xref: /freebsd/sys/net/if_bridge.c (revision c243e4902be8df1e643c76b5f18b68bb77cc5268)
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 int	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 IFC_SIMPLE_DECLARE(bridge, 0);
485 
486 static int
487 bridge_modevent(module_t mod, int type, void *data)
488 {
489 
490 	switch (type) {
491 	case MOD_LOAD:
492 		mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF);
493 		if_clone_attach(&bridge_cloner);
494 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
495 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
496 		    UMA_ALIGN_PTR, 0);
497 		LIST_INIT(&bridge_list);
498 		bridge_input_p = bridge_input;
499 		bridge_output_p = bridge_output;
500 		bridge_dn_p = bridge_dummynet;
501 		bridge_linkstate_p = bridge_linkstate;
502 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
503 		    ifnet_departure_event, bridge_ifdetach, NULL,
504 		    EVENTHANDLER_PRI_ANY);
505 		break;
506 	case MOD_UNLOAD:
507 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
508 		    bridge_detach_cookie);
509 		if_clone_detach(&bridge_cloner);
510 		uma_zdestroy(bridge_rtnode_zone);
511 		bridge_input_p = NULL;
512 		bridge_output_p = NULL;
513 		bridge_dn_p = NULL;
514 		bridge_linkstate_p = NULL;
515 		mtx_destroy(&bridge_list_mtx);
516 		break;
517 	default:
518 		return (EOPNOTSUPP);
519 	}
520 	return (0);
521 }
522 
523 static moduledata_t bridge_mod = {
524 	"if_bridge",
525 	bridge_modevent,
526 	0
527 };
528 
529 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
530 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
531 
532 /*
533  * handler for net.link.bridge.pfil_ipfw
534  */
535 static int
536 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
537 {
538 	int enable = pfil_ipfw;
539 	int error;
540 
541 	error = sysctl_handle_int(oidp, &enable, 0, req);
542 	enable = (enable) ? 1 : 0;
543 
544 	if (enable != pfil_ipfw) {
545 		pfil_ipfw = enable;
546 
547 		/*
548 		 * Disable pfil so that ipfw doesnt run twice, if the user
549 		 * really wants both then they can re-enable pfil_bridge and/or
550 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
551 		 * layer2 type.
552 		 */
553 		if (pfil_ipfw) {
554 			pfil_onlyip = 0;
555 			pfil_bridge = 0;
556 			pfil_member = 0;
557 		}
558 	}
559 
560 	return (error);
561 }
562 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW,
563 	    &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW");
564 
565 /*
566  * bridge_clone_create:
567  *
568  *	Create a new bridge instance.
569  */
570 static int
571 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
572 {
573 	struct bridge_softc *sc, *sc2;
574 	struct ifnet *bifp, *ifp;
575 	int fb, retry;
576 	unsigned long hostid;
577 
578 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
579 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
580 	if (ifp == NULL) {
581 		free(sc, M_DEVBUF);
582 		return (ENOSPC);
583 	}
584 
585 	BRIDGE_LOCK_INIT(sc);
586 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
587 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
588 
589 	/* Initialize our routing table. */
590 	bridge_rtable_init(sc);
591 
592 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
593 
594 	LIST_INIT(&sc->sc_iflist);
595 	LIST_INIT(&sc->sc_spanlist);
596 
597 	ifp->if_softc = sc;
598 	if_initname(ifp, ifc->ifc_name, unit);
599 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
600 	ifp->if_ioctl = bridge_ioctl;
601 	ifp->if_transmit = bridge_transmit;
602 	ifp->if_qflush = bridge_qflush;
603 	ifp->if_init = bridge_init;
604 	ifp->if_type = IFT_BRIDGE;
605 
606 	/*
607 	 * Generate an ethernet address with a locally administered address.
608 	 *
609 	 * Since we are using random ethernet addresses for the bridge, it is
610 	 * possible that we might have address collisions, so make sure that
611 	 * this hardware address isn't already in use on another bridge.
612 	 * The first try uses the hostid and falls back to arc4rand().
613 	 */
614 	fb = 0;
615 	getcredhostid(curthread->td_ucred, &hostid);
616 	for (retry = 1; retry != 0;) {
617 		if (fb || hostid == 0) {
618 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
619 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
620 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
621 		} else {
622 			sc->sc_defaddr[0] = 0x2;
623 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
624 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
625 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
626 			sc->sc_defaddr[4] =  hostid        & 0xff;
627 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
628 		}
629 
630 		fb = 1;
631 		retry = 0;
632 		mtx_lock(&bridge_list_mtx);
633 		LIST_FOREACH(sc2, &bridge_list, sc_list) {
634 			bifp = sc2->sc_ifp;
635 			if (memcmp(sc->sc_defaddr,
636 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0)
637 				retry = 1;
638 		}
639 		mtx_unlock(&bridge_list_mtx);
640 	}
641 
642 	bstp_attach(&sc->sc_stp, &bridge_ops);
643 	ether_ifattach(ifp, sc->sc_defaddr);
644 	/* Now undo some of the damage... */
645 	ifp->if_baudrate = 0;
646 	ifp->if_type = IFT_BRIDGE;
647 
648 	mtx_lock(&bridge_list_mtx);
649 	LIST_INSERT_HEAD(&bridge_list, sc, sc_list);
650 	mtx_unlock(&bridge_list_mtx);
651 
652 	return (0);
653 }
654 
655 /*
656  * bridge_clone_destroy:
657  *
658  *	Destroy a bridge instance.
659  */
660 static void
661 bridge_clone_destroy(struct ifnet *ifp)
662 {
663 	struct bridge_softc *sc = ifp->if_softc;
664 	struct bridge_iflist *bif;
665 
666 	BRIDGE_LOCK(sc);
667 
668 	bridge_stop(ifp, 1);
669 	ifp->if_flags &= ~IFF_UP;
670 
671 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
672 		bridge_delete_member(sc, bif, 0);
673 
674 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
675 		bridge_delete_span(sc, bif);
676 	}
677 
678 	BRIDGE_UNLOCK(sc);
679 
680 	callout_drain(&sc->sc_brcallout);
681 
682 	mtx_lock(&bridge_list_mtx);
683 	LIST_REMOVE(sc, sc_list);
684 	mtx_unlock(&bridge_list_mtx);
685 
686 	bstp_detach(&sc->sc_stp);
687 	ether_ifdetach(ifp);
688 	if_free(ifp);
689 
690 	/* Tear down the routing table. */
691 	bridge_rtable_fini(sc);
692 
693 	BRIDGE_LOCK_DESTROY(sc);
694 	free(sc, M_DEVBUF);
695 }
696 
697 /*
698  * bridge_ioctl:
699  *
700  *	Handle a control request from the operator.
701  */
702 static int
703 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
704 {
705 	struct bridge_softc *sc = ifp->if_softc;
706 	struct ifreq *ifr = (struct ifreq *)data;
707 	struct bridge_iflist *bif;
708 	struct thread *td = curthread;
709 	union {
710 		struct ifbreq ifbreq;
711 		struct ifbifconf ifbifconf;
712 		struct ifbareq ifbareq;
713 		struct ifbaconf ifbaconf;
714 		struct ifbrparam ifbrparam;
715 		struct ifbropreq ifbropreq;
716 	} args;
717 	struct ifdrv *ifd = (struct ifdrv *) data;
718 	const struct bridge_control *bc;
719 	int error = 0;
720 
721 	switch (cmd) {
722 
723 	case SIOCADDMULTI:
724 	case SIOCDELMULTI:
725 		break;
726 
727 	case SIOCGDRVSPEC:
728 	case SIOCSDRVSPEC:
729 		if (ifd->ifd_cmd >= bridge_control_table_size) {
730 			error = EINVAL;
731 			break;
732 		}
733 		bc = &bridge_control_table[ifd->ifd_cmd];
734 
735 		if (cmd == SIOCGDRVSPEC &&
736 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
737 			error = EINVAL;
738 			break;
739 		}
740 		else if (cmd == SIOCSDRVSPEC &&
741 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
742 			error = EINVAL;
743 			break;
744 		}
745 
746 		if (bc->bc_flags & BC_F_SUSER) {
747 			error = priv_check(td, PRIV_NET_BRIDGE);
748 			if (error)
749 				break;
750 		}
751 
752 		if (ifd->ifd_len != bc->bc_argsize ||
753 		    ifd->ifd_len > sizeof(args)) {
754 			error = EINVAL;
755 			break;
756 		}
757 
758 		bzero(&args, sizeof(args));
759 		if (bc->bc_flags & BC_F_COPYIN) {
760 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
761 			if (error)
762 				break;
763 		}
764 
765 		BRIDGE_LOCK(sc);
766 		error = (*bc->bc_func)(sc, &args);
767 		BRIDGE_UNLOCK(sc);
768 		if (error)
769 			break;
770 
771 		if (bc->bc_flags & BC_F_COPYOUT)
772 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
773 
774 		break;
775 
776 	case SIOCSIFFLAGS:
777 		if (!(ifp->if_flags & IFF_UP) &&
778 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
779 			/*
780 			 * If interface is marked down and it is running,
781 			 * then stop and disable it.
782 			 */
783 			BRIDGE_LOCK(sc);
784 			bridge_stop(ifp, 1);
785 			BRIDGE_UNLOCK(sc);
786 		} else if ((ifp->if_flags & IFF_UP) &&
787 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
788 			/*
789 			 * If interface is marked up and it is stopped, then
790 			 * start it.
791 			 */
792 			(*ifp->if_init)(sc);
793 		}
794 		break;
795 
796 	case SIOCSIFMTU:
797 		if (ifr->ifr_mtu < 576) {
798 			error = EINVAL;
799 			break;
800 		}
801 		if (LIST_EMPTY(&sc->sc_iflist)) {
802 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
803 			break;
804 		}
805 		BRIDGE_LOCK(sc);
806 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
807 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
808 				log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)"
809 				    " != %d\n", sc->sc_ifp->if_xname,
810 				    bif->bif_ifp->if_mtu,
811 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
812 				error = EINVAL;
813 				break;
814 			}
815 		}
816 		if (!error)
817 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
818 		BRIDGE_UNLOCK(sc);
819 		break;
820 	default:
821 		/*
822 		 * drop the lock as ether_ioctl() will call bridge_start() and
823 		 * cause the lock to be recursed.
