xref: /freebsd/sys/net/if_bridge.c (revision 9fd69f37d28cfd7438cac3eeb45fe9dd46b4d7dd)
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 #include "opt_carp.h"
83 
84 #include <sys/param.h>
85 #include <sys/mbuf.h>
86 #include <sys/malloc.h>
87 #include <sys/protosw.h>
88 #include <sys/systm.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 #include <sys/rwlock.h>
104 
105 #include <net/bpf.h>
106 #include <net/if.h>
107 #include <net/if_clone.h>
108 #include <net/if_dl.h>
109 #include <net/if_types.h>
110 #include <net/if_var.h>
111 #include <net/pfil.h>
112 #include <net/vnet.h>
113 
114 #include <netinet/in.h> /* for struct arpcom */
115 #include <netinet/in_systm.h>
116 #include <netinet/in_var.h>
117 #include <netinet/ip.h>
118 #include <netinet/ip_var.h>
119 #ifdef INET6
120 #include <netinet/ip6.h>
121 #include <netinet6/ip6_var.h>
122 #endif
123 #if defined(INET) || defined(INET6)
124 #ifdef DEV_CARP
125 #include <netinet/ip_carp.h>
126 #endif
127 #endif
128 #include <machine/in_cksum.h>
129 #include <netinet/if_ether.h> /* for struct arpcom */
130 #include <net/bridgestp.h>
131 #include <net/if_bridgevar.h>
132 #include <net/if_llc.h>
133 #include <net/if_vlan_var.h>
134 
135 #include <net/route.h>
136 #include <netinet/ip_fw.h>
137 #include <netinet/ipfw/ip_fw_private.h>
138 #include <netinet/ip_dummynet.h>
139 
140 /*
141  * Size of the route hash table.  Must be a power of two.
142  */
143 #ifndef BRIDGE_RTHASH_SIZE
144 #define	BRIDGE_RTHASH_SIZE		1024
145 #endif
146 
147 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
148 
149 /*
150  * Maximum number of addresses to cache.
151  */
152 #ifndef BRIDGE_RTABLE_MAX
153 #define	BRIDGE_RTABLE_MAX		100
154 #endif
155 
156 /*
157  * Timeout (in seconds) for entries learned dynamically.
158  */
159 #ifndef BRIDGE_RTABLE_TIMEOUT
160 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
161 #endif
162 
163 /*
164  * Number of seconds between walks of the route list.
165  */
166 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
167 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
168 #endif
169 
170 /*
171  * List of capabilities to possibly mask on the member interface.
172  */
173 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
174 
175 /*
176  * List of capabilities to strip
177  */
178 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
179 
180 /*
181  * Bridge interface list entry.
182  */
183 struct bridge_iflist {
184 	LIST_ENTRY(bridge_iflist) bif_next;
185 	struct ifnet		*bif_ifp;	/* member if */
186 	struct bstp_port	bif_stp;	/* STP state */
187 	uint32_t		bif_flags;	/* member if flags */
188 	int			bif_savedcaps;	/* saved capabilities */
189 	uint32_t		bif_addrmax;	/* max # of addresses */
190 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
191 	uint32_t		bif_addrexceeded;/* # of address violations */
192 };
193 
194 /*
195  * Bridge route node.
196  */
197 struct bridge_rtnode {
198 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
199 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
200 	struct bridge_iflist	*brt_dst;	/* destination if */
201 	unsigned long		brt_expire;	/* expiration time */
202 	uint8_t			brt_flags;	/* address flags */
203 	uint8_t			brt_addr[ETHER_ADDR_LEN];
204 	uint16_t		brt_vlan;	/* vlan id */
205 };
206 #define	brt_ifp			brt_dst->bif_ifp
207 
208 /*
209  * Software state for each bridge.
210  */
211 struct bridge_softc {
212 	struct ifnet		*sc_ifp;	/* make this an interface */
213 	LIST_ENTRY(bridge_softc) sc_list;
214 	struct mtx		sc_mtx;
215 	struct cv		sc_cv;
216 	uint32_t		sc_brtmax;	/* max # of addresses */
217 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
218 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
219 	struct callout		sc_brcallout;	/* bridge callout */
220 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
221 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
222 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
223 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
224 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
225 	uint32_t		sc_rthash_key;	/* key for hash */
226 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
227 	struct bstp_state	sc_stp;		/* STP state */
228 	uint32_t		sc_brtexceeded;	/* # of cache drops */
229 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
230 	u_char			sc_defaddr[6];	/* Default MAC address */
231 };
232 
233 static struct mtx 	bridge_list_mtx;
234 eventhandler_tag	bridge_detach_cookie = NULL;
235 
236 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
237 
238 uma_zone_t bridge_rtnode_zone;
239 
240 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
241 static void	bridge_clone_destroy(struct ifnet *);
242 
243 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
244 static void	bridge_mutecaps(struct bridge_softc *);
245 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
246 		    int);
247 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
248 static void	bridge_init(void *);
249 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
250 static void	bridge_stop(struct ifnet *, int);
251 static void	bridge_start(struct ifnet *);
252 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
253 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
254 		    struct rtentry *);
255 static void	bridge_enqueue(struct bridge_softc *, struct ifnet *,
256 		    struct mbuf *);
257 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
258 
259 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
260 		    struct mbuf *m);
261 
262 static void	bridge_timer(void *);
263 
264 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
265 		    struct mbuf *, int);
266 static void	bridge_span(struct bridge_softc *, struct mbuf *);
267 
268 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
269 		    uint16_t, struct bridge_iflist *, int, uint8_t);
270 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
271 		    uint16_t);
272 static void	bridge_rttrim(struct bridge_softc *);
273 static void	bridge_rtage(struct bridge_softc *);
274 static void	bridge_rtflush(struct bridge_softc *, int);
275 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
276 		    uint16_t);
277 
278 static int	bridge_rtable_init(struct bridge_softc *);
279 static void	bridge_rtable_fini(struct bridge_softc *);
280 
281 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
282 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
283 		    const uint8_t *, uint16_t);
284 static int	bridge_rtnode_insert(struct bridge_softc *,
285 		    struct bridge_rtnode *);
286 static void	bridge_rtnode_destroy(struct bridge_softc *,
287 		    struct bridge_rtnode *);
288 static void	bridge_rtable_expire(struct ifnet *, int);
289 static void	bridge_state_change(struct ifnet *, int);
290 
291 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
292 		    const char *name);
293 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
294 		    struct ifnet *ifp);
295 static void	bridge_delete_member(struct bridge_softc *,
296 		    struct bridge_iflist *, int);
297 static void	bridge_delete_span(struct bridge_softc *,
298 		    struct bridge_iflist *);
299 
300 static int	bridge_ioctl_add(struct bridge_softc *, void *);
301 static int	bridge_ioctl_del(struct bridge_softc *, void *);
302 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
303 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
304 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
305 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
306 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
307 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
308 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
309 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
310 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
311 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
312 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
313 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
314 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
315 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
316 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
317 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
318 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
319 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
320 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
321 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
322 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
323 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
324 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
325 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
326 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
327 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
328 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
329 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
330 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
331 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
332 		    int);
333 static int	bridge_ip_checkbasic(struct mbuf **mp);
334 #ifdef INET6
335 static int	bridge_ip6_checkbasic(struct mbuf **mp);
336 #endif /* INET6 */
337 static int	bridge_fragment(struct ifnet *, struct mbuf *,
338 		    struct ether_header *, int, struct llc *);
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 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 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
362     &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
363 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
364     &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
365 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
366     &pfil_bridge, 0, "Packet filter on the bridge interface");
367 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
368     &pfil_member, 0, "Packet filter on the member interface");
369 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
370     &pfil_local_phys, 0,
371     "Packet filter on the physical interface for locally destined packets");
372 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
373     &log_stp, 0, "Log STP state changes");
374 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
375     &bridge_inherit_mac, 0,
376     "Inherit MAC address from the first bridge member");
377 
378 struct bridge_control {
379 	int	(*bc_func)(struct bridge_softc *, void *);
380 	int	bc_argsize;
381 	int	bc_flags;
382 };
383 
384 #define	BC_F_COPYIN		0x01	/* copy arguments in */
385 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
386 #define	BC_F_SUSER		0x04	/* do super-user check */
387 
388 const struct bridge_control bridge_control_table[] = {
389 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
390 	  BC_F_COPYIN|BC_F_SUSER },
391 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
392 	  BC_F_COPYIN|BC_F_SUSER },
393 
394 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
395 	  BC_F_COPYIN|BC_F_COPYOUT },
396 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
397 	  BC_F_COPYIN|BC_F_SUSER },
398 
399 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
400 	  BC_F_COPYIN|BC_F_SUSER },
401 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
402 	  BC_F_COPYOUT },
403 
404 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
405 	  BC_F_COPYIN|BC_F_COPYOUT },
406 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
407 	  BC_F_COPYIN|BC_F_COPYOUT },
408 
409 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
410 	  BC_F_COPYIN|BC_F_SUSER },
411 
412 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
413 	  BC_F_COPYIN|BC_F_SUSER },
414 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
415 	  BC_F_COPYOUT },
416 
417 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
418 	  BC_F_COPYIN|BC_F_SUSER },
419 
420 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
421 	  BC_F_COPYIN|BC_F_SUSER },
422 
423 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
424 	  BC_F_COPYOUT },
425 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
426 	  BC_F_COPYIN|BC_F_SUSER },
427 
428 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
429 	  BC_F_COPYOUT },
430 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
431 	  BC_F_COPYIN|BC_F_SUSER },
432 
433 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
434 	  BC_F_COPYOUT },
435 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
436 	  BC_F_COPYIN|BC_F_SUSER },
437 
438 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
439 	  BC_F_COPYOUT },
440 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
441 	  BC_F_COPYIN|BC_F_SUSER },
442 
443 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
444 	  BC_F_COPYIN|BC_F_SUSER },
445 
446 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
447 	  BC_F_COPYIN|BC_F_SUSER },
448 
449 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
450 	  BC_F_COPYIN|BC_F_SUSER },
451 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
452 	  BC_F_COPYIN|BC_F_SUSER },
453 
454 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
455 	  BC_F_COPYOUT },
456 
457 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
458 	  BC_F_COPYOUT },
459 
460 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
461 	  BC_F_COPYIN|BC_F_COPYOUT },
462 
463 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
464 	  BC_F_COPYIN|BC_F_SUSER },
465 
466 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
467 	  BC_F_COPYIN|BC_F_SUSER },
468 
469 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
470 	  BC_F_COPYIN|BC_F_SUSER },
471 
472 };
473 const int bridge_control_table_size =
474     sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
475 
476 LIST_HEAD(, bridge_softc) bridge_list;
477 
478 IFC_SIMPLE_DECLARE(bridge, 0);
479 
480 static int
481 bridge_modevent(module_t mod, int type, void *data)
482 {
483 
484 	switch (type) {
485 	case MOD_LOAD:
486 		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
487 		if_clone_attach(&bridge_cloner);
488 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
489 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
490 		    UMA_ALIGN_PTR, 0);
491 		LIST_INIT(&bridge_list);
492 		bridge_input_p = bridge_input;
493 		bridge_output_p = bridge_output;
494 		bridge_dn_p = bridge_dummynet;
495 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
496 		    ifnet_departure_event, bridge_ifdetach, NULL,
497 		    EVENTHANDLER_PRI_ANY);
498 		break;
499 	case MOD_UNLOAD:
500 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
501 		    bridge_detach_cookie);
502 		if_clone_detach(&bridge_cloner);
503 		uma_zdestroy(bridge_rtnode_zone);
504 		bridge_input_p = NULL;
505 		bridge_output_p = NULL;
506 		bridge_dn_p = NULL;
507 		mtx_destroy(&bridge_list_mtx);
508 		break;
509 	default:
510 		return (EOPNOTSUPP);
511 	}
512 	return (0);
513 }
514 
515 static moduledata_t bridge_mod = {
516 	"if_bridge",
517 	bridge_modevent,
518 	0
519 };
520 
521 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
522 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
523 
524 /*
525  * handler for net.link.bridge.pfil_ipfw
526  */
527 static int
528 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
529 {
530 	int enable = pfil_ipfw;
531 	int error;
532 
533 	error = sysctl_handle_int(oidp, &enable, 0, req);
534 	enable = (enable) ? 1 : 0;
535 
536 	if (enable != pfil_ipfw) {
537 		pfil_ipfw = enable;
538 
539 		/*
540 		 * Disable pfil so that ipfw doesnt run twice, if the user
541 		 * really wants both then they can re-enable pfil_bridge and/or
542 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
543 		 * layer2 type.
