct bridge_rtnode *brt;
2993 
2994 	BRIDGE_LOCK(sc);
2995 
2996 	/*
2997 	 * If the age is zero then flush, otherwise set all the expiry times to
2998 	 * age for the interface
2999 	 */
3000 	if (age == 0)
3001 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3002 	else {
3003 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3004 			/* Cap the expiry time to 'age' */
3005 			if (brt->brt_ifp == ifp &&
3006 			    brt->brt_expire > time_uptime + age &&
3007 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3008 				brt->brt_expire = time_uptime + age;
3009 		}
3010 	}
3011 	BRIDGE_UNLOCK(sc);
3012 }
3013 
3014 /*
3015  * bridge_state_change:
3016  *
3017  *	Callback from the bridgestp code when a port changes states.
3018  */
3019 static void
3020 bridge_state_change(struct ifnet *ifp, int state)
3021 {
3022 	struct bridge_softc *sc = ifp->if_bridge;
3023 	static const char *stpstates[] = {
3024 		"disabled",
3025 		"listening",
3026 		"learning",
3027 		"forwarding",
3028 		"blocking",
3029 		"discarding"
3030 	};
3031 
3032 	if (log_stp)
3033 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3034 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3035 }
3036 
3037 /*
3038  * Send bridge packets through pfil if they are one of the types pfil can deal
3039  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3040  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3041  * that interface.
3042  */
3043 static int
3044 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3045 {
3046 	int snap, error, i, hlen;
3047 	struct ether_header *eh1, eh2;
3048 	struct ip *ip;
3049 	struct llc llc1;
3050 	u_int16_t ether_type;
3051 
3052 	snap = 0;
3053 	error = -1;	/* Default error if not error == 0 */
3054 
3055 #if 0
3056 	/* we may return with the IP fields swapped, ensure its not shared */
3057 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3058 #endif
3059 
3060 	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
3061 		return (0); /* filtering is disabled */
3062 
3063 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3064 	if ((*mp)->m_len < i) {
3065 	    *mp = m_pullup(*mp, i);
3066 	    if (*mp == NULL) {
3067 		printf("%s: m_pullup failed\n", __func__);
3068 		return (-1);
3069 	    }
3070 	}
3071 
3072 	eh1 = mtod(*mp, struct ether_header *);
3073 	ether_type = ntohs(eh1->ether_type);
3074 
3075 	/*
3076 	 * Check for SNAP/LLC.
3077 	 */
3078 	if (ether_type < ETHERMTU) {
3079 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3080 
3081 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3082 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3083 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3084 		    llc2->llc_control == LLC_UI) {
3085 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3086 			snap = 1;
3087 		}
3088 	}
3089 
3090 	/*
3091 	 * If we're trying to filter bridge traffic, don't look at anything
3092 	 * other than IP and ARP traffic.  If the filter doesn't understand
3093 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3094 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3095 	 * but of course we don't have an AppleTalk filter to begin with.
3096 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3097 	 * ARP traffic.)
3098 	 */
3099 	switch (ether_type) {
3100 		case ETHERTYPE_ARP:
3101 		case ETHERTYPE_REVARP:
3102 			if (pfil_ipfw_arp == 0)
3103 				return (0); /* Automatically pass */
3104 			break;
3105 
3106 		case ETHERTYPE_IP:
3107 #ifdef INET6
3108 		case ETHERTYPE_IPV6:
3109 #endif /* INET6 */
3110 			break;
3111 		default:
3112 			/*
3113 			 * Check to see if the user wants to pass non-ip
3114 			 * packets, these will not be checked by pfil(9) and
3115 			 * passed unconditionally so the default is to drop.