824 		 */
825 		error = ether_ioctl(ifp, cmd, data);
826 		break;
827 	}
828 
829 	return (error);
830 }
831 
832 /*
833  * bridge_mutecaps:
834  *
835  *	Clear or restore unwanted capabilities on the member interface
836  */
837 static void
838 bridge_mutecaps(struct bridge_softc *sc)
839 {
840 	struct bridge_iflist *bif;
841 	int enabled, mask;
842 
843 	/* Initial bitmask of capabilities to test */
844 	mask = BRIDGE_IFCAPS_MASK;
845 
846 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
847 		/* Every member must support it or its disabled */
848 		mask &= bif->bif_savedcaps;
849 	}
850 
851 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
852 		enabled = bif->bif_ifp->if_capenable;
853 		enabled &= ~BRIDGE_IFCAPS_STRIP;
854 		/* strip off mask bits and enable them again if allowed */
855 		enabled &= ~BRIDGE_IFCAPS_MASK;
856 		enabled |= mask;
857 		bridge_set_ifcap(sc, bif, enabled);
858 	}
859 
860 }
861 
862 static void
863 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
864 {
865 	struct ifnet *ifp = bif->bif_ifp;
866 	struct ifreq ifr;
867 	int error;
868 
869 	bzero(&ifr, sizeof(ifr));
870 	ifr.ifr_reqcap = set;
871 
872 	if (ifp->if_capenable != set) {
873 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
874 		if (error)
875 			if_printf(sc->sc_ifp,
876 			    "error setting interface capabilities on %s\n",
877 			    ifp->if_xname);
878 	}
879 }
880 
881 /*
882  * bridge_lookup_member:
883  *
884  *	Lookup a bridge member interface.
885  */
886 static struct bridge_iflist *
887 bridge_lookup_member(struct bridge_softc *sc, const char *name)
888 {
889 	struct bridge_iflist *bif;
890 	struct ifnet *ifp;
891 
892 	BRIDGE_LOCK_ASSERT(sc);
893 
894 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
895 		ifp = bif->bif_ifp;
896 		if (strcmp(ifp->if_xname, name) == 0)
897 			return (bif);
898 	}
899 
900 	return (NULL);
901 }
902 
903 /*
904  * bridge_lookup_member_if:
905  *
906  *	Lookup a bridge member interface by ifnet*.
907  */
908 static struct bridge_iflist *
909 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
910 {
911 	struct bridge_iflist *bif;
912 
913 	BRIDGE_LOCK_ASSERT(sc);
914 
915 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
916 		if (bif->bif_ifp == member_ifp)
917 			return (bif);
918 	}
919 
920 	return (NULL);
921 }
922 
923 /*
924  * bridge_delete_member:
925  *
926  *	Delete the specified member interface.
927  */
928 static void
929 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
930     int gone)
931 {
932 	struct ifnet *ifs = bif->bif_ifp;
933 	struct ifnet *fif = NULL;
934 
935 	BRIDGE_LOCK_ASSERT(sc);
936 
937 	if (bif->bif_flags & IFBIF_STP)
938 		bstp_disable(&bif->bif_stp);
939 
940 	ifs->if_bridge = NULL;
941 	BRIDGE_XLOCK(sc);
942 	LIST_REMOVE(bif, bif_next);
943 	BRIDGE_XDROP(sc);
944 
945 	/*
946 	 * If removing the interface that gave the bridge its mac address, set
947 	 * the mac address of the bridge to the address of the next member, or
948 	 * to its default address if no members are left.
949 	 */
950 	if (bridge_inherit_mac && sc->sc_ifaddr == ifs) {
951 		if (LIST_EMPTY(&sc->sc_iflist)) {
952 			bcopy(sc->sc_defaddr,
953 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
954 			sc->sc_ifaddr = NULL;
955 		} else {
956 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
957 			bcopy(IF_LLADDR(fif),
958 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
959 			sc->sc_ifaddr = fif;
960 		}
961 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
962 	}
963 
964 	bridge_linkcheck(sc);
965 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
966 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
967 	KASSERT(bif->bif_addrcnt == 0,
968 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
969 
970 	BRIDGE_UNLOCK(sc);
971 	if (!gone) {
972 		switch (ifs->if_type) {
973 		case IFT_ETHER:
974 		case IFT_L2VLAN:
975 			/*
976 			 * Take the interface out of promiscuous mode.
977 			 */
978 			(void) ifpromisc(ifs, 0);
979 			break;
980 
981 		case IFT_GIF:
982 			break;
983 
984 		default:
985 #ifdef DIAGNOSTIC
986 			panic("bridge_delete_member: impossible");
987 #endif
988 			break;
989 		}
990 		/* reneable any interface capabilities */
991 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
992 	}
993 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
994 	BRIDGE_LOCK(sc);
995 	free(bif, M_DEVBUF);
996 }
997 
998 /*
999  * bridge_delete_span:
1000  *
1001  *	Delete the specified span interface.
1002  */
1003 static void
1004 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1005 {
1006 	BRIDGE_LOCK_ASSERT(sc);
1007 
1008 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1009 	    ("%s: not a span interface", __func__));
1010 
1011 	LIST_REMOVE(bif, bif_next);
1012 	free(bif, M_DEVBUF);
1013 }
1014 
1015 static int
1016 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1017 {
1018 	struct ifbreq *req = arg;
1019 	struct bridge_iflist *bif = NULL;
1020 	struct ifnet *ifs;
1021 	int error = 0;
1022 
1023 	ifs = ifunit(req->ifbr_ifsname);
1024 	if (ifs == NULL)
1025 		return (ENOENT);
1026 	if (ifs->if_ioctl == NULL)	/* must be supported */
1027 		return (EINVAL);
1028 
1029 	/* If it's in the span list, it can't be a member. */
1030 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1031 		if (ifs == bif->bif_ifp)
1032 			return (EBUSY);
1033 
1034 	if (ifs->if_bridge == sc)
1035 		return (EEXIST);
1036 
1037 	if (ifs->if_bridge != NULL)
1038 		return (EBUSY);
1039 
1040 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1041 	if (bif == NULL)
1042 		return (ENOMEM);
1043 
1044 	bif->bif_ifp = ifs;
1045 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1046 	bif->bif_savedcaps = ifs->if_capenable;
1047 
1048 	switch (ifs->if_type) {
1049 	case IFT_ETHER:
1050 	case IFT_L2VLAN:
1051 	case IFT_GIF:
1052 		/* permitted interface types */
1053 		break;
1054 	default:
1055 		error = EINVAL;
1056 		goto out;
1057 	}
1058 
1059 	/* Allow the first Ethernet member to define the MTU */
1060 	if (LIST_EMPTY(&sc->sc_iflist))
1061 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1062 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1063 		if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n",
1064 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1065 		error = EINVAL;
1066 		goto out;
1067 	}
1068 
1069 	/*
1070 	 * Assign the interface's MAC address to the bridge if it's the first
1071 	 * member and the MAC address of the bridge has not been changed from
1072 	 * the default randomly generated one.
1073 	 */
1074 	if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1075 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1076 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1077 		sc->sc_ifaddr = ifs;
1078 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1079 	}
1080 
1081 	ifs->if_bridge = sc;
1082 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1083 	/*
1084 	 * XXX: XLOCK HERE!?!
1085 	 *
1086 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1087 	 */
1088 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1089 
1090 	/* Set interface capabilities to the intersection set of all members */
1091 	bridge_mutecaps(sc);
1092 	bridge_linkcheck(sc);
1093 
1094 	/* Place the interface into promiscuous mode */
1095 	switch (ifs->if_type) {
1096 		case IFT_ETHER:
1097 		case IFT_L2VLAN:
1098 			BRIDGE_UNLOCK(sc);
1099 			error = ifpromisc(ifs, 1);
1100 			BRIDGE_LOCK(sc);
1101 			break;
1102 	}
1103 	if (error)
1104 		bridge_delete_member(sc, bif, 0);
1105 out:
1106 	if (error) {
1107 		if (bif != NULL)
1108 			free(bif, M_DEVBUF);
1109 	}
1110 	return (error);
1111 }
1112 
1113 static int
1114 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1115 {
1116 	struct ifbreq *req = arg;
1117 	struct bridge_iflist *bif;
1118 
1119 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1120 	if (bif == NULL)
1121 		return (ENOENT);
1122 
1123 	bridge_delete_member(sc, bif, 0);
1124 
1125 	return (0);
1126 }
1127 
1128 static int
1129 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1130 {
1131 	struct ifbreq *req = arg;
1132 	struct bridge_iflist *bif;
1133 	struct bstp_port *bp;
1134 
1135 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1136 	if (bif == NULL)
1137 		return (ENOENT);
1138 
1139 	bp = &bif->bif_stp;
1140 	req->ifbr_ifsflags = bif->bif_flags;
1141 	req->ifbr_state = bp->bp_state;
1142 	req->ifbr_priority = bp->bp_priority;
1143 	req->ifbr_path_cost = bp->bp_path_cost;
1144 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1145 	req->ifbr_proto = bp->bp_protover;
1146 	req->ifbr_role = bp->bp_role;
1147 	req->ifbr_stpflags = bp->bp_flags;
1148 	req->ifbr_addrcnt = bif->bif_addrcnt;
1149 	req->ifbr_addrmax = bif->bif_addrmax;
1150 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1151 
1152 	/* Copy STP state options as flags */
1153 	if (bp->bp_operedge)
1154 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1155 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1156 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1157 	if (bp->bp_ptp_link)
1158 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1159 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1160 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1161 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1162 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1163 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1164 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1165 	return (0);
1166 }
1167 
1168 static int
1169 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1170 {
1171 	struct ifbreq *req = arg;
1172 	struct bridge_iflist *bif;
1173 	struct bstp_port *bp;
1174 	int error;
1175 
1176 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1177 	if (bif == NULL)
1178 		return (ENOENT);
1179 	bp = &bif->bif_stp;
1180 
1181 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1182 		/* SPAN is readonly */
1183 		return (EINVAL);
1184 
1185 	if (req->ifbr_ifsflags & IFBIF_STP) {
1186 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1187 			error = bstp_enable(&bif->bif_stp);
1188 			if (error)
1189 				return (error);
1190 		}
1191 	} else {
1192 		if ((bif->bif_flags & IFBIF_STP) != 0)
1193 			bstp_disable(&bif->bif_stp);
1194 	}
1195 
1196 	/* Pass on STP flags */
1197 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1198 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1199 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1200 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1201 
1202 	/* Save the bits relating to the bridge */
1203 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1204 
1205 	return (0);
1206 }
1207 
1208 static int
1209 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1210 {
1211 	struct ifbrparam *param = arg;
1212 
1213 	sc->sc_brtmax = param->ifbrp_csize;
1214 	bridge_rttrim(sc);
1215 
1216 	return (0);
1217 }
1218 
1219 static int
1220 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1221 {
1222 	struct ifbrparam *param = arg;
1223 
1224 	param->ifbrp_csize = sc->sc_brtmax;
1225 
1226 	return (0);
1227 }
1228 
1229 static int
1230 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1231 {
1232 	struct ifbifconf *bifc = arg;
1233 	struct bridge_iflist *bif;
1234 	struct ifbreq breq;
1235 	char *buf, *outbuf;
1236 	int count, buflen, len, error = 0;
1237 
1238 	count = 0;
1239 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1240 		count++;
1241 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1242 		count++;
1243 
1244 	buflen = sizeof(breq) * count;
1245 	if (bifc->ifbic_len == 0) {
1246 		bifc->ifbic_len = buflen;
1247 		return (0);
1248 	}
1249 	BRIDGE_UNLOCK(sc);
1250 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1251 	BRIDGE_LOCK(sc);