544 		 */
545 		if (pfil_ipfw) {
546 			pfil_onlyip = 0;
547 			pfil_bridge = 0;
548 			pfil_member = 0;
549 		}
550 	}
551 
552 	return (error);
553 }
554 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
555 	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
556 
557 /*
558  * bridge_clone_create:
559  *
560  *	Create a new bridge instance.
561  */
562 static int
563 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
564 {
565 	struct bridge_softc *sc, *sc2;
566 	struct ifnet *bifp, *ifp;
567 	int retry;
568 
569 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
570 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
571 	if (ifp == NULL) {
572 		free(sc, M_DEVBUF);
573 		return (ENOSPC);
574 	}
575 
576 	BRIDGE_LOCK_INIT(sc);
577 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
578 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
579 
580 	/* Initialize our routing table. */
581 	bridge_rtable_init(sc);
582 
583 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
584 
585 	LIST_INIT(&sc->sc_iflist);
586 	LIST_INIT(&sc->sc_spanlist);
587 
588 	ifp->if_softc = sc;
589 	if_initname(ifp, ifc->ifc_name, unit);
590 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
591 	ifp->if_ioctl = bridge_ioctl;
592 	ifp->if_start = bridge_start;
593 	ifp->if_init = bridge_init;
594 	ifp->if_type = IFT_BRIDGE;
595 	IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
596 	ifp->if_snd.ifq_drv_maxlen = ifqmaxlen;
597 	IFQ_SET_READY(&ifp->if_snd);
598 
599 	/*
600 	 * Generate a random ethernet address with a locally administered
601 	 * address.
602 	 *
603 	 * Since we are using random ethernet addresses for the bridge, it is
604 	 * possible that we might have address collisions, so make sure that
605 	 * this hardware address isn't already in use on another bridge.
606 	 */
607 	for (retry = 1; retry != 0;) {
608 		arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
609 		sc->sc_defaddr[0] &= ~1;	/* clear multicast bit */
610 		sc->sc_defaddr[0] |= 2;		/* set the LAA bit */
611 		retry = 0;
612 		mtx_lock(&bridge_list_mtx);
613 		LIST_FOREACH(sc2, &bridge_list, sc_list) {
614 			bifp = sc2->sc_ifp;
615 			if (memcmp(sc->sc_defaddr,
616 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0)
617 				retry = 1;
618 		}
619 		mtx_unlock(&bridge_list_mtx);
620 	}
621 
622 	bstp_attach(&sc->sc_stp, &bridge_ops);
623 	ether_ifattach(ifp, sc->sc_defaddr);
624 	/* Now undo some of the damage... */
625 	ifp->if_baudrate = 0;
626 	ifp->if_type = IFT_BRIDGE;
627 
628 	mtx_lock(&bridge_list_mtx);
629 	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
630 	mtx_unlock(&bridge_list_mtx);
631 
632 	return (0);
633 }
634 
635 /*
636  * bridge_clone_destroy:
637  *
638  *	Destroy a bridge instance.
639  */
640 static void
641 bridge_clone_destroy(struct ifnet *ifp)
642 {
643 	struct bridge_softc *sc = ifp->if_softc;
644 	struct bridge_iflist *bif;
645 
646 	BRIDGE_LOCK(sc);
647 
648 	bridge_stop(ifp, 1);
649 	ifp->if_flags &= ~IFF_UP;
650 
651 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
652 		bridge_delete_member(sc, bif, 0);
653 
654 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
655 		bridge_delete_span(sc, bif);
656 	}
657 
658 	BRIDGE_UNLOCK(sc);
659 
660 	callout_drain(&sc->sc_brcallout);
661 
662 	mtx_lock(&bridge_list_mtx);
663 	LIST_REMOVE(sc, sc_list);
664 	mtx_unlock(&bridge_list_mtx);
665 
666 	bstp_detach(&sc->sc_stp);
667 	ether_ifdetach(ifp);
668 	if_free_type(ifp, IFT_ETHER);
669 
670 	/* Tear down the routing table. */
671 	bridge_rtable_fini(sc);
672 
673 	BRIDGE_LOCK_DESTROY(sc);
674 	free(sc, M_DEVBUF);
675 }
676 
677 /*
678  * bridge_ioctl:
679  *
680  *	Handle a control request from the operator.
681  */
682 static int
683 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
684 {
685 	struct bridge_softc *sc = ifp->if_softc;
686 	struct ifreq *ifr = (struct ifreq *)data;
687 	struct bridge_iflist *bif;
688 	struct thread *td = curthread;
689 	union {
690 		struct ifbreq ifbreq;
691 		struct ifbifconf ifbifconf;
692 		struct ifbareq ifbareq;
693 		struct ifbaconf ifbaconf;
694 		struct ifbrparam ifbrparam;
695 		struct ifbropreq ifbropreq;
696 	} args;
697 	struct ifdrv *ifd = (struct ifdrv *) data;
698 	const struct bridge_control *bc;
699 	int error = 0;
700 
701 	switch (cmd) {
702 
703 	case SIOCADDMULTI:
704 	case SIOCDELMULTI:
705 		break;
706 
707 	case SIOCGDRVSPEC:
708 	case SIOCSDRVSPEC:
709 		if (ifd->ifd_cmd >= bridge_control_table_size) {
710 			error = EINVAL;
711 			break;
712 		}
713 		bc = &bridge_control_table[ifd->ifd_cmd];
714 
715 		if (cmd == SIOCGDRVSPEC &&
716 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
717 			error = EINVAL;
718 			break;
719 		}
720 		else if (cmd == SIOCSDRVSPEC &&
721 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
722 			error = EINVAL;
723 			break;
724 		}
725 
726 		if (bc->bc_flags & BC_F_SUSER) {
727 			error = priv_check(td, PRIV_NET_BRIDGE);
728 			if (error)
729 				break;
730 		}
731 
732 		if (ifd->ifd_len != bc->bc_argsize ||
733 		    ifd->ifd_len > sizeof(args)) {
734 			error = EINVAL;
735 			break;
736 		}
737 
738 		bzero(&args, sizeof(args));
739 		if (bc->bc_flags & BC_F_COPYIN) {
740 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
741 			if (error)
742 				break;
743 		}
744 
745 		BRIDGE_LOCK(sc);
746 		error = (*bc->bc_func)(sc, &args);
747 		BRIDGE_UNLOCK(sc);
748 		if (error)
749 			break;
750 
751 		if (bc->bc_flags & BC_F_COPYOUT)
752 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
753 
754 		break;
755 
756 	case SIOCSIFFLAGS:
757 		if (!(ifp->if_flags & IFF_UP) &&
758 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
759 			/*
760 			 * If interface is marked down and it is running,
761 			 * then stop and disable it.
762 			 */
763 			BRIDGE_LOCK(sc);
764 			bridge_stop(ifp, 1);
765 			BRIDGE_UNLOCK(sc);
766 		} else if ((ifp->if_flags & IFF_UP) &&
767 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
768 			/*
769 			 * If interface is marked up and it is stopped, then
770 			 * start it.
771 			 */
772 			(*ifp->if_init)(sc);
773 		}
774 		break;
775 
776 	case SIOCSIFMTU:
777 		if (ifr->ifr_mtu < 576) {
778 			error = EINVAL;
779 			break;
780 		}
781 		if (LIST_EMPTY(&sc->sc_iflist)) {
782 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
783 			break;
784 		}
785 		BRIDGE_LOCK(sc);
786 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
787 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
788 				log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
789 				    " != %d\n", sc->sc_ifp->if_xname,
790 				    bif->bif_ifp->if_mtu,
791 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
792 				error = EINVAL;
793 				break;
794 			}
795 		}
796 		if (!error)
797 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
798 		BRIDGE_UNLOCK(sc);
799 		break;
800 	default:
801 		/*
802 		 * drop the lock as ether_ioctl() will call bridge_start() and
803 		 * cause the lock to be recursed.
804 		 */
805 		error = ether_ioctl(ifp, cmd, data);
806 		break;
807 	}
808 
809 	return (error);
810 }
811 
812 /*
813  * bridge_mutecaps:
814  *
815  *	Clear or restore unwanted capabilities on the member interface
816  */
817 static void
818 bridge_mutecaps(struct bridge_softc *sc)
819 {
820 	struct bridge_iflist *bif;
821 	int enabled, mask;
822 
823 	/* Initial bitmask of capabilities to test */
824 	mask = BRIDGE_IFCAPS_MASK;
825 
826 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
827 		/* Every member must support it or its disabled */
828 		mask &= bif->bif_savedcaps;
829 	}
830 
831 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
832 		enabled = bif->bif_ifp->if_capenable;
833 		enabled &= ~BRIDGE_IFCAPS_STRIP;
834 		/* strip off mask bits and enable them again if allowed */
835 		enabled &= ~BRIDGE_IFCAPS_MASK;
836 		enabled |= mask;
837 		bridge_set_ifcap(sc, bif, enabled);
838 	}
839 
840 }
841 
842 static void
843 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
844 {
845 	struct ifnet *ifp = bif->bif_ifp;
846 	struct ifreq ifr;
847 	int error;
848 
849 	bzero(&ifr, sizeof(ifr));
850 	ifr.ifr_reqcap = set;
851 
852 	if (ifp->if_capenable != set) {
853 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
854 		if (error)
855 			if_printf(sc->sc_ifp,
856 			    "error setting interface capabilities on %s\n",
857 			    ifp->if_xname);
858 	}
859 }
860 
861 /*
862  * bridge_lookup_member:
863  *
864  *	Lookup a bridge member interface.
865  */
866 static struct bridge_iflist *
867 bridge_lookup_member(struct bridge_softc *sc, const char *name)
868 {
869 	struct bridge_iflist *bif;
870 	struct ifnet *ifp;
871 
872 	BRIDGE_LOCK_ASSERT(sc);
873 
874 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
875 		ifp = bif->bif_ifp;
876 		if (strcmp(ifp->if_xname, name) == 0)
877 			return (bif);
878 	}
879 
880 	return (NULL);
881 }
882 
883 /*
884  * bridge_lookup_member_if:
885  *
886  *	Lookup a bridge member interface by ifnet*.
887  */
888 static struct bridge_iflist *
889 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
890 {
891 	struct bridge_iflist *bif;
892 
893 	BRIDGE_LOCK_ASSERT(sc);
894 
895 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
896 		if (bif->bif_ifp == member_ifp)
897 			return (bif);
898 	}
899 
900 	return (NULL);
901 }
902 
903 /*
904  * bridge_delete_member:
905  *
906  *	Delete the specified member interface.
907  */
908 static void
909 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
910     int gone)
911 {
912 	struct ifnet *ifs = bif->bif_ifp;
913 	struct ifnet *fif = NULL;
914 
915 	BRIDGE_LOCK_ASSERT(sc);
916 
917 	if (bif->bif_flags & IFBIF_STP)
918 		bstp_disable(&bif->bif_stp);
919 
920 	ifs->if_bridge = NULL;
921 	BRIDGE_XLOCK(sc);
922 	LIST_REMOVE(bif, bif_next);
923 	BRIDGE_XDROP(sc);
924 
925 	/*
926 	 * If removing the interface that gave the bridge its mac address, set
927 	 * the mac address of the bridge to the address of the next member, or
928 	 * to its default address if no members are left.