3116 			 */
3117 			if (pfil_onlyip)
3118 				goto bad;
3119 	}
3120 
3121 	/* Run the packet through pfil before stripping link headers */
3122 	if (PFIL_HOOKED(&V_link_pfil_hook) && pfil_ipfw != 0 &&
3123 			dir == PFIL_OUT && ifp != NULL) {
3124 
3125 		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3126 
3127 		if (*mp == NULL || error != 0) /* packet consumed by filter */
3128 			return (error);
3129 	}
3130 
3131 	/* Strip off the Ethernet header and keep a copy. */
3132 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3133 	m_adj(*mp, ETHER_HDR_LEN);
3134 
3135 	/* Strip off snap header, if present */
3136 	if (snap) {
3137 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3138 		m_adj(*mp, sizeof(struct llc));
3139 	}
3140 
3141 	/*
3142 	 * Check the IP header for alignment and errors
3143 	 */
3144 	if (dir == PFIL_IN) {
3145 		switch (ether_type) {
3146 			case ETHERTYPE_IP:
3147 				error = bridge_ip_checkbasic(mp);
3148 				break;
3149 #ifdef INET6
3150 			case ETHERTYPE_IPV6:
3151 				error = bridge_ip6_checkbasic(mp);
3152 				break;
3153 #endif /* INET6 */
3154 			default:
3155 				error = 0;
3156 		}
3157 		if (error)
3158 			goto bad;
3159 	}
3160 
3161 	error = 0;
3162 
3163 	/*
3164 	 * Run the packet through pfil
3165 	 */
3166 	switch (ether_type) {
3167 	case ETHERTYPE_IP:
3168 		/*
3169 		 * Run pfil on the member interface and the bridge, both can
3170 		 * be skipped by clearing pfil_member or pfil_bridge.
3171 		 *
3172 		 * Keep the order:
3173 		 *   in_if -> bridge_if -> out_if
3174 		 */
3175 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3176 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3177 					dir, NULL);
3178 
3179 		if (*mp == NULL || error != 0) /* filter may consume */
3180 			break;
3181 
3182 		if (pfil_member && ifp != NULL)
3183 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3184 					dir, NULL);
3185 
3186 		if (*mp == NULL || error != 0) /* filter may consume */
3187 			break;
3188 
3189 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3190 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3191 					dir, NULL);
3192 
3193 		if (*mp == NULL || error != 0) /* filter may consume */
3194 			break;
3195 
3196 		/* check if we need to fragment the packet */
3197 		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3198 			i = (*mp)->m_pkthdr.len;
3199 			if (i > ifp->if_mtu) {
3200 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3201 					    &llc1);
3202 				return (error);
3203 			}
3204 		}
3205 
3206 		/* Recalculate the ip checksum. */
3207 		ip = mtod(*mp, struct ip *);
3208 		hlen = ip->ip_hl << 2;
3209 		if (hlen < sizeof(struct ip))
3210 			goto bad;
3211 		if (hlen > (*mp)->m_len) {
3212 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3213 				goto bad;
3214 			ip = mtod(*mp, struct ip *);
3215 			if (ip == NULL)
3216 				goto bad;
3217 		}
3218 		ip->ip_sum = 0;
3219 		if (hlen == sizeof(struct ip))
3220 			ip->ip_sum = in_cksum_hdr(ip);
3221 		else
3222 			ip->ip_sum = in_cksum(*mp, hlen);
3223 
3224 		break;
3225 #ifdef INET6
3226 	case ETHERTYPE_IPV6:
3227 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3228 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3229 					dir, NULL);
3230 
3231 		if (*mp == NULL || error != 0) /* filter may consume */
3232 			break;
3233 
3234 		if (pfil_member && ifp != NULL)
3235 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3236 					dir, NULL);
3237 
3238 		if (*mp == NULL || error != 0) /* filter may consume */
3239 			break;
3240 
3241 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3242 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3243 					dir, NULL);
3244 		break;
3245 #endif
3246 	default:
3247 		error = 0;
3248 		break;
3249 	}
3250 
3251 	if (*mp == NULL)
3252 		return (error);
3253 	if (error != 0)
3254 		goto bad;
3255 
3256 	error = -1;
3257 
3258 	/*
3259 	 * Finally, put everything back the way it was and return
3260 	 */
3261 	if (snap) {
3262 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3263 		if (*mp == NULL)
3264 			return (error);
3265 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3266 	}
3267 
3268 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3269 	if (*mp == NULL)
3270 		return (error);
3271 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3272 
3273 	return (0);
3274 
3275 bad:
3276 	m_freem(*mp);
3277 	*mp = NULL;
3278 	return (error);
3279 }
3280 
3281 /*
3282  * Perform basic checks on header size since
3283  * pfil assumes ip_input has already processed
3284  * it for it.  Cut-and-pasted from ip_input.c.
