xref: /freebsd/sys/net/if_bridge.c (revision ff0ba87247820afbdfdc1b307c803f7923d0e4d3)
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/eventhandler.h>
85 #include <sys/mbuf.h>
86 #include <sys/malloc.h>
87 #include <sys/protosw.h>
88 #include <sys/systm.h>
89 #include <sys/jail.h>
90 #include <sys/time.h>
91 #include <sys/socket.h> /* for net/if.h */
92 #include <sys/sockio.h>
93 #include <sys/ctype.h>  /* string functions */
94 #include <sys/kernel.h>
95 #include <sys/random.h>
96 #include <sys/syslog.h>
97 #include <sys/sysctl.h>
98 #include <vm/uma.h>
99 #include <sys/module.h>
100 #include <sys/priv.h>
101 #include <sys/proc.h>
102 #include <sys/lock.h>
103 #include <sys/mutex.h>
104 
105 #include <net/bpf.h>
106 #include <net/if.h>
107 #include <net/if_clone.h>
108 #include <net/if_dl.h>
109 #include <net/if_types.h>
110 #include <net/if_var.h>
111 #include <net/pfil.h>
112 #include <net/vnet.h>
113 
114 #include <netinet/in.h>
115 #include <netinet/in_systm.h>
116 #include <netinet/in_var.h>
117 #include <netinet/ip.h>
118 #include <netinet/ip_var.h>
119 #ifdef INET6
120 #include <netinet/ip6.h>
121 #include <netinet6/ip6_var.h>
122 #include <netinet6/in6_ifattach.h>
123 #endif
124 #if defined(INET) || defined(INET6)
125 #include <netinet/ip_carp.h>
126 #endif
127 #include <machine/in_cksum.h>
128 #include <netinet/if_ether.h>
129 #include <net/bridgestp.h>
130 #include <net/if_bridgevar.h>
131 #include <net/if_llc.h>
132 #include <net/if_vlan_var.h>
133 
134 #include <net/route.h>
135 
136 /*
137  * Size of the route hash table.  Must be a power of two.
138  */
139 #ifndef BRIDGE_RTHASH_SIZE
140 #define	BRIDGE_RTHASH_SIZE		1024
141 #endif
142 
143 #define	BRIDGE_RTHASH_MASK		(BRIDGE_RTHASH_SIZE - 1)
144 
145 /*
146  * Default maximum number of addresses to cache.
147  */
148 #ifndef BRIDGE_RTABLE_MAX
149 #define	BRIDGE_RTABLE_MAX		2000
150 #endif
151 
152 /*
153  * Timeout (in seconds) for entries learned dynamically.
154  */
155 #ifndef BRIDGE_RTABLE_TIMEOUT
156 #define	BRIDGE_RTABLE_TIMEOUT		(20 * 60)	/* same as ARP */
157 #endif
158 
159 /*
160  * Number of seconds between walks of the route list.
161  */
162 #ifndef BRIDGE_RTABLE_PRUNE_PERIOD
163 #define	BRIDGE_RTABLE_PRUNE_PERIOD	(5 * 60)
164 #endif
165 
166 /*
167  * List of capabilities to possibly mask on the member interface.
168  */
169 #define	BRIDGE_IFCAPS_MASK		(IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM)
170 
171 /*
172  * List of capabilities to strip
173  */
174 #define	BRIDGE_IFCAPS_STRIP		IFCAP_LRO
175 
176 /*
177  * Bridge interface list entry.
178  */
179 struct bridge_iflist {
180 	LIST_ENTRY(bridge_iflist) bif_next;
181 	struct ifnet		*bif_ifp;	/* member if */
182 	struct bstp_port	bif_stp;	/* STP state */
183 	uint32_t		bif_flags;	/* member if flags */
184 	int			bif_savedcaps;	/* saved capabilities */
185 	uint32_t		bif_addrmax;	/* max # of addresses */
186 	uint32_t		bif_addrcnt;	/* cur. # of addresses */
187 	uint32_t		bif_addrexceeded;/* # of address violations */
188 };
189 
190 /*
191  * Bridge route node.
192  */
193 struct bridge_rtnode {
194 	LIST_ENTRY(bridge_rtnode) brt_hash;	/* hash table linkage */
195 	LIST_ENTRY(bridge_rtnode) brt_list;	/* list linkage */
196 	struct bridge_iflist	*brt_dst;	/* destination if */
197 	unsigned long		brt_expire;	/* expiration time */
198 	uint8_t			brt_flags;	/* address flags */
199 	uint8_t			brt_addr[ETHER_ADDR_LEN];
200 	uint16_t		brt_vlan;	/* vlan id */
201 };
202 #define	brt_ifp			brt_dst->bif_ifp
203 
204 /*
205  * Software state for each bridge.
206  */
207 struct bridge_softc {
208 	struct ifnet		*sc_ifp;	/* make this an interface */
209 	LIST_ENTRY(bridge_softc) sc_list;
210 	struct mtx		sc_mtx;
211 	struct cv		sc_cv;
212 	uint32_t		sc_brtmax;	/* max # of addresses */
213 	uint32_t		sc_brtcnt;	/* cur. # of addresses */
214 	uint32_t		sc_brttimeout;	/* rt timeout in seconds */
215 	struct callout		sc_brcallout;	/* bridge callout */
216 	uint32_t		sc_iflist_ref;	/* refcount for sc_iflist */
217 	uint32_t		sc_iflist_xcnt;	/* refcount for sc_iflist */
218 	LIST_HEAD(, bridge_iflist) sc_iflist;	/* member interface list */
219 	LIST_HEAD(, bridge_rtnode) *sc_rthash;	/* our forwarding table */
220 	LIST_HEAD(, bridge_rtnode) sc_rtlist;	/* list version of above */
221 	uint32_t		sc_rthash_key;	/* key for hash */
222 	LIST_HEAD(, bridge_iflist) sc_spanlist;	/* span ports list */
223 	struct bstp_state	sc_stp;		/* STP state */
224 	uint32_t		sc_brtexceeded;	/* # of cache drops */
225 	struct ifnet		*sc_ifaddr;	/* member mac copied from */
226 	u_char			sc_defaddr[6];	/* Default MAC address */
227 };
228 
229 static VNET_DEFINE(struct mtx, bridge_list_mtx);
230 #define	V_bridge_list_mtx	VNET(bridge_list_mtx)
231 eventhandler_tag	bridge_detach_cookie = NULL;
232 
233 int	bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD;
234 
235 uma_zone_t bridge_rtnode_zone;
236 
237 static int	bridge_clone_create(struct if_clone *, int, caddr_t);
238 static void	bridge_clone_destroy(struct ifnet *);
239 
240 static int	bridge_ioctl(struct ifnet *, u_long, caddr_t);
241 static void	bridge_mutecaps(struct bridge_softc *);
242 static void	bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *,
243 		    int);
244 static void	bridge_ifdetach(void *arg __unused, struct ifnet *);
245 static void	bridge_init(void *);
246 static void	bridge_dummynet(struct mbuf *, struct ifnet *);
247 static void	bridge_stop(struct ifnet *, int);
248 static int	bridge_transmit(struct ifnet *, struct mbuf *);
249 static void	bridge_qflush(struct ifnet *);
250 static struct mbuf *bridge_input(struct ifnet *, struct mbuf *);
251 static int	bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *,
252 		    struct rtentry *);
253 static int	bridge_enqueue(struct bridge_softc *, struct ifnet *,
254 		    struct mbuf *);
255 static void	bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int);
256 
257 static void	bridge_forward(struct bridge_softc *, struct bridge_iflist *,
258 		    struct mbuf *m);
259 
260 static void	bridge_timer(void *);
261 
262 static void	bridge_broadcast(struct bridge_softc *, struct ifnet *,
263 		    struct mbuf *, int);
264 static void	bridge_span(struct bridge_softc *, struct mbuf *);
265 
266 static int	bridge_rtupdate(struct bridge_softc *, const uint8_t *,
267 		    uint16_t, struct bridge_iflist *, int, uint8_t);
268 static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *,
269 		    uint16_t);
270 static void	bridge_rttrim(struct bridge_softc *);
271 static void	bridge_rtage(struct bridge_softc *);
272 static void	bridge_rtflush(struct bridge_softc *, int);
273 static int	bridge_rtdaddr(struct bridge_softc *, const uint8_t *,
274 		    uint16_t);
275 
276 static void	bridge_rtable_init(struct bridge_softc *);
277 static void	bridge_rtable_fini(struct bridge_softc *);
278 
279 static int	bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *);
280 static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *,
281 		    const uint8_t *, uint16_t);
282 static int	bridge_rtnode_insert(struct bridge_softc *,
283 		    struct bridge_rtnode *);
284 static void	bridge_rtnode_destroy(struct bridge_softc *,
285 		    struct bridge_rtnode *);
286 static void	bridge_rtable_expire(struct ifnet *, int);
287 static void	bridge_state_change(struct ifnet *, int);
288 
289 static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *,
290 		    const char *name);
291 static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *,
292 		    struct ifnet *ifp);
293 static void	bridge_delete_member(struct bridge_softc *,
294 		    struct bridge_iflist *, int);
295 static void	bridge_delete_span(struct bridge_softc *,
296 		    struct bridge_iflist *);
297 
298 static int	bridge_ioctl_add(struct bridge_softc *, void *);
299 static int	bridge_ioctl_del(struct bridge_softc *, void *);
300 static int	bridge_ioctl_gifflags(struct bridge_softc *, void *);
301 static int	bridge_ioctl_sifflags(struct bridge_softc *, void *);
302 static int	bridge_ioctl_scache(struct bridge_softc *, void *);
303 static int	bridge_ioctl_gcache(struct bridge_softc *, void *);
304 static int	bridge_ioctl_gifs(struct bridge_softc *, void *);
305 static int	bridge_ioctl_rts(struct bridge_softc *, void *);
306 static int	bridge_ioctl_saddr(struct bridge_softc *, void *);
307 static int	bridge_ioctl_sto(struct bridge_softc *, void *);
308 static int	bridge_ioctl_gto(struct bridge_softc *, void *);
309 static int	bridge_ioctl_daddr(struct bridge_softc *, void *);
310 static int	bridge_ioctl_flush(struct bridge_softc *, void *);
311 static int	bridge_ioctl_gpri(struct bridge_softc *, void *);
312 static int	bridge_ioctl_spri(struct bridge_softc *, void *);
313 static int	bridge_ioctl_ght(struct bridge_softc *, void *);
314 static int	bridge_ioctl_sht(struct bridge_softc *, void *);
315 static int	bridge_ioctl_gfd(struct bridge_softc *, void *);
316 static int	bridge_ioctl_sfd(struct bridge_softc *, void *);
317 static int	bridge_ioctl_gma(struct bridge_softc *, void *);
318 static int	bridge_ioctl_sma(struct bridge_softc *, void *);
319 static int	bridge_ioctl_sifprio(struct bridge_softc *, void *);
320 static int	bridge_ioctl_sifcost(struct bridge_softc *, void *);
321 static int	bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *);
322 static int	bridge_ioctl_addspan(struct bridge_softc *, void *);
323 static int	bridge_ioctl_delspan(struct bridge_softc *, void *);
324 static int	bridge_ioctl_gbparam(struct bridge_softc *, void *);
325 static int	bridge_ioctl_grte(struct bridge_softc *, void *);
326 static int	bridge_ioctl_gifsstp(struct bridge_softc *, void *);
327 static int	bridge_ioctl_sproto(struct bridge_softc *, void *);
328 static int	bridge_ioctl_stxhc(struct bridge_softc *, void *);
329 static int	bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *,
330 		    int);
331 static int	bridge_ip_checkbasic(struct mbuf **mp);
332 #ifdef INET6
333 static int	bridge_ip6_checkbasic(struct mbuf **mp);
334 #endif /* INET6 */
335 static int	bridge_fragment(struct ifnet *, struct mbuf *,
336 		    struct ether_header *, int, struct llc *);
337 static void	bridge_linkstate(struct ifnet *ifp);
338 static void	bridge_linkcheck(struct bridge_softc *sc);
339 
340 extern void (*bridge_linkstate_p)(struct ifnet *ifp);
341 
342 /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */
343 #define	VLANTAGOF(_m)	\
344     (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1
345 
346 static struct bstp_cb_ops bridge_ops = {
347 	.bcb_state = bridge_state_change,
348 	.bcb_rtage = bridge_rtable_expire
349 };
350 
351 SYSCTL_DECL(_net_link);
352 static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge");
353 
354 /* only pass IP[46] packets when pfil is enabled */
355 static VNET_DEFINE(int, pfil_onlyip) = 1;
356 #define	V_pfil_onlyip	VNET(pfil_onlyip)
357 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip,
358     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_onlyip), 0,
359     "Only pass IP packets when pfil is enabled");
360 
361 /* run pfil hooks on the bridge interface */
362 static VNET_DEFINE(int, pfil_bridge) = 1;
363 #define	V_pfil_bridge	VNET(pfil_bridge)
364 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge,
365     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_bridge), 0,
366     "Packet filter on the bridge interface");
367 
368 /* layer2 filter with ipfw */
369 static VNET_DEFINE(int, pfil_ipfw);
370 #define	V_pfil_ipfw	VNET(pfil_ipfw)
371 
372 /* layer2 ARP filter with ipfw */
373 static VNET_DEFINE(int, pfil_ipfw_arp);
374 #define	V_pfil_ipfw_arp	VNET(pfil_ipfw_arp)
375 SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp,
376     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_ipfw_arp), 0,
377     "Filter ARP packets through IPFW layer2");
378 
379 /* run pfil hooks on the member interface */
380 static VNET_DEFINE(int, pfil_member) = 1;
381 #define	V_pfil_member	VNET(pfil_member)