929 	 */
930 	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
931 		if (LIST_EMPTY(&sc->sc_iflist)) {
932 			bcopy(sc->sc_defaddr,
933 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
934 			sc->sc_ifaddr = NULL;
935 		} else {
936 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
937 			bcopy(IF_LLADDR(fif),
938 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
939 			sc->sc_ifaddr = fif;
940 		}
941 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
942 	}
943 
944 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
945 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
946 	KASSERT(bif->bif_addrcnt == 0,
947 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
948 
949 	BRIDGE_UNLOCK(sc);
950 	if (!gone) {
951 		switch (ifs->if_type) {
952 		case IFT_ETHER:
953 		case IFT_L2VLAN:
954 			/*
955 			 * Take the interface out of promiscuous mode.
956 			 */
957 			(void) ifpromisc(ifs, 0);
958 			break;
959 
960 		case IFT_GIF:
961 			break;
962 
963 		default:
964 #ifdef DIAGNOSTIC
965 			panic("bridge_delete_member: impossible");
966 #endif
967 			break;
968 		}
969 		/* reneable any interface capabilities */
970 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
971 	}
972 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
973 	BRIDGE_LOCK(sc);
974 	free(bif, M_DEVBUF);
975 }
976 
977 /*
978  * bridge_delete_span:
979  *
980  *	Delete the specified span interface.
981  */
982 static void
983 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
984 {
985 	BRIDGE_LOCK_ASSERT(sc);
986 
987 	KASSERT(bif->bif_ifp->if_bridge == NULL,
988 	    ("%s: not a span interface", __func__));
989 
990 	LIST_REMOVE(bif, bif_next);
991 	free(bif, M_DEVBUF);
992 }
993 
994 static int
995 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
996 {
997 	struct ifbreq *req = arg;
998 	struct bridge_iflist *bif = NULL;
999 	struct ifnet *ifs;
1000 	int error = 0;
1001 
1002 	ifs = ifunit(req->ifbr_ifsname);
1003 	if (ifs == NULL)
1004 		return (ENOENT);
1005 	if (ifs->if_ioctl == NULL)	/* must be supported */
1006 		return (EINVAL);
1007 
1008 	/* If it's in the span list, it can't be a member. */
1009 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1010 		if (ifs == bif->bif_ifp)
1011 			return (EBUSY);
1012 
1013 	if (ifs->if_bridge == sc)
1014 		return (EEXIST);
1015 
1016 	if (ifs->if_bridge != NULL)
1017 		return (EBUSY);
1018 
1019 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1020 	if (bif == NULL)
1021 		return (ENOMEM);
1022 
1023 	bif->bif_ifp = ifs;
1024 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1025 	bif->bif_savedcaps = ifs->if_capenable;
1026 
1027 	switch (ifs->if_type) {
1028 	case IFT_ETHER:
1029 	case IFT_L2VLAN:
1030 	case IFT_GIF:
1031 		/* permitted interface types */
1032 		break;
1033 	default:
1034 		error = EINVAL;
1035 		goto out;
1036 	}
1037 
1038 	/* Allow the first Ethernet member to define the MTU */
1039 	if (LIST_EMPTY(&sc->sc_iflist))
1040 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1041 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1042 		if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1043 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1044 		error = EINVAL;
1045 		goto out;
1046 	}
1047 
1048 	/*
1049 	 * Assign the interface's MAC address to the bridge if it's the first
1050 	 * member and the MAC address of the bridge has not been changed from
1051 	 * the default randomly generated one.
1052 	 */
1053 	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1054 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1055 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1056 		sc->sc_ifaddr = ifs;
1057 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1058 	}
1059 
1060 	ifs->if_bridge = sc;
1061 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1062 	/*
1063 	 * XXX: XLOCK HERE!?!
1064 	 *
1065 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1066 	 */
1067 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1068 
1069 	/* Set interface capabilities to the intersection set of all members */
1070 	bridge_mutecaps(sc);
1071 
1072 	switch (ifs->if_type) {
1073 	case IFT_ETHER:
1074 	case IFT_L2VLAN:
1075 		/*
1076 		 * Place the interface into promiscuous mode.
1077 		 */
1078 		BRIDGE_UNLOCK(sc);
1079 		error = ifpromisc(ifs, 1);
1080 		BRIDGE_LOCK(sc);
1081 		break;
1082 	}
1083 	if (error)
1084 		bridge_delete_member(sc, bif, 0);
1085 out:
1086 	if (error) {
1087 		if (bif != NULL)
1088 			free(bif, M_DEVBUF);
1089 	}
1090 	return (error);
1091 }
1092 
1093 static int
1094 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1095 {
1096 	struct ifbreq *req = arg;
1097 	struct bridge_iflist *bif;
1098 
1099 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1100 	if (bif == NULL)
1101 		return (ENOENT);
1102 
1103 	bridge_delete_member(sc, bif, 0);
1104 
1105 	return (0);
1106 }
1107 
1108 static int
1109 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1110 {
1111 	struct ifbreq *req = arg;
1112 	struct bridge_iflist *bif;
1113 	struct bstp_port *bp;
1114 
1115 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1116 	if (bif == NULL)
1117 		return (ENOENT);
1118 
1119 	bp = &bif->bif_stp;
1120 	req->ifbr_ifsflags = bif->bif_flags;
1121 	req->ifbr_state = bp->bp_state;
1122 	req->ifbr_priority = bp->bp_priority;
1123 	req->ifbr_path_cost = bp->bp_path_cost;
1124 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1125 	req->ifbr_proto = bp->bp_protover;
1126 	req->ifbr_role = bp->bp_role;
1127 	req->ifbr_stpflags = bp->bp_flags;
1128 	req->ifbr_addrcnt = bif->bif_addrcnt;
1129 	req->ifbr_addrmax = bif->bif_addrmax;
1130 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1131 
1132 	/* Copy STP state options as flags */
1133 	if (bp->bp_operedge)
1134 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1135 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1136 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1137 	if (bp->bp_ptp_link)
1138 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1139 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1140 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1141 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1142 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1143 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1144 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1145 	return (0);
1146 }
1147 
1148 static int
1149 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1150 {
1151 	struct ifbreq *req = arg;
1152 	struct bridge_iflist *bif;
1153 	struct bstp_port *bp;
1154 	int error;
1155 
1156 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1157 	if (bif == NULL)
1158 		return (ENOENT);
1159 	bp = &bif->bif_stp;
1160 
1161 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1162 		/* SPAN is readonly */
1163 		return (EINVAL);
1164 
1165 	if (req->ifbr_ifsflags & IFBIF_STP) {
1166 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1167 			error = bstp_enable(&bif->bif_stp);
1168 			if (error)
1169 				return (error);
1170 		}
1171 	} else {
1172 		if ((bif->bif_flags & IFBIF_STP) != 0)
1173 			bstp_disable(&bif->bif_stp);
1174 	}
1175 
1176 	/* Pass on STP flags */
1177 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1178 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1179 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1180 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1181 
1182 	/* Save the bits relating to the bridge */
1183 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1184 
1185 	return (0);
1186 }
1187 
1188 static int
1189 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1190 {
1191 	struct ifbrparam *param = arg;
1192 
1193 	sc->sc_brtmax = param->ifbrp_csize;
1194 	bridge_rttrim(sc);
1195 
1196 	return (0);
1197 }
1198 
1199 static int
1200 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1201 {
1202 	struct ifbrparam *param = arg;
1203 
1204 	param->ifbrp_csize = sc->sc_brtmax;
1205 
1206 	return (0);
1207 }
1208 
1209 static int
1210 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1211 {
1212 	struct ifbifconf *bifc = arg;
1213 	struct bridge_iflist *bif;
1214 	struct ifbreq breq;
1215 	char *buf, *outbuf;
1216 	int count, buflen, len, error = 0;
1217 
1218 	count = 0;
1219 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1220 		count++;
1221 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1222 		count++;
1223 
1224 	buflen = sizeof(breq) * count;
1225 	if (bifc->ifbic_len == 0) {
1226 		bifc->ifbic_len = buflen;
1227 		return (0);
1228 	}
1229 	BRIDGE_UNLOCK(sc);
1230 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1231 	BRIDGE_LOCK(sc);
1232 
1233 	count = 0;
1234 	buf = outbuf;
1235 	len = min(bifc->ifbic_len, buflen);
1236 	bzero(&breq, sizeof(breq));
1237 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1238 		if (len < sizeof(breq))
1239 			break;
1240 
1241 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1242 		    sizeof(breq.ifbr_ifsname));
1243 		/* Fill in the ifbreq structure */
1244 		error = bridge_ioctl_gifflags(sc, &breq);
1245 		if (error)
1246 			break;
1247 		memcpy(buf, &breq, sizeof(breq));
1248 		count++;
1249 		buf += sizeof(breq);
1250 		len -= sizeof(breq);
1251 	}
1252 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1253 		if (len < sizeof(breq))
1254 			break;
1255 
1256 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1257 		    sizeof(breq.ifbr_ifsname));
1258 		breq.ifbr_ifsflags = bif->bif_flags;
1259 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1260 		memcpy(buf, &breq, sizeof(breq));
1261 		count++;
1262 		buf += sizeof(breq);
1263 		len -= sizeof(breq);
1264 	}
1265 
1266 	BRIDGE_UNLOCK(sc);
1267 	bifc->ifbic_len = sizeof(breq) * count;
1268 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1269 	BRIDGE_LOCK(sc);
1270 	free(outbuf, M_TEMP);
1271 	return (error);
1272 }
1273 
1274 static int
1275 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1276 {
1277 	struct ifbaconf *bac = arg;
1278 	struct bridge_rtnode *brt;
1279 	struct ifbareq bareq;
1280 	char *buf, *outbuf;
1281 	int count, buflen, len, error = 0;
1282 
1283 	if (bac->ifbac_len == 0)
1284 		return (0);
1285 
1286 	count = 0;
1287 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1288 		count++;
1289 	buflen = sizeof(bareq) * count;
1290 
1291 	BRIDGE_UNLOCK(sc);
1292 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1293 	BRIDGE_LOCK(sc);
1294 
1295 	count = 0;
1296 	buf = outbuf;
1297 	len = min(bac->ifbac_len, buflen);
1298 	bzero(&bareq, sizeof(bareq));
1299 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1300 		if (len < sizeof(bareq))
1301 			goto out;
1302 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1303 		    sizeof(bareq.ifba_ifsname));
1304 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1305 		bareq.ifba_vlan = brt->brt_vlan;
1306 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1307 				time_uptime < brt->brt_expire)
1308 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1309 		else
1310 			bareq.ifba_expire = 0;
1311 		bareq.ifba_flags = brt->brt_flags;
1312 
1313 		memcpy(buf, &bareq, sizeof(bareq));
1314 		count++;
1315 		buf += sizeof(bareq);
1316 		len -= sizeof(bareq);
1317 	}
1318 out:
1319 	BRIDGE_UNLOCK(sc);
1320 	bac->ifbac_len = sizeof(bareq) * count;
1321 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1322 	BRIDGE_LOCK(sc);
1323 	free(outbuf, M_TEMP);
1324 	return (error);
1325 }
1326 
1327 static int
1328 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1329 {
1330 	struct ifbareq *req = arg;
1331 	struct bridge_iflist *bif;
1332 	int error;
1333 
1334 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1335 	if (bif == NULL)
1336 		return (ENOENT);
1337 
1338 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1339 	    req->ifba_flags);
1340 
1341 	return (error);
1342 }
1343 
1344 static int
1345 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1346 {
1347 	struct ifbrparam *param = arg;
1348 
1349 	sc->sc_brttimeout = param->ifbrp_ctime;
1350 	return (0);
1351 }
1352 
1353 static int
1354 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1355 {
1356 	struct ifbrparam *param = arg;
1357 
1358 	param->ifbrp_ctime = sc->sc_brttimeout;
1359 	return (0);
1360 }
1361 
1362 static int
1363 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1364 {
1365 	struct ifbareq *req = arg;
1366 
