xref: /freebsd/sys/net/if_bridge.c (revision 5686c6c38a3e1cc78804eaf5f880bda23dcf592f)
1 /*	$NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $	*/
2 
3 /*
4  * Copyright 2001 Wasabi Systems, Inc.
5  * All rights reserved.
6  *
7  * Written by Jason R. Thorpe for Wasabi Systems, Inc.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed for the NetBSD Project by
20  *	Wasabi Systems, Inc.
21  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
22  *    or promote products derived from this software without specific prior
23  *    written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
27  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
28  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
29  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35  * POSSIBILITY OF SUCH DAMAGE.
36  */
37 
38 /*
39  * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net)
40  * All rights reserved.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  *
51  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
52  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
53  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
54  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
55  * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
56  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
57  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
58  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
59  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
60  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
61  * POSSIBILITY OF SUCH DAMAGE.
62  *
63  * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp
64  */
65 
66 /*
67  * Network interface bridge support.
68  *
69  * TODO:
70  *
71  *	- Currently only supports Ethernet-like interfaces (Ethernet,
72  *	  802.11, VLANs on Ethernet, etc.)  Figure out a nice way
73  *	  to bridge other types of interfaces (FDDI-FDDI, and maybe
74  *	  consider heterogenous bridges).
75  */
76 
77 #include <sys/cdefs.h>
78 __FBSDID("$FreeBSD$");
79 
80 #include "opt_inet.h"
81 #include "opt_inet6.h"
82 
83 #include <sys/param.h>
84 #include <sys/mbuf.h>
85 #include <sys/malloc.h>
86 #include <sys/protosw.h>
87 #include <sys/systm.h>
88 #include <sys/jail.h>
89 #include <sys/time.h>
90 #include <sys/socket.h> /* for net/if.h */
91 #include <sys/sockio.h>
92 #include <sys/ctype.h>  /* string functions */
93 #include <sys/kernel.h>
94 #include <sys/random.h>
95 #include <sys/syslog.h>
96 #include <sys/sysctl.h>
97 #include <vm/uma.h>
98 #include <sys/module.h>
99 #include <sys/priv.h>
100 #include <sys/proc.h>
101 #include <sys/lock.h>
102 #include <sys/mutex.h>
103 
104 #include <net/bpf.h>
105 #include <net/if.h>
106 #include <net/if_clone.h>
107 #include <net/if_dl.h>
108 #include <net/if_types.h>
109 #include <net/if_var.h>
110 #include <net/pfil.h>
111 #include <net/vnet.h>
112 
113 #include <netinet/in.h> /* for struct arpcom */
114 #include <netinet/in_systm.h>
115 #include <netinet/in_var.h>
116 #include <netinet/ip.h>
117 #include <netinet/ip_var.h>
118 #ifdef INET6
119 #include <netinet/ip6.h>
120 #include <netinet6/ip6_var.h>
121 #endif
122 #if defined(INET) || defined(INET6)
123 #include <netinet/ip_carp.h>
124 #endif
125 #include <machine/in_cksum.h>
126 #include <netinet/if_ether.h> /* for struct arpcom */
127 #include <net/bridgestp.h>
128 #include <net/if_bridgevar.h>
129 #include <net/if_llc.h>
130 #include <net/if_vlan_var.h>
131 
132 #include <net/route.h>
133 
134 /*
135  * Size of the route hash table.  Must be a power of two.
136  */
137 #ifndef BRIDGE_RTHASH_SIZE
138 #define	BRIDGE_RTHASH_SIZE		1024
139 #endif
140 
141 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
142 
143 /*
144  * Default maximum number of addresses to cache.
145  */
146 #ifndef BRIDGE_RTABLE_MAX
147 #define	BRIDGE_RTABLE_MAX		2000
148 #endif
149 
150 /*
151  * Timeout (in seconds) for entries learned dynamically.
152  */
153 #ifndef BRIDGE_RTABLE_TIMEOUT
154 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
155 #endif
156 
157 /*
158  * Number of seconds between walks of the route list.
159  */
160 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
161 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
162 #endif
163 
164 /*
165  * List of capabilities to possibly mask on the member interface.
166  */
167 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
168 
169 /*
170  * List of capabilities to strip
171  */
172 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
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 int	bridge_transmit(struct ifnet *, struct mbuf *);
246 static void	bridge_qflush(struct ifnet *);
247 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
248 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
249 		    struct rtentry *);
250 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
251 		    struct mbuf *);
252 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
253 
254 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
255 		    struct mbuf *m);
256 
257 static void	bridge_timer(void *);
258 
259 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
260 		    struct mbuf *, int);
261 static void	bridge_span(struct bridge_softc *, struct mbuf *);
262 
263 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
264 		    uint16_t, struct bridge_iflist *, int, uint8_t);
265 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
266 		    uint16_t);
267 static void	bridge_rttrim(struct bridge_softc *);
268 static void	bridge_rtage(struct bridge_softc *);
269 static void	bridge_rtflush(struct bridge_softc *, int);
270 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
271 		    uint16_t);
272 
273 static void	bridge_rtable_init(struct bridge_softc *);
274 static void	bridge_rtable_fini(struct bridge_softc *);
275 
276 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
277 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
278 		    const uint8_t *, uint16_t);
279 static int	bridge_rtnode_insert(struct bridge_softc *,
280 		    struct bridge_rtnode *);
281 static void	bridge_rtnode_destroy(struct bridge_softc *,
282 		    struct bridge_rtnode *);
283 static void	bridge_rtable_expire(struct ifnet *, int);
284 static void	bridge_state_change(struct ifnet *, int);
285 
286 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
287 		    const char *name);
288 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
289 		    struct ifnet *ifp);
290 static void	bridge_delete_member(struct bridge_softc *,
291 		    struct bridge_iflist *, int);
292 static void	bridge_delete_span(struct bridge_softc *,
293 		    struct bridge_iflist *);
294 
295 static int	bridge_ioctl_add(struct bridge_softc *, void *);
296 static int	bridge_ioctl_del(struct bridge_softc *, void *);
297 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
298 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
299 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
300 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
301 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
302 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
303 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
304 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
305 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
306 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
307 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
308 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
309 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
310 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
311 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
312 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
313 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
314 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
315 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
316 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
317 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
318 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
319 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
320 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
321 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
322 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
323 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
324 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
325 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
326 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
327 		    int);
328 static int	bridge_ip_checkbasic(struct mbuf **mp);
329 #ifdef INET6
330 static int	bridge_ip6_checkbasic(struct mbuf **mp);
331 #endif /* INET6 */
332 static int	bridge_fragment(struct ifnet *, struct mbuf *,
333 		    struct ether_header *, int, struct llc *);
334 static void	bridge_linkstate(struct ifnet *ifp);
335 static void	bridge_linkcheck(struct bridge_softc *sc);
336 
337 extern void (*bridge_linkstate_p)(struct ifnet *ifp);
338 
339 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
340 #define	VLANTAGOF(_m)	\
341     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
342 
343 static struct bstp_cb_ops bridge_ops = {
344 	.bcb_state = bridge_state_change,
345 	.bcb_rtage = bridge_rtable_expire
346 };
347 
348 SYSCTL_DECL(_net_link);
349 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
350 
351 static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */
352 static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */
353 static int pfil_member = 1; /* run pfil hooks on the member interface */
354 static int pfil_ipfw = 0;   /* layer2 filter with ipfw */
355 static int pfil_ipfw_arp = 0;   /* layer2 filter with ipfw */
356 static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for
357                                    locally destined packets */
358 static int log_stp   = 0;   /* log STP state changes */
359 static int bridge_inherit_mac = 0;   /* share MAC with first bridge member */
360 TUNABLE_INT("net.link.bridge.pfil_onlyip", &pfil_onlyip);
361 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW,
362     &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled");
363 TUNABLE_INT("net.link.bridge.ipfw_arp", &pfil_ipfw_arp);
364 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW,
365     &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2");
366 TUNABLE_INT("net.link.bridge.pfil_bridge", &pfil_bridge);
367 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW,
368     &pfil_bridge, 0, "Packet filter on the bridge interface");
369 TUNABLE_INT("net.link.bridge.pfil_member", &pfil_member);
370 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW,
371     &pfil_member, 0, "Packet filter on the member interface");
372 TUNABLE_INT("net.link.bridge.pfil_local_phys", &pfil_local_phys);
373 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW,
374     &pfil_local_phys, 0,
375     "Packet filter on the physical interface for locally destined packets");
376 TUNABLE_INT("net.link.bridge.log_stp", &log_stp);
377 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW,
378     &log_stp, 0, "Log STP state changes");
379 TUNABLE_INT("net.link.bridge.inherit_mac", &bridge_inherit_mac);
380 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW,
381     &bridge_inherit_mac, 0,
382     "Inherit MAC address from the first bridge member");
383 
384 struct bridge_control {
385 	int	(*bc_func)(struct bridge_softc *, void *);
386 	int	bc_argsize;
387 	int	bc_flags;
388 };
389 
390 #define	BC_F_COPYIN		0x01	/* copy arguments in */
391 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
392 #define	BC_F_SUSER		0x04	/* do super-user check */
393 
394 const struct bridge_control bridge_control_table[] = {
395 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
396 	  BC_F_COPYIN|BC_F_SUSER },
397 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
398 	  BC_F_COPYIN|BC_F_SUSER },
399 
400 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
401 	  BC_F_COPYIN|BC_F_COPYOUT },
402 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
403 	  BC_F_COPYIN|BC_F_SUSER },
404 
405 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
406 	  BC_F_COPYIN|BC_F_SUSER },
407 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
408 	  BC_F_COPYOUT },
409 
410 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
411 	  BC_F_COPYIN|BC_F_COPYOUT },
412 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
413 	  BC_F_COPYIN|BC_F_COPYOUT },
414 
415 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
416 	  BC_F_COPYIN|BC_F_SUSER },
417 
418 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
419 	  BC_F_COPYIN|BC_F_SUSER },
420 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
421 	  BC_F_COPYOUT },
422 
423 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
424 	  BC_F_COPYIN|BC_F_SUSER },
425 
426 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
427 	  BC_F_COPYIN|BC_F_SUSER },
428 
429 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
430 	  BC_F_COPYOUT },
431 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
432 	  BC_F_COPYIN|BC_F_SUSER },
433 
434 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
435 	  BC_F_COPYOUT },
436 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
437 	  BC_F_COPYIN|BC_F_SUSER },
438 
439 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
440 	  BC_F_COPYOUT },
441 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
442 	  BC_F_COPYIN|BC_F_SUSER },
443 
444 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
445 	  BC_F_COPYOUT },
446 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
447 	  BC_F_COPYIN|BC_F_SUSER },
448 
449 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
450 	  BC_F_COPYIN|BC_F_SUSER },
451 
452 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
453 	  BC_F_COPYIN|BC_F_SUSER },
454 
455 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
456 	  BC_F_COPYIN|BC_F_SUSER },
457 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
458 	  BC_F_COPYIN|BC_F_SUSER },
459 
460 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
461 	  BC_F_COPYOUT },
462 
463 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
464 	  BC_F_COPYOUT },
465 
466 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
467 	  BC_F_COPYIN|BC_F_COPYOUT },
468 
469 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
470 	  BC_F_COPYIN|BC_F_SUSER },
471 
472 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
473 	  BC_F_COPYIN|BC_F_SUSER },
474 
475 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
476 	  BC_F_COPYIN|BC_F_SUSER },
477 
478 };
479 const int bridge_control_table_size =
480     sizeof(bridge_control_table) / sizeof(bridge_control_table[0]);
481 
482 LIST_HEAD(, bridge_softc) bridge_list;
483 
484 static struct if_clone *bridge_cloner;
485 static const char bridge_name[] = "bridge";
486 
487 static int
488 bridge_modevent(module_t mod, int type, void *data)
489 {
490 
491 	switch (type) {
492 	case MOD_LOAD:
493 		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
494 		bridge_cloner = if_clone_simple(bridge_name,
495 		    bridge_clone_create, bridge_clone_destroy, 0);
496 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
497 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
498 		    UMA_ALIGN_PTR, 0);
499 		LIST_INIT(&bridge_list);
500 		bridge_input_p = bridge_input;
501 		bridge_output_p = bridge_output;
502 		bridge_dn_p = bridge_dummynet;
503 		bridge_linkstate_p = bridge_linkstate;
504 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
505 		    ifnet_departure_event, bridge_ifdetach, NULL,
506 		    EVENTHANDLER_PRI_ANY);
507 		break;
508 	case MOD_UNLOAD:
509 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
510 		    bridge_detach_cookie);
511 		if_clone_detach(bridge_cloner);
512 		uma_zdestroy(bridge_rtnode_zone);
513 		bridge_input_p = NULL;
514 		bridge_output_p = NULL;
515 		bridge_dn_p = NULL;
516 		bridge_linkstate_p = NULL;
517 		mtx_destroy(&bridge_list_mtx);
518 		break;
519 	default:
520 		return (EOPNOTSUPP);
521 	}
522 	return (0);
523 }
524 
525 static moduledata_t bridge_mod = {
526 	"if_bridge",
527 	bridge_modevent,
528 	0
529 };
530 
531 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
532 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
533 
534 /*
535  * handler for net.link.bridge.ipfw
536  */
537 static int
538 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
539 {
540 	int enable = pfil_ipfw;
541 	int error;
542 
543 	error = sysctl_handle_int(oidp, &enable, 0, req);
544 	enable = (enable) ? 1 : 0;
545 
546 	if (enable != pfil_ipfw) {
547 		pfil_ipfw = enable;
548 
549 		/*
550 		 * Disable pfil so that ipfw doesnt run twice, if the user
551 		 * really wants both then they can re-enable pfil_bridge and/or
552 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
553 		 * layer2 type.