382 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member,
383     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_member), 0,
384     "Packet filter on the member interface");
385 
386 /* run pfil hooks on the physical interface for locally destined packets */
387 static VNET_DEFINE(int, pfil_local_phys);
388 #define	V_pfil_local_phys	VNET(pfil_local_phys)
389 SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys,
390     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(pfil_local_phys), 0,
391     "Packet filter on the physical interface for locally destined packets");
392 
393 /* log STP state changes */
394 static VNET_DEFINE(int, log_stp);
395 #define	V_log_stp	VNET(log_stp)
396 SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp,
397     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(log_stp), 0,
398     "Log STP state changes");
399 
400 /* share MAC with first bridge member */
401 static VNET_DEFINE(int, bridge_inherit_mac);
402 #define	V_bridge_inherit_mac	VNET(bridge_inherit_mac)
403 SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac,
404     CTLFLAG_RWTUN | CTLFLAG_VNET, &VNET_NAME(bridge_inherit_mac), 0,
405     "Inherit MAC address from the first bridge member");
406 
407 static VNET_DEFINE(int, allow_llz_overlap) = 0;
408 #define	V_allow_llz_overlap	VNET(allow_llz_overlap)
409 SYSCTL_INT(_net_link_bridge, OID_AUTO, allow_llz_overlap,
410     CTLFLAG_VNET | CTLFLAG_RW | CTLFLAG_VNET, &VNET_NAME(allow_llz_overlap), 0,
411     "Allow overlap of link-local scope "
412     "zones of a bridge interface and the member interfaces");
413 
414 struct bridge_control {
415 	int	(*bc_func)(struct bridge_softc *, void *);
416 	int	bc_argsize;
417 	int	bc_flags;
418 };
419 
420 #define	BC_F_COPYIN		0x01	/* copy arguments in */
421 #define	BC_F_COPYOUT		0x02	/* copy arguments out */
422 #define	BC_F_SUSER		0x04	/* do super-user check */
423 
424 const struct bridge_control bridge_control_table[] = {
425 	{ bridge_ioctl_add,		sizeof(struct ifbreq),
426 	  BC_F_COPYIN|BC_F_SUSER },
427 	{ bridge_ioctl_del,		sizeof(struct ifbreq),
428 	  BC_F_COPYIN|BC_F_SUSER },
429 
430 	{ bridge_ioctl_gifflags,	sizeof(struct ifbreq),
431 	  BC_F_COPYIN|BC_F_COPYOUT },
432 	{ bridge_ioctl_sifflags,	sizeof(struct ifbreq),
433 	  BC_F_COPYIN|BC_F_SUSER },
434 
435 	{ bridge_ioctl_scache,		sizeof(struct ifbrparam),
436 	  BC_F_COPYIN|BC_F_SUSER },
437 	{ bridge_ioctl_gcache,		sizeof(struct ifbrparam),
438 	  BC_F_COPYOUT },
439 
440 	{ bridge_ioctl_gifs,		sizeof(struct ifbifconf),
441 	  BC_F_COPYIN|BC_F_COPYOUT },
442 	{ bridge_ioctl_rts,		sizeof(struct ifbaconf),
443 	  BC_F_COPYIN|BC_F_COPYOUT },
444 
445 	{ bridge_ioctl_saddr,		sizeof(struct ifbareq),
446 	  BC_F_COPYIN|BC_F_SUSER },
447 
448 	{ bridge_ioctl_sto,		sizeof(struct ifbrparam),
449 	  BC_F_COPYIN|BC_F_SUSER },
450 	{ bridge_ioctl_gto,		sizeof(struct ifbrparam),
451 	  BC_F_COPYOUT },
452 
453 	{ bridge_ioctl_daddr,		sizeof(struct ifbareq),
454 	  BC_F_COPYIN|BC_F_SUSER },
455 
456 	{ bridge_ioctl_flush,		sizeof(struct ifbreq),
457 	  BC_F_COPYIN|BC_F_SUSER },
458 
459 	{ bridge_ioctl_gpri,		sizeof(struct ifbrparam),
460 	  BC_F_COPYOUT },
461 	{ bridge_ioctl_spri,		sizeof(struct ifbrparam),
462 	  BC_F_COPYIN|BC_F_SUSER },
463 
464 	{ bridge_ioctl_ght,		sizeof(struct ifbrparam),
465 	  BC_F_COPYOUT },
466 	{ bridge_ioctl_sht,		sizeof(struct ifbrparam),
467 	  BC_F_COPYIN|BC_F_SUSER },
468 
469 	{ bridge_ioctl_gfd,		sizeof(struct ifbrparam),
470 	  BC_F_COPYOUT },
471 	{ bridge_ioctl_sfd,		sizeof(struct ifbrparam),
472 	  BC_F_COPYIN|BC_F_SUSER },
473 
474 	{ bridge_ioctl_gma,		sizeof(struct ifbrparam),
475 	  BC_F_COPYOUT },
476 	{ bridge_ioctl_sma,		sizeof(struct ifbrparam),
477 	  BC_F_COPYIN|BC_F_SUSER },
478 
479 	{ bridge_ioctl_sifprio,		sizeof(struct ifbreq),
480 	  BC_F_COPYIN|BC_F_SUSER },
481 
482 	{ bridge_ioctl_sifcost,		sizeof(struct ifbreq),
483 	  BC_F_COPYIN|BC_F_SUSER },
484 
485 	{ bridge_ioctl_addspan,		sizeof(struct ifbreq),
486 	  BC_F_COPYIN|BC_F_SUSER },
487 	{ bridge_ioctl_delspan,		sizeof(struct ifbreq),
488 	  BC_F_COPYIN|BC_F_SUSER },
489 
490 	{ bridge_ioctl_gbparam,		sizeof(struct ifbropreq),
491 	  BC_F_COPYOUT },
492 
493 	{ bridge_ioctl_grte,		sizeof(struct ifbrparam),
494 	  BC_F_COPYOUT },
495 
496 	{ bridge_ioctl_gifsstp,		sizeof(struct ifbpstpconf),
497 	  BC_F_COPYIN|BC_F_COPYOUT },
498 
499 	{ bridge_ioctl_sproto,		sizeof(struct ifbrparam),
500 	  BC_F_COPYIN|BC_F_SUSER },
501 
502 	{ bridge_ioctl_stxhc,		sizeof(struct ifbrparam),
503 	  BC_F_COPYIN|BC_F_SUSER },
504 
505 	{ bridge_ioctl_sifmaxaddr,	sizeof(struct ifbreq),
506 	  BC_F_COPYIN|BC_F_SUSER },
507 
508 };
509 const int bridge_control_table_size = nitems(bridge_control_table);
510 
511 static VNET_DEFINE(LIST_HEAD(, bridge_softc), bridge_list);
512 #define	V_bridge_list	VNET(bridge_list)
513 #define	BRIDGE_LIST_LOCK_INIT(x)	mtx_init(&V_bridge_list_mtx,	\
514 					    "if_bridge list", NULL, MTX_DEF)
515 #define	BRIDGE_LIST_LOCK_DESTROY(x)	mtx_destroy(&V_bridge_list_mtx)
516 #define	BRIDGE_LIST_LOCK(x)		mtx_lock(&V_bridge_list_mtx)
517 #define	BRIDGE_LIST_UNLOCK(x)		mtx_unlock(&V_bridge_list_mtx)
518 
519 static VNET_DEFINE(struct if_clone *, bridge_cloner);
520 #define	V_bridge_cloner	VNET(bridge_cloner)
521 
522 static const char bridge_name[] = "bridge";
523 
524 static void
525 vnet_bridge_init(const void *unused __unused)
526 {
527 
528 	BRIDGE_LIST_LOCK_INIT();
529 	LIST_INIT(&V_bridge_list);
530 	V_bridge_cloner = if_clone_simple(bridge_name,
531 	    bridge_clone_create, bridge_clone_destroy, 0);
532 }
533 VNET_SYSINIT(vnet_bridge_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
534     vnet_bridge_init, NULL);
535 
536 static void
537 vnet_bridge_uninit(const void *unused __unused)
538 {
539 
540 	if_clone_detach(V_bridge_cloner);
541 	BRIDGE_LIST_LOCK_DESTROY();
542 }
543 VNET_SYSUNINIT(vnet_bridge_uninit, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
544     vnet_bridge_uninit, NULL);
545 
546 static int
547 bridge_modevent(module_t mod, int type, void *data)
548 {
549 
550 	switch (type) {
551 	case MOD_LOAD:
552 		bridge_rtnode_zone = uma_zcreate("bridge_rtnode",
553 		    sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL,
554 		    UMA_ALIGN_PTR, 0);
555 		bridge_input_p = bridge_input;
556 		bridge_output_p = bridge_output;
557 		bridge_dn_p = bridge_dummynet;
558 		bridge_linkstate_p = bridge_linkstate;
559 		bridge_detach_cookie = EVENTHANDLER_REGISTER(
560 		    ifnet_departure_event, bridge_ifdetach, NULL,
561 		    EVENTHANDLER_PRI_ANY);
562 		break;
563 	case MOD_UNLOAD:
564 		EVENTHANDLER_DEREGISTER(ifnet_departure_event,
565 		    bridge_detach_cookie);
566 		uma_zdestroy(bridge_rtnode_zone);
567 		bridge_input_p = NULL;
568 		bridge_output_p = NULL;
569 		bridge_dn_p = NULL;
570 		bridge_linkstate_p = NULL;
571 		break;
572 	default:
573 		return (EOPNOTSUPP);
574 	}
575 	return (0);
576 }
577 
578 static moduledata_t bridge_mod = {
579 	"if_bridge",
580 	bridge_modevent,
581 	0
582 };
583 
584 DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
585 MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1);
586 
587 /*
588  * handler for net.link.bridge.ipfw
589  */
590 static int
591 sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS)
592 {
593 	int enable = V_pfil_ipfw;
594 	int error;
595 
596 	error = sysctl_handle_int(oidp, &enable, 0, req);
597 	enable &= 1;
598 
599 	if (enable != V_pfil_ipfw) {
600 		V_pfil_ipfw = enable;
601 
602 		/*
603 		 * Disable pfil so that ipfw doesnt run twice, if the user
604 		 * really wants both then they can re-enable pfil_bridge and/or
605 		 * pfil_member. Also allow non-ip packets as ipfw can filter by
606 		 * layer2 type.
607 		 */
608 		if (V_pfil_ipfw) {
609 			V_pfil_onlyip = 0;
610 			V_pfil_bridge = 0;
611 			V_pfil_member = 0;
612 		}
613 	}
614 
615 	return (error);
616 }
617 SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw,
618     CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_VNET,
619     &VNET_NAME(pfil_ipfw), 0, &sysctl_pfil_ipfw, "I",
620     "Layer2 filter with IPFW");
621 
622 /*
623  * bridge_clone_create:
624  *
625  *	Create a new bridge instance.
626  */
627 static int
628 bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params)
629 {
630 	struct bridge_softc *sc, *sc2;
631 	struct ifnet *bifp, *ifp;
632 	int fb, retry;
633 	unsigned long hostid;
634 
635 	sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO);
636 	ifp = sc->sc_ifp = if_alloc(IFT_ETHER);
637 	if (ifp == NULL) {
638 		free(sc, M_DEVBUF);
639 		return (ENOSPC);
640 	}
641 
642 	BRIDGE_LOCK_INIT(sc);
643 	sc->sc_brtmax = BRIDGE_RTABLE_MAX;
644 	sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT;
645 
646 	/* Initialize our routing table. */
647 	bridge_rtable_init(sc);
648 
649 	callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0);
650 
651 	LIST_INIT(&sc->sc_iflist);
652 	LIST_INIT(&sc->sc_spanlist);
653 
654 	ifp->if_softc = sc;
655 	if_initname(ifp, bridge_name, unit);
656 	ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
657 	ifp->if_ioctl = bridge_ioctl;
658 	ifp->if_transmit = bridge_transmit;
659 	ifp->if_qflush = bridge_qflush;
660 	ifp->if_init = bridge_init;
661 	ifp->if_type = IFT_BRIDGE;
662 
663 	/*
664 	 * Generate an ethernet address with a locally administered address.
665 	 *
666 	 * Since we are using random ethernet addresses for the bridge, it is
667 	 * possible that we might have address collisions, so make sure that
668 	 * this hardware address isn't already in use on another bridge.
669 	 * The first try uses the hostid and falls back to arc4rand().
670 	 */
671 	fb = 0;
672 	getcredhostid(curthread->td_ucred, &hostid);
673 	do {
674 		if (fb || hostid == 0) {
675 			arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1);
676 			sc->sc_defaddr[0] &= ~1;/* clear multicast bit */
677 			sc->sc_defaddr[0] |= 2;	/* set the LAA bit */
678 		} else {
679 			sc->sc_defaddr[0] = 0x2;
680 			sc->sc_defaddr[1] = (hostid >> 24) & 0xff;
681 			sc->sc_defaddr[2] = (hostid >> 16) & 0xff;
682 			sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff;
683 			sc->sc_defaddr[4] =  hostid        & 0xff;
684 			sc->sc_defaddr[5] = ifp->if_dunit & 0xff;
685 		}
686 
687 		fb = 1;
688 		retry = 0;
689 		BRIDGE_LIST_LOCK();
690 		LIST_FOREACH(sc2, &V_bridge_list, sc_list) {
691 			bifp = sc2->sc_ifp;
692 			if (memcmp(sc->sc_defaddr,
693 			    IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) {
694 				retry = 1;
695 				break;
696 			}
697 		}
698 		BRIDGE_LIST_UNLOCK();
699 	} while (retry == 1);
700 
701 	bstp_attach(&sc->sc_stp, &bridge_ops);
702 	ether_ifattach(ifp, sc->sc_defaddr);
703 	/* Now undo some of the damage... */
704 	ifp->if_baudrate = 0;
705 	ifp->if_type = IFT_BRIDGE;
706 
707 	BRIDGE_LIST_LOCK();
708 	LIST_INSERT_HEAD(&V_bridge_list, sc, sc_list);
709 	BRIDGE_LIST_UNLOCK();
710 
711 	return (0);
712 }
713 
714 /*
715  * bridge_clone_destroy:
716  *
717  *	Destroy a bridge instance.
718  */
719 static void
720 bridge_clone_destroy(struct ifnet *ifp)
721 {
722 	struct bridge_softc *sc = ifp->if_softc;
723 	struct bridge_iflist *bif;
724 
725 	BRIDGE_LOCK(sc);
726 
727 	bridge_stop(ifp, 1);
728 	ifp->if_flags &= ~IFF_UP;
729 
730 	while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL)
731 		bridge_delete_member(sc, bif, 0);
732 
733 	while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) {
734 		bridge_delete_span(sc, bif);
735 	}
736 
737 	BRIDGE_UNLOCK(sc);
738 
739 	callout_drain(&sc->sc_brcallout);
740 
741 	BRIDGE_LIST_LOCK();
742 	LIST_REMOVE(sc, sc_list);
743 	BRIDGE_LIST_UNLOCK();
744 
745 	bstp_detach(&sc->sc_stp);
746 	ether_ifdetach(ifp);
747 	if_free(ifp);
748 
749 	/* Tear down the routing table. */
750 	bridge_rtable_fini(sc);
751 
752 	BRIDGE_LOCK_DESTROY(sc);
753 	free(sc, M_DEVBUF);
754 }
755 
756 /*
757  * bridge_ioctl:
758  *
759  *	Handle a control request from the operator.