1367 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1368 }
1369 
1370 static int
1371 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1372 {
1373 	struct ifbreq *req = arg;
1374 
1375 	bridge_rtflush(sc, req->ifbr_ifsflags);
1376 	return (0);
1377 }
1378 
1379 static int
1380 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1381 {
1382 	struct ifbrparam *param = arg;
1383 	struct bstp_state *bs = &sc->sc_stp;
1384 
1385 	param->ifbrp_prio = bs->bs_bridge_priority;
1386 	return (0);
1387 }
1388 
1389 static int
1390 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1391 {
1392 	struct ifbrparam *param = arg;
1393 
1394 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1395 }
1396 
1397 static int
1398 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1399 {
1400 	struct ifbrparam *param = arg;
1401 	struct bstp_state *bs = &sc->sc_stp;
1402 
1403 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1404 	return (0);
1405 }
1406 
1407 static int
1408 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1409 {
1410 	struct ifbrparam *param = arg;
1411 
1412 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1413 }
1414 
1415 static int
1416 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1417 {
1418 	struct ifbrparam *param = arg;
1419 	struct bstp_state *bs = &sc->sc_stp;
1420 
1421 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1422 	return (0);
1423 }
1424 
1425 static int
1426 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1427 {
1428 	struct ifbrparam *param = arg;
1429 
1430 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1431 }
1432 
1433 static int
1434 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1435 {
1436 	struct ifbrparam *param = arg;
1437 	struct bstp_state *bs = &sc->sc_stp;
1438 
1439 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1440 	return (0);
1441 }
1442 
1443 static int
1444 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1445 {
1446 	struct ifbrparam *param = arg;
1447 
1448 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1449 }
1450 
1451 static int
1452 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1453 {
1454 	struct ifbreq *req = arg;
1455 	struct bridge_iflist *bif;
1456 
1457 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1458 	if (bif == NULL)
1459 		return (ENOENT);
1460 
1461 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1462 }
1463 
1464 static int
1465 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1466 {
1467 	struct ifbreq *req = arg;
1468 	struct bridge_iflist *bif;
1469 
1470 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1471 	if (bif == NULL)
1472 		return (ENOENT);
1473 
1474 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1475 }
1476 
1477 static int
1478 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1479 {
1480 	struct ifbreq *req = arg;
1481 	struct bridge_iflist *bif;
1482 
1483 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1484 	if (bif == NULL)
1485 		return (ENOENT);
1486 
1487 	bif->bif_addrmax = req->ifbr_addrmax;
1488 	return (0);
1489 }
1490 
1491 static int
1492 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1493 {
1494 	struct ifbreq *req = arg;
1495 	struct bridge_iflist *bif = NULL;
1496 	struct ifnet *ifs;
1497 
1498 	ifs = ifunit(req->ifbr_ifsname);
1499 	if (ifs == NULL)
1500 		return (ENOENT);
1501 
1502 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1503 		if (ifs == bif->bif_ifp)
1504 			return (EBUSY);
1505 
1506 	if (ifs->if_bridge != NULL)
1507 		return (EBUSY);
1508 
1509 	switch (ifs->if_type) {
1510 		case IFT_ETHER:
1511 		case IFT_GIF:
1512 		case IFT_L2VLAN:
1513 			break;
1514 		default:
1515 			return (EINVAL);
1516 	}
1517 
1518 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1519 	if (bif == NULL)
1520 		return (ENOMEM);
1521 
1522 	bif->bif_ifp = ifs;
1523 	bif->bif_flags = IFBIF_SPAN;
1524 
1525 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1526 
1527 	return (0);
1528 }
1529 
1530 static int
1531 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1532 {
1533 	struct ifbreq *req = arg;
1534 	struct bridge_iflist *bif;
1535 	struct ifnet *ifs;
1536 
1537 	ifs = ifunit(req->ifbr_ifsname);
1538 	if (ifs == NULL)
1539 		return (ENOENT);
1540 
1541 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1542 		if (ifs == bif->bif_ifp)
1543 			break;
1544 
1545 	if (bif == NULL)
1546 		return (ENOENT);
1547 
1548 	bridge_delete_span(sc, bif);
1549 
1550 	return (0);
1551 }
1552 
1553 static int
1554 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1555 {
1556 	struct ifbropreq *req = arg;
1557 	struct bstp_state *bs = &sc->sc_stp;
1558 	struct bstp_port *root_port;
1559 
1560 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1561 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1562 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1563 
1564 	root_port = bs->bs_root_port;
1565 	if (root_port == NULL)
1566 		req->ifbop_root_port = 0;
1567 	else
1568 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1569 
1570 	req->ifbop_holdcount = bs->bs_txholdcount;
1571 	req->ifbop_priority = bs->bs_bridge_priority;
1572 	req->ifbop_protocol = bs->bs_protover;
1573 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1574 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1575 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1576 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1577 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1578 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1579 
1580 	return (0);
1581 }
1582 
1583 static int
1584 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1585 {
1586 	struct ifbrparam *param = arg;
1587 
1588 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1589 	return (0);
1590 }
1591 
1592 static int
1593 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1594 {
1595 	struct ifbpstpconf *bifstp = arg;
1596 	struct bridge_iflist *bif;
1597 	struct bstp_port *bp;
1598 	struct ifbpstpreq bpreq;
1599 	char *buf, *outbuf;
1600 	int count, buflen, len, error = 0;
1601 
1602 	count = 0;
1603 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1604 		if ((bif->bif_flags & IFBIF_STP) != 0)
1605 			count++;
1606 	}
1607 
1608 	buflen = sizeof(bpreq) * count;
1609 	if (bifstp->ifbpstp_len == 0) {
1610 		bifstp->ifbpstp_len = buflen;
1611 		return (0);
1612 	}
1613 
1614 	BRIDGE_UNLOCK(sc);
1615 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1616 	BRIDGE_LOCK(sc);
1617 
1618 	count = 0;
1619 	buf = outbuf;
1620 	len = min(bifstp->ifbpstp_len, buflen);
1621 	bzero(&bpreq, sizeof(bpreq));
1622 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1623 		if (len < sizeof(bpreq))
1624 			break;
1625 
1626 		if ((bif->bif_flags & IFBIF_STP) == 0)
1627 			continue;
1628 
1629 		bp = &bif->bif_stp;
1630 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1631 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1632 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1633 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1634 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1635 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1636 
1637 		memcpy(buf, &bpreq, sizeof(bpreq));
1638 		count++;
1639 		buf += sizeof(bpreq);
1640 		len -= sizeof(bpreq);
1641 	}
1642 
1643 	BRIDGE_UNLOCK(sc);
1644 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1645 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1646 	BRIDGE_LOCK(sc);
1647 	free(outbuf, M_TEMP);
1648 	return (error);
1649 }
1650 
1651 static int
1652 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1653 {
1654 	struct ifbrparam *param = arg;
1655 
1656 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1657 }
1658 
1659 static int
1660 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1661 {
1662 	struct ifbrparam *param = arg;
1663 
1664 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1665 }
1666 
1667 /*
1668  * bridge_ifdetach:
1669  *
1670  *	Detach an interface from a bridge.  Called when a member
1671  *	interface is detaching.
1672  */
1673 static void
1674 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1675 {
1676 	struct bridge_softc *sc = ifp->if_bridge;
1677 	struct bridge_iflist *bif;
1678 
1679 	/* Check if the interface is a bridge member */
1680 	if (sc != NULL) {
1681 		BRIDGE_LOCK(sc);
1682 
1683 		bif = bridge_lookup_member_if(sc, ifp);
1684 		if (bif != NULL)
1685 			bridge_delete_member(sc, bif, 1);
1686 
1687 		BRIDGE_UNLOCK(sc);
1688 		return;
1689 	}
1690 
1691 	/* Check if the interface is a span port */
1692 	mtx_lock(&bridge_list_mtx);
1693 	LIST_FOREACH(sc, &bridge_list, sc_list) {
1694 		BRIDGE_LOCK(sc);
1695 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1696 			if (ifp == bif->bif_ifp) {
1697 				bridge_delete_span(sc, bif);
1698 				break;
1699 			}
1700 
1701 		BRIDGE_UNLOCK(sc);
1702 	}
1703 	mtx_unlock(&bridge_list_mtx);
1704 }
1705 
1706 /*
1707  * bridge_init:
1708  *
1709  *	Initialize a bridge interface.
1710  */
1711 static void
1712 bridge_init(void *xsc)
1713 {
1714 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1715 	struct ifnet *ifp = sc->sc_ifp;
1716 
1717 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1718 		return;
1719 
1720 	BRIDGE_LOCK(sc);
1721 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1722 	    bridge_timer, sc);
1723 
1724 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1725 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1726 
1727 	BRIDGE_UNLOCK(sc);
1728 }
1729 
1730 /*
1731  * bridge_stop:
1732  *
1733  *	Stop the bridge interface.
1734  */
1735 static void
1736 bridge_stop(struct ifnet *ifp, int disable)
1737 {
1738 	struct bridge_softc *sc = ifp->if_softc;
1739 
1740 	BRIDGE_LOCK_ASSERT(sc);
1741 
1742 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1743 		return;
1744 
1745 	callout_stop(&sc->sc_brcallout);
1746 	bstp_stop(&sc->sc_stp);
1747 
1748 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1749 
1750 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1751 }
1752 
1753 /*
1754  * bridge_enqueue:
1755  *
1756  *	Enqueue a packet on a bridge member interface.
1757  *
1758  */
1759 static void
1760 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1761 {
1762 	int len, err = 0;
1763 	short mflags;
1764 	struct mbuf *m0;
1765 
1766 	len = m->m_pkthdr.len;
1767 	mflags = m->m_flags;
1768 
1769 	/* We may be sending a fragment so traverse the mbuf */
1770 	for (; m; m = m0) {
1771 		m0 = m->m_nextpkt;
1772 		m->m_nextpkt = NULL;
1773 
1774 		/*
1775 		 * If underlying interface can not do VLAN tag insertion itself
1776 		 * then attach a packet tag that holds it.
1777 		 */
1778 		if ((m->m_flags & M_VLANTAG) &&
1779 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1780 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1781 			if (m == NULL) {
1782 				if_printf(dst_ifp,
1783 				    "unable to prepend VLAN header\n");
1784 				dst_ifp->if_oerrors++;
1785 				continue;
1786 			}
1787 			m->m_flags &= ~M_VLANTAG;
1788 		}
1789 
1790 		if (err == 0)
1791 			dst_ifp->if_transmit(dst_ifp, m);
1792 	}
1793 
1794 	if (err == 0) {
1795 		sc->sc_ifp->if_opackets++;
1796 		sc->sc_ifp->if_obytes += len;
1797 		if (mflags & M_MCAST)
1798 			sc->sc_ifp->if_omcasts++;
1799 	}
1800 }
1801 
1802 /*
1803  * bridge_dummynet:
1804  *
1805  * 	Receive a queued packet from dummynet and pass it on to the output
1806  * 	interface.
1807  *
1808  *	The mbuf has the Ethernet header already attached.
1809  */
1810 static void
1811 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1812 {
1813 	struct bridge_softc *sc;
1814 
1815 	sc = ifp->if_bridge;
1816 
1817 	/*
1818 	 * The packet didnt originate from a member interface. This should only
1819 	 * ever happen if a member interface is removed while packets are
1820 	 * queued for it.
1821 	 */
1822 	if (sc == NULL) {
1823 		m_freem(m);
1824 		return;
1825 	}
1826 
1827 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1828 #ifdef INET6
1829 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1830 #endif
1831 	    ) {
1832 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1833 			return;
1834 		if (m == NULL)
1835 			return;
1836 	}
1837 
1838 	bridge_enqueue(sc, ifp, m);
1839 }
1840 
1841 /*
1842  * bridge_output:
1843  *
1844  *	Send output from a bridge member interface.  This
1845  *	performs the bridging function for locally originated
1846  *	packets.
