xref: /freebsd/sys/net/if_bridge.c (revision b3aaa0cc21c63d388230c7ef2a80abd631ff20d5)
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 #include <sys/vimage.h>
105 
106 #include <net/bpf.h>
107 #include <net/if.h>
108 #include <net/if_clone.h>
109 #include <net/if_dl.h>
110 #include <net/if_types.h>
111 #include <net/if_var.h>
112 #include <net/pfil.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 #include <netinet/vinet.h>
120 #ifdef INET6
121 #include <netinet/ip6.h>
122 #include <netinet6/ip6_var.h>
123 #include <netinet6/vinet6.h>
124 #endif
125 #ifdef DEV_CARP
126 #include <netinet/ip_carp.h>
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/ip_dummynet.h>
138 
139 /*
140  * Size of the route hash table.  Must be a power of two.
141  */
142 #ifndef BRIDGE_RTHASH_SIZE
143 #define	BRIDGE_RTHASH_SIZE		1024
144 #endif
145 
146 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
147 
148 /*
149  * Maximum number of addresses to cache.
150  */
151 #ifndef BRIDGE_RTABLE_MAX
152 #define	BRIDGE_RTABLE_MAX		100
153 #endif
154 
155 /*
156  * Timeout (in seconds) for entries learned dynamically.
157  */
158 #ifndef BRIDGE_RTABLE_TIMEOUT
159 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
160 #endif
161 
162 /*
163  * Number of seconds between walks of the route list.
164  */
165 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
166 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
167 #endif
168 
169 /*
170  * List of capabilities to possibly mask on the member interface.
171  */
172 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
173 
174 /*
175  * Bridge interface list entry.
176  */
177 struct bridge_iflist {
178 	LIST_ENTRY(bridge_iflist) bif_next;
179 	struct ifnet		*bif_ifp;	/* member if */
180 	struct bstp_port	bif_stp;	/* STP state */
181 	uint32_t		bif_flags;	/* member if flags */
182 	int			bif_savedcaps;	/* saved capabilities */
183 	uint32_t		bif_addrmax;	/* max # of addresses */
184 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
185 	uint32_t		bif_addrexceeded;/* # of address violations */
186 };
187 
188 /*
189  * Bridge route node.
190  */
191 struct bridge_rtnode {
192 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
193 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
194 	struct bridge_iflist	*brt_dst;	/* destination if */
195 	unsigned long		brt_expire;	/* expiration time */
196 	uint8_t			brt_flags;	/* address flags */
197 	uint8_t			brt_addr[ETHER_ADDR_LEN];
198 	uint16_t		brt_vlan;	/* vlan id */
199 };
200 #define	brt_ifp			brt_dst->bif_ifp
201 
202 /*
203  * Software state for each bridge.
204  */
205 struct bridge_softc {
206 	struct ifnet		*sc_ifp;	/* make this an interface */
207 	LIST_ENTRY(bridge_softc) sc_list;
208 	struct mtx		sc_mtx;
209 	struct cv		sc_cv;
210 	uint32_t		sc_brtmax;	/* max # of addresses */
211 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
212 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
213 	struct callout		sc_brcallout;	/* bridge callout */
214 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
215 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
216 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
217 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
218 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
219 	uint32_t		sc_rthash_key;	/* key for hash */
220 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
221 	struct bstp_state	sc_stp;		/* STP state */
222 	uint32_t		sc_brtexceeded;	/* # of cache drops */
223 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
224 	u_char			sc_defaddr[6];	/* Default MAC address */
225 };
226 
227 static struct mtx 	bridge_list_mtx;
228 eventhandler_tag	bridge_detach_cookie = NULL;
229 
230 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
231 
232 uma_zone_t bridge_rtnode_zone;
233 
234 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
235 static void	bridge_clone_destroy(struct ifnet *);
236 
237 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
238 static void	bridge_mutecaps(struct bridge_softc *);
239 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
240 		    int);
241 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
242 static void	bridge_init(void *);
243 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
244 static void	bridge_stop(struct ifnet *, int);
245 static void	bridge_start(struct ifnet *);
246 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
247 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
248 		    struct rtentry *);
249 static void	bridge_enqueue(struct bridge_softc *, struct ifnet *,
250 		    struct mbuf *);
251 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
252 
253 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
254 		    struct mbuf *m);
255 
256 static void	bridge_timer(void *);
257 
258 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
259 		    struct mbuf *, int);
260 static void	bridge_span(struct bridge_softc *, struct mbuf *);
261 
262 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
263 		    uint16_t, struct bridge_iflist *, int, uint8_t);
264 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
265 		    uint16_t);
266 static void	bridge_rttrim(struct bridge_softc *);
267 static void	bridge_rtage(struct bridge_softc *);
268 static void	bridge_rtflush(struct bridge_softc *, int);
269 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
270 		    uint16_t);
271 
272 static int	bridge_rtable_init(struct bridge_softc *);
273 static void	bridge_rtable_fini(struct bridge_softc *);
274 
275 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
276 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
277 		    const uint8_t *, uint16_t);
278 static int	bridge_rtnode_insert(struct bridge_softc *,
279 		    struct bridge_rtnode *);
280 static void	bridge_rtnode_destroy(struct bridge_softc *,
281 		    struct bridge_rtnode *);
282 static void	bridge_rtable_expire(struct ifnet *, int);
283 static void	bridge_state_change(struct ifnet *, int);
284 
285 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
286 		    const char *name);
287 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
288 		    struct ifnet *ifp);
289 static void	bridge_delete_member(struct bridge_softc *,
290 		    struct bridge_iflist *, int);
291 static void	bridge_delete_span(struct bridge_softc *,
292 		    struct bridge_iflist *);
293 
294 static int	bridge_ioctl_add(struct bridge_softc *, void *);
295 static int	bridge_ioctl_del(struct bridge_softc *, void *);
296 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
297 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
298 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
299 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
300 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
301 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
302 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
303 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
304 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
305 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
306 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
307 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
308 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
309 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
310 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
311 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
312 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
313 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
314 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
315 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
316 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
317 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
318 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
319 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
320 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
321 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
322 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
323 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
324 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
325 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
326 		    int);
327 static int	bridge_ip_checkbasic(struct mbuf **mp);
328 #ifdef INET6
329 static int	bridge_ip6_checkbasic(struct mbuf **mp);
330 #endif /* INET6 */
331 static int	bridge_fragment(struct ifnet *, struct mbuf *,
332 		    struct ether_header *, int, struct llc *);
333 
334 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
335 #define	VLANTAGOF(_m)	\
336     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
337 
338 static struct bstp_cb_ops bridge_ops = {
339 	.bcb_state = bridge_state_change,
340 	.bcb_rtage = bridge_rtable_expire
341 };
342 
343 SYSCTL_DECL(_net_link);
344 SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
345 
346 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
347 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
348 static int pfil_member = 1; /* run pfil hooks on the member interface */
349 static int pfil_ipfw = 0;   /* layer2 filter with ipfw */
350 static int pfil_ipfw_arp = 0;   /* layer2 filter with ipfw */
351 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
352                                    locally destined packets */
353 static int log_stp   = 0;   /* log STP state changes */
354 static int bridge_inherit_mac = 0;   /* share MAC with first bridge member */
355 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
356     &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
357 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
358     &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
359 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
360     &pfil_bridge, 0, "Packet filter on the bridge interface");
361 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
362     &pfil_member, 0, "Packet filter on the member interface");
363 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
364     &pfil_local_phys, 0,
365     "Packet filter on the physical interface for locally destined packets");
366 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
367     &log_stp, 0, "Log STP state changes");
368 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
369     &bridge_inherit_mac, 0,
370     "Inherit MAC address from the first bridge member");
371 
372 struct bridge_control {
373 	int	(*bc_func)(struct bridge_softc *, void *);
374 	int	bc_argsize;
375 	int	bc_flags;
376 };
377 
378 #define	BC_F_COPYIN		0x01	/* copy arguments in */
379 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
380 #define	BC_F_SUSER		0x04	/* do super-user check */
381 
382 const struct bridge_control bridge_control_table[] = {
383 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
384 	  BC_F_COPYIN|BC_F_SUSER },
385 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
386 	  BC_F_COPYIN|BC_F_SUSER },
387 
388 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
389 	  BC_F_COPYIN|BC_F_COPYOUT },
390 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
391 	  BC_F_COPYIN|BC_F_SUSER },
392 
393 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
394 	  BC_F_COPYIN|BC_F_SUSER },
395 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
396 	  BC_F_COPYOUT },
397 
398 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
399 	  BC_F_COPYIN|BC_F_COPYOUT },
400 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
401 	  BC_F_COPYIN|BC_F_COPYOUT },
402 
403 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
404 	  BC_F_COPYIN|BC_F_SUSER },
405 
406 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
407 	  BC_F_COPYIN|BC_F_SUSER },
408 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
409 	  BC_F_COPYOUT },
410 
411 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
412 	  BC_F_COPYIN|BC_F_SUSER },
413 
414 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
415 	  BC_F_COPYIN|BC_F_SUSER },
416 
417 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
418 	  BC_F_COPYOUT },
419 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
420 	  BC_F_COPYIN|BC_F_SUSER },
421 
422 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
423 	  BC_F_COPYOUT },
424 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
425 	  BC_F_COPYIN|BC_F_SUSER },
426 
427 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
428 	  BC_F_COPYOUT },
429 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
430 	  BC_F_COPYIN|BC_F_SUSER },
431 
432 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
433 	  BC_F_COPYOUT },
434 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
435 	  BC_F_COPYIN|BC_F_SUSER },
436 
437 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
438 	  BC_F_COPYIN|BC_F_SUSER },
439 
440 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
441 	  BC_F_COPYIN|BC_F_SUSER },
442 
443 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
444 	  BC_F_COPYIN|BC_F_SUSER },
445 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
446 	  BC_F_COPYIN|BC_F_SUSER },
447 
448 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
449 	  BC_F_COPYOUT },
450 
451 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
452 	  BC_F_COPYOUT },
453 
454 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
455 	  BC_F_COPYIN|BC_F_COPYOUT },
456 
457 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
458 	  BC_F_COPYIN|BC_F_SUSER },
459 
460 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
461 	  BC_F_COPYIN|BC_F_SUSER },
462 
463 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
464 	  BC_F_COPYIN|BC_F_SUSER },
465 
466 };
467 const int bridge_control_table_size =
468     sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
469 
470 LIST_HEAD(, bridge_softc) bridge_list;
471 
472 IFC_SIMPLE_DECLARE(bridge, 0);
473 
474 static int
475 bridge_modevent(module_t mod, int type, void *data)
476 {
477 
478 	switch (type) {
479 	case MOD_LOAD:
480 		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
481 		if_clone_attach(&bridge_cloner);
482 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
483 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
484 		    UMA_ALIGN_PTR, 0);
485 		LIST_INIT(&bridge_list);
486 		bridge_input_p = bridge_input;
487 		bridge_output_p = bridge_output;
488 		bridge_dn_p = bridge_dummynet;
489 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
490 		    ifnet_departure_event, bridge_ifdetach, NULL,
491 		    EVENTHANDLER_PRI_ANY);
492 		break;
493 	case MOD_UNLOAD:
494 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
495 		    bridge_detach_cookie);
496 		if_clone_detach(&bridge_cloner);
497 		uma_zdestroy(bridge_rtnode_zone);
498 		bridge_input_p = NULL;
499 		bridge_output_p = NULL;
500 		bridge_dn_p = NULL;
501 		mtx_destroy(&bridge_list_mtx);
502 		break;
503 	default:
504 		return (EOPNOTSUPP);
505 	}
506 	return (0);
507 }
508 
509 static moduledata_t bridge_mod = {
510 	"if_bridge",
511 	bridge_modevent,
512 	0
513 };
514 
515 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
516 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
517 
518 /*
519  * handler for net.link.bridge.pfil_ipfw
520  */
521 static int
522 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
523 {
524 	int enable = pfil_ipfw;
525 	int error;
526 
527 	error = sysctl_handle_int(oidp, &enable, 0, req);
528 	enable = (enable) ? 1 : 0;
529 
530 	if (enable != pfil_ipfw) {
531 		pfil_ipfw = enable;
532 
533 		/*
534 		 * Disable pfil so that ipfw doesnt run twice, if the user
535 		 * really wants both then they can re-enable pfil_bridge and/or
536 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
537 		 * layer2 type.