760  */
761 static int
762 bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
763 {
764 	struct bridge_softc *sc = ifp->if_softc;
765 	struct ifreq *ifr = (struct ifreq *)data;
766 	struct bridge_iflist *bif;
767 	struct thread *td = curthread;
768 	union {
769 		struct ifbreq ifbreq;
770 		struct ifbifconf ifbifconf;
771 		struct ifbareq ifbareq;
772 		struct ifbaconf ifbaconf;
773 		struct ifbrparam ifbrparam;
774 		struct ifbropreq ifbropreq;
775 	} args;
776 	struct ifdrv *ifd = (struct ifdrv *) data;
777 	const struct bridge_control *bc;
778 	int error = 0;
779 
780 	switch (cmd) {
781 
782 	case SIOCADDMULTI:
783 	case SIOCDELMULTI:
784 		break;
785 
786 	case SIOCGDRVSPEC:
787 	case SIOCSDRVSPEC:
788 		if (ifd->ifd_cmd >= bridge_control_table_size) {
789 			error = EINVAL;
790 			break;
791 		}
792 		bc = &bridge_control_table[ifd->ifd_cmd];
793 
794 		if (cmd == SIOCGDRVSPEC &&
795 		    (bc->bc_flags & BC_F_COPYOUT) == 0) {
796 			error = EINVAL;
797 			break;
798 		}
799 		else if (cmd == SIOCSDRVSPEC &&
800 		    (bc->bc_flags & BC_F_COPYOUT) != 0) {
801 			error = EINVAL;
802 			break;
803 		}
804 
805 		if (bc->bc_flags & BC_F_SUSER) {
806 			error = priv_check(td, PRIV_NET_BRIDGE);
807 			if (error)
808 				break;
809 		}
810 
811 		if (ifd->ifd_len != bc->bc_argsize ||
812 		    ifd->ifd_len > sizeof(args)) {
813 			error = EINVAL;
814 			break;
815 		}
816 
817 		bzero(&args, sizeof(args));
818 		if (bc->bc_flags & BC_F_COPYIN) {
819 			error = copyin(ifd->ifd_data, &args, ifd->ifd_len);
820 			if (error)
821 				break;
822 		}
823 
824 		BRIDGE_LOCK(sc);
825 		error = (*bc->bc_func)(sc, &args);
826 		BRIDGE_UNLOCK(sc);
827 		if (error)
828 			break;
829 
830 		if (bc->bc_flags & BC_F_COPYOUT)
831 			error = copyout(&args, ifd->ifd_data, ifd->ifd_len);
832 
833 		break;
834 
835 	case SIOCSIFFLAGS:
836 		if (!(ifp->if_flags & IFF_UP) &&
837 		    (ifp->if_drv_flags & IFF_DRV_RUNNING)) {
838 			/*
839 			 * If interface is marked down and it is running,
840 			 * then stop and disable it.
841 			 */
842 			BRIDGE_LOCK(sc);
843 			bridge_stop(ifp, 1);
844 			BRIDGE_UNLOCK(sc);
845 		} else if ((ifp->if_flags & IFF_UP) &&
846 		    !(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
847 			/*
848 			 * If interface is marked up and it is stopped, then
849 			 * start it.
850 			 */
851 			(*ifp->if_init)(sc);
852 		}
853 		break;
854 
855 	case SIOCSIFMTU:
856 		if (ifr->ifr_mtu < 576) {
857 			error = EINVAL;
858 			break;
859 		}
860 		if (LIST_EMPTY(&sc->sc_iflist)) {
861 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
862 			break;
863 		}
864 		BRIDGE_LOCK(sc);
865 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
866 			if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) {
867 				log(LOG_NOTICE, "%s: invalid MTU: %u(%s)"
868 				    " != %d\n", sc->sc_ifp->if_xname,
869 				    bif->bif_ifp->if_mtu,
870 				    bif->bif_ifp->if_xname, ifr->ifr_mtu);
871 				error = EINVAL;
872 				break;
873 			}
874 		}
875 		if (!error)
876 			sc->sc_ifp->if_mtu = ifr->ifr_mtu;
877 		BRIDGE_UNLOCK(sc);
878 		break;
879 	default:
880 		/*
881 		 * drop the lock as ether_ioctl() will call bridge_start() and
882 		 * cause the lock to be recursed.
883 		 */
884 		error = ether_ioctl(ifp, cmd, data);
885 		break;
886 	}
887 
888 	return (error);
889 }
890 
891 /*
892  * bridge_mutecaps:
893  *
894  *	Clear or restore unwanted capabilities on the member interface
895  */
896 static void
897 bridge_mutecaps(struct bridge_softc *sc)
898 {
899 	struct bridge_iflist *bif;
900 	int enabled, mask;
901 
902 	/* Initial bitmask of capabilities to test */
903 	mask = BRIDGE_IFCAPS_MASK;
904 
905 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
906 		/* Every member must support it or its disabled */
907 		mask &= bif->bif_savedcaps;
908 	}
909 
910 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
911 		enabled = bif->bif_ifp->if_capenable;
912 		enabled &= ~BRIDGE_IFCAPS_STRIP;
913 		/* strip off mask bits and enable them again if allowed */
914 		enabled &= ~BRIDGE_IFCAPS_MASK;
915 		enabled |= mask;
916 		bridge_set_ifcap(sc, bif, enabled);
917 	}
918 
919 }
920 
921 static void
922 bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set)
923 {
924 	struct ifnet *ifp = bif->bif_ifp;
925 	struct ifreq ifr;
926 	int error;
927 
928 	bzero(&ifr, sizeof(ifr));
929 	ifr.ifr_reqcap = set;
930 
931 	if (ifp->if_capenable != set) {
932 		error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr);
933 		if (error)
934 			if_printf(sc->sc_ifp,
935 			    "error setting interface capabilities on %s\n",
936 			    ifp->if_xname);
937 	}
938 }
939 
940 /*
941  * bridge_lookup_member:
942  *
943  *	Lookup a bridge member interface.
944  */
945 static struct bridge_iflist *
946 bridge_lookup_member(struct bridge_softc *sc, const char *name)
947 {
948 	struct bridge_iflist *bif;
949 	struct ifnet *ifp;
950 
951 	BRIDGE_LOCK_ASSERT(sc);
952 
953 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
954 		ifp = bif->bif_ifp;
955 		if (strcmp(ifp->if_xname, name) == 0)
956 			return (bif);
957 	}
958 
959 	return (NULL);
960 }
961 
962 /*
963  * bridge_lookup_member_if:
964  *
965  *	Lookup a bridge member interface by ifnet*.
966  */
967 static struct bridge_iflist *
968 bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp)
969 {
970 	struct bridge_iflist *bif;
971 
972 	BRIDGE_LOCK_ASSERT(sc);
973 
974 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
975 		if (bif->bif_ifp == member_ifp)
976 			return (bif);
977 	}
978 
979 	return (NULL);
980 }
981 
982 /*
983  * bridge_delete_member:
984  *
985  *	Delete the specified member interface.
986  */
987 static void
988 bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif,
989     int gone)
990 {
991 	struct ifnet *ifs = bif->bif_ifp;
992 	struct ifnet *fif = NULL;
993 
994 	BRIDGE_LOCK_ASSERT(sc);
995 
996 	if (bif->bif_flags & IFBIF_STP)
997 		bstp_disable(&bif->bif_stp);
998 
999 	ifs->if_bridge = NULL;
1000 	BRIDGE_XLOCK(sc);
1001 	LIST_REMOVE(bif, bif_next);
1002 	BRIDGE_XDROP(sc);
1003 
1004 	/*
1005 	 * If removing the interface that gave the bridge its mac address, set
1006 	 * the mac address of the bridge to the address of the next member, or
1007 	 * to its default address if no members are left.
1008 	 */
1009 	if (V_bridge_inherit_mac && sc->sc_ifaddr == ifs) {
1010 		if (LIST_EMPTY(&sc->sc_iflist)) {
1011 			bcopy(sc->sc_defaddr,
1012 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1013 			sc->sc_ifaddr = NULL;
1014 		} else {
1015 			fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp;
1016 			bcopy(IF_LLADDR(fif),
1017 			    IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1018 			sc->sc_ifaddr = fif;
1019 		}
1020 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1021 	}
1022 
1023 	bridge_linkcheck(sc);
1024 	bridge_mutecaps(sc);	/* recalcuate now this interface is removed */
1025 	bridge_rtdelete(sc, ifs, IFBF_FLUSHALL);
1026 	KASSERT(bif->bif_addrcnt == 0,
1027 	    ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt));
1028 
1029 	BRIDGE_UNLOCK(sc);
1030 	if (!gone) {
1031 		switch (ifs->if_type) {
1032 		case IFT_ETHER:
1033 		case IFT_L2VLAN:
1034 			/*
1035 			 * Take the interface out of promiscuous mode.
1036 			 */
1037 			(void) ifpromisc(ifs, 0);
1038 			break;
1039 
1040 		case IFT_GIF:
1041 			break;
1042 
1043 		default:
1044 #ifdef DIAGNOSTIC
1045 			panic("bridge_delete_member: impossible");
1046 #endif
1047 			break;
1048 		}
1049 		/* reneable any interface capabilities */
1050 		bridge_set_ifcap(sc, bif, bif->bif_savedcaps);
1051 	}
1052 	bstp_destroy(&bif->bif_stp);	/* prepare to free */
1053 	BRIDGE_LOCK(sc);
1054 	free(bif, M_DEVBUF);
1055 }
1056 
1057 /*
1058  * bridge_delete_span:
1059  *
1060  *	Delete the specified span interface.
1061  */
1062 static void
1063 bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif)
1064 {
1065 	BRIDGE_LOCK_ASSERT(sc);
1066 
1067 	KASSERT(bif->bif_ifp->if_bridge == NULL,
1068 	    ("%s: not a span interface", __func__));
1069 
1070 	LIST_REMOVE(bif, bif_next);
1071 	free(bif, M_DEVBUF);
1072 }
1073 
1074 static int
1075 bridge_ioctl_add(struct bridge_softc *sc, void *arg)
1076 {
1077 	struct ifbreq *req = arg;
1078 	struct bridge_iflist *bif = NULL;
1079 	struct ifnet *ifs;
1080 	int error = 0;
1081 
1082 	ifs = ifunit(req->ifbr_ifsname);
1083 	if (ifs == NULL)
1084 		return (ENOENT);
1085 	if (ifs->if_ioctl == NULL)	/* must be supported */
1086 		return (EINVAL);
1087 
1088 	/* If it's in the span list, it can't be a member. */
1089 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1090 		if (ifs == bif->bif_ifp)
1091 			return (EBUSY);
1092 
1093 	if (ifs->if_bridge == sc)
1094 		return (EEXIST);
1095 
1096 	if (ifs->if_bridge != NULL)
1097 		return (EBUSY);
1098 
1099 	switch (ifs->if_type) {
1100 	case IFT_ETHER:
1101 	case IFT_L2VLAN:
1102 	case IFT_GIF:
1103 		/* permitted interface types */
1104 		break;
1105 	default:
1106 		return (EINVAL);
1107 	}
1108 
1109 #ifdef INET6
1110 	/*
1111 	 * Two valid inet6 addresses with link-local scope must not be
1112 	 * on the parent interface and the member interfaces at the
1113 	 * same time.  This restriction is needed to prevent violation
1114 	 * of link-local scope zone.  Attempts to add a member
1115 	 * interface which has inet6 addresses when the parent has
1116 	 * inet6 triggers removal of all inet6 addresses on the member
1117 	 * interface.
1118 	 */
1119 
1120 	/* Check if the parent interface has a link-local scope addr. */
1121 	if (V_allow_llz_overlap == 0 &&
1122 	    in6ifa_llaonifp(sc->sc_ifp) != NULL) {
1123 		/*
1124 		 * If any, remove all inet6 addresses from the member
1125 		 * interfaces.
1126 		 */
1127 		BRIDGE_XLOCK(sc);
1128 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1129  			if (in6ifa_llaonifp(bif->bif_ifp)) {
1130 				BRIDGE_UNLOCK(sc);
1131 				in6_ifdetach(bif->bif_ifp);
1132 				BRIDGE_LOCK(sc);
1133 				if_printf(sc->sc_ifp,
1134 				    "IPv6 addresses on %s have been removed "
1135 				    "before adding it as a member to prevent "
1136 				    "IPv6 address scope violation.\n",
1137 				    bif->bif_ifp->if_xname);
1138 			}
1139 		}
1140 		BRIDGE_XDROP(sc);
1141 		if (in6ifa_llaonifp(ifs)) {
1142 			BRIDGE_UNLOCK(sc);
1143 			in6_ifdetach(ifs);
1144 			BRIDGE_LOCK(sc);
1145 			if_printf(sc->sc_ifp,
1146 			    "IPv6 addresses on %s have been removed "
1147 			    "before adding it as a member to prevent "
1148 			    "IPv6 address scope violation.\n",
1149 			    ifs->if_xname);
1150 		}
1151 	}
1152 #endif
1153 	/* Allow the first Ethernet member to define the MTU */
1154 	if (LIST_EMPTY(&sc->sc_iflist))
1155 		sc->sc_ifp->if_mtu = ifs->if_mtu;
1156 	else if (sc->sc_ifp->if_mtu != ifs->if_mtu) {
1157 		if_printf(sc->sc_ifp, "invalid MTU: %u(%s) != %u\n",
1158 		    ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu);
1159 		return (EINVAL);
1160 	}
1161 
1162 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1163 	if (bif == NULL)
1164 		return (ENOMEM);
1165 
1166 	bif->bif_ifp = ifs;
1167 	bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER;
1168 	bif->bif_savedcaps = ifs->if_capenable;
1169 
1170 	/*
1171 	 * Assign the interface's MAC address to the bridge if it's the first
1172 	 * member and the MAC address of the bridge has not been changed from
1173 	 * the default randomly generated one.
1174 	 */
1175 	if (V_bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) &&
1176 	    !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) {
1177 		bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN);
1178 		sc->sc_ifaddr = ifs;
1179 		EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp);
1180 	}
1181 
1182 	ifs->if_bridge = sc;
1183 	bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp);
1184 	/*
1185 	 * XXX: XLOCK HERE!?!
1186 	 *
1187 	 * NOTE: insert_***HEAD*** should be safe for the traversals.