554 		 */
555 		if (pfil_ipfw) {
556 			pfil_onlyip = 0;
557 			pfil_bridge = 0;
558 			pfil_member = 0;
559 		}
560 	}
561 
562 	return (error);
563 }
564 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
565 	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
566 
567 /*
568  * bridge_clone_create:
569  *
570  *	Create a new bridge instance.
571  */
572 static int
573 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
574 {
575 	struct bridge_softc *sc, *sc2;
576 	struct ifnet *bifp, *ifp;
577 	int fb, retry;
578 	unsigned long hostid;
579 
580 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
581 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
582 	if (ifp == NULL) {
583 		free(sc, M_DEVBUF);
584 		return (ENOSPC);
585 	}
586 
587 	BRIDGE_LOCK_INIT(sc);
588 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
589 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
590 
591 	/* Initialize our routing table. */
592 	bridge_rtable_init(sc);
593 
594 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
595 
596 	LIST_INIT(&sc->sc_iflist);
597 	LIST_INIT(&sc->sc_spanlist);
598 
599 	ifp->if_softc = sc;
600 	if_initname(ifp, bridge_name, unit);
601 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
602 	ifp->if_ioctl = bridge_ioctl;
603 	ifp->if_transmit = bridge_transmit;
604 	ifp->if_qflush = bridge_qflush;
605 	ifp->if_init = bridge_init;
606 	ifp->if_type = IFT_BRIDGE;
607 
608 	/*
609 	 * Generate an ethernet address with a locally administered address.
610 	 *
611 	 * Since we are using random ethernet addresses for the bridge, it is
612 	 * possible that we might have address collisions, so make sure that
613 	 * this hardware address isn't already in use on another bridge.
614 	 * The first try uses the hostid and falls back to arc4rand().
615 	 */
616 	fb = 0;
617 	getcredhostid(curthread->td_ucred, &hostid);
618 	do {
619 		if (fb || hostid == 0) {
620 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
621 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
622 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
623 		} else {
624 			sc->sc_defaddr[0] = 0x2;
625 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
626 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
627 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
628 			sc->sc_defaddr[4] =  hostid        & 0xff;
629 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
630 		}
631 
632 		fb = 1;
633 		retry = 0;
634 		mtx_lock(&bridge_list_mtx);
635 		LIST_FOREACH(sc2, &bridge_list, sc_list) {
636 			bifp = sc2->sc_ifp;
637 			if (memcmp(sc->sc_defaddr,
638 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
639 				retry = 1;
640 				break;
641 			}
642 		}
643 		mtx_unlock(&bridge_list_mtx);
644 	} while (retry == 1);
645 
646 	bstp_attach(&sc->sc_stp, &bridge_ops);
647 	ether_ifattach(ifp, sc->sc_defaddr);
648 	/* Now undo some of the damage... */
649 	ifp->if_baudrate = 0;
650 	ifp->if_type = IFT_BRIDGE;
651 
652 	mtx_lock(&bridge_list_mtx);
653 	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
654 	mtx_unlock(&bridge_list_mtx);
655 
656 	return (0);
657 }
658 
659 /*
660  * bridge_clone_destroy:
661  *
662  *	Destroy a bridge instance.
663  */
664 static void
665 bridge_clone_destroy(struct ifnet *ifp)
666 {
667 	struct bridge_softc *sc = ifp->if_softc;
668 	struct bridge_iflist *bif;
669 
670 	BRIDGE_LOCK(sc);
671 
672 	bridge_stop(ifp, 1);
673 	ifp->if_flags &= ~IFF_UP;
674 
675 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
676 		bridge_delete_member(sc, bif, 0);
677 
678 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
679 		bridge_delete_span(sc, bif);
680 	}
681 
682 	BRIDGE_UNLOCK(sc);
683 
684 	callout_drain(&sc->sc_brcallout);
685 
686 	mtx_lock(&bridge_list_mtx);
687 	LIST_REMOVE(sc, sc_list);
688 	mtx_unlock(&bridge_list_mtx);
689 
690 	bstp_detach(&sc->sc_stp);
691 	ether_ifdetach(ifp);
692 	if_free(ifp);
693 
694 	/* Tear down the routing table. */
695 	bridge_rtable_fini(sc);
696 
697 	BRIDGE_LOCK_DESTROY(sc);
698 	free(sc, M_DEVBUF);
699 }
700 
701 /*
702  * bridge_ioctl:
703  *
704  *	Handle a control request from the operator.
705  */
706 static int
707 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
708 {
709 	struct bridge_softc *sc = ifp->if_softc;
710 	struct ifreq *ifr = (struct ifreq *)data;
711 	struct bridge_iflist *bif;
712 	struct thread *td = curthread;
713 	union {
714 		struct ifbreq ifbreq;
715 		struct ifbifconf ifbifconf;
716 		struct ifbareq ifbareq;
717 		struct ifbaconf ifbaconf;
718 		struct ifbrparam ifbrparam;
719 		struct ifbropreq ifbropreq;
720 	} args;
721 	struct ifdrv *ifd = (struct ifdrv *) data;
722 	const struct bridge_control *bc;
723 	int error = 0;
724 
725 	switch (cmd) {
726 
727 	case SIOCADDMULTI:
728 	case SIOCDELMULTI:
729 		break;
730 
731 	case SIOCGDRVSPEC:
732 	case SIOCSDRVSPEC:
733 		if (ifd->ifd_cmd >= bridge_control_table_size) {
734 			error = EINVAL;
735 			break;
736 		}
737 		bc = &bridge_control_table[ifd->ifd_cmd];
738 
739 		if (cmd == SIOCGDRVSPEC &&
740 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
741 			error = EINVAL;
742 			break;
743 		}
744 		else if (cmd == SIOCSDRVSPEC &&
745 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
746 			error = EINVAL;
747 			break;
748 		}
749 
750 		if (bc->bc_flags & BC_F_SUSER) {
751 			error = priv_check(td, PRIV_NET_BRIDGE);
752 			if (error)
753 				break;
754 		}
755 
756 		if (ifd->ifd_len != bc->bc_argsize ||
757 		    ifd->ifd_len > sizeof(args)) {
758 			error = EINVAL;
759 			break;
760 		}
761 
762 		bzero(&args, sizeof(args));
763 		if (bc->bc_flags & BC_F_COPYIN) {
764 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
765 			if (error)
766 				break;
767 		}
768 
769 		BRIDGE_LOCK(sc);
770 		error = (*bc->bc_func)(sc, &args);
771 		BRIDGE_UNLOCK(sc);
772 		if (error)
773 			break;
774 
775 		if (bc->bc_flags & BC_F_COPYOUT)
776 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
777 
778 		break;
779 
780 	case SIOCSIFFLAGS:
781 		if (!(ifp->if_flags & IFF_UP) &&
782 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
783 			/*
784 			 * If interface is marked down and it is running,
785 			 * then stop and disable it.
786 			 */
787 			BRIDGE_LOCK(sc);
788 			bridge_stop(ifp, 1);
789 			BRIDGE_UNLOCK(sc);
790 		} else if ((ifp->if_flags & IFF_UP) &&
791 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
792 			/*
793 			 * If interface is marked up and it is stopped, then
794 			 * start it.
795 			 */
796 			(*ifp->if_init)(sc);
797 		}
798 		break;
799 
800 	case SIOCSIFMTU:
801 		if (ifr->ifr_mtu < 576) {
802 			error = EINVAL;
803 			break;
804 		}
805 		if (LIST_EMPTY(&sc->sc_iflist)) {
806 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
807 			break;
808 		}
809 		BRIDGE_LOCK(sc);
810 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
811 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
812 				log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
813 				    " != %d\n", sc->sc_ifp->if_xname,
814 				    bif->bif_ifp->if_mtu,
815 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
816 				error = EINVAL;
817 				break;
818 			}
819 		}
820 		if (!error)
821 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
822 		BRIDGE_UNLOCK(sc);
823 		break;
824 	default:
825 		/*
826 		 * drop the lock as ether_ioctl() will call bridge_start() and
827 		 * cause the lock to be recursed.
828 		 */
829 		error = ether_ioctl(ifp, cmd, data);
830 		break;
831 	}
832 
833 	return (error);
834 }
835 
836 /*
837  * bridge_mutecaps:
838  *
839  *	Clear or restore unwanted capabilities on the member interface
840  */
841 static void
842 bridge_mutecaps(struct bridge_softc *sc)
843 {
844 	struct bridge_iflist *bif;
845 	int enabled, mask;
846 
847 	/* Initial bitmask of capabilities to test */
848 	mask = BRIDGE_IFCAPS_MASK;
849 
850 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
851 		/* Every member must support it or its disabled */
852 		mask &= bif->bif_savedcaps;
853 	}
854 
855 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
856 		enabled = bif->bif_ifp->if_capenable;
857 		enabled &= ~BRIDGE_IFCAPS_STRIP;
858 		/* strip off mask bits and enable them again if allowed */
859 		enabled &= ~BRIDGE_IFCAPS_MASK;
860 		enabled |= mask;
861 		bridge_set_ifcap(sc, bif, enabled);
862 	}
863 
864 }
865 
866 static void
867 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
868 {
869 	struct ifnet *ifp = bif->bif_ifp;
870 	struct ifreq ifr;
871 	int error;
872 
873 	bzero(&ifr, sizeof(ifr));
874 	ifr.ifr_reqcap = set;
875 
876 	if (ifp->if_capenable != set) {
877 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
878 		if (error)
879 			if_printf(sc->sc_ifp,
880 			    "error setting interface capabilities on %s\n",
881 			    ifp->if_xname);
882 	}
883 }
884 
885 /*
886  * bridge_lookup_member:
887  *
888  *	Lookup a bridge member interface.
889  */
890 static struct bridge_iflist *
891 bridge_lookup_member(struct bridge_softc *sc, const char *name)
892 {
893 	struct bridge_iflist *bif;
894 	struct ifnet *ifp;
895 
896 	BRIDGE_LOCK_ASSERT(sc);
897 
898 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
899 		ifp = bif->bif_ifp;
900 		if (strcmp(ifp->if_xname, name) == 0)
901 			return (bif);
902 	}
903 
904 	return (NULL);
905 }
906 
907 /*
908  * bridge_lookup_member_if:
909  *
910  *	Lookup a bridge member interface by ifnet*.