3285  * Given how simple the IPv6 version is,
3286  * does the IPv4 version really need to be
3287  * this complicated?
3288  *
3289  * XXX Should we update ipstat here, or not?
3290  * XXX Right now we update ipstat but not
3291  * XXX csum_counter.
3292  */
3293 static int
3294 bridge_ip_checkbasic(struct mbuf **mp)
3295 {
3296 	struct mbuf *m = *mp;
3297 	struct ip *ip;
3298 	int len, hlen;
3299 	u_short sum;
3300 
3301 	if (*mp == NULL)
3302 		return (-1);
3303 
3304 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3305 		if ((m = m_copyup(m, sizeof(struct ip),
3306 			(max_linkhdr + 3) & ~3)) == NULL) {
3307 			/* XXXJRT new stat, please */
3308 			KMOD_IPSTAT_INC(ips_toosmall);
3309 			goto bad;
3310 		}
3311 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3312 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3313 			KMOD_IPSTAT_INC(ips_toosmall);
3314 			goto bad;
3315 		}
3316 	}
3317 	ip = mtod(m, struct ip *);
3318 	if (ip == NULL) goto bad;
3319 
3320 	if (ip->ip_v != IPVERSION) {
3321 		KMOD_IPSTAT_INC(ips_badvers);
3322 		goto bad;
3323 	}
3324 	hlen = ip->ip_hl << 2;
3325 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3326 		KMOD_IPSTAT_INC(ips_badhlen);
3327 		goto bad;
3328 	}
3329 	if (hlen > m->m_len) {
3330 		if ((m = m_pullup(m, hlen)) == 0) {
3331 			KMOD_IPSTAT_INC(ips_badhlen);
3332 			goto bad;
3333 		}
3334 		ip = mtod(m, struct ip *);
3335 		if (ip == NULL) goto bad;
3336 	}
3337 
3338 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3339 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3340 	} else {
3341 		if (hlen == sizeof(struct ip)) {
3342 			sum = in_cksum_hdr(ip);
3343 		} else {
3344 			sum = in_cksum(m, hlen);
3345 		}
3346 	}
3347 	if (sum) {
3348 		KMOD_IPSTAT_INC(ips_badsum);
3349 		goto bad;
3350 	}
3351 
3352 	/* Retrieve the packet length. */
3353 	len = ntohs(ip->ip_len);
3354 
3355 	/*
3356 	 * Check for additional length bogosity
3357 	 */
3358 	if (len < hlen) {
3359 		KMOD_IPSTAT_INC(ips_badlen);
3360 		goto bad;
3361 	}
3362 
3363 	/*
3364 	 * Check that the amount of data in the buffers
3365 	 * is as at least much as the IP header would have us expect.
3366 	 * Drop packet if shorter than we expect.
3367 	 */
3368 	if (m->m_pkthdr.len < len) {
3369 		KMOD_IPSTAT_INC(ips_tooshort);
3370 		goto bad;
3371 	}
3372 
3373 	/* Checks out, proceed */
3374 	*mp = m;
3375 	return (0);
3376 
3377 bad:
3378 	*mp = m;
3379 	return (-1);
3380 }
3381 
3382 #ifdef INET6
3383 /*
3384  * Same as above, but for IPv6.
3385  * Cut-and-pasted from ip6_input.c.
3386  * XXX Should we update ip6stat, or not?
3387  */
3388 static int
3389 bridge_ip6_checkbasic(struct mbuf **mp)
3390 {
3391 	struct mbuf *m = *mp;
3392 	struct ip6_hdr *ip6;
3393 
3394 	/*
3395 	 * If the IPv6 header is not aligned, slurp it up into a new
3396 	 * mbuf with space for link headers, in the event we forward
3397 	 * it.  Otherwise, if it is aligned, make sure the entire base
3398 	 * IPv6 header is in the first mbuf of the chain.