1188 	 */
1189 	LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next);
1190 
1191 	/* Set interface capabilities to the intersection set of all members */
1192 	bridge_mutecaps(sc);
1193 	bridge_linkcheck(sc);
1194 
1195 	/* Place the interface into promiscuous mode */
1196 	switch (ifs->if_type) {
1197 		case IFT_ETHER:
1198 		case IFT_L2VLAN:
1199 			BRIDGE_UNLOCK(sc);
1200 			error = ifpromisc(ifs, 1);
1201 			BRIDGE_LOCK(sc);
1202 			break;
1203 	}
1204 
1205 	if (error) {
1206 		bridge_delete_member(sc, bif, 0);
1207 		free(bif, M_DEVBUF);
1208 	}
1209 	return (error);
1210 }
1211 
1212 static int
1213 bridge_ioctl_del(struct bridge_softc *sc, void *arg)
1214 {
1215 	struct ifbreq *req = arg;
1216 	struct bridge_iflist *bif;
1217 
1218 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1219 	if (bif == NULL)
1220 		return (ENOENT);
1221 
1222 	bridge_delete_member(sc, bif, 0);
1223 
1224 	return (0);
1225 }
1226 
1227 static int
1228 bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg)
1229 {
1230 	struct ifbreq *req = arg;
1231 	struct bridge_iflist *bif;
1232 	struct bstp_port *bp;
1233 
1234 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1235 	if (bif == NULL)
1236 		return (ENOENT);
1237 
1238 	bp = &bif->bif_stp;
1239 	req->ifbr_ifsflags = bif->bif_flags;
1240 	req->ifbr_state = bp->bp_state;
1241 	req->ifbr_priority = bp->bp_priority;
1242 	req->ifbr_path_cost = bp->bp_path_cost;
1243 	req->ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1244 	req->ifbr_proto = bp->bp_protover;
1245 	req->ifbr_role = bp->bp_role;
1246 	req->ifbr_stpflags = bp->bp_flags;
1247 	req->ifbr_addrcnt = bif->bif_addrcnt;
1248 	req->ifbr_addrmax = bif->bif_addrmax;
1249 	req->ifbr_addrexceeded = bif->bif_addrexceeded;
1250 
1251 	/* Copy STP state options as flags */
1252 	if (bp->bp_operedge)
1253 		req->ifbr_ifsflags |= IFBIF_BSTP_EDGE;
1254 	if (bp->bp_flags & BSTP_PORT_AUTOEDGE)
1255 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE;
1256 	if (bp->bp_ptp_link)
1257 		req->ifbr_ifsflags |= IFBIF_BSTP_PTP;
1258 	if (bp->bp_flags & BSTP_PORT_AUTOPTP)
1259 		req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP;
1260 	if (bp->bp_flags & BSTP_PORT_ADMEDGE)
1261 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE;
1262 	if (bp->bp_flags & BSTP_PORT_ADMCOST)
1263 		req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST;
1264 	return (0);
1265 }
1266 
1267 static int
1268 bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg)
1269 {
1270 	struct ifbreq *req = arg;
1271 	struct bridge_iflist *bif;
1272 	struct bstp_port *bp;
1273 	int error;
1274 
1275 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1276 	if (bif == NULL)
1277 		return (ENOENT);
1278 	bp = &bif->bif_stp;
1279 
1280 	if (req->ifbr_ifsflags & IFBIF_SPAN)
1281 		/* SPAN is readonly */
1282 		return (EINVAL);
1283 
1284 	if (req->ifbr_ifsflags & IFBIF_STP) {
1285 		if ((bif->bif_flags & IFBIF_STP) == 0) {
1286 			error = bstp_enable(&bif->bif_stp);
1287 			if (error)
1288 				return (error);
1289 		}
1290 	} else {
1291 		if ((bif->bif_flags & IFBIF_STP) != 0)
1292 			bstp_disable(&bif->bif_stp);
1293 	}
1294 
1295 	/* Pass on STP flags */
1296 	bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0);
1297 	bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0);
1298 	bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0);
1299 	bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0);
1300 
1301 	/* Save the bits relating to the bridge */
1302 	bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK;
1303 
1304 	return (0);
1305 }
1306 
1307 static int
1308 bridge_ioctl_scache(struct bridge_softc *sc, void *arg)
1309 {
1310 	struct ifbrparam *param = arg;
1311 
1312 	sc->sc_brtmax = param->ifbrp_csize;
1313 	bridge_rttrim(sc);
1314 
1315 	return (0);
1316 }
1317 
1318 static int
1319 bridge_ioctl_gcache(struct bridge_softc *sc, void *arg)
1320 {
1321 	struct ifbrparam *param = arg;
1322 
1323 	param->ifbrp_csize = sc->sc_brtmax;
1324 
1325 	return (0);
1326 }
1327 
1328 static int
1329 bridge_ioctl_gifs(struct bridge_softc *sc, void *arg)
1330 {
1331 	struct ifbifconf *bifc = arg;
1332 	struct bridge_iflist *bif;
1333 	struct ifbreq breq;
1334 	char *buf, *outbuf;
1335 	int count, buflen, len, error = 0;
1336 
1337 	count = 0;
1338 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next)
1339 		count++;
1340 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1341 		count++;
1342 
1343 	buflen = sizeof(breq) * count;
1344 	if (bifc->ifbic_len == 0) {
1345 		bifc->ifbic_len = buflen;
1346 		return (0);
1347 	}
1348 	BRIDGE_UNLOCK(sc);
1349 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1350 	BRIDGE_LOCK(sc);
1351 
1352 	count = 0;
1353 	buf = outbuf;
1354 	len = min(bifc->ifbic_len, buflen);
1355 	bzero(&breq, sizeof(breq));
1356 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1357 		if (len < sizeof(breq))
1358 			break;
1359 
1360 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1361 		    sizeof(breq.ifbr_ifsname));
1362 		/* Fill in the ifbreq structure */
1363 		error = bridge_ioctl_gifflags(sc, &breq);
1364 		if (error)
1365 			break;
1366 		memcpy(buf, &breq, sizeof(breq));
1367 		count++;
1368 		buf += sizeof(breq);
1369 		len -= sizeof(breq);
1370 	}
1371 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
1372 		if (len < sizeof(breq))
1373 			break;
1374 
1375 		strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname,
1376 		    sizeof(breq.ifbr_ifsname));
1377 		breq.ifbr_ifsflags = bif->bif_flags;
1378 		breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff;
1379 		memcpy(buf, &breq, sizeof(breq));
1380 		count++;
1381 		buf += sizeof(breq);
1382 		len -= sizeof(breq);
1383 	}
1384 
1385 	BRIDGE_UNLOCK(sc);
1386 	bifc->ifbic_len = sizeof(breq) * count;
1387 	error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len);
1388 	BRIDGE_LOCK(sc);
1389 	free(outbuf, M_TEMP);
1390 	return (error);
1391 }
1392 
1393 static int
1394 bridge_ioctl_rts(struct bridge_softc *sc, void *arg)
1395 {
1396 	struct ifbaconf *bac = arg;
1397 	struct bridge_rtnode *brt;
1398 	struct ifbareq bareq;
1399 	char *buf, *outbuf;
1400 	int count, buflen, len, error = 0;
1401 
1402 	if (bac->ifbac_len == 0)
1403 		return (0);
1404 
1405 	count = 0;
1406 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list)
1407 		count++;
1408 	buflen = sizeof(bareq) * count;
1409 
1410 	BRIDGE_UNLOCK(sc);
1411 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1412 	BRIDGE_LOCK(sc);
1413 
1414 	count = 0;
1415 	buf = outbuf;
1416 	len = min(bac->ifbac_len, buflen);
1417 	bzero(&bareq, sizeof(bareq));
1418 	LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
1419 		if (len < sizeof(bareq))
1420 			goto out;
1421 		strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname,
1422 		    sizeof(bareq.ifba_ifsname));
1423 		memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr));
1424 		bareq.ifba_vlan = brt->brt_vlan;
1425 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
1426 				time_uptime < brt->brt_expire)
1427 			bareq.ifba_expire = brt->brt_expire - time_uptime;
1428 		else
1429 			bareq.ifba_expire = 0;
1430 		bareq.ifba_flags = brt->brt_flags;
1431 
1432 		memcpy(buf, &bareq, sizeof(bareq));
1433 		count++;
1434 		buf += sizeof(bareq);
1435 		len -= sizeof(bareq);
1436 	}
1437 out:
1438 	BRIDGE_UNLOCK(sc);
1439 	bac->ifbac_len = sizeof(bareq) * count;
1440 	error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len);
1441 	BRIDGE_LOCK(sc);
1442 	free(outbuf, M_TEMP);
1443 	return (error);
1444 }
1445 
1446 static int
1447 bridge_ioctl_saddr(struct bridge_softc *sc, void *arg)
1448 {
1449 	struct ifbareq *req = arg;
1450 	struct bridge_iflist *bif;
1451 	int error;
1452 
1453 	bif = bridge_lookup_member(sc, req->ifba_ifsname);
1454 	if (bif == NULL)
1455 		return (ENOENT);
1456 
1457 	error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1,
1458 	    req->ifba_flags);
1459 
1460 	return (error);
1461 }
1462 
1463 static int
1464 bridge_ioctl_sto(struct bridge_softc *sc, void *arg)
1465 {
1466 	struct ifbrparam *param = arg;
1467 
1468 	sc->sc_brttimeout = param->ifbrp_ctime;
1469 	return (0);
1470 }
1471 
1472 static int
1473 bridge_ioctl_gto(struct bridge_softc *sc, void *arg)
1474 {
1475 	struct ifbrparam *param = arg;
1476 
1477 	param->ifbrp_ctime = sc->sc_brttimeout;
1478 	return (0);
1479 }
1480 
1481 static int
1482 bridge_ioctl_daddr(struct bridge_softc *sc, void *arg)
1483 {
1484 	struct ifbareq *req = arg;
1485 
1486 	return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan));
1487 }
1488 
1489 static int
1490 bridge_ioctl_flush(struct bridge_softc *sc, void *arg)
1491 {
1492 	struct ifbreq *req = arg;
1493 
1494 	bridge_rtflush(sc, req->ifbr_ifsflags);
1495 	return (0);
1496 }
1497 
1498 static int
1499 bridge_ioctl_gpri(struct bridge_softc *sc, void *arg)
1500 {
1501 	struct ifbrparam *param = arg;
1502 	struct bstp_state *bs = &sc->sc_stp;
1503 
1504 	param->ifbrp_prio = bs->bs_bridge_priority;
1505 	return (0);
1506 }
1507 
1508 static int
1509 bridge_ioctl_spri(struct bridge_softc *sc, void *arg)
1510 {
1511 	struct ifbrparam *param = arg;
1512 
1513 	return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio));
1514 }
1515 
1516 static int
1517 bridge_ioctl_ght(struct bridge_softc *sc, void *arg)
1518 {
1519 	struct ifbrparam *param = arg;
1520 	struct bstp_state *bs = &sc->sc_stp;
1521 
1522 	param->ifbrp_hellotime = bs->bs_bridge_htime >> 8;
1523 	return (0);
1524 }
1525 
1526 static int
1527 bridge_ioctl_sht(struct bridge_softc *sc, void *arg)
1528 {
1529 	struct ifbrparam *param = arg;
1530 
1531 	return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime));
1532 }
1533 
1534 static int
1535 bridge_ioctl_gfd(struct bridge_softc *sc, void *arg)
1536 {
1537 	struct ifbrparam *param = arg;
1538 	struct bstp_state *bs = &sc->sc_stp;
1539 
1540 	param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8;
1541 	return (0);
1542 }
1543 
1544 static int
1545 bridge_ioctl_sfd(struct bridge_softc *sc, void *arg)
1546 {
1547 	struct ifbrparam *param = arg;
1548 
1549 	return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay));
1550 }
1551 
1552 static int
1553 bridge_ioctl_gma(struct bridge_softc *sc, void *arg)
1554 {
1555 	struct ifbrparam *param = arg;
1556 	struct bstp_state *bs = &sc->sc_stp;
1557 
1558 	param->ifbrp_maxage = bs->bs_bridge_max_age >> 8;
1559 	return (0);
1560 }
1561 
1562 static int
1563 bridge_ioctl_sma(struct bridge_softc *sc, void *arg)
1564 {
1565 	struct ifbrparam *param = arg;
1566 
1567 	return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage));
1568 }
1569 
1570 static int
1571 bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg)
1572 {
1573 	struct ifbreq *req = arg;
1574 	struct bridge_iflist *bif;
1575 
1576 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1577 	if (bif == NULL)
1578 		return (ENOENT);
1579 
1580 	return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority));
1581 }
1582 
1583 static int
1584 bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg)
1585 {
1586 	struct ifbreq *req = arg;
1587 	struct bridge_iflist *bif;
1588 
1589 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1590 	if (bif == NULL)
1591 		return (ENOENT);
1592 
1593 	return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost));
1594 }
1595 
1596 static int
1597 bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg)
1598 {
1599 	struct ifbreq *req = arg;
1600 	struct bridge_iflist *bif;
1601 
1602 	bif = bridge_lookup_member(sc, req->ifbr_ifsname);
1603 	if (bif == NULL)
1604 		return (ENOENT);
1605 
1606 	bif->bif_addrmax = req->ifbr_addrmax;
1607 	return (0);
1608 }
1609 
1610 static int
1611 bridge_ioctl_addspan(struct bridge_softc *sc, void *arg)
1612 {
1613 	struct ifbreq *req = arg;
1614 	struct bridge_iflist *bif = NULL;
1615 	struct ifnet *ifs;
1616 
1617 	ifs = ifunit(req->ifbr_ifsname);
1618 	if (ifs == NULL)
1619 		return (ENOENT);
1620 
1621 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1622 		if (ifs == bif->bif_ifp)
1623 			return (EBUSY);
1624 
1625 	if (ifs->if_bridge != NULL)
1626 		return (EBUSY);
1627 
1628 	switch (ifs->if_type) {
1629 		case IFT_ETHER:
1630 		case IFT_GIF:
1631 		case IFT_L2VLAN:
1632 			break;
1633 		default:
1634 			return (EINVAL);
1635 	}
1636 
1637 	bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO);
1638 	if (bif == NULL)
1639 		return (ENOMEM);
1640 
1641 	bif->bif_ifp = ifs;
1642 	bif->bif_flags = IFBIF_SPAN;
1643 
1644 	LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next);
1645 
1646 	return (0);
1647 }
1648 
1649 static int
1650 bridge_ioctl_delspan(struct bridge_softc *sc, void *arg)
1651 {
1652 	struct ifbreq *req = arg;
1653 	struct bridge_iflist *bif;
1654 	struct ifnet *ifs;
1655 
1656 	ifs = ifunit(req->ifbr_ifsname);
1657 	if (ifs == NULL)
1658 		return (ENOENT);
1659 
1660 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1661 		if (ifs == bif->bif_ifp)
1662 			break;
1663 
1664 	if (bif == NULL)
1665 		return (ENOENT);
1666 
1667 	bridge_delete_span(sc, bif);
1668 
1669 	return (0);
1670 }
1671 
1672 static int
1673 bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg)
1674 {
1675 	struct ifbropreq *req = arg;
1676 	struct bstp_state *bs = &sc->sc_stp;
1677 	struct bstp_port *root_port;
1678 
1679 	req->ifbop_maxage = bs->bs_bridge_max_age >> 8;
1680 	req->ifbop_hellotime = bs->bs_bridge_htime >> 8;
1681 	req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8;
1682 
1683 	root_port = bs->bs_root_port;
1684 	if (root_port == NULL)
1685 		req->ifbop_root_port = 0;
1686 	else
1687 		req->ifbop_root_port = root_port->bp_ifp->if_index;
1688 
1689 	req->ifbop_holdcount = bs->bs_txholdcount;
1690 	req->ifbop_priority = bs->bs_bridge_priority;
1691 	req->ifbop_protocol = bs->bs_protover;
1692 	req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost;
1693 	req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id;
1694 	req->ifbop_designated_root = bs->bs_root_pv.pv_root_id;
1695 	req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id;
1696 	req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec;
1697 	req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec;
1698 
1699 	return (0);
1700 }
1701 
1702 static int
1703 bridge_ioctl_grte(struct bridge_softc *sc, void *arg)
1704 {
1705 	struct ifbrparam *param = arg;
1706 
1707 	param->ifbrp_cexceeded = sc->sc_brtexceeded;
1708 	return (0);
1709 }
1710 
1711 static int
1712 bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg)
1713 {
1714 	struct ifbpstpconf *bifstp = arg;
1715 	struct bridge_iflist *bif;
1716 	struct bstp_port *bp;
1717 	struct ifbpstpreq bpreq;
1718 	char *buf, *outbuf;
1719 	int count, buflen, len, error = 0;
1720 
1721 	count = 0;
1722 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1723 		if ((bif->bif_flags & IFBIF_STP) != 0)
1724 			count++;
1725 	}
1726 
1727 	buflen = sizeof(bpreq) * count;
1728 	if (bifstp->ifbpstp_len == 0) {
1729 		bifstp->ifbpstp_len = buflen;
1730 		return (0);
1731 	}
1732 
1733 	BRIDGE_UNLOCK(sc);
1734 	outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO);
1735 	BRIDGE_LOCK(sc);
1736 
1737 	count = 0;
1738 	buf = outbuf;
1739 	len = min(bifstp->ifbpstp_len, buflen);
1740 	bzero(&bpreq, sizeof(bpreq));
1741 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
1742 		if (len < sizeof(bpreq))
1743 			break;
1744 
1745 		if ((bif->bif_flags & IFBIF_STP) == 0)
1746 			continue;
1747 
1748 		bp = &bif->bif_stp;
1749 		bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff;
1750 		bpreq.ifbp_fwd_trans = bp->bp_forward_transitions;
1751 		bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost;
1752 		bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id;
1753 		bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id;
1754 		bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id;
1755 
1756 		memcpy(buf, &bpreq, sizeof(bpreq));
1757 		count++;
1758 		buf += sizeof(bpreq);
1759 		len -= sizeof(bpreq);
1760 	}
1761 
1762 	BRIDGE_UNLOCK(sc);
1763 	bifstp->ifbpstp_len = sizeof(bpreq) * count;
1764 	error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len);
1765 	BRIDGE_LOCK(sc);
1766 	free(outbuf, M_TEMP);
1767 	return (error);
1768 }
1769 
1770 static int
1771 bridge_ioctl_sproto(struct bridge_softc *sc, void *arg)
1772 {
1773 	struct ifbrparam *param = arg;
1774 
1775 	return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto));
1776 }
1777 
1778 static int
1779 bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg)
1780 {
1781 	struct ifbrparam *param = arg;
1782 
1783 	return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc));
1784 }
1785 
1786 /*
1787  * bridge_ifdetach:
1788  *
1789  *	Detach an interface from a bridge.  Called when a member
1790  *	interface is detaching.