538 		 */
539 		if (pfil_ipfw) {
540 			pfil_onlyip = 0;
541 			pfil_bridge = 0;
542 			pfil_member = 0;
543 		}
544 	}
545 
546 	return (error);
547 }
548 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
549 	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
550 
551 /*
552  * bridge_clone_create:
553  *
554  *	Create a new bridge instance.
555  */
556 static int
557 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
558 {
559 	struct bridge_softc *sc, *sc2;
560 	struct ifnet *bifp, *ifp;
561 	int retry;
562 
563 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
564 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
565 	if (ifp == NULL) {
566 		free(sc, M_DEVBUF);
567 		return (ENOSPC);
568 	}
569 
570 	BRIDGE_LOCK_INIT(sc);
571 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
572 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
573 
574 	/* Initialize our routing table. */
575 	bridge_rtable_init(sc);
576 
577 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
578 
579 	LIST_INIT(&sc->sc_iflist);
580 	LIST_INIT(&sc->sc_spanlist);
581 
582 	ifp->if_softc = sc;
583 	if_initname(ifp, ifc->ifc_name, unit);
584 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
585 	ifp->if_ioctl = bridge_ioctl;
586 	ifp->if_start = bridge_start;
587 	ifp->if_init = bridge_init;
588 	ifp->if_type = IFT_BRIDGE;
589 	IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
590 	ifp->if_snd.ifq_drv_maxlen = ifqmaxlen;
591 	IFQ_SET_READY(&ifp->if_snd);
592 
593 	/*
594 	 * Generate a random ethernet address with a locally administered
595 	 * address.
596 	 *
597 	 * Since we are using random ethernet addresses for the bridge, it is
598 	 * possible that we might have address collisions, so make sure that
599 	 * this hardware address isn't already in use on another bridge.
600 	 */
601 	for (retry = 1; retry != 0;) {
602 		arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
603 		sc->sc_defaddr[0] &= ~1;	/* clear multicast bit */
604 		sc->sc_defaddr[0] |= 2;		/* set the LAA bit */
605 		retry = 0;
606 		mtx_lock(&bridge_list_mtx);
607 		LIST_FOREACH(sc2, &bridge_list, sc_list) {
608 			bifp = sc2->sc_ifp;
609 			if (memcmp(sc->sc_defaddr,
610 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0)
611 				retry = 1;
612 		}
613 		mtx_unlock(&bridge_list_mtx);
614 	}
615 
616 	bstp_attach(&sc->sc_stp, &bridge_ops);
617 	ether_ifattach(ifp, sc->sc_defaddr);
618 	/* Now undo some of the damage... */
619 	ifp->if_baudrate = 0;
620 	ifp->if_type = IFT_BRIDGE;
621 
622 	mtx_lock(&bridge_list_mtx);
623 	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
624 	mtx_unlock(&bridge_list_mtx);
625 
626 	return (0);
627 }
628 
629 /*
630  * bridge_clone_destroy:
631  *
632  *	Destroy a bridge instance.
633  */
634 static void
635 bridge_clone_destroy(struct ifnet *ifp)
636 {
637 	struct bridge_softc *sc = ifp->if_softc;
638 	struct bridge_iflist *bif;
639 
640 	BRIDGE_LOCK(sc);
641 
642 	bridge_stop(ifp, 1);
643 	ifp->if_flags &= ~IFF_UP;
644 
645 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
646 		bridge_delete_member(sc, bif, 0);
647 
648 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
649 		bridge_delete_span(sc, bif);
650 	}
651 
652 	BRIDGE_UNLOCK(sc);
653 
654 	callout_drain(&sc->sc_brcallout);
655 
656 	mtx_lock(&bridge_list_mtx);
657 	LIST_REMOVE(sc, sc_list);
658 	mtx_unlock(&bridge_list_mtx);
659 
660 	bstp_detach(&sc->sc_stp);
661 	ether_ifdetach(ifp);
662 	if_free_type(ifp, IFT_ETHER);
663 
664 	/* Tear down the routing table. */
665 	bridge_rtable_fini(sc);
666 
667 	BRIDGE_LOCK_DESTROY(sc);
668 	free(sc, M_DEVBUF);
669 }
670 
671 /*
672  * bridge_ioctl:
673  *
674  *	Handle a control request from the operator.
675  */
676 static int
677 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
678 {
679 	struct bridge_softc *sc = ifp->if_softc;
680 	struct thread *td = curthread;
681 	union {
682 		struct ifbreq ifbreq;
683 		struct ifbifconf ifbifconf;
684 		struct ifbareq ifbareq;
685 		struct ifbaconf ifbaconf;
686 		struct ifbrparam ifbrparam;
687 		struct ifbropreq ifbropreq;
688 	} args;
689 	struct ifdrv *ifd = (struct ifdrv *) data;
690 	const struct bridge_control *bc;
691 	int error = 0;
692 
693 	switch (cmd) {
694 
695 	case SIOCADDMULTI:
696 	case SIOCDELMULTI:
697 		break;
698 
699 	case SIOCGDRVSPEC:
700 	case SIOCSDRVSPEC:
701 		if (ifd->ifd_cmd >= bridge_control_table_size) {
702 			error = EINVAL;
703 			break;
704 		}
705 		bc = &bridge_control_table[ifd->ifd_cmd];
706 
707 		if (cmd == SIOCGDRVSPEC &&
708 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
709 			error = EINVAL;
710 			break;
711 		}
712 		else if (cmd == SIOCSDRVSPEC &&
713 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
714 			error = EINVAL;
715 			break;
716 		}
717 
718 		if (bc->bc_flags & BC_F_SUSER) {
719 			error = priv_check(td, PRIV_NET_BRIDGE);
720 			if (error)
721 				break;
722 		}
723 
724 		if (ifd->ifd_len != bc->bc_argsize ||
725 		    ifd->ifd_len > sizeof(args)) {
726 			error = EINVAL;
727 			break;
728 		}
729 
730 		bzero(&args, sizeof(args));
731 		if (bc->bc_flags & BC_F_COPYIN) {
732 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
733 			if (error)
734 				break;
735 		}
736 
737 		BRIDGE_LOCK(sc);
738 		error = (*bc->bc_func)(sc, &args);
739 		BRIDGE_UNLOCK(sc);
740 		if (error)
741 			break;
742 
743 		if (bc->bc_flags & BC_F_COPYOUT)
744 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
745 
746 		break;
747 
748 	case SIOCSIFFLAGS:
749 		if (!(ifp->if_flags & IFF_UP) &&
750 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
751 			/*
752 			 * If interface is marked down and it is running,
753 			 * then stop and disable it.
754 			 */
755 			BRIDGE_LOCK(sc);
756 			bridge_stop(ifp, 1);
757 			BRIDGE_UNLOCK(sc);
758 		} else if ((ifp->if_flags & IFF_UP) &&
759 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
760 			/*
761 			 * If interface is marked up and it is stopped, then
762 			 * start it.
763 			 */
764 			(*ifp->if_init)(sc);
765 		}
766 		break;
767 
768 	case SIOCSIFMTU:
769 		/* Do not allow the MTU to be changed on the bridge */
770 		error = EINVAL;
771 		break;
772 
773 	default:
774 		/*
775 		 * drop the lock as ether_ioctl() will call bridge_start() and
776 		 * cause the lock to be recursed.
777 		 */
778 		error = ether_ioctl(ifp, cmd, data);
779 		break;
780 	}
781 
782 	return (error);
783 }
784 
785 /*
786  * bridge_mutecaps:
787  *
788  *	Clear or restore unwanted capabilities on the member interface
789  */
790 static void
791 bridge_mutecaps(struct bridge_softc *sc)
792 {
793 	struct bridge_iflist *bif;
794 	int enabled, mask;
795 
796 	/* Initial bitmask of capabilities to test */
797 	mask = BRIDGE_IFCAPS_MASK;
798 
799 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
800 		/* Every member must support it or its disabled */
801 		mask &= bif->bif_savedcaps;
802 	}
803 
804 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
805 		enabled = bif->bif_ifp->if_capenable;
806 		/* strip off mask bits and enable them again if allowed */
807 		enabled &= ~BRIDGE_IFCAPS_MASK;
808 		enabled |= mask;
809 		/*
810 		 * Receive offload can only be enabled if all members also
811 		 * support send offload.
812 		 */
813 		if ((enabled & IFCAP_TSO) == 0)
814 			enabled &= ~IFCAP_LRO;
815 
816 		bridge_set_ifcap(sc, bif, enabled);
817 	}
818 
819 }
820 
821 static void
822 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
823 {
824 	struct ifnet *ifp = bif->bif_ifp;
825 	struct ifreq ifr;
826 	int error;
827 
828 	bzero(&ifr, sizeof(ifr));
829 	ifr.ifr_reqcap = set;
830 
831 	if (ifp->if_capenable != set) {
832 		IFF_LOCKGIANT(ifp);
833 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
834 		IFF_UNLOCKGIANT(ifp);
835 		if (error)
836 			if_printf(sc->sc_ifp,
837 			    "error setting interface capabilities on %s\n",
838 			    ifp->if_xname);
839 	}
840 }
841 
842 /*
843  * bridge_lookup_member:
844  *
845  *	Lookup a bridge member interface.
846  */
847 static struct bridge_iflist *
848 bridge_lookup_member(struct bridge_softc *sc, const char *name)
849 {
850 	struct bridge_iflist *bif;
851 	struct ifnet *ifp;
852 
853 	BRIDGE_LOCK_ASSERT(sc);
854 
855 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
856 		ifp = bif->bif_ifp;
857 		if (strcmp(ifp->if_xname, name) == 0)
858 			return (bif);
859 	}
860 
861 	return (NULL);
862 }
863 
864 /*
865  * bridge_lookup_member_if:
866  *
867  *	Lookup a bridge member interface by ifnet*.
868  */
869 static struct bridge_iflist *
870 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
871 {
872 	struct bridge_iflist *bif;
873 
874 	BRIDGE_LOCK_ASSERT(sc);
875 
876 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
877 		if (bif->bif_ifp == member_ifp)
878 			return (bif);
879 	}
880 
881 	return (NULL);
882 }
883 
884 /*
885  * bridge_delete_member:
886  *
887  *	Delete the specified member interface.
888  */
889 static void
890 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
891     int gone)
892 {
893 	struct ifnet *ifs = bif->bif_ifp;
894 	struct ifnet *fif = NULL;
895 
896 	BRIDGE_LOCK_ASSERT(sc);
897 
898 	if (!gone) {
899 		switch (ifs->if_type) {
900 		case IFT_ETHER:
901 		case IFT_L2VLAN:
902 			/*
903 			 * Take the interface out of promiscuous mode.
904 			 */
905 			(void) ifpromisc(ifs, 0);
906 			break;
907 
908 		case IFT_GIF:
909 			break;
910 
911 		default:
912 #ifdef DIAGNOSTIC
913 			panic("bridge_delete_member: impossible");
914 #endif
915 			break;
916 		}
917 		/* reneable any interface capabilities */
918 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
919 	}
920 
921 	if (bif->bif_flags & IFBIF_STP)
922 		bstp_disable(&bif->bif_stp);
923 
924 	ifs->if_bridge = NULL;
925 	BRIDGE_XLOCK(sc);
926 	LIST_REMOVE(bif, bif_next);
927 	BRIDGE_XDROP(sc);
928 
929 	/*
930 	 * If removing the interface that gave the bridge its mac address, set
931 	 * the mac address of the bridge to the address of the next member, or
932 	 * to its default address if no members are left.