1252 
1253 	count = 0;
1254 	buf = outbuf;
1255 	len = min(bifc->ifbic_len, buflen);
1256 	bzero(&breq, sizeof(breq));
1257 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1258 		if (len < sizeof(breq))
1259 			break;
1260 
1261 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1262 		    sizeof(breq.ifbr_ifsname));
1263 		/* Fill in the ifbreq structure */
1264 		error = bridge_ioctl_gifflags(sc, &breq);
1265 		if (error)
1266 			break;
1267 		memcpy(buf, &breq, sizeof(breq));
1268 		count++;
1269 		buf += sizeof(breq);
1270 		len -= sizeof(breq);
1271 	}
1272 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1273 		if (len < sizeof(breq))
1274 			break;
1275 
1276 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1277 		    sizeof(breq.ifbr_ifsname));
1278 		breq.ifbr_ifsflags = bif->bif_flags;
1279 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1280 		memcpy(buf, &breq, sizeof(breq));
1281 		count++;
1282 		buf += sizeof(breq);
1283 		len -= sizeof(breq);
1284 	}
1285 
1286 	BRIDGE_UNLOCK(sc);
1287 	bifc->ifbic_len = sizeof(breq) * count;
1288 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1289 	BRIDGE_LOCK(sc);
1290 	free(outbuf, M_TEMP);
1291 	return (error);
1292 }
1293 
1294 static int
1295 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1296 {
1297 	struct ifbaconf *bac = arg;
1298 	struct bridge_rtnode *brt;
1299 	struct ifbareq bareq;
1300 	char *buf, *outbuf;
1301 	int count, buflen, len, error = 0;
1302 
1303 	if (bac->ifbac_len == 0)
1304 		return (0);
1305 
1306 	count = 0;
1307 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1308 		count++;
1309 	buflen = sizeof(bareq) * count;
1310 
1311 	BRIDGE_UNLOCK(sc);
1312 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1313 	BRIDGE_LOCK(sc);
1314 
1315 	count = 0;
1316 	buf = outbuf;
1317 	len = min(bac->ifbac_len, buflen);
1318 	bzero(&bareq, sizeof(bareq));
1319 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1320 		if (len < sizeof(bareq))
1321 			goto out;
1322 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1323 		    sizeof(bareq.ifba_ifsname));
1324 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1325 		bareq.ifba_vlan = brt->brt_vlan;
1326 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1327 				time_uptime < brt->brt_expire)
1328 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1329 		else
1330 			bareq.ifba_expire = 0;
1331 		bareq.ifba_flags = brt->brt_flags;
1332 
1333 		memcpy(buf, &bareq, sizeof(bareq));
1334 		count++;
1335 		buf += sizeof(bareq);
1336 		len -= sizeof(bareq);
1337 	}
1338 out:
1339 	BRIDGE_UNLOCK(sc);
1340 	bac->ifbac_len = sizeof(bareq) * count;
1341 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1342 	BRIDGE_LOCK(sc);
1343 	free(outbuf, M_TEMP);
1344 	return (error);
1345 }
1346 
1347 static int
1348 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1349 {
1350 	struct ifbareq *req = arg;
1351 	struct bridge_iflist *bif;
1352 	int error;
1353 
1354 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1355 	if (bif == NULL)
1356 		return (ENOENT);
1357 
1358 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1359 	    req->ifba_flags);
1360 
1361 	return (error);
1362 }
1363 
1364 static int
1365 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1366 {
1367 	struct ifbrparam *param = arg;
1368 
1369 	sc->sc_brttimeout = param->ifbrp_ctime;
1370 	return (0);
1371 }
1372 
1373 static int
1374 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1375 {
1376 	struct ifbrparam *param = arg;
1377 
1378 	param->ifbrp_ctime = sc->sc_brttimeout;
1379 	return (0);
1380 }
1381 
1382 static int
1383 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1384 {
1385 	struct ifbareq *req = arg;
1386 
1387 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1388 }
1389 
1390 static int
1391 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1392 {
1393 	struct ifbreq *req = arg;
1394 
1395 	bridge_rtflush(sc, req->ifbr_ifsflags);
1396 	return (0);
1397 }
1398 
1399 static int
1400 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1401 {
1402 	struct ifbrparam *param = arg;
1403 	struct bstp_state *bs = &sc->sc_stp;
1404 
1405 	param->ifbrp_prio = bs->bs_bridge_priority;
1406 	return (0);
1407 }
1408 
1409 static int
1410 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1411 {
1412 	struct ifbrparam *param = arg;
1413 
1414 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1415 }
1416 
1417 static int
1418 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1419 {
1420 	struct ifbrparam *param = arg;
1421 	struct bstp_state *bs = &sc->sc_stp;
1422 
1423 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1424 	return (0);
1425 }
1426 
1427 static int
1428 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1429 {
1430 	struct ifbrparam *param = arg;
1431 
1432 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1433 }
1434 
1435 static int
1436 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1437 {
1438 	struct ifbrparam *param = arg;
1439 	struct bstp_state *bs = &sc->sc_stp;
1440 
1441 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1442 	return (0);
1443 }
1444 
1445 static int
1446 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1447 {
1448 	struct ifbrparam *param = arg;
1449 
1450 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1451 }
1452 
1453 static int
1454 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1455 {
1456 	struct ifbrparam *param = arg;
1457 	struct bstp_state *bs = &sc->sc_stp;
1458 
1459 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1460 	return (0);
1461 }
1462 
1463 static int
1464 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1465 {
1466 	struct ifbrparam *param = arg;
1467 
1468 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1469 }
1470 
1471 static int
1472 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1473 {
1474 	struct ifbreq *req = arg;
1475 	struct bridge_iflist *bif;
1476 
1477 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1478 	if (bif == NULL)
1479 		return (ENOENT);
1480 
1481 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1482 }
1483 
1484 static int
1485 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1486 {
1487 	struct ifbreq *req = arg;
1488 	struct bridge_iflist *bif;
1489 
1490 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1491 	if (bif == NULL)
1492 		return (ENOENT);
1493 
1494 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1495 }
1496 
1497 static int
1498 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1499 {
1500 	struct ifbreq *req = arg;
1501 	struct bridge_iflist *bif;
1502 
1503 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1504 	if (bif == NULL)
1505 		return (ENOENT);
1506 
1507 	bif->bif_addrmax = req->ifbr_addrmax;
1508 	return (0);
1509 }
1510 
1511 static int
1512 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1513 {
1514 	struct ifbreq *req = arg;
1515 	struct bridge_iflist *bif = NULL;
1516 	struct ifnet *ifs;
1517 
1518 	ifs = ifunit(req->ifbr_ifsname);
1519 	if (ifs == NULL)
1520 		return (ENOENT);
1521 
1522 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1523 		if (ifs == bif->bif_ifp)
1524 			return (EBUSY);
1525 
1526 	if (ifs->if_bridge != NULL)
1527 		return (EBUSY);
1528 
1529 	switch (ifs->if_type) {
1530 		case IFT_ETHER:
1531 		case IFT_GIF:
1532 		case IFT_L2VLAN:
1533 			break;
1534 		default:
1535 			return (EINVAL);
1536 	}
1537 
1538 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1539 	if (bif == NULL)
1540 		return (ENOMEM);
1541 
1542 	bif->bif_ifp = ifs;
1543 	bif->bif_flags = IFBIF_SPAN;
1544 
1545 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1546 
1547 	return (0);
1548 }
1549 
1550 static int
1551 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1552 {
1553 	struct ifbreq *req = arg;
1554 	struct bridge_iflist *bif;
1555 	struct ifnet *ifs;
1556 
1557 	ifs = ifunit(req->ifbr_ifsname);
1558 	if (ifs == NULL)
1559 		return (ENOENT);
1560 
1561 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1562 		if (ifs == bif->bif_ifp)
1563 			break;
1564 
1565 	if (bif == NULL)
1566 		return (ENOENT);
1567 
1568 	bridge_delete_span(sc, bif);
1569 
1570 	return (0);
1571 }
1572 
1573 static int
1574 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1575 {
1576 	struct ifbropreq *req = arg;
1577 	struct bstp_state *bs = &sc->sc_stp;
1578 	struct bstp_port *root_port;
1579 
1580 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1581 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1582 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1583 
1584 	root_port = bs->bs_root_port;
1585 	if (root_port == NULL)
1586 		req->ifbop_root_port = 0;
1587 	else
1588 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1589 
1590 	req->ifbop_holdcount = bs->bs_txholdcount;
1591 	req->ifbop_priority = bs->bs_bridge_priority;
1592 	req->ifbop_protocol = bs->bs_protover;
1593 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1594 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1595 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1596 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1597 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1598 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1599 
1600 	return (0);
1601 }
1602 
1603 static int
1604 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1605 {
1606 	struct ifbrparam *param = arg;
1607 
1608 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1609 	return (0);
1610 }
1611 
1612 static int
1613 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1614 {
1615 	struct ifbpstpconf *bifstp = arg;
1616 	struct bridge_iflist *bif;
1617 	struct bstp_port *bp;
1618 	struct ifbpstpreq bpreq;
1619 	char *buf, *outbuf;
1620 	int count, buflen, len, error = 0;
1621 
1622 	count = 0;
1623 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1624 		if ((bif->bif_flags & IFBIF_STP) != 0)
1625 			count++;
1626 	}
1627 
1628 	buflen = sizeof(bpreq) * count;
1629 	if (bifstp->ifbpstp_len == 0) {
1630 		bifstp->ifbpstp_len = buflen;
1631 		return (0);
1632 	}
1633 
1634 	BRIDGE_UNLOCK(sc);
1635 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1636 	BRIDGE_LOCK(sc);
1637 
1638 	count = 0;
1639 	buf = outbuf;
1640 	len = min(bifstp->ifbpstp_len, buflen);
1641 	bzero(&bpreq, sizeof(bpreq));
1642 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1643 		if (len < sizeof(bpreq))
1644 			break;
1645 
1646 		if ((bif->bif_flags & IFBIF_STP) == 0)
1647 			continue;
1648 
1649 		bp = &bif->bif_stp;
1650 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1651 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1652 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1653 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1654 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1655 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1656 
1657 		memcpy(buf, &bpreq, sizeof(bpreq));
1658 		count++;
1659 		buf += sizeof(bpreq);
1660 		len -= sizeof(bpreq);
1661 	}
1662 
1663 	BRIDGE_UNLOCK(sc);
1664 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1665 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1666 	BRIDGE_LOCK(sc);
1667 	free(outbuf, M_TEMP);
1668 	return (error);
1669 }
1670 
1671 static int
1672 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1673 {
1674 	struct ifbrparam *param = arg;
1675 
1676 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1677 }
1678 
1679 static int
1680 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1681 {
1682 	struct ifbrparam *param = arg;
1683 
1684 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1685 }
1686 
1687 /*
1688  * bridge_ifdetach:
1689  *
1690  *	Detach an interface from a bridge.  Called when a member
1691  *	interface is detaching.