1847  *
1848  *	The mbuf has the Ethernet header already attached.  We must
1849  *	enqueue or free the mbuf before returning.
1850  */
1851 static int
1852 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1853     struct rtentry *rt)
1854 {
1855 	struct ether_header *eh;
1856 	struct ifnet *dst_if;
1857 	struct bridge_softc *sc;
1858 	uint16_t vlan;
1859 
1860 	if (m->m_len < ETHER_HDR_LEN) {
1861 		m = m_pullup(m, ETHER_HDR_LEN);
1862 		if (m == NULL)
1863 			return (0);
1864 	}
1865 
1866 	eh = mtod(m, struct ether_header *);
1867 	sc = ifp->if_bridge;
1868 	vlan = VLANTAGOF(m);
1869 
1870 	BRIDGE_LOCK(sc);
1871 
1872 	/*
1873 	 * If bridge is down, but the original output interface is up,
1874 	 * go ahead and send out that interface.  Otherwise, the packet
1875 	 * is dropped below.
1876 	 */
1877 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1878 		dst_if = ifp;
1879 		goto sendunicast;
1880 	}
1881 
1882 	/*
1883 	 * If the packet is a multicast, or we don't know a better way to
1884 	 * get there, send to all interfaces.
1885 	 */
1886 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1887 		dst_if = NULL;
1888 	else
1889 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1890 	if (dst_if == NULL) {
1891 		struct bridge_iflist *bif;
1892 		struct mbuf *mc;
1893 		int error = 0, used = 0;
1894 
1895 		bridge_span(sc, m);
1896 
1897 		BRIDGE_LOCK2REF(sc, error);
1898 		if (error) {
1899 			m_freem(m);
1900 			return (0);
1901 		}
1902 
1903 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1904 			dst_if = bif->bif_ifp;
1905 
1906 			if (dst_if->if_type == IFT_GIF)
1907 				continue;
1908 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1909 				continue;
1910 
1911 			/*
1912 			 * If this is not the original output interface,
1913 			 * and the interface is participating in spanning
1914 			 * tree, make sure the port is in a state that
1915 			 * allows forwarding.
1916 			 */
1917 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1918 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1919 				continue;
1920 
1921 			if (LIST_NEXT(bif, bif_next) == NULL) {
1922 				used = 1;
1923 				mc = m;
1924 			} else {
1925 				mc = m_copypacket(m, M_DONTWAIT);
1926 				if (mc == NULL) {
1927 					sc->sc_ifp->if_oerrors++;
1928 					continue;
1929 				}
1930 			}
1931 
1932 			bridge_enqueue(sc, dst_if, mc);
1933 		}
1934 		if (used == 0)
1935 			m_freem(m);
1936 		BRIDGE_UNREF(sc);
1937 		return (0);
1938 	}
1939 
1940 sendunicast:
1941 	/*
1942 	 * XXX Spanning tree consideration here?
1943 	 */
1944 
1945 	bridge_span(sc, m);
1946 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1947 		m_freem(m);
1948 		BRIDGE_UNLOCK(sc);
1949 		return (0);
1950 	}
1951 
1952 	BRIDGE_UNLOCK(sc);
1953 	bridge_enqueue(sc, dst_if, m);
1954 	return (0);
1955 }
1956 
1957 /*
1958  * bridge_start:
1959  *
1960  *	Start output on a bridge.
1961  *
1962  */
1963 static void
1964 bridge_start(struct ifnet *ifp)
1965 {
1966 	struct bridge_softc *sc;
1967 	struct mbuf *m;
1968 	struct ether_header *eh;
1969 	struct ifnet *dst_if;
1970 
1971 	sc = ifp->if_softc;
1972 
1973 	ifp->if_drv_flags |= IFF_DRV_OACTIVE;
1974 	for (;;) {
1975 		IFQ_DEQUEUE(&ifp->if_snd, m);
1976 		if (m == 0)
1977 			break;
1978 		ETHER_BPF_MTAP(ifp, m);
1979 
1980 		eh = mtod(m, struct ether_header *);
1981 		dst_if = NULL;
1982 
1983 		BRIDGE_LOCK(sc);
1984 		if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
1985 			dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1);
1986 		}
1987 
1988 		if (dst_if == NULL)
1989 			bridge_broadcast(sc, ifp, m, 0);
1990 		else {
1991 			BRIDGE_UNLOCK(sc);
1992 			bridge_enqueue(sc, dst_if, m);
1993 		}
1994 	}
1995 	ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
1996 }
1997 
1998 /*
1999  * bridge_forward:
2000  *
2001  *	The forwarding function of the bridge.
2002  *
2003  *	NOTE: Releases the lock on return.
2004  */
2005 static void
2006 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2007     struct mbuf *m)
2008 {
2009 	struct bridge_iflist *dbif;
2010 	struct ifnet *src_if, *dst_if, *ifp;
2011 	struct ether_header *eh;
2012 	uint16_t vlan;
2013 	uint8_t *dst;
2014 	int error;
2015 
2016 	src_if = m->m_pkthdr.rcvif;
2017 	ifp = sc->sc_ifp;
2018 
2019 	ifp->if_ipackets++;
2020 	ifp->if_ibytes += m->m_pkthdr.len;
2021 	vlan = VLANTAGOF(m);
2022 
2023 	if ((sbif->bif_flags & IFBIF_STP) &&
2024 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2025 		goto drop;
2026 
2027 	eh = mtod(m, struct ether_header *);
2028 	dst = eh->ether_dhost;
2029 
2030 	/* If the interface is learning, record the address. */
2031 	if (sbif->bif_flags & IFBIF_LEARNING) {
2032 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2033 		    sbif, 0, IFBAF_DYNAMIC);
2034 		/*
2035 		 * If the interface has addresses limits then deny any source
2036 		 * that is not in the cache.
2037 		 */
2038 		if (error && sbif->bif_addrmax)
2039 			goto drop;
2040 	}
2041 
2042 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2043 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2044 		goto drop;
2045 
2046 	/*
2047 	 * At this point, the port either doesn't participate
2048 	 * in spanning tree or it is in the forwarding state.
2049 	 */
2050 
2051 	/*
2052 	 * If the packet is unicast, destined for someone on
2053 	 * "this" side of the bridge, drop it.
2054 	 */
2055 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2056 		dst_if = bridge_rtlookup(sc, dst, vlan);
2057 		if (src_if == dst_if)
2058 			goto drop;
2059 	} else {
2060 		/*
2061 		 * Check if its a reserved multicast address, any address
2062 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2063 		 * bridge.
2064 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2065 		 */
2066 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2067 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2068 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2069 			goto drop;
2070 
2071 		/* ...forward it to all interfaces. */
2072 		ifp->if_imcasts++;
2073 		dst_if = NULL;
2074 	}
2075 
2076 	/*
2077 	 * If we have a destination interface which is a member of our bridge,
2078 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2079 	 * For broadcast or multicast packets, don't bother because it will
2080 	 * be reinjected into ether_input. We do this before we pass the packets
2081 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2082 	 * drop the packet, or possibly modify it, making it difficult to debug
2083 	 * firewall issues on the bridge.
2084 	 */
2085 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2086 		ETHER_BPF_MTAP(ifp, m);
2087 
2088 	/* run the packet filter */
2089 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2090 #ifdef INET6
2091 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2092 #endif
2093 	    ) {
2094 		BRIDGE_UNLOCK(sc);
2095 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2096 			return;
2097 		if (m == NULL)
2098 			return;
2099 		BRIDGE_LOCK(sc);
2100 	}
2101 
2102 	if (dst_if == NULL) {
2103 		bridge_broadcast(sc, src_if, m, 1);
2104 		return;
2105 	}
2106 
2107 	/*
2108 	 * At this point, we're dealing with a unicast frame
2109 	 * going to a different interface.
2110 	 */
2111 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2112 		goto drop;
2113 
2114 	dbif = bridge_lookup_member_if(sc, dst_if);
2115 	if (dbif == NULL)
2116 		/* Not a member of the bridge (anymore?) */
2117 		goto drop;
2118 
2119 	/* Private segments can not talk to each other */
2120 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2121 		goto drop;
2122 
2123 	if ((dbif->bif_flags & IFBIF_STP) &&
2124 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2125 		goto drop;
2126 
2127 	BRIDGE_UNLOCK(sc);
2128 
2129 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2130 #ifdef INET6
2131 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2132 #endif
2133 	    ) {
2134 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2135 			return;
2136 		if (m == NULL)
2137 			return;
2138 	}
2139 
2140 	bridge_enqueue(sc, dst_if, m);
2141 	return;
2142 
2143 drop:
2144 	BRIDGE_UNLOCK(sc);
2145 	m_freem(m);
2146 }
2147 
2148 /*
2149  * bridge_input:
2150  *
2151  *	Receive input from a member interface.  Queue the packet for
2152  *	bridging if it is not for us.
2153  */
2154 static struct mbuf *
2155 bridge_input(struct ifnet *ifp, struct mbuf *m)
2156 {
2157 	struct bridge_softc *sc = ifp->if_bridge;
2158 	struct bridge_iflist *bif, *bif2;
2159 	struct ifnet *bifp;
2160 	struct ether_header *eh;
2161 	struct mbuf *mc, *mc2;
2162 	uint16_t vlan;
2163 	int error;
2164 
2165 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2166 		return (m);
2167 
2168 	bifp = sc->sc_ifp;
2169 	vlan = VLANTAGOF(m);
2170 
2171 	/*
2172 	 * Implement support for bridge monitoring. If this flag has been
2173 	 * set on this interface, discard the packet once we push it through
2174 	 * the bpf(4) machinery, but before we do, increment the byte and
2175 	 * packet counters associated with this interface.
2176 	 */
2177 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2178 		m->m_pkthdr.rcvif  = bifp;
2179 		ETHER_BPF_MTAP(bifp, m);
2180 		bifp->if_ipackets++;
2181 		bifp->if_ibytes += m->m_pkthdr.len;
2182 		m_freem(m);
2183 		return (NULL);
2184 	}
2185 	BRIDGE_LOCK(sc);
2186 	bif = bridge_lookup_member_if(sc, ifp);
2187 	if (bif == NULL) {
2188 		BRIDGE_UNLOCK(sc);
2189 		return (m);
2190 	}
2191 
2192 	eh = mtod(m, struct ether_header *);
2193 
2194 	bridge_span(sc, m);
2195 
2196 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2197 		/* Tap off 802.1D packets; they do not get forwarded. */
2198 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2199 		    ETHER_ADDR_LEN) == 0) {
2200 			m = bstp_input(&bif->bif_stp, ifp, m);
2201 			if (m == NULL) {
2202 				BRIDGE_UNLOCK(sc);
2203 				return (NULL);
2204 			}
2205 		}
2206 
2207 		if ((bif->bif_flags & IFBIF_STP) &&
2208 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2209 			BRIDGE_UNLOCK(sc);
2210 			return (m);
2211 		}
2212 
2213 		/*
2214 		 * Make a deep copy of the packet and enqueue the copy
2215 		 * for bridge processing; return the original packet for
2216 		 * local processing.
2217 		 */
2218 		mc = m_dup(m, M_DONTWAIT);
2219 		if (mc == NULL) {
2220 			BRIDGE_UNLOCK(sc);
2221 			return (m);
2222 		}
2223 
2224 		/* Perform the bridge forwarding function with the copy. */
2225 		bridge_forward(sc, bif, mc);
2226 
2227 		/*
2228 		 * Reinject the mbuf as arriving on the bridge so we have a
2229 		 * chance at claiming multicast packets. We can not loop back
2230 		 * here from ether_input as a bridge is never a member of a
2231 		 * bridge.