911  */
912 static struct bridge_iflist *
913 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
914 {
915 	struct bridge_iflist *bif;
916 
917 	BRIDGE_LOCK_ASSERT(sc);
918 
919 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
920 		if (bif->bif_ifp == member_ifp)
921 			return (bif);
922 	}
923 
924 	return (NULL);
925 }
926 
927 /*
928  * bridge_delete_member:
929  *
930  *	Delete the specified member interface.
931  */
932 static void
933 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
934     int gone)
935 {
936 	struct ifnet *ifs = bif->bif_ifp;
937 	struct ifnet *fif = NULL;
938 
939 	BRIDGE_LOCK_ASSERT(sc);
940 
941 	if (bif->bif_flags & IFBIF_STP)
942 		bstp_disable(&bif->bif_stp);
943 
944 	ifs->if_bridge = NULL;
945 	BRIDGE_XLOCK(sc);
946 	LIST_REMOVE(bif, bif_next);
947 	BRIDGE_XDROP(sc);
948 
949 	/*
950 	 * If removing the interface that gave the bridge its mac address, set
951 	 * the mac address of the bridge to the address of the next member, or
952 	 * to its default address if no members are left.
953 	 */
954 	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
955 		if (LIST_EMPTY(&sc->sc_iflist)) {
956 			bcopy(sc->sc_defaddr,
957 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
958 			sc->sc_ifaddr = NULL;
959 		} else {
960 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
961 			bcopy(IF_LLADDR(fif),
962 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
963 			sc->sc_ifaddr = fif;
964 		}
965 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
966 	}
967 
968 	bridge_linkcheck(sc);
969 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
970 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
971 	KASSERT(bif->bif_addrcnt == 0,
972 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
973 
974 	BRIDGE_UNLOCK(sc);
975 	if (!gone) {
976 		switch (ifs->if_type) {
977 		case IFT_ETHER:
978 		case IFT_L2VLAN:
979 			/*
980 			 * Take the interface out of promiscuous mode.
981 			 */
982 			(void) ifpromisc(ifs, 0);
983 			break;
984 
985 		case IFT_GIF:
986 			break;
987 
988 		default:
989 #ifdef DIAGNOSTIC
990 			panic("bridge_delete_member: impossible");
991 #endif
992 			break;
993 		}
994 		/* reneable any interface capabilities */
995 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
996 	}
997 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
998 	BRIDGE_LOCK(sc);
999 	free(bif, M_DEVBUF);
1000 }
1001 
1002 /*
1003  * bridge_delete_span:
1004  *
1005  *	Delete the specified span interface.
1006  */
1007 static void
1008 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1009 {
1010 	BRIDGE_LOCK_ASSERT(sc);
1011 
1012 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1013 	    ("%s: not a span interface", __func__));
1014 
1015 	LIST_REMOVE(bif, bif_next);
1016 	free(bif, M_DEVBUF);
1017 }
1018 
1019 static int
1020 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1021 {
1022 	struct ifbreq *req = arg;
1023 	struct bridge_iflist *bif = NULL;
1024 	struct ifnet *ifs;
1025 	int error = 0;
1026 
1027 	ifs = ifunit(req->ifbr_ifsname);
1028 	if (ifs == NULL)
1029 		return (ENOENT);
1030 	if (ifs->if_ioctl == NULL)	/* must be supported */
1031 		return (EINVAL);
1032 
1033 	/* If it's in the span list, it can't be a member. */
1034 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1035 		if (ifs == bif->bif_ifp)
1036 			return (EBUSY);
1037 
1038 	if (ifs->if_bridge == sc)
1039 		return (EEXIST);
1040 
1041 	if (ifs->if_bridge != NULL)
1042 		return (EBUSY);
1043 
1044 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1045 	if (bif == NULL)
1046 		return (ENOMEM);
1047 
1048 	bif->bif_ifp = ifs;
1049 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1050 	bif->bif_savedcaps = ifs->if_capenable;
1051 
1052 	switch (ifs->if_type) {
1053 	case IFT_ETHER:
1054 	case IFT_L2VLAN:
1055 	case IFT_GIF:
1056 		/* permitted interface types */
1057 		break;
1058 	default:
1059 		error = EINVAL;
1060 		goto out;
1061 	}
1062 
1063 	/* Allow the first Ethernet member to define the MTU */
1064 	if (LIST_EMPTY(&sc->sc_iflist))
1065 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1066 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1067 		if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1068 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1069 		error = EINVAL;
1070 		goto out;
1071 	}
1072 
1073 	/*
1074 	 * Assign the interface's MAC address to the bridge if it's the first
1075 	 * member and the MAC address of the bridge has not been changed from
1076 	 * the default randomly generated one.
1077 	 */
1078 	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1079 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1080 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1081 		sc->sc_ifaddr = ifs;
1082 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1083 	}
1084 
1085 	ifs->if_bridge = sc;
1086 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1087 	/*
1088 	 * XXX: XLOCK HERE!?!
1089 	 *
1090 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1091 	 */
1092 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1093 
1094 	/* Set interface capabilities to the intersection set of all members */
1095 	bridge_mutecaps(sc);
1096 	bridge_linkcheck(sc);
1097 
1098 	/* Place the interface into promiscuous mode */
1099 	switch (ifs->if_type) {
1100 		case IFT_ETHER:
1101 		case IFT_L2VLAN:
1102 			BRIDGE_UNLOCK(sc);
1103 			error = ifpromisc(ifs, 1);
1104 			BRIDGE_LOCK(sc);
1105 			break;
1106 	}
1107 	if (error)
1108 		bridge_delete_member(sc, bif, 0);
1109 out:
1110 	if (error) {
1111 		if (bif != NULL)
1112 			free(bif, M_DEVBUF);
1113 	}
1114 	return (error);
1115 }
1116 
1117 static int
1118 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1119 {
1120 	struct ifbreq *req = arg;
1121 	struct bridge_iflist *bif;
1122 
1123 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1124 	if (bif == NULL)
1125 		return (ENOENT);
1126 
1127 	bridge_delete_member(sc, bif, 0);
1128 
1129 	return (0);
1130 }
1131 
1132 static int
1133 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1134 {
1135 	struct ifbreq *req = arg;
1136 	struct bridge_iflist *bif;
1137 	struct bstp_port *bp;
1138 
1139 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1140 	if (bif == NULL)
1141 		return (ENOENT);
1142 
1143 	bp = &bif->bif_stp;
1144 	req->ifbr_ifsflags = bif->bif_flags;
1145 	req->ifbr_state = bp->bp_state;
1146 	req->ifbr_priority = bp->bp_priority;
1147 	req->ifbr_path_cost = bp->bp_path_cost;
1148 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1149 	req->ifbr_proto = bp->bp_protover;
1150 	req->ifbr_role = bp->bp_role;
1151 	req->ifbr_stpflags = bp->bp_flags;
1152 	req->ifbr_addrcnt = bif->bif_addrcnt;
1153 	req->ifbr_addrmax = bif->bif_addrmax;
1154 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1155 
1156 	/* Copy STP state options as flags */
1157 	if (bp->bp_operedge)
1158 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1159 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1160 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1161 	if (bp->bp_ptp_link)
1162 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1163 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1164 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1165 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1166 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1167 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1168 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1169 	return (0);
1170 }
1171 
1172 static int
1173 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1174 {
1175 	struct ifbreq *req = arg;
1176 	struct bridge_iflist *bif;
1177 	struct bstp_port *bp;
1178 	int error;
1179 
1180 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1181 	if (bif == NULL)
1182 		return (ENOENT);
1183 	bp = &bif->bif_stp;
1184 
1185 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1186 		/* SPAN is readonly */
1187 		return (EINVAL);
1188 
1189 	if (req->ifbr_ifsflags & IFBIF_STP) {
1190 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1191 			error = bstp_enable(&bif->bif_stp);
1192 			if (error)
1193 				return (error);
1194 		}
1195 	} else {
1196 		if ((bif->bif_flags & IFBIF_STP) != 0)
1197 			bstp_disable(&bif->bif_stp);
1198 	}
1199 
1200 	/* Pass on STP flags */
1201 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1202 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1203 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1204 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1205 
1206 	/* Save the bits relating to the bridge */
1207 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1208 
1209 	return (0);
1210 }
1211 
1212 static int
1213 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1214 {
1215 	struct ifbrparam *param = arg;
1216 
1217 	sc->sc_brtmax = param->ifbrp_csize;
1218 	bridge_rttrim(sc);
1219 
1220 	return (0);
1221 }
1222 
1223 static int
1224 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1225 {
1226 	struct ifbrparam *param = arg;
1227 
1228 	param->ifbrp_csize = sc->sc_brtmax;
1229 
1230 	return (0);
1231 }
1232 
1233 static int
1234 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1235 {
1236 	struct ifbifconf *bifc = arg;
1237 	struct bridge_iflist *bif;
1238 	struct ifbreq breq;
1239 	char *buf, *outbuf;
1240 	int count, buflen, len, error = 0;
1241 
1242 	count = 0;
1243 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1244 		count++;
1245 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1246 		count++;
1247 
1248 	buflen = sizeof(breq) * count;
1249 	if (bifc->ifbic_len == 0) {
1250 		bifc->ifbic_len = buflen;
1251 		return (0);
1252 	}
1253 	BRIDGE_UNLOCK(sc);
1254 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1255 	BRIDGE_LOCK(sc);
1256 
1257 	count = 0;
1258 	buf = outbuf;
1259 	len = min(bifc->ifbic_len, buflen);
1260 	bzero(&breq, sizeof(breq));
1261 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1262 		if (len < sizeof(breq))
1263 			break;
1264 
1265 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1266 		    sizeof(breq.ifbr_ifsname));
1267 		/* Fill in the ifbreq structure */
1268 		error = bridge_ioctl_gifflags(sc, &breq);
1269 		if (error)
1270 			break;
1271 		memcpy(buf, &breq, sizeof(breq));
1272 		count++;
1273 		buf += sizeof(breq);
1274 		len -= sizeof(breq);
1275 	}
1276 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1277 		if (len < sizeof(breq))
1278 			break;
1279 
1280 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1281 		    sizeof(breq.ifbr_ifsname));
1282 		breq.ifbr_ifsflags = bif->bif_flags;
1283 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1284 		memcpy(buf, &breq, sizeof(breq));
1285 		count++;
1286 		buf += sizeof(breq);
1287 		len -= sizeof(breq);
1288 	}
1289 
1290 	BRIDGE_UNLOCK(sc);
1291 	bifc->ifbic_len = sizeof(breq) * count;
1292 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1293 	BRIDGE_LOCK(sc);
1294 	free(outbuf, M_TEMP);
1295 	return (error);
1296 }
1297 
1298 static int
1299 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1300 {
1301 	struct ifbaconf *bac = arg;
1302 	struct bridge_rtnode *brt;
1303 	struct ifbareq bareq;
1304 	char *buf, *outbuf;
1305 	int count, buflen, len, error = 0;
1306 
1307 	if (bac->ifbac_len == 0)
1308 		return (0);
1309 
1310 	count = 0;
1311 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1312 		count++;
1313 	buflen = sizeof(bareq) * count;
1314 
1315 	BRIDGE_UNLOCK(sc);
1316 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1317 	BRIDGE_LOCK(sc);
1318 
1319 	count = 0;
1320 	buf = outbuf;
1321 	len = min(bac->ifbac_len, buflen);
1322 	bzero(&bareq, sizeof(bareq));
1323 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1324 		if (len < sizeof(bareq))
1325 			goto out;
1326 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1327 		    sizeof(bareq.ifba_ifsname));
1328 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1329 		bareq.ifba_vlan = brt->brt_vlan;
1330 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1331 				time_uptime < brt->brt_expire)