933 	 */
934 	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
935 		if (LIST_EMPTY(&sc->sc_iflist)) {
936 			bcopy(sc->sc_defaddr,
937 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
938 			sc->sc_ifaddr = NULL;
939 		} else {
940 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
941 			bcopy(IF_LLADDR(fif),
942 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
943 			sc->sc_ifaddr = fif;
944 		}
945 	}
946 
947 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
948 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
949 	KASSERT(bif->bif_addrcnt == 0,
950 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
951 
952 	BRIDGE_UNLOCK(sc);
953 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
954 	BRIDGE_LOCK(sc);
955 	free(bif, M_DEVBUF);
956 }
957 
958 /*
959  * bridge_delete_span:
960  *
961  *	Delete the specified span interface.
962  */
963 static void
964 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
965 {
966 	BRIDGE_LOCK_ASSERT(sc);
967 
968 	KASSERT(bif->bif_ifp->if_bridge == NULL,
969 	    ("%s: not a span interface", __func__));
970 
971 	LIST_REMOVE(bif, bif_next);
972 	free(bif, M_DEVBUF);
973 }
974 
975 static int
976 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
977 {
978 	struct ifbreq *req = arg;
979 	struct bridge_iflist *bif = NULL;
980 	struct ifnet *ifs;
981 	int error = 0;
982 
983 	ifs = ifunit(req->ifbr_ifsname);
984 	if (ifs == NULL)
985 		return (ENOENT);
986 	if (ifs->if_ioctl == NULL)	/* must be supported */
987 		return (EINVAL);
988 
989 	/* If it's in the span list, it can't be a member. */
990 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
991 		if (ifs == bif->bif_ifp)
992 			return (EBUSY);
993 
994 	/* Allow the first Ethernet member to define the MTU */
995 	if (ifs->if_type != IFT_GIF) {
996 		if (LIST_EMPTY(&sc->sc_iflist))
997 			sc->sc_ifp->if_mtu = ifs->if_mtu;
998 		else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
999 			if_printf(sc->sc_ifp, "invalid MTU for %s\n",
1000 			    ifs->if_xname);
1001 			return (EINVAL);
1002 		}
1003 	}
1004 
1005 	if (ifs->if_bridge == sc)
1006 		return (EEXIST);
1007 
1008 	if (ifs->if_bridge != NULL)
1009 		return (EBUSY);
1010 
1011 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1012 	if (bif == NULL)
1013 		return (ENOMEM);
1014 
1015 	bif->bif_ifp = ifs;
1016 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1017 	bif->bif_savedcaps = ifs->if_capenable;
1018 
1019 	switch (ifs->if_type) {
1020 	case IFT_ETHER:
1021 	case IFT_L2VLAN:
1022 		/*
1023 		 * Place the interface into promiscuous mode.
1024 		 */
1025 		error = ifpromisc(ifs, 1);
1026 		if (error)
1027 			goto out;
1028 		break;
1029 
1030 	case IFT_GIF:
1031 		break;
1032 
1033 	default:
1034 		error = EINVAL;
1035 		goto out;
1036 	}
1037 
1038 	/*
1039 	 * Assign the interface's MAC address to the bridge if it's the first
1040 	 * member and the MAC address of the bridge has not been changed from
1041 	 * the default randomly generated one.
1042 	 */
1043 	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1044 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1045 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1046 		sc->sc_ifaddr = ifs;
1047 	}
1048 
1049 	ifs->if_bridge = sc;
1050 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1051 	/*
1052 	 * XXX: XLOCK HERE!?!
1053 	 *
1054 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1055 	 */
1056 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1057 
1058 	/* Set interface capabilities to the intersection set of all members */
1059 	bridge_mutecaps(sc);
1060 out:
1061 	if (error) {
1062 		if (bif != NULL)
1063 			free(bif, M_DEVBUF);
1064 	}
1065 	return (error);
1066 }
1067 
1068 static int
1069 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1070 {
1071 	struct ifbreq *req = arg;
1072 	struct bridge_iflist *bif;
1073 
1074 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1075 	if (bif == NULL)
1076 		return (ENOENT);
1077 
1078 	bridge_delete_member(sc, bif, 0);
1079 
1080 	return (0);
1081 }
1082 
1083 static int
1084 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1085 {
1086 	struct ifbreq *req = arg;
1087 	struct bridge_iflist *bif;
1088 	struct bstp_port *bp;
1089 
1090 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1091 	if (bif == NULL)
1092 		return (ENOENT);
1093 
1094 	bp = &bif->bif_stp;
1095 	req->ifbr_ifsflags = bif->bif_flags;
1096 	req->ifbr_state = bp->bp_state;
1097 	req->ifbr_priority = bp->bp_priority;
1098 	req->ifbr_path_cost = bp->bp_path_cost;
1099 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1100 	req->ifbr_proto = bp->bp_protover;
1101 	req->ifbr_role = bp->bp_role;
1102 	req->ifbr_stpflags = bp->bp_flags;
1103 	req->ifbr_addrcnt = bif->bif_addrcnt;
1104 	req->ifbr_addrmax = bif->bif_addrmax;
1105 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1106 
1107 	/* Copy STP state options as flags */
1108 	if (bp->bp_operedge)
1109 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1110 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1111 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1112 	if (bp->bp_ptp_link)
1113 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1114 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1115 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1116 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1117 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1118 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1119 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1120 	return (0);
1121 }
1122 
1123 static int
1124 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1125 {
1126 	struct ifbreq *req = arg;
1127 	struct bridge_iflist *bif;
1128 	struct bstp_port *bp;
1129 	int error;
1130 
1131 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1132 	if (bif == NULL)
1133 		return (ENOENT);
1134 	bp = &bif->bif_stp;
1135 
1136 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1137 		/* SPAN is readonly */
1138 		return (EINVAL);
1139 
1140 	if (req->ifbr_ifsflags & IFBIF_STP) {
1141 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1142 			error = bstp_enable(&bif->bif_stp);
1143 			if (error)
1144 				return (error);
1145 		}
1146 	} else {
1147 		if ((bif->bif_flags & IFBIF_STP) != 0)
1148 			bstp_disable(&bif->bif_stp);
1149 	}
1150 
1151 	/* Pass on STP flags */
1152 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1153 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1154 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1155 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1156 
1157 	/* Save the bits relating to the bridge */
1158 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1159 
1160 	return (0);
1161 }
1162 
1163 static int
1164 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1165 {
1166 	struct ifbrparam *param = arg;
1167 
1168 	sc->sc_brtmax = param->ifbrp_csize;
1169 	bridge_rttrim(sc);
1170 
1171 	return (0);
1172 }
1173 
1174 static int
1175 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1176 {
1177 	struct ifbrparam *param = arg;
1178 
1179 	param->ifbrp_csize = sc->sc_brtmax;
1180 
1181 	return (0);
1182 }
1183 
1184 static int
1185 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1186 {
1187 	struct ifbifconf *bifc = arg;
1188 	struct bridge_iflist *bif;
1189 	struct ifbreq breq;
1190 	char *buf, *outbuf;
1191 	int count, buflen, len, error = 0;
1192 
1193 	count = 0;
1194 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1195 		count++;
1196 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1197 		count++;
1198 
1199 	buflen = sizeof(breq) * count;
1200 	if (bifc->ifbic_len == 0) {
1201 		bifc->ifbic_len = buflen;
1202 		return (0);
1203 	}
1204 	BRIDGE_UNLOCK(sc);
1205 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1206 	BRIDGE_LOCK(sc);
1207 
1208 	count = 0;
1209 	buf = outbuf;
1210 	len = min(bifc->ifbic_len, buflen);
1211 	bzero(&breq, sizeof(breq));
1212 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1213 		if (len < sizeof(breq))
1214 			break;
1215 
1216 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1217 		    sizeof(breq.ifbr_ifsname));
1218 		/* Fill in the ifbreq structure */
1219 		error = bridge_ioctl_gifflags(sc, &breq);
1220 		if (error)
1221 			break;
1222 		memcpy(buf, &breq, sizeof(breq));
1223 		count++;
1224 		buf += sizeof(breq);
1225 		len -= sizeof(breq);
1226 	}
1227 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1228 		if (len < sizeof(breq))
1229 			break;
1230 
1231 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1232 		    sizeof(breq.ifbr_ifsname));
1233 		breq.ifbr_ifsflags = bif->bif_flags;
1234 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1235 		memcpy(buf, &breq, sizeof(breq));
1236 		count++;
1237 		buf += sizeof(breq);
1238 		len -= sizeof(breq);
1239 	}
1240 
1241 	BRIDGE_UNLOCK(sc);
1242 	bifc->ifbic_len = sizeof(breq) * count;
1243 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1244 	BRIDGE_LOCK(sc);
1245 	free(outbuf, M_TEMP);
1246 	return (error);
1247 }
1248 
1249 static int
1250 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1251 {
1252 	struct ifbaconf *bac = arg;
1253 	struct bridge_rtnode *brt;
1254 	struct ifbareq bareq;
1255 	char *buf, *outbuf;
1256 	int count, buflen, len, error = 0;
1257 
1258 	if (bac->ifbac_len == 0)
1259 		return (0);
1260 
1261 	count = 0;
1262 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1263 		count++;
1264 	buflen = sizeof(bareq) * count;
1265 
1266 	BRIDGE_UNLOCK(sc);
1267 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1268 	BRIDGE_LOCK(sc);
1269 
1270 	count = 0;
1271 	buf = outbuf;
1272 	len = min(bac->ifbac_len, buflen);
1273 	bzero(&bareq, sizeof(bareq));
1274 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1275 		if (len < sizeof(bareq))
1276 			goto out;
1277 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1278 		    sizeof(bareq.ifba_ifsname));
1279 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1280 		bareq.ifba_vlan = brt->brt_vlan;
1281 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1282 				time_uptime < brt->brt_expire)
1283 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1284 		else
1285 			bareq.ifba_expire = 0;
1286 		bareq.ifba_flags = brt->brt_flags;
1287 
1288 		memcpy(buf, &bareq, sizeof(bareq));
1289 		count++;
1290 		buf += sizeof(bareq);
1291 		len -= sizeof(bareq);
1292 	}
1293 out:
1294 	BRIDGE_UNLOCK(sc);
1295 	bac->ifbac_len = sizeof(bareq) * count;
1296 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1297 	BRIDGE_LOCK(sc);
1298 	free(outbuf, M_TEMP);
1299 	return (error);
1300 }
1301 
1302 static int
1303 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1304 {
1305 	struct ifbareq *req = arg;
1306 	struct bridge_iflist *bif;
1307 	int error;
1308 
1309 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1310 	if (bif == NULL)
1311 		return (ENOENT);
1312 
1313 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1314 	    req->ifba_flags);
1315 
1316 	return (error);
1317 }
1318 
1319 static int
1320 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1321 {
1322 	struct ifbrparam *param = arg;
1323 
1324 	sc->sc_brttimeout = param->ifbrp_ctime;
1325 	return (0);
1326 }
1327 
1328 static int
1329 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1330 {
1331 	struct ifbrparam *param = arg;
1332 
1333 	param->ifbrp_ctime = sc->sc_brttimeout;
1334 	return (0);
1335 }
1336 
1337 static int
1338 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1339 {
1340 	struct ifbareq *req = arg;
1341 
1342 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1343 }
1344 
1345 static int
1346 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1347 {
1348 	struct ifbreq *req = arg;
1349 
1350 	bridge_rtflush(sc, req->ifbr_ifsflags);
1351 	return (0);
1352 }
1353 
1354 static int
1355 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1356 {
1357 	struct ifbrparam *param = arg;
1358 	struct bstp_state *bs = &sc->sc_stp;
1359 
1360 	param->ifbrp_prio = bs->bs_bridge_priority;
1361 	return (0);
1362 }
1363 
1364 static int