1791  */
1792 static void
1793 bridge_ifdetach(void *arg __unused, struct ifnet *ifp)
1794 {
1795 	struct bridge_softc *sc = ifp->if_bridge;
1796 	struct bridge_iflist *bif;
1797 
1798 	if (ifp->if_flags & IFF_RENAMING)
1799 		return;
1800 
1801 	/* Check if the interface is a bridge member */
1802 	if (sc != NULL) {
1803 		BRIDGE_LOCK(sc);
1804 
1805 		bif = bridge_lookup_member_if(sc, ifp);
1806 		if (bif != NULL)
1807 			bridge_delete_member(sc, bif, 1);
1808 
1809 		BRIDGE_UNLOCK(sc);
1810 		return;
1811 	}
1812 
1813 	/* Check if the interface is a span port */
1814 	BRIDGE_LIST_LOCK();
1815 	LIST_FOREACH(sc, &V_bridge_list, sc_list) {
1816 		BRIDGE_LOCK(sc);
1817 		LIST_FOREACH(bif, &sc->sc_spanlist, bif_next)
1818 			if (ifp == bif->bif_ifp) {
1819 				bridge_delete_span(sc, bif);
1820 				break;
1821 			}
1822 
1823 		BRIDGE_UNLOCK(sc);
1824 	}
1825 	BRIDGE_LIST_UNLOCK();
1826 }
1827 
1828 /*
1829  * bridge_init:
1830  *
1831  *	Initialize a bridge interface.
1832  */
1833 static void
1834 bridge_init(void *xsc)
1835 {
1836 	struct bridge_softc *sc = (struct bridge_softc *)xsc;
1837 	struct ifnet *ifp = sc->sc_ifp;
1838 
1839 	if (ifp->if_drv_flags & IFF_DRV_RUNNING)
1840 		return;
1841 
1842 	BRIDGE_LOCK(sc);
1843 	callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz,
1844 	    bridge_timer, sc);
1845 
1846 	ifp->if_drv_flags |= IFF_DRV_RUNNING;
1847 	bstp_init(&sc->sc_stp);		/* Initialize Spanning Tree */
1848 
1849 	BRIDGE_UNLOCK(sc);
1850 }
1851 
1852 /*
1853  * bridge_stop:
1854  *
1855  *	Stop the bridge interface.
1856  */
1857 static void
1858 bridge_stop(struct ifnet *ifp, int disable)
1859 {
1860 	struct bridge_softc *sc = ifp->if_softc;
1861 
1862 	BRIDGE_LOCK_ASSERT(sc);
1863 
1864 	if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
1865 		return;
1866 
1867 	callout_stop(&sc->sc_brcallout);
1868 	bstp_stop(&sc->sc_stp);
1869 
1870 	bridge_rtflush(sc, IFBF_FLUSHDYN);
1871 
1872 	ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
1873 }
1874 
1875 /*
1876  * bridge_enqueue:
1877  *
1878  *	Enqueue a packet on a bridge member interface.
1879  *
1880  */
1881 static int
1882 bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m)
1883 {
1884 	int len, err = 0;
1885 	short mflags;
1886 	struct mbuf *m0;
1887 
1888 	/* We may be sending a fragment so traverse the mbuf */
1889 	for (; m; m = m0) {
1890 		m0 = m->m_nextpkt;
1891 		m->m_nextpkt = NULL;
1892 		len = m->m_pkthdr.len;
1893 		mflags = m->m_flags;
1894 
1895 		/*
1896 		 * If underlying interface can not do VLAN tag insertion itself
1897 		 * then attach a packet tag that holds it.
1898 		 */
1899 		if ((m->m_flags & M_VLANTAG) &&
1900 		    (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) {
1901 			m = ether_vlanencap(m, m->m_pkthdr.ether_vtag);
1902 			if (m == NULL) {
1903 				if_printf(dst_ifp,
1904 				    "unable to prepend VLAN header\n");
1905 				if_inc_counter(dst_ifp, IFCOUNTER_OERRORS, 1);
1906 				continue;
1907 			}
1908 			m->m_flags &= ~M_VLANTAG;
1909 		}
1910 
1911 		if ((err = dst_ifp->if_transmit(dst_ifp, m))) {
1912 			m_freem(m0);
1913 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
1914 			break;
1915 		}
1916 
1917 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1);
1918 		if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, len);
1919 		if (mflags & M_MCAST)
1920 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OMCASTS, 1);
1921 	}
1922 
1923 	return (err);
1924 }
1925 
1926 /*
1927  * bridge_dummynet:
1928  *
1929  * 	Receive a queued packet from dummynet and pass it on to the output
1930  * 	interface.
1931  *
1932  *	The mbuf has the Ethernet header already attached.
1933  */
1934 static void
1935 bridge_dummynet(struct mbuf *m, struct ifnet *ifp)
1936 {
1937 	struct bridge_softc *sc;
1938 
1939 	sc = ifp->if_bridge;
1940 
1941 	/*
1942 	 * The packet didnt originate from a member interface. This should only
1943 	 * ever happen if a member interface is removed while packets are
1944 	 * queued for it.
1945 	 */
1946 	if (sc == NULL) {
1947 		m_freem(m);
1948 		return;
1949 	}
1950 
1951 	if (PFIL_HOOKED(&V_inet_pfil_hook)
1952 #ifdef INET6
1953 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
1954 #endif
1955 	    ) {
1956 		if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0)
1957 			return;
1958 		if (m == NULL)
1959 			return;
1960 	}
1961 
1962 	bridge_enqueue(sc, ifp, m);
1963 }
1964 
1965 /*
1966  * bridge_output:
1967  *
1968  *	Send output from a bridge member interface.  This
1969  *	performs the bridging function for locally originated
1970  *	packets.
1971  *
1972  *	The mbuf has the Ethernet header already attached.  We must
1973  *	enqueue or free the mbuf before returning.
1974  */
1975 static int
1976 bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa,
1977     struct rtentry *rt)
1978 {
1979 	struct ether_header *eh;
1980 	struct ifnet *dst_if;
1981 	struct bridge_softc *sc;
1982 	uint16_t vlan;
1983 
1984 	if (m->m_len < ETHER_HDR_LEN) {
1985 		m = m_pullup(m, ETHER_HDR_LEN);
1986 		if (m == NULL)
1987 			return (0);
1988 	}
1989 
1990 	eh = mtod(m, struct ether_header *);
1991 	sc = ifp->if_bridge;
1992 	vlan = VLANTAGOF(m);
1993 
1994 	BRIDGE_LOCK(sc);
1995 
1996 	/*
1997 	 * If bridge is down, but the original output interface is up,
1998 	 * go ahead and send out that interface.  Otherwise, the packet
1999 	 * is dropped below.
2000 	 */
2001 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2002 		dst_if = ifp;
2003 		goto sendunicast;
2004 	}
2005 
2006 	/*
2007 	 * If the packet is a multicast, or we don't know a better way to
2008 	 * get there, send to all interfaces.
2009 	 */
2010 	if (ETHER_IS_MULTICAST(eh->ether_dhost))
2011 		dst_if = NULL;
2012 	else
2013 		dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan);
2014 	if (dst_if == NULL) {
2015 		struct bridge_iflist *bif;
2016 		struct mbuf *mc;
2017 		int error = 0, used = 0;
2018 
2019 		bridge_span(sc, m);
2020 
2021 		BRIDGE_LOCK2REF(sc, error);
2022 		if (error) {
2023 			m_freem(m);
2024 			return (0);
2025 		}
2026 
2027 		LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
2028 			dst_if = bif->bif_ifp;
2029 
2030 			if (dst_if->if_type == IFT_GIF)
2031 				continue;
2032 			if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2033 				continue;
2034 
2035 			/*
2036 			 * If this is not the original output interface,
2037 			 * and the interface is participating in spanning
2038 			 * tree, make sure the port is in a state that
2039 			 * allows forwarding.
2040 			 */
2041 			if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) &&
2042 			    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2043 				continue;
2044 
2045 			if (LIST_NEXT(bif, bif_next) == NULL) {
2046 				used = 1;
2047 				mc = m;
2048 			} else {
2049 				mc = m_copypacket(m, M_NOWAIT);
2050 				if (mc == NULL) {
2051 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2052 					continue;
2053 				}
2054 			}
2055 
2056 			bridge_enqueue(sc, dst_if, mc);
2057 		}
2058 		if (used == 0)
2059 			m_freem(m);
2060 		BRIDGE_UNREF(sc);
2061 		return (0);
2062 	}
2063 
2064 sendunicast:
2065 	/*
2066 	 * XXX Spanning tree consideration here?
2067 	 */
2068 
2069 	bridge_span(sc, m);
2070 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) {
2071 		m_freem(m);
2072 		BRIDGE_UNLOCK(sc);
2073 		return (0);
2074 	}
2075 
2076 	BRIDGE_UNLOCK(sc);
2077 	bridge_enqueue(sc, dst_if, m);
2078 	return (0);
2079 }
2080 
2081 /*
2082  * bridge_transmit:
2083  *
2084  *	Do output on a bridge.
2085  *
2086  */
2087 static int
2088 bridge_transmit(struct ifnet *ifp, struct mbuf *m)
2089 {
2090 	struct bridge_softc *sc;
2091 	struct ether_header *eh;
2092 	struct ifnet *dst_if;
2093 	int error = 0;
2094 
2095 	sc = ifp->if_softc;
2096 
2097 	ETHER_BPF_MTAP(ifp, m);
2098 
2099 	eh = mtod(m, struct ether_header *);
2100 
2101 	BRIDGE_LOCK(sc);
2102 	if (((m->m_flags & (M_BCAST|M_MCAST)) == 0) &&
2103 	    (dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1)) != NULL) {
2104 		BRIDGE_UNLOCK(sc);
2105 		error = bridge_enqueue(sc, dst_if, m);
2106 	} else
2107 		bridge_broadcast(sc, ifp, m, 0);
2108 
2109 	return (error);
2110 }
2111 
2112 /*
2113  * The ifp->if_qflush entry point for if_bridge(4) is no-op.
2114  */
2115 static void
2116 bridge_qflush(struct ifnet *ifp __unused)
2117 {
2118 }
2119 
2120 /*
2121  * bridge_forward:
2122  *
2123  *	The forwarding function of the bridge.
2124  *
2125  *	NOTE: Releases the lock on return.
2126  */
2127 static void
2128 bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif,
2129     struct mbuf *m)
2130 {
2131 	struct bridge_iflist *dbif;
2132 	struct ifnet *src_if, *dst_if, *ifp;
2133 	struct ether_header *eh;
2134 	uint16_t vlan;
2135 	uint8_t *dst;
2136 	int error;
2137 
2138 	src_if = m->m_pkthdr.rcvif;
2139 	ifp = sc->sc_ifp;
2140 
2141 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
2142 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2143 	vlan = VLANTAGOF(m);
2144 
2145 	if ((sbif->bif_flags & IFBIF_STP) &&
2146 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2147 		goto drop;
2148 
2149 	eh = mtod(m, struct ether_header *);
2150 	dst = eh->ether_dhost;
2151 
2152 	/* If the interface is learning, record the address. */
2153 	if (sbif->bif_flags & IFBIF_LEARNING) {
2154 		error = bridge_rtupdate(sc, eh->ether_shost, vlan,
2155 		    sbif, 0, IFBAF_DYNAMIC);
2156 		/*
2157 		 * If the interface has addresses limits then deny any source
2158 		 * that is not in the cache.
2159 		 */
2160 		if (error && sbif->bif_addrmax)
2161 			goto drop;
2162 	}
2163 
2164 	if ((sbif->bif_flags & IFBIF_STP) != 0 &&
2165 	    sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING)
2166 		goto drop;
2167 
2168 	/*
2169 	 * At this point, the port either doesn't participate
2170 	 * in spanning tree or it is in the forwarding state.
2171 	 */
2172 
2173 	/*
2174 	 * If the packet is unicast, destined for someone on
2175 	 * "this" side of the bridge, drop it.
2176 	 */
2177 	if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) {
2178 		dst_if = bridge_rtlookup(sc, dst, vlan);
2179 		if (src_if == dst_if)
2180 			goto drop;
2181 	} else {
2182 		/*
2183 		 * Check if its a reserved multicast address, any address
2184 		 * listed in 802.1D section 7.12.6 may not be forwarded by the
2185 		 * bridge.
2186 		 * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F
2187 		 */
2188 		if (dst[0] == 0x01 && dst[1] == 0x80 &&
2189 		    dst[2] == 0xc2 && dst[3] == 0x00 &&
2190 		    dst[4] == 0x00 && dst[5] <= 0x0f)
2191 			goto drop;
2192 
2193 		/* ...forward it to all interfaces. */
2194 		if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
2195 		dst_if = NULL;
2196 	}
2197 
2198 	/*
2199 	 * If we have a destination interface which is a member of our bridge,
2200 	 * OR this is a unicast packet, push it through the bpf(4) machinery.