1692  */
1693 static void
1694 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1695 {
1696 	struct bridge_softc *sc = ifp->if_bridge;
1697 	struct bridge_iflist *bif;
1698 
1699 	/* Check if the interface is a bridge member */
1700 	if (sc != NULL) {
1701 		BRIDGE_LOCK(sc);
1702 
1703 		bif = bridge_lookup_member_if(sc, ifp);
1704 		if (bif != NULL)
1705 			bridge_delete_member(sc, bif, 1);
1706 
1707 		BRIDGE_UNLOCK(sc);
1708 		return;
1709 	}
1710 
1711 	/* Check if the interface is a span port */
1712 	mtx_lock(&bridge_list_mtx);
1713 	LIST_FOREACH(sc, &bridge_list, sc_list) {
1714 		BRIDGE_LOCK(sc);
1715 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1716 			if (ifp == bif->bif_ifp) {
1717 				bridge_delete_span(sc, bif);
1718 				break;
1719 			}
1720 
1721 		BRIDGE_UNLOCK(sc);
1722 	}
1723 	mtx_unlock(&bridge_list_mtx);
1724 }
1725 
1726 /*
1727  * bridge_init:
1728  *
1729  *	Initialize a bridge interface.
1730  */
1731 static void
1732 bridge_init(void *xsc)
1733 {
1734 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1735 	struct ifnet *ifp = sc->sc_ifp;
1736 
1737 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1738 		return;
1739 
1740 	BRIDGE_LOCK(sc);
1741 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1742 	    bridge_timer, sc);
1743 
1744 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1745 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1746 
1747 	BRIDGE_UNLOCK(sc);
1748 }
1749 
1750 /*
1751  * bridge_stop:
1752  *
1753  *	Stop the bridge interface.
1754  */
1755 static void
1756 bridge_stop(struct ifnet *ifp, int disable)
1757 {
1758 	struct bridge_softc *sc = ifp->if_softc;
1759 
1760 	BRIDGE_LOCK_ASSERT(sc);
1761 
1762 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1763 		return;
1764 
1765 	callout_stop(&sc->sc_brcallout);
1766 	bstp_stop(&sc->sc_stp);
1767 
1768 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1769 
1770 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1771 }
1772 
1773 /*
1774  * bridge_enqueue:
1775  *
1776  *	Enqueue a packet on a bridge member interface.
1777  *
1778  */
1779 static int
1780 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1781 {
1782 	int len, err = 0;
1783 	short mflags;
1784 	struct mbuf *m0;
1785 
1786 	len = m->m_pkthdr.len;
1787 	mflags = m->m_flags;
1788 
1789 	/* We may be sending a fragment so traverse the mbuf */
1790 	for (; m; m = m0) {
1791 		m0 = m->m_nextpkt;
1792 		m->m_nextpkt = NULL;
1793 
1794 		/*
1795 		 * If underlying interface can not do VLAN tag insertion itself
1796 		 * then attach a packet tag that holds it.
1797 		 */
1798 		if ((m->m_flags & M_VLANTAG) &&
1799 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1800 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1801 			if (m == NULL) {
1802 				if_printf(dst_ifp,
1803 				    "unable to prepend VLAN header\n");
1804 				dst_ifp->if_oerrors++;
1805 				continue;
1806 			}
1807 			m->m_flags &= ~M_VLANTAG;
1808 		}
1809 
1810 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1811 			m_freem(m0);
1812 			break;
1813 		}
1814 	}
1815 
1816 	if (err == 0) {
1817 		sc->sc_ifp->if_opackets++;
1818 		sc->sc_ifp->if_obytes += len;
1819 		if (mflags & M_MCAST)
1820 			sc->sc_ifp->if_omcasts++;
1821 	}
1822 
1823 	return (err);
1824 }
1825 
1826 /*
1827  * bridge_dummynet:
1828  *
1829  * 	Receive a queued packet from dummynet and pass it on to the output
1830  * 	interface.
1831  *
1832  *	The mbuf has the Ethernet header already attached.
1833  */
1834 static void
1835 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1836 {
1837 	struct bridge_softc *sc;
1838 
1839 	sc = ifp->if_bridge;
1840 
1841 	/*
1842 	 * The packet didnt originate from a member interface. This should only
1843 	 * ever happen if a member interface is removed while packets are
1844 	 * queued for it.
1845 	 */
1846 	if (sc == NULL) {
1847 		m_freem(m);
1848 		return;
1849 	}
1850 
1851 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1852 #ifdef INET6
1853 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1854 #endif
1855 	    ) {
1856 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1857 			return;
1858 		if (m == NULL)
1859 			return;
1860 	}
1861 
1862 	bridge_enqueue(sc, ifp, m);
1863 }
1864 
1865 /*
1866  * bridge_output:
1867  *
1868  *	Send output from a bridge member interface.  This
1869  *	performs the bridging function for locally originated
1870  *	packets.
1871  *
1872  *	The mbuf has the Ethernet header already attached.  We must
1873  *	enqueue or free the mbuf before returning.
1874  */
1875 static int
1876 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1877     struct rtentry *rt)
1878 {
1879 	struct ether_header *eh;
1880 	struct ifnet *dst_if;
1881 	struct bridge_softc *sc;
1882 	uint16_t vlan;
1883 
1884 	if (m->m_len < ETHER_HDR_LEN) {
1885 		m = m_pullup(m, ETHER_HDR_LEN);
1886 		if (m == NULL)
1887 			return (0);
1888 	}
1889 
1890 	eh = mtod(m, struct ether_header *);
1891 	sc = ifp->if_bridge;
1892 	vlan = VLANTAGOF(m);
1893 
1894 	BRIDGE_LOCK(sc);
1895 
1896 	/*
1897 	 * If bridge is down, but the original output interface is up,
1898 	 * go ahead and send out that interface.  Otherwise, the packet
1899 	 * is dropped below.
1900 	 */
1901 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1902 		dst_if = ifp;
1903 		goto sendunicast;
1904 	}
1905 
1906 	/*
1907 	 * If the packet is a multicast, or we don't know a better way to
1908 	 * get there, send to all interfaces.
1909 	 */
1910 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
1911 		dst_if = NULL;
1912 	else
1913 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
1914 	if (dst_if == NULL) {
1915 		struct bridge_iflist *bif;
1916 		struct mbuf *mc;
1917 		int error = 0, used = 0;
1918 
1919 		bridge_span(sc, m);
1920 
1921 		BRIDGE_LOCK2REF(sc, error);
1922 		if (error) {
1923 			m_freem(m);
1924 			return (0);
1925 		}
1926 
1927 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1928 			dst_if = bif->bif_ifp;
1929 
1930 			if (dst_if->if_type == IFT_GIF)
1931 				continue;
1932 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
1933 				continue;
1934 
1935 			/*
1936 			 * If this is not the original output interface,
1937 			 * and the interface is participating in spanning
1938 			 * tree, make sure the port is in a state that
1939 			 * allows forwarding.
1940 			 */
1941 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
1942 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
1943 				continue;
1944 
1945 			if (LIST_NEXT(bif, bif_next) == NULL) {
1946 				used = 1;
1947 				mc = m;
1948 			} else {
1949 				mc = m_copypacket(m, M_DONTWAIT);
1950 				if (mc == NULL) {
1951 					sc->sc_ifp->if_oerrors++;
1952 					continue;
1953 				}
1954 			}
1955 
1956 			bridge_enqueue(sc, dst_if, mc);
1957 		}
1958 		if (used == 0)
1959 			m_freem(m);
1960 		BRIDGE_UNREF(sc);
1961 		return (0);
1962 	}
1963 
1964 sendunicast:
1965 	/*
1966 	 * XXX Spanning tree consideration here?
1967 	 */
1968 
1969 	bridge_span(sc, m);
1970 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
1971 		m_freem(m);
1972 		BRIDGE_UNLOCK(sc);
1973 		return (0);
1974 	}
1975 
1976 	BRIDGE_UNLOCK(sc);
1977 	bridge_enqueue(sc, dst_if, m);
1978 	return (0);
1979 }
1980 
1981 /*
1982  * bridge_transmit:
1983  *
1984  *	Do output on a bridge.
1985  *
1986  */
1987 static int
1988 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
1989 {
1990 	struct bridge_softc *sc;
1991 	int error = 0;
1992 
1993 	sc = ifp->if_softc;
1994 
1995 	ETHER_BPF_MTAP(ifp, m);
1996 
1997 
1998 	BRIDGE_LOCK(sc);
1999 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2000 		struct ether_header *eh;
2001 		struct ifnet *dst_if;
2002 
2003 		eh = mtod(m, struct ether_header *);
2004 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1);
2005 		BRIDGE_UNLOCK(sc);
2006 		error = bridge_enqueue(sc, dst_if, m);
2007 	} else
2008 		bridge_broadcast(sc, ifp, m, 0);
2009 
2010 	return (error);
2011 }
2012 
2013 /*
2014  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2015  */
2016 static void
2017 bridge_qflush(struct ifnet *ifp __unused)
2018 {
2019 }
2020 
2021 /*
2022  * bridge_forward:
2023  *
2024  *	The forwarding function of the bridge.
2025  *
2026  *	NOTE: Releases the lock on return.