1365 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1366 {
1367 	struct ifbrparam *param = arg;
1368 
1369 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1370 }
1371 
1372 static int
1373 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1374 {
1375 	struct ifbrparam *param = arg;
1376 	struct bstp_state *bs = &sc->sc_stp;
1377 
1378 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1379 	return (0);
1380 }
1381 
1382 static int
1383 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1384 {
1385 	struct ifbrparam *param = arg;
1386 
1387 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1388 }
1389 
1390 static int
1391 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1392 {
1393 	struct ifbrparam *param = arg;
1394 	struct bstp_state *bs = &sc->sc_stp;
1395 
1396 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1397 	return (0);
1398 }
1399 
1400 static int
1401 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1402 {
1403 	struct ifbrparam *param = arg;
1404 
1405 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1406 }
1407 
1408 static int
1409 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1410 {
1411 	struct ifbrparam *param = arg;
1412 	struct bstp_state *bs = &sc->sc_stp;
1413 
1414 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1415 	return (0);
1416 }
1417 
1418 static int
1419 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1420 {
1421 	struct ifbrparam *param = arg;
1422 
1423 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1424 }
1425 
1426 static int
1427 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1428 {
1429 	struct ifbreq *req = arg;
1430 	struct bridge_iflist *bif;
1431 
1432 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1433 	if (bif == NULL)
1434 		return (ENOENT);
1435 
1436 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1437 }
1438 
1439 static int
1440 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1441 {
1442 	struct ifbreq *req = arg;
1443 	struct bridge_iflist *bif;
1444 
1445 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1446 	if (bif == NULL)
1447 		return (ENOENT);
1448 
1449 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1450 }
1451 
1452 static int
1453 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1454 {
1455 	struct ifbreq *req = arg;
1456 	struct bridge_iflist *bif;
1457 
1458 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1459 	if (bif == NULL)
1460 		return (ENOENT);
1461 
1462 	bif->bif_addrmax = req->ifbr_addrmax;
1463 	return (0);
1464 }
1465 
1466 static int
1467 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1468 {
1469 	struct ifbreq *req = arg;
1470 	struct bridge_iflist *bif = NULL;
1471 	struct ifnet *ifs;
1472 
1473 	ifs = ifunit(req->ifbr_ifsname);
1474 	if (ifs == NULL)
1475 		return (ENOENT);
1476 
1477 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1478 		if (ifs == bif->bif_ifp)
1479 			return (EBUSY);
1480 
1481 	if (ifs->if_bridge != NULL)
1482 		return (EBUSY);
1483 
1484 	switch (ifs->if_type) {
1485 		case IFT_ETHER:
1486 		case IFT_GIF:
1487 		case IFT_L2VLAN:
1488 			break;
1489 		default:
1490 			return (EINVAL);
1491 	}
1492 
1493 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1494 	if (bif == NULL)
1495 		return (ENOMEM);
1496 
1497 	bif->bif_ifp = ifs;
1498 	bif->bif_flags = IFBIF_SPAN;
1499 
1500 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1501 
1502 	return (0);
1503 }
1504 
1505 static int
1506 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1507 {
1508 	struct ifbreq *req = arg;
1509 	struct bridge_iflist *bif;
1510 	struct ifnet *ifs;
1511 
1512 	ifs = ifunit(req->ifbr_ifsname);
1513 	if (ifs == NULL)
1514 		return (ENOENT);
1515 
1516 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1517 		if (ifs == bif->bif_ifp)
1518 			break;
1519 
1520 	if (bif == NULL)
1521 		return (ENOENT);
1522 
1523 	bridge_delete_span(sc, bif);
1524 
1525 	return (0);
1526 }
1527 
1528 static int
1529 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1530 {
1531 	struct ifbropreq *req = arg;
1532 	struct bstp_state *bs = &sc->sc_stp;
1533 	struct bstp_port *root_port;
1534 
1535 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1536 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1537 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1538 
1539 	root_port = bs->bs_root_port;
1540 	if (root_port == NULL)
1541 		req->ifbop_root_port = 0;
1542 	else
1543 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1544 
1545 	req->ifbop_holdcount = bs->bs_txholdcount;
1546 	req->ifbop_priority = bs->bs_bridge_priority;
1547 	req->ifbop_protocol = bs->bs_protover;
1548 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1549 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1550 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1551 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1552 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1553 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1554 
1555 	return (0);
1556 }
1557 
1558 static int
1559 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1560 {
1561 	struct ifbrparam *param = arg;
1562 
1563 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1564 	return (0);
1565 }
1566 
1567 static int
1568 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1569 {
1570 	struct ifbpstpconf *bifstp = arg;
1571 	struct bridge_iflist *bif;
1572 	struct bstp_port *bp;
1573 	struct ifbpstpreq bpreq;
1574 	char *buf, *outbuf;
1575 	int count, buflen, len, error = 0;
1576 
1577 	count = 0;
1578 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1579 		if ((bif->bif_flags & IFBIF_STP) != 0)
1580 			count++;
1581 	}
1582 
1583 	buflen = sizeof(bpreq) * count;
1584 	if (bifstp->ifbpstp_len == 0) {
1585 		bifstp->ifbpstp_len = buflen;
1586 		return (0);
1587 	}
1588 
1589 	BRIDGE_UNLOCK(sc);
1590 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1591 	BRIDGE_LOCK(sc);
1592 
1593 	count = 0;
1594 	buf = outbuf;
1595 	len = min(bifstp->ifbpstp_len, buflen);
1596 	bzero(&bpreq, sizeof(bpreq));
1597 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1598 		if (len < sizeof(bpreq))
1599 			break;
1600 
1601 		if ((bif->bif_flags & IFBIF_STP) == 0)
1602 			continue;
1603 
1604 		bp = &bif->bif_stp;
1605 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1606 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1607 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1608 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1609 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1610 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1611 
1612 		memcpy(buf, &bpreq, sizeof(bpreq));
1613 		count++;
1614 		buf += sizeof(bpreq);
1615 		len -= sizeof(bpreq);
1616 	}
1617 
1618 	BRIDGE_UNLOCK(sc);
1619 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1620 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1621 	BRIDGE_LOCK(sc);
1622 	free(outbuf, M_TEMP);
1623 	return (error);
1624 }
1625 
1626 static int
1627 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1628 {
1629 	struct ifbrparam *param = arg;
1630 
1631 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1632 }
1633 
1634 static int
1635 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1636 {
1637 	struct ifbrparam *param = arg;
1638 
1639 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1640 }
1641 
1642 /*
1643  * bridge_ifdetach:
1644  *
1645  *	Detach an interface from a bridge.  Called when a member
1646  *	interface is detaching.
1647  */
1648 static void
1649 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1650 {
1651 	struct bridge_softc *sc = ifp->if_bridge;
1652 	struct bridge_iflist *bif;
1653 
1654 	/* Check if the interface is a bridge member */
1655 	if (sc != NULL) {
1656 		BRIDGE_LOCK(sc);
1657 
1658 		bif = bridge_lookup_member_if(sc, ifp);
1659 		if (bif != NULL)
1660 			bridge_delete_member(sc, bif, 1);
1661 
1662 		BRIDGE_UNLOCK(sc);
1663 		return;
1664 	}
1665 
1666 	/* Check if the interface is a span port */
1667 	mtx_lock(&bridge_list_mtx);
1668 	LIST_FOREACH(sc, &bridge_list, sc_list) {
1669 		BRIDGE_LOCK(sc);
1670 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1671 			if (ifp == bif->bif_ifp) {
1672 				bridge_delete_span(sc, bif);
1673 				break;
1674 			}
1675 
1676 		BRIDGE_UNLOCK(sc);
1677 	}
1678 	mtx_unlock(&bridge_list_mtx);
1679 }
1680 
1681 /*
1682  * bridge_init:
1683  *
1684  *	Initialize a bridge interface.
1685  */
1686 static void
1687 bridge_init(void *xsc)
1688 {
1689 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1690 	struct ifnet *ifp = sc->sc_ifp;
1691 
1692 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1693 		return;
1694 
1695 	BRIDGE_LOCK(sc);
1696 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1697 	    bridge_timer, sc);
1698 
1699 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1700 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1701 
1702 	BRIDGE_UNLOCK(sc);
1703 }
1704 
1705 /*
1706  * bridge_stop:
1707  *
1708  *	Stop the bridge interface.
1709  */
1710 static void
1711 bridge_stop(struct ifnet *ifp, int disable)
1712 {
1713 	struct bridge_softc *sc = ifp->if_softc;
1714 
1715 	BRIDGE_LOCK_ASSERT(sc);
1716 
1717 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1718 		return;
1719 
1720 	callout_stop(&sc->sc_brcallout);
1721 	bstp_stop(&sc->sc_stp);
1722 
1723 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1724 
1725 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1726 }
1727 
1728 /*
1729  * bridge_enqueue:
1730  *
1731  *	Enqueue a packet on a bridge member interface.
1732  *
1733  */
1734 static void
1735 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1736 {
1737 	int len, err = 0;
1738 	short mflags;
1739 	struct mbuf *m0;
1740 
1741 	len = m->m_pkthdr.len;
1742 	mflags = m->m_flags;
1743 
1744 	/* We may be sending a fragment so traverse the mbuf */
1745 	for (; m; m = m0) {
1746 		m0 = m->m_nextpkt;
1747 		m->m_nextpkt = NULL;
1748 
1749 		/*
1750 		 * If underlying interface can not do VLAN tag insertion itself
1751 		 * then attach a packet tag that holds it.
1752 		 */
1753 		if ((m->m_flags & M_VLANTAG) &&
1754 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1755 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1756 			if (m == NULL) {
1757 				if_printf(dst_ifp,
1758 				    "unable to prepend VLAN header\n");
1759 				dst_ifp->if_oerrors++;
1760 				continue;
1761 			}
1762 			m->m_flags &= ~M_VLANTAG;
1763 		}
1764 
1765 		if (err == 0)
1766 			IFQ_ENQUEUE(&dst_ifp->if_snd, m, err);
1767 	}
1768 
1769 	if (err == 0) {
1770 
1771 		sc->sc_ifp->if_opackets++;
1772 		sc->sc_ifp->if_obytes += len;
1773 
1774 		dst_ifp->if_obytes += len;
1775 
1776 		if (mflags & M_MCAST) {
1777 			sc->sc_ifp->if_omcasts++;
1778 			dst_ifp->if_omcasts++;
1779 		}
1780 	}
1781 
1782 	if ((dst_ifp->if_drv_flags & IFF_DRV_OACTIVE) == 0)
1783 		(*dst_ifp->if_start)(dst_ifp);
1784 }
1785 
1786 /*
1787  * bridge_dummynet:
1788  *
1789  * 	Receive a queued packet from dummynet and pass it on to the output
1790  * 	interface.
1791  *
1792  *	The mbuf has the Ethernet header already attached.
1793  */
1794 static void
1795 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1796 {
1797 	struct bridge_softc *sc;
1798 
1799 	sc = ifp->if_bridge;
1800 
1801 	/*
1802 	 * The packet didnt originate from a member interface. This should only
1803 	 * ever happen if a member interface is removed while packets are
1804 	 * queued for it.
1805 	 */
1806 	if (sc == NULL) {
1807 		m_freem(m);
1808 		return;
1809 	}
1810 
1811 	if (PFIL_HOOKED(&inet_pfil_hook)
1812 #ifdef INET6
1813 	    || PFIL_HOOKED(&inet6_pfil_hook)
1814 #endif
1815 	    ) {
1816 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1817 			return;
1818 		if (m == NULL)
1819 			return;
1820 	}
1821 
1822 	bridge_enqueue(sc, ifp, m);
1823 }
1824 
1825 /*
1826  * bridge_output:
1827  *
1828  *	Send output from a bridge member interface.  This
1829  *	performs the bridging function for locally originated
1830  *	packets.
1831  *
1832  *	The mbuf has the Ethernet header already attached.  We must
1833  *	enqueue or free the mbuf before returning.