2201 	 * For broadcast or multicast packets, don't bother because it will
2202 	 * be reinjected into ether_input. We do this before we pass the packets
2203 	 * through the pfil(9) framework, as it is possible that pfil(9) will
2204 	 * drop the packet, or possibly modify it, making it difficult to debug
2205 	 * firewall issues on the bridge.
2206 	 */
2207 	if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0)
2208 		ETHER_BPF_MTAP(ifp, m);
2209 
2210 	/* run the packet filter */
2211 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2212 #ifdef INET6
2213 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2214 #endif
2215 	    ) {
2216 		BRIDGE_UNLOCK(sc);
2217 		if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0)
2218 			return;
2219 		if (m == NULL)
2220 			return;
2221 		BRIDGE_LOCK(sc);
2222 	}
2223 
2224 	if (dst_if == NULL) {
2225 		bridge_broadcast(sc, src_if, m, 1);
2226 		return;
2227 	}
2228 
2229 	/*
2230 	 * At this point, we're dealing with a unicast frame
2231 	 * going to a different interface.
2232 	 */
2233 	if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2234 		goto drop;
2235 
2236 	dbif = bridge_lookup_member_if(sc, dst_if);
2237 	if (dbif == NULL)
2238 		/* Not a member of the bridge (anymore?) */
2239 		goto drop;
2240 
2241 	/* Private segments can not talk to each other */
2242 	if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)
2243 		goto drop;
2244 
2245 	if ((dbif->bif_flags & IFBIF_STP) &&
2246 	    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2247 		goto drop;
2248 
2249 	BRIDGE_UNLOCK(sc);
2250 
2251 	if (PFIL_HOOKED(&V_inet_pfil_hook)
2252 #ifdef INET6
2253 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2254 #endif
2255 	    ) {
2256 		if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0)
2257 			return;
2258 		if (m == NULL)
2259 			return;
2260 	}
2261 
2262 	bridge_enqueue(sc, dst_if, m);
2263 	return;
2264 
2265 drop:
2266 	BRIDGE_UNLOCK(sc);
2267 	m_freem(m);
2268 }
2269 
2270 /*
2271  * bridge_input:
2272  *
2273  *	Receive input from a member interface.  Queue the packet for
2274  *	bridging if it is not for us.
2275  */
2276 static struct mbuf *
2277 bridge_input(struct ifnet *ifp, struct mbuf *m)
2278 {
2279 	struct bridge_softc *sc = ifp->if_bridge;
2280 	struct bridge_iflist *bif, *bif2;
2281 	struct ifnet *bifp;
2282 	struct ether_header *eh;
2283 	struct mbuf *mc, *mc2;
2284 	uint16_t vlan;
2285 	int error;
2286 
2287 	if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
2288 		return (m);
2289 
2290 	bifp = sc->sc_ifp;
2291 	vlan = VLANTAGOF(m);
2292 
2293 	/*
2294 	 * Implement support for bridge monitoring. If this flag has been
2295 	 * set on this interface, discard the packet once we push it through
2296 	 * the bpf(4) machinery, but before we do, increment the byte and
2297 	 * packet counters associated with this interface.
2298 	 */
2299 	if ((bifp->if_flags & IFF_MONITOR) != 0) {
2300 		m->m_pkthdr.rcvif  = bifp;
2301 		ETHER_BPF_MTAP(bifp, m);
2302 		if_inc_counter(bifp, IFCOUNTER_IPACKETS, 1);
2303 		if_inc_counter(bifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
2304 		m_freem(m);
2305 		return (NULL);
2306 	}
2307 	BRIDGE_LOCK(sc);
2308 	bif = bridge_lookup_member_if(sc, ifp);
2309 	if (bif == NULL) {
2310 		BRIDGE_UNLOCK(sc);
2311 		return (m);
2312 	}
2313 
2314 	eh = mtod(m, struct ether_header *);
2315 
2316 	bridge_span(sc, m);
2317 
2318 	if (m->m_flags & (M_BCAST|M_MCAST)) {
2319 		/* Tap off 802.1D packets; they do not get forwarded. */
2320 		if (memcmp(eh->ether_dhost, bstp_etheraddr,
2321 		    ETHER_ADDR_LEN) == 0) {
2322 			bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */
2323 			BRIDGE_UNLOCK(sc);
2324 			return (NULL);
2325 		}
2326 
2327 		if ((bif->bif_flags & IFBIF_STP) &&
2328 		    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2329 			BRIDGE_UNLOCK(sc);
2330 			return (m);
2331 		}
2332 
2333 		/*
2334 		 * Make a deep copy of the packet and enqueue the copy
2335 		 * for bridge processing; return the original packet for
2336 		 * local processing.
2337 		 */
2338 		mc = m_dup(m, M_NOWAIT);
2339 		if (mc == NULL) {
2340 			BRIDGE_UNLOCK(sc);
2341 			return (m);
2342 		}
2343 
2344 		/* Perform the bridge forwarding function with the copy. */
2345 		bridge_forward(sc, bif, mc);
2346 
2347 		/*
2348 		 * Reinject the mbuf as arriving on the bridge so we have a
2349 		 * chance at claiming multicast packets. We can not loop back
2350 		 * here from ether_input as a bridge is never a member of a
2351 		 * bridge.
2352 		 */
2353 		KASSERT(bifp->if_bridge == NULL,
2354 		    ("loop created in bridge_input"));
2355 		mc2 = m_dup(m, M_NOWAIT);
2356 		if (mc2 != NULL) {
2357 			/* Keep the layer3 header aligned */
2358 			int i = min(mc2->m_pkthdr.len, max_protohdr);
2359 			mc2 = m_copyup(mc2, i, ETHER_ALIGN);
2360 		}
2361 		if (mc2 != NULL) {
2362 			mc2->m_pkthdr.rcvif = bifp;
2363 			(*bifp->if_input)(bifp, mc2);
2364 		}
2365 
2366 		/* Return the original packet for local processing. */
2367 		return (m);
2368 	}
2369 
2370 	if ((bif->bif_flags & IFBIF_STP) &&
2371 	    bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) {
2372 		BRIDGE_UNLOCK(sc);
2373 		return (m);
2374 	}
2375 
2376 #if (defined(INET) || defined(INET6))
2377 #   define OR_CARP_CHECK_WE_ARE_DST(iface) \
2378 	|| ((iface)->if_carp \
2379 	    && (*carp_forus_p)((iface), eh->ether_dhost))
2380 #   define OR_CARP_CHECK_WE_ARE_SRC(iface) \
2381 	|| ((iface)->if_carp \
2382 	    && (*carp_forus_p)((iface), eh->ether_shost))
2383 #else
2384 #   define OR_CARP_CHECK_WE_ARE_DST(iface)
2385 #   define OR_CARP_CHECK_WE_ARE_SRC(iface)
2386 #endif
2387 
2388 #ifdef INET6
2389 #   define OR_PFIL_HOOKED_INET6 \
2390 	|| PFIL_HOOKED(&V_inet6_pfil_hook)
2391 #else
2392 #   define OR_PFIL_HOOKED_INET6
2393 #endif
2394 
2395 #define GRAB_OUR_PACKETS(iface) \
2396 	if ((iface)->if_type == IFT_GIF) \
2397 		continue; \
2398 	/* It is destined for us. */ \
2399 	if (memcmp(IF_LLADDR((iface)), eh->ether_dhost,  ETHER_ADDR_LEN) == 0 \
2400 	    OR_CARP_CHECK_WE_ARE_DST((iface))				\
2401 	    ) {								\
2402 		if ((iface)->if_type == IFT_BRIDGE) {			\
2403 			ETHER_BPF_MTAP(iface, m);			\
2404 			if_inc_counter(iface, IFCOUNTER_IPACKETS, 1);				\
2405 			if_inc_counter(iface, IFCOUNTER_IBYTES, m->m_pkthdr.len);		\
2406 			/* Filter on the physical interface. */		\
2407 			if (V_pfil_local_phys &&			\
2408 			    (PFIL_HOOKED(&V_inet_pfil_hook)		\
2409 			     OR_PFIL_HOOKED_INET6)) {			\
2410 				if (bridge_pfil(&m, NULL, ifp,		\
2411 				    PFIL_IN) != 0 || m == NULL) {	\
2412 					BRIDGE_UNLOCK(sc);		\
2413 					return (NULL);			\
2414 				}					\
2415 				eh = mtod(m, struct ether_header *);	\
2416 			}						\
2417 		}							\
2418 		if (bif->bif_flags & IFBIF_LEARNING) {			\
2419 			error = bridge_rtupdate(sc, eh->ether_shost,	\
2420 			    vlan, bif, 0, IFBAF_DYNAMIC);		\
2421 			if (error && bif->bif_addrmax) {		\
2422 				BRIDGE_UNLOCK(sc);			\
2423 				m_freem(m);				\
2424 				return (NULL);				\
2425 			}						\
2426 		}							\
2427 		m->m_pkthdr.rcvif = iface;				\
2428 		BRIDGE_UNLOCK(sc);					\
2429 		return (m);						\
2430 	}								\
2431 									\
2432 	/* We just received a packet that we sent out. */		\
2433 	if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \
2434 	    OR_CARP_CHECK_WE_ARE_SRC((iface))			\
2435 	    ) {								\
2436 		BRIDGE_UNLOCK(sc);					\
2437 		m_freem(m);						\
2438 		return (NULL);						\
2439 	}
2440 
2441 	/*
2442 	 * Unicast.  Make sure it's not for the bridge.
2443 	 */
2444 	do { GRAB_OUR_PACKETS(bifp) } while (0);
2445 
2446 	/*
2447 	 * Give a chance for ifp at first priority. This will help when	the
2448 	 * packet comes through the interface like VLAN's with the same MACs
2449 	 * on several interfaces from the same bridge. This also will save
2450 	 * some CPU cycles in case the destination interface and the input
2451 	 * interface (eq ifp) are the same.
2452 	 */
2453 	do { GRAB_OUR_PACKETS(ifp) } while (0);
2454 
2455 	/* Now check the all bridge members. */
2456 	LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) {
2457 		GRAB_OUR_PACKETS(bif2->bif_ifp)
2458 	}
2459 
2460 #undef OR_CARP_CHECK_WE_ARE_DST
2461 #undef OR_CARP_CHECK_WE_ARE_SRC
2462 #undef OR_PFIL_HOOKED_INET6
2463 #undef GRAB_OUR_PACKETS
2464 
2465 	/* Perform the bridge forwarding function. */
2466 	bridge_forward(sc, bif, m);
2467 
2468 	return (NULL);
2469 }
2470 
2471 /*
2472  * bridge_broadcast:
2473  *
2474  *	Send a frame to all interfaces that are members of
2475  *	the bridge, except for the one on which the packet
2476  *	arrived.
2477  *
2478  *	NOTE: Releases the lock on return.
2479  */
2480 static void
2481 bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if,
2482     struct mbuf *m, int runfilt)
2483 {
2484 	struct bridge_iflist *dbif, *sbif;
2485 	struct mbuf *mc;
2486 	struct ifnet *dst_if;
2487 	int error = 0, used = 0, i;
2488 
2489 	sbif = bridge_lookup_member_if(sc, src_if);
2490 
2491 	BRIDGE_LOCK2REF(sc, error);
2492 	if (error) {
2493 		m_freem(m);
2494 		return;
2495 	}
2496 
2497 	/* Filter on the bridge interface before broadcasting */
2498 	if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2499 #ifdef INET6
2500 	    || PFIL_HOOKED(&V_inet6_pfil_hook)
2501 #endif
2502 	    )) {
2503 		if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0)
2504 			goto out;
2505 		if (m == NULL)
2506 			goto out;
2507 	}
2508 
2509 	LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) {
2510 		dst_if = dbif->bif_ifp;
2511 		if (dst_if == src_if)
2512 			continue;
2513 
2514 		/* Private segments can not talk to each other */
2515 		if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE))
2516 			continue;
2517 
2518 		if ((dbif->bif_flags & IFBIF_STP) &&
2519 		    dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING)
2520 			continue;
2521 
2522 		if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 &&
2523 		    (m->m_flags & (M_BCAST|M_MCAST)) == 0)
2524 			continue;
2525 
2526 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2527 			continue;
2528 
2529 		if (LIST_NEXT(dbif, bif_next) == NULL) {
2530 			mc = m;
2531 			used = 1;
2532 		} else {
2533 			mc = m_dup(m, M_NOWAIT);
2534 			if (mc == NULL) {
2535 				if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2536 				continue;
2537 			}
2538 		}
2539 
2540 		/*
2541 		 * Filter on the output interface. Pass a NULL bridge interface
2542 		 * pointer so we do not redundantly filter on the bridge for
2543 		 * each interface we broadcast on.
2544 		 */
2545 		if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook)
2546 #ifdef INET6
2547 		    || PFIL_HOOKED(&V_inet6_pfil_hook)
2548 #endif
2549 		    )) {
2550 			if (used == 0) {
2551 				/* Keep the layer3 header aligned */
2552 				i = min(mc->m_pkthdr.len, max_protohdr);
2553 				mc = m_copyup(mc, i, ETHER_ALIGN);
2554 				if (mc == NULL) {
2555 					if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2556 					continue;
2557 				}
2558 			}
2559 			if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0)
2560 				continue;
2561 			if (mc == NULL)
2562 				continue;
2563 		}
2564 
2565 		bridge_enqueue(sc, dst_if, mc);
2566 	}
2567 	if (used == 0)
2568 		m_freem(m);
2569 
2570 out:
2571 	BRIDGE_UNREF(sc);
2572 }
2573 
2574 /*
2575  * bridge_span:
2576  *
2577  *	Duplicate a packet out one or more interfaces that are in span mode,
2578  *	the original mbuf is unmodified.
2579  */
2580 static void
2581 bridge_span(struct bridge_softc *sc, struct mbuf *m)
2582 {
2583 	struct bridge_iflist *bif;
2584 	struct ifnet *dst_if;
2585 	struct mbuf *mc;
2586 
2587 	if (LIST_EMPTY(&sc->sc_spanlist))
2588 		return;
2589 
2590 	LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) {
2591 		dst_if = bif->bif_ifp;
2592 
2593 		if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0)
2594 			continue;
2595 
2596 		mc = m_copypacket(m, M_NOWAIT);
2597 		if (mc == NULL) {
2598 			if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1);
2599 			continue;
2600 		}
2601 
2602 		bridge_enqueue(sc, dst_if, mc);
2603 	}
2604 }
2605 
2606 /*
2607  * bridge_rtupdate:
2608  *
2609  *	Add a bridge routing entry.
2610  */
2611 static int
2612 bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan,
2613     struct bridge_iflist *bif, int setflags, uint8_t flags)
2614 {
2615 	struct bridge_rtnode *brt;
2616 	int error;
2617 
2618 	BRIDGE_LOCK_ASSERT(sc);
2619 
2620 	/* Check the source address is valid and not multicast. */
2621 	if (ETHER_IS_MULTICAST(dst) ||
2622 	    (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 &&
2623 	     dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0)
2624 		return (EINVAL);
2625 
2626 	/* 802.1p frames map to vlan 1 */
2627 	if (vlan == 0)
2628 		vlan = 1;
2629 
2630 	/*
2631 	 * A route for this destination might already exist.  If so,
2632 	 * update it, otherwise create a new one.