2027  */
2028 static void
2029 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2030     struct mbuf *m)
2031 {
2032 	struct bridge_iflist *dbif;
2033 	struct ifnet *src_if, *dst_if, *ifp;
2034 	struct ether_header *eh;
2035 	uint16_t vlan;
2036 	uint8_t *dst;
2037 	int error;
2038 
2039 	src_if = m->m_pkthdr.rcvif;
2040 	ifp = sc->sc_ifp;
2041 
2042 	ifp->if_ipackets++;
2043 	ifp->if_ibytes += m->m_pkthdr.len;
2044 	vlan = VLANTAGOF(m);
2045 
2046 	if ((sbif->bif_flags & IFBIF_STP) &&
2047 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2048 		goto drop;
2049 
2050 	eh = mtod(m, struct ether_header *);
2051 	dst = eh->ether_dhost;
2052 
2053 	/* If the interface is learning, record the address. */
2054 	if (sbif->bif_flags & IFBIF_LEARNING) {
2055 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2056 		    sbif, 0, IFBAF_DYNAMIC);
2057 		/*
2058 		 * If the interface has addresses limits then deny any source
2059 		 * that is not in the cache.
2060 		 */
2061 		if (error && sbif->bif_addrmax)
2062 			goto drop;
2063 	}
2064 
2065 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2066 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2067 		goto drop;
2068 
2069 	/*
2070 	 * At this point, the port either doesn't participate
2071 	 * in spanning tree or it is in the forwarding state.
2072 	 */
2073 
2074 	/*
2075 	 * If the packet is unicast, destined for someone on
2076 	 * "this" side of the bridge, drop it.
2077 	 */
2078 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2079 		dst_if = bridge_rtlookup(sc, dst, vlan);
2080 		if (src_if == dst_if)
2081 			goto drop;
2082 	} else {
2083 		/*
2084 		 * Check if its a reserved multicast address, any address
2085 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2086 		 * bridge.
2087 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2088 		 */
2089 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2090 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2091 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2092 			goto drop;
2093 
2094 		/* ...forward it to all interfaces. */
2095 		ifp->if_imcasts++;
2096 		dst_if = NULL;
2097 	}
2098 
2099 	/*
2100 	 * If we have a destination interface which is a member of our bridge,
2101 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2102 	 * For broadcast or multicast packets, don't bother because it will
2103 	 * be reinjected into ether_input. We do this before we pass the packets
2104 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2105 	 * drop the packet, or possibly modify it, making it difficult to debug
2106 	 * firewall issues on the bridge.
2107 	 */
2108 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2109 		ETHER_BPF_MTAP(ifp, m);
2110 
2111 	/* run the packet filter */
2112 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2113 #ifdef INET6
2114 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2115 #endif
2116 	    ) {
2117 		BRIDGE_UNLOCK(sc);
2118 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2119 			return;
2120 		if (m == NULL)
2121 			return;
2122 		BRIDGE_LOCK(sc);
2123 	}
2124 
2125 	if (dst_if == NULL) {
2126 		bridge_broadcast(sc, src_if, m, 1);
2127 		return;
2128 	}
2129 
2130 	/*
2131 	 * At this point, we're dealing with a unicast frame
2132 	 * going to a different interface.
2133 	 */
2134 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2135 		goto drop;
2136 
2137 	dbif = bridge_lookup_member_if(sc, dst_if);
2138 	if (dbif == NULL)
2139 		/* Not a member of the bridge (anymore?) */
2140 		goto drop;
2141 
2142 	/* Private segments can not talk to each other */
2143 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2144 		goto drop;
2145 
2146 	if ((dbif->bif_flags & IFBIF_STP) &&
2147 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2148 		goto drop;
2149 
2150 	BRIDGE_UNLOCK(sc);
2151 
2152 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2153 #ifdef INET6
2154 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2155 #endif
2156 	    ) {
2157 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2158 			return;
2159 		if (m == NULL)
2160 			return;
2161 	}
2162 
2163 	bridge_enqueue(sc, dst_if, m);
2164 	return;
2165 
2166 drop:
2167 	BRIDGE_UNLOCK(sc);
2168 	m_freem(m);
2169 }
2170 
2171 /*
2172  * bridge_input:
2173  *
2174  *	Receive input from a member interface.  Queue the packet for
2175  *	bridging if it is not for us.
2176  */
2177 static struct mbuf *
2178 bridge_input(struct ifnet *ifp, struct mbuf *m)
2179 {
2180 	struct bridge_softc *sc = ifp->if_bridge;
2181 	struct bridge_iflist *bif, *bif2;
2182 	struct ifnet *bifp;
2183 	struct ether_header *eh;
2184 	struct mbuf *mc, *mc2;
2185 	uint16_t vlan;
2186 	int error;
2187 
2188 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2189 		return (m);
2190 
2191 	bifp = sc->sc_ifp;
2192 	vlan = VLANTAGOF(m);
2193 
2194 	/*
2195 	 * Implement support for bridge monitoring. If this flag has been
2196 	 * set on this interface, discard the packet once we push it through
2197 	 * the bpf(4) machinery, but before we do, increment the byte and
2198 	 * packet counters associated with this interface.
2199 	 */
2200 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2201 		m->m_pkthdr.rcvif  = bifp;
2202 		ETHER_BPF_MTAP(bifp, m);
2203 		bifp->if_ipackets++;
2204 		bifp->if_ibytes += m->m_pkthdr.len;
2205 		m_freem(m);
2206 		return (NULL);
2207 	}
2208 	BRIDGE_LOCK(sc);
2209 	bif = bridge_lookup_member_if(sc, ifp);
2210 	if (bif == NULL) {
2211 		BRIDGE_UNLOCK(sc);
2212 		return (m);
2213 	}
2214 
2215 	eh = mtod(m, struct ether_header *);
2216 
2217 	bridge_span(sc, m);
2218 
2219 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2220 		/* Tap off 802.1D packets; they do not get forwarded. */
2221 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2222 		    ETHER_ADDR_LEN) == 0) {
2223 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2224 			BRIDGE_UNLOCK(sc);
2225 			return (NULL);
2226 		}
2227 
2228 		if ((bif->bif_flags & IFBIF_STP) &&
2229 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2230 			BRIDGE_UNLOCK(sc);
2231 			return (m);
2232 		}
2233 
2234 		/*
2235 		 * Make a deep copy of the packet and enqueue the copy
2236 		 * for bridge processing; return the original packet for
2237 		 * local processing.
2238 		 */
2239 		mc = m_dup(m, M_DONTWAIT);
2240 		if (mc == NULL) {
2241 			BRIDGE_UNLOCK(sc);
2242 			return (m);
2243 		}
2244 
2245 		/* Perform the bridge forwarding function with the copy. */
2246 		bridge_forward(sc, bif, mc);
2247 
2248 		/*
2249 		 * Reinject the mbuf as arriving on the bridge so we have a
2250 		 * chance at claiming multicast packets. We can not loop back
2251 		 * here from ether_input as a bridge is never a member of a
2252 		 * bridge.
2253 		 */
2254 		KASSERT(bifp->if_bridge == NULL,
2255 		    ("loop created in bridge_input"));
2256 		mc2 = m_dup(m, M_DONTWAIT);
2257 		if (mc2 != NULL) {
2258 			/* Keep the layer3 header aligned */
2259 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2260 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2261 		}
2262 		if (mc2 != NULL) {
2263 			mc2->m_pkthdr.rcvif = bifp;
2264 			(*bifp->if_input)(bifp, mc2);
2265 		}
2266 
2267 		/* Return the original packet for local processing. */
2268 		return (m);
2269 	}
2270 
2271 	if ((bif->bif_flags & IFBIF_STP) &&
2272 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2273 		BRIDGE_UNLOCK(sc);
2274 		return (m);
2275 	}
2276 
2277 #if (defined(INET) || defined(INET6))
2278 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2279 	|| ((iface)->if_carp \
2280 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2281 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2282 	|| ((iface)->if_carp \
2283 	    && (*carp_forus_p)((iface), eh->ether_shost))
2284 #else
2285 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2286 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2287 #endif
2288 
2289 #ifdef INET6
2290 #   define OR_PFIL_HOOKED_INET6 \
2291 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2292 #else
2293 #   define OR_PFIL_HOOKED_INET6
2294 #endif
2295 
2296 #define GRAB_OUR_PACKETS(iface) \
2297 	if ((iface)->if_type == IFT_GIF) \
2298 		continue; \
2299 	/* It is destined for us. */ \
2300 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2301 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2302 	    ) {								\
2303 		if ((iface)->if_type == IFT_BRIDGE) {			\
2304 			ETHER_BPF_MTAP(iface, m);			\
2305 			iface->if_ipackets++;				\
2306 			/* Filter on the physical interface. */		\
2307 			if (pfil_local_phys &&				\
2308 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2309 			     OR_PFIL_HOOKED_INET6)) {			\
2310 				if (bridge_pfil(&m, NULL, ifp,		\
2311 				    PFIL_IN) != 0 || m == NULL) {	\
2312 					BRIDGE_UNLOCK(sc);		\
2313 					return (NULL);			\
2314 				}					\
2315 			}						\
2316 		}							\
2317 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2318 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2319 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2320 			if (error && bif->bif_addrmax) {		\
2321 				BRIDGE_UNLOCK(sc);			\
2322 				m_freem(m);				\
2323 				return (NULL);				\
2324 			}						\
2325 		}							\
2326 		m->m_pkthdr.rcvif = iface;				\
2327 		BRIDGE_UNLOCK(sc);					\
2328 		return (m);						\
2329 	}								\
2330 									\
2331 	/* We just received a packet that we sent out. */		\
2332 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2333 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2334 	    ) {								\
2335 		BRIDGE_UNLOCK(sc);					\
2336 		m_freem(m);						\
2337 		return (NULL);						\
2338 	}
2339 
2340 	/*
2341 	 * Unicast.  Make sure it's not for the bridge.
2342 	 */
2343 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2344 
2345 	/*
2346 	 * Give a chance for ifp at first priority. This will help when	the
2347 	 * packet comes through the interface like VLAN's with the same MACs
2348 	 * on several interfaces from the same bridge. This also will save
2349 	 * some CPU cycles in case the destination interface and the input
2350 	 * interface (eq ifp) are the same.
2351 	 */
2352 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2353 
2354 	/* Now check the all bridge members. */
2355 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2356 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2357 	}
2358 
2359 #undef OR_CARP_CHECK_WE_ARE_DST
2360 #undef OR_CARP_CHECK_WE_ARE_SRC
2361 #undef OR_PFIL_HOOKED_INET6
2362 #undef GRAB_OUR_PACKETS
2363 
2364 	/* Perform the bridge forwarding function. */
2365 	bridge_forward(sc, bif, m);
2366 
2367 	return (NULL);
2368 }
2369 
2370 /*
2371  * bridge_broadcast:
2372  *
2373  *	Send a frame to all interfaces that are members of
2374  *	the bridge, except for the one on which the packet
2375  *	arrived.