1834  */
1835 static int
1836 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1837     struct rtentry *rt)
1838 {
1839 	struct ether_header *eh;
1840 	struct ifnet *dst_if;
1841 	struct bridge_softc *sc;
1842 	uint16_t vlan;
1843 
1844 	if (m->m_len < ETHER_HDR_LEN) {
1845 		m = m_pullup(m, ETHER_HDR_LEN);
1846 		if (m == NULL)
1847 			return (0);
1848 	}
1849 
1850 	eh = mtod(m, struct ether_header *);
1851 	sc = ifp->if_bridge;
1852 	vlan = VLANTAGOF(m);
1853 
1854 	BRIDGE_LOCK(sc);
1855 
1856 	/*
1857 	 * If bridge is down, but the original output interface is up,
1858 	 * go ahead and send out that interface.  Otherwise, the packet
1859 	 * is dropped below.
1860 	 */
1861 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1862 		dst_if = ifp;
1863 		goto sendunicast;
1864 	}
1865 
1866 	/*
1867 	 * If the packet is a multicast, or we don't know a better way to
1868 	 * get there, send to all interfaces.
1869 	 */
1870 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1871 		dst_if = NULL;
1872 	else
1873 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1874 	if (dst_if == NULL) {
1875 		struct bridge_iflist *bif;
1876 		struct mbuf *mc;
1877 		int error = 0, used = 0;
1878 
1879 		bridge_span(sc, m);
1880 
1881 		BRIDGE_LOCK2REF(sc, error);
1882 		if (error) {
1883 			m_freem(m);
1884 			return (0);
1885 		}
1886 
1887 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1888 			dst_if = bif->bif_ifp;
1889 
1890 			if (dst_if->if_type == IFT_GIF)
1891 				continue;
1892 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1893 				continue;
1894 
1895 			/*
1896 			 * If this is not the original output interface,
1897 			 * and the interface is participating in spanning
1898 			 * tree, make sure the port is in a state that
1899 			 * allows forwarding.
1900 			 */
1901 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1902 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1903 				continue;
1904 
1905 			if (LIST_NEXT(bif, bif_next) == NULL) {
1906 				used = 1;
1907 				mc = m;
1908 			} else {
1909 				mc = m_copypacket(m, M_DONTWAIT);
1910 				if (mc == NULL) {
1911 					sc->sc_ifp->if_oerrors++;
1912 					continue;
1913 				}
1914 			}
1915 
1916 			bridge_enqueue(sc, dst_if, mc);
1917 		}
1918 		if (used == 0)
1919 			m_freem(m);
1920 		BRIDGE_UNREF(sc);
1921 		return (0);
1922 	}
1923 
1924 sendunicast:
1925 	/*
1926 	 * XXX Spanning tree consideration here?
1927 	 */
1928 
1929 	bridge_span(sc, m);
1930 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1931 		m_freem(m);
1932 		BRIDGE_UNLOCK(sc);
1933 		return (0);
1934 	}
1935 
1936 	BRIDGE_UNLOCK(sc);
1937 	bridge_enqueue(sc, dst_if, m);
1938 	return (0);
1939 }
1940 
1941 /*
1942  * bridge_start:
1943  *
1944  *	Start output on a bridge.
1945  *
1946  */
1947 static void
1948 bridge_start(struct ifnet *ifp)
1949 {
1950 	struct bridge_softc *sc;
1951 	struct mbuf *m;
1952 	struct ether_header *eh;
1953 	struct ifnet *dst_if;
1954 
1955 	sc = ifp->if_softc;
1956 
1957 	ifp->if_drv_flags |= IFF_DRV_OACTIVE;
1958 	for (;;) {
1959 		IFQ_DEQUEUE(&ifp->if_snd, m);
1960 		if (m == 0)
1961 			break;
1962 		ETHER_BPF_MTAP(ifp, m);
1963 
1964 		eh = mtod(m, struct ether_header *);
1965 		dst_if = NULL;
1966 
1967 		BRIDGE_LOCK(sc);
1968 		if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
1969 			dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1);
1970 		}
1971 
1972 		if (dst_if == NULL)
1973 			bridge_broadcast(sc, ifp, m, 0);
1974 		else {
1975 			BRIDGE_UNLOCK(sc);
1976 			bridge_enqueue(sc, dst_if, m);
1977 		}
1978 	}
1979 	ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
1980 }
1981 
1982 /*
1983  * bridge_forward:
1984  *
1985  *	The forwarding function of the bridge.
1986  *
1987  *	NOTE: Releases the lock on return.
1988  */
1989 static void
1990 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
1991     struct mbuf *m)
1992 {
1993 	struct bridge_iflist *dbif;
1994 	struct ifnet *src_if, *dst_if, *ifp;
1995 	struct ether_header *eh;
1996 	uint16_t vlan;
1997 	uint8_t *dst;
1998 	int error;
1999 
2000 	src_if = m->m_pkthdr.rcvif;
2001 	ifp = sc->sc_ifp;
2002 
2003 	ifp->if_ipackets++;
2004 	ifp->if_ibytes += m->m_pkthdr.len;
2005 	vlan = VLANTAGOF(m);
2006 
2007 	if ((sbif->bif_flags & IFBIF_STP) &&
2008 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2009 		goto drop;
2010 
2011 	eh = mtod(m, struct ether_header *);
2012 	dst = eh->ether_dhost;
2013 
2014 	/* If the interface is learning, record the address. */
2015 	if (sbif->bif_flags & IFBIF_LEARNING) {
2016 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2017 		    sbif, 0, IFBAF_DYNAMIC);
2018 		/*
2019 		 * If the interface has addresses limits then deny any source
2020 		 * that is not in the cache.
2021 		 */
2022 		if (error && sbif->bif_addrmax)
2023 			goto drop;
2024 	}
2025 
2026 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2027 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2028 		goto drop;
2029 
2030 	/*
2031 	 * At this point, the port either doesn't participate
2032 	 * in spanning tree or it is in the forwarding state.
2033 	 */
2034 
2035 	/*
2036 	 * If the packet is unicast, destined for someone on
2037 	 * "this" side of the bridge, drop it.
2038 	 */
2039 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2040 		dst_if = bridge_rtlookup(sc, dst, vlan);
2041 		if (src_if == dst_if)
2042 			goto drop;
2043 	} else {
2044 		/*
2045 		 * Check if its a reserved multicast address, any address
2046 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2047 		 * bridge.
2048 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2049 		 */
2050 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2051 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2052 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2053 			goto drop;
2054 
2055 		/* ...forward it to all interfaces. */
2056 		ifp->if_imcasts++;
2057 		dst_if = NULL;
2058 	}
2059 
2060 	/*
2061 	 * If we have a destination interface which is a member of our bridge,
2062 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2063 	 * For broadcast or multicast packets, don't bother because it will
2064 	 * be reinjected into ether_input. We do this before we pass the packets
2065 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2066 	 * drop the packet, or possibly modify it, making it difficult to debug
2067 	 * firewall issues on the bridge.
2068 	 */
2069 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2070 		ETHER_BPF_MTAP(ifp, m);
2071 
2072 	/* run the packet filter */
2073 	if (PFIL_HOOKED(&inet_pfil_hook)
2074 #ifdef INET6
2075 	    || PFIL_HOOKED(&inet6_pfil_hook)
2076 #endif
2077 	    ) {
2078 		BRIDGE_UNLOCK(sc);
2079 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2080 			return;
2081 		if (m == NULL)
2082 			return;
2083 		BRIDGE_LOCK(sc);
2084 	}
2085 
2086 	if (dst_if == NULL) {
2087 		bridge_broadcast(sc, src_if, m, 1);
2088 		return;
2089 	}
2090 
2091 	/*
2092 	 * At this point, we're dealing with a unicast frame
2093 	 * going to a different interface.
2094 	 */
2095 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2096 		goto drop;
2097 
2098 	dbif = bridge_lookup_member_if(sc, dst_if);
2099 	if (dbif == NULL)
2100 		/* Not a member of the bridge (anymore?) */
2101 		goto drop;
2102 
2103 	/* Private segments can not talk to each other */
2104 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2105 		goto drop;
2106 
2107 	if ((dbif->bif_flags & IFBIF_STP) &&
2108 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2109 		goto drop;
2110 
2111 	BRIDGE_UNLOCK(sc);
2112 
2113 	if (PFIL_HOOKED(&inet_pfil_hook)
2114 #ifdef INET6
2115 	    || PFIL_HOOKED(&inet6_pfil_hook)
2116 #endif
2117 	    ) {
2118 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2119 			return;
2120 		if (m == NULL)
2121 			return;
2122 	}
2123 
2124 	bridge_enqueue(sc, dst_if, m);
2125 	return;
2126 
2127 drop:
2128 	BRIDGE_UNLOCK(sc);
2129 	m_freem(m);
2130 }
2131 
2132 /*
2133  * bridge_input:
2134  *
2135  *	Receive input from a member interface.  Queue the packet for
2136  *	bridging if it is not for us.
2137  */
2138 static struct mbuf *
2139 bridge_input(struct ifnet *ifp, struct mbuf *m)
2140 {
2141 	struct bridge_softc *sc = ifp->if_bridge;
2142 	struct bridge_iflist *bif, *bif2;
2143 	struct ifnet *bifp;
2144 	struct ether_header *eh;
2145 	struct mbuf *mc, *mc2;
2146 	uint16_t vlan;
2147 	int error;
2148 
2149 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2150 		return (m);
2151 
2152 	bifp = sc->sc_ifp;
2153 	vlan = VLANTAGOF(m);
2154 
2155 	/*
2156 	 * Implement support for bridge monitoring. If this flag has been
2157 	 * set on this interface, discard the packet once we push it through
2158 	 * the bpf(4) machinery, but before we do, increment the byte and
2159 	 * packet counters associated with this interface.
2160 	 */
2161 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2162 		m->m_pkthdr.rcvif  = bifp;
2163 		ETHER_BPF_MTAP(bifp, m);
2164 		bifp->if_ipackets++;
2165 		bifp->if_ibytes += m->m_pkthdr.len;
2166 		m_freem(m);
2167 		return (NULL);
2168 	}
2169 	BRIDGE_LOCK(sc);
2170 	bif = bridge_lookup_member_if(sc, ifp);
2171 	if (bif == NULL) {
2172 		BRIDGE_UNLOCK(sc);
2173 		return (m);
2174 	}
2175 
2176 	eh = mtod(m, struct ether_header *);
2177 
2178 	bridge_span(sc, m);
2179 
2180 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2181 		/* Tap off 802.1D packets; they do not get forwarded. */
2182 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2183 		    ETHER_ADDR_LEN) == 0) {
2184 			m = bstp_input(&bif->bif_stp, ifp, m);
2185 			if (m == NULL) {
2186 				BRIDGE_UNLOCK(sc);
2187 				return (NULL);
2188 			}
2189 		}
2190 
2191 		if ((bif->bif_flags & IFBIF_STP) &&
2192 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2193 			BRIDGE_UNLOCK(sc);
2194 			return (m);
2195 		}
2196 
2197 		/*
2198 		 * Make a deep copy of the packet and enqueue the copy
2199 		 * for bridge processing; return the original packet for
2200 		 * local processing.
2201 		 */
2202 		mc = m_dup(m, M_DONTWAIT);
2203 		if (mc == NULL) {
2204 			BRIDGE_UNLOCK(sc);
2205 			return (m);
2206 		}
2207 
2208 		/* Perform the bridge forwarding function with the copy. */
2209 		bridge_forward(sc, bif, mc);
2210 
2211 		/*
2212 		 * Reinject the mbuf as arriving on the bridge so we have a
2213 		 * chance at claiming multicast packets. We can not loop back
2214 		 * here from ether_input as a bridge is never a member of a
2215 		 * bridge.