1332 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1333 		else
1334 			bareq.ifba_expire = 0;
1335 		bareq.ifba_flags = brt->brt_flags;
1336 
1337 		memcpy(buf, &bareq, sizeof(bareq));
1338 		count++;
1339 		buf += sizeof(bareq);
1340 		len -= sizeof(bareq);
1341 	}
1342 out:
1343 	BRIDGE_UNLOCK(sc);
1344 	bac->ifbac_len = sizeof(bareq) * count;
1345 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1346 	BRIDGE_LOCK(sc);
1347 	free(outbuf, M_TEMP);
1348 	return (error);
1349 }
1350 
1351 static int
1352 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1353 {
1354 	struct ifbareq *req = arg;
1355 	struct bridge_iflist *bif;
1356 	int error;
1357 
1358 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1359 	if (bif == NULL)
1360 		return (ENOENT);
1361 
1362 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1363 	    req->ifba_flags);
1364 
1365 	return (error);
1366 }
1367 
1368 static int
1369 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1370 {
1371 	struct ifbrparam *param = arg;
1372 
1373 	sc->sc_brttimeout = param->ifbrp_ctime;
1374 	return (0);
1375 }
1376 
1377 static int
1378 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1379 {
1380 	struct ifbrparam *param = arg;
1381 
1382 	param->ifbrp_ctime = sc->sc_brttimeout;
1383 	return (0);
1384 }
1385 
1386 static int
1387 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1388 {
1389 	struct ifbareq *req = arg;
1390 
1391 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1392 }
1393 
1394 static int
1395 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1396 {
1397 	struct ifbreq *req = arg;
1398 
1399 	bridge_rtflush(sc, req->ifbr_ifsflags);
1400 	return (0);
1401 }
1402 
1403 static int
1404 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1405 {
1406 	struct ifbrparam *param = arg;
1407 	struct bstp_state *bs = &sc->sc_stp;
1408 
1409 	param->ifbrp_prio = bs->bs_bridge_priority;
1410 	return (0);
1411 }
1412 
1413 static int
1414 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1415 {
1416 	struct ifbrparam *param = arg;
1417 
1418 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1419 }
1420 
1421 static int
1422 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1423 {
1424 	struct ifbrparam *param = arg;
1425 	struct bstp_state *bs = &sc->sc_stp;
1426 
1427 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1428 	return (0);
1429 }
1430 
1431 static int
1432 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1433 {
1434 	struct ifbrparam *param = arg;
1435 
1436 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1437 }
1438 
1439 static int
1440 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1441 {
1442 	struct ifbrparam *param = arg;
1443 	struct bstp_state *bs = &sc->sc_stp;
1444 
1445 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1446 	return (0);
1447 }
1448 
1449 static int
1450 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1451 {
1452 	struct ifbrparam *param = arg;
1453 
1454 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1455 }
1456 
1457 static int
1458 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1459 {
1460 	struct ifbrparam *param = arg;
1461 	struct bstp_state *bs = &sc->sc_stp;
1462 
1463 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1464 	return (0);
1465 }
1466 
1467 static int
1468 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1469 {
1470 	struct ifbrparam *param = arg;
1471 
1472 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1473 }
1474 
1475 static int
1476 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1477 {
1478 	struct ifbreq *req = arg;
1479 	struct bridge_iflist *bif;
1480 
1481 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1482 	if (bif == NULL)
1483 		return (ENOENT);
1484 
1485 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1486 }
1487 
1488 static int
1489 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1490 {
1491 	struct ifbreq *req = arg;
1492 	struct bridge_iflist *bif;
1493 
1494 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1495 	if (bif == NULL)
1496 		return (ENOENT);
1497 
1498 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1499 }
1500 
1501 static int
1502 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1503 {
1504 	struct ifbreq *req = arg;
1505 	struct bridge_iflist *bif;
1506 
1507 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1508 	if (bif == NULL)
1509 		return (ENOENT);
1510 
1511 	bif->bif_addrmax = req->ifbr_addrmax;
1512 	return (0);
1513 }
1514 
1515 static int
1516 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1517 {
1518 	struct ifbreq *req = arg;
1519 	struct bridge_iflist *bif = NULL;
1520 	struct ifnet *ifs;
1521 
1522 	ifs = ifunit(req->ifbr_ifsname);
1523 	if (ifs == NULL)
1524 		return (ENOENT);
1525 
1526 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1527 		if (ifs == bif->bif_ifp)
1528 			return (EBUSY);
1529 
1530 	if (ifs->if_bridge != NULL)
1531 		return (EBUSY);
1532 
1533 	switch (ifs->if_type) {
1534 		case IFT_ETHER:
1535 		case IFT_GIF:
1536 		case IFT_L2VLAN:
1537 			break;
1538 		default:
1539 			return (EINVAL);
1540 	}
1541 
1542 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1543 	if (bif == NULL)
1544 		return (ENOMEM);
1545 
1546 	bif->bif_ifp = ifs;
1547 	bif->bif_flags = IFBIF_SPAN;
1548 
1549 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1550 
1551 	return (0);
1552 }
1553 
1554 static int
1555 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1556 {
1557 	struct ifbreq *req = arg;
1558 	struct bridge_iflist *bif;
1559 	struct ifnet *ifs;
1560 
1561 	ifs = ifunit(req->ifbr_ifsname);
1562 	if (ifs == NULL)
1563 		return (ENOENT);
1564 
1565 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1566 		if (ifs == bif->bif_ifp)
1567 			break;
1568 
1569 	if (bif == NULL)
1570 		return (ENOENT);
1571 
1572 	bridge_delete_span(sc, bif);
1573 
1574 	return (0);
1575 }
1576 
1577 static int
1578 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1579 {
1580 	struct ifbropreq *req = arg;
1581 	struct bstp_state *bs = &sc->sc_stp;
1582 	struct bstp_port *root_port;
1583 
1584 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1585 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1586 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1587 
1588 	root_port = bs->bs_root_port;
1589 	if (root_port == NULL)
1590 		req->ifbop_root_port = 0;
1591 	else
1592 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1593 
1594 	req->ifbop_holdcount = bs->bs_txholdcount;
1595 	req->ifbop_priority = bs->bs_bridge_priority;
1596 	req->ifbop_protocol = bs->bs_protover;
1597 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1598 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1599 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1600 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1601 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1602 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1603 
1604 	return (0);
1605 }
1606 
1607 static int
1608 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1609 {
1610 	struct ifbrparam *param = arg;
1611 
1612 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1613 	return (0);
1614 }
1615 
1616 static int
1617 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1618 {
1619 	struct ifbpstpconf *bifstp = arg;
1620 	struct bridge_iflist *bif;
1621 	struct bstp_port *bp;
1622 	struct ifbpstpreq bpreq;
1623 	char *buf, *outbuf;
1624 	int count, buflen, len, error = 0;
1625 
1626 	count = 0;
1627 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1628 		if ((bif->bif_flags & IFBIF_STP) != 0)
1629 			count++;
1630 	}
1631 
1632 	buflen = sizeof(bpreq) * count;
1633 	if (bifstp->ifbpstp_len == 0) {
1634 		bifstp->ifbpstp_len = buflen;
1635 		return (0);
1636 	}
1637 
1638 	BRIDGE_UNLOCK(sc);
1639 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1640 	BRIDGE_LOCK(sc);
1641 
1642 	count = 0;
1643 	buf = outbuf;
1644 	len = min(bifstp->ifbpstp_len, buflen);
1645 	bzero(&bpreq, sizeof(bpreq));
1646 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1647 		if (len < sizeof(bpreq))
1648 			break;
1649 
1650 		if ((bif->bif_flags & IFBIF_STP) == 0)
1651 			continue;
1652 
1653 		bp = &bif->bif_stp;
1654 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1655 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1656 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1657 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1658 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1659 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1660 
1661 		memcpy(buf, &bpreq, sizeof(bpreq));
1662 		count++;
1663 		buf += sizeof(bpreq);
1664 		len -= sizeof(bpreq);
1665 	}
1666 
1667 	BRIDGE_UNLOCK(sc);
1668 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1669 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1670 	BRIDGE_LOCK(sc);
1671 	free(outbuf, M_TEMP);
1672 	return (error);
1673 }
1674 
1675 static int
1676 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1677 {
1678 	struct ifbrparam *param = arg;
1679 
1680 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1681 }
1682 
1683 static int
1684 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1685 {
1686 	struct ifbrparam *param = arg;
1687 
1688 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1689 }
1690 
1691 /*
1692  * bridge_ifdetach:
1693  *
1694  *	Detach an interface from a bridge.  Called when a member
1695  *	interface is detaching.
1696  */
1697 static void
1698 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1699 {
1700 	struct bridge_softc *sc = ifp->if_bridge;
1701 	struct bridge_iflist *bif;
1702 
1703 	if (ifp->if_flags & IFF_RENAMING)
1704 		return;
1705 
1706 	/* Check if the interface is a bridge member */
1707 	if (sc != NULL) {
1708 		BRIDGE_LOCK(sc);
1709 
1710 		bif = bridge_lookup_member_if(sc, ifp);
1711 		if (bif != NULL)
1712 			bridge_delete_member(sc, bif, 1);
1713 
1714 		BRIDGE_UNLOCK(sc);
1715 		return;
1716 	}
1717 
1718 	/* Check if the interface is a span port */
1719 	mtx_lock(&bridge_list_mtx);
1720 	LIST_FOREACH(sc, &bridge_list, sc_list) {
1721 		BRIDGE_LOCK(sc);
1722 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1723 			if (ifp == bif->bif_ifp) {
1724 				bridge_delete_span(sc, bif);
1725 				break;
1726 			}
1727 
1728 		BRIDGE_UNLOCK(sc);
1729 	}
1730 	mtx_unlock(&bridge_list_mtx);
1731 }
1732 
1733 /*
1734  * bridge_init:
1735  *
1736  *	Initialize a bridge interface.
1737  */
1738 static void
1739 bridge_init(void *xsc)
1740 {
1741 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1742 	struct ifnet *ifp = sc->sc_ifp;
1743 
1744 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1745 		return;
1746 
1747 	BRIDGE_LOCK(sc);
1748 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1749 	    bridge_timer, sc);
1750 
1751 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1752 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1753 
1754 	BRIDGE_UNLOCK(sc);
1755 }
1756 
1757 /*
1758  * bridge_stop:
1759  *
1760  *	Stop the bridge interface.
1761  */
1762 static void
1763 bridge_stop(struct ifnet *ifp, int disable)
1764 {
1765 	struct bridge_softc *sc = ifp->if_softc;
1766 
1767 	BRIDGE_LOCK_ASSERT(sc);
1768 
1769 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1770 		return;
1771 
1772 	callout_stop(&sc->sc_brcallout);
1773 	bstp_stop(&sc->sc_stp);
1774 
1775 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1776 
1777 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1778 }
1779 
1780 /*
1781  * bridge_enqueue:
1782  *
1783  *	Enqueue a packet on a bridge member interface.
1784  *
1785  */
1786 static int
1787 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1788 {
1789 	int len, err = 0;
1790 	short mflags;
1791 	struct mbuf *m0;
1792 
1793 	/* We may be sending a fragment so traverse the mbuf */
1794 	for (; m; m = m0) {
1795 		m0 = m->m_nextpkt;
1796 		m->m_nextpkt = NULL;
1797 		len = m->m_pkthdr.len;
1798 		mflags = m->m_flags;
1799 
1800 		/*
1801 		 * If underlying interface can not do VLAN tag insertion itself
1802 		 * then attach a packet tag that holds it.