2232 		 */
2233 		KASSERT(bifp->if_bridge == NULL,
2234 		    ("loop created in bridge_input"));
2235 		mc2 = m_dup(m, M_DONTWAIT);
2236 		if (mc2 != NULL) {
2237 			/* Keep the layer3 header aligned */
2238 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2239 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2240 		}
2241 		if (mc2 != NULL) {
2242 			mc2->m_pkthdr.rcvif = bifp;
2243 			(*bifp->if_input)(bifp, mc2);
2244 		}
2245 
2246 		/* Return the original packet for local processing. */
2247 		return (m);
2248 	}
2249 
2250 	if ((bif->bif_flags & IFBIF_STP) &&
2251 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2252 		BRIDGE_UNLOCK(sc);
2253 		return (m);
2254 	}
2255 
2256 #if (defined(INET) || defined(INET6)) && defined(DEV_CARP)
2257 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2258 	|| ((iface)->if_carp \
2259 	    && carp_forus((iface)->if_carp, eh->ether_dhost))
2260 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2261 	|| ((iface)->if_carp \
2262 	    && carp_forus((iface)->if_carp, eh->ether_shost))
2263 #else
2264 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2265 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2266 #endif
2267 
2268 #ifdef INET6
2269 #   define OR_PFIL_HOOKED_INET6 \
2270 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2271 #else
2272 #   define OR_PFIL_HOOKED_INET6
2273 #endif
2274 
2275 #define GRAB_OUR_PACKETS(iface) \
2276 	if ((iface)->if_type == IFT_GIF) \
2277 		continue; \
2278 	/* It is destined for us. */ \
2279 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2280 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2281 	    ) {								\
2282 		if ((iface)->if_type == IFT_BRIDGE) {			\
2283 			ETHER_BPF_MTAP(iface, m);			\
2284 			iface->if_ipackets++;				\
2285 			/* Filter on the physical interface. */		\
2286 			if (pfil_local_phys &&				\
2287 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2288 			     OR_PFIL_HOOKED_INET6)) {			\
2289 				if (bridge_pfil(&m, NULL, ifp,		\
2290 				    PFIL_IN) != 0 || m == NULL) {	\
2291 					BRIDGE_UNLOCK(sc);		\
2292 					return (NULL);			\
2293 				}					\
2294 			}						\
2295 		}							\
2296 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2297 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2298 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2299 			if (error && bif->bif_addrmax) {		\
2300 				BRIDGE_UNLOCK(sc);			\
2301 				m_freem(m);				\
2302 				return (NULL);				\
2303 			}						\
2304 		}							\
2305 		m->m_pkthdr.rcvif = iface;				\
2306 		BRIDGE_UNLOCK(sc);					\
2307 		return (m);						\
2308 	}								\
2309 									\
2310 	/* We just received a packet that we sent out. */		\
2311 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2312 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2313 	    ) {								\
2314 		BRIDGE_UNLOCK(sc);					\
2315 		m_freem(m);						\
2316 		return (NULL);						\
2317 	}
2318 
2319 	/*
2320 	 * Unicast.  Make sure it's not for the bridge.
2321 	 */
2322 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2323 
2324 	/*
2325 	 * Give a chance for ifp at first priority. This will help when	the
2326 	 * packet comes through the interface like VLAN's with the same MACs
2327 	 * on several interfaces from the same bridge. This also will save
2328 	 * some CPU cycles in case the destination interface and the input
2329 	 * interface (eq ifp) are the same.
2330 	 */
2331 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2332 
2333 	/* Now check the all bridge members. */
2334 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2335 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2336 	}
2337 
2338 #undef OR_CARP_CHECK_WE_ARE_DST
2339 #undef OR_CARP_CHECK_WE_ARE_SRC
2340 #undef OR_PFIL_HOOKED_INET6
2341 #undef GRAB_OUR_PACKETS
2342 
2343 	/* Perform the bridge forwarding function. */
2344 	bridge_forward(sc, bif, m);
2345 
2346 	return (NULL);
2347 }
2348 
2349 /*
2350  * bridge_broadcast:
2351  *
2352  *	Send a frame to all interfaces that are members of
2353  *	the bridge, except for the one on which the packet
2354  *	arrived.
2355  *
2356  *	NOTE: Releases the lock on return.
2357  */
2358 static void
2359 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2360     struct mbuf *m, int runfilt)
2361 {
2362 	struct bridge_iflist *dbif, *sbif;
2363 	struct mbuf *mc;
2364 	struct ifnet *dst_if;
2365 	int error = 0, used = 0, i;
2366 
2367 	sbif = bridge_lookup_member_if(sc, src_if);
2368 
2369 	BRIDGE_LOCK2REF(sc, error);
2370 	if (error) {
2371 		m_freem(m);
2372 		return;
2373 	}
2374 
2375 	/* Filter on the bridge interface before broadcasting */
2376 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2377 #ifdef INET6
2378 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2379 #endif
2380 	    )) {
2381 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2382 			goto out;
2383 		if (m == NULL)
2384 			goto out;
2385 	}
2386 
2387 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2388 		dst_if = dbif->bif_ifp;
2389 		if (dst_if == src_if)
2390 			continue;
2391 
2392 		/* Private segments can not talk to each other */
2393 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2394 			continue;
2395 
2396 		if ((dbif->bif_flags & IFBIF_STP) &&
2397 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2398 			continue;
2399 
2400 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2401 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2402 			continue;
2403 
2404 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2405 			continue;
2406 
2407 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2408 			mc = m;
2409 			used = 1;
2410 		} else {
2411 			mc = m_dup(m, M_DONTWAIT);
2412 			if (mc == NULL) {
2413 				sc->sc_ifp->if_oerrors++;
2414 				continue;
2415 			}
2416 		}
2417 
2418 		/*
2419 		 * Filter on the output interface. Pass a NULL bridge interface
2420 		 * pointer so we do not redundantly filter on the bridge for
2421 		 * each interface we broadcast on.
2422 		 */
2423 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2424 #ifdef INET6
2425 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2426 #endif
2427 		    )) {
2428 			if (used == 0) {
2429 				/* Keep the layer3 header aligned */
2430 				i = min(mc->m_pkthdr.len, max_protohdr);
2431 				mc = m_copyup(mc, i, ETHER_ALIGN);
2432 				if (mc == NULL) {
2433 					sc->sc_ifp->if_oerrors++;
2434 					continue;
2435 				}
2436 			}
2437 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2438 				continue;
2439 			if (mc == NULL)
2440 				continue;
2441 		}
2442 
2443 		bridge_enqueue(sc, dst_if, mc);
2444 	}
2445 	if (used == 0)
2446 		m_freem(m);
2447 
2448 out:
2449 	BRIDGE_UNREF(sc);
2450 }
2451 
2452 /*
2453  * bridge_span:
2454  *
2455  *	Duplicate a packet out one or more interfaces that are in span mode,
2456  *	the original mbuf is unmodified.
2457  */
2458 static void
2459 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2460 {
2461 	struct bridge_iflist *bif;
2462 	struct ifnet *dst_if;
2463 	struct mbuf *mc;
2464 
2465 	if (LIST_EMPTY(&sc->sc_spanlist))
2466 		return;
2467 
2468 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2469 		dst_if = bif->bif_ifp;
2470 
2471 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2472 			continue;
2473 
2474 		mc = m_copypacket(m, M_DONTWAIT);
2475 		if (mc == NULL) {
2476 			sc->sc_ifp->if_oerrors++;
2477 			continue;
2478 		}
2479 
2480 		bridge_enqueue(sc, dst_if, mc);
2481 	}
2482 }
2483 
2484 /*
2485  * bridge_rtupdate:
2486  *
2487  *	Add a bridge routing entry.
2488  */
2489 static int
2490 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2491     struct bridge_iflist *bif, int setflags, uint8_t flags)
2492 {
2493 	struct bridge_rtnode *brt;
2494 	int error;
2495 
2496 	BRIDGE_LOCK_ASSERT(sc);
2497 
2498 	/* Check the source address is valid and not multicast. */
2499 	if (ETHER_IS_MULTICAST(dst) ||
2500 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2501 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2502 		return (EINVAL);
2503 
2504 	/* 802.1p frames map to vlan 1 */
2505 	if (vlan == 0)
2506 		vlan = 1;
2507 
2508 	/*
2509 	 * A route for this destination might already exist.  If so,
2510 	 * update it, otherwise create a new one.
2511 	 */
2512 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2513 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2514 			sc->sc_brtexceeded++;
2515 			return (ENOSPC);
2516 		}
2517 		/* Check per interface address limits (if enabled) */
2518 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2519 			bif->bif_addrexceeded++;
2520 			return (ENOSPC);
2521 		}
2522 
2523 		/*
2524 		 * Allocate a new bridge forwarding node, and
2525 		 * initialize the expiration time and Ethernet
2526 		 * address.
2527 		 */
2528 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2529 		if (brt == NULL)
2530 			return (ENOMEM);
2531 
2532 		if (bif->bif_flags & IFBIF_STICKY)
2533 			brt->brt_flags = IFBAF_STICKY;
2534 		else
2535 			brt->brt_flags = IFBAF_DYNAMIC;
2536 
2537 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2538 		brt->brt_vlan = vlan;
2539 
2540 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2541 			uma_zfree(bridge_rtnode_zone, brt);
2542 			return (error);
2543 		}
2544 		brt->brt_dst = bif;
2545 		bif->bif_addrcnt++;
2546 	}
2547 
2548 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2549 	    brt->brt_dst != bif) {
2550 		brt->brt_dst->bif_addrcnt--;
2551 		brt->brt_dst = bif;
2552 		brt->brt_dst->bif_addrcnt++;
2553 	}
2554 
2555 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2556 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2557 	if (setflags)
2558 		brt->brt_flags = flags;
2559 
2560 	return (0);
2561 }
2562 
2563 /*
2564  * bridge_rtlookup:
2565  *
2566  *	Lookup the destination interface for an address.
2567  */
2568 static struct ifnet *
2569 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2570 {
2571 	struct bridge_rtnode *brt;
2572 
2573 	BRIDGE_LOCK_ASSERT(sc);
2574 
2575 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2576 		return (NULL);
2577 
2578 	return (brt->brt_ifp);
2579 }
2580 
2581 /*
2582  * bridge_rttrim:
2583  *
2584  *	Trim the routine table so that we have a number
2585  *	of routing entries less than or equal to the
2586  *	maximum number.
2587  */
2588 static void
2589 bridge_rttrim(struct bridge_softc *sc)
2590 {
2591 	struct bridge_rtnode *brt, *nbrt;
2592 
2593 	BRIDGE_LOCK_ASSERT(sc);
2594 
2595 	/* Make sure we actually need to do this. */
2596 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2597 		return;
2598 
2599 	/* Force an aging cycle; this might trim enough addresses. */
2600 	bridge_rtage(sc);
2601 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2602 		return;
2603 
2604 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2605 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2606 			bridge_rtnode_destroy(sc, brt);
2607 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2608 				return;
2609 		}
2610 	}
2611 }
2612 
2613 /*
2614  * bridge_timer:
2615  *
2616  *	Aging timer for the bridge.
2617  */
2618 static void
2619 bridge_timer(void *arg)
2620 {
2621 	struct bridge_softc *sc = arg;
2622 
2623 	BRIDGE_LOCK_ASSERT(sc);
2624 
2625 	bridge_rtage(sc);
2626 
2627 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2628 		callout_reset(&sc->sc_brcallout,
2629 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2630 }
2631 
2632 /*
2633  * bridge_rtage:
2634  *
2635  *	Perform an aging cycle.
2636  */
2637 static void
2638 bridge_rtage(struct bridge_softc *sc)
2639 {
2640 	struct bridge_rtnode *brt, *nbrt;
2641 
2642 	BRIDGE_LOCK_ASSERT(sc);
2643 
2644 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2645 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2646 			if (time_uptime >= brt->brt_expire)
2647 				bridge_rtnode_destroy(sc, brt);
2648 		}
2649 	}
2650 }
2651 
2652 /*
2653  * bridge_rtflush:
2654  *
2655  *	Remove all dynamic addresses from the bridge.