2216 		 */
2217 		KASSERT(bifp->if_bridge == NULL,
2218 		    ("loop created in bridge_input"));
2219 		mc2 = m_dup(m, M_DONTWAIT);
2220 		if (mc2 != NULL) {
2221 			/* Keep the layer3 header aligned */
2222 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2223 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2224 		}
2225 		if (mc2 != NULL) {
2226 			mc2->m_pkthdr.rcvif = bifp;
2227 			(*bifp->if_input)(bifp, mc2);
2228 		}
2229 
2230 		/* Return the original packet for local processing. */
2231 		return (m);
2232 	}
2233 
2234 	if ((bif->bif_flags & IFBIF_STP) &&
2235 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2236 		BRIDGE_UNLOCK(sc);
2237 		return (m);
2238 	}
2239 
2240 #ifdef DEV_CARP
2241 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2242 	|| ((iface)->if_carp \
2243 	    && carp_forus((iface)->if_carp, eh->ether_dhost))
2244 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2245 	|| ((iface)->if_carp \
2246 	    && carp_forus((iface)->if_carp, eh->ether_shost))
2247 #else
2248 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2249 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2250 #endif
2251 
2252 #ifdef INET6
2253 #   define OR_PFIL_HOOKED_INET6 \
2254 	|| PFIL_HOOKED(&inet6_pfil_hook)
2255 #else
2256 #   define OR_PFIL_HOOKED_INET6
2257 #endif
2258 
2259 #define GRAB_OUR_PACKETS(iface) \
2260 	if ((iface)->if_type == IFT_GIF) \
2261 		continue; \
2262 	/* It is destined for us. */ \
2263 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2264 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2265 	    ) {								\
2266 		if ((iface)->if_type == IFT_BRIDGE) {			\
2267 			ETHER_BPF_MTAP(iface, m);			\
2268 			iface->if_ipackets++;				\
2269 			/* Filter on the physical interface. */		\
2270 			if (pfil_local_phys &&				\
2271 			    (PFIL_HOOKED(&inet_pfil_hook)		\
2272 			     OR_PFIL_HOOKED_INET6)) {			\
2273 				if (bridge_pfil(&m, NULL, ifp,		\
2274 				    PFIL_IN) != 0 || m == NULL) {	\
2275 					BRIDGE_UNLOCK(sc);		\
2276 					return (NULL);			\
2277 				}					\
2278 			}						\
2279 		}							\
2280 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2281 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2282 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2283 			if (error && bif->bif_addrmax) {		\
2284 				BRIDGE_UNLOCK(sc);			\
2285 				m_freem(m);				\
2286 				return (NULL);				\
2287 			}						\
2288 		}							\
2289 		m->m_pkthdr.rcvif = iface;				\
2290 		BRIDGE_UNLOCK(sc);					\
2291 		return (m);						\
2292 	}								\
2293 									\
2294 	/* We just received a packet that we sent out. */		\
2295 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2296 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2297 	    ) {								\
2298 		BRIDGE_UNLOCK(sc);					\
2299 		m_freem(m);						\
2300 		return (NULL);						\
2301 	}
2302 
2303 	/*
2304 	 * Unicast.  Make sure it's not for the bridge.
2305 	 */
2306 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2307 
2308 	/*
2309 	 * Give a chance for ifp at first priority. This will help when	the
2310 	 * packet comes through the interface like VLAN's with the same MACs
2311 	 * on several interfaces from the same bridge. This also will save
2312 	 * some CPU cycles in case the destination interface and the input
2313 	 * interface (eq ifp) are the same.
2314 	 */
2315 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2316 
2317 	/* Now check the all bridge members. */
2318 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2319 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2320 	}
2321 
2322 #undef OR_CARP_CHECK_WE_ARE_DST
2323 #undef OR_CARP_CHECK_WE_ARE_SRC
2324 #undef OR_PFIL_HOOKED_INET6
2325 #undef GRAB_OUR_PACKETS
2326 
2327 	/* Perform the bridge forwarding function. */
2328 	bridge_forward(sc, bif, m);
2329 
2330 	return (NULL);
2331 }
2332 
2333 /*
2334  * bridge_broadcast:
2335  *
2336  *	Send a frame to all interfaces that are members of
2337  *	the bridge, except for the one on which the packet
2338  *	arrived.
2339  *
2340  *	NOTE: Releases the lock on return.
2341  */
2342 static void
2343 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2344     struct mbuf *m, int runfilt)
2345 {
2346 	struct bridge_iflist *dbif, *sbif;
2347 	struct mbuf *mc;
2348 	struct ifnet *dst_if;
2349 	int error = 0, used = 0, i;
2350 
2351 	sbif = bridge_lookup_member_if(sc, src_if);
2352 
2353 	BRIDGE_LOCK2REF(sc, error);
2354 	if (error) {
2355 		m_freem(m);
2356 		return;
2357 	}
2358 
2359 	/* Filter on the bridge interface before broadcasting */
2360 	if (runfilt && (PFIL_HOOKED(&inet_pfil_hook)
2361 #ifdef INET6
2362 	    || PFIL_HOOKED(&inet6_pfil_hook)
2363 #endif
2364 	    )) {
2365 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2366 			goto out;
2367 		if (m == NULL)
2368 			goto out;
2369 	}
2370 
2371 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2372 		dst_if = dbif->bif_ifp;
2373 		if (dst_if == src_if)
2374 			continue;
2375 
2376 		/* Private segments can not talk to each other */
2377 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2378 			continue;
2379 
2380 		if ((dbif->bif_flags & IFBIF_STP) &&
2381 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2382 			continue;
2383 
2384 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2385 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2386 			continue;
2387 
2388 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2389 			continue;
2390 
2391 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2392 			mc = m;
2393 			used = 1;
2394 		} else {
2395 			mc = m_dup(m, M_DONTWAIT);
2396 			if (mc == NULL) {
2397 				sc->sc_ifp->if_oerrors++;
2398 				continue;
2399 			}
2400 		}
2401 
2402 		/*
2403 		 * Filter on the output interface. Pass a NULL bridge interface
2404 		 * pointer so we do not redundantly filter on the bridge for
2405 		 * each interface we broadcast on.
2406 		 */
2407 		if (runfilt && (PFIL_HOOKED(&inet_pfil_hook)
2408 #ifdef INET6
2409 		    || PFIL_HOOKED(&inet6_pfil_hook)
2410 #endif
2411 		    )) {
2412 			if (used == 0) {
2413 				/* Keep the layer3 header aligned */
2414 				i = min(mc->m_pkthdr.len, max_protohdr);
2415 				mc = m_copyup(mc, i, ETHER_ALIGN);
2416 				if (mc == NULL) {
2417 					sc->sc_ifp->if_oerrors++;
2418 					continue;
2419 				}
2420 			}
2421 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2422 				continue;
2423 			if (mc == NULL)
2424 				continue;
2425 		}
2426 
2427 		bridge_enqueue(sc, dst_if, mc);
2428 	}
2429 	if (used == 0)
2430 		m_freem(m);
2431 
2432 out:
2433 	BRIDGE_UNREF(sc);
2434 }
2435 
2436 /*
2437  * bridge_span:
2438  *
2439  *	Duplicate a packet out one or more interfaces that are in span mode,
2440  *	the original mbuf is unmodified.
2441  */
2442 static void
2443 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2444 {
2445 	struct bridge_iflist *bif;
2446 	struct ifnet *dst_if;
2447 	struct mbuf *mc;
2448 
2449 	if (LIST_EMPTY(&sc->sc_spanlist))
2450 		return;
2451 
2452 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2453 		dst_if = bif->bif_ifp;
2454 
2455 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2456 			continue;
2457 
2458 		mc = m_copypacket(m, M_DONTWAIT);
2459 		if (mc == NULL) {
2460 			sc->sc_ifp->if_oerrors++;
2461 			continue;
2462 		}
2463 
2464 		bridge_enqueue(sc, dst_if, mc);
2465 	}
2466 }
2467 
2468 /*
2469  * bridge_rtupdate:
2470  *
2471  *	Add a bridge routing entry.
2472  */
2473 static int
2474 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2475     struct bridge_iflist *bif, int setflags, uint8_t flags)
2476 {
2477 	struct bridge_rtnode *brt;
2478 	int error;
2479 
2480 	BRIDGE_LOCK_ASSERT(sc);
2481 
2482 	/* Check the source address is valid and not multicast. */
2483 	if (ETHER_IS_MULTICAST(dst) ||
2484 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2485 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2486 		return (EINVAL);
2487 
2488 	/* 802.1p frames map to vlan 1 */
2489 	if (vlan == 0)
2490 		vlan = 1;
2491 
2492 	/*
2493 	 * A route for this destination might already exist.  If so,
2494 	 * update it, otherwise create a new one.
2495 	 */
2496 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2497 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2498 			sc->sc_brtexceeded++;
2499 			return (ENOSPC);
2500 		}
2501 		/* Check per interface address limits (if enabled) */
2502 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2503 			bif->bif_addrexceeded++;
2504 			return (ENOSPC);
2505 		}
2506 
2507 		/*
2508 		 * Allocate a new bridge forwarding node, and
2509 		 * initialize the expiration time and Ethernet
2510 		 * address.
2511 		 */
2512 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2513 		if (brt == NULL)
2514 			return (ENOMEM);
2515 
2516 		if (bif->bif_flags & IFBIF_STICKY)
2517 			brt->brt_flags = IFBAF_STICKY;
2518 		else
2519 			brt->brt_flags = IFBAF_DYNAMIC;
2520 
2521 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2522 		brt->brt_vlan = vlan;
2523 
2524 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2525 			uma_zfree(bridge_rtnode_zone, brt);
2526 			return (error);
2527 		}
2528 		brt->brt_dst = bif;
2529 		bif->bif_addrcnt++;
2530 	}
2531 
2532 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2533 	    brt->brt_dst != bif) {
2534 		brt->brt_dst->bif_addrcnt--;
2535 		brt->brt_dst = bif;
2536 		brt->brt_dst->bif_addrcnt++;
2537 	}
2538 
2539 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2540 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2541 	if (setflags)
2542 		brt->brt_flags = flags;
2543 
2544 	return (0);
2545 }
2546 
2547 /*
2548  * bridge_rtlookup:
2549  *
2550  *	Lookup the destination interface for an address.
2551  */
2552 static struct ifnet *
2553 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2554 {
2555 	struct bridge_rtnode *brt;
2556 
2557 	BRIDGE_LOCK_ASSERT(sc);
2558 
2559 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2560 		return (NULL);
2561 
2562 	return (brt->brt_ifp);
2563 }
2564 
2565 /*
2566  * bridge_rttrim:
2567  *
2568  *	Trim the routine table so that we have a number
2569  *	of routing entries less than or equal to the
2570  *	maximum number.
2571  */
2572 static void
2573 bridge_rttrim(struct bridge_softc *sc)
2574 {
2575 	struct bridge_rtnode *brt, *nbrt;
2576 
2577 	BRIDGE_LOCK_ASSERT(sc);
2578 
2579 	/* Make sure we actually need to do this. */
2580 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2581 		return;
2582 
2583 	/* Force an aging cycle; this might trim enough addresses. */
2584 	bridge_rtage(sc);
2585 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2586 		return;
2587 
2588 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2589 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2590 			bridge_rtnode_destroy(sc, brt);
2591 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2592 				return;
2593 		}
2594 	}
2595 }
2596 
2597 /*
2598  * bridge_timer:
2599  *
2600  *	Aging timer for the bridge.
2601  */
2602 static void
2603 bridge_timer(void *arg)
2604 {
2605 	struct bridge_softc *sc = arg;
2606 
2607 	BRIDGE_LOCK_ASSERT(sc);
2608 
2609 	bridge_rtage(sc);
2610 
2611 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2612 		callout_reset(&sc->sc_brcallout,
2613 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2614 }
2615 
2616 /*
2617  * bridge_rtage:
2618  *
2619  *	Perform an aging cycle.