3399 	 */
3400 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3401 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3402 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3403 			    (max_linkhdr + 3) & ~3)) == NULL) {
3404 			/* XXXJRT new stat, please */
3405 			IP6STAT_INC(ip6s_toosmall);
3406 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3407 			goto bad;
3408 		}
3409 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3410 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3411 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3412 			IP6STAT_INC(ip6s_toosmall);
3413 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3414 			goto bad;
3415 		}
3416 	}
3417 
3418 	ip6 = mtod(m, struct ip6_hdr *);
3419 
3420 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3421 		IP6STAT_INC(ip6s_badvers);
3422 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3423 		goto bad;
3424 	}
3425 
3426 	/* Checks out, proceed */
3427 	*mp = m;
3428 	return (0);
3429 
3430 bad:
3431 	*mp = m;
3432 	return (-1);
3433 }
3434 #endif /* INET6 */
3435 
3436 /*
3437  * bridge_fragment:
3438  *
3439  *	Return a fragmented mbuf chain.
3440  */
3441 static int
3442 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3443     int snap, struct llc *llc)
3444 {
3445 	struct mbuf *m0;
3446 	struct ip *ip;
3447 	int error = -1;
3448 
3449 	if (m->m_len < sizeof(struct ip) &&
3450 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3451 		goto out;
3452 	ip = mtod(m, struct ip *);
3453 
3454 	m->m_pkthdr.csum_flags |= CSUM_IP;
3455 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3456 	if (error)
3457 		goto out;
3458 
3459 	/* walk the chain and re-add the Ethernet header */
3460 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3461 		if (error == 0) {
3462 			if (snap) {
3463 				M_PREPEND(m0, sizeof(struct llc), M_NOWAIT);
3464 				if (m0 == NULL) {
3465 					error = ENOBUFS;
3466 					continue;
3467 				}
3468 				bcopy(llc, mtod(m0, caddr_t),
3469 				    sizeof(struct llc));
3470 			}
3471 			M_PREPEND(m0, ETHER_HDR_LEN, M_NOWAIT);
3472 			if (m0 == NULL) {
3473 				error = ENOBUFS;
3474 				continue;
3475 			}
3476 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3477 		} else
3478 			m_freem(m);
3479 	}
3480 
3481 	if (error == 0)
3482 		KMOD_IPSTAT_INC(ips_fragmented);
3483 
3484 	return (error);
3485 
3486 out:
3487 	if (m != NULL)
3488 		m_freem(m);
3489 	return (error);
3490 }
3491 
3492 static void
3493 bridge_linkstate(struct ifnet *ifp)
3494 {
3495 	struct bridge_softc *sc = ifp->if_bridge;
3496 	struct bridge_iflist *bif;
3497 
3498 	BRIDGE_LOCK(sc);
3499 	bif = bridge_lookup_member_if(sc, ifp);
3500 	if (bif == NULL) {
3501 		BRIDGE_UNLOCK(sc);
3502 		return;
3503 	}
3504 	bridge_linkcheck(sc);
3505 	BRIDGE_UNLOCK(sc);
3506 
3507 	bstp_linkstate(&bif->bif_stp);
3508 }
3509 
3510 static void
3511 bridge_linkcheck(struct bridge_softc *sc)
3512 {
3513 	struct bridge_iflist *bif;
3514 	int new_link, hasls;
3515 
3516 	BRIDGE_LOCK_ASSERT(sc);
3517 	new_link = LINK_STATE_DOWN;
3518 	hasls = 0;
3519 	/* Our link is considered up if at least one of our ports is active */
3520 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3521 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3522 			hasls++;
3523 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3524 			new_link = LINK_STATE_UP;
3525 			break;
3526 		}
3527 	}
3528 	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3529 		/* If no interfaces support link-state then we default to up */
3530 		new_link = LINK_STATE_UP;
3531 	}
3532 	if_link_state_change(sc->sc_ifp, new_link);
3533 }
3534