1803 		 */
1804 		if ((m->m_flags & M_VLANTAG) &&
1805 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1806 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1807 			if (m == NULL) {
1808 				if_printf(dst_ifp,
1809 				    "unable to prepend VLAN header\n");
1810 				dst_ifp->if_oerrors++;
1811 				continue;
1812 			}
1813 			m->m_flags &= ~M_VLANTAG;
1814 		}
1815 
1816 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1817 			m_freem(m0);
1818 			sc->sc_ifp->if_oerrors++;
1819 			break;
1820 		}
1821 
1822 		sc->sc_ifp->if_opackets++;
1823 		sc->sc_ifp->if_obytes += len;
1824 		if (mflags & M_MCAST)
1825 			sc->sc_ifp->if_omcasts++;
1826 	}
1827 
1828 	return (err);
1829 }
1830 
1831 /*
1832  * bridge_dummynet:
1833  *
1834  * 	Receive a queued packet from dummynet and pass it on to the output
1835  * 	interface.
1836  *
1837  *	The mbuf has the Ethernet header already attached.
1838  */
1839 static void
1840 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1841 {
1842 	struct bridge_softc *sc;
1843 
1844 	sc = ifp->if_bridge;
1845 
1846 	/*
1847 	 * The packet didnt originate from a member interface. This should only
1848 	 * ever happen if a member interface is removed while packets are
1849 	 * queued for it.
1850 	 */
1851 	if (sc == NULL) {
1852 		m_freem(m);
1853 		return;
1854 	}
1855 
1856 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1857 #ifdef INET6
1858 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1859 #endif
1860 	    ) {
1861 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1862 			return;
1863 		if (m == NULL)
1864 			return;
1865 	}
1866 
1867 	bridge_enqueue(sc, ifp, m);
1868 }
1869 
1870 /*
1871  * bridge_output:
1872  *
1873  *	Send output from a bridge member interface.  This
1874  *	performs the bridging function for locally originated
1875  *	packets.
1876  *
1877  *	The mbuf has the Ethernet header already attached.  We must
1878  *	enqueue or free the mbuf before returning.
1879  */
1880 static int
1881 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1882     struct rtentry *rt)
1883 {
1884 	struct ether_header *eh;
1885 	struct ifnet *dst_if;
1886 	struct bridge_softc *sc;
1887 	uint16_t vlan;
1888 
1889 	if (m->m_len < ETHER_HDR_LEN) {
1890 		m = m_pullup(m, ETHER_HDR_LEN);
1891 		if (m == NULL)
1892 			return (0);
1893 	}
1894 
1895 	eh = mtod(m, struct ether_header *);
1896 	sc = ifp->if_bridge;
1897 	vlan = VLANTAGOF(m);
1898 
1899 	BRIDGE_LOCK(sc);
1900 
1901 	/*
1902 	 * If bridge is down, but the original output interface is up,
1903 	 * go ahead and send out that interface.  Otherwise, the packet
1904 	 * is dropped below.
1905 	 */
1906 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1907 		dst_if = ifp;
1908 		goto sendunicast;
1909 	}
1910 
1911 	/*
1912 	 * If the packet is a multicast, or we don't know a better way to
1913 	 * get there, send to all interfaces.
1914 	 */
1915 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1916 		dst_if = NULL;
1917 	else
1918 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1919 	if (dst_if == NULL) {
1920 		struct bridge_iflist *bif;
1921 		struct mbuf *mc;
1922 		int error = 0, used = 0;
1923 
1924 		bridge_span(sc, m);
1925 
1926 		BRIDGE_LOCK2REF(sc, error);
1927 		if (error) {
1928 			m_freem(m);
1929 			return (0);
1930 		}
1931 
1932 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1933 			dst_if = bif->bif_ifp;
1934 
1935 			if (dst_if->if_type == IFT_GIF)
1936 				continue;
1937 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1938 				continue;
1939 
1940 			/*
1941 			 * If this is not the original output interface,
1942 			 * and the interface is participating in spanning
1943 			 * tree, make sure the port is in a state that
1944 			 * allows forwarding.
1945 			 */
1946 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1947 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1948 				continue;
1949 
1950 			if (LIST_NEXT(bif, bif_next) == NULL) {
1951 				used = 1;
1952 				mc = m;
1953 			} else {
1954 				mc = m_copypacket(m, M_NOWAIT);
1955 				if (mc == NULL) {
1956 					sc->sc_ifp->if_oerrors++;
1957 					continue;
1958 				}
1959 			}
1960 
1961 			bridge_enqueue(sc, dst_if, mc);
1962 		}
1963 		if (used == 0)
1964 			m_freem(m);
1965 		BRIDGE_UNREF(sc);
1966 		return (0);
1967 	}
1968 
1969 sendunicast:
1970 	/*
1971 	 * XXX Spanning tree consideration here?
1972 	 */
1973 
1974 	bridge_span(sc, m);
1975 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1976 		m_freem(m);
1977 		BRIDGE_UNLOCK(sc);
1978 		return (0);
1979 	}
1980 
1981 	BRIDGE_UNLOCK(sc);
1982 	bridge_enqueue(sc, dst_if, m);
1983 	return (0);
1984 }
1985 
1986 /*
1987  * bridge_transmit:
1988  *
1989  *	Do output on a bridge.
1990  *
1991  */
1992 static int
1993 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
1994 {
1995 	struct bridge_softc *sc;
1996 	struct ether_header *eh;
1997 	struct ifnet *dst_if;
1998 	int error = 0;
1999 
2000 	sc = ifp->if_softc;
2001 
2002 	ETHER_BPF_MTAP(ifp, m);
2003 
2004 	eh = mtod(m, struct ether_header *);
2005 
2006 	BRIDGE_LOCK(sc);
2007 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2008 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2009 		BRIDGE_UNLOCK(sc);
2010 		error = bridge_enqueue(sc, dst_if, m);
2011 	} else
2012 		bridge_broadcast(sc, ifp, m, 0);
2013 
2014 	return (error);
2015 }
2016 
2017 /*
2018  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2019  */
2020 static void
2021 bridge_qflush(struct ifnet *ifp __unused)
2022 {
2023 }
2024 
2025 /*
2026  * bridge_forward:
2027  *
2028  *	The forwarding function of the bridge.
2029  *
2030  *	NOTE: Releases the lock on return.
2031  */
2032 static void
2033 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2034     struct mbuf *m)
2035 {
2036 	struct bridge_iflist *dbif;
2037 	struct ifnet *src_if, *dst_if, *ifp;
2038 	struct ether_header *eh;
2039 	uint16_t vlan;
2040 	uint8_t *dst;
2041 	int error;
2042 
2043 	src_if = m->m_pkthdr.rcvif;
2044 	ifp = sc->sc_ifp;
2045 
2046 	ifp->if_ipackets++;
2047 	ifp->if_ibytes += m->m_pkthdr.len;
2048 	vlan = VLANTAGOF(m);
2049 
2050 	if ((sbif->bif_flags & IFBIF_STP) &&
2051 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2052 		goto drop;
2053 
2054 	eh = mtod(m, struct ether_header *);
2055 	dst = eh->ether_dhost;
2056 
2057 	/* If the interface is learning, record the address. */
2058 	if (sbif->bif_flags & IFBIF_LEARNING) {
2059 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2060 		    sbif, 0, IFBAF_DYNAMIC);
2061 		/*
2062 		 * If the interface has addresses limits then deny any source
2063 		 * that is not in the cache.
2064 		 */
2065 		if (error && sbif->bif_addrmax)
2066 			goto drop;
2067 	}
2068 
2069 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2070 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2071 		goto drop;
2072 
2073 	/*
2074 	 * At this point, the port either doesn't participate
2075 	 * in spanning tree or it is in the forwarding state.
2076 	 */
2077 
2078 	/*
2079 	 * If the packet is unicast, destined for someone on
2080 	 * "this" side of the bridge, drop it.
2081 	 */
2082 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2083 		dst_if = bridge_rtlookup(sc, dst, vlan);
2084 		if (src_if == dst_if)
2085 			goto drop;
2086 	} else {
2087 		/*
2088 		 * Check if its a reserved multicast address, any address
2089 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2090 		 * bridge.
2091 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2092 		 */
2093 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2094 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2095 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2096 			goto drop;
2097 
2098 		/* ...forward it to all interfaces. */
2099 		ifp->if_imcasts++;
2100 		dst_if = NULL;
2101 	}
2102 
2103 	/*
2104 	 * If we have a destination interface which is a member of our bridge,
2105 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2106 	 * For broadcast or multicast packets, don't bother because it will
2107 	 * be reinjected into ether_input. We do this before we pass the packets
2108 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2109 	 * drop the packet, or possibly modify it, making it difficult to debug
2110 	 * firewall issues on the bridge.
2111 	 */
2112 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2113 		ETHER_BPF_MTAP(ifp, m);
2114 
2115 	/* run the packet filter */
2116 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2117 #ifdef INET6
2118 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2119 #endif
2120 	    ) {
2121 		BRIDGE_UNLOCK(sc);
2122 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2123 			return;
2124 		if (m == NULL)
2125 			return;
2126 		BRIDGE_LOCK(sc);
2127 	}
2128 
2129 	if (dst_if == NULL) {
2130 		bridge_broadcast(sc, src_if, m, 1);
2131 		return;
2132 	}
2133 
2134 	/*
2135 	 * At this point, we're dealing with a unicast frame
2136 	 * going to a different interface.
2137 	 */
2138 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2139 		goto drop;
2140 
2141 	dbif = bridge_lookup_member_if(sc, dst_if);
2142 	if (dbif == NULL)
2143 		/* Not a member of the bridge (anymore?) */
2144 		goto drop;
2145 
2146 	/* Private segments can not talk to each other */
2147 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2148 		goto drop;
2149 
2150 	if ((dbif->bif_flags & IFBIF_STP) &&
2151 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2152 		goto drop;
2153 
2154 	BRIDGE_UNLOCK(sc);
2155 
2156 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2157 #ifdef INET6
2158 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2159 #endif
2160 	    ) {
2161 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2162 			return;
2163 		if (m == NULL)
2164 			return;
2165 	}
2166 
2167 	bridge_enqueue(sc, dst_if, m);
2168 	return;
2169 
2170 drop:
2171 	BRIDGE_UNLOCK(sc);
2172 	m_freem(m);
2173 }
2174 
2175 /*
2176  * bridge_input:
2177  *
2178  *	Receive input from a member interface.  Queue the packet for
2179  *	bridging if it is not for us.
2180  */
2181 static struct mbuf *
2182 bridge_input(struct ifnet *ifp, struct mbuf *m)
2183 {
2184 	struct bridge_softc *sc = ifp->if_bridge;
2185 	struct bridge_iflist *bif, *bif2;
2186 	struct ifnet *bifp;
2187 	struct ether_header *eh;
2188 	struct mbuf *mc, *mc2;
2189 	uint16_t vlan;
2190 	int error;
2191 
2192 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2193 		return (m);
2194 
2195 	bifp = sc->sc_ifp;
2196 	vlan = VLANTAGOF(m);
2197 
2198 	/*
2199 	 * Implement support for bridge monitoring. If this flag has been
2200 	 * set on this interface, discard the packet once we push it through
2201 	 * the bpf(4) machinery, but before we do, increment the byte and
2202 	 * packet counters associated with this interface.
2203 	 */
2204 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2205 		m->m_pkthdr.rcvif  = bifp;
2206 		ETHER_BPF_MTAP(bifp, m);
2207 		bifp->if_ipackets++;
2208 		bifp->if_ibytes += m->m_pkthdr.len;
2209 		m_freem(m);
2210 		return (NULL);
2211 	}
2212 	BRIDGE_LOCK(sc);
2213 	bif = bridge_lookup_member_if(sc, ifp);
2214 	if (bif == NULL) {
2215 		BRIDGE_UNLOCK(sc);
2216 		return (m);
2217 	}
2218 
2219 	eh = mtod(m, struct ether_header *);
2220 
2221 	bridge_span(sc, m);
2222 
2223 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2224 		/* Tap off 802.1D packets; they do not get forwarded. */
2225 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2226 		    ETHER_ADDR_LEN) == 0) {
2227 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2228 			BRIDGE_UNLOCK(sc);
2229 			return (NULL);
2230 		}
2231 
2232 		if ((bif->bif_flags & IFBIF_STP) &&
2233 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2234 			BRIDGE_UNLOCK(sc);
2235 			return (m);
2236 		}
2237 
2238 		/*
2239 		 * Make a deep copy of the packet and enqueue the copy
2240 		 * for bridge processing; return the original packet for
2241 		 * local processing.