2656  */
2657 static void
2658 bridge_rtflush(struct bridge_softc *sc, int full)
2659 {
2660 	struct bridge_rtnode *brt, *nbrt;
2661 
2662 	BRIDGE_LOCK_ASSERT(sc);
2663 
2664 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2665 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2666 			bridge_rtnode_destroy(sc, brt);
2667 	}
2668 }
2669 
2670 /*
2671  * bridge_rtdaddr:
2672  *
2673  *	Remove an address from the table.
2674  */
2675 static int
2676 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2677 {
2678 	struct bridge_rtnode *brt;
2679 	int found = 0;
2680 
2681 	BRIDGE_LOCK_ASSERT(sc);
2682 
2683 	/*
2684 	 * If vlan is zero then we want to delete for all vlans so the lookup
2685 	 * may return more than one.
2686 	 */
2687 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2688 		bridge_rtnode_destroy(sc, brt);
2689 		found = 1;
2690 	}
2691 
2692 	return (found ? 0 : ENOENT);
2693 }
2694 
2695 /*
2696  * bridge_rtdelete:
2697  *
2698  *	Delete routes to a speicifc member interface.
2699  */
2700 static void
2701 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2702 {
2703 	struct bridge_rtnode *brt, *nbrt;
2704 
2705 	BRIDGE_LOCK_ASSERT(sc);
2706 
2707 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2708 		if (brt->brt_ifp == ifp && (full ||
2709 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2710 			bridge_rtnode_destroy(sc, brt);
2711 	}
2712 }
2713 
2714 /*
2715  * bridge_rtable_init:
2716  *
2717  *	Initialize the route table for this bridge.
2718  */
2719 static int
2720 bridge_rtable_init(struct bridge_softc *sc)
2721 {
2722 	int i;
2723 
2724 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2725 	    M_DEVBUF, M_NOWAIT);
2726 	if (sc->sc_rthash == NULL)
2727 		return (ENOMEM);
2728 
2729 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2730 		LIST_INIT(&sc->sc_rthash[i]);
2731 
2732 	sc->sc_rthash_key = arc4random();
2733 
2734 	LIST_INIT(&sc->sc_rtlist);
2735 
2736 	return (0);
2737 }
2738 
2739 /*
2740  * bridge_rtable_fini:
2741  *
2742  *	Deconstruct the route table for this bridge.
2743  */
2744 static void
2745 bridge_rtable_fini(struct bridge_softc *sc)
2746 {
2747 
2748 	KASSERT(sc->sc_brtcnt == 0,
2749 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2750 	free(sc->sc_rthash, M_DEVBUF);
2751 }
2752 
2753 /*
2754  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2755  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2756  */
2757 #define	mix(a, b, c)							\
2758 do {									\
2759 	a -= b; a -= c; a ^= (c >> 13);					\
2760 	b -= c; b -= a; b ^= (a << 8);					\
2761 	c -= a; c -= b; c ^= (b >> 13);					\
2762 	a -= b; a -= c; a ^= (c >> 12);					\
2763 	b -= c; b -= a; b ^= (a << 16);					\
2764 	c -= a; c -= b; c ^= (b >> 5);					\
2765 	a -= b; a -= c; a ^= (c >> 3);					\
2766 	b -= c; b -= a; b ^= (a << 10);					\
2767 	c -= a; c -= b; c ^= (b >> 15);					\
2768 } while (/*CONSTCOND*/0)
2769 
2770 static __inline uint32_t
2771 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2772 {
2773 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2774 
2775 	b += addr[5] << 8;
2776 	b += addr[4];
2777 	a += addr[3] << 24;
2778 	a += addr[2] << 16;
2779 	a += addr[1] << 8;
2780 	a += addr[0];
2781 
2782 	mix(a, b, c);
2783 
2784 	return (c & BRIDGE_RTHASH_MASK);
2785 }
2786 
2787 #undef mix
2788 
2789 static int
2790 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2791 {
2792 	int i, d;
2793 
2794 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2795 		d = ((int)a[i]) - ((int)b[i]);
2796 	}
2797 
2798 	return (d);
2799 }
2800 
2801 /*
2802  * bridge_rtnode_lookup:
2803  *
2804  *	Look up a bridge route node for the specified destination. Compare the
2805  *	vlan id or if zero then just return the first match.
2806  */
2807 static struct bridge_rtnode *
2808 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2809 {
2810 	struct bridge_rtnode *brt;
2811 	uint32_t hash;
2812 	int dir;
2813 
2814 	BRIDGE_LOCK_ASSERT(sc);
2815 
2816 	hash = bridge_rthash(sc, addr);
2817 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2818 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2819 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2820 			return (brt);
2821 		if (dir > 0)
2822 			return (NULL);
2823 	}
2824 
2825 	return (NULL);
2826 }
2827 
2828 /*
2829  * bridge_rtnode_insert:
2830  *
2831  *	Insert the specified bridge node into the route table.  We
2832  *	assume the entry is not already in the table.
2833  */
2834 static int
2835 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2836 {
2837 	struct bridge_rtnode *lbrt;
2838 	uint32_t hash;
2839 	int dir;
2840 
2841 	BRIDGE_LOCK_ASSERT(sc);
2842 
2843 	hash = bridge_rthash(sc, brt->brt_addr);
2844 
2845 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2846 	if (lbrt == NULL) {
2847 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2848 		goto out;
2849 	}
2850 
2851 	do {
2852 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2853 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2854 			return (EEXIST);
2855 		if (dir > 0) {
2856 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2857 			goto out;
2858 		}
2859 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2860 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2861 			goto out;
2862 		}
2863 		lbrt = LIST_NEXT(lbrt, brt_hash);
2864 	} while (lbrt != NULL);
2865 
2866 #ifdef DIAGNOSTIC
2867 	panic("bridge_rtnode_insert: impossible");
2868 #endif
2869 
2870 out:
2871 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2872 	sc->sc_brtcnt++;
2873 
2874 	return (0);
2875 }
2876 
2877 /*
2878  * bridge_rtnode_destroy:
2879  *
2880  *	Destroy a bridge rtnode.
2881  */
2882 static void
2883 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2884 {
2885 	BRIDGE_LOCK_ASSERT(sc);
2886 
2887 	LIST_REMOVE(brt, brt_hash);
2888 
2889 	LIST_REMOVE(brt, brt_list);
2890 	sc->sc_brtcnt--;
2891 	brt->brt_dst->bif_addrcnt--;
2892 	uma_zfree(bridge_rtnode_zone, brt);
2893 }
2894 
2895 /*
2896  * bridge_rtable_expire:
2897  *
2898  *	Set the expiry time for all routes on an interface.
2899  */
2900 static void
2901 bridge_rtable_expire(struct ifnet *ifp, int age)
2902 {
2903 	struct bridge_softc *sc = ifp->if_bridge;
2904 	struct bridge_rtnode *brt;
2905 
2906 	BRIDGE_LOCK(sc);
2907 
2908 	/*
2909 	 * If the age is zero then flush, otherwise set all the expiry times to
2910 	 * age for the interface
2911 	 */
2912 	if (age == 0)
2913 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2914 	else {
2915 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2916 			/* Cap the expiry time to 'age' */
2917 			if (brt->brt_ifp == ifp &&
2918 			    brt->brt_expire > time_uptime + age &&
2919 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2920 				brt->brt_expire = time_uptime + age;
2921 		}
2922 	}
2923 	BRIDGE_UNLOCK(sc);
2924 }
2925 
2926 /*
2927  * bridge_state_change:
2928  *
2929  *	Callback from the bridgestp code when a port changes states.
2930  */
2931 static void
2932 bridge_state_change(struct ifnet *ifp, int state)
2933 {
2934 	struct bridge_softc *sc = ifp->if_bridge;
2935 	static const char *stpstates[] = {
2936 		"disabled",
2937 		"listening",
2938 		"learning",
2939 		"forwarding",
2940 		"blocking",
2941 		"discarding"
2942 	};
2943 
2944 	if (log_stp)
2945 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2946 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2947 }
2948 
2949 /*
2950  * Send bridge packets through pfil if they are one of the types pfil can deal
2951  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
2952  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2953  * that interface.
2954  */
2955 static int
2956 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2957 {
2958 	int snap, error, i, hlen;
2959 	struct ether_header *eh1, eh2;
2960 	struct ip_fw_args args;
2961 	struct ip *ip;
2962 	struct llc llc1;
2963 	u_int16_t ether_type;
2964 
2965 	snap = 0;
2966 	error = -1;	/* Default error if not error == 0 */
2967 
2968 #if 0
2969 	/* we may return with the IP fields swapped, ensure its not shared */
2970 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2971 #endif
2972 
2973 	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2974 		return (0); /* filtering is disabled */
2975 
2976 	i = min((*mp)->m_pkthdr.len, max_protohdr);
2977 	if ((*mp)->m_len < i) {
2978 	    *mp = m_pullup(*mp, i);
2979 	    if (*mp == NULL) {
2980 		printf("%s: m_pullup failed\n", __func__);
2981 		return (-1);
2982 	    }
2983 	}
2984 
2985 	eh1 = mtod(*mp, struct ether_header *);
2986 	ether_type = ntohs(eh1->ether_type);
2987 
2988 	/*
2989 	 * Check for SNAP/LLC.
2990 	 */
2991 	if (ether_type < ETHERMTU) {
2992 		struct llc *llc2 = (struct llc *)(eh1 + 1);
2993 
2994 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
2995 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
2996 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
2997 		    llc2->llc_control == LLC_UI) {
2998 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
2999 			snap = 1;
3000 		}
3001 	}
3002 
3003 	/*
3004 	 * If we're trying to filter bridge traffic, don't look at anything
3005 	 * other than IP and ARP traffic.  If the filter doesn't understand
3006 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3007 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3008 	 * but of course we don't have an AppleTalk filter to begin with.
3009 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3010 	 * ARP traffic.)
3011 	 */
3012 	switch (ether_type) {
3013 		case ETHERTYPE_ARP:
3014 		case ETHERTYPE_REVARP:
3015 			if (pfil_ipfw_arp == 0)
3016 				return (0); /* Automatically pass */
3017 			break;
3018 
3019 		case ETHERTYPE_IP:
3020 #ifdef INET6
3021 		case ETHERTYPE_IPV6:
3022 #endif /* INET6 */
3023 			break;
3024 		default:
3025 			/*
3026 			 * Check to see if the user wants to pass non-ip
3027 			 * packets, these will not be checked by pfil(9) and
3028 			 * passed unconditionally so the default is to drop.
3029 			 */
3030 			if (pfil_onlyip)
3031 				goto bad;
3032 	}
3033 
3034 	/* Strip off the Ethernet header and keep a copy. */
3035 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3036 	m_adj(*mp, ETHER_HDR_LEN);
3037 
3038 	/* Strip off snap header, if present */
3039 	if (snap) {
3040 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3041 		m_adj(*mp, sizeof(struct llc));
3042 	}
3043 
3044 	/*
3045 	 * Check the IP header for alignment and errors
3046 	 */
3047 	if (dir == PFIL_IN) {
3048 		switch (ether_type) {
3049 			case ETHERTYPE_IP:
3050 				error = bridge_ip_checkbasic(mp);
3051 				break;
3052 #ifdef INET6
3053 			case ETHERTYPE_IPV6:
3054 				error = bridge_ip6_checkbasic(mp);
3055 				break;
3056 #endif /* INET6 */
3057 			default:
3058 				error = 0;
3059 		}
3060 		if (error)
3061 			goto bad;
3062 	}
3063 
3064 	/* XXX this section is also in if_ethersubr.c */
3065 	// XXX PFIL_OUT or DIR_OUT ?