2633 	 */
2634 	if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) {
2635 		if (sc->sc_brtcnt >= sc->sc_brtmax) {
2636 			sc->sc_brtexceeded++;
2637 			return (ENOSPC);
2638 		}
2639 		/* Check per interface address limits (if enabled) */
2640 		if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) {
2641 			bif->bif_addrexceeded++;
2642 			return (ENOSPC);
2643 		}
2644 
2645 		/*
2646 		 * Allocate a new bridge forwarding node, and
2647 		 * initialize the expiration time and Ethernet
2648 		 * address.
2649 		 */
2650 		brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO);
2651 		if (brt == NULL)
2652 			return (ENOMEM);
2653 
2654 		if (bif->bif_flags & IFBIF_STICKY)
2655 			brt->brt_flags = IFBAF_STICKY;
2656 		else
2657 			brt->brt_flags = IFBAF_DYNAMIC;
2658 
2659 		memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN);
2660 		brt->brt_vlan = vlan;
2661 
2662 		if ((error = bridge_rtnode_insert(sc, brt)) != 0) {
2663 			uma_zfree(bridge_rtnode_zone, brt);
2664 			return (error);
2665 		}
2666 		brt->brt_dst = bif;
2667 		bif->bif_addrcnt++;
2668 	}
2669 
2670 	if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC &&
2671 	    brt->brt_dst != bif) {
2672 		brt->brt_dst->bif_addrcnt--;
2673 		brt->brt_dst = bif;
2674 		brt->brt_dst->bif_addrcnt++;
2675 	}
2676 
2677 	if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2678 		brt->brt_expire = time_uptime + sc->sc_brttimeout;
2679 	if (setflags)
2680 		brt->brt_flags = flags;
2681 
2682 	return (0);
2683 }
2684 
2685 /*
2686  * bridge_rtlookup:
2687  *
2688  *	Lookup the destination interface for an address.
2689  */
2690 static struct ifnet *
2691 bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2692 {
2693 	struct bridge_rtnode *brt;
2694 
2695 	BRIDGE_LOCK_ASSERT(sc);
2696 
2697 	if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL)
2698 		return (NULL);
2699 
2700 	return (brt->brt_ifp);
2701 }
2702 
2703 /*
2704  * bridge_rttrim:
2705  *
2706  *	Trim the routine table so that we have a number
2707  *	of routing entries less than or equal to the
2708  *	maximum number.
2709  */
2710 static void
2711 bridge_rttrim(struct bridge_softc *sc)
2712 {
2713 	struct bridge_rtnode *brt, *nbrt;
2714 
2715 	BRIDGE_LOCK_ASSERT(sc);
2716 
2717 	/* Make sure we actually need to do this. */
2718 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2719 		return;
2720 
2721 	/* Force an aging cycle; this might trim enough addresses. */
2722 	bridge_rtage(sc);
2723 	if (sc->sc_brtcnt <= sc->sc_brtmax)
2724 		return;
2725 
2726 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2727 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2728 			bridge_rtnode_destroy(sc, brt);
2729 			if (sc->sc_brtcnt <= sc->sc_brtmax)
2730 				return;
2731 		}
2732 	}
2733 }
2734 
2735 /*
2736  * bridge_timer:
2737  *
2738  *	Aging timer for the bridge.
2739  */
2740 static void
2741 bridge_timer(void *arg)
2742 {
2743 	struct bridge_softc *sc = arg;
2744 
2745 	BRIDGE_LOCK_ASSERT(sc);
2746 
2747 	bridge_rtage(sc);
2748 
2749 	if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING)
2750 		callout_reset(&sc->sc_brcallout,
2751 		    bridge_rtable_prune_period * hz, bridge_timer, sc);
2752 }
2753 
2754 /*
2755  * bridge_rtage:
2756  *
2757  *	Perform an aging cycle.
2758  */
2759 static void
2760 bridge_rtage(struct bridge_softc *sc)
2761 {
2762 	struct bridge_rtnode *brt, *nbrt;
2763 
2764 	BRIDGE_LOCK_ASSERT(sc);
2765 
2766 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2767 		if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) {
2768 			if (time_uptime >= brt->brt_expire)
2769 				bridge_rtnode_destroy(sc, brt);
2770 		}
2771 	}
2772 }
2773 
2774 /*
2775  * bridge_rtflush:
2776  *
2777  *	Remove all dynamic addresses from the bridge.
2778  */
2779 static void
2780 bridge_rtflush(struct bridge_softc *sc, int full)
2781 {
2782 	struct bridge_rtnode *brt, *nbrt;
2783 
2784 	BRIDGE_LOCK_ASSERT(sc);
2785 
2786 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2787 		if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
2788 			bridge_rtnode_destroy(sc, brt);
2789 	}
2790 }
2791 
2792 /*
2793  * bridge_rtdaddr:
2794  *
2795  *	Remove an address from the table.
2796  */
2797 static int
2798 bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2799 {
2800 	struct bridge_rtnode *brt;
2801 	int found = 0;
2802 
2803 	BRIDGE_LOCK_ASSERT(sc);
2804 
2805 	/*
2806 	 * If vlan is zero then we want to delete for all vlans so the lookup
2807 	 * may return more than one.
2808 	 */
2809 	while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) {
2810 		bridge_rtnode_destroy(sc, brt);
2811 		found = 1;
2812 	}
2813 
2814 	return (found ? 0 : ENOENT);
2815 }
2816 
2817 /*
2818  * bridge_rtdelete:
2819  *
2820  *	Delete routes to a speicifc member interface.
2821  */
2822 static void
2823 bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full)
2824 {
2825 	struct bridge_rtnode *brt, *nbrt;
2826 
2827 	BRIDGE_LOCK_ASSERT(sc);
2828 
2829 	LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) {
2830 		if (brt->brt_ifp == ifp && (full ||
2831 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC))
2832 			bridge_rtnode_destroy(sc, brt);
2833 	}
2834 }
2835 
2836 /*
2837  * bridge_rtable_init:
2838  *
2839  *	Initialize the route table for this bridge.
2840  */
2841 static void
2842 bridge_rtable_init(struct bridge_softc *sc)
2843 {
2844 	int i;
2845 
2846 	sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE,
2847 	    M_DEVBUF, M_WAITOK);
2848 
2849 	for (i = 0; i < BRIDGE_RTHASH_SIZE; i++)
2850 		LIST_INIT(&sc->sc_rthash[i]);
2851 
2852 	sc->sc_rthash_key = arc4random();
2853 	LIST_INIT(&sc->sc_rtlist);
2854 }
2855 
2856 /*
2857  * bridge_rtable_fini:
2858  *
2859  *	Deconstruct the route table for this bridge.
2860  */
2861 static void
2862 bridge_rtable_fini(struct bridge_softc *sc)
2863 {
2864 
2865 	KASSERT(sc->sc_brtcnt == 0,
2866 	    ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt));
2867 	free(sc->sc_rthash, M_DEVBUF);
2868 }
2869 
2870 /*
2871  * The following hash function is adapted from "Hash Functions" by Bob Jenkins
2872  * ("Algorithm Alley", Dr. Dobbs Journal, September 1997).
2873  */
2874 #define	mix(a, b, c)							\
2875 do {									\
2876 	a -= b; a -= c; a ^= (c >> 13);					\
2877 	b -= c; b -= a; b ^= (a << 8);					\
2878 	c -= a; c -= b; c ^= (b >> 13);					\
2879 	a -= b; a -= c; a ^= (c >> 12);					\
2880 	b -= c; b -= a; b ^= (a << 16);					\
2881 	c -= a; c -= b; c ^= (b >> 5);					\
2882 	a -= b; a -= c; a ^= (c >> 3);					\
2883 	b -= c; b -= a; b ^= (a << 10);					\
2884 	c -= a; c -= b; c ^= (b >> 15);					\
2885 } while (/*CONSTCOND*/0)
2886 
2887 static __inline uint32_t
2888 bridge_rthash(struct bridge_softc *sc, const uint8_t *addr)
2889 {
2890 	uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key;
2891 
2892 	b += addr[5] << 8;
2893 	b += addr[4];
2894 	a += addr[3] << 24;
2895 	a += addr[2] << 16;
2896 	a += addr[1] << 8;
2897 	a += addr[0];
2898 
2899 	mix(a, b, c);
2900 
2901 	return (c & BRIDGE_RTHASH_MASK);
2902 }
2903 
2904 #undef mix
2905 
2906 static int
2907 bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b)
2908 {
2909 	int i, d;
2910 
2911 	for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) {
2912 		d = ((int)a[i]) - ((int)b[i]);
2913 	}
2914 
2915 	return (d);
2916 }
2917 
2918 /*
2919  * bridge_rtnode_lookup:
2920  *
2921  *	Look up a bridge route node for the specified destination. Compare the
2922  *	vlan id or if zero then just return the first match.
2923  */
2924 static struct bridge_rtnode *
2925 bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan)
2926 {
2927 	struct bridge_rtnode *brt;
2928 	uint32_t hash;
2929 	int dir;
2930 
2931 	BRIDGE_LOCK_ASSERT(sc);
2932 
2933 	hash = bridge_rthash(sc, addr);
2934 	LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) {
2935 		dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr);
2936 		if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0))
2937 			return (brt);
2938 		if (dir > 0)
2939 			return (NULL);
2940 	}
2941 
2942 	return (NULL);
2943 }
2944 
2945 /*
2946  * bridge_rtnode_insert:
2947  *
2948  *	Insert the specified bridge node into the route table.  We
2949  *	assume the entry is not already in the table.
2950  */
2951 static int
2952 bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt)
2953 {
2954 	struct bridge_rtnode *lbrt;
2955 	uint32_t hash;
2956 	int dir;
2957 
2958 	BRIDGE_LOCK_ASSERT(sc);
2959 
2960 	hash = bridge_rthash(sc, brt->brt_addr);
2961 
2962 	lbrt = LIST_FIRST(&sc->sc_rthash[hash]);
2963 	if (lbrt == NULL) {
2964 		LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash);
2965 		goto out;
2966 	}
2967 
2968 	do {
2969 		dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr);
2970 		if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan)
2971 			return (EEXIST);
2972 		if (dir > 0) {
2973 			LIST_INSERT_BEFORE(lbrt, brt, brt_hash);
2974 			goto out;
2975 		}
2976 		if (LIST_NEXT(lbrt, brt_hash) == NULL) {
2977 			LIST_INSERT_AFTER(lbrt, brt, brt_hash);
2978 			goto out;
2979 		}
2980 		lbrt = LIST_NEXT(lbrt, brt_hash);
2981 	} while (lbrt != NULL);
2982 
2983 #ifdef DIAGNOSTIC
2984 	panic("bridge_rtnode_insert: impossible");
2985 #endif
2986 
2987 out:
2988 	LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list);
2989 	sc->sc_brtcnt++;
2990 
2991 	return (0);
2992 }
2993 
2994 /*
2995  * bridge_rtnode_destroy:
2996  *
2997  *	Destroy a bridge rtnode.
2998  */
2999 static void
3000 bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt)
3001 {
3002 	BRIDGE_LOCK_ASSERT(sc);
3003 
3004 	LIST_REMOVE(brt, brt_hash);
3005 
3006 	LIST_REMOVE(brt, brt_list);
3007 	sc->sc_brtcnt--;
3008 	brt->brt_dst->bif_addrcnt--;
3009 	uma_zfree(bridge_rtnode_zone, brt);
3010 }
3011 
3012 /*
3013  * bridge_rtable_expire:
3014  *
3015  *	Set the expiry time for all routes on an interface.
3016  */
3017 static void
3018 bridge_rtable_expire(struct ifnet *ifp, int age)
3019 {
3020 	struct bridge_softc *sc = ifp->if_bridge;
3021 	struct bridge_rtnode *brt;
3022 
3023 	BRIDGE_LOCK(sc);
3024 
3025 	/*
3026 	 * If the age is zero then flush, otherwise set all the expiry times to
3027 	 * age for the interface
3028 	 */
3029 	if (age == 0)
3030 		bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN);
3031 	else {
3032 		LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) {
3033 			/* Cap the expiry time to 'age' */
3034 			if (brt->brt_ifp == ifp &&
3035 			    brt->brt_expire > time_uptime + age &&
3036 			    (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)
3037 				brt->brt_expire = time_uptime + age;
3038 		}
3039 	}
3040 	BRIDGE_UNLOCK(sc);
3041 }
3042 
3043 /*
3044  * bridge_state_change:
3045  *
3046  *	Callback from the bridgestp code when a port changes states.
3047  */
3048 static void
3049 bridge_state_change(struct ifnet *ifp, int state)
3050 {
3051 	struct bridge_softc *sc = ifp->if_bridge;
3052 	static const char *stpstates[] = {
3053 		"disabled",
3054 		"listening",
3055 		"learning",
3056 		"forwarding",
3057 		"blocking",
3058 		"discarding"
3059 	};
3060 
3061 	if (V_log_stp)
3062 		log(LOG_NOTICE, "%s: state changed to %s on %s\n",
3063 		    sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname);
3064 }
3065 
3066 /*
3067  * Send bridge packets through pfil if they are one of the types pfil can deal
3068  * with, or if they are ARP or REVARP.  (pfil will pass ARP and REVARP without
3069  * question.) If *bifp or *ifp are NULL then packet filtering is skipped for
3070  * that interface.
3071  */
3072 static int
3073 bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir)
3074 {
3075 	int snap, error, i, hlen;
3076 	struct ether_header *eh1, eh2;
3077 	struct ip *ip;
3078 	struct llc llc1;
3079 	u_int16_t ether_type;
3080 
3081 	snap = 0;
3082 	error = -1;	/* Default error if not error == 0 */
3083 
3084 #if 0
3085 	/* we may return with the IP fields swapped, ensure its not shared */
3086 	KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__));
3087 #endif
3088 
3089 	if (V_pfil_bridge == 0 && V_pfil_member == 0 && V_pfil_ipfw == 0)
3090 		return (0); /* filtering is disabled */
3091 
3092 	i = min((*mp)->m_pkthdr.len, max_protohdr);
3093 	if ((*mp)->m_len < i) {
3094 	    *mp = m_pullup(*mp, i);
3095 	    if (*mp == NULL) {
3096 		printf("%s: m_pullup failed\n", __func__);
3097 		return (-1);
3098 	    }
3099 	}
3100 
3101 	eh1 = mtod(*mp, struct ether_header *);
3102 	ether_type = ntohs(eh1->ether_type);
3103 
3104 	/*
3105 	 * Check for SNAP/LLC.