2242 		 */
2243 		mc = m_dup(m, M_NOWAIT);
2244 		if (mc == NULL) {
2245 			BRIDGE_UNLOCK(sc);
2246 			return (m);
2247 		}
2248 
2249 		/* Perform the bridge forwarding function with the copy. */
2250 		bridge_forward(sc, bif, mc);
2251 
2252 		/*
2253 		 * Reinject the mbuf as arriving on the bridge so we have a
2254 		 * chance at claiming multicast packets. We can not loop back
2255 		 * here from ether_input as a bridge is never a member of a
2256 		 * bridge.
2257 		 */
2258 		KASSERT(bifp->if_bridge == NULL,
2259 		    ("loop created in bridge_input"));
2260 		mc2 = m_dup(m, M_NOWAIT);
2261 		if (mc2 != NULL) {
2262 			/* Keep the layer3 header aligned */
2263 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2264 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2265 		}
2266 		if (mc2 != NULL) {
2267 			mc2->m_pkthdr.rcvif = bifp;
2268 			(*bifp->if_input)(bifp, mc2);
2269 		}
2270 
2271 		/* Return the original packet for local processing. */
2272 		return (m);
2273 	}
2274 
2275 	if ((bif->bif_flags & IFBIF_STP) &&
2276 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2277 		BRIDGE_UNLOCK(sc);
2278 		return (m);
2279 	}
2280 
2281 #if (defined(INET) || defined(INET6))
2282 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2283 	|| ((iface)->if_carp \
2284 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2285 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2286 	|| ((iface)->if_carp \
2287 	    && (*carp_forus_p)((iface), eh->ether_shost))
2288 #else
2289 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2290 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2291 #endif
2292 
2293 #ifdef INET6
2294 #   define OR_PFIL_HOOKED_INET6 \
2295 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2296 #else
2297 #   define OR_PFIL_HOOKED_INET6
2298 #endif
2299 
2300 #define GRAB_OUR_PACKETS(iface) \
2301 	if ((iface)->if_type == IFT_GIF) \
2302 		continue; \
2303 	/* It is destined for us. */ \
2304 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2305 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2306 	    ) {								\
2307 		if ((iface)->if_type == IFT_BRIDGE) {			\
2308 			ETHER_BPF_MTAP(iface, m);			\
2309 			iface->if_ipackets++;				\
2310 			iface->if_ibytes += m->m_pkthdr.len;		\
2311 			/* Filter on the physical interface. */		\
2312 			if (pfil_local_phys &&				\
2313 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2314 			     OR_PFIL_HOOKED_INET6)) {			\
2315 				if (bridge_pfil(&m, NULL, ifp,		\
2316 				    PFIL_IN) != 0 || m == NULL) {	\
2317 					BRIDGE_UNLOCK(sc);		\
2318 					return (NULL);			\
2319 				}					\
2320 				eh = mtod(m, struct ether_header *);	\
2321 			}						\
2322 		}							\
2323 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2324 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2325 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2326 			if (error && bif->bif_addrmax) {		\
2327 				BRIDGE_UNLOCK(sc);			\
2328 				m_freem(m);				\
2329 				return (NULL);				\
2330 			}						\
2331 		}							\
2332 		m->m_pkthdr.rcvif = iface;				\
2333 		BRIDGE_UNLOCK(sc);					\
2334 		return (m);						\
2335 	}								\
2336 									\
2337 	/* We just received a packet that we sent out. */		\
2338 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2339 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2340 	    ) {								\
2341 		BRIDGE_UNLOCK(sc);					\
2342 		m_freem(m);						\
2343 		return (NULL);						\
2344 	}
2345 
2346 	/*
2347 	 * Unicast.  Make sure it's not for the bridge.
2348 	 */
2349 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2350 
2351 	/*
2352 	 * Give a chance for ifp at first priority. This will help when	the
2353 	 * packet comes through the interface like VLAN's with the same MACs
2354 	 * on several interfaces from the same bridge. This also will save
2355 	 * some CPU cycles in case the destination interface and the input
2356 	 * interface (eq ifp) are the same.
2357 	 */
2358 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2359 
2360 	/* Now check the all bridge members. */
2361 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2362 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2363 	}
2364 
2365 #undef OR_CARP_CHECK_WE_ARE_DST
2366 #undef OR_CARP_CHECK_WE_ARE_SRC
2367 #undef OR_PFIL_HOOKED_INET6
2368 #undef GRAB_OUR_PACKETS
2369 
2370 	/* Perform the bridge forwarding function. */
2371 	bridge_forward(sc, bif, m);
2372 
2373 	return (NULL);
2374 }
2375 
2376 /*
2377  * bridge_broadcast:
2378  *
2379  *	Send a frame to all interfaces that are members of
2380  *	the bridge, except for the one on which the packet
2381  *	arrived.
2382  *
2383  *	NOTE: Releases the lock on return.
2384  */
2385 static void
2386 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2387     struct mbuf *m, int runfilt)
2388 {
2389 	struct bridge_iflist *dbif, *sbif;
2390 	struct mbuf *mc;
2391 	struct ifnet *dst_if;
2392 	int error = 0, used = 0, i;
2393 
2394 	sbif = bridge_lookup_member_if(sc, src_if);
2395 
2396 	BRIDGE_LOCK2REF(sc, error);
2397 	if (error) {
2398 		m_freem(m);
2399 		return;
2400 	}
2401 
2402 	/* Filter on the bridge interface before broadcasting */
2403 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2404 #ifdef INET6
2405 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2406 #endif
2407 	    )) {
2408 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2409 			goto out;
2410 		if (m == NULL)
2411 			goto out;
2412 	}
2413 
2414 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2415 		dst_if = dbif->bif_ifp;
2416 		if (dst_if == src_if)
2417 			continue;
2418 
2419 		/* Private segments can not talk to each other */
2420 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2421 			continue;
2422 
2423 		if ((dbif->bif_flags & IFBIF_STP) &&
2424 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2425 			continue;
2426 
2427 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2428 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2429 			continue;
2430 
2431 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2432 			continue;
2433 
2434 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2435 			mc = m;
2436 			used = 1;
2437 		} else {
2438 			mc = m_dup(m, M_NOWAIT);
2439 			if (mc == NULL) {
2440 				sc->sc_ifp->if_oerrors++;
2441 				continue;
2442 			}
2443 		}
2444 
2445 		/*
2446 		 * Filter on the output interface. Pass a NULL bridge interface
2447 		 * pointer so we do not redundantly filter on the bridge for
2448 		 * each interface we broadcast on.
2449 		 */
2450 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2451 #ifdef INET6
2452 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2453 #endif
2454 		    )) {
2455 			if (used == 0) {
2456 				/* Keep the layer3 header aligned */
2457 				i = min(mc->m_pkthdr.len, max_protohdr);
2458 				mc = m_copyup(mc, i, ETHER_ALIGN);
2459 				if (mc == NULL) {
2460 					sc->sc_ifp->if_oerrors++;
2461 					continue;
2462 				}
2463 			}
2464 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2465 				continue;
2466 			if (mc == NULL)
2467 				continue;
2468 		}
2469 
2470 		bridge_enqueue(sc, dst_if, mc);
2471 	}
2472 	if (used == 0)
2473 		m_freem(m);
2474 
2475 out:
2476 	BRIDGE_UNREF(sc);
2477 }
2478 
2479 /*
2480  * bridge_span:
2481  *
2482  *	Duplicate a packet out one or more interfaces that are in span mode,
2483  *	the original mbuf is unmodified.
2484  */
2485 static void
2486 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2487 {
2488 	struct bridge_iflist *bif;
2489 	struct ifnet *dst_if;
2490 	struct mbuf *mc;
2491 
2492 	if (LIST_EMPTY(&sc->sc_spanlist))
2493 		return;
2494 
2495 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2496 		dst_if = bif->bif_ifp;
2497 
2498 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2499 			continue;
2500 
2501 		mc = m_copypacket(m, M_NOWAIT);
2502 		if (mc == NULL) {
2503 			sc->sc_ifp->if_oerrors++;
2504 			continue;
2505 		}
2506 
2507 		bridge_enqueue(sc, dst_if, mc);
2508 	}
2509 }
2510 
2511 /*
2512  * bridge_rtupdate:
2513  *
2514  *	Add a bridge routing entry.
2515  */
2516 static int
2517 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2518     struct bridge_iflist *bif, int setflags, uint8_t flags)
2519 {
2520 	struct bridge_rtnode *brt;
2521 	int error;
2522 
2523 	BRIDGE_LOCK_ASSERT(sc);
2524 
2525 	/* Check the source address is valid and not multicast. */
2526 	if (ETHER_IS_MULTICAST(dst) ||
2527 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2528 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2529 		return (EINVAL);
2530 
2531 	/* 802.1p frames map to vlan 1 */
2532 	if (vlan == 0)
2533 		vlan = 1;
2534 
2535 	/*
2536 	 * A route for this destination might already exist.  If so,
2537 	 * update it, otherwise create a new one.
2538 	 */
2539 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2540 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2541 			sc->sc_brtexceeded++;
2542 			return (ENOSPC);
2543 		}
2544 		/* Check per interface address limits (if enabled) */
2545 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2546 			bif->bif_addrexceeded++;
2547 			return (ENOSPC);
2548 		}
2549 
2550 		/*
2551 		 * Allocate a new bridge forwarding node, and
2552 		 * initialize the expiration time and Ethernet
2553 		 * address.
2554 		 */
2555 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2556 		if (brt == NULL)
2557 			return (ENOMEM);
2558 
2559 		if (bif->bif_flags & IFBIF_STICKY)
2560 			brt->brt_flags = IFBAF_STICKY;
2561 		else
2562 			brt->brt_flags = IFBAF_DYNAMIC;
2563 
2564 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2565 		brt->brt_vlan = vlan;
2566 
2567 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2568 			uma_zfree(bridge_rtnode_zone, brt);
2569 			return (error);
2570 		}
2571 		brt->brt_dst = bif;
2572 		bif->bif_addrcnt++;
2573 	}
2574 
2575 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2576 	    brt->brt_dst != bif) {
2577 		brt->brt_dst->bif_addrcnt--;
2578 		brt->brt_dst = bif;
2579 		brt->brt_dst->bif_addrcnt++;
2580 	}
2581 
2582 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2583 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2584 	if (setflags)
2585 		brt->brt_flags = flags;
2586 
2587 	return (0);
2588 }
2589 
2590 /*
2591  * bridge_rtlookup:
2592  *
2593  *	Lookup the destination interface for an address.
2594  */
2595 static struct ifnet *
2596 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2597 {
2598 	struct bridge_rtnode *brt;
2599 
2600 	BRIDGE_LOCK_ASSERT(sc);
2601 
2602 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2603 		return (NULL);
2604 
2605 	return (brt->brt_ifp);
2606 }
2607 
2608 /*
2609  * bridge_rttrim:
2610  *
2611  *	Trim the routine table so that we have a number
2612  *	of routing entries less than or equal to the
2613  *	maximum number.
2614  */
2615 static void
2616 bridge_rttrim(struct bridge_softc *sc)
2617 {
2618 	struct bridge_rtnode *brt, *nbrt;
2619 
2620 	BRIDGE_LOCK_ASSERT(sc);
2621 
2622 	/* Make sure we actually need to do this. */
2623 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2624 		return;
2625 
2626 	/* Force an aging cycle; this might trim enough addresses. */
2627 	bridge_rtage(sc);
2628 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2629 		return;
2630 
2631 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2632 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2633 			bridge_rtnode_destroy(sc, brt);
2634 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2635 				return;
2636 		}
2637 	}
2638 }
2639 
2640 /*
2641  * bridge_timer:
2642  *
2643  *	Aging timer for the bridge.