3066 	if (V_ip_fw_chk_ptr && pfil_ipfw != 0 &&
3067 			dir == PFIL_OUT && ifp != NULL) {
3068 		struct m_tag *mtag;
3069 
3070 		error = -1;
3071 		/* fetch the start point from existing tags, if any */
3072 		mtag = m_tag_locate(*mp, MTAG_IPFW_RULE, 0, NULL);
3073 		if (mtag == NULL) {
3074 			args.rule.slot = 0;
3075 		} else {
3076 			struct dn_pkt_tag *dn_tag;
3077 
3078 			/* XXX can we free the tag after use ? */
3079 			mtag->m_tag_id = PACKET_TAG_NONE;
3080 			dn_tag = (struct dn_pkt_tag *)(mtag + 1);
3081 			/* packet already partially processed ? */
3082 			if (dn_tag->rule.slot != 0 && V_fw_one_pass)
3083 				goto ipfwpass;
3084 			args.rule = dn_tag->rule;
3085 		}
3086 
3087 		args.m = *mp;
3088 		args.oif = ifp;
3089 		args.next_hop = NULL;
3090 		args.eh = &eh2;
3091 		args.inp = NULL;	/* used by ipfw uid/gid/jail rules */
3092 		i = V_ip_fw_chk_ptr(&args);
3093 		*mp = args.m;
3094 
3095 		if (*mp == NULL)
3096 			return (error);
3097 
3098 		if (ip_dn_io_ptr && (i == IP_FW_DUMMYNET)) {
3099 
3100 			/* put the Ethernet header back on */
3101 			M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3102 			if (*mp == NULL)
3103 				return (error);
3104 			bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3105 
3106 			/*
3107 			 * Pass the pkt to dummynet, which consumes it. The
3108 			 * packet will return to us via bridge_dummynet().
3109 			 */
3110 			args.oif = ifp;
3111 			ip_dn_io_ptr(mp, DIR_FWD | PROTO_IFB, &args);
3112 			return (error);
3113 		}
3114 
3115 		if (i != IP_FW_PASS) /* drop */
3116 			goto bad;
3117 	}
3118 
3119 ipfwpass:
3120 	error = 0;
3121 
3122 	/*
3123 	 * Run the packet through pfil
3124 	 */
3125 	switch (ether_type) {
3126 	case ETHERTYPE_IP:
3127 		/*
3128 		 * before calling the firewall, swap fields the same as
3129 		 * IP does. here we assume the header is contiguous
3130 		 */
3131 		ip = mtod(*mp, struct ip *);
3132 
3133 		ip->ip_len = ntohs(ip->ip_len);
3134 		ip->ip_off = ntohs(ip->ip_off);
3135 
3136 		/*
3137 		 * Run pfil on the member interface and the bridge, both can
3138 		 * be skipped by clearing pfil_member or pfil_bridge.
3139 		 *
3140 		 * Keep the order:
3141 		 *   in_if -> bridge_if -> out_if
3142 		 */
3143 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3144 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3145 					dir, NULL);
3146 
3147 		if (*mp == NULL || error != 0) /* filter may consume */
3148 			break;
3149 
3150 		if (pfil_member && ifp != NULL)
3151 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3152 					dir, NULL);
3153 
3154 		if (*mp == NULL || error != 0) /* filter may consume */
3155 			break;
3156 
3157 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3158 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3159 					dir, NULL);
3160 
3161 		if (*mp == NULL || error != 0) /* filter may consume */
3162 			break;
3163 
3164 		/* check if we need to fragment the packet */
3165 		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3166 			i = (*mp)->m_pkthdr.len;
3167 			if (i > ifp->if_mtu) {
3168 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3169 					    &llc1);
3170 				return (error);
3171 			}
3172 		}
3173 
3174 		/* Recalculate the ip checksum and restore byte ordering */
3175 		ip = mtod(*mp, struct ip *);
3176 		hlen = ip->ip_hl << 2;
3177 		if (hlen < sizeof(struct ip))
3178 			goto bad;
3179 		if (hlen > (*mp)->m_len) {
3180 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3181 				goto bad;
3182 			ip = mtod(*mp, struct ip *);
3183 			if (ip == NULL)
3184 				goto bad;
3185 		}
3186 		ip->ip_len = htons(ip->ip_len);
3187 		ip->ip_off = htons(ip->ip_off);
3188 		ip->ip_sum = 0;
3189 		if (hlen == sizeof(struct ip))
3190 			ip->ip_sum = in_cksum_hdr(ip);
3191 		else
3192 			ip->ip_sum = in_cksum(*mp, hlen);
3193 
3194 		break;
3195 #ifdef INET6
3196 	case ETHERTYPE_IPV6:
3197 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3198 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3199 					dir, NULL);
3200 
3201 		if (*mp == NULL || error != 0) /* filter may consume */
3202 			break;
3203 
3204 		if (pfil_member && ifp != NULL)
3205 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3206 					dir, NULL);
3207 
3208 		if (*mp == NULL || error != 0) /* filter may consume */
3209 			break;
3210 
3211 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3212 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3213 					dir, NULL);
3214 		break;
3215 #endif
3216 	default:
3217 		error = 0;
3218 		break;
3219 	}
3220 
3221 	if (*mp == NULL)
3222 		return (error);
3223 	if (error != 0)
3224 		goto bad;
3225 
3226 	error = -1;
3227 
3228 	/*
3229 	 * Finally, put everything back the way it was and return
3230 	 */
3231 	if (snap) {
3232 		M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT);
3233 		if (*mp == NULL)
3234 			return (error);
3235 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3236 	}
3237 
3238 	M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3239 	if (*mp == NULL)
3240 		return (error);
3241 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3242 
3243 	return (0);
3244 
3245 bad:
3246 	m_freem(*mp);
3247 	*mp = NULL;
3248 	return (error);
3249 }
3250 
3251 /*
3252  * Perform basic checks on header size since
3253  * pfil assumes ip_input has already processed
3254  * it for it.  Cut-and-pasted from ip_input.c.
3255  * Given how simple the IPv6 version is,
3256  * does the IPv4 version really need to be
3257  * this complicated?
3258  *
3259  * XXX Should we update ipstat here, or not?
3260  * XXX Right now we update ipstat but not
3261  * XXX csum_counter.
3262  */
3263 static int
3264 bridge_ip_checkbasic(struct mbuf **mp)
3265 {
3266 	struct mbuf *m = *mp;
3267 	struct ip *ip;
3268 	int len, hlen;
3269 	u_short sum;
3270 
3271 	if (*mp == NULL)
3272 		return (-1);
3273 
3274 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3275 		if ((m = m_copyup(m, sizeof(struct ip),
3276 			(max_linkhdr + 3) & ~3)) == NULL) {
3277 			/* XXXJRT new stat, please */
3278 			KMOD_IPSTAT_INC(ips_toosmall);
3279 			goto bad;
3280 		}
3281 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3282 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3283 			KMOD_IPSTAT_INC(ips_toosmall);
3284 			goto bad;
3285 		}
3286 	}
3287 	ip = mtod(m, struct ip *);
3288 	if (ip == NULL) goto bad;
3289 
3290 	if (ip->ip_v != IPVERSION) {
3291 		KMOD_IPSTAT_INC(ips_badvers);
3292 		goto bad;
3293 	}
3294 	hlen = ip->ip_hl << 2;
3295 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3296 		KMOD_IPSTAT_INC(ips_badhlen);
3297 		goto bad;
3298 	}
3299 	if (hlen > m->m_len) {
3300 		if ((m = m_pullup(m, hlen)) == 0) {
3301 			KMOD_IPSTAT_INC(ips_badhlen);
3302 			goto bad;
3303 		}
3304 		ip = mtod(m, struct ip *);
3305 		if (ip == NULL) goto bad;
3306 	}
3307 
3308 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3309 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3310 	} else {
3311 		if (hlen == sizeof(struct ip)) {
3312 			sum = in_cksum_hdr(ip);
3313 		} else {
3314 			sum = in_cksum(m, hlen);
3315 		}
3316 	}
3317 	if (sum) {
3318 		KMOD_IPSTAT_INC(ips_badsum);
3319 		goto bad;
3320 	}
3321 
3322 	/* Retrieve the packet length. */
3323 	len = ntohs(ip->ip_len);
3324 
3325 	/*
3326 	 * Check for additional length bogosity
3327 	 */
3328 	if (len < hlen) {
3329 		KMOD_IPSTAT_INC(ips_badlen);
3330 		goto bad;
3331 	}
3332 
3333 	/*
3334 	 * Check that the amount of data in the buffers
3335 	 * is as at least much as the IP header would have us expect.
3336 	 * Drop packet if shorter than we expect.
3337 	 */
3338 	if (m->m_pkthdr.len < len) {
3339 		KMOD_IPSTAT_INC(ips_tooshort);
3340 		goto bad;
3341 	}
3342 
3343 	/* Checks out, proceed */
3344 	*mp = m;
3345 	return (0);
3346 
3347 bad:
3348 	*mp = m;
3349 	return (-1);
3350 }
3351 
3352 #ifdef INET6
3353 /*
3354  * Same as above, but for IPv6.
3355  * Cut-and-pasted from ip6_input.c.
3356  * XXX Should we update ip6stat, or not?
3357  */
3358 static int
3359 bridge_ip6_checkbasic(struct mbuf **mp)
3360 {
3361 	struct mbuf *m = *mp;
3362 	struct ip6_hdr *ip6;
3363 
3364 	/*
3365 	 * If the IPv6 header is not aligned, slurp it up into a new
3366 	 * mbuf with space for link headers, in the event we forward
3367 	 * it.  Otherwise, if it is aligned, make sure the entire base
3368 	 * IPv6 header is in the first mbuf of the chain.
3369 	 */
3370 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3371 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3372 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3373 			    (max_linkhdr + 3) & ~3)) == NULL) {
3374 			/* XXXJRT new stat, please */
3375 			V_ip6stat.ip6s_toosmall++;
3376 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3377 			goto bad;
3378 		}
3379 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3380 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3381 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3382 			V_ip6stat.ip6s_toosmall++;
3383 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3384 			goto bad;
3385 		}
3386 	}
3387 
3388 	ip6 = mtod(m, struct ip6_hdr *);
3389 
3390 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3391 		V_ip6stat.ip6s_badvers++;
3392 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3393 		goto bad;
3394 	}
3395 
3396 	/* Checks out, proceed */
3397 	*mp = m;
3398 	return (0);
3399 
3400 bad:
3401 	*mp = m;
3402 	return (-1);
3403 }
3404 #endif /* INET6 */
3405 
3406 /*
3407  * bridge_fragment:
3408  *
3409  *	Return a fragmented mbuf chain.
3410  */
3411 static int
3412 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3413     int snap, struct llc *llc)
3414 {
3415 	struct mbuf *m0;
3416 	struct ip *ip;
3417 	int error = -1;
3418 
3419 	if (m->m_len < sizeof(struct ip) &&
3420 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3421 		goto out;
3422 	ip = mtod(m, struct ip *);
3423 
3424 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist,
3425 		    CSUM_DELAY_IP);
3426 	if (error)
3427 		goto out;
3428 
3429 	/* walk the chain and re-add the Ethernet header */
3430 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3431 		if (error == 0) {
3432 			if (snap) {
3433 				M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT);
3434 				if (m0 == NULL) {
3435 					error = ENOBUFS;
3436 					continue;
3437 				}
3438 				bcopy(llc, mtod(m0, caddr_t),
3439 				    sizeof(struct llc));
3440 			}
3441 			M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT);
3442 			if (m0 == NULL) {
3443 				error = ENOBUFS;
3444 				continue;
3445 			}
3446 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3447 		} else
3448 			m_freem(m);
3449 	}
3450 
3451 	if (error == 0)
3452 		KMOD_IPSTAT_INC(ips_fragmented);
3453 
3454 	return (error);
3455 
3456 out:
3457 	if (m != NULL)
3458 		m_freem(m);
3459 	return (error);
3460 }
3461