3106 	 */
3107 	if (ether_type < ETHERMTU) {
3108 		struct llc *llc2 = (struct llc *)(eh1 + 1);
3109 
3110 		if ((*mp)->m_len >= ETHER_HDR_LEN + 8 &&
3111 		    llc2->llc_dsap == LLC_SNAP_LSAP &&
3112 		    llc2->llc_ssap == LLC_SNAP_LSAP &&
3113 		    llc2->llc_control == LLC_UI) {
3114 			ether_type = htons(llc2->llc_un.type_snap.ether_type);
3115 			snap = 1;
3116 		}
3117 	}
3118 
3119 	/*
3120 	 * If we're trying to filter bridge traffic, don't look at anything
3121 	 * other than IP and ARP traffic.  If the filter doesn't understand
3122 	 * IPv6, don't allow IPv6 through the bridge either.  This is lame
3123 	 * since if we really wanted, say, an AppleTalk filter, we are hosed,
3124 	 * but of course we don't have an AppleTalk filter to begin with.
3125 	 * (Note that since pfil doesn't understand ARP it will pass *ALL*
3126 	 * ARP traffic.)
3127 	 */
3128 	switch (ether_type) {
3129 		case ETHERTYPE_ARP:
3130 		case ETHERTYPE_REVARP:
3131 			if (V_pfil_ipfw_arp == 0)
3132 				return (0); /* Automatically pass */
3133 			break;
3134 
3135 		case ETHERTYPE_IP:
3136 #ifdef INET6
3137 		case ETHERTYPE_IPV6:
3138 #endif /* INET6 */
3139 			break;
3140 		default:
3141 			/*
3142 			 * Check to see if the user wants to pass non-ip
3143 			 * packets, these will not be checked by pfil(9) and
3144 			 * passed unconditionally so the default is to drop.
3145 			 */
3146 			if (V_pfil_onlyip)
3147 				goto bad;
3148 	}
3149 
3150 	/* Run the packet through pfil before stripping link headers */
3151 	if (PFIL_HOOKED(&V_link_pfil_hook) && V_pfil_ipfw != 0 &&
3152 			dir == PFIL_OUT && ifp != NULL) {
3153 
3154 		error = pfil_run_hooks(&V_link_pfil_hook, mp, ifp, dir, NULL);
3155 
3156 		if (*mp == NULL || error != 0) /* packet consumed by filter */
3157 			return (error);
3158 	}
3159 
3160 	/* Strip off the Ethernet header and keep a copy. */
3161 	m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2);
3162 	m_adj(*mp, ETHER_HDR_LEN);
3163 
3164 	/* Strip off snap header, if present */
3165 	if (snap) {
3166 		m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1);
3167 		m_adj(*mp, sizeof(struct llc));
3168 	}
3169 
3170 	/*
3171 	 * Check the IP header for alignment and errors
3172 	 */
3173 	if (dir == PFIL_IN) {
3174 		switch (ether_type) {
3175 			case ETHERTYPE_IP:
3176 				error = bridge_ip_checkbasic(mp);
3177 				break;
3178 #ifdef INET6
3179 			case ETHERTYPE_IPV6:
3180 				error = bridge_ip6_checkbasic(mp);
3181 				break;
3182 #endif /* INET6 */
3183 			default:
3184 				error = 0;
3185 		}
3186 		if (error)
3187 			goto bad;
3188 	}
3189 
3190 	error = 0;
3191 
3192 	/*
3193 	 * Run the packet through pfil
3194 	 */
3195 	switch (ether_type) {
3196 	case ETHERTYPE_IP:
3197 		/*
3198 		 * Run pfil on the member interface and the bridge, both can
3199 		 * be skipped by clearing pfil_member or pfil_bridge.
3200 		 *
3201 		 * Keep the order:
3202 		 *   in_if -> bridge_if -> out_if
3203 		 */
3204 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3205 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3206 					dir, NULL);
3207 
3208 		if (*mp == NULL || error != 0) /* filter may consume */
3209 			break;
3210 
3211 		if (V_pfil_member && ifp != NULL)
3212 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp,
3213 					dir, NULL);
3214 
3215 		if (*mp == NULL || error != 0) /* filter may consume */
3216 			break;
3217 
3218 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3219 			error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp,
3220 					dir, NULL);
3221 
3222 		if (*mp == NULL || error != 0) /* filter may consume */
3223 			break;
3224 
3225 		/* check if we need to fragment the packet */
3226 		if (V_pfil_member && ifp != NULL && dir == PFIL_OUT) {
3227 			i = (*mp)->m_pkthdr.len;
3228 			if (i > ifp->if_mtu) {
3229 				error = bridge_fragment(ifp, *mp, &eh2, snap,
3230 					    &llc1);
3231 				return (error);
3232 			}
3233 		}
3234 
3235 		/* Recalculate the ip checksum. */
3236 		ip = mtod(*mp, struct ip *);
3237 		hlen = ip->ip_hl << 2;
3238 		if (hlen < sizeof(struct ip))
3239 			goto bad;
3240 		if (hlen > (*mp)->m_len) {
3241 			if ((*mp = m_pullup(*mp, hlen)) == 0)
3242 				goto bad;
3243 			ip = mtod(*mp, struct ip *);
3244 			if (ip == NULL)
3245 				goto bad;
3246 		}
3247 		ip->ip_sum = 0;
3248 		if (hlen == sizeof(struct ip))
3249 			ip->ip_sum = in_cksum_hdr(ip);
3250 		else
3251 			ip->ip_sum = in_cksum(*mp, hlen);
3252 
3253 		break;
3254 #ifdef INET6
3255 	case ETHERTYPE_IPV6:
3256 		if (V_pfil_bridge && dir == PFIL_OUT && bifp != NULL)
3257 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3258 					dir, NULL);
3259 
3260 		if (*mp == NULL || error != 0) /* filter may consume */
3261 			break;
3262 
3263 		if (V_pfil_member && ifp != NULL)
3264 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp,
3265 					dir, NULL);
3266 
3267 		if (*mp == NULL || error != 0) /* filter may consume */
3268 			break;
3269 
3270 		if (V_pfil_bridge && dir == PFIL_IN && bifp != NULL)
3271 			error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp,
3272 					dir, NULL);
3273 		break;
3274 #endif
3275 	default:
3276 		error = 0;
3277 		break;
3278 	}
3279 
3280 	if (*mp == NULL)
3281 		return (error);
3282 	if (error != 0)
3283 		goto bad;
3284 
3285 	error = -1;
3286 
3287 	/*
3288 	 * Finally, put everything back the way it was and return
3289 	 */
3290 	if (snap) {
3291 		M_PREPEND(*mp, sizeof(struct llc), M_NOWAIT);
3292 		if (*mp == NULL)
3293 			return (error);
3294 		bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc));
3295 	}
3296 
3297 	M_PREPEND(*mp, ETHER_HDR_LEN, M_NOWAIT);
3298 	if (*mp == NULL)
3299 		return (error);
3300 	bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN);
3301 
3302 	return (0);
3303 
3304 bad:
3305 	m_freem(*mp);
3306 	*mp = NULL;
3307 	return (error);
3308 }
3309 
3310 /*
3311  * Perform basic checks on header size since
3312  * pfil assumes ip_input has already processed
3313  * it for it.  Cut-and-pasted from ip_input.c.
3314  * Given how simple the IPv6 version is,
3315  * does the IPv4 version really need to be
3316  * this complicated?
3317  *
3318  * XXX Should we update ipstat here, or not?
3319  * XXX Right now we update ipstat but not
3320  * XXX csum_counter.
3321  */
3322 static int
3323 bridge_ip_checkbasic(struct mbuf **mp)
3324 {
3325 	struct mbuf *m = *mp;
3326 	struct ip *ip;
3327 	int len, hlen;
3328 	u_short sum;
3329 
3330 	if (*mp == NULL)
3331 		return (-1);
3332 
3333 	if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3334 		if ((m = m_copyup(m, sizeof(struct ip),
3335 			(max_linkhdr + 3) & ~3)) == NULL) {
3336 			/* XXXJRT new stat, please */
3337 			KMOD_IPSTAT_INC(ips_toosmall);
3338 			goto bad;
3339 		}
3340 	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
3341 		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
3342 			KMOD_IPSTAT_INC(ips_toosmall);
3343 			goto bad;
3344 		}
3345 	}
3346 	ip = mtod(m, struct ip *);
3347 	if (ip == NULL) goto bad;
3348 
3349 	if (ip->ip_v != IPVERSION) {
3350 		KMOD_IPSTAT_INC(ips_badvers);
3351 		goto bad;
3352 	}
3353 	hlen = ip->ip_hl << 2;
3354 	if (hlen < sizeof(struct ip)) { /* minimum header length */
3355 		KMOD_IPSTAT_INC(ips_badhlen);
3356 		goto bad;
3357 	}
3358 	if (hlen > m->m_len) {
3359 		if ((m = m_pullup(m, hlen)) == 0) {
3360 			KMOD_IPSTAT_INC(ips_badhlen);
3361 			goto bad;
3362 		}
3363 		ip = mtod(m, struct ip *);
3364 		if (ip == NULL) goto bad;
3365 	}
3366 
3367 	if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
3368 		sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
3369 	} else {
3370 		if (hlen == sizeof(struct ip)) {
3371 			sum = in_cksum_hdr(ip);
3372 		} else {
3373 			sum = in_cksum(m, hlen);
3374 		}
3375 	}
3376 	if (sum) {
3377 		KMOD_IPSTAT_INC(ips_badsum);
3378 		goto bad;
3379 	}
3380 
3381 	/* Retrieve the packet length. */
3382 	len = ntohs(ip->ip_len);
3383 
3384 	/*
3385 	 * Check for additional length bogosity
3386 	 */
3387 	if (len < hlen) {
3388 		KMOD_IPSTAT_INC(ips_badlen);
3389 		goto bad;
3390 	}
3391 
3392 	/*
3393 	 * Check that the amount of data in the buffers
3394 	 * is as at least much as the IP header would have us expect.
3395 	 * Drop packet if shorter than we expect.
3396 	 */
3397 	if (m->m_pkthdr.len < len) {
3398 		KMOD_IPSTAT_INC(ips_tooshort);
3399 		goto bad;
3400 	}
3401 
3402 	/* Checks out, proceed */
3403 	*mp = m;
3404 	return (0);
3405 
3406 bad:
3407 	*mp = m;
3408 	return (-1);
3409 }
3410 
3411 #ifdef INET6
3412 /*
3413  * Same as above, but for IPv6.
3414  * Cut-and-pasted from ip6_input.c.
3415  * XXX Should we update ip6stat, or not?
3416  */
3417 static int
3418 bridge_ip6_checkbasic(struct mbuf **mp)
3419 {
3420 	struct mbuf *m = *mp;
3421 	struct ip6_hdr *ip6;
3422 
3423 	/*
3424 	 * If the IPv6 header is not aligned, slurp it up into a new
3425 	 * mbuf with space for link headers, in the event we forward
3426 	 * it.  Otherwise, if it is aligned, make sure the entire base
3427 	 * IPv6 header is in the first mbuf of the chain.
3428 	 */
3429 	if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) {
3430 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3431 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
3432 			    (max_linkhdr + 3) & ~3)) == NULL) {
3433 			/* XXXJRT new stat, please */
3434 			IP6STAT_INC(ip6s_toosmall);
3435 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3436 			goto bad;
3437 		}
3438 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
3439 		struct ifnet *inifp = m->m_pkthdr.rcvif;
3440 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
3441 			IP6STAT_INC(ip6s_toosmall);
3442 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
3443 			goto bad;
3444 		}
3445 	}
3446 
3447 	ip6 = mtod(m, struct ip6_hdr *);
3448 
3449 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
3450 		IP6STAT_INC(ip6s_badvers);
3451 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
3452 		goto bad;
3453 	}
3454 
3455 	/* Checks out, proceed */
3456 	*mp = m;
3457 	return (0);
3458 
3459 bad:
3460 	*mp = m;
3461 	return (-1);
3462 }
3463 #endif /* INET6 */
3464 
3465 /*
3466  * bridge_fragment:
3467  *
3468  *	Return a fragmented mbuf chain.
3469  */
3470 static int
3471 bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh,
3472     int snap, struct llc *llc)
3473 {
3474 	struct mbuf *m0;
3475 	struct ip *ip;
3476 	int error = -1;
3477 
3478 	if (m->m_len < sizeof(struct ip) &&
3479 	    (m = m_pullup(m, sizeof(struct ip))) == NULL)
3480 		goto out;
3481 	ip = mtod(m, struct ip *);
3482 
3483 	m->m_pkthdr.csum_flags |= CSUM_IP;
3484 	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist);
3485 	if (error)
3486 		goto out;
3487 
3488 	/* walk the chain and re-add the Ethernet header */
3489 	for (m0 = m; m0; m0 = m0->m_nextpkt) {
3490 		if (error == 0) {
3491 			if (snap) {
3492 				M_PREPEND(m0, sizeof(struct llc), M_NOWAIT);
3493 				if (m0 == NULL) {
3494 					error = ENOBUFS;
3495 					continue;
3496 				}
3497 				bcopy(llc, mtod(m0, caddr_t),
3498 				    sizeof(struct llc));
3499 			}
3500 			M_PREPEND(m0, ETHER_HDR_LEN, M_NOWAIT);
3501 			if (m0 == NULL) {
3502 				error = ENOBUFS;
3503 				continue;
3504 			}
3505 			bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN);
3506 		} else
3507 			m_freem(m);
3508 	}
3509 
3510 	if (error == 0)
3511 		KMOD_IPSTAT_INC(ips_fragmented);
3512 
3513 	return (error);
3514 
3515 out:
3516 	if (m != NULL)
3517 		m_freem(m);
3518 	return (error);
3519 }
3520 
3521 static void
3522 bridge_linkstate(struct ifnet *ifp)
3523 {
3524 	struct bridge_softc *sc = ifp->if_bridge;
3525 	struct bridge_iflist *bif;
3526 
3527 	BRIDGE_LOCK(sc);
3528 	bif = bridge_lookup_member_if(sc, ifp);
3529 	if (bif == NULL) {
3530 		BRIDGE_UNLOCK(sc);
3531 		return;
3532 	}
3533 	bridge_linkcheck(sc);
3534 	BRIDGE_UNLOCK(sc);
3535 
3536 	bstp_linkstate(&bif->bif_stp);
3537 }
3538 
3539 static void
3540 bridge_linkcheck(struct bridge_softc *sc)
3541 {
3542 	struct bridge_iflist *bif;
3543 	int new_link, hasls;
3544 
3545 	BRIDGE_LOCK_ASSERT(sc);
3546 	new_link = LINK_STATE_DOWN;
3547 	hasls = 0;
3548 	/* Our link is considered up if at least one of our ports is active */
3549 	LIST_FOREACH(bif, &sc->sc_iflist, bif_next) {
3550 		if (bif->bif_ifp->if_capabilities & IFCAP_LINKSTATE)
3551 			hasls++;
3552 		if (bif->bif_ifp->if_link_state == LINK_STATE_UP) {
3553 			new_link = LINK_STATE_UP;
3554 			break;
3555 		}
3556 	}
3557 	if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) {
3558 		/* If no interfaces support link-state then we default to up */
3559 		new_link = LINK_STATE_UP;
3560 	}
3561 	if_link_state_change(sc->sc_ifp, new_link);
3562 }
3563