2644  */
2645 static void
2646 bridge_timer(void *arg)
2647 {
2648 	struct bridge_softc *sc = arg;
2649 
2650 	BRIDGE_LOCK_ASSERT(sc);
2651 
2652 	bridge_rtage(sc);
2653 
2654 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2655 		callout_reset(&sc->sc_brcallout,
2656 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2657 }
2658 
2659 /*
2660  * bridge_rtage:
2661  *
2662  *	Perform an aging cycle.
2663  */
2664 static void
2665 bridge_rtage(struct bridge_softc *sc)
2666 {
2667 	struct bridge_rtnode *brt, *nbrt;
2668 
2669 	BRIDGE_LOCK_ASSERT(sc);
2670 
2671 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2672 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2673 			if (time_uptime >= brt->brt_expire)
2674 				bridge_rtnode_destroy(sc, brt);
2675 		}
2676 	}
2677 }
2678 
2679 /*
2680  * bridge_rtflush:
2681  *
2682  *	Remove all dynamic addresses from the bridge.
2683  */
2684 static void
2685 bridge_rtflush(struct bridge_softc *sc, 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 (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2693 			bridge_rtnode_destroy(sc, brt);
2694 	}
2695 }
2696 
2697 /*
2698  * bridge_rtdaddr:
2699  *
2700  *	Remove an address from the table.
2701  */
2702 static int
2703 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2704 {
2705 	struct bridge_rtnode *brt;
2706 	int found = 0;
2707 
2708 	BRIDGE_LOCK_ASSERT(sc);
2709 
2710 	/*
2711 	 * If vlan is zero then we want to delete for all vlans so the lookup
2712 	 * may return more than one.
2713 	 */
2714 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2715 		bridge_rtnode_destroy(sc, brt);
2716 		found = 1;
2717 	}
2718 
2719 	return (found ? 0 : ENOENT);
2720 }
2721 
2722 /*
2723  * bridge_rtdelete:
2724  *
2725  *	Delete routes to a speicifc member interface.
2726  */
2727 static void
2728 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2729 {
2730 	struct bridge_rtnode *brt, *nbrt;
2731 
2732 	BRIDGE_LOCK_ASSERT(sc);
2733 
2734 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2735 		if (brt->brt_ifp == ifp && (full ||
2736 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2737 			bridge_rtnode_destroy(sc, brt);
2738 	}
2739 }
2740 
2741 /*
2742  * bridge_rtable_init:
2743  *
2744  *	Initialize the route table for this bridge.
2745  */
2746 static void
2747 bridge_rtable_init(struct bridge_softc *sc)
2748 {
2749 	int i;
2750 
2751 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2752 	    M_DEVBUF, M_WAITOK);
2753 
2754 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2755 		LIST_INIT(&sc->sc_rthash[i]);
2756 
2757 	sc->sc_rthash_key = arc4random();
2758 	LIST_INIT(&sc->sc_rtlist);
2759 }
2760 
2761 /*
2762  * bridge_rtable_fini:
2763  *
2764  *	Deconstruct the route table for this bridge.
2765  */
2766 static void
2767 bridge_rtable_fini(struct bridge_softc *sc)
2768 {
2769 
2770 	KASSERT(sc->sc_brtcnt == 0,
2771 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2772 	free(sc->sc_rthash, M_DEVBUF);
2773 }
2774 
2775 /*
2776  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2777  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2778  */
2779 #define	mix(a, b, c)							\
2780 do {									\
2781 	a -= b; a -= c; a ^= (c >> 13);					\
2782 	b -= c; b -= a; b ^= (a << 8);					\
2783 	c -= a; c -= b; c ^= (b >> 13);					\
2784 	a -= b; a -= c; a ^= (c >> 12);					\
2785 	b -= c; b -= a; b ^= (a << 16);					\
2786 	c -= a; c -= b; c ^= (b >> 5);					\
2787 	a -= b; a -= c; a ^= (c >> 3);					\
2788 	b -= c; b -= a; b ^= (a << 10);					\
2789 	c -= a; c -= b; c ^= (b >> 15);					\
2790 } while (/*CONSTCOND*/0)
2791 
2792 static __inline uint32_t
2793 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2794 {
2795 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2796 
2797 	b += addr[5] << 8;
2798 	b += addr[4];
2799 	a += addr[3] << 24;
2800 	a += addr[2] << 16;
2801 	a += addr[1] << 8;
2802 	a += addr[0];
2803 
2804 	mix(a, b, c);
2805 
2806 	return (c & BRIDGE_RTHASH_MASK);
2807 }
2808 
2809 #undef mix
2810 
2811 static int
2812 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2813 {
2814 	int i, d;
2815 
2816 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2817 		d = ((int)a[i]) - ((int)b[i]);
2818 	}
2819 
2820 	return (d);
2821 }
2822 
2823 /*
2824  * bridge_rtnode_lookup:
2825  *
2826  *	Look up a bridge route node for the specified destination. Compare the
2827  *	vlan id or if zero then just return the first match.
2828  */
2829 static struct bridge_rtnode *
2830 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2831 {
2832 	struct bridge_rtnode *brt;
2833 	uint32_t hash;
2834 	int dir;
2835 
2836 	BRIDGE_LOCK_ASSERT(sc);
2837 
2838 	hash = bridge_rthash(sc, addr);
2839 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2840 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2841 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2842 			return (brt);
2843 		if (dir > 0)
2844 			return (NULL);
2845 	}
2846 
2847 	return (NULL);
2848 }
2849 
2850 /*
2851  * bridge_rtnode_insert:
2852  *
2853  *	Insert the specified bridge node into the route table.  We
2854  *	assume the entry is not already in the table.
2855  */
2856 static int
2857 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2858 {
2859 	struct bridge_rtnode *lbrt;
2860 	uint32_t hash;
2861 	int dir;
2862 
2863 	BRIDGE_LOCK_ASSERT(sc);
2864 
2865 	hash = bridge_rthash(sc, brt->brt_addr);
2866 
2867 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2868 	if (lbrt == NULL) {
2869 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2870 		goto out;
2871 	}
2872 
2873 	do {
2874 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2875 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2876 			return (EEXIST);
2877 		if (dir > 0) {
2878 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2879 			goto out;
2880 		}
2881 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2882 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2883 			goto out;
2884 		}
2885 		lbrt = LIST_NEXT(lbrt, brt_hash);
2886 	} while (lbrt != NULL);
2887 
2888 #ifdef DIAGNOSTIC
2889 	panic("bridge_rtnode_insert: impossible");
2890 #endif
2891 
2892 out:
2893 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2894 	sc->sc_brtcnt++;
2895 
2896 	return (0);
2897 }
2898 
2899 /*
2900  * bridge_rtnode_destroy:
2901  *
2902  *	Destroy a bridge rtnode.
2903  */
2904 static void
2905 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2906 {
2907 	BRIDGE_LOCK_ASSERT(sc);
2908 
2909 	LIST_REMOVE(brt, brt_hash);
2910 
2911 	LIST_REMOVE(brt, brt_list);
2912 	sc->sc_brtcnt--;
2913 	brt->brt_dst->bif_addrcnt--;
2914 	uma_zfree(bridge_rtnode_zone, brt);
2915 }
2916 
2917 /*
2918  * bridge_rtable_expire:
2919  *
2920  *	Set the expiry time for all routes on an interface.
2921  */
2922 static void
2923 bridge_rtable_expire(struct ifnet *ifp, int age)
2924 {
2925 	struct bridge_softc *sc = ifp->if_bridge;
2926 	struct bridge_rtnode *brt;
2927 
2928 	BRIDGE_LOCK(sc);
2929 
2930 	/*
2931 	 * If the age is zero then flush, otherwise set all the expiry times to
2932 	 * age for the interface
2933 	 */
2934 	if (age == 0)
2935 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2936 	else {
2937 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2938 			/* Cap the expiry time to 'age' */
2939 			if (brt->brt_ifp == ifp &&
2940 			    brt->brt_expire > time_uptime + age &&
2941 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2942 				brt->brt_expire = time_uptime + age;
2943 		}
2944 	}
2945 	BRIDGE_UNLOCK(sc);
2946 }
2947 
2948 /*
2949  * bridge_state_change:
2950  *
2951  *	Callback from the bridgestp code when a port changes states.
2952  */
2953 static void
2954 bridge_state_change(struct ifnet *ifp, int state)
2955 {
2956 	struct bridge_softc *sc = ifp->if_bridge;
2957 	static const char *stpstates[] = {
2958 		"disabled",
2959 		"listening",
2960 		"learning",
2961 		"forwarding",
2962 		"blocking",
2963 		"discarding"
2964 	};
2965 
2966 	if (log_stp)
2967 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2968 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2969 }
2970 
2971 /*
2972  * Send bridge packets through pfil if they are one of the types pfil can deal
2973  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
2974  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2975  * that interface.
2976  */
2977 static int
2978 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2979 {
2980 	int snap, error, i, hlen;
2981 	struct ether_header *eh1, eh2;
2982 	struct ip *ip;
2983 	struct llc llc1;
2984 	u_int16_t ether_type;
2985 
2986 	snap = 0;
2987 	error = -1;	/* Default error if not error == 0 */
2988 
2989 #if 0
2990 	/* we may return with the IP fields swapped, ensure its not shared */
2991 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2992 #endif
2993 
2994 	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2995 		return (0); /* filtering is disabled */
2996 
2997 	i = min((*mp)->m_pkthdr.len, max_protohdr);
2998 	if ((*mp)->m_len < i) {
2999 	    *mp = m_pullup(*mp, i);
3000 	    if (*mp == NULL) {
3001 		printf("%s: m_pullup failed\n", __func__);
3002 		return (-1);
3003 	    }
3004 	}
3005 
3006 	eh1 = mtod(*mp, struct ether_header *);
3007 	ether_type = ntohs(eh1->ether_type);
3008 
3009 	/*
3010 	 * Check for SNAP/LLC.
3011 	 */
3012 	if (ether_type < ETHERMTU) {
3013 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3014 
3015 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3016 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3017 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3018 		    llc2->llc_control == LLC_UI) {
3019 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3020 			snap = 1;
3021 		}
3022 	}
3023 
3024 	/*
3025 	 * If we're trying to filter bridge traffic, don't look at anything
3026 	 * other than IP and ARP traffic.  If the filter doesn't understand
3027 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3028 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3029 	 * but of course we don't have an AppleTalk filter to begin with.
3030 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3031 	 * ARP traffic.)
3032 	 */
3033 	switch (ether_type) {
3034 		case ETHERTYPE_ARP:
3035 		case ETHERTYPE_REVARP:
3036 			if (pfil_ipfw_arp == 0)
3037 				return (0); /* Automatically pass */
3038 			break;
3039 
3040 		case ETHERTYPE_IP:
3041 #ifdef INET6
3042 		case ETHERTYPE_IPV6:
3043 #endif /* INET6 */
3044 			break;
3045 		default:
3046 			/*
3047 			 * Check to see if the user wants to pass non-ip
3048 			 * packets, these will not be checked by pfil(9) and
3049 			 * passed unconditionally so the default is to drop.