2376  *
2377  *	NOTE: Releases the lock on return.
2378  */
2379 static void
2380 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2381     struct mbuf *m, int runfilt)
2382 {
2383 	struct bridge_iflist *dbif, *sbif;
2384 	struct mbuf *mc;
2385 	struct ifnet *dst_if;
2386 	int error = 0, used = 0, i;
2387 
2388 	sbif = bridge_lookup_member_if(sc, src_if);
2389 
2390 	BRIDGE_LOCK2REF(sc, error);
2391 	if (error) {
2392 		m_freem(m);
2393 		return;
2394 	}
2395 
2396 	/* Filter on the bridge interface before broadcasting */
2397 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2398 #ifdef INET6
2399 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2400 #endif
2401 	    )) {
2402 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2403 			goto out;
2404 		if (m == NULL)
2405 			goto out;
2406 	}
2407 
2408 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2409 		dst_if = dbif->bif_ifp;
2410 		if (dst_if == src_if)
2411 			continue;
2412 
2413 		/* Private segments can not talk to each other */
2414 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2415 			continue;
2416 
2417 		if ((dbif->bif_flags & IFBIF_STP) &&
2418 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2419 			continue;
2420 
2421 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2422 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2423 			continue;
2424 
2425 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2426 			continue;
2427 
2428 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2429 			mc = m;
2430 			used = 1;
2431 		} else {
2432 			mc = m_dup(m, M_DONTWAIT);
2433 			if (mc == NULL) {
2434 				sc->sc_ifp->if_oerrors++;
2435 				continue;
2436 			}
2437 		}
2438 
2439 		/*
2440 		 * Filter on the output interface. Pass a NULL bridge interface
2441 		 * pointer so we do not redundantly filter on the bridge for
2442 		 * each interface we broadcast on.
2443 		 */
2444 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2445 #ifdef INET6
2446 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2447 #endif
2448 		    )) {
2449 			if (used == 0) {
2450 				/* Keep the layer3 header aligned */
2451 				i = min(mc->m_pkthdr.len, max_protohdr);
2452 				mc = m_copyup(mc, i, ETHER_ALIGN);
2453 				if (mc == NULL) {
2454 					sc->sc_ifp->if_oerrors++;
2455 					continue;
2456 				}
2457 			}
2458 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2459 				continue;
2460 			if (mc == NULL)
2461 				continue;
2462 		}
2463 
2464 		bridge_enqueue(sc, dst_if, mc);
2465 	}
2466 	if (used == 0)
2467 		m_freem(m);
2468 
2469 out:
2470 	BRIDGE_UNREF(sc);
2471 }
2472 
2473 /*
2474  * bridge_span:
2475  *
2476  *	Duplicate a packet out one or more interfaces that are in span mode,
2477  *	the original mbuf is unmodified.
2478  */
2479 static void
2480 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2481 {
2482 	struct bridge_iflist *bif;
2483 	struct ifnet *dst_if;
2484 	struct mbuf *mc;
2485 
2486 	if (LIST_EMPTY(&sc->sc_spanlist))
2487 		return;
2488 
2489 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2490 		dst_if = bif->bif_ifp;
2491 
2492 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2493 			continue;
2494 
2495 		mc = m_copypacket(m, M_DONTWAIT);
2496 		if (mc == NULL) {
2497 			sc->sc_ifp->if_oerrors++;
2498 			continue;
2499 		}
2500 
2501 		bridge_enqueue(sc, dst_if, mc);
2502 	}
2503 }
2504 
2505 /*
2506  * bridge_rtupdate:
2507  *
2508  *	Add a bridge routing entry.
2509  */
2510 static int
2511 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2512     struct bridge_iflist *bif, int setflags, uint8_t flags)
2513 {
2514 	struct bridge_rtnode *brt;
2515 	int error;
2516 
2517 	BRIDGE_LOCK_ASSERT(sc);
2518 
2519 	/* Check the source address is valid and not multicast. */
2520 	if (ETHER_IS_MULTICAST(dst) ||
2521 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2522 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2523 		return (EINVAL);
2524 
2525 	/* 802.1p frames map to vlan 1 */
2526 	if (vlan == 0)
2527 		vlan = 1;
2528 
2529 	/*
2530 	 * A route for this destination might already exist.  If so,
2531 	 * update it, otherwise create a new one.
2532 	 */
2533 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2534 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2535 			sc->sc_brtexceeded++;
2536 			return (ENOSPC);
2537 		}
2538 		/* Check per interface address limits (if enabled) */
2539 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2540 			bif->bif_addrexceeded++;
2541 			return (ENOSPC);
2542 		}
2543 
2544 		/*
2545 		 * Allocate a new bridge forwarding node, and
2546 		 * initialize the expiration time and Ethernet
2547 		 * address.
2548 		 */
2549 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2550 		if (brt == NULL)
2551 			return (ENOMEM);
2552 
2553 		if (bif->bif_flags & IFBIF_STICKY)
2554 			brt->brt_flags = IFBAF_STICKY;
2555 		else
2556 			brt->brt_flags = IFBAF_DYNAMIC;
2557 
2558 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2559 		brt->brt_vlan = vlan;
2560 
2561 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2562 			uma_zfree(bridge_rtnode_zone, brt);
2563 			return (error);
2564 		}
2565 		brt->brt_dst = bif;
2566 		bif->bif_addrcnt++;
2567 	}
2568 
2569 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2570 	    brt->brt_dst != bif) {
2571 		brt->brt_dst->bif_addrcnt--;
2572 		brt->brt_dst = bif;
2573 		brt->brt_dst->bif_addrcnt++;
2574 	}
2575 
2576 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2577 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2578 	if (setflags)
2579 		brt->brt_flags = flags;
2580 
2581 	return (0);
2582 }
2583 
2584 /*
2585  * bridge_rtlookup:
2586  *
2587  *	Lookup the destination interface for an address.
2588  */
2589 static struct ifnet *
2590 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2591 {
2592 	struct bridge_rtnode *brt;
2593 
2594 	BRIDGE_LOCK_ASSERT(sc);
2595 
2596 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2597 		return (NULL);
2598 
2599 	return (brt->brt_ifp);
2600 }
2601 
2602 /*
2603  * bridge_rttrim:
2604  *
2605  *	Trim the routine table so that we have a number
2606  *	of routing entries less than or equal to the
2607  *	maximum number.
2608  */
2609 static void
2610 bridge_rttrim(struct bridge_softc *sc)
2611 {
2612 	struct bridge_rtnode *brt, *nbrt;
2613 
2614 	BRIDGE_LOCK_ASSERT(sc);
2615 
2616 	/* Make sure we actually need to do this. */
2617 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2618 		return;
2619 
2620 	/* Force an aging cycle; this might trim enough addresses. */
2621 	bridge_rtage(sc);
2622 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2623 		return;
2624 
2625 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2626 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2627 			bridge_rtnode_destroy(sc, brt);
2628 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2629 				return;
2630 		}
2631 	}
2632 }
2633 
2634 /*
2635  * bridge_timer:
2636  *
2637  *	Aging timer for the bridge.
2638  */
2639 static void
2640 bridge_timer(void *arg)
2641 {
2642 	struct bridge_softc *sc = arg;
2643 
2644 	BRIDGE_LOCK_ASSERT(sc);
2645 
2646 	bridge_rtage(sc);
2647 
2648 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2649 		callout_reset(&sc->sc_brcallout,
2650 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2651 }
2652 
2653 /*
2654  * bridge_rtage:
2655  *
2656  *	Perform an aging cycle.
2657  */
2658 static void
2659 bridge_rtage(struct bridge_softc *sc)
2660 {
2661 	struct bridge_rtnode *brt, *nbrt;
2662 
2663 	BRIDGE_LOCK_ASSERT(sc);
2664 
2665 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2666 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2667 			if (time_uptime >= brt->brt_expire)
2668 				bridge_rtnode_destroy(sc, brt);
2669 		}
2670 	}
2671 }
2672 
2673 /*
2674  * bridge_rtflush:
2675  *
2676  *	Remove all dynamic addresses from the bridge.
2677  */
2678 static void
2679 bridge_rtflush(struct bridge_softc *sc, int full)
2680 {
2681 	struct bridge_rtnode *brt, *nbrt;
2682 
2683 	BRIDGE_LOCK_ASSERT(sc);
2684 
2685 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2686 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2687 			bridge_rtnode_destroy(sc, brt);
2688 	}
2689 }
2690 
2691 /*
2692  * bridge_rtdaddr:
2693  *
2694  *	Remove an address from the table.
2695  */
2696 static int
2697 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2698 {
2699 	struct bridge_rtnode *brt;
2700 	int found = 0;
2701 
2702 	BRIDGE_LOCK_ASSERT(sc);
2703 
2704 	/*
2705 	 * If vlan is zero then we want to delete for all vlans so the lookup
2706 	 * may return more than one.
2707 	 */
2708 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2709 		bridge_rtnode_destroy(sc, brt);
2710 		found = 1;
2711 	}
2712 
2713 	return (found ? 0 : ENOENT);
2714 }
2715 
2716 /*
2717  * bridge_rtdelete:
2718  *
2719  *	Delete routes to a speicifc member interface.
2720  */
2721 static void
2722 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2723 {
2724 	struct bridge_rtnode *brt, *nbrt;
2725 
2726 	BRIDGE_LOCK_ASSERT(sc);
2727 
2728 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2729 		if (brt->brt_ifp == ifp && (full ||
2730 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2731 			bridge_rtnode_destroy(sc, brt);
2732 	}
2733 }
2734 
2735 /*
2736  * bridge_rtable_init:
2737  *
2738  *	Initialize the route table for this bridge.
2739  */
2740 static int
2741 bridge_rtable_init(struct bridge_softc *sc)
2742 {
2743 	int i;
2744 
2745 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2746 	    M_DEVBUF, M_NOWAIT);
2747 	if (sc->sc_rthash == NULL)
2748 		return (ENOMEM);
2749 
2750 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2751 		LIST_INIT(&sc->sc_rthash[i]);
2752 
2753 	sc->sc_rthash_key = arc4random();
2754 
2755 	LIST_INIT(&sc->sc_rtlist);
2756 
2757 	return (0);
2758 }
2759 
2760 /*
2761  * bridge_rtable_fini:
2762  *
2763  *	Deconstruct the route table for this bridge.