2620  */
2621 static void
2622 bridge_rtage(struct bridge_softc *sc)
2623 {
2624 	struct bridge_rtnode *brt, *nbrt;
2625 
2626 	BRIDGE_LOCK_ASSERT(sc);
2627 
2628 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2629 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2630 			if (time_uptime >= brt->brt_expire)
2631 				bridge_rtnode_destroy(sc, brt);
2632 		}
2633 	}
2634 }
2635 
2636 /*
2637  * bridge_rtflush:
2638  *
2639  *	Remove all dynamic addresses from the bridge.
2640  */
2641 static void
2642 bridge_rtflush(struct bridge_softc *sc, int full)
2643 {
2644 	struct bridge_rtnode *brt, *nbrt;
2645 
2646 	BRIDGE_LOCK_ASSERT(sc);
2647 
2648 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2649 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2650 			bridge_rtnode_destroy(sc, brt);
2651 	}
2652 }
2653 
2654 /*
2655  * bridge_rtdaddr:
2656  *
2657  *	Remove an address from the table.
2658  */
2659 static int
2660 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2661 {
2662 	struct bridge_rtnode *brt;
2663 	int found = 0;
2664 
2665 	BRIDGE_LOCK_ASSERT(sc);
2666 
2667 	/*
2668 	 * If vlan is zero then we want to delete for all vlans so the lookup
2669 	 * may return more than one.
2670 	 */
2671 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2672 		bridge_rtnode_destroy(sc, brt);
2673 		found = 1;
2674 	}
2675 
2676 	return (found ? 0 : ENOENT);
2677 }
2678 
2679 /*
2680  * bridge_rtdelete:
2681  *
2682  *	Delete routes to a speicifc member interface.
2683  */
2684 static void
2685 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2686 {
2687 	struct bridge_rtnode *brt, *nbrt;
2688 
2689 	BRIDGE_LOCK_ASSERT(sc);
2690 
2691 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2692 		if (brt->brt_ifp == ifp && (full ||
2693 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2694 			bridge_rtnode_destroy(sc, brt);
2695 	}
2696 }
2697 
2698 /*
2699  * bridge_rtable_init:
2700  *
2701  *	Initialize the route table for this bridge.
2702  */
2703 static int
2704 bridge_rtable_init(struct bridge_softc *sc)
2705 {
2706 	int i;
2707 
2708 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2709 	    M_DEVBUF, M_NOWAIT);
2710 	if (sc->sc_rthash == NULL)
2711 		return (ENOMEM);
2712 
2713 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2714 		LIST_INIT(&sc->sc_rthash[i]);
2715 
2716 	sc->sc_rthash_key = arc4random();
2717 
2718 	LIST_INIT(&sc->sc_rtlist);
2719 
2720 	return (0);
2721 }
2722 
2723 /*
2724  * bridge_rtable_fini:
2725  *
2726  *	Deconstruct the route table for this bridge.
2727  */
2728 static void
2729 bridge_rtable_fini(struct bridge_softc *sc)
2730 {
2731 
2732 	KASSERT(sc->sc_brtcnt == 0,
2733 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2734 	free(sc->sc_rthash, M_DEVBUF);
2735 }
2736 
2737 /*
2738  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2739  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2740  */
2741 #define	mix(a, b, c)							\
2742 do {									\
2743 	a -= b; a -= c; a ^= (c >> 13);					\
2744 	b -= c; b -= a; b ^= (a << 8);					\
2745 	c -= a; c -= b; c ^= (b >> 13);					\
2746 	a -= b; a -= c; a ^= (c >> 12);					\
2747 	b -= c; b -= a; b ^= (a << 16);					\
2748 	c -= a; c -= b; c ^= (b >> 5);					\
2749 	a -= b; a -= c; a ^= (c >> 3);					\
2750 	b -= c; b -= a; b ^= (a << 10);					\
2751 	c -= a; c -= b; c ^= (b >> 15);					\
2752 } while (/*CONSTCOND*/0)
2753 
2754 static __inline uint32_t
2755 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2756 {
2757 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2758 
2759 	b += addr[5] << 8;
2760 	b += addr[4];
2761 	a += addr[3] << 24;
2762 	a += addr[2] << 16;
2763 	a += addr[1] << 8;
2764 	a += addr[0];
2765 
2766 	mix(a, b, c);
2767 
2768 	return (c & BRIDGE_RTHASH_MASK);
2769 }
2770 
2771 #undef mix
2772 
2773 static int
2774 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2775 {
2776 	int i, d;
2777 
2778 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2779 		d = ((int)a[i]) - ((int)b[i]);
2780 	}
2781 
2782 	return (d);
2783 }
2784 
2785 /*
2786  * bridge_rtnode_lookup:
2787  *
2788  *	Look up a bridge route node for the specified destination. Compare the
2789  *	vlan id or if zero then just return the first match.
2790  */
2791 static struct bridge_rtnode *
2792 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2793 {
2794 	struct bridge_rtnode *brt;
2795 	uint32_t hash;
2796 	int dir;
2797 
2798 	BRIDGE_LOCK_ASSERT(sc);
2799 
2800 	hash = bridge_rthash(sc, addr);
2801 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2802 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2803 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2804 			return (brt);
2805 		if (dir > 0)
2806 			return (NULL);
2807 	}
2808 
2809 	return (NULL);
2810 }
2811 
2812 /*
2813  * bridge_rtnode_insert:
2814  *
2815  *	Insert the specified bridge node into the route table.  We
2816  *	assume the entry is not already in the table.
2817  */
2818 static int
2819 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2820 {
2821 	struct bridge_rtnode *lbrt;
2822 	uint32_t hash;
2823 	int dir;
2824 
2825 	BRIDGE_LOCK_ASSERT(sc);
2826 
2827 	hash = bridge_rthash(sc, brt->brt_addr);
2828 
2829 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2830 	if (lbrt == NULL) {
2831 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2832 		goto out;
2833 	}
2834 
2835 	do {
2836 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2837 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2838 			return (EEXIST);
2839 		if (dir > 0) {
2840 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2841 			goto out;
2842 		}
2843 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2844 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2845 			goto out;
2846 		}
2847 		lbrt = LIST_NEXT(lbrt, brt_hash);
2848 	} while (lbrt != NULL);
2849 
2850 #ifdef DIAGNOSTIC
2851 	panic("bridge_rtnode_insert: impossible");
2852 #endif
2853 
2854 out:
2855 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2856 	sc->sc_brtcnt++;
2857 
2858 	return (0);
2859 }
2860 
2861 /*
2862  * bridge_rtnode_destroy:
2863  *
2864  *	Destroy a bridge rtnode.
2865  */
2866 static void
2867 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2868 {
2869 	BRIDGE_LOCK_ASSERT(sc);
2870 
2871 	LIST_REMOVE(brt, brt_hash);
2872 
2873 	LIST_REMOVE(brt, brt_list);
2874 	sc->sc_brtcnt--;
2875 	brt->brt_dst->bif_addrcnt--;
2876 	uma_zfree(bridge_rtnode_zone, brt);
2877 }
2878 
2879 /*
2880  * bridge_rtable_expire:
2881  *
2882  *	Set the expiry time for all routes on an interface.
2883  */
2884 static void
2885 bridge_rtable_expire(struct ifnet *ifp, int age)
2886 {
2887 	struct bridge_softc *sc = ifp->if_bridge;
2888 	struct bridge_rtnode *brt;
2889 
2890 	BRIDGE_LOCK(sc);
2891 
2892 	/*
2893 	 * If the age is zero then flush, otherwise set all the expiry times to
2894 	 * age for the interface
2895 	 */
2896 	if (age == 0)
2897 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2898 	else {
2899 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2900 			/* Cap the expiry time to 'age' */
2901 			if (brt->brt_ifp == ifp &&
2902 			    brt->brt_expire > time_uptime + age &&
2903 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2904 				brt->brt_expire = time_uptime + age;
2905 		}
2906 	}
2907 	BRIDGE_UNLOCK(sc);
2908 }
2909 
2910 /*
2911  * bridge_state_change:
2912  *
2913  *	Callback from the bridgestp code when a port changes states.
2914  */
2915 static void
2916 bridge_state_change(struct ifnet *ifp, int state)
2917 {
2918 	struct bridge_softc *sc = ifp->if_bridge;
2919 	static const char *stpstates[] = {
2920 		"disabled",
2921 		"listening",
2922 		"learning",
2923 		"forwarding",
2924 		"blocking",
2925 		"discarding"
2926 	};
2927 
2928 	if (log_stp)
2929 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2930 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2931 }
2932 
2933 /*
2934  * Send bridge packets through pfil if they are one of the types pfil can deal
2935  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
2936  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2937  * that interface.
2938  */
2939 static int
2940 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2941 {
2942 	int snap, error, i, hlen;
2943 	struct ether_header *eh1, eh2;
2944 	struct ip_fw_args args;
2945 	struct ip *ip;
2946 	struct llc llc1;
2947 	u_int16_t ether_type;
2948 
2949 	snap = 0;
2950 	error = -1;	/* Default error if not error == 0 */
2951 
2952 #if 0
2953 	/* we may return with the IP fields swapped, ensure its not shared */
2954 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2955 #endif
2956 
2957 	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2958 		return (0); /* filtering is disabled */
2959 
2960 	i = min((*mp)->m_pkthdr.len, max_protohdr);
2961 	if ((*mp)->m_len < i) {
2962 	    *mp = m_pullup(*mp, i);
2963 	    if (*mp == NULL) {
2964 		printf("%s: m_pullup failed\n", __func__);
2965 		return (-1);
2966 	    }
2967 	}
2968 
2969 	eh1 = mtod(*mp, struct ether_header *);
2970 	ether_type = ntohs(eh1->ether_type);
2971 
2972 	/*
2973 	 * Check for SNAP/LLC.
2974 	 */
2975 	if (ether_type < ETHERMTU) {
2976 		struct llc *llc2 = (struct llc *)(eh1 + 1);
2977 
2978 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
2979 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
2980 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
2981 		    llc2->llc_control == LLC_UI) {
2982 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
2983 			snap = 1;
2984 		}
2985 	}
2986 
2987 	/*
2988 	 * If we're trying to filter bridge traffic, don't look at anything
2989 	 * other than IP and ARP traffic.  If the filter doesn't understand
2990 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
2991 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
2992 	 * but of course we don't have an AppleTalk filter to begin with.
2993 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
2994 	 * ARP traffic.)
2995 	 */
2996 	switch (ether_type) {
2997 		case ETHERTYPE_ARP:
2998 		case ETHERTYPE_REVARP:
2999 			if (pfil_ipfw_arp == 0)
3000 				return (0); /* Automatically pass */
3001 			break;
3002 
3003 		case ETHERTYPE_IP:
3004 #ifdef INET6
3005 		case ETHERTYPE_IPV6:
3006 #endif /* INET6 */
3007 			break;
3008 		default:
3009 			/*
3010 			 * Check to see if the user wants to pass non-ip
3011 			 * packets, these will not be checked by pfil(9) and
3012 			 * passed unconditionally so the default is to drop.