3050 			 */
3051 			if (pfil_onlyip)
3052 				goto bad;
3053 	}
3054 
3055 	/* Run the packet through pfil before stripping link headers */
3056 	if (PFIL_HOOKED(&V_link_pfil_hook) && pfil_ipfw != 0 &&
3057 			dir == PFIL_OUT && ifp != NULL) {
3058 
3059 		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3060 
3061 		if (*mp == NULL || error != 0) /* packet consumed by filter */
3062 			return (error);
3063 	}
3064 
3065 	/* Strip off the Ethernet header and keep a copy. */
3066 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3067 	m_adj(*mp, ETHER_HDR_LEN);
3068 
3069 	/* Strip off snap header, if present */
3070 	if (snap) {
3071 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3072 		m_adj(*mp, sizeof(struct llc));
3073 	}
3074 
3075 	/*
3076 	 * Check the IP header for alignment and errors
3077 	 */
3078 	if (dir == PFIL_IN) {
3079 		switch (ether_type) {
3080 			case ETHERTYPE_IP:
3081 				error = bridge_ip_checkbasic(mp);
3082 				break;
3083 #ifdef INET6
3084 			case ETHERTYPE_IPV6:
3085 				error = bridge_ip6_checkbasic(mp);
3086 				break;
3087 #endif /* INET6 */
3088 			default:
3089 				error = 0;
3090 		}
3091 		if (error)
3092 			goto bad;
3093 	}
3094 
3095 	error = 0;
3096 
3097 	/*
3098 	 * Run the packet through pfil
3099 	 */
3100 	switch (ether_type) {
3101 	case ETHERTYPE_IP:
3102 		/*
3103 		 * Run pfil on the member interface and the bridge, both can
3104 		 * be skipped by clearing pfil_member or pfil_bridge.
3105 		 *
3106 		 * Keep the order:
3107 		 *   in_if -> bridge_if -> out_if
3108 		 */
3109 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3110 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3111 					dir, NULL);
3112 
3113 		if (*mp == NULL || error != 0) /* filter may consume */
3114 			break;
3115 
3116 		if (pfil_member && ifp != NULL)
3117 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3118 					dir, NULL);
3119 
3120 		if (*mp == NULL || error != 0) /* filter may consume */
3121 			break;
3122 
3123 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3124 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3125 					dir, NULL);
3126 
3127 		if (*mp == NULL || error != 0) /* filter may consume */
3128 			break;
3129 
3130 		/* check if we need to fragment the packet */
3131 		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3132 			i = (*mp)->m_pkthdr.len;
3133 			if (i > ifp->if_mtu) {
3134 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3135 					    &llc1);
3136 				return (error);
3137 			}
3138 		}
3139 
3140 		/* Recalculate the ip checksum. */
3141 		ip = mtod(*mp, struct ip *);
3142 		hlen = ip->ip_hl << 2;
3143 		if (hlen < sizeof(struct ip))
3144 			goto bad;
3145 		if (hlen > (*mp)->m_len) {
3146 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3147 				goto bad;
3148 			ip = mtod(*mp, struct ip *);
3149 			if (ip == NULL)
3150 				goto bad;
3151 		}
3152 		ip->ip_sum = 0;
3153 		if (hlen == sizeof(struct ip))
3154 			ip->ip_sum = in_cksum_hdr(ip);
3155 		else
3156 			ip->ip_sum = in_cksum(*mp, hlen);
3157 
3158 		break;
3159 #ifdef INET6
3160 	case ETHERTYPE_IPV6:
3161 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3162 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3163 					dir, NULL);
3164 
3165 		if (*mp == NULL || error != 0) /* filter may consume */
3166 			break;
3167 
3168 		if (pfil_member && ifp != NULL)
3169 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3170 					dir, NULL);
3171 
3172 		if (*mp == NULL || error != 0) /* filter may consume */
3173 			break;
3174 
3175 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3176 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3177 					dir, NULL);
3178 		break;
3179 #endif
3180 	default:
3181 		error = 0;
3182 		break;
3183 	}
3184 
3185 	if (*mp == NULL)
3186 		return (error);
3187 	if (error != 0)
3188 		goto bad;
3189 
3190 	error = -1;
3191 
3192 	/*
3193 	 * Finally, put everything back the way it was and return
3194 	 */
3195 	if (snap) {
3196 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3197 		if (*mp == NULL)
3198 			return (error);
3199 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3200 	}
3201 
3202 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3203 	if (*mp == NULL)
3204 		return (error);
3205 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3206 
3207 	return (0);
3208 
3209 bad:
3210 	m_freem(*mp);
3211 	*mp = NULL;
3212 	return (error);
3213 }
3214 
3215 /*
3216  * Perform basic checks on header size since
3217  * pfil assumes ip_input has already processed
3218  * it for it.  Cut-and-pasted from ip_input.c.
3219  * Given how simple the IPv6 version is,
3220  * does the IPv4 version really need to be
3221  * this complicated?
3222  *
3223  * XXX Should we update ipstat here, or not?
3224  * XXX Right now we update ipstat but not
3225  * XXX csum_counter.
3226  */
3227 static int
3228 bridge_ip_checkbasic(struct mbuf **mp)
3229 {
3230 	struct mbuf *m = *mp;
3231 	struct ip *ip;
3232 	int len, hlen;
3233 	u_short sum;
3234 
3235 	if (*mp == NULL)
3236 		return (-1);
3237 
3238 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3239 		if ((m = m_copyup(m, sizeof(struct ip),
3240 			(max_linkhdr + 3) & ~3)) == NULL) {
3241 			/* XXXJRT new stat, please */
3242 			KMOD_IPSTAT_INC(ips_toosmall);
3243 			goto bad;
3244 		}
3245 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3246 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3247 			KMOD_IPSTAT_INC(ips_toosmall);
3248 			goto bad;
3249 		}
3250 	}
3251 	ip = mtod(m, struct ip *);
3252 	if (ip == NULL) goto bad;
3253 
3254 	if (ip->ip_v != IPVERSION) {
3255 		KMOD_IPSTAT_INC(ips_badvers);
3256 		goto bad;
3257 	}
3258 	hlen = ip->ip_hl << 2;
3259 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3260 		KMOD_IPSTAT_INC(ips_badhlen);
3261 		goto bad;
3262 	}
3263 	if (hlen > m->m_len) {
3264 		if ((m = m_pullup(m, hlen)) == 0) {
3265 			KMOD_IPSTAT_INC(ips_badhlen);
3266 			goto bad;
3267 		}
3268 		ip = mtod(m, struct ip *);
3269 		if (ip == NULL) goto bad;
3270 	}
3271 
3272 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3273 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3274 	} else {
3275 		if (hlen == sizeof(struct ip)) {
3276 			sum = in_cksum_hdr(ip);
3277 		} else {
3278 			sum = in_cksum(m, hlen);
3279 		}
3280 	}
3281 	if (sum) {
3282 		KMOD_IPSTAT_INC(ips_badsum);
3283 		goto bad;
3284 	}
3285 
3286 	/* Retrieve the packet length. */
3287 	len = ntohs(ip->ip_len);
3288 
3289 	/*
3290 	 * Check for additional length bogosity
3291 	 */
3292 	if (len < hlen) {
3293 		KMOD_IPSTAT_INC(ips_badlen);
3294 		goto bad;
3295 	}
3296 
3297 	/*
3298 	 * Check that the amount of data in the buffers
3299 	 * is as at least much as the IP header would have us expect.
3300 	 * Drop packet if shorter than we expect.
3301 	 */
3302 	if (m->m_pkthdr.len < len) {
3303 		KMOD_IPSTAT_INC(ips_tooshort);
3304 		goto bad;
3305 	}
3306 
3307 	/* Checks out, proceed */
3308 	*mp = m;
3309 	return (0);
3310 
3311 bad:
3312 	*mp = m;
3313 	return (-1);
3314 }
3315 
3316 #ifdef INET6
3317 /*
3318  * Same as above, but for IPv6.
3319  * Cut-and-pasted from ip6_input.c.
3320  * XXX Should we update ip6stat, or not?
3321  */
3322 static int
3323 bridge_ip6_checkbasic(struct mbuf **mp)
3324 {
3325 	struct mbuf *m = *mp;
3326 	struct ip6_hdr *ip6;
3327 
3328 	/*
3329 	 * If the IPv6 header is not aligned, slurp it up into a new
3330 	 * mbuf with space for link headers, in the event we forward
3331 	 * it.  Otherwise, if it is aligned, make sure the entire base
3332 	 * IPv6 header is in the first mbuf of the chain.
3333 	 */
3334 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3335 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3336 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3337 			    (max_linkhdr + 3) & ~3)) == NULL) {
3338 			/* XXXJRT new stat, please */
3339 			IP6STAT_INC(ip6s_toosmall);
3340 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3341 			goto bad;
3342 		}
3343 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3344 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3345 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3346 			IP6STAT_INC(ip6s_toosmall);
3347 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3348 			goto bad;
3349 		}
3350 	}
3351 
3352 	ip6 = mtod(m, struct ip6_hdr *);
3353 
3354 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3355 		IP6STAT_INC(ip6s_badvers);
3356 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3357 		goto bad;
3358 	}
3359 
3360 	/* Checks out, proceed */
3361 	*mp = m;
3362 	return (0);
3363 
3364 bad:
3365 	*mp = m;
3366 	return (-1);
3367 }
3368 #endif /* INET6 */
3369 
3370 /*
3371  * bridge_fragment:
3372  *
3373  *	Return a fragmented mbuf chain.
3374  */
3375 static int
3376 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3377     int snap, struct llc *llc)
3378 {
3379 	struct mbuf *m0;
3380 	struct ip *ip;
3381 	int error = -1;
3382 
3383 	if (m->m_len < sizeof(struct ip) &&
3384 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3385 		goto out;
3386 	ip = mtod(m, struct ip *);
3387 
3388 	m->m_pkthdr.csum_flags |= CSUM_IP;
3389 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3390 	if (error)
3391 		goto out;
3392 
3393 	/* walk the chain and re-add the Ethernet header */
3394 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3395 		if (error == 0) {
3396 			if (snap) {
3397 				M_PREPEND(m0, sizeof(struct llc), M_NOWAIT);
3398 				if (m0 == NULL) {
3399 					error = ENOBUFS;
3400 					continue;
3401 				}
3402 				bcopy(llc, mtod(m0, caddr_t),
3403 				    sizeof(struct llc));
3404 			}
3405 			M_PREPEND(m0, ETHER_HDR_LEN, M_NOWAIT);
3406 			if (m0 == NULL) {
3407 				error = ENOBUFS;
3408 				continue;
3409 			}
3410 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3411 		} else
3412 			m_freem(m);
3413 	}
3414 
3415 	if (error == 0)
3416 		KMOD_IPSTAT_INC(ips_fragmented);
3417 
3418 	return (error);
3419 
3420 out:
3421 	if (m != NULL)
3422 		m_freem(m);
3423 	return (error);
3424 }
3425 
3426 static void
3427 bridge_linkstate(struct ifnet *ifp)
3428 {
3429 	struct bridge_softc *sc = ifp->if_bridge;
3430 	struct bridge_iflist *bif;
3431 
3432 	BRIDGE_LOCK(sc);
3433 	bif = bridge_lookup_member_if(sc, ifp);
3434 	if (bif == NULL) {
3435 		BRIDGE_UNLOCK(sc);
3436 		return;
3437 	}
3438 	bridge_linkcheck(sc);
3439 	BRIDGE_UNLOCK(sc);
3440 
3441 	bstp_linkstate(&bif->bif_stp);
3442 }
3443 
3444 static void
3445 bridge_linkcheck(struct bridge_softc *sc)
3446 {
3447 	struct bridge_iflist *bif;
3448 	int new_link, hasls;
3449 
3450 	BRIDGE_LOCK_ASSERT(sc);
3451 	new_link = LINK_STATE_DOWN;
3452 	hasls = 0;
3453 	/* Our link is considered up if at least one of our ports is active */
3454 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3455 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3456 			hasls++;
3457 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3458 			new_link = LINK_STATE_UP;
3459 			break;
3460 		}
3461 	}
3462 	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3463 		/* If no interfaces support link-state then we default to up */
3464 		new_link = LINK_STATE_UP;
3465 	}
3466 	if_link_state_change(sc->sc_ifp, new_link);
3467 }
3468