2764  */
2765 static void
2766 bridge_rtable_fini(struct bridge_softc *sc)
2767 {
2768 
2769 	KASSERT(sc->sc_brtcnt == 0,
2770 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2771 	free(sc->sc_rthash, M_DEVBUF);
2772 }
2773 
2774 /*
2775  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2776  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2777  */
2778 #define	mix(a, b, c)							\
2779 do {									\
2780 	a -= b; a -= c; a ^= (c >> 13);					\
2781 	b -= c; b -= a; b ^= (a << 8);					\
2782 	c -= a; c -= b; c ^= (b >> 13);					\
2783 	a -= b; a -= c; a ^= (c >> 12);					\
2784 	b -= c; b -= a; b ^= (a << 16);					\
2785 	c -= a; c -= b; c ^= (b >> 5);					\
2786 	a -= b; a -= c; a ^= (c >> 3);					\
2787 	b -= c; b -= a; b ^= (a << 10);					\
2788 	c -= a; c -= b; c ^= (b >> 15);					\
2789 } while (/*CONSTCOND*/0)
2790 
2791 static __inline uint32_t
2792 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2793 {
2794 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2795 
2796 	b += addr[5] << 8;
2797 	b += addr[4];
2798 	a += addr[3] << 24;
2799 	a += addr[2] << 16;
2800 	a += addr[1] << 8;
2801 	a += addr[0];
2802 
2803 	mix(a, b, c);
2804 
2805 	return (c & BRIDGE_RTHASH_MASK);
2806 }
2807 
2808 #undef mix
2809 
2810 static int
2811 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2812 {
2813 	int i, d;
2814 
2815 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2816 		d = ((int)a[i]) - ((int)b[i]);
2817 	}
2818 
2819 	return (d);
2820 }
2821 
2822 /*
2823  * bridge_rtnode_lookup:
2824  *
2825  *	Look up a bridge route node for the specified destination. Compare the
2826  *	vlan id or if zero then just return the first match.
2827  */
2828 static struct bridge_rtnode *
2829 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2830 {
2831 	struct bridge_rtnode *brt;
2832 	uint32_t hash;
2833 	int dir;
2834 
2835 	BRIDGE_LOCK_ASSERT(sc);
2836 
2837 	hash = bridge_rthash(sc, addr);
2838 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2839 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2840 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2841 			return (brt);
2842 		if (dir > 0)
2843 			return (NULL);
2844 	}
2845 
2846 	return (NULL);
2847 }
2848 
2849 /*
2850  * bridge_rtnode_insert:
2851  *
2852  *	Insert the specified bridge node into the route table.  We
2853  *	assume the entry is not already in the table.
2854  */
2855 static int
2856 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2857 {
2858 	struct bridge_rtnode *lbrt;
2859 	uint32_t hash;
2860 	int dir;
2861 
2862 	BRIDGE_LOCK_ASSERT(sc);
2863 
2864 	hash = bridge_rthash(sc, brt->brt_addr);
2865 
2866 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2867 	if (lbrt == NULL) {
2868 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2869 		goto out;
2870 	}
2871 
2872 	do {
2873 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2874 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2875 			return (EEXIST);
2876 		if (dir > 0) {
2877 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2878 			goto out;
2879 		}
2880 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2881 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2882 			goto out;
2883 		}
2884 		lbrt = LIST_NEXT(lbrt, brt_hash);
2885 	} while (lbrt != NULL);
2886 
2887 #ifdef DIAGNOSTIC
2888 	panic("bridge_rtnode_insert: impossible");
2889 #endif
2890 
2891 out:
2892 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2893 	sc->sc_brtcnt++;
2894 
2895 	return (0);
2896 }
2897 
2898 /*
2899  * bridge_rtnode_destroy:
2900  *
2901  *	Destroy a bridge rtnode.
2902  */
2903 static void
2904 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
2905 {
2906 	BRIDGE_LOCK_ASSERT(sc);
2907 
2908 	LIST_REMOVE(brt, brt_hash);
2909 
2910 	LIST_REMOVE(brt, brt_list);
2911 	sc->sc_brtcnt--;
2912 	brt->brt_dst->bif_addrcnt--;
2913 	uma_zfree(bridge_rtnode_zone, brt);
2914 }
2915 
2916 /*
2917  * bridge_rtable_expire:
2918  *
2919  *	Set the expiry time for all routes on an interface.
2920  */
2921 static void
2922 bridge_rtable_expire(struct ifnet *ifp, int age)
2923 {
2924 	struct bridge_softc *sc = ifp->if_bridge;
2925 	struct bridge_rtnode *brt;
2926 
2927 	BRIDGE_LOCK(sc);
2928 
2929 	/*
2930 	 * If the age is zero then flush, otherwise set all the expiry times to
2931 	 * age for the interface
2932 	 */
2933 	if (age == 0)
2934 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
2935 	else {
2936 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
2937 			/* Cap the expiry time to 'age' */
2938 			if (brt->brt_ifp == ifp &&
2939 			    brt->brt_expire > time_uptime + age &&
2940 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2941 				brt->brt_expire = time_uptime + age;
2942 		}
2943 	}
2944 	BRIDGE_UNLOCK(sc);
2945 }
2946 
2947 /*
2948  * bridge_state_change:
2949  *
2950  *	Callback from the bridgestp code when a port changes states.
2951  */
2952 static void
2953 bridge_state_change(struct ifnet *ifp, int state)
2954 {
2955 	struct bridge_softc *sc = ifp->if_bridge;
2956 	static const char *stpstates[] = {
2957 		"disabled",
2958 		"listening",
2959 		"learning",
2960 		"forwarding",
2961 		"blocking",
2962 		"discarding"
2963 	};
2964 
2965 	if (log_stp)
2966 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
2967 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
2968 }
2969 
2970 /*
2971  * Send bridge packets through pfil if they are one of the types pfil can deal
2972  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
2973  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
2974  * that interface.
2975  */
2976 static int
2977 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
2978 {
2979 	int snap, error, i, hlen;
2980 	struct ether_header *eh1, eh2;
2981 	struct ip *ip;
2982 	struct llc llc1;
2983 	u_int16_t ether_type;
2984 
2985 	snap = 0;
2986 	error = -1;	/* Default error if not error == 0 */
2987 
2988 #if 0
2989 	/* we may return with the IP fields swapped, ensure its not shared */
2990 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
2991 #endif
2992 
2993 	if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0)
2994 		return (0); /* filtering is disabled */
2995 
2996 	i = min((*mp)->m_pkthdr.len, max_protohdr);
2997 	if ((*mp)->m_len < i) {
2998 	    *mp = m_pullup(*mp, i);
2999 	    if (*mp == NULL) {
3000 		printf("%s: m_pullup failed\n", __func__);
3001 		return (-1);
3002 	    }
3003 	}
3004 
3005 	eh1 = mtod(*mp, struct ether_header *);
3006 	ether_type = ntohs(eh1->ether_type);
3007 
3008 	/*
3009 	 * Check for SNAP/LLC.
3010 	 */
3011 	if (ether_type < ETHERMTU) {
3012 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3013 
3014 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3015 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3016 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3017 		    llc2->llc_control == LLC_UI) {
3018 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3019 			snap = 1;
3020 		}
3021 	}
3022 
3023 	/*
3024 	 * If we're trying to filter bridge traffic, don't look at anything
3025 	 * other than IP and ARP traffic.  If the filter doesn't understand
3026 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3027 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3028 	 * but of course we don't have an AppleTalk filter to begin with.
3029 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3030 	 * ARP traffic.)
3031 	 */
3032 	switch (ether_type) {
3033 		case ETHERTYPE_ARP:
3034 		case ETHERTYPE_REVARP:
3035 			if (pfil_ipfw_arp == 0)
3036 				return (0); /* Automatically pass */
3037 			break;
3038 
3039 		case ETHERTYPE_IP:
3040 #ifdef INET6
3041 		case ETHERTYPE_IPV6:
3042 #endif /* INET6 */
3043 			break;
3044 		default:
3045 			/*
3046 			 * Check to see if the user wants to pass non-ip
3047 			 * packets, these will not be checked by pfil(9) and
3048 			 * passed unconditionally so the default is to drop.
3049 			 */
3050 			if (pfil_onlyip)
3051 				goto bad;
3052 	}
3053 
3054 	/* Run the packet through pfil before stripping link headers */
3055 	if (PFIL_HOOKED(&V_link_pfil_hook) && pfil_ipfw != 0 &&
3056 			dir == PFIL_OUT && ifp != NULL) {
3057 
3058 		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3059 
3060 		if (*mp == NULL || error != 0) /* packet consumed by filter */
3061 			return (error);
3062 	}
3063 
3064 	/* Strip off the Ethernet header and keep a copy. */
3065 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3066 	m_adj(*mp, ETHER_HDR_LEN);
3067 
3068 	/* Strip off snap header, if present */
3069 	if (snap) {
3070 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3071 		m_adj(*mp, sizeof(struct llc));
3072 	}
3073 
3074 	/*
3075 	 * Check the IP header for alignment and errors
3076 	 */
3077 	if (dir == PFIL_IN) {
3078 		switch (ether_type) {
3079 			case ETHERTYPE_IP:
3080 				error = bridge_ip_checkbasic(mp);
3081 				break;
3082 #ifdef INET6
3083 			case ETHERTYPE_IPV6:
3084 				error = bridge_ip6_checkbasic(mp);
3085 				break;
3086 #endif /* INET6 */
3087 			default:
3088 				error = 0;
3089 		}
3090 		if (error)
3091 			goto bad;
3092 	}
3093 
3094 	error = 0;
3095 
3096 	/*
3097 	 * Run the packet through pfil
3098 	 */
3099 	switch (ether_type) {
3100 	case ETHERTYPE_IP:
3101 		/*
3102 		 * before calling the firewall, swap fields the same as
3103 		 * IP does. here we assume the header is contiguous
3104 		 */
3105 		ip = mtod(*mp, struct ip *);
3106 
3107 		ip->ip_len = ntohs(ip->ip_len);
3108 		ip->ip_off = ntohs(ip->ip_off);
3109 
3110 		/*
3111 		 * Run pfil on the member interface and the bridge, both can
3112 		 * be skipped by clearing pfil_member or pfil_bridge.