3013 			 */
3014 			if (pfil_onlyip)
3015 				goto bad;
3016 	}
3017 
3018 	/* Strip off the Ethernet header and keep a copy. */
3019 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3020 	m_adj(*mp, ETHER_HDR_LEN);
3021 
3022 	/* Strip off snap header, if present */
3023 	if (snap) {
3024 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3025 		m_adj(*mp, sizeof(struct llc));
3026 	}
3027 
3028 	/*
3029 	 * Check the IP header for alignment and errors
3030 	 */
3031 	if (dir == PFIL_IN) {
3032 		switch (ether_type) {
3033 			case ETHERTYPE_IP:
3034 				error = bridge_ip_checkbasic(mp);
3035 				break;
3036 #ifdef INET6
3037 			case ETHERTYPE_IPV6:
3038 				error = bridge_ip6_checkbasic(mp);
3039 				break;
3040 #endif /* INET6 */
3041 			default:
3042 				error = 0;
3043 		}
3044 		if (error)
3045 			goto bad;
3046 	}
3047 
3048 	if (IPFW_LOADED && pfil_ipfw != 0 && dir == PFIL_OUT && ifp != NULL) {
3049 		INIT_VNET_INET(curvnet);
3050 
3051 		error = -1;
3052 		args.rule = ip_dn_claim_rule(*mp);
3053 		if (args.rule != NULL && V_fw_one_pass)
3054 			goto ipfwpass; /* packet already partially processed */
3055 
3056 		args.m = *mp;
3057 		args.oif = ifp;
3058 		args.next_hop = NULL;
3059 		args.eh = &eh2;
3060 		args.inp = NULL;	/* used by ipfw uid/gid/jail rules */
3061 		i = ip_fw_chk_ptr(&args);
3062 		*mp = args.m;
3063 
3064 		if (*mp == NULL)
3065 			return (error);
3066 
3067 		if (DUMMYNET_LOADED && (i == IP_FW_DUMMYNET)) {
3068 
3069 			/* put the Ethernet header back on */
3070 			M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3071 			if (*mp == NULL)
3072 				return (error);
3073 			bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3074 
3075 			/*
3076 			 * Pass the pkt to dummynet, which consumes it. The
3077 			 * packet will return to us via bridge_dummynet().
3078 			 */
3079 			args.oif = ifp;
3080 			ip_dn_io_ptr(mp, DN_TO_IFB_FWD, &args);
3081 			return (error);
3082 		}
3083 
3084 		if (i != IP_FW_PASS) /* drop */
3085 			goto bad;
3086 	}
3087 
3088 ipfwpass:
3089 	error = 0;
3090 
3091 	/*
3092 	 * Run the packet through pfil
3093 	 */
3094 	switch (ether_type) {
3095 	case ETHERTYPE_IP:
3096 		/*
3097 		 * before calling the firewall, swap fields the same as
3098 		 * IP does. here we assume the header is contiguous
3099 		 */
3100 		ip = mtod(*mp, struct ip *);
3101 
3102 		ip->ip_len = ntohs(ip->ip_len);
3103 		ip->ip_off = ntohs(ip->ip_off);
3104 
3105 		/*
3106 		 * Run pfil on the member interface and the bridge, both can
3107 		 * be skipped by clearing pfil_member or pfil_bridge.
3108 		 *
3109 		 * Keep the order:
3110 		 *   in_if -> bridge_if -> out_if
3111 		 */
3112 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3113 			error = pfil_run_hooks(&inet_pfil_hook, mp, bifp,
3114 					dir, NULL);
3115 
3116 		if (*mp == NULL || error != 0) /* filter may consume */
3117 			break;
3118 
3119 		if (pfil_member && ifp != NULL)
3120 			error = pfil_run_hooks(&inet_pfil_hook, mp, ifp,
3121 					dir, NULL);
3122 
3123 		if (*mp == NULL || error != 0) /* filter may consume */
3124 			break;
3125 
3126 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3127 			error = pfil_run_hooks(&inet_pfil_hook, mp, bifp,
3128 					dir, NULL);
3129 
3130 		if (*mp == NULL || error != 0) /* filter may consume */
3131 			break;
3132 
3133 		/* check if we need to fragment the packet */
3134 		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3135 			i = (*mp)->m_pkthdr.len;
3136 			if (i > ifp->if_mtu) {
3137 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3138 					    &llc1);
3139 				return (error);
3140 			}
3141 		}
3142 
3143 		/* Recalculate the ip checksum and restore byte ordering */
3144 		ip = mtod(*mp, struct ip *);
3145 		hlen = ip->ip_hl << 2;
3146 		if (hlen < sizeof(struct ip))
3147 			goto bad;
3148 		if (hlen > (*mp)->m_len) {
3149 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3150 				goto bad;
3151 			ip = mtod(*mp, struct ip *);
3152 			if (ip == NULL)
3153 				goto bad;
3154 		}
3155 		ip->ip_len = htons(ip->ip_len);
3156 		ip->ip_off = htons(ip->ip_off);
3157 		ip->ip_sum = 0;
3158 		if (hlen == sizeof(struct ip))
3159 			ip->ip_sum = in_cksum_hdr(ip);
3160 		else
3161 			ip->ip_sum = in_cksum(*mp, hlen);
3162 
3163 		break;
3164 #ifdef INET6
3165 	case ETHERTYPE_IPV6:
3166 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3167 			error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp,
3168 					dir, NULL);
3169 
3170 		if (*mp == NULL || error != 0) /* filter may consume */
3171 			break;
3172 
3173 		if (pfil_member && ifp != NULL)
3174 			error = pfil_run_hooks(&inet6_pfil_hook, mp, ifp,
3175 					dir, NULL);
3176 
3177 		if (*mp == NULL || error != 0) /* filter may consume */
3178 			break;
3179 
3180 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3181 			error = pfil_run_hooks(&inet6_pfil_hook, mp, bifp,
3182 					dir, NULL);
3183 		break;
3184 #endif
3185 	default:
3186 		error = 0;
3187 		break;
3188 	}
3189 
3190 	if (*mp == NULL)
3191 		return (error);
3192 	if (error != 0)
3193 		goto bad;
3194 
3195 	error = -1;
3196 
3197 	/*
3198 	 * Finally, put everything back the way it was and return
3199 	 */
3200 	if (snap) {
3201 		M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT);
3202 		if (*mp == NULL)
3203 			return (error);
3204 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3205 	}
3206 
3207 	M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3208 	if (*mp == NULL)
3209 		return (error);
3210 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3211 
3212 	return (0);
3213 
3214 bad:
3215 	m_freem(*mp);
3216 	*mp = NULL;
3217 	return (error);
3218 }
3219 
3220 /*
3221  * Perform basic checks on header size since
3222  * pfil assumes ip_input has already processed
3223  * it for it.  Cut-and-pasted from ip_input.c.
3224  * Given how simple the IPv6 version is,
3225  * does the IPv4 version really need to be
3226  * this complicated?
3227  *
3228  * XXX Should we update ipstat here, or not?
3229  * XXX Right now we update ipstat but not
3230  * XXX csum_counter.
3231  */
3232 static int
3233 bridge_ip_checkbasic(struct mbuf **mp)
3234 {
3235 	INIT_VNET_INET(curvnet);
3236 	struct mbuf *m = *mp;
3237 	struct ip *ip;
3238 	int len, hlen;
3239 	u_short sum;
3240 
3241 	if (*mp == NULL)
3242 		return (-1);
3243 
3244 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3245 		if ((m = m_copyup(m, sizeof(struct ip),
3246 			(max_linkhdr + 3) & ~3)) == NULL) {
3247 			/* XXXJRT new stat, please */
3248 			V_ipstat.ips_toosmall++;
3249 			goto bad;
3250 		}
3251 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3252 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3253 			V_ipstat.ips_toosmall++;
3254 			goto bad;
3255 		}
3256 	}
3257 	ip = mtod(m, struct ip *);
3258 	if (ip == NULL) goto bad;
3259 
3260 	if (ip->ip_v != IPVERSION) {
3261 		V_ipstat.ips_badvers++;
3262 		goto bad;
3263 	}
3264 	hlen = ip->ip_hl << 2;
3265 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3266 		V_ipstat.ips_badhlen++;
3267 		goto bad;
3268 	}
3269 	if (hlen > m->m_len) {
3270 		if ((m = m_pullup(m, hlen)) == 0) {
3271 			V_ipstat.ips_badhlen++;
3272 			goto bad;
3273 		}
3274 		ip = mtod(m, struct ip *);
3275 		if (ip == NULL) goto bad;
3276 	}
3277 
3278 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3279 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3280 	} else {
3281 		if (hlen == sizeof(struct ip)) {
3282 			sum = in_cksum_hdr(ip);
3283 		} else {
3284 			sum = in_cksum(m, hlen);
3285 		}
3286 	}
3287 	if (sum) {
3288 		V_ipstat.ips_badsum++;
3289 		goto bad;
3290 	}
3291 
3292 	/* Retrieve the packet length. */
3293 	len = ntohs(ip->ip_len);
3294 
3295 	/*
3296 	 * Check for additional length bogosity
3297 	 */
3298 	if (len < hlen) {
3299 		V_ipstat.ips_badlen++;
3300 		goto bad;
3301 	}
3302 
3303 	/*
3304 	 * Check that the amount of data in the buffers
3305 	 * is as at least much as the IP header would have us expect.
3306 	 * Drop packet if shorter than we expect.
3307 	 */
3308 	if (m->m_pkthdr.len < len) {
3309 		V_ipstat.ips_tooshort++;
3310 		goto bad;
3311 	}
3312 
3313 	/* Checks out, proceed */
3314 	*mp = m;
3315 	return (0);
3316 
3317 bad:
3318 	*mp = m;
3319 	return (-1);
3320 }
3321 
3322 #ifdef INET6
3323 /*
3324  * Same as above, but for IPv6.
3325  * Cut-and-pasted from ip6_input.c.
3326  * XXX Should we update ip6stat, or not?
3327  */
3328 static int
3329 bridge_ip6_checkbasic(struct mbuf **mp)
3330 {
3331 	INIT_VNET_INET6(curvnet);
3332 	struct mbuf *m = *mp;
3333 	struct ip6_hdr *ip6;
3334 
3335 	/*
3336 	 * If the IPv6 header is not aligned, slurp it up into a new
3337 	 * mbuf with space for link headers, in the event we forward
3338 	 * it.  Otherwise, if it is aligned, make sure the entire base
3339 	 * IPv6 header is in the first mbuf of the chain.
3340 	 */
3341 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3342 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3343 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3344 			    (max_linkhdr + 3) & ~3)) == NULL) {
3345 			/* XXXJRT new stat, please */
3346 			V_ip6stat.ip6s_toosmall++;
3347 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3348 			goto bad;
3349 		}
3350 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3351 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3352 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3353 			V_ip6stat.ip6s_toosmall++;
3354 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3355 			goto bad;
3356 		}
3357 	}
3358 
3359 	ip6 = mtod(m, struct ip6_hdr *);
3360 
3361 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3362 		V_ip6stat.ip6s_badvers++;
3363 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3364 		goto bad;
3365 	}
3366 
3367 	/* Checks out, proceed */
3368 	*mp = m;
3369 	return (0);
3370 
3371 bad:
3372 	*mp = m;
3373 	return (-1);
3374 }
3375 #endif /* INET6 */
3376 
3377 /*
3378  * bridge_fragment:
3379  *
3380  *	Return a fragmented mbuf chain.
3381  */
3382 static int
3383 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3384     int snap, struct llc *llc)
3385 {
3386 	INIT_VNET_INET(curvnet);
3387 	struct mbuf *m0;
3388 	struct ip *ip;
3389 	int error = -1;
3390 
3391 	if (m->m_len < sizeof(struct ip) &&
3392 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3393 		goto out;
3394 	ip = mtod(m, struct ip *);
3395 
3396 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist,
3397 		    CSUM_DELAY_IP);
3398 	if (error)
3399 		goto out;
3400 
3401 	/* walk the chain and re-add the Ethernet header */
3402 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3403 		if (error == 0) {
3404 			if (snap) {
3405 				M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT);
3406 				if (m0 == NULL) {
3407 					error = ENOBUFS;
3408 					continue;
3409 				}
3410 				bcopy(llc, mtod(m0, caddr_t),
3411 				    sizeof(struct llc));
3412 			}
3413 			M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT);
3414 			if (m0 == NULL) {
3415 				error = ENOBUFS;
3416 				continue;
3417 			}
3418 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3419 		} else
3420 			m_freem(m);
3421 	}
3422 
3423 	if (error == 0)
3424 		V_ipstat.ips_fragmented++;
3425 
3426 	return (error);
3427 
3428 out:
3429 	if (m != NULL)
3430 		m_freem(m);
3431 	return (error);
3432 }
3433