3113 		 *
3114 		 * Keep the order:
3115 		 *   in_if -> bridge_if -> out_if
3116 		 */
3117 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3118 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3119 					dir, NULL);
3120 
3121 		if (*mp == NULL || error != 0) /* filter may consume */
3122 			break;
3123 
3124 		if (pfil_member && ifp != NULL)
3125 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3126 					dir, NULL);
3127 
3128 		if (*mp == NULL || error != 0) /* filter may consume */
3129 			break;
3130 
3131 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3132 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3133 					dir, NULL);
3134 
3135 		if (*mp == NULL || error != 0) /* filter may consume */
3136 			break;
3137 
3138 		/* check if we need to fragment the packet */
3139 		if (pfil_member && ifp != NULL && dir == PFIL_OUT) {
3140 			i = (*mp)->m_pkthdr.len;
3141 			if (i > ifp->if_mtu) {
3142 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3143 					    &llc1);
3144 				return (error);
3145 			}
3146 		}
3147 
3148 		/* Recalculate the ip checksum and restore byte ordering */
3149 		ip = mtod(*mp, struct ip *);
3150 		hlen = ip->ip_hl << 2;
3151 		if (hlen < sizeof(struct ip))
3152 			goto bad;
3153 		if (hlen > (*mp)->m_len) {
3154 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3155 				goto bad;
3156 			ip = mtod(*mp, struct ip *);
3157 			if (ip == NULL)
3158 				goto bad;
3159 		}
3160 		ip->ip_len = htons(ip->ip_len);
3161 		ip->ip_off = htons(ip->ip_off);
3162 		ip->ip_sum = 0;
3163 		if (hlen == sizeof(struct ip))
3164 			ip->ip_sum = in_cksum_hdr(ip);
3165 		else
3166 			ip->ip_sum = in_cksum(*mp, hlen);
3167 
3168 		break;
3169 #ifdef INET6
3170 	case ETHERTYPE_IPV6:
3171 		if (pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3172 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3173 					dir, NULL);
3174 
3175 		if (*mp == NULL || error != 0) /* filter may consume */
3176 			break;
3177 
3178 		if (pfil_member && ifp != NULL)
3179 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3180 					dir, NULL);
3181 
3182 		if (*mp == NULL || error != 0) /* filter may consume */
3183 			break;
3184 
3185 		if (pfil_bridge && dir == PFIL_IN && bifp != NULL)
3186 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3187 					dir, NULL);
3188 		break;
3189 #endif
3190 	default:
3191 		error = 0;
3192 		break;
3193 	}
3194 
3195 	if (*mp == NULL)
3196 		return (error);
3197 	if (error != 0)
3198 		goto bad;
3199 
3200 	error = -1;
3201 
3202 	/*
3203 	 * Finally, put everything back the way it was and return
3204 	 */
3205 	if (snap) {
3206 		M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT);
3207 		if (*mp == NULL)
3208 			return (error);
3209 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3210 	}
3211 
3212 	M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT);
3213 	if (*mp == NULL)
3214 		return (error);
3215 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3216 
3217 	return (0);
3218 
3219 bad:
3220 	m_freem(*mp);
3221 	*mp = NULL;
3222 	return (error);
3223 }
3224 
3225 /*
3226  * Perform basic checks on header size since
3227  * pfil assumes ip_input has already processed
3228  * it for it.  Cut-and-pasted from ip_input.c.
3229  * Given how simple the IPv6 version is,
3230  * does the IPv4 version really need to be
3231  * this complicated?
3232  *
3233  * XXX Should we update ipstat here, or not?
3234  * XXX Right now we update ipstat but not
3235  * XXX csum_counter.
3236  */
3237 static int
3238 bridge_ip_checkbasic(struct mbuf **mp)
3239 {
3240 	struct mbuf *m = *mp;
3241 	struct ip *ip;
3242 	int len, hlen;
3243 	u_short sum;
3244 
3245 	if (*mp == NULL)
3246 		return (-1);
3247 
3248 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3249 		if ((m = m_copyup(m, sizeof(struct ip),
3250 			(max_linkhdr + 3) & ~3)) == NULL) {
3251 			/* XXXJRT new stat, please */
3252 			KMOD_IPSTAT_INC(ips_toosmall);
3253 			goto bad;
3254 		}
3255 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3256 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3257 			KMOD_IPSTAT_INC(ips_toosmall);
3258 			goto bad;
3259 		}
3260 	}
3261 	ip = mtod(m, struct ip *);
3262 	if (ip == NULL) goto bad;
3263 
3264 	if (ip->ip_v != IPVERSION) {
3265 		KMOD_IPSTAT_INC(ips_badvers);
3266 		goto bad;
3267 	}
3268 	hlen = ip->ip_hl << 2;
3269 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3270 		KMOD_IPSTAT_INC(ips_badhlen);
3271 		goto bad;
3272 	}
3273 	if (hlen > m->m_len) {
3274 		if ((m = m_pullup(m, hlen)) == 0) {
3275 			KMOD_IPSTAT_INC(ips_badhlen);
3276 			goto bad;
3277 		}
3278 		ip = mtod(m, struct ip *);
3279 		if (ip == NULL) goto bad;
3280 	}
3281 
3282 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3283 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3284 	} else {
3285 		if (hlen == sizeof(struct ip)) {
3286 			sum = in_cksum_hdr(ip);
3287 		} else {
3288 			sum = in_cksum(m, hlen);
3289 		}
3290 	}
3291 	if (sum) {
3292 		KMOD_IPSTAT_INC(ips_badsum);
3293 		goto bad;
3294 	}
3295 
3296 	/* Retrieve the packet length. */
3297 	len = ntohs(ip->ip_len);
3298 
3299 	/*
3300 	 * Check for additional length bogosity
3301 	 */
3302 	if (len < hlen) {
3303 		KMOD_IPSTAT_INC(ips_badlen);
3304 		goto bad;
3305 	}
3306 
3307 	/*
3308 	 * Check that the amount of data in the buffers
3309 	 * is as at least much as the IP header would have us expect.
3310 	 * Drop packet if shorter than we expect.
3311 	 */
3312 	if (m->m_pkthdr.len < len) {
3313 		KMOD_IPSTAT_INC(ips_tooshort);
3314 		goto bad;
3315 	}
3316 
3317 	/* Checks out, proceed */
3318 	*mp = m;
3319 	return (0);
3320 
3321 bad:
3322 	*mp = m;
3323 	return (-1);
3324 }
3325 
3326 #ifdef INET6
3327 /*
3328  * Same as above, but for IPv6.
3329  * Cut-and-pasted from ip6_input.c.
3330  * XXX Should we update ip6stat, or not?
3331  */
3332 static int
3333 bridge_ip6_checkbasic(struct mbuf **mp)
3334 {
3335 	struct mbuf *m = *mp;
3336 	struct ip6_hdr *ip6;
3337 
3338 	/*
3339 	 * If the IPv6 header is not aligned, slurp it up into a new
3340 	 * mbuf with space for link headers, in the event we forward
3341 	 * it.  Otherwise, if it is aligned, make sure the entire base
3342 	 * IPv6 header is in the first mbuf of the chain.
3343 	 */
3344 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3345 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3346 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3347 			    (max_linkhdr + 3) & ~3)) == NULL) {
3348 			/* XXXJRT new stat, please */
3349 			V_ip6stat.ip6s_toosmall++;
3350 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3351 			goto bad;
3352 		}
3353 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3354 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3355 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3356 			V_ip6stat.ip6s_toosmall++;
3357 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3358 			goto bad;
3359 		}
3360 	}
3361 
3362 	ip6 = mtod(m, struct ip6_hdr *);
3363 
3364 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3365 		V_ip6stat.ip6s_badvers++;
3366 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3367 		goto bad;
3368 	}
3369 
3370 	/* Checks out, proceed */
3371 	*mp = m;
3372 	return (0);
3373 
3374 bad:
3375 	*mp = m;
3376 	return (-1);
3377 }
3378 #endif /* INET6 */
3379 
3380 /*
3381  * bridge_fragment:
3382  *
3383  *	Return a fragmented mbuf chain.
3384  */
3385 static int
3386 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3387     int snap, struct llc *llc)
3388 {
3389 	struct mbuf *m0;
3390 	struct ip *ip;
3391 	int error = -1;
3392 
3393 	if (m->m_len < sizeof(struct ip) &&
3394 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3395 		goto out;
3396 	ip = mtod(m, struct ip *);
3397 
3398 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist,
3399 		    CSUM_DELAY_IP);
3400 	if (error)
3401 		goto out;
3402 
3403 	/* walk the chain and re-add the Ethernet header */
3404 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3405 		if (error == 0) {
3406 			if (snap) {
3407 				M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT);
3408 				if (m0 == NULL) {
3409 					error = ENOBUFS;
3410 					continue;
3411 				}
3412 				bcopy(llc, mtod(m0, caddr_t),
3413 				    sizeof(struct llc));
3414 			}
3415 			M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT);
3416 			if (m0 == NULL) {
3417 				error = ENOBUFS;
3418 				continue;
3419 			}
3420 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3421 		} else
3422 			m_freem(m);
3423 	}
3424 
3425 	if (error == 0)
3426 		KMOD_IPSTAT_INC(ips_fragmented);
3427 
3428 	return (error);
3429 
3430 out:
3431 	if (m != NULL)
3432 		m_freem(m);
3433 	return (error);
3434 }
3435 
3436 static void
3437 bridge_linkstate(struct ifnet *ifp)
3438 {
3439 	struct bridge_softc *sc = ifp->if_bridge;
3440 	struct bridge_iflist *bif;
3441 
3442 	BRIDGE_LOCK(sc);
3443 	bif = bridge_lookup_member_if(sc, ifp);
3444 	if (bif == NULL) {
3445 		BRIDGE_UNLOCK(sc);
3446 		return;
3447 	}
3448 	bridge_linkcheck(sc);
3449 	BRIDGE_UNLOCK(sc);
3450 
3451 	bstp_linkstate(&bif->bif_stp);
3452 }
3453 
3454 static void
3455 bridge_linkcheck(struct bridge_softc *sc)
3456 {
3457 	struct bridge_iflist *bif;
3458 	int new_link, hasls;
3459 
3460 	BRIDGE_LOCK_ASSERT(sc);
3461 	new_link = LINK_STATE_DOWN;
3462 	hasls = 0;
3463 	/* Our link is considered up if at least one of our ports is active */
3464 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3465 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3466 			hasls++;
3467 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3468 			new_link = LINK_STATE_UP;
3469 			break;
3470 		}
3471 	}
3472 	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3473 		/* If no interfaces support link-state then we default to up */
3474 		new_link = LINK_STATE_UP;
3475 	}
3476 	if_link_state_change(sc->sc_ifp, new_link);
3477 }
3478